Effectiveness of Tax and Price

IARC
HANDBOOKS OF CANCER PREVENTION

Tobacco Control
Volume 14
Effectiveness of Tax and Price

Policies for Tobacco Control
2011
IARC HANDBOOKS OF CANCER PREVENTION
Tobacco Control
Volume 14
Effectiveness of Tax and Price

Policies for Tobacco Control
2011
This Handbook was produced within th EC-funded grant (HEALTH-F2-2009-223323) : Pricing Policies and Control of Tobacco in Europe (PPACTE)
International Agency for Research on Cancer
The International Agency for Research on Cancer (IARC) was established in 1965 by the World Health Assembly,
as an independently funded organization within the framework of the World Health Organization. The headquarters of
the Agency are in Lyon, France.
The Agency conducts a programme of research concentrating particularly on the epidemiology of cancer and the
study of potential carcinogens in the human environment. Its field studies are supplemented by biological and chemical
research carried out in the Agency’s laboratories in Lyon and, through collaborative research agreements, in national
research institutions in many countries. The Agency also conducts a programme for the education and training of
personnel for cancer research.
The publications of the Agency contribute to the dissemination of authoritative information on different aspects of
cancer research. Information about IARC publications, and how to order them, is available via the Internet at: http://
www.iarc.fr/en/publications/index.php.
This publication represents the views and opinions of an IARC Working Group on the Effectiveness of Tax and Price
Policies for Tobacco Control which met in Lyon, France, 17 May–22 May 2010.
The IARC Handbook Volume 14 was funded by the European Commission FP7 Grant Agreement
HEALTH-F2-2009-223323 (through the project ‘Pricing Policies and Control of Tobacco in Europe’ (PPACTE)).
i
IARC Handbooks of Cancer Prevention
Published by the International Agency for Research on Cancer,

150 cours Albert Thomas, 69372 Lyon Cedex 08, France
© International Agency for Research on Cancer, 2011
Distributed by
WHO Press, World Health Organization, 20 Avenue Appia, 1211 Geneva 27, Switzerland (tel: +41 22 791 3264;;
fax:+41 22 791 4857;; email: bookorders@who.int).
Publications of the World Health Organization enjoy copyright protection in accordance with the provisions
of Protocol 2 of the Universal Copyright Convention. All rights reserved.
The designations employed and the presentation of the material in this publication do not imply the expression of any
opinion whatsoever on the part of the Secretariat of the World Health Organization concerning the legal status of any
country, territory, city, or area or of its authorities, or concerning the delimitation of its frontiers or boundaries.
The mention of specific companies or of certain manufacturers’ products does not imply that they are endorsed
or recommended by the World Health Organization in preference to others of a similar nature that are not mentioned.
Errors and omissions excepted, the names of proprietary products are distinguished by initial capital letters.
The authors alone are responsible for the views expressed in this publication.
The International Agency for Research on Cancer welcomes requests for permission to reproduce or translate its
publications, in part or in full. Requests for permission to reproduce or translate IARC publications – whether for sale
or for noncommercial distribution – should be addressed to WHO Press, at the above address
(fax: +41 22 791 4806;; email: permissions@who.int).
Format for Bibliographic Citation
IARC Handbooks of Cancer Prevention, Tobacco Control, Vol. 14: Effectiveness of Tax and Price Policies for Tobacco
Control (2011: Lyon, France)
IARC Library Cataloguing-in-Publication Data
Effectiveness of tax and price policies for tobacco control / IARC Working Group on the Effectiveness of Tax
and Price Policies for Tobacco Control (2011 : Lyon, France)
(IARC Handbooks of Cancer Prevention ;; 14)
1. Neoplasms – prevention & control 2. Public Policy 3. Smoking – economics
4. Smoking – prevention & control 5. Taxes– legislation & jurisprudence
I. IARC Working Group the Effectiveness of Tax and Price Policies for Tobacco Control II. Series
ISSN 1027-5622 (NLM Classification QZ 39)

ISBN 978-92-832-3014-4
ii
Table of contents
List of participants......................................................................................................................................................................iv
Acknowledgements....................................................................................................................................................................vi
Preface.......................................................................................................................................................................................vii
Chapter 1
Overview of Handbook volume 14.............................................................................................................................................1
Chapter 2
Overview of tobacco taxation....................................................................................................................................................9
Chapter 3
Tobacco industry pricing, price-related marketing and lobbying strategies........................................................................31
Chapter 4
Tax, price and aggregate demand for tobacco products.......................................................................................................91
Chapter 5
Tax, price and adult tobacco use...........................................................................................................................................137
Chapter 6
Tax, price and tobacco use among young people...............................................................................................................201
Chapter 7
Tax, price and tobacco use among the poor........................................................................................................................259
Chapter 8
Tax avoidance and tax evasion.............................................................................................................................................297
Chapter 9
Health and economic impact of tobacco taxation................................................................................................................327
Chapter 10
Summary.................................................................................................................................................................................349
Chapter 11
Evaluation................................................................................................................................................................................355
Chapter 12
Recommendations..................................................................................................................................................................357
Working procedures...............................................................................................................................................................359
iii
List of participants
Frank Chaloupka (Chair) Silvano Gallus Carlo La Vecchia

Economics, College of Business Department of Epidemiology Department of Epidemiology
Administration Istituto di Ricerche Farmacologiche Istituto di Ricerche Farmacologiche
Health Policy and Administration Mario Negri Mario Negri
University of Illinois at Chicago Via La Masa, 19 Via La Masa, 19
601 S. Morgan St, Room 2103 20156 Milano 20156 Milan
Chicago, Il 60607–7121 Italy Italy
United States of America
Anna Gilmore David Levy
Evan Blecher School for Health HBSA of Pacific Institute for
American Cancer Society University of Bath Research and Evaluation
250 Williams Street, NW Claverton Down 11720 Beltsville Dr, Suite 900
Atlanta, Georgia 30303–1002 Bath BA2 7AY Calverton 20705–3102
United States of America United Kingdom United States of America
Luke Clancy Fiona Godfrey Nigar Nargis

Tobacco Free Research Institute The Union (IUATLD) Department of Economics
Digital Depot 88 rue des Sources University of Dhaka
Thomas Street Weimerskirch Dhaka-1000,
Dublin 8 L2542 Luxembourg Bangladesh
Ireland Luxembourg
Lien Nguyen
Laura Currie Teh-Wei Hu Centre for Health and Social
Tobacco Free Research Institute 214 University Hall Economics – CHESS
Digital Depot School of Public Health National Institute for Health and
Thomas Street University of California Welfare
Dublin 8 Berkeley, CA PO Box 30, FI-00271 Helsinki,
Ireland United States of America Finland
Sofia Delipalla Roberto Iglesias (Unable to attend) Zeynep Onder

Tobacco Free Initiative Aliança do Controle do Tabagismo/ Bilkent University
World Health Organization Alliance for Tobacco Control Faculty of Business Administration
20, avenue Appia Avenida Nossa Senhora de Bilkent, Ankara 06800
CH-1211 Geneva 27 Copacabana 330/404 Turkey
Switzerland 22020–001 Rio de Janeiro – RJ
Brazil Markku Pekurinen
Esteve Fernandez (Unable to attend)
Tobacco Research Unit Luk Joossens Centre for Health and Social
Cancer Prevention and Control Framework Convention Alliance Economics – CHESS
Department Belgian Foundation against Cancer National Institute for Health and
Catalan Institute of Oncology 479 Chaussée de Louvain Welfare
Av. Gran Via s/n Km. 2.7 B-1030 Brussels PO Box 30, FI-00271 Helsinki,
08907 L’Hospitalet (Barcelona) Belgium Finland
Spain
iv
List of participants
Anne-Marie Perucic Frank van Driessche1 IARC Secretariat

Tobacco Free Initiative European Commission
World Health Organization Directorate general for Taxation and John Daniel, Editor
20, avenue Appia the Customs Union Sharon Grant, Copyright
CH-1211 Geneva 27 Excise duties and transport, Farhad Islami, Rapporteur
Switzerland environment and energy taxes Nina Kamstrup-Larsen (Post-
Rue de la loi 200 meeting)
Hana Ross B 1049 Brussels Maria León, Responsible Officer
International Tobacco Control Belgium Qian Li, Rapporteur
Research Joachim Schüz (Section Head)
American Cancer Society Corné van Walbeek Kurt Straif
250 Williams Street, NW School of Economics
Atlanta, GA 30303–1002 Room 4.75 Leslie Social Science Technical and administrative
United States of America Building assistance
University of Cape Town
Katherine Smith (Unable to attend) Cape Town, Christine Bassier (Secretarial)
School for Health South Africa François Deloche (Bibliography)
University of Bath Roland Dray (Photography)
Bath BA2 7AY Ayda Yurekli (Unable to attend) Sylvia Moutinho (Bibliography)
United Kingdom Tobacco Free Initiative Stéphanie Royannais (Layout)
World Health Organization
John A. Tauras 20, avenue Appia
Department of Economics CH-1211 Geneva 27 1
Invited specialist (an invited specialist is an expert in
University of Illinois at Chicago Switzerland the subject under assessment, but also has declared
601 S. Morgan a conflicting interest. Under such circumstances, the
Chicago, IL 60607–7121 expert does not participate in the evaluation process
United States of America while his expertise guides the description of general
remarks on the intervention under assessment).
v
Acknowledgements
The Working Group acknowledges the invaluable help received during the gathering of the published evidence and the
preparation of the working drafts from Michal Stoklosa and Naw Htee Khu (International Tobacco Control Research at
the American Cancer Society), Cathy Flower and Lizz Jennings (University of Bath) and Lyn Reed (University of Cape
Town).
The Working Group is grateful to Sylvia Moutinho (Communications Group, IARC) for her continued assistance with the
database of references prior, during and after the Handbook Meeting.
vi
Preface
Tobacco smoking causes cancer to 7R UHSRUW SHULRGLFDOO\ WR WKH peer-reviewed the evidence in an
over 15 organ sites, and exposure Conference of the Parties the interval of approximately 8 months
to secondhand smoke and parental country-specific rates of tobacco and later finalized the draft chapters
smoking cause cancer in non- taxes and the trends in tobacco in a six-day meeting of experts in
smokers and in the offspring use. Lyon, France in May 2010. The
(Secretan et al., 2009). Tobacco use Adherence to the treaty will main chapters present the evidence,
represents the largest preventable induce countries to revise existent and include tables presenting key
cause of cancer worldwide. In tobacco control policies and set descriptors of the studies reviewed
particular, tobacco smoking is priorities, including the use of taxes. on the effect of taxes on aggregated
pandemic and covers all ages, Handbook Volume 14 can inform demand for tobacco, adult tobacco
affecting over a billion people. The policy-makers in those countries to use, use among young people
eradication of tobacco use can only fulfil the demands of WHO FCTC and use among the poor. Volume
be achieved by preventing children Article 6. 14 includes a summary chapter
and adolescents from starting use The use of taxes to increase the describing the key findings and
today. Quitting smoking will reduce retail price of tobacco products can conclusions from each chapter. The
disease and mortality in a shorter affect both initiation and cessation. evaluation and recommendations
time span, as the risk of several The examination of the effectiveness for research and public health are
cancers decreases with increasing of this intervention for tobacco control presented in additional, succinct
time since cessation even after is the central theme of Volume 14 chapters.
several decades of smoking (IARC, in the IARC Handbooks of Cancer Two WHO publications provide
2007). Prevention series. complementary data regarding the
Article 6 of the World Health IARC Handbook Volume 14 use of taxes on tobacco products
Organization Framework Convention presents an evidence-based as a tobacco control intervention.
on Tobacco Control (WHO FCTC) is evaluation of the literature published The WHO Report on the Global
one of the instruments in the treaty up to May 2010 on the effectiveness Tobacco Epidemic (2008) presents
aimed at reducing the demand of of tax and price policies in reducing the MPOWER package, a set of
tobacco use in the population (WHO, the prevalence and consumption of six policy interventions, including
2005). It directs ratifying nations to tobacco use (Chaloupka et al., 2011). increases in taxes, promoted to
adopt three measures: The volume also covers the impact of reduce tobacco use and associated
7R LPSOHPHQW WD[ SROLFLHV DQG cross-border shopping, smuggling, mortality. This report includes
if appropriate, price policies on and the tobacco industry’s use of the level of implementation of tax
tobacco products to achieve a discount prices and other strategies policies across the world, including
reduction in the consumption of on reducing the effectiveness of the price of the most popular brand
tobacco in the population;; taxes. The volume was authored of cigarettes, the amount of taxes as
7RSURKLELWRUOLPLWWKHVDOHDQG by a Working Group of experts a percent of the retail price, the type
importation of tax-free and duty- from 12 countries who gathered, of tax applied and the affordability of
free tobacco products;; and lastly, critically analysed, synthesized and the most popular brand. A second
vii
publication, WHO Technical Manual References

on Tobacco Tax Administration,
focuses on tax administration and
Chaloupka FJ, Straif K, Leon ME (2011).
offers the best practices to countries Effectiveness of tax and price policies in
World Health Organization (2005). WHO
around the world that have signed tobacco control. Tob Control, 20:235–238.
Framework Convention on Tobacco Control.
doi:10.1136/tc.2010.039982 PMID:21115556
the WHO-FCTC and are required Geneva, World Health Organization.
to implement effective tax and price IARC (2007). IARC Handbooks of Cancer
World Health Organization (2008). WHO report
Prevention, Tobacco Control, Volume 11:
policies (WHO, 2010). Reversal of Risk After Quitting Smoking. Lyon,
on the global tobacco epidemic 2008: the
MPOWER package. Geneva, World Health
This IARC Handbook provides International Agency for Research on Cancer.
Organization.
an in-depth and up-to-date critical
Secretan B, Straif K, Baan R et al.;; WHO
review on the effectiveness of various World Health Organization (2010).
International Agency for Research on Cancer
WHO Technical Manual on Tobacco Tax
tax-related interventions to curb the Monograph Working Group (2009). A review
Administration. Geneva, World Health
of human carcinogens–Part E: tobacco, areca
global tobacco epidemic. Together Organization.
nut, alcohol, coal smoke, and salted fish.
with the two WHO publications this Lancet Oncol, 10:1033–1034.doi:10.1016/
S1470-2045(09)70326-2 PMID:19891056
will allow policy makers to base their
decisions on the latest scientific
evidence and choose the most
effective interventions.
viii
Chapter 1
Overview of Handbook Volume 14
The scientific evaluation of the effective and important means of “all forms of illicit trade in tobacco
effectiveness of cancer prevention reducing tobacco consumption products, including smuggling, illicit
interventions is part of the mission of by various segments of the manufacturing and counterfeiting”
the International Agency for Research population, in particular young (WHO, 2005).
on Cancer (IARC). This commitment persons. Maximizing the effectiveness of
is reflected in the IARC Handbooks 2. Without prejudice to the tax and price policies in achieving
of Cancer Prevention that focus on sovereign right of the Parties reductions in tobacco use and its
the evaluation of tobacco control to determine and establish health consequences requires a
policies. Handbook Volume 12 was their taxation policies, each clear understanding of the impact
the first in the Tobacco Control series Party should take account of of these policies on initiation and
providing a framework for evaluating its national health objectives escalation of tobacco use, cessation
tobacco control policies (IARC, concerning tobacco control and and relapse, tobacco product
2008). Handbook Volume 13 was adopt or maintain, as appropriate, consumption, substitutability of
the first of the series to focus on a measures which may include: tobacco products, tax avoidance
specific policy, with a comprehensive a. Implementing tax policies and evasion, and related outcomes.
evaluation of the effectiveness of and, where appropriate, price At the same time, those making or
smoke-free legislations in reducing policies, on tobacco products advocating tobacco tax and price
exposure to secondhand tobacco so as to contribute to the health policies must understand the impact
smoke (SHS), tobacco use among objectives aimed at reducing of these policies on other outcomes,
youth and adults, and other outcomes tobacco consumption;; and including their impact on government
(IARC, 2009). b. Prohibiting or restricting, as revenues, employment and inflation.
This Handbook, Volume 14, appropriate, sales to and/or This Handbook provides a thorough
extends the discussion on the importations by international review of this evidence.
impact of public policies on a major travellers of tax- and duty-free
cause of cancer—tobacco use— tobacco products. Price of tobacco and tobacco use
by presenting, summarizing and 3. The Parties shall provide rates of
evaluating the research evidence taxation for tobacco products and One of the most fundamental
on the effectiveness of tax and price trends in tobacco consumption laws of economics is that of the
policies for tobacco control. Tax and in their periodic reports to the downward-sloping demand curve,
price policies are central to the World Conference of the Parties in which states that increases in the
Health Organization’s Framework accordance with Article 21. price of a given product will lead to
Convention on Tobacco Control Given the potential role of tax and reductions in the quantity demanded
(WHO FCTC). Article 6 of the treaty, price differentials in contributing to of that product, while reductions in
the first to address a specific tobacco tax avoidance and evasion, tax and price will lead to increases in quantity
control policy, states (WHO, 2005): price policies are further addressed demanded. The extensive empirical
1. The Parties recognize that in Article 15 of the WHO FCTC, research on the demand for tobacco
price and tax measures are an which calls on countries to eliminate products confirms that the law of the
1
downward-sloping demand curve structure of the market (monopolistic, cross-sectional collection is the
applies to tobacco products. Much oligopolistic or competitive) as well lack of price variation within most
of this research comes from the as the business and political agenda countries, resulting in an inability
United States and other high-income of tobacco companies. to estimate the price elasticity of
countries, but over the past decade, The availability and quality of demand. Employing self-reported
numerous studies of the demand for data present another challenge, cigarette prices collected as part of
tobacco products have been done in particularly for low- and middle- the survey and which usually exhibit
low- and middle-income countries. income countries. Aggregate time- some degree of variation is not an
Higher prices influence the series data on cigarette sales are ideal option, as the price an individual
demand for tobacco products in most readily available, even in low- pays is likely to be related to their
two ways. First, they reduce the resource countries. However, since smoking behaviour, potentially
prevalence of tobacco use by they represent tax-paid cigarette biasing the resulting estimate of price
discouraging non-users from taking sales rather than actual consumption, elasticity. Individual-level data thus
up tobacco use, by encouraging the presence of cross-border should be augmented with externally
existing users to quit, and by helping shopping, cigarette smuggling, illicit collected data on price, tobacco
former users stay quit. Second, production and other forms of tax control public policies and other
higher tobacco prices reduce the avoidance and evasion can bias important determinants of demand.
consumption of tobacco products upward the estimates of the effects In addition, individual-level data can
among those who continue to use of taxes and prices on cigarette also suffer from reporting biases
tobacco after a price increase. demand. such as underreporting of tobacco
The strength of these responses is Another complication when consumption.
measured by the price elasticity of analysing aggregate data results from These and other challenges in
the demand for tobacco products, the fact that tobacco product prices estimating the impact of taxes and
which is defined as the percentage are determined by the interaction prices on tobacco use are described
change in consumption that results of supply and demand. If this is not in more detail in the following
from a 1% price increase. The accounted for, the resulting estimates chapters. Many of the methodological
seminal World Bank publication of the price elasticity of demand can challenges in estimating the impact of
Curbing the Epidemic concluded be biased. In addition, macro-level tobacco prices on tobacco use have
that the price elasticity of cigarette studies cannot distinguish between been overcome by employing diverse
GHPDQGLVDURXQGíIRUGHYHORSHG several behavioural changes that and sophisticated econometric and
FRXQWULHV DQG EHWZHHQ í DQG lead to change in tobacco demand, other statistical methods. In addition,
í IRU GHYHORSLQJ FRXQWULHV -KD such as initiation, cessation or some recent data (such as the CDC/
and Chaloupka, 1999). change in quantity of tobacco WHO Global Adult Tobacco Surveys
There are numerous product consumed. Neither can and the International Tobacco Control
methodological challenges in these studies examine differences Policy Evaluation Project surveys)
estimating the impact of tobacco in responsiveness to changes in have been collected specifically
prices on tobacco use. Given the wide price among different population for studying tobacco use and the
range of tobacco products, selecting subgroups defined by age, gender, impact of tobacco control policies,
and/or developing an appropriate race/ethnicity, socioeconomic status which helps to overcome the typical
price measure can be difficult. Some and other characteristics of the challenges of modeling the demand
studies have used tobacco product population. for tobacco products.
excise tax rates as proxies for Individual-level data collected
tobacco product prices, because tax in surveys overcome some of the Tobacco tax as a public policy
in most countries represents a larger challenges associated with the
share of retail price and is also the use of aggregate data, but they Nearly every country in the world
primary policy tool for manipulating are more expensive to gather, so taxes tobacco products. Almost all
tobacco product prices. However, their availability (and, if available, countries levy excise or other taxes
a tax increase may not always their quality) can be limited in that are specific to tobacco products,
result in a corresponding change in lower-resource countries. The many apply duties on imported
tobacco retail prices. The degree most common problem with using tobacco products, and many apply
of pass-through depends on the individual-level data from a single value-added or sales taxes on these
2
products. In 1776, Adam Smith, Over time, as the research inelasticity of demand for tobacco
the father of modern economics, evidence demonstrating that higher products and the relatively low share
discussed the appropriateness of tobacco product taxes and prices of tax in price in most countries mean
taxing tobacco products in his classic reduce tobacco use has accumulated, that significant increases in tobacco
volume An Inquiry into the Nature the role of tobacco taxes has changed taxes will generate significant
and Causes of the Wealth of Nations: and increasingly these taxes are increases in government revenues.
“Sugar, rum and tobacco used as a policy tool to improve Advances in tobacco farming and
are commodities which public health by reducing tobacco tobacco product manufacturing lead
are no where necessaries consumption and accounting for the to job losses in the tobacco sector
of life, which are become external cost of smoking. A significant during periods of stable tobacco
objects of almost universal increase in tobacco product taxes and taxes, while tobacco-dependent jobs
consumption, and which are prices has been demonstrated to be lost as a result of higher taxes are
therefore extremely proper the single most effective and cost- replaced by jobs in other sectors as
subjects of taxation. [...] In effective intervention for reducing those deterred from using tobacco
the mean time the people tobacco use, particularly among the products spend the money that they
might be relieved from some young and the poor;; thus the central once spent on tobacco on other goods
of the most burdensome role of tobacco tax and price policies and services and the government
taxes;; from those which are in the WHO FCTC. spends new tax revenues, creating
imposed either upon the jobs in other sectors.
necessaries of life, or upon Challenges of using tax Nevertheless, it is critical to gain
the materials of manufacture. and price to control tobacco use the necessary political support to
The labouring poor would increase tobacco taxes, because the
thus be enabled to live better, Despite the public health rationale industry opposes higher taxes most
to work cheaper, and to for increasing tobacco taxes to severely, as is evident for example
send their goods cheaper to reduce tobacco use and its health from examining documents found
market. The cheapness of and economic consequences, in legal discovery during lawsuits
their goods would increase some dispute the social benefits against the industry and made
the demand for them, and of this intervention. Opponents of publicly available. The strategy for
consequently for the labour higher tobacco taxes question their securing that support will vary from
of those who produced them. revenue-generating potential and the country to country. One strategy
This increase in the demand sustainability of these revenues. They used successfully in a growing
for labour, would both point out to the possible negative number of countries has been the
increase the numbers and economic impact of higher tobacco dedication of revenues from tobacco
improve the circumstances taxes, particularly when it comes taxes to other tobacco control or
of the labouring poor. Their to tobacco-related employment, health promotion activities, including
consumption would increase, the prosperity of sectors indirectly some specifically targeting the poor.
and together with it the associated with tobacco business, This Handbook reviews the evidence
revenue arising from all those upward pressure on inflation, the on the revenue and economic impact
articles of their consumption negative distributional impact of of tobacco taxes, their effects on the
upon which the taxes might higher tobacco taxes on the poor, poor, public support for these taxes,
be allowed to remain.” and the danger of tax avoidance and and the effects of taxes dedicated for
Historically, and still the case in tax evasion in response to higher tobacco control on tobacco use and
many countries, the primary aim of taxes. Government interference its consequences.
tobacco taxation was to generate in the decision whether or not to
government revenue. Tobacco consume tobacco is portrayed as an Outline of the Handbook
products are generally good infringement on individuals’ freedom
candidates for taxation, given that to choose. The production of this Handbook
they are typically produced by a However, many obstacles cited involved several steps, beginning
small number of manufacturers, as barriers to implementation of with the selection of the topic. Given
have relatively few substitutes, and higher tobacco-product taxes the centrality of tax and price policies
exhibit relatively inelastic demand. are misleading. For example, the to the WHO FCTC, the large and
3
growing research evidence on the people (Chapter 6);; tax, price and analysis. They also may use the term
impact of taxes and prices on tobacco tobacco use among the poor (Chapter “confounder” to signify a generic
use and related outcomes, and 7);; tax avoidance and tax evasion control variable in the model. An
IARC’s participation in the research (Chapter 8);; and the economic and “ecological study,” the term used by
project entitled Pricing Policies and health impact of tobacco taxation epidemiologists, can be translated as
the Control of Tobacco in Europe (Chapter 9). Each chapter assessed “macro-level study” for an economist,
(PPACTE), a grant proposal funded the quality and limitations of the data while the term “ceteris paribus” (other
by the European Commission, the when conducting its critical review. things being equal) would need to be
topic of tax and price policies was A strength of this volume is explained to an epidemiologist that
selected. Specifically, the goal of the involvement of researchers is not familiar with this expression.
Handbook Volume 14 is to present from a large number of countries This Handbook clarifies terms used
and assess the scientific evidence representing a variety of scientific when describing the evidence, and
on the effectiveness of tax and price disciplines, including economists, where possible, uses terminology
policies in tobacco control. epidemiologists, public health that is common across disciplines.
In collaboration with WHO and and public policy experts. This Therefore, it contributes to diminishing
IARC, a preliminary outline for the collaborative effort provided an communication barriers among the
Handbook was developed by the opportunity to bridge the gap in research community.
Chair of the Working Group (WG) terminology employed by various It is hoped that the research
of Volume 14. Twenty-one scientists disciplines and to find a common evidence presented in this Handbook
and policy experts from 12 countries, vocabulary to interpret and discuss will support policymakers, public
including high-, middle- and low- the f indings in the published scientific health professionals, and tobacco
income countries from multiple literature. Econometric terms can be control advocates in their effort to
regions, and including several novel for an epidemiologist. The term champion the use of tobacco taxes
researchers from the PPACTE “endogeneity,” a situation when an as a means to control tobacco use
project, agreed to participate as explanatory variable is correlated as well as to generate resources
members of the WG. In the fall of with the error term, for example, is for tobacco control programs. The
2009, experts revised and expanded less familiar to an epidemiologist WHO FCTC is entering into force in
the preliminary outline for the volume who may describe the same a progressively increasing number
and identified relevant literature;; condition as “reversed causality” or of countries (173 Parties as of June
in the spring of 2010 draft chapters as a situation where the exposure 2011). The conclusions presented
presenting and critically reviewing will “cause” the intermediate variable. in this volume will provide research
this literature were prepared. For example, if income is the background for discussions at
From 17 to 22 May 2010, IARC exposure/or independent variable the Conference of Parties to the
gathered 18 members of the WG and health is the outcome/the WHO-FCTC when it considers the
in Lyon, France to finalize this dependent variable, smoking would development of guidelines for taxing
Handbook on the evidence for be an intermediate variable between tobacco products. This Handbook will
the effectiveness of tax and price income and health, because people offer an evidence-based context for
policies in tobacco control. Pertinent with higher income are less likely to the research findings generated by
work published up to the week of the smoke. An economist may express the European Commission-funded
Handbook Meeting was established the same concept by saying that project PPACTE. PPACTE addresses
as eligible for inclusion a priori. At smoking is endogenous to income. A European Union policy needs related
the meeting, the WG revised the “confounder,” a term commonly used to the WHO FCTC process and has
chapters which are organized into in epidemiology referring to a factor potential to contribute to tobacco
the following domains: overview associated with an exposure and control not only in Europe but also
of tobacco taxation (Chapter 2);; the outcome, but not on the causal abroad. By embracing a global
tobacco industry pricing strategies pathway between the exposure view, this Handbook builds a bridge
and tax-related lobbying (Chapter 3);; and the outcome, is not used in between tobacco control policies in
tax, price and aggregate demand for econometrics. Economists sometime Europe and in the rest of the world.
tobacco (Chapter 4);; tax, price and use the term “confounders” to refer
adult tobacco use (Chapter 5);; tax, to problematic unmeasured “omitted
price and tobacco use among young variables” that are not included in the
4
Chapter 2. Overview of tobacco tax regimes and stresses the Public policies can influence
taxation importance of applying similar tax tobacco retail prices by multiple
levels for reducing the incentive for means, such as setting tobacco taxes,
Tobacco products are subject to product substitution. It demonstrates regulating prices and limiting price-
different types of taxes and to that sizeable increases in taxes on related marketing. These include,
different tax levels. The types of tobacco products under a properly for example, the implementation of
taxes, the tax structure in place, and designed tax system will result in minimum pricing policies, the use of
the enforcement of tobacco tax policy concomitant increases in retail prices. specific as opposed to ad valorem
reflect political, social and economic In view of the important revenue- tax structure, and bans on price-
considerations and can cause generating potential of tobacco reducing marketing techniques.
significant retail price differentials. taxes, particularly for low-resource Chapter 3 examines industry
In many cases, tax level is countries, Chapter 2 discusses the pricing strategies, price-related
related to income, with low-income strategy of using a share of tobacco marketing efforts, price manipulations
countries having low taxes and vice tax revenue to strengthen weak and market segmentation. It
versa. High-income countries tend health systems in these countries. describes industry efforts to reduce
to favour specific (per-unit) excise the impact of taxation on tobacco
tax regimes, while low- and middle- Chapter 3. Industry pricing use by trying to change tobacco
income countries rely more on ad strategies and other pricing tax policy, including tax structure
valorem (value-based) excise taxes. policies and tax levels, by opposing tobacco
Specific excise taxes generally result tax earmarking, and by engaging
in higher tobacco product prices. Recognizing the powerful impact of in anti-tax lobbying activities.
Some countries have implemented taxes and prices on the demand for These activities have some similar
more complex taxation regimes in tobacco products, tobacco companies characteristics across countries and
an attempt to find a balance between have demonstrated zeal in influencing the targeted level of government,
budgetary, health, and free market tax policies to minimize their impact even though most evidence comes
competition objectives. For example, on tobacco use and on their profits. from North America.
several countries are using a part Tobacco companies have The tobacco industry’s tax
or all of tobacco tax revenues for engaged in aggressive and well- lobbying efforts could be curtailed
funding health or tobacco control funded lobbying activities all over the by using the provisions of Article
activities. world. Their lobbying practices and 5.3 of the WHO FCTC, which aims
The tax level and the tax regime strategies may vary with the level of to limit tobacco industry influence
have implications for consumer government they want to influence. on tobacco control policy-making.
behaviour, the behaviour of the The impact of tobacco taxes However, very little is known about
tobacco industry, and the effectiveness on prices can be modified by the the effectiveness of policies that ban
of tobacco tax as a public policy industry’s response to the tax industry price-related marketing or
measure. Specific excise taxes can increase. Since most tobacco product introduce minimum pricing.
increase tobacco companies’ pricing markets are highly concentrated,
power, raise their profits and increase recent significant tax increases have Chapter 4. Tax, price and
market concentration. Differential led to price increases larger than aggregated demand for tobacco
rates on different types of tobacco justified by the tax increase alone,
products or even on items within the generating higher profit margins for A large body of the empirical
same product category result in price the tobacco industry. research employs macro- level data
gaps and opportunities for product Tobacco companies also use a to study the impact of tax and price
substitution to lower-taxed products variety of marketing techniques that on demand for tobacco products.
and brands. reduce prices on selected tobacco There is significant variation in the
Chapter 2 describes the products. These tactics can soften theoretical models as well as the
mechanisms of using tobacco the impact of tobacco tax increases econometric methods applied to
taxes to influence retail prices of specifically targeting youth and the these data. Initial evidence based on
tobacco products. It discusses poor. Price-related marketing efforts data from the USA and the United
the effectiveness of tax policy in are also designed to acquire market Kingdom that has demonstrated a
reducing tobacco use under different share from the competition. negative relationship between the
5
price and the demand for tobacco Since the demand for tobacco and improved over time. Of particular
products has been augmented by a products responds inversely to interest are the magnitude of impact
growing number of studies from low- tobacco price and positively to measured by the price elasticity
and middle- income countries. disposable income, Chapter 4 of demand and the relative impact
Chapter 4 summarizes the introduces the concept of affordability, of price/tax on prevalence versus
evidence and contrasts the results which captures the effect of price intensity of cigarette use. Gender
based on the older and more recent and income on tobacco demand differences in price responsiveness
studies using aggregate-level data. simultaneously. It points out that in to demand for tobacco have also
The focus is on the comparison a growing economy, the price of been of interest to researchers using
of price and income elasticities of tobacco would have to increase at the micro-level data on tobacco use.
tobacco demand, both over time the same rate as income to maintain The chapter summarizes the studies
and across country income groups. a given affordability level and to focusing on tobacco products other
Chapter 4 also points to the strengths prevent consumption from rising. than cigarettes and assesses the
and limitations of using aggregated Chapter 4 also briefly discusses level of substitution (measured by
and individual-level data to assess studies on substitution between the cross-price elasticity of demand)
the demand for tobacco products, tobacco and other harmful among tobacco products.
and the impact of price and tax substances (e.g. alcohol) and Studies using US data are of
policies on this demand. between different types of tobacco particular importance due to the
The price elasticity of demand products in response to changes in extensive subnational and temporal
measures the response rate (in relative tax and price levels. variation in taxes and prices.
percentage terms) of an aggregate Research evidence from other high-
measure of consumption (e.g. Chapter 5. Tax, price income countries is less common
cigarette sales) to a one-percent and adult tobacco use due to limited price variation within a
increase in the price of tobacco. single country.
The important issue is whether the Household and individual-level data The evidence from low- and
demand for tobacco can be classified have also been used to study the middle-income countries is primarily
as price-elastic (the price elasticity is impact of tobacco taxes and prices based on household expenditure
greater than one in absolute value) or on the demand for tobacco products. data. However, the quality of data and
as price-inelastic (the price elasticity These data, collected by various the appropriateness of methods used
is less than one in absolute value). survey methods, allow us to examine in these studies vary.
The magnitude of the price elasticity the differential impact of tobacco Research on the attitudes and
has implications for the impact of control measures, including tobacco perceptions of adults towards
prices/taxes increases on public taxes/prices, on tobacco use among increasing cigarette prices/taxes is
health and tax revenue. population subgroups with similar also discussed in Chapter 5.
Income is another important characteristics based on age,
determinant of the demand for gender, socioeconomic status, etc. In Chapter 6. Tax, price and tobacco
tobacco products. Since tobacco is addition, these micro-level data can use among young people
considered a normal good, economic distinguish between the impact of
theory predicts that as aggregate price/tax on prevalence and intensity Economic theory predicts that
income increases, aggregate of use, and allow researchers to tobacco use among young people
demand for tobacco increases as study behavioural changes such would be more responsive to
well. Therefore, income elasticity (the as tobacco use initiation, uptake, changes in prices/taxes compared
percentage change in consumption cessation, and switching from to adults, due to youth’s lower
of a product that results from a everyday to some-day tobacco use. disposable income and their lower
one-percent increase in income) Chapter 5 focuses on studies that level of addiction thanks to shorter
is expected to be positive. It is less examine the effect of price and tax smoking history. The effect of peer
clear whether there is a difference on adult tobacco use. It describes pressure is also expected to increase
in the income elasticity observed empirical methods employed to the responsiveness of youth to
in countries with different levels of assess the impact of price/tax on tobacco prices/taxes.
income or whether income elasticity individual level demand for tobacco Chapter 6 summarizes the
changes over time. and how these methods have evolved literature on the demand for tobacco
6
products among young people and regressive. Third, the health impact review indicated that there is much
presents the results by country of consuming tobacco increases confusion and improper use of these
income groups. It contrasts total medical expenditures and reduces terms. Tax avoidance refers to legal
price elasticity of demand, price labour productivity. This puts methods of circumventing tobacco
elasticity of smoking prevalence and pressure on the budgets of low- taxes, and tax evasion refers to illegal
price elasticity of smoking intensity income families and reduces their methods of circumventing tobacco
with results from studies on older income-generating potential due to taxes. Illicit trade includes both
population subgroups. The effects morbidity and premature mortality. legally produced products illegally
of price on youth smoking uptake, Chapter 7 presents the evidence traded across borders (smuggling)
smoking initiation, smoking cessation, on the demand for tobacco products and illegally manufactured products.
and on demand of other tobacco among the poor and compares it with Most tax avoidance activities include
products are discussed as well. Of the demand among more affluent the payment of some tobacco taxes.
particular interest is the literature segments of the population. It points Tax evasion involves both small
on youth smoking initiation, since to differences in the price elasticity and large quantities of tobacco
different types of data seem to lead of tobacco demand among high- products, and usually no taxes are
to different conclusions. The quality and low-income populations within a paid. These activities may involve
of the data and the measurement country, compares and contrasts the criminal networks or other large-
error inherent in retrospective evidence from low-, middle- and high- scale operations.
information lie at the heart of this income countries, and discusses the Chapter 8 categorizes tax
discussion. The chapter points to factors that influence the extent of evasion and tax avoidance activities
the gap in research evidence when differences in price elasticity across into various types and discusses
it come to stages of youth smoking different socioeconomic groups. their determinants based on the
uptake, peer and family influences Lower disposable income among most recent empirical evidence. It
on tobacco use among youth people, the poor would suggest that they focuses in particular on the role of
and the impact of price/tax on youth are more sensitive to changes in tobacco tax/price differentials, tax
smoking cessation. prices and taxes compared to more administration, enforcement, and
affluent populations. In that case, degree of punishment in motivating
Chapter 7. Tax, price and tobacco tax increases would be progressive these two types of activities. In
use among the poor and help the poor to reduce their addition, Chapter 8 examines the
tobacco tax expenditures. However, role of tax evasion and tax avoidance
There are at least three links the evidence as far as the relative in health disparities and the role of
between tobacco use and poverty. magnitude of the price elasticity is these activities in undermining other
First, spending on tobacco has an mixed and varies from country to tobacco control measures.
opportunity cost when resources country. For example, in settings Next, the chapter presents
from other goods and services are where there is ready access to low- estimates of the extent of tax evasion
diverted to obtaining tobacco. Lower- taxed or untaxed and inexpensive globally, regionally, and in key
income households are particularly tobacco products, low-income countries, as well as the impact of this
vulnerable to this diversion, as tobacco users may be less sensitive activity on tax revenues and public
tobacco may replace food and other to changes in prices due to the low health. The authors also contrast the
essential products and services cost of substitution. Future data scale of tax evasion with the scale
(e.g. health care, education) for the collection efforts may need to provide of tax avoidance. This is particularly
entire family. Second, tobacco use more information on tobacco tax challenging due to the illegal nature
prevalence is higher among the regressivity to address this important of these activities. Various research
low-income, low-education groups public policy issue. methodologies trying to overcome
in the majority of countries (with the these challenges are discussed,
exception of some upper middle- Chapter 8. Tax avoidance and tax as well as their strengths and
income countries). This means evasion weaknesses.
that tobacco excise taxes could Despite the underground nature
be disproportionately collected Chapter 8 starts by defining what is of these activities, effective strategies
from people with lower income, considered tax evasion and what is and policies to control tax avoidance
and therefore would be labelled as merely tax avoidance. The literature and tax evasion exist. These address
7
both the supply side and the demand The pros and cons of earmarking levels and incomes. A portion of
side of the market. A review of tobacco tax revenue are also tobacco tax revenues should be
country experiences with these discussed based on the experiences used to fund comprehensive tobacco
policies and case studies of countries of jurisdictions that have earmarked control programs and other health
that have successfully implemented revenues for tobacco control and/or promotion activities, given that such
an integrated set of actions to curb health promotion programs. programs lead to further reductions
both tax avoidance and tax evasion in tobacco use and associated
conclude this chapter. Summary of findings improvements in population health.
of the Handbook The WG also recommends the
Chapter 9. Economics and health implementation of a multinational
impact of tobacco taxation For each chapter the Working surveillance and monitoring system
Group examined all the evidence, so that data on tobacco use,
Higher tobacco taxes/prices can assessing the quality of the data tobacco taxes and prices, price-
be expected to have a large impact and appropriateness of the methods reducing marketing and lobbying
on society’s well-being. The most employed to generate the results, efforts of tobacco companies, tax
important is their impact on heath and then voted on the strength of the avoidance and evasion, and tax
status, life expectancy, labour evidence presented in the Evaluation administration and enforcement
productivity and overall economic chapter. activities can be collected regularly.
performance. Additional benefits The scale for the quality of Such targeted tobacco control data
include lower smoking-related evidence ranged from “sufficient” to collection efforts would improve the
healthcare expenditures. Further “strong” to “limited” to “inadequate quality of the research evidence
health and productivity gains can or no evidence” and “evidence gathered to date, particularly for
be expected through reductions in suggesting a lack of an effect.” For 12 low- and middle-income countries.
secondhand smoke exposure and of 18 conclusions in this volume, the The full recommendations of the
reduced maternal smoking. strength of the evidence was at the Working Group are detailed in the
The extent to which tobacco highest level, while four concluding Recommendations chapter.
product taxes and prices contribute statements were supported by
to these health economic gains on “strong evidence” and two remaining
the population level is summarized statements were supported only
in Chapter 9. The chapter starts with by “limited evidence”. There is
a brief discussion of a conceptual sufficient evidence that higher
framework for describing the impact tobacco excise taxes and prices
tobacco taxes have on lives saved reduce overall tobacco consumption
and disease incidence (in smokers and prevalence of tobacco use. This
and secondhand smokers) and on is achieved by the impact price/tax
healthcare cost savings. Chapter 9 has on promoting cessation among
References
examines not only the direct impact of current users, preventing initiation
lower tobacco use, but also considers and uptake among young people,
the association between tobacco use and lowering consumption among IARC (2008). IARC Handbooks of Cancer
and alcohol, tobacco use and illicit continuing tobacco users. Prevention: Tobacco Control, Vol. 12: Methods
for Evaluating Tobacco Control Policies. IARC,
drug abuse, and between tobacco The consensus among the Lyon.
use and obesity. members of the Working Group on
IARC (2009). IARC Handbooks of Cancer
The chapter further presents the strength of the evidence resulted Prevention: Tobacco Control, Vol. 13: Evaluating
evidence on the effects of higher in several public health and research the effectiveness of smoke-free policies. IARC,
Lyon.
tobacco taxes on tobacco industry recommendations. The most
employment (including tobacco effective tobacco tax public health Jha P, Chaloupka FJ (1999). Curbing the
farming) as well as on non-tobacco policy would promote a relatively epidemic. Governments and the Economics
of Tobacco Control. Washington D.C., World
industry employment, on government simple tobacco excise tax structure Bank.
revenue, on tobacco tax revenue, and that emphasizes specific tax and
World Health Organization (2005). WHO
on the Consumer Price Index. involves regular tax increases that Framework Convention on Tobacco Control.
outpace growth in general price Geneva, World Health Organization.
8
Chapter 2
Overview to tobacco taxation
Introduction
Taxes are usually raised to produces negative externalities categories: consumption taxes and
provide revenues for government (McCarten and Stotsky 1995;; World customs duties.
expenditure. These taxes take many Health Organization, 2010).
forms—for example, income taxes, This chapter provides an Consumption taxes
payroll taxes, customs duties, excise introduction to the different types of
taxes, sales taxes or value-added taxes applied on tobacco products, Consumption taxes are taxes on
taxes (VAT). Indirect taxes are with a special focus on excise taxes. spending on goods and services.
taxes levied on the consumption of The reasons behind levying such The term refers to a tax system with
specific goods (for example, excise taxes are then discussed, including a taxable base of consumption. The
taxes on tobacco or alcohol) or on the political, social and economic main consumption taxes are value
practically all goods (VAT). The arguments often used against added tax (VAT) or retail sales taxes
interest in raising taxes on products tobacco tax increases when the and excise duties. These are indirect
such as tobacco is based on their issue is discussed by policymakers. taxes, meaning that they are not
potential to raise large amounts of The different approaches to excise levied directly on the income of the
revenues relatively easily, but also taxation are then reviewed, with a consumer or earner. These taxes are
because they enable correcting for description of the structures applied due to the revenue authorities by the
the negative externalities tobacco on tobacco products and the pros and supplier of the goods or services;;
use generates (negative health cons of each one of them. The impact however they are ultimately borne by
impact of both tobacco consumption of taxes on the price of tobacco the final consumer. They are called
and exposure to tobacco smoke) products is also considered briefly. regressive because they are not
and discourage its use because of An overview is made of the levels based on the ability-to-pay principle.
its destructive impact. The rationale of taxation globally (with a focus on Consumption taxes apply to all
behind raising taxes on products cigarettes because of better availability supplies or releases for consumption
such as tobacco lies in the particular of data). The issue of earmarking on the territory of a jurisdiction and,
aspects of the product: i) production or dedicating tobacco taxes for normally, also to imports of tobacco
is dominated by a few companies, specific programmes, particularly products. Tobacco products that are
which makes supervision and health-related programmes, is also exported are normally not subject to
tax collection by the government discussed. Finally, the last section consumption taxes.
relatively easy;; ii) the demand for highlights the main issues covered in VAT, retail sales taxes and excise
this product is relatively inelastic – the chapter. duties have different characteristics:
tobacco users are addicted to the Value added tax is a general
products and therefore have little Description of taxes consumption tax that applies,
sensitivity towards a price change;; in principle, to all commercial
iii) the product is not considered a Taxes on tobacco products can activities involving the production
basic necessity;; and iv) the product be classified into two general and distribution of goods and the
9
provision of services. It is charged as (World Health Organization, 2010). tobacco products. Often they aim to
a percentage of price, which means In contrast to VAT, excise duties finance various programmes through
that the actual tax burden is visible are usually levied at the stage of earmarking but nonetheless act
at each stage in the production and production or importation—and not as excises (e.g. the health tax on
distribution chain. It is collected distribution—and they target the tobacco products in Romania).
fractionally, via a system of partial consumption or the use of specific
payments whereby taxable persons products. The most commonly applied Customs duties
(i.e. VAT-registered businesses) excise duties are those on alcoholic
deduct from the VAT they have beverages, manufactured tobacco Customs duties (also called tariffs)
collected the amount of tax they products and energy products (motor are taxes levied on imports of goods
have paid to other taxable persons fuels and heating fuels, such as petrol (and, sometimes, on exports) by the
on purchases for their business and gasoline, electricity, natural gas, customs authorities of a country,
activities. This mechanism ensures coal and coke). mainly to raise state revenue, and/
that the amount of tax will be the Excisable goods have the or to protect domestic industries
same, independent of the number of following common characteristics: from more efficient or predatory
intermediate transactions before the demand is price inelastic;; production, competitors from abroad. Again, the
final supply to the consumer. VAT distribution and sales can be closely duty may be specific or ad valorem.
is due to the revenue authorities by supervised by the government;; Specific customs duties are based
the seller of the goods, who is the and they are associated with upon the weight, dimensions, or
“taxable person,” but it is actually negative externalities (e.g. health or some other criteria of the item
paid by the buyer to the seller as part environmental) or are considered (such as the size of the engine in
of the price. luxury goods. the case of automobiles). The ad
VAT is a multistage sales tax There are two types of excise valorem customs duties are levied
that applies at several stages of the duties on tobacco products: specific on importer’s CIF (cost, insurance
production/distribution chain for a and ad valorem. A specific excise and freight) value, as opposed to
product or service. However, a few duty is a fixed monetary amount of ad valorem excise duties which are
countries have single-stage sales tax per quantity, volume, or weight levied on the manufacturer’s price or
taxes that apply only at one stage. of tobacco products (e.g. per piece, the retail selling price. Consequently
The most common single-stage tax is pack, carton, kilogram). An ad the impact of a customs duty on the
the retail sales tax which is charged valorem excise duty, on the other final consumers’ price will be less
only on the sale of an item to its final hand, is levied as a percentage of than that of an excise duty, because
end user (e.g. the United States). some measure of value of the tobacco the CIF value at importation can be
Multistage taxes ease the products (e.g. the manufacturer’s considerably lower than the e.g. the
enforcement of higher tax levels price or the retail selling price). final retail selling price. For example,
(rates), as the taxes are collected Excises on tobacco are levied total tax as percentage of retail price
fractionally. Norway, Denmark, in most countries around the world. is 50% in Saudi Arabia, Bahrain, and
Sweden and Hungary have the Only a few countries do not levy an Qatar despite 100% import duties
highest VAT rate at 25% (http:// excise on tobacco products (e.g. (World Health Organization, 2010).
ec.europa.eu/taxation _customs/ Benin, Cook Islands, Maldives, Saudi Almost all countries levy – usually
taxation/vat/how_vat_works/rates/ Arabia, Grenada) (World Health an ad valorem – tariff on imported
index_en.htm). More and more, Organization, 2009). However, the tobacco products. Again, the practice
conventional sales taxes are being type (specific versus ad valorem), varies greatly among countries, with
replaced by more broadly-based rates and base of the tax vary rates for example of 100% in Guyana
value added taxes. considerably across countries. and 83% in Egypt (World Health
Most countries around the Other, supplementary taxes Organization, 2010).
world levy a VAT or another broad- on tobacco products are named Customs duties aim to raise
based consumption tax on tobacco differently in different countries;; state revenue, and/or to protect
products. Only a few countries do not however they may act as excise domestic industries and not to
apply such a tax on tobacco products duties despite their names (e.g. the influence the consumers’ price/
(e.g. Yemen, Egypt, Maldives, stamp duty in Brazil). Some countries behaviour. Moreover, relying on
Fiji, the Comoros, and Grenada) levy several additional taxes on higher import duties as a way of
10
Overview of tobacco taxation
generating revenues or increasing Cigarettes are the most Cigars are handmade or machine
the price of tobacco products may consumed tobacco product. made but at lower speed and higher
not be an appropriate policy given Worldwide cigarette consumption cost compared to cigarettes. This
the trade liberalization and the accounts for approximately 80% is reflected in the taxation regime,
bilateral, multilateral or global trade or more of the total production of whereby cigars or cigarillos are
agreements which provide for a tobacco leaves (http://www.fao.org/ often taxed at a considerably lower
phasing out of such duties. docrep/006/Y4956e/y4956e04.htm). level than cigarettes. However,
In the European Union, cigarettes new products have appeared over
Description of taxed products account for approximately 92% of the last years (e.g. “eco cigarillos”
the total sales of tobacco products in the EU, “small cigars” in the US)
In principle, excise duties are levies (European Commission, 2010a). which are manufactured at low cost
on manufactured tobacco and not on However, in some areas such as and marketed as alternatives for
raw tobacco leaves. Manufactured southeastern Asia, substitutes cigarettes but taxed at a sometimes
tobacco includes products which are like bidis and kreteks have a more considerably, lower rate.
entirely or partly made of tobacco important market share.
for the purpose of smoking, sniffing, Bidis are the Indian-southeastern Loose smoking tobacco
sucking or chewing. It includes rolls Asian version of cigarettes. They are
of tobacco such as cigarettes;; bidis, made by rolling a dried, rectangular Fine-cut tobacco is loose tobacco
kreteks, cigars and cigarillos;; loose piece of tendu or temburni leaf which consumers primarily use to
smoking tobacco such as fine-cut (plants native to Asia) with sun- make cigarettes, either by rolling it by
tobacco, pipe and water pipe tobacco, dried, flaked tobacco (approximately hand into cigarette paper (roll-your-
as well as smokeless tobacco such 0.2–0.3 g) into a conical shape and own, RYO) or using fabricated filter
as snus (for sucking), nasal snuff (for tied with a piece of thread. The bidi tubes and a making device (make-
sniffing) and chewing tobacco. industry has a large number of small- your-own, MYO).
In most countries, the various scale industries, with a significant Although worldwide fine-cut
categories of manufactured tobacco share of bidis being handmade. tobacco (together with pipe tobacco)
carry different levels of taxation, Bidis account for around 85% of total is estimated to be only around
reflecting differences in the fiscal smoking tobacco consumption in 1% to 2% of the tobacco market,
policy objectives as well as in the India, with the remainder consisting in some regions it has a more
perceived tax-bearing capacity of of cigarette consumption (John et al., important market share (Euromonitor
the different product categories. 2010). Historically, excises on bidis International, 2009). Fine-cut
In particular, hand-made or have been close to zero. A lobbying tobacco comprises approximately 8%
more labour intensive products, argument of the bidis industry is that of the total sales of tobacco products
products made mainly by small- a tax increase will affect employment in the European Union (European
and medium-sized enterprises, as and tobacco-related trade (Ray and Commission, 2010a). The core
well as products predominantly Gupta, 2009). markets in the EU are Germany, the
consumed by consumers in the lower Kreteks, sometimes referred to as Netherlands, Belgium, Luxembourg,
income groups, often benefit from clove cigarettes, are the Indonesian France and the United Kingdom,
preferential tax treatment (World version of cigarettes and by far the covering in volume 80% of the EU
Health Organization, 2010). most widely-smoked form of tobacco fine-cut market. In Luxembourg
in Indonesia. They typically contain and the Netherlands, fine-cut even
Rolls of smoking tobacco a mixture of approximately 60–80% accounts for more than 50% of total
tobacco, 20–40% ground cloves, consumption of tobacco (European
Cigarettes are basically rolls of clove oil and other additives. Commission, 2010a).
tobacco wrapped in paper tubes Cigars and cigarillos are rolls The fine-cut tobacco
capable of being smoked as they are. of tobacco with an outer wrapper manufacturing process is relatively
Manufacturing cigarettes is a capital- of natural tobacco or rolls with a labour-intensive as compared to
intensive, fast-paced and highly threshed, non-cut, blend filler and cigarettes. There are many small-
automated process. Machines may with an outer wrapper of the normal and medium-sized, often family-
produce between 8000 and 20 000 colour of a cigar, of reconstituted owned, enterprises producing fine-
cigarettes every minute. tobacco, covering the product in full. cut tobacco. In addition, fine-cut
11
tobacco is predominantly consumed Smokeless tobacco products a variety of dissolvable tobacco

by consumers in the lower income products and snus, in addition to
groups. Historically, it has been Taxation of smokeless tobacco the more traditional moist snuff and
taxed at a significantly lower level products has received comparatively chewing tobacco products. The
than cigarettes. little attention in most countries. issue of how to tax all these products
Pipe tobacco is loose tobacco Smokeless tobacco is a major remains an open question for further
processed in a different way to consumer’s choice in some markets study.
make it capable of being burned such as Sweden, Norway and India, Finally, to avoid loopholes,
in a pipe. In many countries, it is a and is widespread in countries such countries may also tax other
niche market product with low and as the USA (IARC, 2007). Basically, manufactured tobacco, such as
steadily declining volumes. Because there are three major forms of oral tobacco refuse put up for retail sale
of its generally more traditional and smokeless tobacco products: or all other tobacco which has been
more labour intensive manufacturing Tobacco alone with aroma and cut, split, twisted or pressed and is
processes, inter alia, it often has an flavouring includes products that capable of being smoked without
even lower tax level than fine-cut are sucked, chewed or both. For further industrial processing.
tobacco. Pipe tobacco is taxed as example, snuff, which is chopped into
«other smoking tobacco» (European particles like large coffee grounds Objectives of tobacco taxation
Commission, 2010a). As a result, and moistened, is used by holding
also in this market, new products between gum and cheek. Swedish In many countries tobacco is taxed
have appeared which are taxed as snus, which is a variant of snuff more heavily than other goods. There
pipe tobacco, but are marketed as, processed differently and typically are at least three reasons for this.
and in direct competition to, fine-cut moister, is sucked. Snus exists in
tobacco. two packaging formats, loose snus Revenue objectives
Water pipe tobacco is another and portion-packed snus. Chewed
form of smoking tobacco widely used products, shredded like short cut Historically, revenue generation
in southwestern Asia and the eastern grass, generally mildly acidic, are has been the primary aim of
Mediterranean area. However, its intended to be chewed throughout tobacco taxation of most, if not all
consumption is increasing in other the day as desired, for example governments. Taxes on tobacco
regions, such as the EU (Unpublished loose-leaf. products are a very efficient revenue
data from Internal Reports from Tobacco with other components raiser given the large sales volumes,
Member States to the EC;; Knishkowy includes products that contain lime, the relatively inelastic demand
and Amitai, 2005). Recent data sodium bicarbonate, ash or other (consumers are not price-sensitive
published by the Eurobarometer additives and which can be either due to addiction) and the lack of
indicate that 9% of smokers use chewed or sucked, for example chimó close substitutes. They satisfy
water pipes occasionally (European and shammah. the so-called “Ramsey Rule” for
Commission, 2010b). Little Betel quid with tobacco includes economically efficient consumption
information is available with regards areca nut, slaked lime, catechu, and taxes—because of the relative
to excises on tobacco products for tobacco, and comprises products inelasticity of demand, they can
water pipes. The tax rates seem to that can be chewed or/and sucked, generate considerable revenues
vary widely, from 2% of the producer such as gutka. while creating fewer distortions in the
price in Libyan Arab Jamahirya, to The tax treatment of smokeless market than would result from taxes
15% in Syrian Arab Republic, to 58% tobacco differs widely among on goods and services with more
of retail price in Turkey and 108% in countries;; often it is not taxed, while elastic demand. Moreover, given the
Lebanon ( World Health Organization, some countries apply differential rates small number of producers and the
2010). In the EU, water pipe tobacco for, e.g., snuff and chewing tobacco large sales volumes, tobacco taxes
is taxed like pipe tobacco (European (World Health Organization, 2010). are relatively easy to collect, at low
legislation Directive 2010/12;; Smokeless tobacco is becoming administration and enforcement cost,
see «other smoking tobacco» in a more important policy issue in particular as compared to general
European Commission, 2010a). because of the appearance of new consumption taxes and income
smokeless tobacco products. These taxes. Table 2.1 shows the share of
new smokeless products include excises on tobacco as a percentage
12
of total tax revenues in countries in Table 2.1. The share of excises on tobacco as a percentage of total tax
the EU. revenues in the EU Member States and two other selected countries in 2005
Given the size of total revenues
Country Share of tobacco excise
in some countries, even a share of
1–2% represents a significant source Sweden 0.7%
of revenue in absolute monetary Slovenia 0.8%
amounts. However, to date, and Denmark 0.9%
in particular in more developed Finland 1.2%
economies, VAT and other general
Netherlands 1.5%
sales taxes have the capacity to raise
Belgium 1.8%
much larger revenues from a more
Lithuania 1.8%
widely spread tax base. Therefore
the retention and the increase of Ukraine 1.8%
excise duties on tobacco products Austria 2.0%

are also justified by reasons other France 2.1%
than budgetary. United Kingdom 2.2%
Italy 2.3%
Health objectives Latvia 2.3%
Ireland 2.6%
To discourage consumption
Germany 2.8%
of the product
Spain 2.9%
Tobacco use is the leading cause of Hungary 3.2%
preventable death, and is estimated Estonia 3.4%

to kill more than 5 million people Cyprus 3.6%
each year worldwide. If current Portugal 3.7%
trends persist, tobacco will kill more Slovakia 4.0%
than 8 million people worldwide by
Czech Republic 4.0%
the year 2030, with 80% of these
Poland 4.8%
premature deaths in low- and middle-
Malta 4.9%
income countries (World Health
Organization, 2008). It is the biggest Greece 5.6%
single form of avoidable death and Romania 5.8%

one of the leading causes of illness Bulgaria 6.8%
and mortality. Luxembourg 7.3%
Taxation forms part of an overall Indonesia (2007) 8.4%
strategy of tobacco use prevention
and dissuasion that also includes Source: European Commission (2008a) (SEC/2008/2266);; see tobacco products legislation;; impact assessment
other measures intended to reduce
demand, such as protection from of estimates of the magnitude of the from taking up smoking (especially
exposure to tobacco smoke, effects of price on overall tobacco among young people), encourage
advertising bans, regulation of the consumption, it is clear that a price smoking cessation, reduce the
contents, etc. Price increases of increase will lead to a reduction number of ex-smokers who resume
tobacco are considered to be the in consumption (see Chapter 4 for the habit, and reduce in the long run
most effective and cost-effective the value of these estimates). More the average cigarette consumption
single measure to prevent and importantly, the impact of higher among continuing smokers.
reduce tobacco use. Over one prices is likely to be greatest on young Article 6 of the World Health
hundred studies have examined the people, who are more responsive Organization’s Framework Convention
impact of tobacco taxes and prices to price rises than older people. on Tobacco Control (WHO FCTC)
on overall tobacco use. While these In addition, price increases are an recommends tax policies so as to
studies have produced a wide range effective policy tool to prevent people contribute to health objectives aimed
13
at reducing tobacco consumption to environmental tobacco smoke Political, social and economic
(World Health Organization, (passive smoking). The second considerations
2005). More and more countries or comprises collectively-borne costs,
jurisdictions are using tobacco taxes such as the cost of publicly-funded In principle, the three aforementioned
as a way to promote public health by medical treatment for smoking- objectives are complementary.
reducing tobacco use and the death related conditions, and other Increases in tobacco taxes aiming to
and disease it causes (e.g. the EU, public expenditure costs. The third raise additional revenue will contribute
Norway, New York and California in category of externalities is, in effect, to a reduction in consumption and
the United States, Pakistan, etc.). A a tax revenue externality, namely in the external cost of smoking. Vice
health-driven taxation policy aims the loss of income and consumption versa, given the inelastic demand and
to increase the overall tax and price taxes as a result of reduction in the the high share of the tax in the retail
levels of tobacco products, increase consumer’s income and expenditure, price, tax policies aimed at reducing
in particular the tax and price levels especially through premature tobacco consumption or at recouping
of the cheaper brands, and reduce death and a higher rate of sickness externalities will, all other things being
the price gap between low-priced absence. The consequences or the unchanged, entail revenue increase.
and premium brands to discourage costs for the individual consumer’s Nonetheless, when determining
down-trading (smokers switching to own health, income, and so on, are their taxation policy governments
cheaper brands as a result of tax and no externalities. will take into account other, at
price increases). Most estimates distinguish times competing, considerations.
between the gross costs of smoking This section does not aim to list
To recoup what economists call (higher costs of medical treatment, exhaustively all political, social or
“negative externalities.” etc., as a result of conditions caused economic considerations that may be
by smoking), and the net costs, taken into account in determining a
Negative externalities are the costs which offset against the gross costs taxation policy, but focuses briefly on
borne by society collectively, or by a range of cost savings (mainly the most frequently used arguments
individuals other than the individual public expenditure effects, such as against tax increases.
tobacco consumer. Theories savings on retirement pensions)
suggest that these external costs arising because of the premature Poverty
associated with the consumption of death of smokers. Some literature
tobacco, such as the costs to treat suggests that in high taxing countries Concerns about the affordability of
smoking-related diseases, warrant smokers pay their way—in other cigarettes, in particular for the poor,
supplementary taxes on tobacco. words, that the supplementary taxes are arguments for those who oppose
Generally, these costs are not on tobacco overcompensate the tobacco tax increases. Consumption
reflected in the price of the tobacco net external cost (Cnossen, 2006;; taxes are regressive because
products. The purpose of externality Manning et al., 1989). However, they are not based on the ability-
taxation is to confront the individual this is a controversial area, inter to-pay principle. Assuming equal
decision-maker with the external alia, because of the treatment of consumption patterns, tobacco taxes
costs of their decision, on the same costs borne by family members. will account for a greater share of
basis as if these costs were private Family members of a smoker may income for the poor than for the rich
costs (so-called “internalisation”). experience considerable costs, (see Chapter 7). This regressivity will
Because the consumer pays for including ill health, and pain and be more pronounced in countries
the societal cost, he is assumed to distress as a result of the illness and where the tobacco consumption is
make a more economically efficient premature death of the smoker. This greater among lower than among
decision on whether and how much harm inflicted on family members— higher incomes. A “pro-poor policy”
tobacco to consume. which is hardly quantifiable—is often can keep taxes on tobacco low in
Negative externalities fall into not considered as an external cost general, but can also keep taxes low
three broad categories. The first (Smith, 2007), as family members on the products/brands most widely
consists of direct externalities are assumed to care for each other’s used by the poor while more heavily
experienced by other individuals, welfare to the extent that the welfare taxing more expensive products or
including the adverse health effects of the household can be considered brands (e.g. fine-cut tobacco versus
experienced by those exposed as a single entity. cigarettes).
14
However, a policy aimed at Employment rates to tobacco products depending

keeping tobacco affordable for lower on the source or type of tobacco
incomes is likely to end up with a In general, only jobs in tobacco contained in the product or on other
disproportionate share of the health farming, leaf processing, warehousing product characteristics, or indirectly
and economic burden of tobacco and manufacturing are fully by applying an ad valorem excise
consumption on the poor. On the dependent on tobacco. Other sectors, duty where foreign brands are more
other hand, although the initial tax such as retailers who sell tobacco expensive than local. This can be
is regressive, tax increases can be among many other products are only an infringement of Article III (2) of
progressive (Chaloupka et al., 2000). partly and indirectly involved. To date, the General Agreement on Tariffs
To the extent that lower incomes are tobacco manufacturing is a capital- and Trade legal text referred to as
more sensitive to price increases intensive sector and is relatively GATT, 1947 (http://www.wto.org/
than higher incomes, tax increases small in terms of numbers of people english/docs_e/legal_e/gatt47_01_e.
will entail a higher reduction in employed, e.g. 60 000 in the EU-25 htm#articleIII), now embedded in the
consumption among this population in 2003 (Commission Staff Working World Trade Organization, according
while having less of an impact Document Impact Assessment to which internal taxes shall not
on higher-income populations. Accompanying the Proposal for a be applied to protect domestic
Consequently, the burden of the Council Directive amending Council production, or of similar provisions
actual supplementary taxes paid as a Directive 95/59/EC, 92/79/EEC and laid down in regional free trade
result of a tax increase will be greater 92/80/EEC on the structure and rates arrangements.
on those with higher incomes. In the of excise duty applied to manufactured
long run the poor benefit from an tobacco (SEC/2008/2266)).
increase in the quality of their health In general, any tobacco- Other constraints for tax and price
and economic welfare. dependent job lost in response to increases
the reduced demand for tobacco
Inflation products will be offset by new jobs As aforementioned, the health and
in other sectors because the money budgetary objectives are to a large
Consumption taxes may have an spent on tobacco will be shifted to extent complementary. Decision-
inflationary effect. At times the more labour-intensive goods and makers will also take into account other
inflationary impact of increases services (see Chapter 9). To address considerations, such as the impact on
of tobacco taxes is raised as an employment concerns in tobacco- inflation, employment, affordability
argument to oppose increase of dependent sectors, programs have and the interests of the domestic
tobacco taxes, in particular where been adopted to ease the transition to tobacco growers and manufacturers.
government policy is to keep inflation another economic activity, e.g. crop However, the ability to increase
low. However, as opposed to broad- diversification for tobacco farmers or prices and revenues by means of tax
based consumption taxes, the relative product diversification for retailers. At increases has certain constraints,
weight of expenditure on cigarettes times, the employment argument has as any tax increase may entail a
in the consumer price index should been used to justify reduced rates on change in other variables affecting
not be overestimated, and in general products other than cigarettes, which the expected revenue increase or
the impact on inflation of an increase are perceived as being more labour reduction in consumption. The most
of tobacco taxes will be relatively intensive (e.g. handmade kreteks;; pertinent ones are the manufacturers’
small (see Chapter 9). To the extent bidi rolling, fine cut tobacco). pricing policy, the consumers’
that concerns about the impact on behaviour and the share of the non-
inflation are a barrier to tax increases, Protection of domestic tobacco domestic duty paid consumption in
excluding tobacco products from the growers and manufacturers total domestic consumption.
baskets of goods used in developing
key price indices used for the Some countries levy a lower tax on The share of the non-domestic duty
indexation of wages and pension local tobacco products to protect paid consumption (NDDP):
payments, such as for instance in domestic tobacco growers and tax evasion and tax avoidance
France, Belgium and Luxembourg, tobacco manufacturers from outside
would greatly reduce these concerns competitors. This can be done by A part of the tobacco market will
(World Health Organization, 2010). directly applying different excise escape domestic taxation because
15
of illicit trade and cross-border Consequently, the tax increase objective that a country wants to
shopping. Cross-border shopping in will not result in the expected price achieve with the tobacco tax, and in
neighbouring low-taxing countries increase and related reduction of part on whose perspective is being
or jurisdictions is legal as long as consumption. Moreover, in the case used to evaluate their effects (revenue,
the quantitative restrictions laid of ad valorem duties, this may even health, manufacturers, consumers).
down in the traveller allowances are affect the expected revenue increase To have the best elements of both, it
respected. Illicit trade covers mainly (see further and Chapter 3). is possible to combine an ad valorem
smuggling, illicit manufacturing with a specific tax. A so-called mixed
and counterfeit. Smuggling refers Consumer behaviour structure applies an ad valorem and
to products illegally traded across specific duty to all tobacco products.
borders. Large-scale organized As cigarettes may be sold at different Mixed systems can give preference to
smuggling involves the illegal price points (low, medium-priced, more ad valorem or to more specific
transportation, distribution and sale premium), consumers may switch duties depending on the desired
of large consignments of cigarettes to cheaper cigarettes or to other effects.
and other tobacco products. Small- tobacco products (e.g. fine-cut Ad valorem excise duties can also
scale smuggling involves the instead of cigarettes) as a result be combined with a minimum tax floor.
purchase, by individuals or small of a tax increase. For example, a Minimum excise duties are similar to
groups, of tobacco products in low-tax Klynveld Peat Marwick Goerdeler specific excise duties, and are a fixed
jurisdictions in amounts that exceed (KPMG, 2005) study commissioned monetary amount per quantity or
the limits set by customs regulations, by the European Commission volume that applies if the ad valorem
for resale or just use in high-tax concluded that the market share of excise falls below a minimum floor. In
jurisdictions. Illicit manufacturing cheap cigarettes has soared in most other words the ad valorem cannot
refers to the production of tobacco EU Member States as a result of tax be less than the minimum tax floor;;
products contrary to taxation laws hikes triggering less tax revenue and/ lower-priced products will be taxed at
or other laws (such as licensing or or mitigating the downward the effect the specific minimum rate, and higher-
monopoly-related laws) that restrict on the consumption of cigarettes. priced products will be taxed at the ad
the manufacture of tobacco products. Again, an ad valorem structure will valorem rate. The effects of a minimum
Counterfeit tobacco production is a make price and tax policies more tax floor are similar to those of specific
form of illegal manufacturing in which vulnerable to changes in consumer duties. Finally, more complex taxation
the manufactured products bear a behaviour (see further). systems may even combine a mixed
trademark without the consent of the structure with a minimum duty.
owner of the trademark. Counterfeit Structure of the taxes In summary, this leads to five
and illegally manufactured products structures:
can be sold on the domestic market There are two types of excise taxes: 1) Purely specific;; with a tax base
or smuggled into another jurisdiction specific and ad valorem. A specific per unit, e.g. per 1000 cigarettes;; per
(see Chapter 8). excise tax is a fixed monetary amount 1 kg tobacco.
Significant differences in taxes of tax per quantity, volume, or weight 2) Purely ad valorem;; the tax
and prices of tobacco products of tobacco products. An ad valorem base is the value of the products (e.g.
between countries and jurisdictions excise tax is levied as a percentage ex factory price;; retail price).
have created an environment for of the price of the tobacco products. 3) Mixed;; a combination of both
tax-induced cross-border shopping Countries may have either an ad ad valorem and specific duty.
and illicit trade. Worldwide illicit trade valorem or a specific structure. 4) A combination of an ad valorem
is estimated at 11.6% of the global Specific and ad valorem taxes have duty for medium-priced and/or
cigarette market in 2007 (Joossens different effects on prices, profits premium brands and a specific duty
et al., 2009). and competitive positions of tobacco for cheaper brands. The ad valorem
producers, tax revenues, quality and excise applies on the value of the
Manufacturers’ pricing policy variety of products, administration products;; however, if the ad valorem
and distribution of income. They will excise falls below a minimum floor, a
Manufacturers may absorb the contribute in a different way to the specific tax applies.
tax increase, partly or completely, achievement of health objectives. 5) A combination of a mixed
by reducing their profit margin. The relative merits depend on the duty for medium priced and/or
16
premium brands and a specific duty Table 2.2. Excise structure for cigarettes in 2008
for cheaper brands. A mixed excise
applies;; however if the mixed excise Structure Number of countries (total 182)
falls below a minimum f loor, a specific Purely specific system 55

tax applies. Purely ad valorem system 60
Only 55 countries rely on
Mixture of both excises 48
specific duties only;; most (108)
apply at least to some extent an No excise 19
ad valorem duty (see Table 2.2) Source: World Health Organization (2009). WHO report on the Global Tobacco Epidemic 2009: implementing smoke-
(World Health Organization, 2009). free environments. Geneva, World Health Organization
No complete data are available as
concerns the number of countries system (inter alia, the United States The specific component may not be
applying a minimum tax. Twenty- of America, Canada, Australia, New less than 5% or more than 55% of
four EU Member States (European Zealand, Japan, Singapore). the amount of the total tax burden
Commission, 2010a) as well as (proportional and specific excise
the Russian Federation and the There are also similarities by duty plus VAT). As from 1 January
Ukraine combine a mixed structure region. Most countries in Latin 2011, the specific component of the
with a minimum tax. Turkey applies America and the Caribbean region excise duty on cigarettes may not
an ad valorem system combined as well as in Africa apply purely ad be more than 76.5% of the amount
with a minimum tax (World Health valorem systems. North America of the total tax burden and as
Organization, 2010). and a large number of countries in from 1 January 2014, the specific
The preference of countries for a the East Asian and Pacific region component of the excise duty on
particular structure shows similarities rely on purely specific systems. cigarettes may not be less than 7.5%
by income groups. Most low-income Mixed systems are mainly applied (Council Directive 2010/12/EU of 16
countries rely on ad valorem taxes. in Europe, and in China, Indonesia, February 2010 amending Directives
Conversely, almost all high-income Pakistan, Malaysia and Thailand 92/79/EEC, 92/80/EEC and 95/59/
countries apply purely specific or (World Health Organization, 2009). EC on the structure and rates of
mixed systems. Within the high- Countries applying a mixed excise duty applied on manufactured
income countries, the European structure may rely mainly on ad tobacco and Directive 2008/118/EC).
Region (except Norway) relies on valorem or mainly on specific duties. The upper limit has been increased,
a mixed system, and almost all In the EU, Member States must allowing Member States relying more
other countries on a purely specific apply a mixed structure (Fig. 2.1). on specific duties in the context
Figure 2.1. Share of ad valorem and specific taxes in total excise duties: EU27
Figure generated by the Working Group based on data published in the excise duty tables in European Commission (2010a). Excise duty tables 2010, Part III- Manufactured Tobacco.
Brussels, European Commission. http://ec.europa.eu/taxation_customs/taxation/excise_duties/tobacco_products/rates/index_en.htm). Abbreviations: AT, Austria;; BE, Belgium;; BG,
Bulgaria;; CY, Cyprus;; CZ, Czech Republic;; DE, Germany;; DK, Denmark;; EE, Estonia;; EL, Greece;; ES, Spain;; FI, Finland;; FR, France;; HU, Hungary;; IE, Ireland;; IT, Italy;; LT, Lithuania;;
LU, Luxembourg;; LV, Latvia;; MT, Malta;; NL, Netherlands;; PL, Poland;; PT, Portugal;; RO, Romania;; SE, Sweden;; SI, Slovenia;; SK, Slovakia;; UK, United Kingdom.
17
of a health-driven taxation policy. Specific duties may have several However, rates do cluster by income
To maintain a minimum level of merits as concerns their impact on groups, and some common features
harmonization of the tax structure, prices, consumption and revenues. can be seen on an average level
the lower limit should, in theory, have when moving from one income
been increased to the same extent. Changing an ad valorem structure group to another. Figure 2.2 shows
However, due to strong opposition to a more specific structure the average levels of the price, tax
of several countries applying high amount and tax share of the price of
ad valorem duties, a compromise Objectives may change over time. the most-sold brands of cigarettes
was reached at only 7.5%. As a Where an ad valorem structure was by income groups based on World
consequence, there is flexibility preferred in the past, specific duties Bank classification in 2008 (http://
resulting in widely differing practices, may now fit better to achieve revenue data.worldbank.org/about/country-
reflecting the differing objectives and health objectives. However, a classifications/country-and-lending-
that the countries want to achieve by change from an ad valorem structure groups). There is a clear downward
means of tobacco taxation. into a specific will have a certain cost: trend in the tax and price level as
Within these systems different tax the tax burden on the lower-priced the level of income of countries goes
rates may apply for different products cigarettes would be increased, but down. At the extremes the total tax
(e.g. filtered and unfiltered cigarettes, the change to specific taxation would rate for low-income groups is almost
cigarettes versus fine-cut tobacco) result in a reduction of the tax burden 40% lower than the tax rate of high-
which render the tax structures very on higher priced brands. This might income groups.
complex. have undesired effects on consumer Figures 2.3 and 2.4 show details
There are many factors that can prices of premium brands and on the of tax levels by countries and by WHO
influence the choice of one structure profits and competitive positions of region. The rates shown in Figure 2.3
over another. Regulators will consider tobacco manufacturers. Producers apply to the most-sold brand;; those
the effects on tax revenues, prices, of premium brands would get a tax rates of course vary within a country
profits and competitive positions subsidy, while their competitors when price differentials are large
of tobacco producers, quality and would potentially face a significant among brands and particularly when
variety of products as well as the increase of the tax burden on their differential excise tax rates are applied
ability to administer the tax. products. To avoid these effects, a on different brands and products. For
Because specific duties are the switch to a more specific tax structure example, in Bangladesh, the total tax
same for all cigarettes, independent could be done in the context of tax share of the retail price is 47% for the
of the price, they will reduce the increases, i.e. by increasing the cheapest brand pack of 20 cigarettes
relative price differentials between specific element while leaving the ad and 87% for a pack of Marlboro
high- and low-taxed cigarettes. This valorem unchanged. Another option cigarettes. In the Philippines the total
may lead consumers to switch to is the introduction of a minimum tax tax share of the cheapest brand pack
higher-priced cigarettes, assuming floor that does not affect the tax of cigarettes is 54%, and goes up to
that more expensive cigarettes are burden on higher priced brands. 76% for the Marlboro brand (World
considered to be of a higher quality Health Organization, 2010).
(and thus more appealing). They Tax rates
have an upgrading effect that favours Differential taxation between
high quality, which may lead to a Overview tobacco products
higher average price.
Ad valorem duties are a Tax rates vary greatly across Most countries apply different rates
percentage of the price of cigarettes countries, ranging from countries on tobacco products, either to protect
and will maintain the relative (pre-tax) with no excise imposed on tobacco their domestic industry (e.g. India
price differentials between high- and products (e.g. Benin, Cook Islands, where much lower taxes are applied
low-taxed cigarettes. Consequently Grenada, Maldives, Saudi Arabia) to on bidis compared with cigarettes),
there will be more price competition an excise tax representing more than or because of the perception that
under an ad valorem system, which 75% of the retail price of a pack of the the products are of a different nature
may entail a lower average price. most sold brand of cigarettes (e.g. (e.g. cigars, water pipe tobacco,
They have a multiplier effect that Cuba, Fiji, Poland and Seychelles) chewing tobacco versus cigarettes).
favours low quality. (World Health Organization, 2010).
18
Figure 2.2. Price and tax average for a pack of the most-sold brand of cigarettes by income groups, 2008§
Source: Adapted from World Health Organization (2009). WHO report on the Global Tobacco Epidemic 2009: implementing smoke-free environments. Geneva, World Health Organization.
§
July 2008 World Bank classification of countries by income
Figure 2.3. Cigarette price, excises, and other taxes as in 2008, by region
WHO African Region (AFRO)
19
WHO Region of the Americas (AMRO)
WHO Eastern Mediterranean Region (EMRO)
WHO South-East Asia Region (SEARO)
20
WHO Western Pacific Region (WPRO)
WHO European Region (EURO)
Source: World Health Organization (2009). WHO report on the Global Tobacco Epidemic 2009: implementing smoke-free environments. Geneva, World Health Organization.
Notes: Price of the most sold brand in the country converted into US dollars using official (principal or market) exchange rates at end of time period;; total tax share includes specific
excise, ad valorem excise, value added tax (VAT), imported tax duty (if the most popular brand in the country is imported), and others (if applicable);; un-weighted arithmetic average.
21
Figure 2.4. Taxes on cigarettes in the European Union - 1 January 2010
Figure generated by the Working Group based on data published in the excise duty tables in European Commission (2010a). Excise duty tables 2010, Part III- Manufactured Tobacco.
Brussels, European Commission (data from 2010 used to produce the graph available upon request);;
http://ec.europa.eu/taxation_customs/taxation/excise_duties/tobacco_products/rates/index_en.htm).
Abbreviations: AT, Austria;; BE, Belgium;; BG, Bulgaria;; CY, Cyprus;; CZ, Czech Republic;; DE, Germany;; DK, Denmark;; EE, Estonia;; EL, Greece;; ES, Spain;; FI, Finland;; FR, France;;
HU, Hungary;; IE, Ireland;; IT, Italy;; LT, Lithuania;; LU, Luxembourg;; LV, Latvia;; MT, Malta;; NL, Netherlands;; PL, Poland;; PT, Portugal;; RO, Romania;; SE, Sweden;; SI, Slovenia;; SK,
Slovakia;; UK, United Kingdom.
Little data is collected on the tax rise to product substitution. Mainly On the other hand, chewing
rates applied on products other than as a result of increased taxation, the tobacco is almost always taxed at a
cigarettes on a global scale. Data quantities of cigarettes released in significantly lower rate than cigarettes
collected by the WHO Reports on the the EU-251 decreased by around 14% (e.g. Algeria, Bangladesh, Guatemala,
Global Tobacco Epidemic can provide between 2002 and 2008. Conversely, Nigeria, Pakistan, Sri Lanka and the
some examples of tax rates for some the quantities of fine-cut tobacco Bolivarian Republic of Venezuela).
countries and some tobacco products increased in the same period by To the extent that these products can
(World Health Organization, 2009). around 18% (European Commission, be substituted for cigarettes, tobacco
Table 2.3 provides tax rates of roll- 2010a). taxation can be rendered ineffective
your-own (RYO) tobacco and chewing Moreover, in contrast to ready- if other products remain taxed at a
tobacco as well as cigarettes for made cigarettes, which are taxed much lower rate.
selected countries (where data was on a per-stick basis, specific duties
available). For comparison purposes, on fine-cut tobacco are levied on a Differential taxation within the
the tax rate was applied for the price per-kilogram basis. The conversion same tobacco products
of 20 g of RYO and chewing tobacco, rate between kilogram and sticks is
assuming that a pack of 20 cigarettes a controversial issue. Traditionally, 1 Several countries also apply
would weigh 20 g (1 g of tobacco per kg is supposed to correspond to 1000 differential tax rates within the same
cigarette). For RYO, tax rates seem to sticks. However, the average weight of tobacco product. For example, about
be similar to those for cigarettes but a cigarette is estimated to be around 20% of countries that collect excise
can also vary widely, with the rates on 0.75 gr. Moreover, new products taxes on cigarettes impose different
cigarettes being higher (e.g. Georgia, called “volume tobacco” have gained rates on the cigarettes consumed
Mongolia, Thailand and the United a significant market share. Each kilo in the local market. The differential
Kingdom). In the EU, the minimum of “volume tobacco”2 used can yield a rates vary depending on different
rates for fine-cut are around 50% of savings of at least a half-kilo of cut filler. characteristics of cigarettes: this is
the minimum rates for cigarettes. The Consequently, the real tax burden as usually the retail price level, but can
gap between the level of taxation of well as the increase in volumes of fine- also be the production volume, the
cigarettes and fine-cut tobacco gives cut tobacco may be underestimated. sales volume, the type of cigarette
1
No full data available for RO and LT
2
In contrast to classical fine-cut tobacco, the raw material undergoes an expansion process which leads to an increase of the tobacco volume. One expansion process is called
DIET (dried ice expanded tobacco), but there are other processes/patents (IMPEX, IMCON) with comparable effects on the volume of the tobacco. All expansion processes fall
under the generic term “volume tobacco.”
22
Table 2.3. Tax share of the price of 20g of RYO, chewing tobacco of products and can easily encourage
and 20 cigarettes, selected countries, 2008
substitution of consumption to
lower-taxed products when a tax
Country Cigarettes RYO (20g) Chewing tobacco (20g)
is increased instead of reduction
Paraguay 19% 19% in consumption or quitting, thereby
El Salvador 31% 61% defeating the purpose of tax policies
Nigeria 32% 32% 5% from a public health perspective. From
a tax administration’s perspective,
Mongolia 37% 13%
applying differential tax rates to
Malaysia 48% 46%
the same type of product is not
Pakistan 52% 14% advisable because it makes it more
Georgia 55% 15% difficult for tax collectors to assess
Eritrea 55% 55% the manufacturer’s tax liability, and it
India 55% 33% provides incentives for manufacturers
to avoid taxation by making sure their
Guatemala 57% 11%
products fall in the lower end of the
Australia 62% 67%
tax rate.
Swaziland 62% 20%
Samoa 63% 64% The impact of tax rates and tax
Japan 63% 20% structure
Thailand 64% 15%
Impact on prices
Thailand 64% 25%
Canada 65% 24%
If public health is a concern in the
Uruguay 66% 77%
tobacco tax policy, policy-makers need
Bangladesh 67% 15% to know if the tax increase will actually
Madagascar 67% 44% lead to a price increase and to a
Algeria 68% 15% consequent reduction in consumption.
New Zealand 69% 68%
Under perfect competition the
consumer price cannot increase by
Sri Lanka 72% 13%
more than the amount of tax increase.
Norway 73% 78%
However, under imperfect competition
Italy 75% 45% taxes may be under or over-shifted;;
Myanmar 75% 25% that is, the consumer price may rise
Netherlands 76% 38% by less or more than the amount of the
Venezuela 78% 8%
tax. While either specific or ad valorem
may be over-shifted, this is more likely
Czech Rep 79% 77%
to happen with specific (Delipalla and
UK 80% 55%
Keen, 1992). For empirical evidence,
Source: World Health Organization, unpublished data see Chapter 3.
The tobacco industry works like
(with or without filters, hand or (machine- and hand-made), the rate an oligopoly with a large number of
machine made), the packaging varied between 65 to 310 Rupiahs per consumers and a few producers.
(hard/soft), the cigarette length, etc. stick in 2010. In Brazil, the excise tax Seventy percent of the world’s cigarette
(World Health Organization, 2010). varied between 0.764 to 1.397 Reals market is controlled by 4 multinationals
Tax rates can vary substantially. For per pack of 20 cigarettes depending (Euromonitor International, 2009).
example, in Indonesia tax rates vary on the size of the cigarettes and the In China, however, the market is
by tobacco product but also within type of package (unpublished data controlled by only one company
the same product depending on the received from governments reporting owned by the government. In these
retail price and production capacity of their tax rates). These differential rates non-competitive markets, instead
manufacturers. In the case of kreteks lead to price gaps within the same type of absorbing a tax increase and
23
reducing its profits, the industry Impact on revenues (budgetary Impact on variety of products
can easily pass it on to consumers stability)
(leading to an increase in price) Specific duties provide incentives for
without fears that this would lead Both excises may have an impact more appealing and higher-priced
to a switching to cheaper products on prices;; increases in ad valorem products as well as a greater variety
offered by a competitive company. duties are likely to decrease prices, of products. Upgrading effects
In addition, with the growing support and vice versa for specific duties. of specific taxes will reduce the
to tobacco control policies and the Since specific duties are independent relative tax burden on high quality
expected decrease in consumption of changes in price, they generally products, which provides incentives
in the future, some suggest that produce a more stable stream of for higher quality and greater
the industry will tend to raise prices revenue. Revenue from ad valorem variety of products. Conversely, the
now and save its profits instead of duties is dependent on prices, and multiplier effect under ad valorem
keeping prices low to stimulate future may vary over time depending on the duties provides a disincentive to
consumption (Becker et al., 1994). consumer and producer behaviour. costly quality improvements. From
Of course, such industry Moreover, forecasting revenue in an a tobacco control and prevention
behaviour leads to larger increases accurate way may be very difficult perspective, a higher quality and
in the price of cigarettes, which is as one must predict changes in greater variety of products are not
good for public health. However, this consumer and producer behaviour. desirable, as these increase the
means more revenues and power to In conclusion, specific duties may attractiveness of tobacco products.
the tobacco industry, revenues that entail higher prices and more stable
would be more useful if they ended revenues but tend to favour higher Industry profits and the share
up in the coffers of the government priced and more appealing brands. of taxes in the retail price
instead. Conversely, ad valorem taxes
In the case of the China National have several merits concerning their The theory suggests that profits
Tobacco Co. (CNTC) government- impact on product quality and variety, are relatively higher under specific
owned monopoly, the relation the share of tax in the retail price, taxation (Delipalla and Keen, 1992).
between taxes and prices is very and their ability to keep pace with Specific duties will entail a relatively
different. The tax policy is not used inflation. lower tax burden on premium brands
as an instrument to influence price, Chapter 9 provides an in-depth as compared to cheaper brands.
since the government determines analysis of the impact of tax rates on Ad valorem duties ensure the same
both the tax and price levels. tax revenues. tax burden for all price categories.
If we compare data at the global In particular in premium markets,
level, the average price of a pack of Adjustment of taxes for inflation specific duties may entail a lower
the most-sold brand of cigarettes average price/tax ratio. In high-
is higher in countries relying only The real value of a specific tax will income countries with similar price
on specific excise (2.46 US$) than be eroded by inflation and therefore levels, countries relying on specific
in countries relying only on ad must be periodically adjusted. duties have on average a lower share
valorem (1.29 US$). The same is Conversely, as they are value-based, of tax in the retail price. On average,
true for countries applying a mixed the real value of ad valorem taxes will the pre-tax prices are in all income
system but relying more on specific be preserved, as the tax will increase groups the highest in countries with
(3.87 US$) than on ad valorem to the extent that tobacco prices specific duties only, and the lowest
(3.14 US$). Similar conclusions follow inflation. An advantage of a in countries with mixed systems.
hold also if we account for the minimum tax floor (which is similar Countries with mixed systems
income level of countries (World to a specific duty) is that it will be have, apart from the low-income
Health Organization, 2010). This is automatically adjusted for inflation if economies, the highest share of tax
not meant to hold as an empirical it is set as a percentage of the excise in the retail prices.
justification for the above statement, due on a premium price category Tax authorities may prefer ad
but it indicates that specific excises subject to ad valorem taxation (e.g. valorem duties to maximize the share
seem to lead to higher prices. the most popular price category of tax in the retail price. Assuming
(MPPC) in most EU Member States). identical price levels, this would lead
to higher tax revenues. However,
24
ad valorem duties may render price Administration and enforcement The choice between ad valorem
and tax policies more vulnerable to duties and specific duties is a long-
industry pricing policies and changes If the administrative capacity of a standing and controversial issue. To
in consumer behaviour. country is weak, specific duties may have the best elements of both, it is
have the advantage that it is easier possible to combine an ad valorem
Other considerations: to determine the physical quantity with a specific tax. A so-called mixed
industry characteristics (number of cigarettes, weight of structure applies an ad valorem
the tobacco) than the value of the and a specific duty to all tobacco
Specific duties tend to favour more products. Ad valorem duties require products. Mixed systems can give
expensive products. Conversely ad some monitoring of prices for preference to more ad valorem or
valorem duties will entail a higher reasons of tax collection. In some to more specific duties depending
absolute amount of tax for premium countries (e.g. Philippines and the on the desired effects. The more
market segments as compared to Russian Federation), ad valorem the structure relies on the specific
cheaper brands. The relative price duties have involved valuation duties, the more the upgrading effect
differentials between pre-tax prices problems because manufacturers will come into play. The more the
will under ad valorem systems be had the potential to sell their products structure relies on ad valorem duties,
better reflected in the tax-inclusive to a related wholesale company at the more the multiplicator effect will
consumer prices. Depending artificially low prices (transfer pricing) come into play and the more excise
on the industry characteristics, to reduce excise duties ( World Health duties will be automatically adjusted
governments may prefer one tax Organization, 2010). However, in for inflation.
over the other. In particular in highly these countries ad valorem duties Ad valorem excise duties or
concentrated markets, where around were levied on ex factory prices. In mixed excise duties can also be
80% of the price of cigarettes the EU, where ad valorem duties are combined with a minimum tax floor.
consists of tax (excise and VAT) levied on the maximum retail selling Minimum excise duties are similar
and where there is to a large extent price to the final consumer, such to specific excise duties and are a
a ban on product advertising, a problems have not been reported. fixed monetary amount per quantity
certain level of ad valorem duties On the other hand, under specific or volume that ensures a minimum
will contribute to a more competitive taxation the manufacturer can tax floor. The effects of a minimum
environment, usually translated into manipulate the length of the cigarette tax floor are similar to specific duties;;
lower pre-tax prices. To balance the or the size of the pack to reduce however they tend to apply only to
tax burden on the different market the excise duties. For example, in lower- and medium-priced brands.
segments, countries may opt for a the EU, to reduce the excise duties Premium brands will remain taxed
mixed system. on cigarettes, new products have at the ad valorem rate or mixed
However, the appropriate been marketed consisting of rolls rate, which ensures a higher tax
balance is a very controversial of 18 cm and separate filter tubes burden as compared to a purely
issue. In some countries ad valorem (Commission of the European specific system and a disincentive for
duties are advocated to protect the Communities, COM(2008) 460/2) quality improvements. Over the last
cheaper domestic brands against (European Commission, 2008b). The decade, 24 EU Member States have
more expensive but more attractive rolls are subsequently cut into three introduced such a minimum tax floor.
international brands. Others pieces by the consumer and inserted In most cases they are a percentage
countries with plants manufacturing in the filter tubes;; as for tax purposes (in general between 95% and 105%)
international brands will for a similar the maximum length of cigarettes is 9 of the taxes due on the weighted
reason prefer specific duties. cm, three cigarettes are taxed as t wo. average price or on the most popular
Moreover, the current tax structure Moreover in many countries, in reply price category (unpublished data:
is often a holdover from a past when to tax increases, manufacturers have Internal Reports from EU Member
tobacco manufacturing played a more reduced the number of cigarettes per States to the EC). If these reference
important role (e.g. the southern EU pack (e.g. 17, 18 or 19 instead of 20 prices change, the minimum tax floor
Member States still rely mainly on ad pieces) to maintain the same price will be adjusted. As the ad valorem
valorem duties, although, apart from per pack (unpublished data: Internal duty will apply to these reference
Greece, the domestic production has Reports from EU Member States to prices, the minimum tax floor will at
disappeared over time). the EC). least partly be adjusted for inflation.
25
In conclusion, provided the minimum 9 provides an overview of studies of countries having high taxes and vice
tax f loor is set at an appropriate level, the effectiveness of earmarking on versa. Simultaneously, high-income
this structure will combine a high price tobacco use and health improvement. countries tend to favour specific
and tax level for lower- and medium- Recently, the Taskforce on excise tax structures, while low- and
priced brands with a relatively high Innovative International Financing for middle-income countries rely more
tax price ratio for premium brands. Health Systems—established from on ad valorem excise taxes.
Table 2.4 summarizes the 2008 to 2009 and chaired by United Given the structure of tobacco
differences between structures of Kingdom Prime Minister and World market globally, specific excise taxes
excise duties and their impact. Bank President—recommended generally result in higher tobacco
exploring the technical viability of product prices. In these markets,
Earmarking of tobacco tax solidarity levies on tobacco. Assuming specific excises can increase
revenues that a broad range of countries would tobacco companies’ pricing power,
participate in the increase of tobacco raise profits and increase market
With the challenges governments of taxes, the revenues generated concentration.
low- and middle-income countries would be substantial (Taskforce on Some countries have designed
face to finance their health systems, Innovative International Financing more complex taxation structures in
and the additional problems the for Health Systems, 2009). This an attempt to find a balance between
recent financial crisis has imposed, demonstrates the increasing political budgetary, health and competition
developing innovative means of support for such type of financing and objectives.
financing has become an essential the importance of tobacco taxation, About 75% of the world’s tobacco
issue to address to ensure particularly for countries with limited product market is accounted for
sustainability and autonomy of health resources. The benefit of such by cigarettes. However, in some
care. Earmarking or dedicating practices will be double: reducing countries tobacco products other
tobacco taxes for health programmes consumption and funding health than cigarettes have an important,
is a promising efficient way of raising systems, both leading to improved sometimes significant, market share.
resources internally. public health. Tax levels are often much lower
Earmarked taxes are general or on these products as compared
special taxes committed to support, or Conclusions to cigarettes. Differential rates are
fully fund, pre-specified expenditure applied on different tobacco products
items (Gwilliam and Shalizi, 1996). Worldwide different types of taxes and even sometimes within the same
Earmarking can be weak (purely apply to tobacco products, with product category, resulting in price
a formal undertaking to make the different tax levels (rates) contributing gaps and opportunities for product
system more transparent and to inform to significant price differentials. substitution to lower-taxed products.
the taxpayer of the cost of a service) Historically, revenue generation has Applying a similar tax level would
or strong (revenue determines been the primary aim of tobacco reduce the incentive for substitution
expenditure), wide (covering a whole taxation. However, more and more the and increase the effectiveness of
spending programme) or narrow retention and the increase of excise taxation policy in reducing tobacco
(specific project within a programme) duties on tobacco products aims to use.
(Wilkinson, 1994). improve public health by reducing Given their important revenue-
Countries around the world tobacco consumption and the external generating potential, in some
earmark part or all of their tobacco cost of smoking. Nonetheless, countries, parts or all of tobacco tax
taxes to a specific programme when determining taxation policy, revenues have been used to fund
or activities (e.g. Ecuador, Egypt, governments will take into account health or tobacco control activities.
Estonia, Finland, Korea and Thailand). other, at times competing, political, This practice can be adopted in
This is also the case for several social or economic considerations. low-resource countries as a way to
US states like Arizona, California, These considerations are often strengthen existing health systems,
Massachusetts and Oregon. More reflected in the applicable tax as proposed by the Taskforce on
than 20 countries specifically structure and rates. Innovative International Financing for
use their tobacco taxes for health Moreover, the tax levels are Health Systems (2009).
programmes, mainly for tobacco in general directly related to the
control or health promotion. Chapter income levels, with high-income
26
Table 2.4. Comparison of structures for excise duties
Ad valorem with a specific Mixed with a specific

Effect on prices Specific excise Ad valorem Mixed
minimum floor minimum floor
Tax/price structure
Impact on prices Tends to entail relatively Relatively lower prices The effect will depend on The minimum tax functions as a specific duty and ensures
higher prices compared to specific which element (ad valorem or a relatively higher price level for low- (and medium-) priced
Tax increases may lead to Tax increases may entail specific) prevails products
“over shifting” or upwards down trading or price
product substitutability reductions (“under shifting”)
Inflation Real value will be eroded by The real value will be Real value of the specific Real value of the minimum Real value of the specific tax
inflation, unless adjusted in preserved as prices increase element will be eroded by floor will be eroded by and the minimum floor will be
line with inflation to the extent that tobacco inflation inflation eroded by inflation
prices follow inflation
The minimum floor may be (partly) adjusted for inflation if it is
a percentage of the excise due on WAP or of a premium price
category subject to ad valorem taxation (e.g. the MPPC in
most EU Member States).
Consumers: Ad valorem with a specific Mixed with a specific

Specific excise Ad valorem Mixed
quality and variety minimum floor minimum floor
Impact on product quality and Upgrading effects tend to The multiplier effect provides The effect will depend on The multiplier effect of the ad The multiplier effect of the ad
variety reduce the relative tax on a disincentive to costly quality which element (ad valorem or valorem element provides a valorem element provides a
high quality products, which improvements specific) prevails disincentive to costly quality disincentive to costly quality
provide an incentive for improvements improvements
higher quality and greater
variety of products

Revenue Specific excise Ad valorem Mixed
Budgetary stability/ ability to More stable as compared to Vulnerable to changes in
forecast ad valorem. Easy to forecast consumers and producers’ More specific or a minimum tax floor will entail more budgetary stability
behaviour Difficult to forecast
The real value of taxes and The excise needs to be The excise may have to be The excise may have to be The excise may have The excise may have to be
prices periodically adjusted for periodically adjusted for down periodically adjusted for to be periodically adjusted periodically adjusted for
inflation trading or price reductions inflation, down trading or for inflation, down trading or inflation, for down trading or
price reductions price reductions price reductions
27
28
Table 2.4. Comparison of structures for excise duties

Revenue Specific excise Ad valorem Mixed
The more the structure relies on specific or minimum duties the less vulnerable it becomes for
down trading or price reductions;; but the more the tax can be eroded by inflation
Reduction of tax induced With the same tax level at WAP, a situation of purely specific Mixed tax structures and tax structures including a minimum floor are less vulnerable to cross-
cross-border operations taxation in one country/jurisdiction and purely ad valorem border flow induced by the tax structure
(private imports and illicit taxation in a neighbouring country could result in cross-
trade) border flow for premium brands from one country to a second
country, with cheap brands flowing in the opposite direction
Administrative requirements Low as compared to other indirect taxes (e.g. VAT, sales tax).
No price monitoring required Requires price monitoring Requires price monitoring Requires price monitoring Requires price monitoring
for tax purposes;; only the
volume or weight has to be
ascertained

Manufacturers Specific excise Ad valorem Mixed
Impact of taxes on the profits Will entail a higher tax/price Will entail a higher absolute The gap between tax/price The minimum may ensure a The minimum floor ensures
of the tobacco producers ratio for cheaper market amount of tax for premium ratio for the cheaper and the high absolute amount of tax a high tax/price ratio for
segments and higher profits market segments, because premium market segments and consequently a high tax/ cheaper market segments.
for the manufacturer any increase in the producer will be smaller as compared price ratio for cheaper market The ad valorem element will
price will be taxed to a purely specific system segments. apply to premium market
The ad valorem ensures the segments
desired tax/price ratio for
premium market segments
(equal to the ad valorem rate)
Market shares – Protection of Tends to favour premium May protect the cheaper More specific or a minimum tax floor tend to favour more expensive brands
domestic producers brands domestic brands against
more expensive international
brands
WAP: Weighted Average Price;; MPPC: Most Popular Price Category

References
Becker GS, Grossman M, Murphy KM (1994). European Commission (2010b). Special McCarten WJ, Stotsky J (1995). Excise taxes.
An empirical analysis of cigarette addiction. Eurobarometer 332/Wave 72.3 - Tobacco. In: Shome P, ed., Tax Policy Handbook.
Am Econ Rev, 84:396–418. Special Eurobarometer 332. Brussels, Washington DC, International Monetary Fund,
European Commission. 100–103.
Chaloupka FJ, Hu TW, Warner KE et al.
(2000). Taxation of tobacco products. In: Gwilliam KM, Shalizi ZM (1996). Road funds, Ray C, Gupta PC (2009). Bidis and smokeless
Jha P, Chaloupka F, eds., Tobacco control in user charges and taxes. Environmentally tobacco. Curr Sci, 96:1234–1334.
developing countries. Oxford, U.K., Oxford Sustainable Development.The World Bank.
University Press, Chapter 10;; 237–272. Smith S (2007). Taxation and regulation of
IARC (2007). Smokeless tobacco and some alcohol, tobacco and gambling.
Cnossen S (2006). Tobacco taxation in the tobacco-specific N-nitrosamines. IARC
European Union. CESifo Working, Paper Series Monogr Eval Carcinog Risks Hum, 89:1–592. Taskforce on Innovative International Financing
N°1718. The Hague, Maastricht University, PMID:18335640 for Health Systems (2009). More money for
CPB Netherlands Bureau for Economic Policy health, and more health for the money. London,
Analysis John R, Rao K, Rao G et al. (2010). Economics International Health Partnership
of tobaco and tobacco taxation in India. Paris,
Delipalla S, Keen M (1992). The Comparison International Union Against Tuberculosis and Wilkinson M (1994). Paying for public
between ad valorem and specific taxation under Lung Disease. spending: is there a role for earmarked
imperfect competition. J Public Econ, 49:351– taxes? Fiscal Studies, 15:119–135
367 doi:10.1016/0047-2727(92)90073-O. Joossens L, Merriman D, Ross H et al. (2009). doi:10.1111/j.1475-5890.1994.tb00213.x.
How eliminating the global illicit cigarette trade
Euromonitor International (2009). Global would increase tax revenue and save lives. World Health Organization (2005). WHO
Tobacco: The Power of Cigarette Pricing. Paris, International Union against Tuberculosis Framework Convention on Tobacco Control.
and Lung Disease. Geneva, World Health Organization.
European Commission (2008a). European
Commission Taxation and Customs Union: Knishkowy B, Amitai Y (2005). Water-pipe World Health Organization (2008). WHO report
Tobacco products legislation. Available (narghile) smoking: an emerging health on the global tobacco epidemic 2008: the
from: http://ec.europa.eu/taxation_customs/ risk behavior. Pediatrics, 116:e113–e119. MPOWER package. Geneva, World Health
taxation/excise_duties/tobacco_ products/ doi:10.1542/peds.2004-2173 PMID:15995011 Organization.
legislation/index_en.htm
KPMG (2005). Study on the collection and World Health Organization (2009). WHO
European Commission (2008b). Report on the interpretation of data concerning the release report on the Global Tobacco Epidemic 2009:
structure and rates of excise’ duty applied on for consumption of cigarettes and fine-cut implementing smoke-free environments.
cigarettes and other manufactured tobacco tobacco for rolling of cigarettes. Brussels, Geneva, World Health Organization.
products. COM 460/2008. Brussels, European European Commission.
Commission. World Health Organization (2010).
Manning WG, Keeler EB, Newhouse JP et WHO Technical Manual on Tobacco Tax
European Commission (2010a). Excise duty al. (1989). The taxes of sin. Do smokers and Administration. Geneva, World Health
tables 2010, Part III- Manufactured Tobacco. drinkers pay their way? JAMA, 261:1604–1609. Organization.
Brussels, European Commission. doi:10.1001/jama.261.11.1604 PMID:2918654
29
30
Chapter 3
Tobacco industry pricing, price-related marketing
and lobbying
Introduction
The effectiveness of tobacco taxation Tobacco Monopoly Administration unbranded products upon redemption
and other price-related policies and Chinese National Tobacco Co. of coupons or other proof of purchase,
depends in large part on how tobacco kept retail cigarette prices at the as in the “Marlboro Miles” and
companies respond through both same level as they were before the “Camel Cash” programmes). From
their lobbying efforts to prevent tax tax increase. the tobacco companies’ perspective,
increases or influence excise policies In addition to its direct pricing such price-reducing marketing
and their pricing and price-reducing strategies, tobacco companies can efforts have advantages over direct
marketing strategies. In general, influence price through a variety of pricing strategies in that they can be
tobacco product tax increases will price-reducing marketing efforts. more narrowly targeted to particular
lead to increases in the price of these These include: the distribution of segments of the market. For example,
products, with the extent of the price free samples of their products, the price-reducing promotions can be
increase varying based on how much distribution of coupons through used subnationally to reduce the
or little of the tax increase the industry print advertising and direct mail, immediate impact of subnational tax
decides to pass on to consumers on packaging, and via their web increases, can be applied to a limited
(which may vary brand by brand sites;; retail value-added promotions number of brands (e.g. those used
within the same market) and whether that involve free cigarettes (e.g. by populations of particular interest,
it raises prices on top. Two 2009 buy-one-get-one-free offers);; and such as young people), or directed to
tax increases illustrate this clearly. direct price discounts implemented more price-sensitive consumers (e.g.
In April 2009, the federal cigarette through payments to distributors distribution of coupons via direct mail
excise tax in the United States and retailers (e.g. through “buy- to selected users).
was increased by US$0.6167 per downs” trade programs). In addition, Finally, tobacco companies can
pack, with US cigarette companies other marketing techniques provide use their political influence to lobby
passing on the full amount of the tax added value at the same price policymakers to oppose tobacco tax
increase and raising prices further and can be thought of as price- increases or support tax rollbacks,
(e.g. Philip Morris USA raised prices reducing marketing, broadly defined. alter tax structures in a way favourable
on its leading brands by US$0.71 These include: retail value-added to at least some companies, resist
per pack and on other brands by promotions that involve non-tobacco earmarking of tobacco tax revenues
US$0.78 per pack). In contrast, in items (e.g. a free lighter with the for tobacco control programs, and/or
May 2009, China modified its tax purchase of two packs of cigarettes);; minimize the use and impact of other
structure and increased taxes on and specialty item distribution (e.g. policies that aim to increase tobacco
most cigarette brands, but the State giveaways of various branded or product prices. Similarly, tobacco
31
companies work behind closed used to drive searches are available Others were more fragmented,
doors with Ministries of Finance and in Appendix. with several smaller companies
others to provide guidance on tax We also consulted several key competing more aggressively with
levels and tax structures, as well as experts in the field (many of whom others. Still others were oligopolistic,
other aspects of tax administration are involved in authoring this volume) dominated by a few large firms that
(Gilmore et al., 2007). Such overt and put a call on GLOBALink, the together controlled nearly the entire
and covert activities can weaken most commonly used information- market. As a result, pricing strategies
the effectiveness of tax and price exchange point for tobacco control varied widely, with more competitive
policies in reducing tobacco use and researchers and practitioners, to markets likely to see lower prices and
its consequences. identify other papers. Although our more price competition than more
This chapter provides a review search strategy should have captured concentrated markets.
of these issues. The first section most studies that explore the issues Over the past two decades, the
provides a discussion of industry covered in this chapter in any detail, markets for tobacco products have
pricing strategies generally and of it is possible that some references become increasingly concentrated
how tobacco product prices change have been missed, particularly those and globalized as a result of
in response to changes in tobacco in languages other than English. privatization of government-run
taxes specifically. This is followed While we had aimed to limit the tobacco monopolies, reduced
by a review of the variety of price- evidence to scientific publications barriers to trade in tobacco and
related marketing strategies used obtained via the search strategies tobacco products, increased direct
by tobacco companies, and the above, the evidence base, investments by multinational tobacco
limited empirical evidence on the particularly in relation to parts (a) companies, and consolidation
use of price-reducing promotions (tobacco industry pricing strategies), through mergers and acquisitions.
in response to tax increases and (b) (price-related marketing) and (d) Together, worldwide, the top five
other tobacco control policies. We (interventions to reduce industry cigarette companies (Chinese
then turn to a discussion of industry price manipulation and lobbying), National Tobacco Co. (CNTC),
lobbying efforts to shape policy was limited, with most of the empirical Philip Morris International (PMI),
decisions concerning tobacco excise studies limited to the USA and a few British American Tobacco (BAT),
taxes, providing a systematic review other high- or high-middle income Japan Tobacco (JT), and Imperial
of the limited empirical evidence on countries. For this reason, the first t wo Tobacco Group (ITG)) account for
these activities between 1985 and sections explain industry behaviour more than 80% of global cigarette
2010. The final section discusses using theory and observation before production and consumption
interventions to reduce industry- turning to the empirical evidence, (Euromonitor International, 2009).
related price manipulation and and the fourth section outlines policy Excluding CNTC, which operates
lobbying. options, highlighting any relevant almost entirely in China with little
academic literature. competition from other companies,
Methods the top four companies account for
Tobacco industry pricing about 70% of global production and
A systematic approach was taken to strategies consumption. As shown in Table
searching for literature relevant to 3.1 for leading cigarette markets
this chapter. In total, 13 electronic Background: Market structure globally and in Europe, in most
databases were searched between countries cigarette markets are
October 2009 and March 2010. Historically, industry pricing dominated by some combination of
These covered academic research strategies have varied considerably these four companies. The 3-firm
and grey literature (i.e. manuscripts, across countries given the significant concentration ratio illustrates the
reports, technical notes or similar differences in the structures of highly concentrated nature of these
documents produced and distributed tobacco product markets in different markets—other than in Indonesia,
by governmental agencies, academic countries. Many markets were just three firms control at least 80%
centers or other non-commercial controlled by a domestic, often of the market, and in many markets
publishing entities). Full details of government-run, tobacco monopoly the top three firms account for more
the electronic databases and web whose primary objective may not than 90% (Gilmore et al., 2010).
sites, and the keywords, and terms have been profit maximization.
32
Table 3.1. Cigarette market shares (%) by global brand owner for the major cigarette markets, 2008
Company Brazil Canada China Germany India Indonesia Italy Japan Russian Federation UK USA
Philip Morris International Inc 9.7 21.3 0.1 36.2 12.1 22.6 52.9 24.4 25.4 6.3
British American Tobacco Plc 86.3 59.0 0.6 20.0 2.7 24.1 10.2 19.6 8.1
Japan Tobacco Inc 10.8 0.2 5.0 1.3 16.2 64.9 36.9 38.8 0.4
Imperial Tobacco Group Plc 25.6 2.9 9.2 43.9 4.0
China National Tobacco Corp (CNTC) 98
ITC Group* 58.3
Golden Tobacco Ltd 10.9
VST Industries Ltd 9.2
Godfrey Phillips India Ltd** 0.4
Gudang Garam Tbk Ltd** 28.3
Djarum PT 13.8
Bentoel Internasional investama Tbk PT 5.9
Nojorono Tobacco Indonesia PT 5.5
Philip Morris USA Inc 48.4
Reynolds American Inc* 26.5
Lorillard Inc 10.1
Liggett Vector Brands Inc 1.8
Societé industrielle des Tabacs du Cameroun 1.4

SA
Donskoi Tabak OAO 3.7
Private label 9.9 1.7
Others 2.6 8.9 1.2 3.3 7.8 21.2 4 0.5 5.2 1.3 8.8
Total 100 100 100 100 100 100 100 100 100 100 100
No. of companies with market share >10% 1 3 1 3 3 3 3 3 3 2 3
No. of companies with market share over 25% 1 1 1 2 1 1 1 1 2 2 2
3-firm concentration ratio (cumulative share 97.4 91.1 98.8 81.8 81.3 64.7 93.2 99.5 81.9 90.8 85
of market of 3 biggest companies by market
share)
Table reproduced with permission from Gilmore et al., 2010. Source: Euromonitor International from trade sources/national statistics. Data obtained: 23 September 2009
# Data given for the world’s largest cigarette markets (China, Russia, USA, Japan, Indonesia, Ukraine, Brazil, India), plus the 2 largest European markets (Italy and Germany) and the UK
NB – where companies other than those listed have a market share of 1% or less, their share has been added to the «other» category
Tobbaco industry pricing, price-related marketing and lobbying
*Part owned by BAT. **Part owned by Philip Morris International
33
This market concentration gives the agree on a shared excise framework. tending to decrease with the trend
tobacco companies greater pricing Historically the southern European toward greater industry consolidation
power and profitability, which in countries had state monopolies or described above.
turn may present problems to public small national industries and thus Unfortunately, there has been little
health as further explored below sought ad valorem systems to both research to date on pricing strategies
(Gilmore et al., 2010). protect their cheaper cigarettes in tobacco product markets, with the
Similar patterns are emerging in made using domestic tobacco and exception of some limited research
markets for other tobacco products, to enable them to compete with largely from the USA but also
although to a lesser extent than in higher-priced premium brands from covering Jamaica, Europe and South
cigarette markets and with greater the larger producers (Gilmore and Africa on the relationships between
variability across countries. Globally, McKee, 2004a). By contrast, the tobacco taxes and prices (discussed
non-cigarette product markets are transnationals favoured a wholly below). Instead, a few stylized facts
becoming increasingly concentrated, specific system to help protect their emerge from observations of pricing
and distinctions between cigarette more expensive brands. A similar strategies in various countries, from
companies and other tobacco product pattern continues to this day with, industry analyst and market reports
companies are disappearing as for example, PMI (whose high-price and from internal tobacco company
major cigarette manufacturers have brand Marlboro is the leader in many documents released following a
acquired leading smokeless tobacco markets) favouring a fully specific series of litigation cases in the USA,
product manufacturers (e.g. Altria’s system, Imperial Tobacco (which has which led to several major tobacco
(parent company for Philip Morris a mixed portfolio) favouring a mixed companies being required to make
USA) acquisition of US Smokeless ad valorem and specific system, and internal documents public.
Tobacco Co. in 2008 and Reynolds the small Greek companies lobbying Limited price competition. The
American’s acquisition of Conwood for a fully ad valorem system. This oligopolistic nature of most tobacco
in 2006) and have begun using the issue is covered further in the third product markets (Table 3.1) tends
same brands across product lines section. to result in relatively stable prices
(e.g. Marlboro and Camel cigarettes with limited price competition,
and snus) (McNeill and Sweanor, Tobacco industry pricing particularly in more mature markets
2009). The acquisition of independent strategies in high-income countries where
smokeless manufacturers will, by companies are better established.
reducing competition between In general, tobacco product markets This appears less true for markets
cigarettes and smokeless products, that have large numbers of producers in low- and middle-income countries,
increase the market power (their producing slightly differentiated particularly where the transnational
ability of to set prices) of the resulting products will see a relatively high tobacco companies are relatively
multiproduct tobacco companies. degree of price competition, with new players in the market and more
However, in some countries, where prices largely reflecting the costs likely to use price as one strategy
a much wider variety of tobacco of production and distribution for gaining market share for their
products is available, the markets (including taxes). At the other brands (Vateesatokit et al., 2000;;
for non-cigarette tobacco products extreme, markets that are entirely Szyilágyi and Chapman, 2003).
remain fragmented and highly or nearly monopolized by a single, However, as companies become
competitive (e.g. India, where there profit-maximizing firm will see prices more established, the extent of price
are thousands of small bidi producers). well above costs and relatively little competition appears to diminish,
price competition. However, most particularly in countries where
Background: Individual company tobacco product markets in most incomes are rising more rapidly and
interests countries are at neither extreme, but consumers, influenced by industry
are instead highly concentrated with marketing, increasingly switch to
Different companies vary in their a small number of highly profitable higher-priced brands. To the extent
brand portfolios and thus in their companies accounting for most sales that price competition does remain, it
preference for excise structures. (see Table 3.2). As a result, the nature tends to appear in the form of multiple
This is illustrated through industry of price competition can vary widely, price tiers (e.g. premium, mid-priced
lobbying on tobacco taxes in Europe, both across countries and over time, and economy brands). The existence
where different Member States must although globally competition is of multiple price tiers, more likely in
34
Table 3.2. Profitability (measured using EBITA Margin (%)) for Europe’s two major tobacco companies
and comparator European consumer staple companies
Company 2004 2005 2006 2007 2008* 2009* 2010* 2011*
Tobacco companies
British American Tobacco 24.0 28.1 28.7 30.0 30.7 31.1 32.1 33.7
Imperial Tobacco Group 40.2 41.5 42.9 45.0 28.2 37.7 39.4 39.5
Food companies
Cadbury 15.6 15.9 14.4 13.5 12.0 13.0 13.8 14.9
Danone 12.7 13.1 13.3 12.1 14.4 16.9 15.7 15.9
Nestle 12.7 12.9 13.5 14.0 14.3 14.4 13.0 13.2
Premier Foods 12.9 13.7 13.8 12.5 11.9 12.0 11.9 11.7
Consumer products companies
Unilever NV 15.5 14.8 14.3 14.5 14.6 14.7 14.9 15.1
Henkel 9.4 9.7 10.2 10.5 10.3 9.0 10.6 11.6
L’Oreal 15.3 15.6 16.4 16.6 15.5 14.3 14.9 15.5
Reckitt Benckiser 19.3 20.1 21.5 22.6 23.4 23.9 23.2 23.6
Beverage companies
Heineken NV 13.6 13.1 13.0 14.8 13.2 13.5 14.0 14.4
SABMiller 18.1 20.2 16.9 16.8 16.6 16.8 17.3 18.5
Carlsberg 8.8 8.7 9.6 11.5 13.2 16.0 16.3 17.1
Diageo 28.7 29.0 28.2 28.3 28.5 28.9 31.5 31.8
Table reproduced with permission from Gilmore et al., 2010

* Estimated values
35
markets with wholly ad valorem or The report goes on to describe The tobacco manufacturers involved
mixed tax systems (as explained in how Philip Morris had historically were Imperial Tobacco and Gallaher,
Chapter 2), will facilitate downtrading been one of the followers in the US who jointly account for around 80%
to cheaper brands, which may be cigarette marketing, matching the of the United Kingdom market (see
of particular importance in markets price changes initiated by American Table 3.1). Each manufacturer had
at the end stage of the tobacco Tobacco Co. or R.J. Reynolds, but a series of individual arrangements
epidemic when smoking increasingly noting that by the 1970s, Philip Morris with each retailer whereby the retail
becomes a habit of the more had become the leader: price of a tobacco brand was linked
economically deprived. However, in “We no longer follow the to that of a competing manufacturer’s
oligopoly markets, companies tend to market;; whether we initiate brand (Office of Fair Trading. http://
have brands in every price tier;; thus a price increase or not, our www.oft.gov.uk/shared_oft/press_
the existence of multiple price tiers decision is a key factor in release_attachments/Parties_and_
will not necessarily result in genuine establishing a new industry fines.pdf;; http://www.guardian.
competition. price level, and we must c o.uk /business/2010/apr/16/of t-
Price leadership. In more examine any price move in levies-225m-for- cigarette -price -
concentrated, oligopolistic markets, light of our own judgement of manipulation).
pricing strategies tend to be the appropriate level.” Prices below short-run, industry
characterized by price leadership, The report further describes profit maximizing level. While prices in
with one firm (usually the firm with how the relative absence of price highly concentrated tobacco product
highest market share) initiating a competition in the market allowed markets are above the competitive
change in prices and others almost prices to remain above competitive level, they have historically been well
immediately matching that change. levels, generating high profits which below the level that would maximize
As Chaloupka (2007) describes, were then used to support the other short-run industry profits. This is
cigarette pricing in the USA has been marketing activities through which likely to be explained by two factors.
characterized by this type of strategy US cigarette companies competed First, prices well above competitive
for the past century, with the “leader” more directly. levels would create opportunities
changing over time, but the strategy A similar pattern of price for new firms to gain a foothold in
staying constant. This is further leadership, limited price competition the market by selling cigarettes at
described in a 1976 report on “Pricing and high profitability is seen in relatively low prices. As Chaloupka
Policy” from Philip Morris’ Business Europe, although markets where PMI (2007) describes, this happened in
Planning and Analysis division: is the market leader (most markets in US cigarette markets in the 1930s,
“The cigarette industry is western Europe) tend to differ from when the leading companies at the
characterized by economists the United Kingdom and Irish markets time (American Tobacco Co., R.J.
as a ‘kinky oligopoly’. This because PMI is less prepared to Reynolds and Liggett & Myers) set
charming term implies that lose market share by increasing prices well above competitive levels
the general price level is prices, and thus price increases (US$0.14–0.15 per pack), using
determined by a small number have traditionally been lower in profits to compete primarily through
of firms (price leaders);; that these markets than the United their advertising campaigns. This
no economic advantage can Kingdom and Ireland (Spielman and created an opportunity for new firms
be obtained by any one firm Loveless, 2008a). This makes the to enter the market and for existing
pricing below the general United Kingdom market particularly small firms to gain market share by
price level;; and that major interesting as a case study of industry selling at considerably lower (but
disadvantages accrue to a attempts to maximize short-term still profitable) prices. Many did so
firm which attempts to price profitability, as outlined below. A with their “ten-cent” brands before
above the general level. In further issue worth noting is that of the industry leaders responded by
short, the general price level price-fixing, recently documented in cutting their prices. The ensuing price
results from some sparring the United Kingdom. The Office of Fair war drove most of the new firms from
among the potential price Trading recently fined two tobacco the market, but two – Philip Morris
leaders, after which the rest manufacturers and ten leading and Brown & Williamson – survived,
of the industry accepts the retailers £225m for price-fixing, going on to become major players in
resulting price structure.” decisions which are being appealed. the US cigarette market.
36
More recently, the tobacco dimmer and companies try to profit With respect to industry structure,
transnationals attempted to gain more from the existing pool of users economic theory suggests that
market share in newly opened in the short run. tobacco taxes will be fully passed on
markets (e.g. Thailand and Hungary This pattern is, for example, to users through an equivalent price
in the 1990s) by price discounting observed in western Europe where increase in a perfectly competitive
(Vateesatokit et al., 2000;; Szyilágyi competition in most markets is limited market with constant long-run costs
and Chapman, 2003). In addition, (see Table 3.1), giving the industry of production. At the other extreme,
there were numerous new entrants considerable pricing power and a profit-maximizing monopolist will
to the US market after the 1998 making these markets the industry’s pass on only part of the tax in price,
Master Settlement Agreement (MSA) most profitable globally (Spielman, with the pass-through greater as
in part because loopholes in the 2008a, 2008b;; Euromonitor demand is more inelastic.
agreement created opportunities International, 2009). The United In an oligopolistic market,
for non-participating companies Kingdom market, as outlined above, however, how much or little of the tax
to enter the market and undercut is one of the most interesting, and is passed through can vary based
prices. In most established markets, analyst reports suggest that the on how the industry reacts. Harris
however, marketing restrictions make industry has been able to increase (1987), for example, suggests that
it increasingly difficult for new market prices sufficiently to offset both companies in a highly concentrated
entrants to compete effectively with volume declines and negative mix tobacco product market can use
established brands, and reputational (downtrading to cheaper brands) a tobacco tax increase as an
and litigation risks provide further (Spielman, 2008a, 2008b). The fact opportunity for a coordinated price
barriers to market entry (Gilmore et that this may have been facilitated by increase that leads to an overshifting
al., 2010). tax policies is further outlined below. of the tax, with prices rising by more
A second factor explaining price Furthermore, by virtue of having than the amount of the tax. Using
levels below the short-run industry diverse brand portfolios in multiple data from the years around the
profit-maximizing level, and one price tiers, companies can utilize doubling of the US federal cigarette
perhaps more relevant currently in both strategies—keeping some excise tax (from US$0.08 to US$0.16
at least some declining markets, is brands cheap to “hook” new smokers per pack) on 1 January 1983, Harris
the trade-off between short-run and while maintaining prices on higher- observed that industry prices rose
long-run profitability. As Becker and end brands to offset the negative mix by more than manufacturing costs
his colleagues (1994) describe, the and maintain profits. It would appear in the months leading up to the tax
greater price sensitivity of young such practices may be occurring in increase, with the rise in industry
people (as discussed below in the United Kingdom, for example, prices starting after the tax increase
Chapter 6) and the addictive nature because despite a marked growth in was announced. In the end, he
of tobacco use may lead companies discount brands (Devlin et al., 2003, concludes that prices rose by more
with market power to set prices lower Talking Retail, 2006) the industry is than twice the amount of the tax
than the short-run profit-maximizing managing to grow its profits, and the increase.
level to “get more consumers ‘hooked’ United Kingdom market is among Harris’ (1987) hypothesis
on the addictive good” so that long- the most profitable (Spielman et al., appears to be confirmed by internal
run profits will be higher than they 2008a, 2008b). industry documents. When asked for
would be if prices were set higher his thoughts about how to pass on
in the short run and fewer young Tobacco tax increases and prices an anticipated additional increase
people took up tobacco use. To the in the US federal cigarette tax in
extent that tobacco product markets The effectiveness of tobacco tax 1987, Philip Morris economist Myron
are on the decline in some countries increases in reducing tobacco use Johnston recalled the industry’s
(e.g. in response to adoption and and its consequences depends, pricing strategies around the 1983
implementation of effective tobacco in part, on how increases in taxes increase (Johnston, 1987):
control strategies, including higher are passed on to users in price “Last time, of course, we
taxes), it becomes more likely that increases. The extent to which prices increased prices five times
industry prices will rise to nearer the rise following a tax increase depends between February of 1982
short-run industry profit-maximizing on multiple factors, most notably and January of 1983. In less
level as future prospects grow industry structure and tax structure. than a year, the price went
37
from $20.20 to $26.90 per paid and will be less likely to increases in South Africa (since
thousand ($2.70 more than suffer from ‘sticker shock.’ 1994) and Jamaica (in 2005), while
the tax), and this fact was As a result, they should be Delipalla and O’Donnell (2001) found
not lost on consumers, who less likely to use the price undershifting of cigarette taxes in
could legitimately blame increase as an incentive to 12 European Union countries in the
the manufacturers for the stop smoking or reduce their early 1990s, particularly in markets
price increases. While price consumption.” that were less concentrated.
increases of this magnitude Experiences with several federal Although there are no recent
might have been tolerated tax increases in the 1990s and empirical studies in the European
during the rapid escalation early 2000s show that wholesale Union, as noted above the market
in the overall inflation rate cigarette prices rose by the amount has seen considerable consolidation
between 1977 and 1981, of the tax increases or, at times, in recent years (Hedley, 2007), and
the increase in the price of being absorbed by the industry (e.g. other evidence suggests that the
cigarettes in 1982-83 was the US$0.05 increase in 2002). undershifting observed in the 1990s
made even more dramatic Most recently, however, the much (Delipalla and O’Donnell, 2001) has
by the fact that the overall larger (US$0.6167 per pack) 1 April reversed (Spielman and Loveless,
rate of inflation was slowing 2009 increase in the US federal 2008b). Although the countries in
considerably.” cigarette excise tax did lead to eastern Europe are in general at
Industry reactions, particularly some overshifting;; Philip Morris, for a different stage of the tobacco
by the dominant firm, however, can example, raised prices on its ‘growth epidemic with volumes of consumers
vary with circumstances. In the same and support brands’ (including and tobacco sold still increasing,
memo, Johnston goes on to note Marlboro) by US$0.71 per pack profits are also increasing as a
how the overshifting of the 1983 tax and on its ‘non-support brands’ by result of industry price increases
increase had a particularly negative US$0.78 per pack. Given its role as and consumers trading up to more
impact on Philip Morris because of the market leader in the USA over the expensive brands (Euromonitor
its greater impact on smoking among past few decades, other US cigarette International, 2009).
young people, where Marlboro’s companies have generally followed With regards to tax structure, both
market share was greatest. This led Philip Morris’ lead in responding to theoretical and empirical evidence
him to write (Johnston, 1987): the US tax increases. suggest that specific excise taxes
“I have been asked for my The limited empirical evidence tend to increase consumer prices
views as to how we should on this issue, almost entirely from relatively more than ad valorem
pass on the price increase high-income countries, reflects this excises, and hence lead to relatively
in the event of an increase in variability. Older US studies found higher reductions in consumption
the excise tax. My choice is to that cigarette taxes led to less-than- (Delipalla and Keen, 1992;; Delipalla
do what I suggested to Wally comparable price increases (Sumner and O’Donnell, 2001;; Chaloupka
McDowell in 1982: Pass on and Ward, 1981;; Bishop and Yoo, et al., 2010). The higher impact of
the increase in one fell swoop 1985;; Ashenfelter and Sullivan, specific excise taxes on consumer
and make it clear to smokers 1987);; others found that taxes prices is consistent with a greater
that the government is solely were fully passed on (Sumner and possibility of overshifting of specific
responsible for the price Wohlgenant, 1985);; and still others taxation relatively to ad valorem
increase, advertise to that found that prices rose by more than (Delipalla and Keen, 1992). Under
effect, suggest that people the tax (Barzel, 1976;; Johnson, 1978;; specific taxation, any increase
stock up to avoid the price Sumner, 1981). Most recent US- in producer price will go to the
increase, and recommend based analysis, however, find that producer as revenue and therefore
that they refrigerate their cigarette taxes are overshifted in the would increase producers incentive
cigarettes ‘to preserve their USA, leading to increases in retail to raise price. Economic theory
freshness.’... Then when cigarette prices greater than the tax also shows that industry profits
people exhaust their supply (Barnett et al., 1995;; Keeler et al., are relatively higher under specific
and go to the store to buy 1996;; Hanson and Sullivan, 2009;; taxation (Delipalla and Keen, 1992).
more, they will be less likely Sullivan, 2010). van Walbeek (2010) Moreover, a tax increase may lead
to remember what they last observes similar overshifting of tax to an increase in profits when tax
38
overshifting takes place: as a higher of marketing activities that lower the the cost of the items distributed, but
tax increases consumer price and price of or otherwise add value to subtracting any payments received
reduces demand, for profits to rise, their products. These activities are for the item. The costs associated
the after-tax mark up must rise. clearly defined by the US Federal with distributing non-cigarette items
All this suggests that relying Trade Commission (FTC) in its in connection with sampling or
heavily on specific excise taxation is annual reports to the US Congress retail-value-added programmes are
likely to increase market concentration on cigarette and smokeless tobacco reported in those categories, not as
and industry profits in the long run. company marketing activities specialty item distribution.
With respect to the public health (Federal Trade Commission, 2009a, 6SHFLDOW\LWHPGLVWULEXWLRQQRQ
perspective, while specific taxes offer 2009b);; as defined for cigarettes, branded: all costs of distributing any
greater public health benefits, there these include: item (other than cigarettes, items the
is a danger that the greater industry 3ULFHGLVFRXQWVSULFHGLVFRXQWV sole function of which is to advertise
profits that emerge may in turn paid to cigarette retailers or or promote cigarettes, or written or
be used to damage public health, wholesalers to reduce the price of electronic publications), whether
because these profits can be used cigarettes to consumers, including distributed by sale, redemption of
to increase industry marketing, or off-invoice discounts, buy-downs, coupons, or otherwise, that does not
fund lobbying or litigation efforts to voluntary price reductions and trade bear the name, logo, or an image
challenge public health policies. For programs, but excluding retail-value- of any portion of the package of
this reason, along with increases in added expenditures for promotions any brand or variety of cigarettes,
the specific tax (and its associated involving free cigarettes and including the cost of the items
great effect on price), governments expenditures involving coupons. distributed, but subtracting any
may find they need to implement 6DPSOLQJ VDPSOLQJ RI payments received for the item. The
other policies to counteract the cigarettes, including the cost of the costs associated with distributing
tobacco industry’s increased market cigarettes, all associated excise non-cigarette items in connection
power. taxes, and increased costs under the with sampling or retail-value-added
As described in the previous Master Settlement Agreement, and programmes are reported in those
chapter, this hypothesis is supported the cost of organizing, promoting, categories, not as specialty item
by descriptive data on taxes and and conducting sampling. Sampling distribution
prices globally, with prices higher includes the distribution of cigarettes 5HWDLOYDOXHDGGHG ERQXV
on average in countries that rely on for consumer testing or evaluation cigarettes: Retail-value-added
specific taxes than in those relying when consumers are able to smoke expenditures for promotions
on ad valorem taxes (World Health the cigarettes outside of a facility involving free cigarettes (e.g. buy
Organization, 2010). Chaloupka and operated by the Co., but not the cost two packs, get one free), whether or
colleagues (2010), in their empirical of actual clinical testing or market not the free cigarettes are physically
analysis using data from EU countries research associated with such bundled together with the purchased
over the period from 1998 through cigarette distributions. Sampling also cigarettes, including all expenditures
2007, provide additional support for includes the distribution of coupons and costs associated with the value
this hypothesis. They conclude that for free cigarettes, when no purchase added to the purchase of cigarettes
average cigarette prices are higher or payment is required to obtain the (e.g. excise taxes paid for the free
in EU countries relying more heavily coupons or cigarettes. cigarettes and increased costs under
on specific cigarette excises than on 6SHFLDOW\ LWHP GLVWULEXWLRQ the Master Settlement Agreement).
ad valorem excises. branded: all costs of distributing any 5HWDLOYDOXHDGGHG QRQ
item (other than cigarettes, items the cigarette bonus: Retail-value-added
Tobacco industry price-related sole function of which is to advertise expenditures for promotions involving
marketing efforts or promote cigarettes, or written or free non-cigarette items (e.g. buy
electronic publications), whether two packs, get a cigarette lighter),
Background distributed by sale, redemption of including all expenditures and costs
coupons, or otherwise, that bears associated with the value added to
In addition to directly setting prices the name, logo, or an image of any the purchase of cigarettes.
to tobacco products, tobacco portion of the package of any brand &RXSRQV $OO FRVWV DVVRFLDWHG
companies can engage in a variety or variety of cigarettes, including with coupons for the reduction of
39
the retail cost of cigarettes, whether (an issue explored in a later section). declining to about US$12.5 billion in
redeemed at the point-of-sale or by Various of the marketing techniques 2006. This rise in marketing spending
mail, including all costs associated outlined above may therefore be is even more dramatic when one
with advertising or promotion, used throughout much of the world. considers that US cigarette sales
design, printing, distribution, and For example, the United Kingdom during this period fell by over 43%.
redemption. However, when coupons Tobacco Advertising and Promotions Second, price-related marketing
are distributed for free cigarettes and Act (which was fully implemented has gone from accounting for a minor
no purchase or payment is required to in 2005) bans, inter alia, the use share of overall cigarette company
obtain the coupons or the cigarettes, of voucher schemes and direct marketing expenditures in the 1970s
these activities are considered to be consumer communication. However, to accounting for most of marketing
sampling and not couponing. recent evidence indicates both that spending in the 1990s and 2000s.
The Federal Trade Commission tobacco companies still provide In 2006, for example, spending on
(2009b) defines similar marketing display cabinets and incentives for marketing activities that directly
activities for smokeless tobacco displaying their products (Rooke et reduced cigarette prices (price
products in the USA;; the reporting al., 2010), and that the use of price- discounts, coupons, retail value
of these marketing expenditures based marketing has increased added promotions involving free
is required by federal legislation in following full implementation of this cigarettes, and distribution of free
the USA. Canadian legislation also legislation (Moodie and Hastings, samples) amounted to more than 85%
requires tobacco companies to report 2009). Although price-based, this of total cigarette company marketing
to the federal government on their marketing is not necessarily price- expenditures, as compared to an
promotional expenditures (http:// reducing and includes, for example, estimated less than 19% in 1975. In
laws - lois.justic e.gc.c a /en /SOR- price-marked packaging for discount contrast, spending on promotions
2000-273/FullText.html);; however, brands visible from the display involving merchandise (non-cigarette
such mandated reporting is rare and cabinets (Moodie and Hastings, retail value added promotions and
the detailed information contained is 2009). specialty item distribution) rose from
rarely made publicly available. Unfortunately, with the exception a minimal share of total spending
These and other marketing of the USA, detailed data on how in the 1970s to a peak of about 18
practices are prohibited in some extensively the industry uses these percent in 1994, before declining to
countries. Globally, WHO Report on price-related marketing strategies less than 2% in recent years (largely
the Global Tobacco Epidemic (World is virtually non-existent. However, as a result of the limits on this type
Health Organization, 2009) data trends in the use of price-related of marketing included in the Master
indicate that 79 countries, including marketing strategies in the USA are Settlement Agreement of 1998).
many LMICs, have implemented some informative given that these illustrate
form of restriction on price-based how tobacco companies have Price-related marketing and
marketing. For example, a 2003 EU adapted their marketing strategies tobacco use among young people
Council Recommendation (2003/54/ over time in response to new evidence
EC) recommended that Member on the impact of price on tobacco use In his testimony on price-related
States adopt various tobacco control and to implementation of restrictions marketing as part of the US tobacco
measures including those to prohibit on other marketing activities. litigation (USA v. Philip Morris et al.),
“the use and communication of sales Figure 3.1 illustrates trends Chaloupka (2004) described how the
promotion, such as a discount, a free in cigarette company marketing shift in tobacco company marketing
gift, a premium or an opportunity to expenditures in the USA from strategies towards increased use
participate in a promotional contest 1975 through 2006, as well as of price-reducing and other price-
or game.” A subsequent evaluation changes in the composition of these related promotions accelerated
(SEC(2009)1621) suggested that most expenditures over time. Two striking following the publication of early
Member States had implemented the trends are apparent from this figure. economic research on the influence
recommended measures but failed to First, overall cigarette company of price on smoking behaviour,
specify the proportion implementing. marketing expenditures in the USA particularly among young people
However, few countries have have risen dramatically, from just over (Lewit et al., 1981;; Lewitt and Coate,
comprehensive enough bans to cover US$491 million in 1975 to a peak of 1982). In their review of internal
all forms of price-based promotions over US$15.1 billion in 2003, before industry documents (now available
40
online at http://legacy.library.ucsf. Reynolds, wrote about the short and from teenagers. But, the loss
edu/), Chaloupka and colleagues long-run impact of cigarette price of younger adult males and
(2002) describe how the two studies increases based on the findings from teenagers is more important
by Lewit and colleagues generated the Lewit and colleagues studies to the long term, drying up
considerable attention among (Burrows, 1982a): the supply of new smokers to
leading US cigarette companies. “In terms of the immediate replace the old. This is not a
Of most interest were the findings impact, the effect of price fixed loss to the industry: its
on the greater price sensitivity of on males 35+ is the most importance increases with
smoking among young people, important. Half (50%) of the time. In ten years, increased
given the importance of youth and total drop in industry volume rate per day would have been
young adults to future profitability. is attributable to males 35+, expected to raise this group’s
For example, in a 1982 memo, Diane compared to 24% from consumption by more than
Burrows, a market researcher at R.J. younger adult males and 7% 50%.”
In related memos and reports,
Figure 3.1. Cigarette company marketing expenditures, by category, 1975-2006 Burrows (1982b, 1984) discussed the
marketing implications of the greater
price sensitivity and importance
of imagery to young people. For
example, in a 1984 Strategic
Research Report, Burrows wrote:
“Tactically, extended periods
of closely targeted pack
promotions (B1G1F (buy-
one-get-one-free), sampling)
in selected sites (e.g.,
convenience stores, military
exchanges, special events)
could lead to brand loyalty from
repeated trials. This should
be considered an investment
program.” (Burrows, 1984).
As Chaloupka and colleagues
(2002) describe, the combination of
price-reducing promotions with more
traditional advertising campaigns
Sources: Federal Trade Commission, 2009a;; Working Group’s calculations. played a key role in the growth of R.J.
Notes to Figure 3.1:
Reynolds’ Camel brand among young
$GYHUWLVLQJ LQFOXGHV QHZVSDSHUV DQG PDJD]LQHV RXWGRRU DQG WUDQVLW DQG SRLQWRIVDOH UHSRUWHG VHSDUDWHO\
beginning in 1975). people (most notably young males) in
3XEOLF(QWHUWDLQPHQWLQFOXGHVJHQHUDODXGLHQFHDQGDGXOWRQO\SXEOLFHQWHUWDLQPHQWUHSRUWHGLQDVLQJOHFDWHJRU\LQ the late 1980s and early 1990s. For
earlier years and reported separately beginning in 2002. example, a 1986 memo reported on
3ODFHPHQWLQFOXGHVSURPRWLRQDODOORZDQFHVSDLGWRUHWDLOHUVZKROHVDOHUVDQGRWKHUVUHSRUWHGVHSDUDWHO\EHJLQQLQJ the effectiveness of these efforts,
in 2002 and estimated for earlier years based on the percentage of the combined promotional allowances category
accounted for by these categories in 2002. which included ‘six pack’ promotions
3ULFH LQFOXGHV SULFH GLVFRXQWV DQG UHWDLOYDOXHDGGHG ERQXV FLJDUHWWHV UHSRUWHG VHSDUDWHO\ EHJLQQLQJ LQ (buy-three-get-three-free), the
coupons (reported separately beginning in 1997);; and sampling distribution. Estimates for earlier years are based on “Camel Cash” program, and other
shares in the previously aggregated categories that included these in the first years data are reported for disaggregated
categories. retail value added promotions:
0HUFKDQGLVH LQFOXGHV EUDQGHG DQG QRQEUDQGHG VSHFLDOW\ LWHP GLVWULEXWLRQ UHSRUWHG DV D VLQJOH FDWHJRU\ EHIRUH “The major factor contributing
2002 and separately beginning in 2002);; and retail-value-added non-cigarette bonus (reported separately in 2002 and to CAMEL’s dramatic growth
estimated for earlier years based on the share of combined retail-value added as reported in 2002).
among Mid-West 18–24 year
2WKHULQFOXGHVDOORWKHUFDWHJRULHVUHSRUWHGE\)7&LQFOXGLQJGLUHFWPDLOWHOHSKRQH,QWHUQHWFRPSDQ\ZHEVLWHVDQG
other), and other;; in earlier years, a portion of the F TC-reported other and direct mail expenditures are allocated to other old males appears to be the
categories (e.g. coupons and retail value added) based on shares of expenditures in the first year that expenditures in increased level of Mid-West
more disaggregated categories are reported.
promotional support, and in
41
particular, CAMEL’s targeted to at least partly offset the impact more is spent on comprehensive
promotions (which were of tax increases and other tobacco tobacco control programmes than
implemented the same time control efforts. Based on their review in states spending less (Slater et al.,
as the boost in CAMEL’s share of internal documents, Chaloupka 2001;; Loomis et al., 2006). Finally,
and completed just prior to the and colleagues (2002) described how Keeler and colleagues (1996) found
downward trend).” (Creighton, companies increased distribution that cigarette companies price-
1986). of coupons and use of multipack discriminated by setting lower prices
The changes over time in the promotions to soften the impact of in states with stronger state and local
composition of cigarette marketing state and federal tax increases, with tobacco control policies.
expenditures (shown in Figure 3.1) some of the print ads used to distribute
suggest that the importance of price- coupons referring to them as “tax Tobacco industry tax-related
related marketing efforts described relief.” Similar efforts were observed lobbying efforts
in these selected documents was after the April 2009 federal tobacco
recognized throughout the industry. tax increases, with Philip Morris, for Background
example, contacting smokers to let
Tobacco company price them know that they were increasing In addition to the systematic
and price-related marketing prices because of the tax increase but searches for relevant literature, we
and tobacco control that they could “register at Marlboro. also took a systematic approach to
com to become eligible for cigarette reviewing studies included in this
Feighery et al. (2004) examined coupons and special offers.” section. Hence, we tried to identify
the extent and impact of retailer Similarly, Ruel and colleagues all empirical studies (qualitative and
participation in industry incentive (2004), using observationally collected quantitative) focusing significantly on
programmes in 468 stores in 15 US data on cigarette marketing, and tobacco industry tax-related lobbying
states. They found that cigarette Loomis and colleagues (2006), using activities, and then applied inclusion/
companies engaged approximately scanner-based data, documented exclusion criteria to potentially
two thirds of tobacco retailers in an the increased prevalence of various relevant studies (see Figure 3.2).
incentive program. The study shows price-reducing promotions following Of the 2638 abstracts/titles located
that stores that participate in cigarette the significant price increases for this chapter overall, 2443 were
company incentive programmes resulting from the pass-through of excluded from this section on the basis
may have cheaper cigarette prices costs associated with the MSA and of the title and abstract alone. The
(and more prominent placement the subsequent implementation of the remaining 195 papers were retrieved
of cigarettes and advertising) than marketing restrictions included in the for full paper analysis to assess their
do stores that do not participate. agreement. Similarly, researchers relevance to this section, at which
Although participation in a Philip have used observationally collected stage 164 further studies were
Morris incentive programme was and scanner-based data to conclude excluded. Full analysis was conducted
not significantly related to the price that price-reducing promotions are on the 31 remaining studies that met
of (Philip Morris brand) Marlboro more prevalent in states where the full inclusion criteria.
cigarettes, the price of Lorillard’s
Newport cigarettes was significantly Figure 3.2. Inclusion and exclusion criteria
lower in stores participating in a
Lorillard incentive programme. The To be included in this section, studies had to:
study was however somewhat limited
- Significantly focus on tobacco industry lobbying activities that related to taxation
in its ability to detect links between (studies were included that did not primarily focus on this issue, but they had to focus on
incentive programmes and price it significantly and meet the other inclusion criteria;; studies in which tax-related lobbying
activities were only a minor/tangential issue were excluded);;
reductions because the questions
were limited to participation in - Draw on empirical evidence to support any claims made about tobacco industry tax
related lobbying;;
company- rather than brand-specific
incentive programmes. - Be written in English or another language spoken by a member of the IARC review team;;
Limited research from the USA - Be concerned with lobbying from 1985 onwards (studies which were only concerned with
documents tobacco companies’ use a period prior to 1985 were excluded;; studies which concerned periods both before and
after this date were included).
of price and price-related marketing
42
The data from these studies were Figure 3.3. Critical appraisal criteria
extracted and the studies were
critically appraised using critical These criteria were used to assess all of the included studies. The results of this process
appraisal criteria for qualitative are presented under the critical appraisal section of the results in Table 3.3.
papers, adapted from Rees et al. 1) How clear is/are the research question(s) and/or aim(s)?
(2006) and the United Kingdom 2) Was the methodology appropriate for addressing the stated aims of the study?
Public Health Resource Unit (2006)
3) Where applicable, was the recruitment/search strategy appropriate and/or was an
(see Figure 3.3). The appraisal and adequate sample obtained to support the claims being made (i.e. was the data collection
inclusion/exclusion criteria, were adequate and appropriate)?
piloted by two reviewers on five 4) Were the methods of data analysis appropriate to the subject matter?
relevant papers and then revised 5) Is the description of the findings provided in enough detail and depth to allow
accordingly. interpretation of the meanings and context of what is being studied? [Are data presented
to support interpretations, etc?]
As is common practice with
reviews of qualitative data (Smith et 6) Are the conclusions justified by the results?
al., 2009), the critical appraisal criteria 7) If applicable, are the theoretical developments justified by the results?
were used for descriptive purposes 8) Have the limitations of the study and their impact on the findings been considered?
only and to highlight variations in the 9) Do researchers discuss whether or how the f indings c an be transferred to other c ontexts
quality of studies (i.e. no quality score or consider other ways in which the research may be used?
was calculated but this assessment 10) If the answer to 9 is ‘yes’, do you agree these suggestions are appropriate, based on
the research?
allows us to comment on the quality
of the included studies). Narrative Source : Rees et al. ;; 2006 ;; United Kingdom Public Health Resource Unit
synthesis was performed to combine
the qualitative and quantitative Glantz, 1999;; Goldman and Glantz, se, as opposed to being concerned
evidence. Results are tabulated (see 1999;; Balbach et al., 2000;; Givel and about significant tax increases which
Table 3.3) as well as summarized in Glantz, 2001;; Morley et al., 2002;; would be used to fund tobacco control
the following text. Yerger and Malone, 2002;; Szilágyi activities. Nevertheless, we can
and Chapman, 2003;; Alamar and distinguish arguments and strategies
Results Glantz, 2004;; Levenstein et al., 2005;; used to counter the increases from
Balbach et al., 2006;; Givel, 2006, those use to counter/undermine the
The majority of 31 studies identified 2007;; Gilmore et al., 2007;; Nakkash, proposed earmarking.
focused on North America (22 2007;; Campbell and Balbach, 2008, The other main topic (covered in
focused on the USA, or states 2009;; Balbach and Campbell, 2009;; seven studies) concerned tobacco
within the USA, and two focused on Lum et al., 2009;; Raebeck et al., industry efforts to influence excise
Canada), although it should be noted 2009) or, in some cases, to lower structures (O’Sullivan and Chapman,
that several of the studies concerning existing taxes (Joossens and Raw, 2000;; Szilágyi and Chapman, 2003;;
state-level excise propositions related 2003;; Breton et al., 2006;; Kelton Gilmore and McKee, 2004b;; Gilmore
to the same policy development (Table and Givel, 2008). Of these, only four et al., 2005;; Delnevo et al., 2007;;
3.4). The rest covered a diverse set of covered policy influence outside North Gilmore et al., 2007;; Nakkash, 2007),
countries (Table 3.5). We found no America (Joossens and Raw, 2003;; and these have a very different
studies relating to tobacco industry Szilágyi and Chapman, 2003;; Gilmore geographic profile (Table 3.5). As
tax-related lobbying in South America, et al., 2007;; Nakkash, 2007). As Table there are far fewer studies on this
South Asia, Australasia or Africa or 3.3 summarizes, most of the US issue, and as this is not the main focus
examining influence on supranational studies focused on policy proposals of most of these studies, the picture
(e.g. European Union) tax policy. that included plans to substantially that emerges is far less clear than
The studies covered two main increase taxes and to earmark that relating to industry activities to
topics. The majority (27) discussed (hypothecate) all or most of the influence excise rates. Nevertheless,
tobacco industry efforts to counter revenue raised by these tax increases between them, these studies provide
proposed excise tax increases (Begay for tobacco control programmes. some important insights into how
et al., 1993;; Moon et al., 1993;; Koh, This means it is difficult to ascertain different companies have sought to
1996;; Traynor and Glantz, 1996;; to what extent the industry was influence tobacco excise structures
Heiser and Begay, 1997;; Bialous and concerned about tax increases per to their advantage.
43
44
Table 3.3. Summary of studies concerning tobacco industry (TI) tax-related lobbying. Closely related studies are assessed together.
Numbers in the column “CA criteria met” represent the quality criteria that each study met;; see* at the end of the table for further details.
Study Methods CA criteria met* Key findings
Aim of TI Arguments used Mechanisms employed Success of TI
Studies concerned with tobacco industry efforts to influence tobacco tax rates (and, where applicable, associated plans for earmarking of taxes)
Begay et al., (1993) - Analysis of reports 1-3,6,9, 10 To prevent a significant Not specified in this article - Found the TI was - Proposition 99 passed
by the California Fair tax increase being investing heavily in the in 1988, so the TI failed
Political Practices proposed in the McCain Californian legislature, to prevent it, but did
Commission, statements bill spending political money experience success in
filed with the California more intensively there diverting funds.
secretary of state’s than in the US Congress. - Claims “Although there
Political Reform Division, - Powerful individuals in is little evidence showing
and data on Proposition the legislature seem to that the tobacco industry
99 expenditures from have been particularly has «bought» legislators’
various official budget targeted, e.g. “In 1991- votes, early research
documents. 1992, the Speaker strongly suggests that
- Also conducted received $221 367, tobacco industry’s
“a multivariate making him the largest campaign contributions
simultaneous equations single legislative recipient are influencing the
statistical analysis of tobacco industry behaviour of California
of data on campaign contributions in the United legislators in matters
contributions from States” (explains the related to tobacco policy-
the tobacco industry Speaker of the Assembly, making, independent
to legislators in the the single most powerful of constituents’ support
1991-1992 election, member of the legislature for tobacco control.”
their records on tobacco and second only to the Notes the multivariate
control policy as scored governor in influence). simultaneous equations
by tobacco control - TI lobbying expenditures statistical analysis of data
advocates, members’ grew 10-fold from 1985– on campaign contributions
personal characteristics, 86 to 1987–88, when from the tobacco industry
and constituents’ Proposition 99 passed. to legislators indicated
attitudes towards - The TI also reorganized that the TI had “a
tobacco control.” its lobbying efforts, hiring statistically detectable
large, private firms to effect on legislative
lobby legislators and state behaviour”.
officials since Proposition - “Only 14.7% of revenues
99 passed, rather than were spent on tobacco
relying on organizations education and prevention,
like the Tobacco Institute, not the mandated 20%.
as they previously had. The underfunding
Some of the firms hired amounts to $174.7 million
also worked for medical redirected to medical
groups which TI wanted care programs from fiscal
to side with in efforts to years 1989-1990 through
divert revenue. 1993-1994, despite clear
- “In 1989, the Tobacco language in Proposition
Institute offered to 99 specifying how the
Begay et al., (1993) contribute $250 000 to money should be spent.”
(contd) the California Medical
Association to divert
Health Education Account
funds to medical care
programmes through a
Medical Association-
sponsored ballot initiative.”
Moon et al., (1993) Methodology is not 1,9,10 To prevent a proposal - The tax was framed as - Opponents to proposal On November 6, 1990,
described but the article for a state-wide tobacco «a trap set by ‘special ‘operated under the name Montana voters defeated
appears to be based on tax initiative in Montana, interests’ to subsidize «Citizens Against More Initiative 115 by 59% to
an analysis of various USA, which would have those people who live in Tax and Bureaucracy»’ 41%, so the TI succeeded.
data sources, including added an extra $0.25 to cities”. and they ‘had $1.47 million
media coverage of cigarettes, with money - TI highlighted that at their disposal.’ ‘[M]ore
the issue and other to be dedicated to TC passage of the initiative than 88% of the money
academic articles programmes would give Montana the came from Philip Morris,
highest cigarette tax in the R. J. Reynolds, Brown
nation. & Williamson, and the
- Undertook Tobacco Institute.’
advertisements during the - While the initiative
last week of the campaign, was still in draft form,
which built on above opponents conducted
(informing voters that numerous telephone
although they could not do tracking surveys to
anything about outrageous measure the possible
property taxes, they could impact of various
vote against the "selective campaign themes.
sales tax increase") Increased state taxes and
bureaucracy emerged
as pivotal issues, and
extensive television,
radio, and newspaper
advertisements
emphasized these
concerns.’
- Linked the proposed
increase tobacco tax with
proposed property tax
increases in rural areas
that were happening at the
same time (so framed as
part of broader debates
about tax rises);;
- Brochures were sent
to businesses asking
whether the business
could afford the increase.
45
46
Moon et al., (1993) - Multiple mailings were

(contd) sent to voters from a
Helena tobacco retailer
portraying the increased
tax as «a trap set by
‘special interests’ to
subsidize those people

who live in cities».’
“The Montana Alliance
for Progressive Policy
(MAPP), a liberal group
opposing the increase
as an unfair tax on poor
people, received $22,500
from a tobacco industry
lobby group to produce
and send brochures to its
50,000 members urging
rejection”.
Koh (1996) Methodology is not 1,9,10 To prevent the passage - The tax was unfair - The TI outspent the - TI failed to disrupt the
described but the article of Question 1, a proposal and regressive, Coalition for a Healthy signature gathering (two
appears to be based on to raise the tobacco tax in disproportionately Future by 10:1. rounds). In round one, the
an analysis of various Massachusetts by $0.25 affecting people with low - TI put forward an sheer volume put TI off
data sources, including to fund tobacco education incomes alternative, compromise questioning signatures.
media coverage of programmes - On that basis, it would proposal. - TI alternative proposal
the issue, personal contribute to class - TI legally challenged ignored.
correspondence with warfare (argued the tax earmarking of funds. - TI legal challenge on
some key players and pitted upper middle class - TI organised and funded earmarking failed.
other academic articles liberals – mostly white – a group to lobby against - The proposition passed
against lower middle class the tax increases, called so TI failed to prevent it.
working people – mostly ‘The Committee Against - Overall, state-wide
minority). Unfair Taxes’ tobacco consumption
- Warned that tax- - An ultra-conservative, subsequently declined by
generated monies would out of state Catholic more than three times the
be diverted from their organisation working national average.
intended purposes, as had with the EU to distribute - However, TI political
occurred in California. thousands of letters pressure led to governor
- Argued the tax would claiming that the Q1 funds and legislature diverting
harm small business in the would be used to fund more than $220 million out
state by motivating people abortion counselling and of the tobacco education
to buy cigarettes outside condom distribution in account, prompting a
the state. schools. Unclear if/how TI subsequent round of
connected. lawsuits.
Koh (1996) (contd) - Claimed the tax was - TI stepped up its media
unfair on smokers campaign in final weeks
before vote.
- TI focused on tax as
a generic category,
rather than specifically
discussing tobacco taxes.
- Employed a key former
legislator
Traynor and Glantz, Analysis of published 1,2,5-7,9,10 - To defeat Proposition - Tried to move the issue - TI invested significant - TI’s legal challenge
(1996) reports, public 99 in California (which away from smoking by resources into lobbying failed.
documents, personal proposed to add $0.25 focusing on generic Californian legislature. - But polls indicated the
correspondence, to state cigarette tax and increased taxation Worked with healthcare TI’s ads did have an
internal memos, polling use 20% of the revenue arguments. organisations to ensure effect, reducing public
data, and interviews on tobacco education and - Emphasised government funds were diverted away support for the initiative.
with representatives prevention programmes);; mismanagement of tax from TC programmes to - However, various factors
form organizations - Once Proposition 99 money medical care programmes. undermined one of TI’s
that participated in had based, the aim was - Suggested $0.25 was - Hired a political most popular arguments,
the Proposition 99 to divert earmarked funds too much of an increase consulting firm and formed that concerning smuggling
campaign. away from tobacco control on a single product. the Californians Against and police work, including
activities/programmes. - Claimed the increased Unfair Tax Increases an official report that
tax would lead to (CAUTI). concluded the effect on
increased smuggling and - CAUTI circulated its own smuggling was likely
small-scale smuggling. petition in 1988, as the to be negligible, which
- Built on this to claim pro Proposition 99 effort criticized CAUTI’s ads.
police attention would be was underway to gather Police groups which had
diverted away from other supportive signatures, supported the TI position
business and that more called ‘The Tobacco Tax subsequently dropped
money would be spent on Ripoff’. This was not their support. In addition,
guns and drugs. official and was merely it was revealed that the
- Argued tobacco tax used as a way of reducing ‘undercover policeman’
was just another way for the pool of available paid in the TI’s most effective
special interests, such as signature-gatherers (by ad had nothing more than
physicians, to become paying them more) and a desk job for the policy
richer. to confuse voters about and was also a part-time
- Framed tobacco tax as a Prop 99. actor. All of this largely
regressive tax that would - CAUTI held a press destroyed the small-
negatively affect blue- conference to present scale smuggling/crime
collar workers. its argument that the arguments the TI was
- Argued Proposition proposed tobacco tax using.
99 violated California’s was just another way for - In the end, Proposition
1978 property tax-cutting special interests, such as 99 passed in 1988, so the
initiative and other tax physicians, to become TI failed to prevent it.
rules. richer.
47
48
Traynor and Glantz, - TI, through CAUTI, - However, the TI was
(1996) (contd) saturated TV, radio and influential in helping
print media with paid ads ensure the subsequent
to promote TI position. underfunding of tobacco
Their most effective education and prevention
ad featured an alleged program.
undercover police officer

discussing the crime
implications arguments
(more time would be spent
on smuggled cigarettes,
hence less time on other
issues, and the increased
criminal money from
smuggling, etc would be
spent on drugs and guns).
- TI legally challenged
Proposition 99 on
the basis it violated
California’s 1978 property
tax-cutting initiative and
other tax laws.
- CAUTI secured the
endorsements of the
Californian Sheriffs’
Association and the
California Peace Officers’
Association, which
gave credibility to their
crime/smuggling related
arguments.
Heiser and Begay, Analysis of internal 1-3,5,9,10 To challenge/undermine - Cigarette tax is - Legally challenged - Legal challenges failed.
(1997) memos, meeting the development, then regressive;; the draft proposal, on - Signatures expert was
minutes, newspaper implementation of, - The proposed tax would basis it violated the unable to disqualify
articles, other internal Question 1, a proposal to hurt the state economy state constitution, and enough signatures to
documents, letters, raise the tobacco tax in because it would increase filed subsequent legal challenge the ballot.
newsletters, news and Massachusetts by $0.25 cross-border shopping;; challenges at later stages. - TI ad campaign was
press releases (many of to fund tobacco education The legislature would use - Used a signatures effective, however, in
which were provided programmes, in order to the extra money raised for expert to try to disqualify shifting public opinion,
reduce smoking related purposes other than those signatures supporting the according to poll data.
deaths in the state intended (e.g. to balance ballot proposal;;
budget);;
Heiser and Begay, through the American - The proposal was not - During the second - Q1 nevertheless passed
(1997) (contd) Cancer Society). Also legal required signature drive, so TI failed to prevent it.
used available polling the TI tried to disrupt However, funds were
data and undertook matters by offering two subsequently diverted,
some interviews compromises – both of with only just over 25%
(although these are which involved smaller tax being spent on tobacco
not discussed in the increase, with fewer funds education, apparently
Methods section) being allocated to tobacco justifying one of the TI’s
control programmes;; arguments (article does
- Funded a Committee not make any specific
Against Unfair Taxes to claims about TI’s influence
oppose the initiative;; over this diversion).
- Undertook a major
advertising initiative (TV,
radio and direct mail) to
garner public support for
opposing initiative
Bialous and Glantz, Case study using data 1-3,5-7,9,10 To prevent Proposition - Framed initiative as an - Used front-groups - - The Proposition passed
(1999) from semi-structured 200 in Arizona and then attempt by proponents to “Enough is Enough” and so TI failed to prevent it.
interviews “with key divert funds away from divert large amounts of “No More Taxes”, which - The TI also failed to
players in the initiative” TC programmes (NB taxpayer money to their were 99.96% financed divert funds away from TC
and written records the Proposition aimed own benefit. by Philip Morris and programmes.
(documentary data to increase the cigarette - Used the diversion of the Tobacco Institute, - However, the tobacco
gathered from written excise tax by US$0.40, health education funds respectively, to campaign education programme
records and newspaper with proportional into medical services by against the initiative. experienced a range of
clippings). increases in the tax on the California legislature - Increased lobbying of the other problems and was
other tobacco products. as an example of how the state legislature, with the never fully implemented as
Increased revenues tobacco tax funds were number of paid tobacco planned. Authors suggest
were earmarked: 23% going to benefit only the industry lobbyists rising TI lobbying played a role
for tobacco prevention/ hospitals.’ from approximately four in this
education programs, - Claimed that California’s to 18.
5% for tobacco-related tobacco control - Once Proposition had
diseases and prevention programme was misusing been passed, the TI
research, 70% to provide public funds legally challenged some of
medical care for the the contracts agreed with
poor, and 2% to offset health education funds.
future loss of tobacco - Threatened local level
tax revenues by the boards with legal action
Arizona Department of over tobacco control
Corrections) programs.
- Tried to pass pre-
emptive legislation to limit
local TC programs
49
50
Goldman and Glantz, Interviews with key 1-3,5-7,9,10 - To prevent Measure 44 - Taxing smokers to - Legally challenged - The legal challenge
(1999) informants and analysis becoming law, notably pay for healthcare for the initiative on various initially met with some
of public documents, by preventing it from everyone is unfair. grounds (language used;; success but was
internal memoranda, and gaining a majority public - Measure 44 contained failure to specify all eventually ignored as
news reports vote. Ballot Measure 44 no controls on how money the types of taxes that submitted late.
was a proposal to raise would be spent and were would be raised;; and a - Other TI lobbying efforts
cigarette taxes in Oregon likely to be wasted on misleading summary). also initially met with some
by US$0.30 per pack bureaucracy, etc, rather - "While signatures were success but later failed.
and to use the funds to than being spent on health being collected to qualify - Although TI outspent
support an Oregon Health care and health education;; Measure 44 for the ballot, Measure 44’s supporters
Plan and tobacco control - Healthcare providers the tobacco industry 7 to 1, the initiative passed
(it proposed a dedicated and insurance companies began contributing with 56% of the vote.
10% of the revenues were being greedy in money to a different ballot - TI also failed to divert
would be used for tobacco wanting these taxes;; measure campaign”, tobacco control funds
control). - The Measure would which the authors suggest to other uses or limit the
- Once Measure 44 had allow cuts in spending was designed “to disrupt scope of the program.
passed, to prevent the on programmes that the signature-gathering - Nevertheless, only
funds from being spent were currently funded by process and to dilute 10% of the revenues
on tobacco control tobacco tax [NB authors support for Measure 44” were devoted to tobacco
programmes say this was incorrect] and “to divert the energy control activities, even
and financial resources though public health
of the health insurers groups provided 37%
away from Measure 44.” of the campaign money
The money the industry (notes that this was partly
gave to this alternative
because public health
campaign, meant the
groups were not involved
signature-gatherers could
in the early phases of
be paid significantly more
the tobacco tax effort
than those for Measure
and therefore missed an
44, which enabled more
opportunity to affect the
signature-gatherers
allocation of funds).
to focus on the other
Overall, article concludes:
campaign and the tobacco
‘Despite being outspent
control lobby, eventually,
more than 7 to 1, Measure
to have to pay more for
44 passed with 56% of the
signatures.
vote on 5 November 1996.
- TI formed and funded
‘The end result was an
a registered campaign
committee, called 11% decline in per capita
Fairness Matters to cigarette consumption in
Oregonians Committee Oregon since 1996 and
(FMOC). Of the US$4 614 a decline of 35 000 in the
292 in cash contributed to number of Oregonians
FMOC to fight for Measure who smoke’.
Goldman and Glantz, 39 and against Measure - Philip Morris concluded

(1999) (contd) 44, all but US$30 came they ought to use third
from the Tobacco Institute. parties more heavily in
- Much of the above future campaigns in order
money was used to run to make voters ‘more
ads on radio, TV and via persuadable’ of their views
direct mail (TI significantly
outspent the public
health campaigners on
advertising).
- TI hid its involvement
in some advertising
campaigns (though
advertisers were
eventually required to
declare this).
- TI hired a former Oregon
governor.
- Used the support of a
dentist (i.e. medical care
worker).
- Once measure had
been passed, TI lobbied
legislators about how
the money should be
spent, hoping to weaken
the anti-tobacco ad
programme. This included
lobbying a Subcommittee
which consisted of three
senators and which had
to (by majority) had to
approve the budget
Balbach et al., (2000) Analysis of relevant 2-7, 9,10 - To defeat Proposition - Not made clear in this - Direct lobbying of Proposition 99 passed in
documents (published 99 in California (which article, which merely legislature (investing huge 1988, so the TI failed to
reports, public proposed to add US$0.25 notes that the TI built on financial resources). prevent it.
documents, personal to state cigarette tax and anti-tax, anti-regulation - Worked with private - However, the tobacco
correspondence, use 20% of the revenue and freedom-of-choice healthcare and groups education and prevention
newspaper accounts, on tobacco education and themes representing medics programmes which had
press releases, internal prevention programmes);; to ensure funds were been earmarked for
memos and some - Once Proposition 99 diverted away from receiving 20% of the
industry documents). had based, the aim was tobacco control education revenue did not receive
Files made available by to divert earmarked funds programmes and towards their full allocation, which
various public health / away from tobacco control medical care programmes was instead diverted
tobacco control groups, activities/programmes (which went against the (largely to healthcare
and lawsuits in 1994 and specifications of the public organizations).
51
52
Balbach et al., (2000) 1995. Interviews were ballot). This provided - Following lawsuits by
(contd) also undertaken with ‘crucial political coverage public health groups, and
38 “key participants” for the tobacco industry extensive media coverage
(although TI refused to and the politicians that of the issue, the health
be interviewed) supported it’ education account finally
- Helped establish the received its full allocation
Californians for Smoker’s in 1996

Rights (CSR) group to
promote TI arguments
and target tobacco control
programmes.
- Paid key officials (money
often went via advertising
agencies and law firms
so did not appear to be TI
money)
Givel and Glantz, (2001) Analysis of tobacco 1-3,6,9,10 Oppose all tobacco excise No arguments are For many of the lobbying Found that 20 states
industry documents, tax increases proposed at mentioned specifically tactics/mechanisms had low rates of tobacco
plus reviewed existing state level in USA in relation to taxation but outlined, it is unclear excise taxation (less than
case studies of TI state study notes tendency if they are related to US$0.25 per cigarette
level lobbying, plus data of TI to frame issues as taxation. On the issue pack). Concludes that this
from on the status of important because they of tobacco excises suggests TI has been able
state tobacco control impact on public / other specifically, the following to maintain a “favorable
legislation from the US sectors, rather than are highlighted: market.”
Centers for Disease because they impact on - work with and fund anti-
Control and Prevention, the TI itself tax groups
National Center for pre- - contribute funds to
emption data Chronic "national groups not
Disease Prevention and directly related to
Health Promotion, Office tobacco that make policy
of Smoking and Health. recommendations, which
Also used 1990 Coalition can effect state legislation
on Smoking or Health while allowing the tobacco
pre-emption data lobby to remain behind the
scenes", e.g., funding for
Women Involved in Farm
Economics (WIFE) and
the National Taxpayers
Conference (NTC). Both
of these groups "assisted
the tobacco lobby […] to
fight tobacco excise tax
increases"
Givel and Glantz, (2001) - Forming relationships

(contd) with minority and women’s
political caucuses “to
build coalitions to counter
tobacco excise taxes and
promote smokers’ rights
legislation”.
- Lobbying via National
Smokers’ Alliance (TI-
funded front group)
Morley et al., (2002) Analyzed TI 1-6,8 Not assessed in this Not assessed in this The findings from Not assessed in this
expenditures by state, article, which examines article this study support the article (although authors
using data from tobacco Tobacco Institute lobbying, hypothesis that in the note the Tobacco Institute
industry document but findings suggest the 1990s tobacco control was disbanded in 1998,
websites, and TI industry was particularly activities such as raising which suggests it was not
publication the Tax concerned about policy cigarette excise taxes, seen as successful at this
burden on tobacco. developments that were attracted TI resources to point)
Also collected cigarette likely to lead to increased undermine these efforts.
excise tax rates per cigarette taxes and - California and New
pack for each state for earmarked cigarette taxes York were ranked highest
1991 and 1997, to rank in 1990s in average Tobacco
states on the basis of an Institute spending, and
increase in excise tax Minnesota, Arizona and
rate over this Massachusetts also
ranked highly (all states
that had had public ballots
on raising tobacco excise
taxes and dedicated
some of the increased
revenue to TC).
On the whole, study finds
Tobacco Institute spending
correlated with state efforts
to introduce higher (and
often earmarked) tobacco
excise taxes as well as
to introduce other tobacco
control measures such as
public smoking restrictions
(relationship was stronger
for TI activities relating to
tobacco tax than smoking
restrictions, which authors
suggest may be because
the TI relied more on third
parties to lobby against
53
54
Morley et al., (2002) smoking restrictions than

(contd) it did on tobacco excise
issue)
Yerger and Malone, Analysis of tobacco 1-3,5-7,9,10 To avoid any significant Argued excise taxes - African American The success of the TI in
(2002) industry documents tobacco tax increases were regressive and groups with which the avoiding significant tax
and relevant secondary disproportionately unfair, TI had ties were used increases is not made
data sources including particularly to minorities to oppose tobacco tax clear in this study but the
newspaper and journal increase. For example, authors imply that it was
articles the Congressional Black successful in influencing
Caucus (CBC) shifted its the position on this issue
1981 support for a 10% of African American
increase in the tobacco groups and leaders
excise tax to
help restore funding
for social programmes
to a position opposing
increasing tobacco excise
taxes in 1984
(for undetermined
reasons).
- The TI then “sought
to leverage the caucus’
changed position.”
- Many other African
American groups were
also identified as being
willing to support this
position.
- CBC commissioned a
task force to review the
impact of excise taxes on
the poor, blacks, and other
minorities. The task force
issued a 1987 report,
which was used to lobby
politicians with (and which
the TI helped promote).
- TI produced economic
studies to support its
claims about taxes.
- TI relied on support
of Congress reps who
received TI money
Joossens and Raw, Analysis of tobacco 2,5-7,9,10 To maximise profits, both - Tobacco smuggling is - TI lobbies governments Does not assess success
(2003) industry documents tough legal and illegal caused by market forces, to reduce tobacco tax on in detail, although notes
plus draws on figures markets, whilst using the i.e. the price differences the basis that doing so will that taxes were lowered
concerning smuggling spectre of smuggling as between countries, solve smuggling problems in Sweden and Canada,
that are quoted in an argument for keeping which create an incentive and increase government following concerns about
published articles, taxes low to smuggle cigarettes revenues smuggling. Notes this did
in data provided by from cheaper to more not adequately deal with
customs, tax and health expensive countries smuggling
authorities
Alamar and Glantz, Analysis of tobacco 1-3,5-7, 9,10 - To prevent a significant - Attempted to frame the - Used Wall Street - TI was successful
(2004) industry documents. tax increase being bill as the ‘‘largest tax analysts to present (McCain Bill was defeated
proposed in the McCain increase in history’’ and a arguments about in June 1998)
Bill. bureaucratic mess. smuggling (TI briefed
- Claimed the increase analysts, who then
would lead to increased presented themselves as
levels of illicit trade and independent)
associated law and order
problems.
Levenstein et al., (2005) Analysis of tobacco 1,2,5-7,9,10 To prevent the passage of Tobacco excise taxes - TI set up the LMC - The LMC was “generally
industry documents two types of legislation – were framed as unfair and - From mid-1980s to late- successful in gaining
and union publications, excise taxes on tobacco regressive 1990s, the LMC “worked labour support in New
newspapers, and other products and workplace to elicit support in New York for positions on
general publications smoking restrictions. York by framing issues excise taxes and,
relevant to the case It aimed to prevent in terms that made them especially, worksite
study, plus examined questions on these issues salient to unions.” smoking restrictions.
New York case law and from qualifying for ballot - The unions sided with However, by the late
legislative history, and and to defeat those TI in hope “that such 1990s, the support had
analysed interviews with proposals that made it to cooperation would be largely evaporated, with
key personnel (unclear if the ballot of advantage to them in trade unions in New York
interviews were already their efforts to protect either in support of, or
undertaken as part of and strengthen their at least neutral on,” both
other project or if were organisation. The TI and issues.
part of this study) LMC did their homework, From 1985, tobacco
understood the concerns excise tax ‘increased only
of labor, and appeared slowly and marginally’ but
ready to champion these in 1999 it was doubled.
concerns.” This included - LMC succeeded in
assisting unions even on dividing public health and
issues of “no concern to labor for a while.
the TI”. - However, the LMC
largely failed to attract
support from public sector
55
56
Levenstein et al., (2005) - The TI’s focus was on unions in New York state
(contd) the leaders of unions / as they were dependent
labour groups, rather than on state budgets so could
their members as it was not be counted on to resist
these people who had any action that might help
political influence. replenish state coffers.
- The LMC lobbied and Hence, “While regularly

briefed elected officials at siding with the LMC on
all levels of government, the issue of smokefree
and worked to discourage worksites,” the New York
liberal and labour public sector unions did
coalitions from taking not tend to denounce
anti-tobacco positions, proposals for cigarette tax
built support for industry increases.
positions throughout the - The TI eventually gave
labour movement, and up trying to win public
provided general public sector union support on
communications this issue, which authors
decide was a “fatal error”.
- “By the late 1990s […]
most of labor in New York
had shifted to support for
anti-tobacco policies.”
Shift in support started
with excise but then
moved on to other TC
issues
Balbach and Campbell, Analysis of tobacco 1-7, 9,10 - Overall aim was to - That cigarette taxes - In 1984, the Tobacco - TI was successful in
(2009) and Balbach et industry documents and prevent tobacco excise are regressive and that Institute established influencing CLUW’s
al., (2006) other related documents increases. regressive and unfair to the Labor Management position on tobacco
(CLUW News, the Intermediate aims working women Committee (LMC), which excise taxes (CLUW did
in-house newspaper of included discouraging was composed of the campaign against such
CLUW) liberal and labour groups Institute and five unions increases & against
from taking anti-tobacco representing tobacco earmarking).
positions (including on industry employees. The - Authors acknowledge
tobacco excise) and LMC’s functions included: “it is difficult to know
shifting the focus of the (i) Lobbying and briefing how much or if the
debate away from the elected officials at all CLUW’s involvement in
effects of tax increases levels of government;; the policy process made
on cigarette consumption (ii) discouraging liberal a difference,” but say
and onto the effects on the and labor coalitions “there are indications
people paying cigarette from taking anti-tobacco that organized labor was
taxes positions, including on important in federal level
tobacco excise taxes;; policy making”.
Balbach and Campbell, (iii) building support

(2009) and Balbach et for industry positions
al., (2006) (contd) throughout the labor
movement;; and (iv)
facilitated general public
communication on
relevant issues.
- The TI provided financial
(at least $221 500
between 1988 and 1996)
and in-kind support to
the Coalition of Labor
Union Women (CLUW) in
order to engage CLUW’s
interest in smoke-free
worksite legislation and
tax increases.
- The Tobacco Institute
facilitated its relationships
with CLUW (and similar
groups) by working via the
LMC.
- The Institute supported
the production of a series
of studies by labour
groups, including CLUW,
which demonstrated
the regressive nature of
tobacco taxes.
- The Institute monitored
press coverage and
advised CLUW on how to
promote study messages
to the media.
- CLUW and similar
organizations were
encouraged to oppose
earmarked tobacco taxes,
such as those proposed
in Clinton universal
healthcare proposals.
- The TI was aware many
labour organizations,
including CLUW, were
generally supportive
of higher taxes so they
encouraged them to
57
58
Balbach and Campbell, push for higher taxes on

(2009) and Balbach et corporations, rather than
al., (2006) (contd) cigarettes. This included
assisting the Citizens for
Tax Justice group with the
promotion of its annual
examination of ‘‘corporate

freeloaders’’ – those
corporations that pay no,
or little, taxes.
- TI framed tobacco
taxation as part of generic
consumer excise taxes
(which also including
petrol and alcohol)
Breton et al., (2006) Analysis based of: 1,2,5,6,8 To reduce tobacco excise Initially, TI challenged tax - Various groups - Initially, the TI and its
(i) 569 articles from taxes: “When the cost increases primarily on supported the TI position allies had little influence
four Québec daily (in constant dollars) of economic grounds, with (retail sales sector and – the TI was seen to be
newspapers published tobacco caught up in arguments concerning the journalists). Unclear how in decline anyway in
between 1986 and 1998;; 1984 with its 1970 level, industry’s profitability. much the TI was linked to Canada.
(ii) 28 semi-structured the tobacco industry - Also argued taxes were groups/individuals in these - However, framing the
interviews and union leaders unfair on smokers. sectors. The Québec Food issue as a ‘contraband
with key informants, representing workers - Later focused on Retailers Association problem’ succeeded
and (iii) more than 200 in this sector began to arguments that the tax was a particularly active in winning support
government documents/ challenge decisions to increases were fuelling group, which called a from a broad coalition,
transcriptions of increase the taxes.” (So smuggling (including high press conference in 1992 including retailers, media
parliamentary debates initial aim was to avoid economic cost of this to and demanded “a 70% commentators and
relating to the problem of further tax increases). government and retailers, reduction in tobacco the representatives of
smoking However, “In early 1992, and failure to help control taxes to put an end to employers’ organizations.
a plea for a drastic tobacco use). smuggling”. Authors claim - The health lobby
reduction in tobacco taxes - Argued against export TI was involved in funding received less media
emerged.’’ taxes and better policing this group. coverage and was subject
as solutions to the - A group specifically to criticism that they had
smuggling problem. campaigning on this an ‘extremist’ stance.
- Presented taxes as issue, the Mouvement - Eventually, “a massive
excessive and TC groups pour l’abolition des taxes reduction in the Québec
as having dominated réservées aux cigarettes and federal taxes” was
policy. (MATRAC), was set achieved, not just in
- Following tax decreases, up but authors do not Québec but also ‘in the
the TI argued smoking comment on whether TI taxes levied by five other
rises in young people was involved in funding/ provinces, including
were unrelated to the establishing this group. Ontario’ and also at a
change in price federal level.
Breton et al., (2006) - Claims the TI was - This led to an increase
(contd) able to mobilize union in smoking rates amongst
representatives and its young people.
employees. - Political elites responded
- TI helped organize to this by implementing
small retailers, whose a range of strong TC
comments quoted in strategies (although
the media gained public they did not re-raise the
sympathy. tobacco tax).
- Ran an “effective media - Concludes, the TI’s
strategy”. success was mixed: “the
- Managed to get police smuggling crisis was an
support on smuggling event that, despite its
issue. unfortunate repercussions
- Disseminated the on tobacco taxes, helped
findings of surveys on put the tobacco problem in
different facets of the Québec, especially among
problem. young people, on the
- Arguments were government’s agenda.”
presented as being on
behalf of citizens, not
the TI
Givel, (2006) Analyzed data from the 1,2,5-7 To avoid tobacco excise Argued tobacco use is Study only highlights - “From 1990 to 2003,
State Tobacco Activities increases a matter of individual TI mechanisms/tactics there was a sharp
Tracking and Evaluation choice and it is therefore generally so the following mobilization by health
System, Tobacco Map reasonable to expect do not only concern tax- advocates in all states
Reports created by freedom from excessive related lobbying: and a significant rise in
the U.S. Centers for government regulations - Ongoing employment of new legislation to control
Disease Control and and taxes well established contract tobacco use. The tobacco
Prevention, National lobbyists, who are industry, nevertheless,
Center for Chronic managed and coordinated was able to generally keep
Disease Prevention and in a top-down manner by state tobacco taxes low
Health Promotion, Office TI. “The state contract and counter significant
of Smoking and Health lobbyists often have regulatory threats to
(CDC). Supplemented decades of experience tobacco sales.”
this with data from the and expertise in working - Concludes “the policy
Coalition on Smoking with state legislatures monopoly favoring the
or Health, American to advance or block tobacco industry did not
Lung Association, and legislation.” significantly change,
Federation of Tax. Also - Direct and indirect despite the symbolic
appears to also use campaign contributions;; appearance of punctuation
some TI documents - gifts and honoraria to in the policy system.”
(although this is not legislators;;
mentioned in description - occasional alliances with
of methods) other allied interest groups
(such as the hospitality
industry to counter clean
59
60
Givel, (2006) (contd) indoor air ordinances);;

- The development of
front groups allowed the
tobacco industry and
its political allies to act
without being publicly,
including the National

Smokers Alliance
(primarily funded by Philip
Morris).
- Tactics of the front
groups often included
orchestrated grassroots
advocacy campaigns
through the mail or phone
calls to pressure policy
makers.
- Providing testimony
before legislative bodies
Givel, (2007) Analysis of tobacco 1–3, 5–10 - To keep tobacco taxes - The NSA (TI front group - In 1993, TI established - “Despite the use of
industry documents, low and avoid undue – see next cell) argued the National Smokers traditional lobbying in
plus relevant newspaper regulations (avoid that adult tobacco use Alliance (NSA), “a public conjunction with the use
reports (Lexis Nexus), altogether or keep should not be hindered relations created front of public relations efforts,
all relevant articles weak), which formed by rigorous regulatory group funded by the attempts by the NSA to
from the Americans for part of broader efforts controls or higher tobacco tobacco industry, which dominate public policy
Nonsmokers’ Rights to “maintain, enhance, taxes. operated nationally from to weaken or neutralize
database (http://www. and protect the industry’s - Called into question the 1994 to 1999 to advocate stronger tobacco
tidatabase.org/), profits and market shares” potential health effects for adults using tobacco regulations and taxes
magazine articles, web of tobacco use and the products without vigorous were effective only for
pages, journal articles, viability of proposed regulation or increased some campaigns.”
and books tobacco tax increases. tobacco taxes.” - “From January 1994
- Presented smoking as - Burson-Marsteller, to June 1999, the NSA’s
socially acceptable for the PR firm, helped set record of political victories
adults this up for Philip Morris and shaping of public
(with some subsequent policies was mixed. At
financial assistance from the national level, the
other TTCs) and some NSA played an important
Burson-Marsteller staff supporting role in winning
worked for the NSA. four major campaigns and
- The NSA worked losing one. At the state
to “to ‘‘generate’’ the level, the NSA played an
appearance of public important supporting role
support through public, in losing three campaigns
Givel, (2007) (contd) relations, through the and winning one. At
engineering of consent the local level, the NSA
theory, by using editorials played an important
newspaper articles, and supporting role in winning
pro-tobacco columns that 19 campaigns and losing
called for the free use of 12.” (Not all campaigns
tobacco.” concerned tax increases).
- Also organized national - “[T]he NSA was
and regional press unable to successfully
conferences, issued press discredit tobacco control
releases, monitored media researcher Stanton
coverage of tobacco, Glantz.”
made radio talk show
appearances, issued
alerts to NSA members,
provided public speakers,
issued advertisements,
and corresponded with
legislators.
- Wrongly targeted
University of California–
San Francisco tobacco
control researcher and
Professor of Medicine
Stanton Glantz, alleging,
but not proving, Glantz
misused grant funds for
illegal political purposes
and lobbying.
- Filed a Senate ethics
complaint against
Sen. John McCain,
alleging he had used
“his Senate franking
privilege to bolster his
run for the presidency”
(PM disagreed with this
and withdrew financial
assistance for NSA over
this issue)
Campbell and Balbach, Analysis of tobacco 1-3,5-10 To challenge policy Tobacco taxes are - TI created and funded - In exchange for funding,
(2008) industry documents. interest in significantly regressive the Consumer Tax various labour/progressive
raising tobacco taxes by Alliance (CTA) in 1989 to groups did publicly
turning labour and middle- build opposition to excise take positions against
class opinion against taxes. The CTA focused increasing tobacco excise
prospective excise on garnering support from taxes.
61
62
Campbell and Balbach, tax increases. Initial focus credible liberal allies, - “The Tobacco Institute
(2008) (contd) was on federal increases particularly genuine public was pleased with the
but in 1991, efforts interest groups (groups impact of the [TV]
turned to state-based tax traditionally perceived to advertising [sponsored by
proposals be at odds with the TI). NB the CTA], as measured
The CTA appears to have by tracking polls
been dissolved in 1993. commissioned to measure

- Also focused on their effectiveness” which
garnering media coverage found they increased
of the issue (this included people’s opposition to
sponsorship of TV ads, tobacco excise taxes.
plus a media tour by CTA - Tobacco excise taxes
Executive Director). were increased in 1990
- By working via the CTA and 1993 at USA federal
(which did not divulge level, but the increase was
the industry’s role in its much less than originally
formation), the TI was able predicted (8 cents, rather
to obscure its own role in than 16 cents) and was
these efforts. introduced in two phases
(authors accept that it is
not possible to say whether
or how CTA activities
affected this outcome).
Kelton and Givel, (2008) Analyses TI documents, 1,2,4-7,9,10 To maintain/expand sales TI blamed rampant - TI “promoted smuggling - Smuggling hindered
newspaper reports, despite tax increases smuggling on excessive schemes” not only to Canadian tobacco tax
journal articles, scholarly and to use the issue of taxation increase profits but also increases from furthering
reports, court cases, smuggling as part of to “provide an argument a consistent reduction
government data, court campaign to reduce taxes for tobacco taxation in consumption rates in
testimony, and federal reduction.” Effectively, 1990s.
and state statutes the TI “secretly pushed - The Canadian
and regulations, plus for increased tobacco government responded
relevant information from smuggling so that it could to the influx of smuggled
websites argue against higher tobacco that culminated
taxes as a motivation to after the tax increases by,
smuggle.” in 1994, returning taxes
- “Because of the federal to their original level.
excise and export tax “The intention of the tax
breaks that apply on U.S. reduction was to alleviate
Native American land, for the growing illegal activity,
more than 10 years the but the illegal activity was
tobacco industry utilized more than just a result
this land as smuggling of high taxation. The
routes to avoid newly Canadian government was
Kelton and Givel, (2008) imposed Canadian taxes ill-prepared and unable
(contd) on tobacco products. to reduce or eliminate
By 1993, this tobacco tobacco smuggling.”
smuggling accounted for - “The tax rollback has
30 to 40 percent of the prevented the Canadian
total Canadian tobacco government from earning
market.” public finances estimated
in the billions of dollars
from tobacco taxes that
would have alleviated
the societal cost of
Canadian smokers. Lower
cigarette prices have also
injured the public health
by increasing smoking
prevalence to levels that
would not have existed if
the tobacco tax increases
had met their expected
goals.”
Campbell and Balbach, Analysis of tobacco 2,3,5-7,9,10 - To change the debate - That cigarette taxes - Tobacco taxation was - CTJ initially responded
(2009) industry documents and from the effects of are regressive and that framed as part of a broader to TI approach by saying
media reports (located cigarettes to the people regressive tax increases category of consumer tobacco excise was not
through searches on paying cigarette taxes. are unfair. excise taxes, also including a priority issue for the
Lexis—Nexus) - To discourage liberal and - That this is true even petrol and alcohol. organisation, but agreed
labour groups from taking when taxes are earmarked -Tobacco Institute’s to work with the TI when
anti-tobacco positions, for healthcare spending. strategy included it received funding,
including on tobacco - That there is little recruiting "organized labor, from 1984 onwards.
excise increases evidence that increasing minorities, and other liberal Relationship deepened
taxes on alcohol and groups" to provide early in 1986, as more funding
cigarettes reduces warnings of legislative tax was provided. CTJ took
consumption initiatives, help tobacco anti tobacco tax positions
industry lobbyists gain and lobbied on this issue,
access to legislators who including by testifying
were not industry allies, before the Senate Finance
demonstrate constituent Committee against an
support for pro-tobacco increase in the federal tax
votes, and testify on the on cigarettes.
industry’s behalf. The - TI appears to have been
Institute was successful pleased with activities on
in forging relationships tax undertaken by both
with—and providing CA and CTJ, including
significant financial on earmarking issue
support to two prominent (which included Clinton’s
progressive organizations. healthcare proposals).
63
64
Campbell and Balbach, Citizens for Tax Justice - The Tobacco Institute
(2009) (contd) (CT) and Citizen Action, was dissolved in 1998
to oppose cigarette excise as part of the Master
tax increases as part of its Settlement but “interest
policy efforts.” in cigarette excise
- As other articles outline, taxes remains high” and
the LMC was used to “controversy continues

obscure connections over the economic
between the Tobacco hardship they may cause
Institute and the labour for low-income smokers”
groups it worked with, and on “the fairness of
such as CTJ and CA. raising revenue from
- TI supported CTJ to one population subgroup
hold a conference in (smokers) for programs
1987, which focused on with broader social
promoting the arguments benefits.”
that had been used at
federal level to state level.
- CTJ lobbied Senate
Finance Committee.
- CTJ position pieces
covered in media.
- As tobacco tax concerns
moved increasingly to
state level, the TI (through
LMC) recruited Citizen
Action in a similar way
to CTJ and for similar
reasons (CA was better
organised at state level
though, being more of a
grassroots organisation).
- TI also recruited other
progressive tax groups.
- CA and CTJ were also
used to help the TI lobby
against proposals that
tobacco tax increases
would be earmarked for
health care spending
Lum et al., (2009) Analyzed proposed 1-3,5-10 To avoid tobacco excise - Argued proposed tax - TI developed a voter - “The industry effectively
ballot measures in US tax increases and, in increases did not dedicate segmentation model to influenced early voters.”
states (used various particular, to avoid enough to tobacco control determine which tobacco - Tobacco tax ballot
documentary sources) earmarked taxes and that hospitals and tax increases it could measures commonly
and TI documents. health maintenance defeat in USA states. allocated substantial funds
Used public and private organizations would profit - After doing market to medical services,
polls (obtained from (i.e. ‘‘only a small amount research on the issue, despite lack of public
proponents of the ballot of the tax increase will go sought to present tobacco support for this. Tobacco
measures, pollster to education about the taxes as part of broader tax companies “are becoming
websites or the tobacco hazards of smoking.’’). rise issues. more successful in making
industry documents) - Argued tax increases - The Tobacco Institute this use of funds an issue.”
to identify early levels are unfair […] organized its Tobacco - Between 1988 and 1998
of support for the tax - Built on popular themes Industry Ballot Issues the TI mounted extensive
proposals and how of ‘‘anti-tax/ HMO Committee, which provided opposition campaigns
this support compared scheme’’ and ‘‘what ‘‘oversight of all prevention, to all nine tobacco tax
with the final election they aren’t telling you’’ preparation, and execution ballot measures that
outcomes as effective arguments of tactics relating to ballot were proposed, but only
against ballot measures issues, as well as the place defeated four (44%) of
to discuss new strategies them. From 1998 to 2008
and technologies related to it only challenged five out
battles in this arena’’. of 13 of the tobacco tax
- TI was advised to: lobby measures and defeated
legislatures to ‘‘reform four (so 80% of those
initiative and referendum it challenged). Thus it is
laws to make qualification becoming more successful
of ballot issues more in campaigns it conducts
difficult’’, ‘‘encourage third but probably at least
party ballot issues which partially because it has
threaten […] opposition and become more selective.
impede their progress’’, - Claims TI “learned to
mount ‘‘legal challenges combine the argument
which complicate that the tax would
opposition progress’’ and primarily benefit hospitals
prevent ‘the opposition and HMOs with lack of
from using inappropriate funding for benefits for
funding sources for their smokers. When either of
political activities’. the two arguments was
- TI also conducted public used alone, the tobacco
relations campaigns, industry lost three out of
conducted benchmark four elections, but when
research and built they were combined,
partnerships with other they won three out of four
organizations that might be elections.”
needed in a campaign. - Claims TI spending alone
- In the 2000s, the industry does not explain outcomes
became much more of state proposals to
selective in which increase tobacco taxes.
65
66
Lum et al., (2009) (contd) campaigns it opposed

(although, after deciding
not to fight the tax
proposals in Washington,
it did also realize that not
opposing such increases
was likely to encourage

states to take action so
began also taking this into
consideration.)
- Drew on its “essentially
unlimited financial
resources”
Raebeck et al., (2009) Analysis of tobacco 1-3,5-7,9,10 - The goals of the LMC - Argued excise taxes are - TI provided significant - The TI did appear to be
industry documents and included lobbying elected regressive funding to African and successful in influencing
academic articles officials, discouraging American and Latino the position on tobacco
liberal and labour labour organizations and excise of some key African
groups from taking also sponsored special American and Latino
anti-tobacco positions, projects within these labour organizations
and building support for organizations, including (notes at least one key
industry positions among studies, brochures, and group appeared to shift
organized labour. Notes conference events. its position on tobacco
tobacco taxes and smoke- - Helped support these control issues between
free legislation were seen organizations’ priorities, 1984 and 1988). This is
to be most effective at even where they did not despite the fact that other
reducing smoking rates directly relate to tobacco. research the authors
(and implies, therefore, - Framed discussions of cite indicates these
that these were perceived tobacco taxes as part of groups’ core population
by the Tobacco Institute to debates about broader, constituencies did support
be the biggest threats) ‘consumer’ excise taxes. tobacco tax increases.
- Framed issues so that - According to a 1990
they would be more Institute briefing, the
appealing to labour and LMC’s use of allies,
minority groups which such as the African and
the TI was trying to work American and Latino
with, e.g. presented Labor Organizations
policies (tobacco taxation it was supporting,
increases and smokefree ‘‘demonstrate[d] to
legislation) as particularly legislators––particularly
detrimental to organized the liberal Democrats [sic]
labour and, specifically, to who are most likely to
people of colour. support increases to fund
- To facilitate working with social programs–– that
these kinds of labour consumer excise taxes
Raebeck et al., (2009) groups, the Tobacco are: Unfair Regressive
(contd) Institute established the Inconsistent with tax
Tobacco Industry Labor reform.’’
Management Committee - In 1993, when Clinton’s
(LMC) in 1984 (see healthcare plan included
above). a US$1 per pack cigarette
- The Tobacco Institute tax, the Tobacco Institute
then used the coalitions mobilized these groups to
it had formed with lobby the Congressional
these organizations “to Black and Hispanic
appeal to politicians Caucuses, respectively.
who were traditionally - Claims the TI’s
more responsive to such “reframing of ideas was
arguments, specifically particularly successful
Democrats, liberals, and because the points were
politicians of color.” valid – i.e. excise taxes
- The Institute, through are regressive…”
the LMC, supported the - Authors conclude:
release of studies, letters, “Whether the involvement
and op-eds opposing of [these organizations]
excise taxes. These were had a significant impact
often credited to the labor on tobacco policy
organizations that the decisions is unclear, but
TI was supporting via both organizations had
the LMC, rather than to political influence among
the TI, even though they policymakers, organized
were “usually written labor, and people of color.”
by the Institute or its
consultants”.
- The Tobacco Institute
paid a PR firm, Ogilvy,
Adams & Rinehart, to help
write a publication that
was officially by one of
the labour organizations
it was supporting, called,
‘‘Fair Taxes: Still a Dream
for African-Americans”,
and then promoted it ‘‘the
black and labor media’’.
- TI paid for studies that
highlighted regressive
nature of consumer taxes
(framing tobacco taxes as
part of this) and its impact
on these groups (e.g.
industry sponsored
67
68
Raebeck et al., (2009) study claimed “An African-

(contd) American family, with
both parents working, two
children and an income
of $25 000, will pay an
almost six times larger
share of its income on

federal consumer excise
taxes than a family making
$250 000 a year”).
Studies concerned with tobacco industry efforts to influence tobacco excise structures (** indicates that studies also touch briefly on lobbying around tobacco excise rates)
O’Sullivan and Analysis of tobacco 2-5,8 TTCs were "frustrated in In 1985, Clive Turner On tobacco tax issues Not assessed in this
Chapman, (2000) industry documents their export attempts to (managing director of the (which is not the main article (though presumably
China by tax barriers" and Asian Tobacco Council) focus of this study), the unsuccessful, given
wanted to promote a flat “sought to demonstrate to authors claim the TI: Chinese state monopolies
excise tax structure "to Asian governments that - Lobbied American trade still dominate).
reduce price differentials taxation across the Asian associations on the issue
between imported region was “over the top, of exports.
and locally produced and needlessly punitive” - Lobbied Chinese, US
cigarettes." and that excessive and UK officials and
taxation would “further politicians
- TTCs sought to persuade reduce the taxable base” - Tried to involve the
Chinese government that, for governments leading to Chinese monopoly in TI
in order to optimise their “a revenue reduction” organisations
revenue derived from the - TTCs tried to gain USA - Philip Morris proposed
cigarette industry, they support for its arguments some joint ventures.
should institute a “simple (and in so doing were
system, a flat excise tax careful to frame the
based on per thousand exports as being for
cigarettes.’’ existing smokers, rather
than potentially increasing
smoking rates)
Szilágyi and Chapman, Analysis of tobacco 1-3,5-7,9,10 - To keep tax increases - Tax increases will - Efforts to draw public - Philip Morris failed to
(2003)** industry documents low (BAT and Philip increase black market attention to high tobacco prevent a tax increase,
and media coverage of Morris);; - While companies taxes;; even though it achieved
the tax harmonisation - To secure a beneficial increased their prices - Use of economic significant political support
issues. This was tax structure (Philip to ensure their profits, impact studies to support on the issue. Officials who
supplemented by Morris lobbied for specific publicly they decried the arguments (e.g. a study were sympathetic to Philip
interviews with a finance structure, though other taxes imposed to them was produced by KPMG Morris’ position blamed
ministry official and an companies lobbied for to support arguments this failure on the “divided
investigative journalist other structures, e.g. BAT against introducing EU position of the industry as
lobbied for ad valorem);; tobacco tax excise levels well as heavy lobbying by
Szilágyi and Chapman, - To prevent the at time of Hungary’s BAT”.
(2003)** (contd) introduction of tax accession to EU, and - In line with TI hopes, the
stickers;; PM commissioned tax burden on cigarettes
- To get rid of import a study from the in Hungary decreased in
duties Institute for Economic real terms between 1990
Research (Italy), on ‘the and 1996.
consequences of rapid - Even so, retail prices
alignment to the EU’s of cigarettes increased
minimum tax on cigarettes dynamically as a result of
in five accession pricing policies instated by
countries’, including the industry itself.
Hungary). Such reports - The Hungarian budget
were largely sent to has still one of the lowest
relevant officials, although contributions from tobacco
sometimes also reported taxes among countries
in media;; of the former communist
- Created a coalition bloc.
against the proposed - TTCs were successful
tax stickers (which TTCs in “developing a sphere of
would have had to pay supporters within relevant
for);; ministries (with special
- Attempted to ‘divert regard to the agricultural
legislators’ attention’ and finance portfolios
away from proposed tax and MPs sympathetic
increases by offering to TTCs views) and
the creation of a special third parties (hospitality
fund to support tobacco industry, tobacco growers,
growing in Hungary (to advertising industry) ready
be levied via specific tax to react without delay in
increases on cigarettes) the case of government
attempts to raising
tobacco taxes.”
- More recently, in 1990s,
TI sources admit that
sales volumes have been
“severely affected” by
significant tax increases.
- Authors imply TI
helped shift government
position on derogation
period to meet EU
tax harmonisation
requirements, which
required tax increases.
69
70
Gilmore and McKee, Analysis of tobacco 1-3,5-7,9,10 In the context of - BAT argued the excise - BAT employed various Not assessed in this
(2004b) industry documents and persuading government to regimes it favoured would arguments to suggest that article (at least not in
supplementary data privatise state monopolies increase fiscal and excise it was a better TTC to deal relation to tax issues)
from tobacco industry (preferably through closed revenues received by with than its competitors.
journals, newspapers, deals), TTCs sought to targeted governments This included presenting
routine data, and other secure excise structures (even when BAT was itself itself as a helpful and
published reports that were favourable to involved in tax evasion supportive corporation
their products via smuggling and has which could help advise
in other circumstances a government on things like
track record of lobbying excise regimes.
for tax reductions, which - BAT attempted to
have the potential to establish good political
reduce revenues) connections/contacts to
promote these arguments,
sometimes working via
consultants.
Gilmore et al., (2005) Analysis of tobacco 1-10 BAT sought to acquire No relevant information on Not specified in relation At the time of writing, the
industry documents and a monopoly position this issue is provided by to BAT’s tax-related Moldovan state tobacco
supplementary data, (through an exclusive this study (focuses instead proposals (probably companies had still not
such as newspapers deal) and then bolster its on arguments used in because privatisation did been privatised, so BAT
position by influencing relation to privatisation) not proceed) had not succeeded
excise rules so that they
would “uniquely favour its
products” (which would
encourage smokers to
switch from filterless to
more expensive filter
brands, despite low
incomes).
- It hoped not to increase
its market share to over
80% within 4 years, by
“closing the market to
external competition” and
undertaking marketing
campaign.
- As part of efforts to
close the market to
external competition, BAT
wanted to use a system of
Banderols. It also wanted
to secure tax exemptions
for BAT products
Delnevo C et al., (2007) Statistical modeling, 1,2,6,9,10 To change the taxation - US Smokeless Tobacco No information on this Changes were
using ACNielsen data, of moist snuff from an ad argued the new tax issue in this study implemented, even
of what is likely to valorem to a weight-based system would raise ‘an though authors claim
happen to prices of moist system additional $2 million both arguments are
snuff and tax revenues in revenue’ and would flawed;; the authors
following implementation discourage ‘youth from calculate the revised
of changes promoted by buying it,’ by raising the tax system is actually
US Smokeless Tobacco price of the cheapest likely to encourage youth
in New Jersey available snuff purchasing
Gilmore et al., (2007)** Analysis of tobacco 1-7,9,10 Having secured a - Argued against the - Lobbying of government - BAT achieved a
industry documents, privatization deal in imposition of high taxation officials responsible for reduction of approximately
other documentary data Uzbekistan, BAT sought rates on the basis taxation policy. 50% in the excise tax on
and some interviews to: that high rates would - BAT offered to help cigarettes.
- introduce protective encourage smuggling officials in charge of - BAT achieved an excise
import taxes;; (even though BAT was tobacco taxation, who system it believed would
- equalize excise on involved in smuggling). were unsure of the issues benefit its brands and
imports and domestic - Argued its excise system involved and who were disadvantage those of its
production;; proposals would minimize inexperienced. Effectively, competitors (particularly
- ensure the proper smuggling and optimize BAT created “a symbiotic Philip Morris), this
control and collection government revenue relationship” between included the introduction
of taxes, particularly on (even though BAT internal BAT and the Ministry of of tax stamps in 1996, a
competitors’ import (e.g. docs indicate the driving Finance. licensing requirement for
through the introduction force was to achieve the - BAT staff produced import and export, as well
of a tax stamp system, best possible market papers on smuggling, as wholesale and retail
from which BAT would position for BAT and that which fed into the above distribution.
hopefully be exempted) the company was aware negotiations. - “The price of cigarettes
- achieve a reduction its recommendations - BAT bought up local in Uzbekistan is now the
in the excise tax on perhaps weren’t always TI to achieve market lowest of all countries
cigarettes (even though the best deal for the dominance and political in the World Health
rates were already low);; government) influence Organization’s Europe
- achieve the region, including
implementation of an those with which
excise system to benefit it is economically
BAT’s brands and comparable”.
disadvantage those of its Overall: “BAT thoroughly
competitors (particularly redesigned Uzbekistan’s
Philip Morris). This tobacco taxation system
involved a mixed excise to advance corporate
system, which would objectives” (achieved all
ensure high-end brands its objectives, bar one
(such as Marlboro) were – which it still hoped to
hit through maintaining an achieve at the end of the
ad valorem tax on inputs, study)
71
72
Gilmore et al., (2007)** and which also involved

(contd) having a specific element
for imports to try to reduce
cheaper imports (ideally,
BAT hoped this rate would
not apply to them).
- an ad valorem domestic

system of 40% on
filtered and 25% on plain
cigarettes. This was of
benefit to BAT, which
was focusing on the local
filterless brand Astra, and
planned to later introduce
other cheap local brands.
- a requirement for a
license to import and
export cigarettes, as well
as to provide wholesale
and retail distribution
Nakkash, (2007)** Analysis of tobacco 1-6,8-10 - TTCs lobbied to - BAT claimed its plan (This review only notes - BAT appears to have
industry documents, plus influence tobacco taxation would “afford some mechanisms listed been successful in
secondary documentary policy in Lebanon. protection to local in relation to TI tax influencing tax structure:
sources and 20 semi- - PM sought to achieve a manufacture;; […] swell the lobbying – more general “In May 1995, Macleod
structured interviews simplified excise system, government’s tax revenue mechanisms of TI reported an unconfirmed
with key informants with a unified flat charge [and]… […] improve influence/lobbying are report that the Lebanese
per case, effectively balance of payments also discussed in this Regie has finally come
changing the system from through import thesis): down on the side of an
ad valorem to specific substitution/local value - Lobbying of key officials, ad valorem structure
excise (circa early-mid added.” e.g. BAT set out to “[p] and has submitted a
1980s). - A BAT study warned repare lobby plan for recommendation to the
- PM and BAT were both that “significant and too Faoud Seniora Minister of Ministry of Finance”
concerned about possible frequent tax increases will Finance (known to BATCo - But taxes were
government intentions encourage a resurgence consultant).” increased: “In April
to raise tax rates as a of contraband on the one - Says BAT plan for excise 1999, tobacco taxes
control measure, and hand and discourage any reform had support from substantially increased
Philip Morris lobbied on outflow of goods from the IMF and implies BAT from 54% to 90% plus an
“the dangers of excessive market on the other.” Tax helped secure this support additional 48% as local
taxation and how a rise reform, it was claimed, consumption tax (total
in tax levels could run would “curb smuggling” 138%).”
counter to their prime - BAT argued against the - However, the April 1999
objective.” specific system Philip tax increase was later
Morris supported revoked, following claims
Nakkash, (2007)** - BAT wanted to avoid by implying it could that the rise resulted in
(contd) the specific structure exacerbate smuggling and further smuggling and
being promoted by Philip reduce the government’s losses to the Treasury
Morris and instead secure revenue.
the introduction of a tax - TI argued reducing
structure which would taxes could help reduce
render a similar return smuggling
on both premium and
low priced imports, as it
believed this would be
beneficial to its products
over the longer term.
- BAT also wanted to lobby
against any restrictions
on “imports through
increases in import duties”
* The following critical appraisal (CA) criteria were used to appraise all of the included studies (the numbers listed for each study in the CA column in Table 3.3, above, indicate the criteria for which that study met positively):
1. How clear is/are the research question(s) and/or aim(s)?
2. Was the methodology appropriate for addressing the stated aims of the study?
3. Where applicable, was the recruitment/search strategy appropriate and/or was an adequate sample obtained to support the claims being made (i.e. was the data collection adequate and appropriate)?
4. Were the methods of data analysis appropriate to the subject matter?
5. Is the description of the findings provided in enough detail and depth to allow interpretation of the meanings and context of what is being studied? [Are data presented to support interpretations, etc?]
6. Are the conclusions justified by the results?
7. If applicable, are the theoretical developments justified by the results?
8. Have the limitations of the study and their impact on the findings been considered?
9. Do researchers discuss whether or how the findings can be transferred to other contexts or consider other ways in which the research may be used?
10. If the answer to 9 is ‘yes’, do you agree these suggestions are appropriate, based on the research?
** indicates that studies also touch briefly on lobbying around tobacco excise rates
73
Table 3.4. Summary of the foci of North American studies to highlight Tobacco industry efforts to keep
potential overlap
tobacco taxes low and prevent
earmarking
North American policy development /
References focusing
tobacco industry tactic (categorised Total
on this development
according to main focus of article) Tobacco industry arguments to keep
tobacco taxes low
TI alliances with labour and minority (Yerger and Malone, 2002;; Levenstein et 6
groups in USA to counter tax increases al., 2005;; Balbach and Campbell, 2009;;
(mostly earmarked) Balbach et al., 2006;; Campbell The arguments employed by the
and Balbach, 2009;; Raebeck et al., 2009)
industry to prevent tobacco excise
TI establishment and use of front (Givel, 2007;; Campbell and Balbach, 2 increases or achieve reductions in
groups to counter tax increases (mostly 2008) current rates were relatively consistent
earmarked)
and (in order of prominence with
Various – US federal- and state-level (Givel and Glantz, 2001;; Morley et al., 5
lobbying to counter tax increases (some 2002;; Alamar and Glantz, 2004;; Givel,
which they featured in the studies in
of which covered various specific policy 2006;; Lum et al., 2009) this review) involved:
developments in this table – many were
earmarked)
Emphasizing the regressive nature of
Proposition 99 in California (earmarked) (Begay et al., 1993;; Traynor and Glantz, 3
tobacco taxation
1996;; Balbach et al., 2000)
Proposition 200 in Arizona (earmarked) (Bialous and Glantz, 1999) 1
The argument employed by the
Measure 44 in Oregon (earmarked) (Goldman and Glantz, 1999) 1
industry most often in the studies
Question 1 in Massachusetts (Koh, 1996;; Heiser and Begay, 1997) 2 included in this review involved
(earmarked)
pointing out that tobacco excises
Failed initiative in Montana (earmarked) (Moon et al., 1993) 1
are regressive and consequently
Tax increases in Canada in early 1990s (Breton et al., 2006;; Kelton and Givel, 2
claiming that higher taxes are unfair
2008)
on poorer and more marginal groups
TI lobbying of excise structure for moist (Delnevo et al., 2007) 1
tobacco in society (Koh, 1996;; Traynor and
Glantz, 1996;; Heiser and Begay,
Total 24 1997;; Yerger and Malone, 2002;;
Levenstein et al., 2005;; Balbach et
al., 2006;; Campbell and Balbach,
2008, 2009;; Balbach and Campbell,
Table 3.5. Geographic breakdown of studies 2009;; Raebeck et al., 2009). In
some cases, the industry went
No. of studies No. of studies as far as arguing that tobacco tax
No. of studies examining industry examining industry
Geographic focus increases would contribute to class
overall influence on excise influence on excise
rates structures warfare, pitting upper-middle-class
USA liberals (mostly white) against lower-
22 21 1
middle-class working people (mostly
Canada 2 2 0 minority) (Koh, 1996). By employing
EU countries* 2* 2* 1 (Hungary) this argument, the tobacco industry
was able to forge partnerships with
Former Soviet Union 3 1 3
non-traditional allies, such as labour
China 1 0 1 groups (unions, etc.) and organizations
Lebanon 1 1 1
representing minority ethnic groups
in their campaigns against tobacco
Total 31 27 7
tax increases (see below). Although
*One covered Hungary which was not at that time an EU Member State;; the other covered the
tobacco taxes represent a higher
United Kingdom and Sweden percentage of the income of poorer
consumers (Chaloupka et al., 2001;;
Cnossen, 2006) and because poorer
groups tend to consume more
74
tobacco products, the tax burden on the studies refer generically to the smokers with excessive taxes (Givel,
these groups is higher (Chaloupka “industry”, it should not of course 2006, 2007). Such arguments ignore
et al., 2001;; Cnossen, 2006), this be assumed that every tobacco the fact that nicotine is addictive
argument ignores research which company, major or minor, took part (Advisory Group of the Royal College
finds that tobacco tax increases tend in the smuggling, or other activities of Physicians, 2000) and, therefore,
to be progressive (Chaloupka et al., that are the subject matter of these that smoking is not a ‘choice’.
2001;; Gruber & Koszegi, 2008). studies and the others referred to in Indeed, research has repeatedly
This is explained further in Chapter this chapter. demonstrated that most smokers do
7. When promoting this argument, In one study (Traynor and Glantz, not want to smoke (Robinson and
one study (published as two articles) 1996), the authors found that the Bugler, 2008;; Gallup, 2010).
even reported that the industry had industry used arguments about
involved itself in a campaign calling for smuggling to claim that tobacco tax Denying links between price and
tax increases on large corporations increases would therefore lead to consumption
rather than on consumer products broader crime-related problems, such
such as tobacco (Balbach et al., 2006;; as more money being spent on guns One study in the USA found that the
Balbach and Campbell, 2009). This and drugs, with police attention being industry was denying the existence
is despite the fact that large tobacco diverted away from other crimes to of credible evidence that increasing
companies have themselves been tackle tobacco smuggling. Working taxes on cigarettes reduces
involved in efforts to avoid paying through a front group it had helped consumption (Campbell and Balbach,
high rates of corporate tax (Knorr, establish, Californians Against Unfair 2009). Another, in Canada, found that
1985;; Batuke, 1992;; Burgess, 1993;; Tax Increases (CAUTI), the industry following tax decreases there was
O’Callaghan, 1998) and are members was able to secure endorsements an increase in smoking rates among
of organizations, such as the US for these arguments from the young people, but the industry argued
Chamber of Commerce (Murray, Californian Sheriffs’ Association that these were unrelated to the
1989;; Molinelli, 1999), that are and the California Peace Officers’ change in price (Breton et al., 2006).
actively pushing for lower corporate Association, which lent credibility Such arguments were made despite
taxes (US Chamber of Commerce, to such claims (Traynor and Glantz, the overwhelming evidence on this
2011). 1996). However, various factors issue (see Chapters 4 to 7) and the
eventually undermined this element fact, as detailed earlier in this chapter,
Linking higher taxes to illicit trade of the industry’s campaign, including that the tobacco company internal
and organized crime an official report that concluded the documents show how the industry
effect of the increase on smuggling was fully aware of the relationship
The second most-popular argument was likely to be negligible, and between price and consumption and
employed by the tobacco industry which criticized an advertising was using such evidence to plan the
in this review involved claiming that campaign that CAUTI had funded nature and timing of price-related
tax increases lead to higher rates which highlighted this argument marketing campaigns (Chaloupka et
of illicit trade (Koh, 1996;; Traynor (the report led to police groups that al., 2002).
and Glantz, 1996;; Heiser and had supported the industry position
Begay, 1997;; Joossens and Raw, dropping their support) (Traynor and Tobacco industry arguments
2003;; Szilágyi and Chapman, 2003;; Glantz, 1996). to prevent earmarking
Alamar and Glantz, 2004;; Breton
et al., 2006;; Gilmore et al., 2007;; Claiming tobacco taxes are unfair to Where proposals to increase tobacco
Nakkash, 2007;; Kelton and Givel, smokers taxes included plans to earmark the
2008). Such arguments were use taxes to pay for health programmes
in spite of extensive evidence that The industry also argued, quite (particularly tobacco control initiatives
the industry has itself been involved simply, that high tobacco taxes were and health care programmes), the
in global smuggling operations unfair to smokers (Koh, 1996;; Breton industry also consistently argued that
(see Chapter 8) and specifically in et al., 2006;; Lum et al., 2009) and, by (again, arguments are listed in order
promoting smuggling following tax claiming that smoking was a matter of prominence in the studies included
increases, as explored further below of individual (adult) choice, insisted in this review):
(Kelton and Givel, 2008). Although it was unreasonable to burden
75
The extra funds would be diverted/ Claiming that earmarking would see Figure 3.4 (Moon et al., 1993;;
misused result in spending cuts for specific Koh, 1996;; Traynor and Glantz, 1996;;
programmes Heiser and Begay, 1997;; Bialous and
Many of the studies found the tobacco Glantz, 1999;; Goldman and Glantz,
industry, and various groups it lobbied The same study found evidence of 1999;; Balbach et al., 2000;; Givel and
through, claimed that earmarked the industry arguing that earmarking Glantz, 2001;; Levenstein et al., 2005;;
funds would be used in ways which would lead to cuts in spending on Balbach et al., 2006;; Givel, 2006,
the public did not support or which programmes that were currently 2007;; Campbell and Balbach, 2008;;
differed from those described in the funded by tobacco tax, even though Raebeck et al., 2009).
original proposal (Moon et al., 1993;; this was not true (Goldman and Such groups were used both to
Koh, 1996;; Traynor and Glantz, 1996;; Glantz, 1999). promote arguments directly (Moon
Heiser and Begay, 1997;; Bialous et al., 1993;; Koh, 1996;; Traynor and
and Glantz, 1999;; Goldman and Tobacco industry tactics to keep Glantz, 1996;; Heiser and Begay,
Glantz, 1999;; Lum et al., 2009). In tobacco taxes low and prevent 1997;; Bialous and Glantz, 1999;;
the context of the USA, where health earmarking Goldman and Glantz, 1999;; Balbach
care is largely private, the industry et al., 2000;; Givel and Glantz, 2001;;
used proposals to use some of the The tactics/mechanisms employed Levenstein et al., 2005;; Givel, 2006,
earmarked funds for health care of by the tobacco industry campaigns 2007;; Raebeck et al., 2009) and to
poorer groups to emphasize this to counter excise increases mirror help recruit credible allies (Traynor
issue, framing these developments industry activities on a range of other and Glantz, 1996;; Levenstein et al.,
as a diversion of the funds to doctors, issues (see, for example, Feldman 2005;; Balbach et al., 2006;; Campbell
hospitals and health care companies and Bayer, 2004;; World Health and Balbach, 2008;; Balbach and
rather than to the uninsured public Organization, 2009). These included Campbell, 2009), a tactic discussed
(Bialous and Glantz, 1999;; Traynor (in order with which they feature in in more detail in the following section.
and Glantz, 1996). Such efforts the studies included in this review):
were helped by the fact that health Working with credible allies
care/insurance organizations often Use of front groups
did want to divert the funds towards In addition to creating front groups,
healthcare costs and also worked To obscure its own economic the industry was able to ally itself
(sometimes in collaboration with the interests in promoting many of the with a wide range of pre-existing and
tobacco industry) to try to achieve arguments outlined above and credible public interest groups not
this (Begay et al., 1993;; Traynor & thereby increase their credibility, normally associated with tobacco,
Glantz, 1996;; Balbach et al., 2000). the studies reveal that the industry thereby increasing the perceived
frequently established “front groups”;; credibility of key arguments (Begay et
Arguing that policies requiring
smokers to subsidise policies
Figure 3.4. Front groups identified in the studies
benefiting others are unfair
One study found the industry framed 7REDFFR,QVWLWXWH¶V/DERU0DQDJHPHQW&RPPLWWHHPDGHXSRIXQLRQJURXSV

the use of earmarked funds from associated with the tobacco industry) (Levenstein et al., 2005;; Balbach et al., 2006;;
tobacco taxes to pay for healthcare- Balbach and Campbell, 2009;; Campbell and Balbach, 2009;; Raebeck et al., 2009),
related costs as equating to a tax on &DOLIRUQLDQVIRU6PRNHU¶V5LJKWV%DOEDFKet al., 2000),
smokers to pay for services for others, (QRXJKLV(QRXJK%LDORXVDQG*ODQW]

which it argued was unfair to smokers 1R0RUH7D[HV%LDORXVDQG*ODQW]
(Goldman and Glantz, 1999). &RQVXPHU7D[$OOLDQFH&DPSEHOODQG%DOEDFK
1DWLRQDO6PRNHUV¶$OOLDQFH*LYHODQG*ODQW]*LYHO
7KH)DLUQHVV0DWWHUVWR2UHJRQLDQV&RPPLWWHH*ROGPDQDQG*ODQW]
7KH&RPPLWWHH$JDLQVW8QIDLU7D[HV.RK+HLVHUDQG%HJD\
&LWL]HQV$JDLQVW0RUH7D[DQG%XUHDXFUDF\0RRQet al., 1993),
&DOLIRUQLDQV$JDLQVW8QIDLU7D[,QFUHDVHV7UD\QRUDQG*ODQW]
76
al., 1993;; Moon et al., 1993;; Traynor groups representing medics and police and was also a part-time actor
and Glantz, 1996;; Balbach et al., healthcare organizations in the USA (Traynor and Glantz, 1996).
2000;; Givel and Glantz, 2001;; Yerger (Begay et al., 1993;; Balbach et al.,
and Malone, 2002;; Levenstein et al., 2000). Mounting legal and other official
2005;; Balbach et al., 2006;; Breton challenges
et al., 2006;; Givel, 2006;; Campbell Traditional lobbying
and Balbach, 2008, 2009;; Lum et Seven studies reported industry
al., 2009;; Raebeck et al., 2009). Ten of the studies also reported efforts to mount legal or other official
By emphasizing the regressive evidence of the tobacco industry challenges to proposed and existing
nature of tobacco taxes, the using more traditional lobbying excise legislation (Koh, 1996;; Traynor
industry was able to entice various techniques by targeting key decision- and Glantz, 1996;; Heiser and Begay,
labour and minority ethnic groups makers, both directly and indirectly 1997;; Bialous and Glantz, 1999;;
(which it commonly approached (e.g. via consultants, campaign Goldman and Glantz, 1999;; Givel,
via the Tobacco Institute’s Labour groups, business organizations, etc) 2007;; Lum et al., 2009). For example,
Management Committee and an (Begay et al., 1993;; Traynor and where earmarked tax increases were
industry-created front group known Glantz, 1996;; Bialous and Glantz, being pursued via a public ballot
as the Consumer Tax Alliance;; see 1999;; Balbach et al., 2000;; Morley et system (a method which gained
below) to argue against tobacco tax al., 2002;; Joossens and Raw, 2003;; popularity among public health
increases (Yerger and Malone, 2002;; Levenstein et al., 2005;; Balbach et communities in eligible American
Levenstein et al., 2005;; Balbach et al., 2006;; Givel, 2006, 2007). states from the late 1980s onwards,
al., 2006;; Campbell and Balbach, in light of an apparent reluctance
2008, 2009;; Balbach and Campbell, Media campaigns by state legislators to pass such
2009;; Raebeck et al., 2009). legislation), one study reported that
Groups that were successfully Nine of the studies found evidence the industry legally challenged the
recruited to argue against tobacco of the tobacco industry using mass draft proposal on the basis that it
tax increases included the Coalition media and other publicity campaigns violated the state constitution, filed
of Labour Union Women (CLUW) to raise public awareness of policy subsequent legal challenges at
(Balbach et al., 2006;; Balbach and proposals and create/increase public later stages and also tried to use
Campbell, 2009), Citizens for Tax support for the industry positions a signatures expert to disqualify
Justice (Campbell and Balbach, (Koh, 1996;; Traynor and Glantz, signatures supporting the ballot
2009), Citizen Action (Campbell and 1996;; Heiser and Begay, 1997;; proposal (Heiser and Begay, 1997).
Balbach, 2009), Women Involved in Goldman and Glantz, 1999;; Breton As a result of this latter tactic, several
Farm Economics (Givel and Glantz, et al., 2006;; Givel, 2007;; Campbell studies reported that the supporters
2001), the National Taxpayers and Balbach, 2008;; Campbell and of later policy proposals insured
Conference (Givel and Glantz, 2001), Balbach, 2009;; Raebeck et al., themselves against this by collecting
the Congressional Black Caucus 2009). One study reported that the substantially more signatures than
(Yerger and Malone, 2002) and the most effective media campaign the law required (Traynor and Glantz,
National Black Police Association undertaken by industry groups 1996).
(Yerger and Malone, 2002). The involved an advert featuring an
industry also used claims that tobacco alleged undercover police officer Commissioning supportive/
tax increases would lead to increases discussing the crime implications of informative research
in smuggling to ensure the support of the proposed tobacco tax increases
business groups such as retailers’ (more time would be spent on Six of the studies reported evidence
associations (Moon et al., 1993;; smuggled cigarettes, hence less time of the industry commissioning
Breton et al., 2006) and police groups on other issues, and the increased studies to support and inform
(Traynor and Glantz, 1996;; Breton et criminal money from smuggling, industry positions on tobacco excise
al., 2006). Finally, in seeking to divert etc. would be spent on drugs and debates (Moon et al., 1993;; Yerger
earmarked funds away from tobacco guns) (Traynor and Glantz, 1996). and Malone, 2002;; Balbach et al.,
control programmes (see below), the However, it was eventually revealed 2006;; Breton et al., 2006;; Lum et
industry was able to ally itself with that the “undercover policeman” had al., 2009;; Raebeck et al., 2009).
private healthcare organizations and nothing more than a desk job in the This included market-research to
77
assess: (i) whether it was worth subsidies for uninsured groups (Begay increase which the industry managed
challenging a given proposal, and et al., 1993;; Koh, 1996;; Traynor and to defeat, the industry linked the
(ii) how best to challenge particular Glantz, 1996;; Heiser and Begay, 1997;; proposed tobacco tax increase with
proposals. For example, one study Bialous and Glantz, 1999;; Goldman proposals for property tax increases
found the industry was undertaking and Glantz, 1999;; Balbach et al., in rural areas that were happening
telephone research to assess the 2000). This approach not only helped simultaneously (Moon et al., 1993).
possible impact of various campaign the industry to limit state tobacco
themes (Moon et al., 1993). A review control activities that had potential Proposing alternative legislation
of state-level efforts to achieve to further reduce consumption, but
earmarked tobacco excise increases also provided evidence to support the Four studies found evidence of
via public ballots in the USA found industry argument that the funds are industry efforts to promote alternative
the industry had increasingly begun likely to be used in ways other than (weaker or irrelevant) proposals/public
to undertake a significant amount of those stated and/or which the public ballots to distract attention and energy
research before deciding whether to did not support (see above). from the original proposal (Koh, 1996;;
invest significantly in efforts to defeat Traynor and Glantz, 1996;; Heiser
the proposed legislation (Lum et al., Paying or providing gifts to policy- and Begay, 1997;; Goldman and
2009). In addition, the industry paid makers Glantz, 1999). One study (Goldman
for studies that supported some of and Glantz, 1999) found that while
the arguments outlined in the above Several of the studies found evidence signatures were being collected to
section, such as a series of studies by of the industry paying or giving gifts qualify the proposal for a ballot the
labour groups and political caucuses, to officials and political parties or tobacco industry began contributing
including CLUW, which highlighted campaigns to help attract political money to a different ballot measure
the regressive nature of tobacco support for the industry’s positions on campaign in order “to disrupt the
taxes (Yerger and Malone, 2002;; taxation (Begay et al., 1993;; Balbach signature-gathering process”, for
Balbach et al., 2006;; Raebeck et al., et al., 2000;; Yerger and Malone, the Oregon-based proposal (known
2009). The Tobacco Institute also 2002;; Givel, 2006). Key officials most as Measure 44), and “to divert the
produced some economic studies to likely to be able to influence decisions energy and financial resources of the
support its claims about taxes (Yerger about tobacco taxation were targeted health insurers”. Likewise, a study
and Malone, 2002). (Begay et al., 1993;; Balbach et al., of a similar proposal in California
2000), and funds were often provided reported that an industry-funded
Employing consultants via third parties, such as advertising group circulated its own petition at the
and public relations staff/firms agencies, so it was not obvious that same time as signatures were being
they originated from the industry gathered in support of the earmarked
Several studies demonstrate that, (Balbach et al., 2000). tax proposal as a way of reducing
to gain advice on and assistance the pool of available paid signature-
with lobbying, the tobacco industry Confusing debates about tobacco tax gatherers (by paying them more) and
has used a range of consultancy increases with broader tax debates confusing voters about the official
and public relations firms and other proposal (Traynor and Glantz, 1996).
lobbyists (Begay et al., 1993;; Koh, Four studies reported industry efforts Two other studies of similar state-level
1996;; Traynor and Glantz, 1996;; to frame debates about tobacco policy proposals found the industry
Bialous and Glantz, 1999;; Givel, excise increases within broader put forward compromise proposals
2006, 2007;; Raebeck et al., 2009). debates about general tax increases that involved smaller tax increases
to confuse the issue and to help and fewer funds being allocated to
Working to divert earmarked funds garner opposition to proposals for tobacco control programmes (Koh,
tobacco tax increases (Moon et al., 1996;; Heiser and Begay, 1997).
Some of the studies found that where 1993;; Balbach et al., 2006;; Lum et
earmarked tobacco excise increases al., 2009;; Raebeck et al., 2009). This Using friendly “experts”
were being pursued, the tobacco tactic was apparently supported by the
industry had worked to divert funds industry’s own market research (Lum Three studies found evidence of
away from tobacco control measures et al., 2009). In one of the few studies the industry using friendly experts
to other causes such as healthcare focusing on a policy proposal for a tax to present industry positions and
78
bolster the credibility of industry have been relatively successful, has in completely preventing excise
arguments (Traynor and Glantz, with the Canadian government increases (Begay et al., 1993;; Koh,
1996;; Goldman and Glantz, 1999;; responding “to the influx of smuggled 1996;; Traynor and Glantz, 1996;;
Alamar and Glantz, 2004). Examples tobacco that culminated after the Heiser and Begay, 1997;; Balbach
included Wall Street analysts who tax increases by, in 1994, returning et al., 2000;; Lum et al., 2009). The
had been directly briefed by the taxes to their original level” (Kelton industry’s success in doing this
tobacco industry before presenting and Givel, 2008). The study goes on in turn provides support for one
their opinions to officials but who did to note that although the intention of the industry’s most frequently
not disclose this (Alamar and Glantz, of the tax reduction was to alleviate used arguments used to counter
2004), a high-profile dentist (Goldman the growing illicit trade in tobacco future proposals for earmarked tax
and Glantz, 1999) and an alleged products, it did not succeed because increases, namely that revenue from
undercover police officer (who was ‘the illegal activity was more than just earmarked taxes will be diverted or
later found to be a desk officer and a result of high taxation’ (Kelton and misused. Indeed, part of the reason
part-time actor, as described above) Givel, 2008). this argument has been so effective is
(Traynor and Glantz, 1996). that there is a great deal of evidence
to support this claim;; in several of the
Trying to undermine tobacco control The success of tobacco industry US states that introduced earmarked
experts efforts to keep tobacco taxes low tax increases, the revenue was
and prevent earmarking used for purposes other than those
One study reported that the tobacco described in the original proposal. In
industry had wrongly targeted and In the USA, the industry has had some cases, the industry was found
tried to undermine the credibility of a some success in preventing proposed to have played a key role in helping
key tobacco control academic (Givel, earmarked excise increases (Moon to ensure funds were diverted, which
2007). This involved alleging (but not et al., 1993;; Lum et al., 2009), helped the industry from the point of
proving) that University of California– although many such proposals did view of limiting spending on tobacco
San Francisco Professor of Medicine pass (Begay et al., 1993;; Koh, 1996;; control activities and providing
Stanton Glantz had misused grant Traynor and Glantz, 1996;; Heiser and evidence to shore up this argument
funds for illegal political purposes Begay, 1997;; Bialous and Glantz, in future debates. However, the
and lobbying. 1999;; Goldman and Glantz, 1999;; industry alone cannot be blamed for
Balbach et al., 2000;; Lum et al., the success of this argument;; all of
Stimulating smuggling in the event of 2009). Key contextual factors that the studies that highlighted this issue
tax increases appear to have assisted passage found that a range of other actors,
and implementation of earmarked notably healthcare providers, worked
One of the two studies of tobacco tobacco tax increases include: (i) to divert funds away from tobacco
excise increases in Canada in coherent and adequately funded control programmes. As Bialous
the early 1990s, which were later campaigns by coalitions favouring and Glantz (1999) point out, this
revoked, claims that the industry the tobacco tax increase (in many highlights the importance of focusing
itself helped promote smuggling (via of the studies included here, these on the longer-term implementation
Native American reservations, which campaigns were partially funded of tobacco control policies as well
are not subject to the same taxation by voluntary health groups, such as as on their development and official
laws) during the period in which the the American Cancer Society, and acceptance into law.
tax increases were in place (Kelton healthcare organizations);; and (ii) However, when assessing the
and Givel, 2008). The authors claim budgetary deficits, which worked success of the tobacco industry in
the industry did this both because it to increase legislative support for preventing the numerous state-level
enabled them to maintain or increase proposals. proposals for earmarked tax increases
profit margins despite the tax-related Overall, the studies included in that emerged in the USA from the late
price increases and because it this review suggest that the industry 1980s onwards, it should be noted
provided support for the industry’s has been more successful in that there appears to have been a
claim that the tax increases led to a diverting earmarked funds away from research and publication bias towards
rise in illicit trade (Kelton and Givel, tobacco control activities, usually proposals that succeeded. Hence,
2008). This approach appears to at the implementation stage, than it this review includes seven studies of
79
successful proposals (three studies In contrast to the studies that the other studies in this section
of Proposition 99 in California, two concerning USA state-level efforts of the review reveal almost nothing
of Question 1 in Massachusetts, to increase and earmark tobacco about, because very few comment
and one each of Proposition 200 in taxes, most of the studies on industry on the extent to which different
Arizona and Measure 44 in Oregon) efforts to prevent national/federal tobacco companies worked together
and only one of a similar initiative in level tax increases, or to secure tax to influence policy. This company
Montana which was unsuccessful. decreases at this level, found the differentiation, however, features
This is a ratio of seven studies of four industry was successful. Studies somewhat more in the studies
successful proposals compared to report the industry was successful exploring industry efforts to influence
only one study of one unsuccessful in defeating both the McCain Bill tobacco excise structures (see
proposal. Yet, Lum and colleagues (Alamar and Glantz, 2004) and the following section).
(2009) report that between 1988 and Clinton administration’s healthcare The remaining studies concerning
2008 the industry attempted to defeat proposals in the USA (Campbell tobacco industry efforts to prevent/
14 out of 22 state-level earmarked and Balbach, 2009), although the undermine tobacco excise increases
tax proposals and was successful tobacco industry was clearly not the all focus on assessing how the
in 8 cases, giving a 57% success only interested party involved in the industry worked to persuade credible
rate (8 out of 14 attempts) where it defeat of these bills. The industry was interest groups to support the industry
attempted to challenge and a 36% also found to have helped reduce position on tobacco taxes (Yerger
rate at blocking state-level proposals the levels of federal tobacco tax and Malone, 2002;; Levenstein et al.,
for tobacco tax increases overall (8 increases implemented in 1990 and 2005;; Balbach et al., 2006;; Campbell
out of 22 proposals). 1993 (Campbell and Balbach, 2008). and Balbach, 2008, 2009;; Balbach
In terms of assessing which Studies also report the industry was and Campbell, 2009;; Raebeck et
arguments and tactics were most successful in achieving tax decreases al., 2009). These studies, which all
successful in defeating these in Canada (Joossens & Raw, 2003;; focus on the USA, report that the
state-level tax increases, Lum and Breton et al., 2006;; Kelton and Givel, industry was generally successful
colleagues (2009) claim that the 2008), Sweden (Joossens and Raw, in recruiting credible, existing public
amount of resources invested in 2003), Lebanon (Nakkash, 2007) and interest groups with an interest in
lobbying by the industry was a poor some former Soviet Union countries the issue between the 1980s and
indicator of success. Instead, the (Gilmore et al., 2007). In all of these early 2000s. In most cases, these
authors argue that industry success successful lobbying efforts, the groups already had at least some
was associated with the extent studies report that a key argument commitment to lobbying against
to which it was guided by market employed by the industry involved regressive taxation, although most
research (the authors report that after claiming that tax increases would fuel were not actively lobbying on tobacco
1998, the industry made increasing or were already fuelling illicit trade. tax issues before being encouraged
use of voter segmentation models and This argument was also identified by to do so by tobacco industry groups.
market research to decide whether one of the USA state-level studies as Recruiting the support of such
state-level excise proposals were being particularly effective (Traynor organizations was achieved by
worth challenging;; once the industry and Glantz, 1996). taking three main approaches: (1)
took this approach, its success rate However, the illicit trade argument emphasizing arguments concerning
increased). The same study suggests was also used in the only example of the regressive nature of tobacco
that the industry fared better when a study reporting that the industry did taxes, to demonstrate why this was
it combined arguments concerning not succeed in preventing/limiting tax an issue of interest to labour and civil
the diversion of funds to health care increases at the national level. This rights organizations;; (2) approaching
providers with arguments relating study involved industry lobbying in the groups indirectly via a union
to the unfairness of the proposals Hungary in the 1990s (Szilágyi and organization that, despite having
on smokers, pointing out that “when Chapman, 2003);; in this example, been established and funded by the
either of the t wo arguments was used Philip Morris’ efforts to prevent an Tobacco Institute, was not obviously a
alone, the tobacco industry lost three excise increase appear to have failed tobacco industry organization;; and (3)
out of four elections, but when they because of the “divided position of offering to support targeted groups in
were combined, they won three out the industry” (Szilágyi and Chapman, a range of ways, including on issues
of four elections” (Lum et al., 2009). 2003). This is an interesting factor that the industry had no interest in,
80
to develop the relationship. All of 2007) to protect a dominant market McKee 2004b.;; Gilmore et al., 2007).
the studies suggest the industry’s position. In contrast, when BAT and In this context, officials are therefore
success in recruiting these kinds of Philip Morris had not yet achieved a likely to have been far more open to
groups in the USA is likely to have significant market share in a country, tobacco company arguments than
aided efforts to prevent tax increases they both lobbied for lower import might otherwise have been the case.
because it provided far greater duties/fewer restrictions (O’Sullivan It is unclear to what extent the
credibility to arguments concerning and Chapman, 2000;; Szilágyi and approaches taken to lobbying on this
the unfair and regressive nature of Chapman, 2003;; Nakkash, 2007). issue were informed by the fact that
tobacco excise taxes. This allowed As far as it is possible to discern most of the countries involved were
the industry to claim that one of its any patterns from this limited data, at that time undergoing a process of
key arguments was widely supported, arguments that different tobacco political and economic transition and/
and it also facilitated the lobbying of companies use to promote their or were undemocratic (O’Sullivan
officials and politicians who would not preferred structures seem to focus on and Chapman, 2000;; Szilágyi and
normally be expected to side with the claims that their particular approach Chapman, 2003;; Gilmore and
industry (e.g. liberal Democrats). will increase government revenue McKee, 2004b;; Gilmore et al., 2007;;
(Gilmore and McKee, 2004b;; Delnevo Nakkash, 2007). More research is
Studies concerned with tobacco et al., 2007;; Gilmore et al., 2007;; required to better understand what
industry efforts to influence Nakkash, 2007) and reduce illicit different parts of the tobacco industry
tobacco excise structures trade (Gilmore and McKee, 2004b;; aim to achieve in relation to tobacco
Gilmore et al., 2007;; Nakkash, 2007). excise structures and how they lobby
From the limited available evidence The lobbying on this issue that was on this issue beyond the countries
concerning tobacco industry revealed by these studies tended to be that have been studied.
efforts to influence tobacco excise behind-the-scenes, mainly involving
structures, we know that different direct lobbying of relevant officials, Discussion and conclusions
companies favour different structures although this was sometimes directed
in different contexts. From the studies via third parties such as consultants This systematic review suggests that
included in this review, it appears (Gilmore and McKee, 2004b;; Gilmore the different parts of the tobacco
Philip Morris (which produces the et al., 2007). Two studies suggest industry tend uniformly to lobby
high-end Marlboro cigarettes) tends particular companies may have also against significant tobacco excise
to favour specific taxes (Szilágyi and actively sought support for their increases and, in particular, against
Chapman, 2003;; Nakkash, 2007), preferred excise structures from tobacco excise increases that are
whereas British American Tobacco international financial organizations to be earmarked for tobacco control
(which, historically at least, had a such as the IMF (Nakkash, 2007) activities. In challenging such
far more diverse brand portfolio and American trade organizations proposals, the arguments employed
including mid-price and cheaper, (O’Sullivan and Chapman, 2000). In most often and most successfully by
local brands) tends to support mixed former Soviet Union countries, BAT the industry involve claims that tax
excise structures, incorporating an appears to have been able to create increases are socially regressive,
ad valorem element (Szilágyi and a symbiotic relationship between and that they increase illicit tobacco
Chapman, 2003;; Nakkash, 2007;; itself and finance ministries, in which trade and therefore contribute to
Gilmore et al., 2007). One reason finance ministers perceived the broader crime problems. Yet these
BAT may prefer a mixed structure is company as a source of expert advice and the other most frequently used
because of the competitive advantage on excise issues (Gilmore and McKee, arguments (e.g. that tobacco taxes
it confers over Philip Morris (Gilmore 2004b;; Gilmore et al., 2007). This is are unfair on smokers and that tax
et al., 2007;; Nakkash, 2007). likely to have been made far easier by rises do not lead to reductions in
Other attempts to influence excise the fact that officials in former Soviet consumption) all have flaws, as
structures in ways which would give Union countries were, at that time, outlined above and explained further
one company an advantage over unused to regulating and dealing with in other chapters. The most effective
others included promoting the use free-market corporations and had argument used against earmarked
of tax stamps (Gilmore et al., 2005;; little if any experience of dealing with tax increases has been to claim that
Gilmore et al., 2007) and lobbying tobacco excise, which had previously the revenue from earmarked taxes
for high import duties (Gilmore et al., been dealt with centrally (Gilmore and will be diverted or misused, and
81
because this has occurred, in part as on tax-related lobbying is needed mass media campaigns (Axeland et
a result of tobacco industry efforts, in other geographical areas and in al., 1994), legal challenges (Neuman
such arguments carry weight. particular to better understand what et al., 2002), the commissioning and
Favoured industry tactics to different parts of the tobacco industry use of supportive research (Barnes
influence policy decisions about aim to achieve in relation to tobacco and Bero, 1996;; Grüning et al., 2006),
tobacco excise increases include excise structures. the use of professional consultants
traditional lobbying of relevant One argument that anecdotal and public relations services (Ong
officials (direct and indirect), allying evidence reports the industry and Glantz, 2001, Smith et al., 2010),
with credible political groups that employing to counter tax increases paying or providing gifts to policy-
have (or can be persuaded to (Chaloupka et al., 2001;; Kyriss et al. makers (Saloojee and Dagli, 2000),
have) a shared interest in tobacco 2008) but which was not highlighted proposing alternative (weaker)
taxes (particularly on the basis of in the included studies is that higher legislation (Neuman et al., 2002),
the perceived regressive nature tax rates will lead to a decline in and using friendly “experts” (Grüning
of tobacco tax increases), and consumption and thereby contribute et al., 2006).
establishing industry “ front groups” to to job losses in the tobacco sector, There was also no evidence
lobby against tobacco tax increases. reducing the industry’s contribution of tobacco industry support for
From the limited available to the economy. It is possible that this incremental tax increases (other than
evidence concerning tobacco argument was used by the industry in relation to support for alternative,
industry efforts to influence tobacco in some of the policy developments weaker proposals than those on
excise structures (seven papers in covered by this review but did not offer). However, as studies of
seven different countries/regions), feature in the studies. Alternatively, tobacco company pricing strategies
we observe that different companies it may not have been employed suggest that tobacco companies may
favour different tax structures, either because: (i) it undermined use tax increases as opportunities
with Philip Morris tending to prefer another argument which was used, to increase their own profit margins
100% specific structures and BAT that tax increases would not reduce (Van Walbeek, 2000), it should not
tending to prefer mixed structures consumption;; or (ii) the industry be assumed that the industry will
(predominantly specific but also deemed it less persuasive in the always lobby against tax increases,
including a substantial ad valorem context of growing evidence that in only that it is likely to do so where
element). As far as it is possible to all but the most tobacco dependent such increases are significant and/or
discern any patterns from this limited economies, the tobacco industry earmarked.
data, tobacco companies appear does not make a net positive
to promote their preferred structure economic contribution (Jha and Policies to limit industry price
by claiming that it will increase Chaloupka, 1999;; Gruber & Koszegi, manipulation
government revenue and reduce 2008) (for further discussion see
illicit trade. Chapter 9). Article 6 of the WHO Framework
As highlighted earlier, there is a Most of the tactics the industry Convention on Tobacco Control
significant geographical bias in the used in its efforts both to counter (FCTC) calls on Parties to adopt
studies that met the inclusion criteria (planned and actual) tobacco tax and maintain tax and price policies
for this review, with far more evidence increases and to try to influence tax that will contribute to the health
being available in relation to North structures mirror those reported by objectives. To achieve this, this
America than anywhere else. Given studies of tobacco industry efforts chapter illustrates that governments
that North America is one of the to influence policy more generally. need to understand and address
world’s most economically developed For example, the industry’s use of industry pricing and price-related
and wealthiest regions, and that front groups (Apollonio and Bero, marketing strategies. The example
tobacco excise rates are lower there 2007;; Smith et al., 2010) and third- of the USA indicates that collecting
than in other high-income countries party allies (Ong and Glantz, 2001;; data (ideally monthly to track industry
(World Health Organization, 2010), Neuman et al., 2002;; Smith et al., responses to policy interventions) on
it should not be assumed that the 2010) have both been widely reported, industry marketing expenditure by
industry has (or will) necessarily take as have more traditional lobbying category (thus including price-based
the same approaches elsewhere. It is approaches focusing on targeting promotions) would be a useful first
clear, therefore, that further research officials (Barnoya and Glantz, 2002), step and one that would be in line with
82
Article 5.3 of the FCTC (see Figure Figure 3.5. Extract from the Guidelines for implementation of Article 5.3,
3.5). Furthermore, governments WHO-FCTC
also require detailed brand or price-
category specific data on tobacco “Parties should require the tobacco industry and those working to further its interests
pricing to determine effective to periodically submit information on tobacco production, manufacture, market share,
market expenditure and any other activity, including lobbying, philanthropy, political
policies, and such data should be contributions and all other activities not prohibited or not yet prohibited under Article 13 of
made available to researchers to the convention” (Recommendation 5.2). “Parties should require rules for the disclosure or
identify the empirical research deficit registration of the tobacco industry activities, affiliated organizations and individual acting
on their behalf including lobbyists” (Recommendation 5.3).
identified in this chapter.
Beyond data collection and
research, the chapter illustrates related promotion—price-reducing Price regulation and minimum
the need to develop and implement promotions (e.g. voucher schemes), pricing
effective strategies to prevent price- price-based promotions (e.g. price-
related marketing and industry marked packaging) and retailer The broader economic literature
pricing schemes from undermining incentive schemes. details two reasons for using
tax and other tobacco control efforts. In this respect it is worth noting that minimum prices: to facilitate
Potential actions fall into two broad while bans on all in-store promotions, collusion (price fixing) between firms
categories: (1) bans or restrictions including placing cigarettes below the by preventing price wars and to help
on price-related promotions and (2) counter, would limit the opportunity maintain profit margins thus providing
regulation of tobacco prices including for price-based promotions to be sufficient funds for marketing. In line
the setting of minimum retail prices communicated to the consumer (and with this, most minimum pricing rules
on tobacco products. This section thus likely limit the effectiveness of on tobacco have been implemented
reviews the (very limited) available retailer incentive schemes), as long to protect businesses rather than
literature exploring each of these as packaging remains unregulated, health. Yet minimum prices may also
policy approaches. Then in light of prices could be communicated to be used for public health purposes—
the evidence presented above on consumers in this way as recently to prevent low-cost selling from
industry efforts to influence and documented (Moodie & Hastings, existing firms and to prevent new
undermine taxation policies, it turns 2009). In other words, preventing all firms coming into the market with
to explore interventions to reduce price-based marketing would require lower-priced offers—and several
industry influence on policy in the form bans on point-of-sale marketing countries have now implemented
of Article 5.3 of the WHO Framework and standardised packaging in line such rules for this purpose. Yet, as
Convention on Tobacco Control. with Article 13 of the WHO FCTC we explore below using a recent
guidelines. example from the EU, such rules
Restrict or ban price-related Although, as outlined in section may find themselves in contravention
promotional activities (b), a large number of countries of trade treaty rules (including the EU
have now implemented some form Treaty) on non-discrimination (see
Comprehensive bans on price-related of control on price-related marketing also Chapter 2).
and price-reducing promotions and we only found one (US-based) Minimum pricing rules on
retailer incentive schemes would, study that specifically evaluated tobacco now exist in a variety of
if enforced, inevitably help prevent the impact of restrictions on price- jurisdictions. These include half of
these forms industry promotion. The based marketing, and this was limited US states (where they originated
evidence presented in this chapter enough in its focus and design to between the 1940s and 1960s and
would suggest, in line with broader preclude inclusion (Feighery et al., were meant to protect businesses
evidence that the industry simply 2009). It is important to note, however, and not public health) (Michael,
shifts its marketing expenditure from that there is good evidence for the 2000), a small number of EU Member
one area to another when faced effectiveness of comprehensive States1 (where the rationale again
with marketing restrictions (Saffer, marketing restrictions on reducing was largely to protect retailers),
2000), that such schemes would consumption (Blecher, 2008). and Malaysia where they were
need to cover every type of price- implemented in January 2010 (for
1
However, following a recent ECJ ruling, these minimum pricing rules are now or have already been ended
83
public health purposes, to reduce In the USA advocates of minimum price-discounts (F. Van Driessche,
cigarette purchases by children prices argue that minimum prices personal communication).
and low-income smokers) along are based on percentage mark ups Where an EU Member State
with prohibition of words that mean on wholesale prices (in the EU, most chooses to introduce minimum
discount or cheap sale. The US minimum prices are a percentage (e.g. prices, it must comply with its
minimum pricing rules are limited, 95%) of the weighted average price) obligations under the EU Treaties,
however, as most allow cigarette and, therefore minimum prices will more particularly with the rules on
company promotional incentives rise as the manufacturer’s price rise the free movement of goods (Articles
offered to retailers, such as buy- over time (thereby, keeping up with 34 and 36 of the Treaty on the
downs and Retail Leaders Programs, inflation). In contrast, excise taxes, Functioning of the European Union),
to be added into the formula that sets when wholly specific (e.g. cigarette the rules on competition (Articles
the minimum price. excises in the USA) lose their value 101 TFEU in conjunction with Article
To evaluate the effect of minimum as inflation occurs. However, this is 4, paragraph 3 of the Treaty of the
pricing on tobacco control, two not a problem when specific taxes European Union) and relevant case-
questions must be answered: (i) do are inflation-adjusted on a regular law of the Courts of the European
minimum price laws raise cigarette basis or when ad valorem taxes are Union. However, national rules
prices and prevent cut-price offers and imposed in addition (as is required fixing retail prices may be justified
(ii) are they effective in the presence in the EU). Opponents of minimum on grounds of, inter alia, protection
of promotional incentives? The first pricing argue that excise taxes of human health, provided that they
question looks at the effectiveness generate government revenues while do not constitute means of arbitrary
of minimum pricing in general, but prices raised through minimum price discrimination or a disguised
we found no studies addressing laws benefit the wholesaler and/or restriction on trade between Member
this topic. The second examines the retailer. That is, minimum prices States (Article 36 TFEU) and that
effectiveness when minimum price benefit tobacco sellers rather than they are proportionate to the aim
laws allow promotional incentives governments and publicly-funded pursued. Proportionality requires a
to be deducted from formulas for programmes. This is an issue that measure to be necessary to achieve
establishing the minimum cigarette deserves more attention, as minimum the objective, and that the objective
price (as is the case in most US pricing policies may leave the cannot be achieved by any less
states). The only study on this topic industry with more money to invest trade-restrictive means, for example,
(Feighery et al., 2005) suggests that in marketing research and innovative whether the same public health aim
when such incentives are included, ways to attract more customers. could not be equally ensured by
minimum pricing is not effective;; In the EU Member States that increasing excise duties. The EU
net cigarette prices did not differ implemented minimum prices directive that covers the structure
between stores in eight states with on cigarettes (Austria, France, of tobacco taxation (95/59/EC) also
minimum price laws and seven states Italy, Belgium and Ireland), their stipulates that manufacturers or
without, a finding likely explained by implementation had been advocated importers of tobacco “shall be free to
the fact that nearly all the minimum by the Ministries of Finance (because determine the maximum retail selling
price laws in the study allowed of their high ad valorem excise duties price” for their products.
promotional incentives. Separate which create greater price ranges) It is perhaps unsurprising
comparisons with New York, the as well as by the manufacturers of therefore that the European Court
one state which did not allow these premium brands (e.g. PMI). Moreover, of Justice found that the setting
incentives, showed that cigarette the retail sector supported the of a minimum retail selling price
prices were significantly higher in measures as their margin is, in most for tobacco in Austria, France and
New York than in the 24 other states, Member States, a fixed percentage of Ireland was illegal under the EU
although the number of New York the maximum retail price determined Treaties2 as it limited the freedom
stores participating in the study was by the manufacturer or importer. of producers and importers to
small, and the study did not control Only in Ireland were minimum prices determine their maximum retail
for differences in cost of living across an initiative of the Ministry of Health selling prices, distorted competition
states. to prevent the industry offering and simply benefited cigarette
2
Judgment of the Court (Third Chamber), 4 March, 2010, in cases C-197/08 Commission v France, C-198/08 Commission v Austria and C- 221/08 Commission v Ireland.
84
manufacturers by safeguarding their A key concern the European Tobacco Control agreed upon Article
profit margins. However, the Court Commission had with minimum 5.3, which requires that “in setting
stated that the Member States could pricing was that it distorted and implementing their public health
prohibit the sale of manufactured competition and benefitted tobacco policies with respect to tobacco
tobacco products at a price below the manufacturers by further increasing control, Parties shall act to protect
sum of the cost price and all taxes their profits. Yet other forms of price- those policies from commercial and
(i.e. they could prevent producers or regulation are allowed under the EU other vested interests of the tobacco
importers from selling them at a loss treaty and widely used, for example, industry in accordance with national
by absorbing, even temporarily, the in the utilities sector. A proposal for law.” If properly implemented, Article
impact of taxes on the retail selling price regulation in the tobacco sector 5.3 has the potential to reduce
price of manufactured tobacco (Gilmore et al., 2010) has recently industry influence on policy and thus
products). Belgium has removed its been made that would be compatible to prevent the inappropriate influence
minimum pricing rule and a further with EU treaty rules and would, in on excise policy documented above.
case is pending against Italy. theory at least, address many of the However, implementation and
The above-mentioned cases key problems outlined in this chapter. enforcement will be difficult and
were based on the application of An alternative might be a windfall tax require considerable political will.
Article 9 of EU Directive 95/59/EC on profits with a provision to prevent The industry has already been using
which relates to the price determined this being passed onto consumers. complex arguments to justify its
by the manufacturer or the importer. A surcharge on tobacco industry inclusion in the policy process (Smith
In other judgments (on minimum profits was implemented in Canada et al., 2009) and it is particularly likely
prices for alcohol), the court has in 1994 to recoup losses from to do so in relation to taxation policy
ruled that national rules fixing retail industry involvement in smuggling. on the basis that can be considered
prices could constitute measures Although this surcharge was passed primarily an economic rather than
having an equivalent effect to onto customers, its implementation health policy. It is therefore also
quantitative restrictions on imports illustrates that such interventions are essential that policymakers and the
(Article 34). This would be the case feasible. The problems with industry public are made aware of the flawed
if imported products were placed at a pricing outlined in this chapter nature of the arguments the tobacco
disadvantage in relation to domestic suggest that such proposals should industry uses against excise policies
products, for example, because be given serious consideration. (as outlined in this chapter). Finally,
competitive advantage conferred by ensuring that earmarked taxes, once
lower cost prices were eliminated. Preventing industry lobbying passed, are properly implemented
However this implies that minimum (i.e. that money is not diverted from
prices are not per definition against In light of overwhelming evidence the proposed policy areas) will
the Treaty. of the tobacco industry’s efforts to eliminate one argument from the
undermine public policy, the parties to industry’s armament.
the WHO Framework Convention on
85
References
Advisory Group of the Royal College of Batuke (1992) Representative Offices Chaloupka FJ, Peck RM, Tauras JA, et al.
Physicians (2000). Nicotine Addiction in Britain. Company British American Tobacco: British (2010). Cigarette excise taxation: The impact of
American Tobacco. tax structure on prices, revenues, and cigarette
Alamar BC, Glantz SA (2004). The tobacco smoking . NBER Working Paper No. 16287
industry’s use of Wall Street analysts in shaping Becker GS, Grossman M, Murphy KM (1994). August 2010 JEL No. H2,I18
policy. Tob Control, 13:223–227.doi:10.1136/ An empirical analysis of cigarette addiction.
tc.2003.006643 PMID:15333876 Am Econ Rev, 84:396–418. Chaloupka FJ (2007). Cigarettes: old firms
facing new challenges. In: Tremblay VJ,
Apollonio DE, Bero LA (2007). The creation Begay ME, Traynor M, Glantz SA (1993). The Tremblay CH, eds., Industry and firm studies,
of industry front groups: the tobacco industry tobacco industry, state politics, and tobacco 4th Edition. New York, ME Sharpe: 80–118.
and “get government off our back”. Am J education in California. Am J Public Health,
Public Health, 97:419–427.doi:10.2105/ 83:1214 –1221.doi:10.2105/AJPH.83.9.1214 Chaloupka FJ, Cummings KM, Morley CP,
AJPH.2005.081117 PMID:17267719 PMID:8362994 Horan JK (2002). Tax, price and cigarette
smoking: evidence from the tobacco
Ashenfelter O, Sullivan D (1987). Nonparametric Bialous SA, Glantz SA (1999). Arizona’s documents and implications for tobacco
tests of market structure: an application to the tobacco control initiative illustrates the need company marketing strategies. Tob Control, 11
cigarette industry. J Ind Econ, 35:483–498 for continuing oversight by tobacco control Suppl 1;;I62–I72.doi:10.1136/tc.11.suppl_1.i62
doi:10.2307/2098584. advocates. Tob Control, 8:141–151.doi:10.1136/ PMID:11893816
tc.8.2.141 PMID:10478397
Axeland R, Bayard SP, Jinot J (1994). Setting Chaloupka FJ, Wakefield M, Czart C (2001).
the record straight: secondhand smoke is a Bishop JA, Yoo JH (1985). “Healh scare”, Taxing tobacco: the impact of tobacco taxes
preventable health risk. Tob Control, 3:263– excise taxes and advertising ban in the on cigarette smoking and other tobacco use.
267 doi:10.1136/tc.3.3.263. cigarette demand and supply. South Econ J, In: Rabin RL, Sugarman SD, eds., Regulating
52:402–411 doi:10.2307/1059626. Tobacco. Oxford, Oxford University Press:
Balbach ED, Herzberg A, Barbeau EM (2006). 39–71.
Political coalitions and working women: Blecher E (2008). The impact of tobacco
how the tobacco industry built a relationship advertising bans on consumption in Chaloupka FJ (2004). Testimony. Civil action
with the Coalition of Labor Union Women. developing countries. J Health Econ, 27:930– n°99-CV-02496 (6K). United States of America
J Epidemiol Community Health, 60 Suppl 9 42 .d o i:10.1016 / j. j h e a l e c o. 2 0 0 8 . 0 2 . 010
2 ;; 2 7– 3 2 . d o i:10 .113 6 / j e c h . 2 0 0 6 . 0 4 6 2 76 PMID:18440661 Cnossen S (2006). Tobacco taxation in the
PMID:17708008 European Union. Paper Series N°1718.
Breton E, Richard L, Gagnon F et al. (2006).
Balbach ED, Traynor MP, Glantz SA (2000). Fighting a tobacco-tax rollback: a political Creighton FV (1986). “Camel growth among
The implementation of California’s tobacco tax analysis of the 1994 cigarette c ontraband crisis males 18-24 years old in the mid-west”. R.J.
initiative: the critical role of outsider strategies in in Canada. J Public Health Policy, 27:77–99. Reynolds memorandum.
protecting Proposition 99. J Health Polit Policy PMID:16681189
Law, 25:689–715.doi:10.1215/03616878-25-4- Delipalla S, O’Donnell O (2001). Estimating
689 PMID:10979517 Burgess GJ (1993) Andean Pact Company tax incidence, market power and market
British American Tobacco: British American conduct: The European cigarette industry. Int
Balbach ED, Campbell RB (2009). Union Tobacco. J Ind Organ, 19:885–908 doi:10.1016/S0167-
women, the tobacco industry, and excise taxes: 7187(99)00057-0.
a lesson in unintended consequences. Am J Burrows DS (1982a) “NBER Models of
Prev Med, 37 Suppl;;S121–S125.doi:10.1016/j. Price Sensitivity by Age/Sex.” RJ Reynolds Delipalla S, Keen M (1992). The Comparison
amepre.2009.05.011 PMID:19591750 memorandum. between ad valorem and specific taxation under
imperfect competition. J Public Econ, 49:351–
Barnes DE, Bero LA (1996). Industry-funded Burrows DS (1982b) “Marketing Implications of 367 doi:10.1016/0047-2727(92)90073-O.
research and conflict of interest: an analysis the NBER Models”. RJ Reynolds Memorandum.
of research sponsored by the tobacco industry Delnevo C, Lewis MJ, Foulds J (2007). Taxing
through the Center for Indoor Air Research. J Burrows DS (1984) Younger adult smoker: moist snuff by weight ain’t worth spit. Tob
Health Polit Policy Law, 21(3):515-542. strategies and opportunities. R.J. Reynolds Control, 16:69.doi:10.1136/tc.2006.018127
Tobacco Company. PMID:17297079
Barnett PG, Keeler TE, Hu TW (1995).
Oligopoly structure and the incidence of Campbell R, Balbach ED (2008). Mobilising Delnevo CD, Hrywna M (2007). “A whole
cigarette excise taxes. J Public Econ, 57:457– public opinion for the tobacco industry: the ’nother smoke” or a cigarette in disguise: how
470 doi:10.1016/0047-2727(95)80006-U. Consumer Tax Alliance and excise taxes. RJ Reynolds reframed the image of little cigars.
Tob Control, 17:351–356.doi:10.1136/ Am J Public Health, 97:1368–1375.doi:10.2105/
Barnoya J, Glantz S (2002). Tobacco industry tc.2008.025338 PMID:18687706 AJPH.2006.101063 PMID:17600253
success in preventing regulation of secondhand
smoke in Latin America: the “Latin Project”. Tob Campbell RB, Balbach ED (2009). Building Devlin E, Eadie D, Angus K (2003). Discount
Control, 11:305–314.doi:10.1136/tc.11.4.305 alliances in unlikely places: progressive allies Brands (report prepared for NHS Health
PMID:12432156 and the Tobacco Institute’s coalition strategy Scotland). Glasgow: Center for Tobacco
on cigarette excise taxes. Am J Public Health, Control Research
Barzel Y (1976). An alternative approach to the 99:1188–1196.doi:10.2105/AJPH.2008.143131
analysis of taxation. J Polit Econ, 8 4:1177–1197 PMID:19443832
doi:10.1086/260507.
86
Euromonitor International (2009). Global Gilmore AB, Branston JR, Sweanor D (2010). Keeler TE, Hu TW, Barnett PG et al. (1996).
Tobacco: The Power of Cigarette Pricing. The case for OFSMOKE: how tobacco price Do cigarette producers price-discriminate by
Euromonitor International. http://www. regulation is needed to promote the health state? An empirical analysis of local cigarette
e ur o m o nito r.c o m /G l o bal _To bac c o _T h e _ of markets, government revenue and the pricing and taxation. J Health Econ, 15:499–
Power_of_Cigarette_Pricing. public. Tob Control, 19:423–430.doi:10.1136/ 512.doi:10.1016/S0167- 629 6(9 6) 0 0 49 8 - 5
tc.2009.034470 PMID:20876078 PMID:10164041
Federal Trade Commission (2009a). Federal
Trade Commission Cigarette Report for 2006. Givel M (2007). Consent and counter- Kelton MH Jr, Givel MS (2008). Public policy
Washington DC mobilization: the case of the national implications of tobacco industry smuggling
smokers alliance. J Health Commun, through Native American reservations into
Federal Trade Commission (2009b). Federal 12:339–357.doi:10.1080/10810730701326002 Canada. Int J Health Serv, 38:471–487.
Trade Commission Smokeless Tobacco Report PMID:17558787 doi:10.2190/HS.38.3.f PMID:18724578
for 2006. Washington DC
Givel M (2006). Punctuated equilibrium in Knorr GA (1985) Remarks by Gene A Knorr
Feighery EC, Schleicher NC, Ribisl K M, Rogers Limbo: the tobacco lobby and U.S. State 001200 Presentation to Pm Board Company
T (2009). An examination of the effect on Policymaking from 1990 to 2003. Policy Stud Philip Morris: Philip Morris.
cigarette prices and promotions of Philip Morris J, 34:405–418 doi:10.1111/j.1541-0072.2006.
USA penalties to stores that sell cigarettes to 00179.x. Koh HK (1996). An analysis of the successful
minors. Tob Control, 18:502–504.doi:10.1136/ 1992 Massachusetts tobacco tax initiative.
tc.2008.029116 PMID:19648133 Givel MS, Glantz SA (2001). Tobacco lobby Tob Control, 5:220–225.doi:10.1136/tc.5.3.220
political influence on US state legislatures PMID:9035358
Feighery EC, Ribisl KM, Schleicher NC et al. in the 1990s. Tob Control, 10:124–134.
(2005). How do minimum cigarette price laws doi:10.1136/tc.10.2.124 PMID:11387532 Kyriss T, Potschke-Langer M, Gruning T
affect cigarette prices at the retail level? Tob (2008). Der Verband der Cigarettenindustrie –
Control, 14:80–85.doi:10.1136/tc.2004.008656 Goldman LK, Glantz SA (1999). The passage Verhinderung wirksamer Tabakkontrollpolitik in
PMID:15791016 and initial implementation of Oregon’s Measure Deutschland. Gesundheitswesen, 70:315–324
44. Tob Control, 8:311–322.doi:10.1136/tc.8.3. doi:10.1055/s-2008-1078752. PMID:18604770
Feighery EC, Ribisl KM, Schleicher NC, Clark 311 PMID:10599577
PI (2004). Retailer participation in cigarette Levenstein C, Delaurier GF, Ahmed S, Balbach
company incentive programs is related to Gruber J, Koszegi B (2008). A Modern ED (2005). Labor and the tobacco institute’s
increased levels of cigarette advertising and EConomic View of Tobacco taxation. labor management committee in new york
cheaper cigarette prices in stores. Prev Med, state: the rise and fall of a political coalition.
38:876–884.doi:10.1016/j.ypmed.2003.12.027 Grüning T, Gilmore AB, McKee M (2006). New Solut, 15:135–152. PMID:17208826
PMID:15193911 Tobacco industry influence on science and
scientists in Germany. Am J Public Health, Lewit EM, Coate D, Grossman M (1981).
Feldman E, Bayer R (2004). Unfiltered: 96:20 –32.doi:10.2105/AJPH.2004.061507 The effects of government regulation on
conflicts over tobacco policy and public health. PMID:16317203 teenage smoking. J Law Econ, 24:545–549
Cambridge, MA, Harvard University Press. doi:10.1086/466999.
Hanson A, Sullivan R (2009). The incidence of
Gallup (2010) Tobacco and Smoking. http:// tobacco taxation: evidence from geographic Lewit EM, Coate D (1982). The potential
www.gallup.com/poll/1717/tobacco-smoking. micro-level data. Natl Tax J. In press. for using excise taxes to reduce smoking. J
aspx Health Econ, 1:121–145.doi:10.1016/0167-
Harris JE (1987). The 1983 increase in the 6296(82)90011-X PMID:10263952
Gilmore AB, McKee M (2004a). Tobacco-control federal cigarette excise tax. In: Summers LH,
policy in the European Union. In: Feldman ed., Tax policy and the economy, Volume 1. Loomis BR, Farrelly MC, Nonnemaker JM,
E, Bayer R, eds., Unfiltered: Conflicts over Cambridge, MA, MIT Press Mann NH (2006). Point of purchase cigarette
tobacco policy and public health. Cambridge, promotions before and after the Master
Massachussetts, Harvard University Press: Hedley D (2007). Consolidation Endgame in Settlement Agreement: exploring retail scanner
219–254. sight – but is there one more big throw of the data. Tob Control, 15:140–142.doi:10.1136/
dice? Euromonitor International tc.2005.011262 PMID:16565464
Gilmore AB, McKee M (2004b). Moving East:
how the transnational tobacco industry gained Heiser PF, Begay ME (1997). The campaign Lum KL, Barnes RL, Glantz SA (2009).
entry to the emerging markets of the former to raise the tobacco tax in Massachusetts. Enacting tobacco taxes by direct popular vote
Soviet Union-part II: an overview of priorities Am J Public Health, 87:968–973.doi:10.2105/ in the United States: lessons from 20 years
and tactics used to establish a manufacturing AJPH.87.6.968 PMID:9224178 of experience. Tob Control, 18:377–386.
presence. Tob Control, 13:151–160. doi:10.1136/tc.2009.029843 PMID:19556615
doi:10.1136/tc.2003.005207 PMID:15175532 Johnson TR (1978). Additional evidence on the
effects of alternative taxes on cigarette prices. McNeill A, Sweanor D (2009). Beneficence
Gilmore AB, Collin J, Townsend J (2007). J Polit Econ, 86:325–328 doi:10.1086/260671. or maleficence–Big Tobacco and smokeless
Transnational tobacco company influence products. Addiction, 104:167–168.doi:10.1111/
on tax policy during privatization of a state Johnston M (1987). Handling an excise tax j.1360-0443.2008.02483.x PMID:19149806
monopoly: British American Tobacco and increase. Philip Morris. Bates n°2022216179-
Uzbekistan. Am J Public Health, 97:2001– 2022216180 Michael J (2000). The unfair cigarette sales
2 0 0 9 . d o i : 1 0 . 2 1 0 5 / A J P H . 2 0 0 5 . 0 7 8 3 7 8 Act, an information brief. St Paul, Minnesota,
PMID:17138915 Joossens L, Raw M (2003). Turning off the tap: Minnesota House of Representatives,
the real solution to cigarette smuggling. Int J Research Department.
Gilmore AB, Radu-Loghin C, Zatushevski Tuberc Lung Dis, 7:214–222. PMID:12661834
I, McKee M (2005). Pushing up smoking Molinelli K (1999). Company Philip Morris:
incidence: plans for a privatised tobacco Philip Morris. Bates numbers: 2076396933
industry in Moldova. Lancet, 365:1354– http://legacy.library.ucsf.edu/tid/owi61b00
1359.doi:10.1016/S0140 - 6736(05)61035 - 5
PMID:15823388
87
Moodie C, Hastings G (2009). Making the Rooke C, Cheeseman H, Dockrell M et Szilágyi T, Chapman S (2003). Tobacco
pack the hero, tobacco industry response to al. (2010). Tobacco point-of-sale displays industry efforts to keep cigarettes affordable:
marketing restrictions in the UK: findings from in England: a snapshot survey of current a case study from Hungary. Cent Eur J Public
a long-term audit. Int J Ment Health Addict, practices. Tob Control, 19:279–284. Health, 11:223–228. PMID:14768787
8:1557. doi:10.1136/tc.2009.034447 PMID:20472576
Talking Retail (2006). Budget Brand Soars.
Moon RW, Males MA, Nelson DE (1993). Ruel E, Mani N, Sandoval A et al. (2004). After http://www.talkingretail.com
The 1990 Montana initiative to increase the Master Settlement Agreement: trends in the
cigarette taxes: lessons for other states and American tobacco retail environment from 1999 Traynor MP, Glantz SA (1996). California’s
localities. J Public Health Policy, 14:19–33. to 2002. Health Promot Pract, 5 Suppl;;99S– tobacco tax initiative: the development and
doi:10.2307/3342824 PMID:8486749 11 0 S . d o i : 1 0 .117 7/ 1 5 2 4 8 3 9 9 0 4 2 6 4 6 0 3 passage of Proposition 9 9. J Health Polit Policy
PMID:15231103 Law, 21:543–585.doi:10.1215/03616878-21-3-
Morley CP, Cummings KM, Hyland A et al. 543 PMID:8784688
(2002). Tobacco Institute lobbying at the state Saffer H (2000). Chapter 9: Tobacco advertising
and local levels of government in the 1990s. and promotion. In: Jha P, Chaloupka FJ, eds., United Kingdom Public Health Resource Unit
Tob Control, 11 Suppl 1;;I102–I109.doi:10.1136/ Tobacco control in developing countries. New (2006). Critical Appraisal Skills Programme
tc.11.suppl_1.i102 PMID:11893820 York, Oxford University Press on behalf of (CASP): 10 questions to help you make sense
WHO and the World Bank of qualitative research. Oxford, England
Murray W (1989). Remarks by William Murray
Vice Chairman of the Board Philip Morris Saloojee Y, Dagli E (2000). Tobacco industry US Chamber of Commerce (2011) Jobs
Companies Inc. at the 890000 Philip Morris tactics for resisting public policy on health. Agenda: Economy & Taxes. URL: http://www.
Legal Conference Ritz-Carlton Hotel Naples, Bull World Health Organ, 78(7):902-910. uschamber.com/taxes (last accessed on 12th
Florida. PMID:10686738 September 2011).
Nakkash R (2007). Tobacco industry strategies Slater S, Chaloupka FJ, Wakefield M (2001). Van Walbeek CP (2000). Industry responses
in lebanon: an analysis of internal tobacco State variation in retail promotions and to the recent tobacco excise rate increases in
industry documents. advertising for Marlboro cigarettes. Tob South Africa. Preliminary Report. Report No.
Control, 10:337–339.doi:10.1136/tc.10.4.337 Research Relase No. 2.
Neuman M, Bitton A, Glantz S (2002). Tobacco PMID:11740024
industry strategies for influencing European van Walbeek C (2010). A simulation model
Community tobacco advertising legislation. Smith KE, Fooks G, Collin J et al. (2010). to predict the fiscal and public health impact
Lancet, 359:1323–1330.doi:10.1016/S0140- “Working the system”–British American of a change in cigarette excise taxes. Tob
6736(02)08275-2 PMID:11965294 tobacco’s influence on the European union Control, 19:31–36.doi:10.1136/tc.2008.028779
treaty and its implications for policy: an analysis PMID:19850550
O’Callaghan E (1998). Minutes of a meeting of internal tobacco industry documents.
held on 15 december 1998 Company British PLoS Med, 7:e1000202.doi:10.1371/journal. Vateesatokit P, Hughes B, Ritthphakdee B
American tobacco: British american Tobacco. pmed.1000202 PMID:20084098 (2000). Thailand: winning battles, but the
war’s far from over. Tob Control, 9:122–127.
O’Sullivan B, Chapman S (2000). Eyes on Smith KE, Bambra C, Joyce KE et al. (2009). doi:10.1136/tc.9.2.122 PMID:10841840
the prize: transnational tobacco companies Partners in health? A systematic review of the
in China 1976–1997. Tob Control, 9:292–302. impact of organizational partnerships on public World Health Organization (2010). WHO
doi:10.1136/tc.9.3.292 PMID:10982573 health outcomes in England between 1997 technical manual manual on tobacco tax
and 2008. J Public Health (Oxf), 31:210–221. administration. Geneva, WHO
Ong EK, Glantz SA (2001). Constructing doi:10.1093/pubmed/fdp002 PMID:19182048
“sound science” and “good epidemiology”: World Health Organization (2009). WHO
tobacco, lawyers, and public relations firms. Spielman A, Loveless J. (2008a). The Startling report on the Global Tobacco Epidemic 2009:
Am J Public Health, 91:1749–1757.doi:10.2105/ business of tobacco. 6 March 2008. Citigroup implementing smoke-free environments.
AJPH.91.11.1749 PMID:11684593 Global Markets Ltd. Geneva, WHO
Public Health Resource Unit (2006). Critical Spielman A, Loveless J. (2008b). Lighting up: Yerger VB, Malone RE (2002). African
Appraisal Skills Programme (CASP): 10 The Tobacco Sector. 24 June 2008. Citigroup American leadership groups: smoking with the
questions to help you make sense of qualitative Global Markets Ltd. enemy. Tob Control, 11:336–345.doi:10.1136/
research. Public Health Resource Unit, tc.11.4.336 PMID:12432159
England. Sullivan R (2010). The effect of cigarette
taxation on prices: an empirical analysis using
Raebeck A, Campbell R, Balbach E (2009). city-level data.
Unhealthy Partnerships: The Tobacco Industry
and African American and Latino Labor Sumner DA (1981). Measurement of
Organizations. J Immigr Minor Health . monopoly behavior: an application to the
cigarette industry. J Polit Econ, 89:1010–1019
Rees R, Kavanagh J, Harden A et al. (2006). doi:10.1086/261017.
Young people and physical activity: a systematic
review matching their views to effective Sumner DA, Ward R (1981). Tax changes and
interventions. Health Educ Res, 21:806–825. cigarette prices. J Polit Econ, 89:1261–1265
doi:10.1093/her/cyl120 PMID:17041020 doi:10.1086/261034.
Robinson S, Bugler C (2008). General Lifestyle Sumner DA, Wohlgenant MK (1985). Effects
Survey 2008. Smoking and drinking among of an increase in the federal excise tax on
adults. Newport, United Kingdom, Office for cigarettes. Am J Agric Econ, 67:235–242
National Statistics. doi:10.2307/1240674.
88
Appendix
Literature Search strategy and results ECLAS (European Commission Web of Knowledge
(searches undertaken between Library Catalogue) Searched on Medline and BIOSIS
October 2009 and March 2010) Searched for: ‘(tobacco OR cigar* for: ‘(tobacco OR cigar* OR snus
OR snus OR «Philip Morris» OR JTI OR «Philip Morris» OR JTI OR
Electronic databases OR «R.J. Reynolds» OR Gallaher) «R.J. Reynolds» OR Gallaher) AND
Academic Search Premier, ECLAS AND (taxation OR tax OR excise OR (taxation OR tax OR excise OR
(European Commission Library price OR pricing) AND (polic* OR price OR pricing) AND (polic* OR
Catalogue), ESRC Society Today, intervention OR lobb* OR influence)’ intervention OR lobb* OR influence)’
Intute, Web of Knowledge, Ovid in ‘All Fields’ in ‘Title’ or ‘Topic’
(Embase, Econlit, Social Policy &
Practice), Index to Theses, IBSS, Results: 36 references Results: 480 references
Business Source Premier, European
Sources Online, JSTOR, National ESRC Society Today Searched in all other Web of
Library for Health, Periodicals Searched for: ‘(tobacco OR cigar* Knowledge databases for: ‘(tobacco
Archive Online OR snus OR «Philip Morris» OR JTI OR cigar* OR snus OR «Philip
OR «R.J. Reynolds» OR Gallaher) Morris» OR JTI OR «R.J. Reynolds»
Websites AND (taxation OR tax OR excise OR OR Gallaher) AND (taxation OR tax
GLOBALink, http://www.escholarship. price OR pricing) AND (polic* OR OR excise OR price OR pricing) AND
org/uc/ctcre_tcpmus, intervention OR lobb* OR influence)’ (polic* OR intervention OR lobb* OR
http://escholarship.org/uc/ctcre_tcpmi in ‘ESRC Awards & Outputs’ influence)’ in ‘Title’ or ‘Topic’
Search Strategies Results: 16 references Results: 563 references

Academic Search Premier
Searched for: ‘(tobacco OR cigar* Intute Ovid (Embase, Econlit, Social Policy
OR snus OR «Philip Morris» OR JTI Searched for: ‘(tobacco OR cigar* & Practice)
OR «R.J. Reynolds» OR Gallaher) OR snus OR «Philip Morris» OR JTI Searched for: ‘(tobacco OR cigar*
AND (taxation OR tax OR excise OR OR «R.J. Reynolds» OR Gallaher) OR snus OR «Philip Morris» OR JTI
price OR pricing) AND (polic* OR AND (taxation OR tax OR excise OR OR «R.J. Reynolds» OR Gallaher)
intervention OR lobb* OR influence)’ price OR pricing) AND (polic* OR in ‘Abstract’ AND ‘(taxation OR tax
in ‘Title, abstract or keywords’. intervention OR lobb* OR influence)’ OR excise OR price OR pricing)’ in
in ‘All Fields’ Abstract AND ‘(polic* OR intervention
Results: 407 references OR lobb* OR influence)’ in Abstract
Results: 3 references
89
Searched for: ‘(tobacco OR cigar* Business Source Premier National Library for Health
OR snus OR «Philip Morris» OR JTI Searched for: ‘(tobacco OR cigar* Searched for: ‘(tobacco OR cigar*
OR «R.J. Reynolds» OR Gallaher)’ OR snus OR «Philip Morris» OR JTI OR snus OR «Philip Morris» OR JTI
in Title AND ‘(taxation OR tax OR OR «R.J. Reynolds» OR Gallaher) OR «R.J. Reynolds» OR Gallaher)
excise OR price OR pricing)’ in Title AND (taxation OR tax OR excise OR AND (taxation OR tax OR excise OR
AND ‘(polic* OR intervention OR price OR pricing) AND (polic* OR price OR pricing) AND (polic* OR
lobb* OR influence)’ in Title intervention OR lobb* OR influence)’ intervention OR lobb* OR influence)’
in Title, Abstract & Keywords in Evidence Based Reviews,
Results: 420 Specialist Collections and Books and
Results: 398 Journals
Index to Theses
Searched for: ‘(tobacco OR cigar* European Sources Online Results: 15
OR snus OR «Philip Morris» OR JTI Searched for ‘tobacco tax’ and
OR «R.J. Reynolds» OR Gallaher) ‘tobacco excise’ in ‘keywords’ and in Periodicals Archive Online
AND (taxation OR tax OR excise OR ‘title’ Searched for: ‘(tobacco OR cigar*
price OR pricing) AND (polic* OR OR snus OR «Philip Morris» OR JTI
intervention OR lobb* OR influence)’ Results: 9 OR «R.J. Reynolds» OR Gallaher)
in ‘All Fields’ AND (taxation OR tax OR excise OR
JSTOR price OR pricing) AND (polic* OR
Results: 21 (tobacco OR cigar* OR snus OR intervention OR lobb* OR influence)’
«Philip Morris» OR JTI OR «R.J. in Article, Title & Keyword(s)
IBSS Reynolds» OR Gallaher) AND
Searched for: ‘(tobacco OR cigar* (taxation OR tax OR excise OR Results: 9
OR snus OR «Philip Morris» OR JTI price OR pricing) AND (polic* OR
OR «R.J. Reynolds» OR Gallaher) intervention OR lobb* OR influence)
AND (taxation OR tax OR excise OR in Abstract & Title
price OR pricing) AND (polic* OR
intervention OR lobb* OR influence)’ Results: 89
in ‘All Fields’
Results: 172
90
Chapter 4
Tax, price and aggregate demand
for tobacco products
Introduction
Ever since the detrimental health tobacco control, but since the early demand studies that are estimated
impact of tobacco smoking was 1990s several developing countries, with aggregate data. Aggregate
scientifically established in the among them Bangladesh, Brazil, data is constructed by, in principle,
1950s and 1960s, the medical India, Poland, South Africa and adding the relevant data of individuals
and public health community has Thailand, have implemented effective for the relevant variables (such as
called for interventions aimed at tobacco control strategies (De Beyer consumption, income and advertising
reducing smoking. This call was and Waverley Brigden, 2003). expenditure). Typically aggregate
particularly strong in the United Of the tobacco control data are produced by government
States, the United Kingdom, Canada, interventions that are available, some agencies. A common theme in all
Australia and New Zealand, and (e.g. health warnings, restrictions these studies is that cigarette prices
these countries have implemented on tar and nicotine content) have are a crucial determinant of the
strong and effective tobacco control limited economic content, and as demand for cigarettes. The price
strategies. In the 1960s and 1970s, such, economists have little to say elasticity of demand is typically the
tobacco control strategy was driven about these interventions. Other metric of interest. The more price-
primarily by the dissemination interventions (e.g. restrictions on elastic the demand, the more effective
of information. In the 1980s and smoking in hospitality establishments, is a cigarette price increase in
especially the 1990s the strategy advertising bans and restrictions reducing cigarette consumption. The
became far more comprehensive, of sales to minors) have definite relevant policy variable is the excise
incorporating restrictions on public economic consequences (for tax;; by increasing the excise tax, the
smoking, restrictions on advertising, example, the alleged loss of revenue government can raise the retail price
legal challenges to the industry, suffered by the hospitality industry of cigarettes. The policy implications of
and focused tax increases. Most after the imposition of clean indoor air knowing the price elasticity of demand
other developed countries have legislation), but in the popular debate, are: (1) it allows one to predict the
subsequently implemented similar the focus is usually on the non- likely decrease in consumption if the
strategies, with the result that per- economic aspects, such as freedom price of cigarettes were to increase by
capita tobacco consumption in the of choice and freedom of expression. a certain percentage, and (2) it allows
developed world has been decreasing The focus of this Handbook is not on one to estimate the fiscal impact of a
since the early 1980s (Gajalakshmi these tobacco control interventions, change in the excise tax on cigarettes.
et al., 2000). Tobacco use in low- thus they are not discussed further. The empirical literature on the
and middle-income countries has The primary theme of this book demand for cigarettes is sizeable,
lagged tobacco use in high-income is that increases in the price of and has several different focus
countries, although there has been cigarettes are particularly effective in points. In particular, in the past
a strong increase in tobacco use in reducing the demand for cigarettes. two decades there has been an
the past half century. As a result, This chapter provides a theoretical explosion in the number of studies
low- and middle-income countries framework used in other chapters and that are based on individual- or
have lagged developed countries in reviews the literature that considers household-level data. These studies
91
are discussed in Chapters 5 and 6. currency (effectively making imported consumption. As an improvement on
The current chapter is limited to a products more expensive) decreases backward-looking demand models,
review of studies based on aggregate aggregate imports. forward-looking rational addiction
data only. This literature grew rapidly The law of demand is derived models arose in the late 1980s.
before the 1990s, but has largely from a constrained utility maximizing These were subsequently revised, at
been eclipsed by studies based on framework. Given consumers’ least on a theoretical level, by models
micro-level data since then. In this preferences, which are presented by of addictive behaviour that assumed
chapter studies are categorized by the utility function, the imposition of a that smokers were subject to time-
geographical area: (1) the United budget constraint allows one to derive inconsistent behaviour. This section
States, (2) other developed countries the demand function for a product, describes this progression.
and (3) developing countries. The where the quantity demanded is a Underlying conventional demand
rationale for this categorization negative function of the price of that models is the assumption that
is that studies from each of these product. Empirically the interesting consumers are fully rational and
geographic regions have tended aspect is typically not whether the sovereign in their decision-making. As
to focus on different themes in the negative relationship between price such, they are the best judges of their
tobacco control literature. and quantity demanded exists, but own behaviour and on what goods
The structure of the chapter is as rather how responsive the quantity and services to spend their money.
follows. The theoretical framework demanded is to changes in the price. Within this conventional framework
underlying the demand for tobacco This responsiveness is captured in tobacco is considered a good like
is considered in the first section. the price elasticity of demand. The any other good. A chosen behaviour
The following section provides a price elasticity of demand, which is a priori assumed optimal because
short overview of the econometric is a number without units, indicates a person has rationally chosen it and
strategies that have been used to by what percentage the quantity the government has no reason, in the
estimate the demand for tobacco demanded changes in response to a absence of market failures, to interfere
using aggregate data. Next the 1% change in the price. with this revealed preference.
relevant empirical literature is The relationship between However, if a product is addictive,
reviewed, focusing especially on cigarette price and consumption has the assumption of consumer
price and income elasticities of opened up a lively methodological sovereignty applies no more. The
demand. The chapter then concludes debate. One major source of addictiveness of the product “forces”
with a consideration of the issue of contention is the theoretical modelling a consumer to buy a product that
affordability and how this influences of addictive products, since the she might not have bought had she
the demand for cigarettes. assumptions underlying the models not been addicted to it. Some early
employed have fundamentally theorists have postulated that, given
Theoretical frameworks different implications for the optimal the addictive nature of nicotine, it was
for the demand for tobacco tax level (Guhl and Hughes, 2006). not conducive to standard economic
Modelling tobacco consumption analysis (Chaloupka, 1991). It also
The negative relationship between based on economic models of choice implies that the demand for tobacco
the price of a product and the quantity has undergone continuous evolution products does not respond to changes
demanded is a fundamental concept in response to expanding knowledge in the price, and is thus perfectly
in economics. This relationship, the and insights into addictive behaviour. price-inelastic (U.S. Department of
law of demand, applies not only Initially the demand for tobacco was Health and Human Services, 2000).
for goods and services, but also modelled with a contemporaneous If this view were true, it would imply
to factors of production, in both specification. This means that current that increases in the excise tax would
micro and macro contexts. For demand is specified as a function of be impotent as a measure to reduce
example, within a microeconomic current prices and current values of all tobacco consumption. However,
context, an increase in the price of other explanatory variables. Addictive as will be clear from this and other
a product decreases the quantity behaviour was initially modelled chapters in this book, this view does
demanded of that product, if all through backward-looking “myopic” not have any empirical support.
other factors are held constant. demand models, meaning that Before the 1990s, empirical
Similarly, within a macroeconomic previous consumption (or the previous studies that investigated the
context, an appreciation of the period’s prices) influences current determinants of demand for tobacco
92
Tax, price and aggregate demand for tobacco products
were, in terms of their theoretical A rationally addicted person’s withdrawal (Chaloupka, 1991).
specification, not much different from current consumption, on the other Tolerance implies that a given
studies that investigated the demand hand, is determined not only by quantity of current consumption
for other consumer products. Broadly past consumption (which in turn yields less additional satisfaction as
speaking, tobacco consumption was is determined by past prices), but the cumulative past consumption of
specified as a function of the price also by future prices. For instance, the product increases. This implies
of the tobacco product in question, if consumers of addictive goods that, to obtain the same amount of
an income variable, the prices of predict that the price of the product additional satisfaction, consumers
substitutes and complements, an will increase in future, that knowledge would have to consume increasingly
advertising variable, and often will cause his/her consumption in the larger quantities of the product.
some dummy variables aimed at current period to decrease. Reinforcement means that current
capturing the impact of anti-tobacco Importantly, the full price of consumption of the product increases
measures. In some cases a lagged the product includes not only the future consumption, and that past
dependent variable was included in monetary price of the product, but consumption increases current
the regression equation to capture also the negative health effects and consumption. Stated differently,
“habit persistence.” Other than the legal sanctions associated with the more one consumes, the more
improving the statistical fit of the consumption. For example, the full one wants to consume. Withdrawal
regression equation, it also allowed price of drugs includes the fines, means that the smoker’s total
one to differentiate between short- prison sentences and increased utility/satisfaction falls if cigarette
run and long-run price elasticities of mortality associated with drug use, consumption is reduced.
demand, as will be pointed out in the over and above the purchase price of Furthermore, Becker and
next section. the drugs. Similarly, for alcohol, the Murphy (1988) show that the rational
Empirical tobacco demand full price includes the increased risk addiction model can explain the
analyses in the past two decades of accidents, domestic violence and observation that heavily addicted
have largely been underpinned by the increased morbidity and mortality, smokers often quit their consumption
rational addiction framework formally while for tobacco, the full price abruptly, i.e. “cold turkey.” The
introduced by Gary Becker and includes social disapproval, increased explanation lies in the fact that if the
Kevin Murphy in 1988. The model morbidity and premature death, over addiction is strong, reinforcement
of rational addiction has become the and above the prices paid for these in consumption is larger than had
standard approach to modelling the products. As an illustration, Becker the addiction been weak. As the
consumption of addictive goods. It and Murphy (1988) argue that the US level of reinforcement increases,
has been applied to coffee (Olekalns Surgeon General’s report published so does the degree of (adjacent)
and Bardsley, 1996), alcohol (Waters in January 1964 (U.S. Department complementarity between current
and Sloan, 1995;; Grossman et al., of Health Education and Welfare, and future consumption. A person
1998) and tobacco (Chaloupka, 1964) greatly increased the full who wishes to reduce consumption
1991;; Becker et al., 1994). The price of tobacco smoking. Between slowly would be incurring a loss of
essence of the rational addiction 1964 and 1975 per-capita cigarette utility over a long time period. On the
model is that people with stable consumption in the US decreased by other hand, by quitting abruptly, the
preferences may rationally decide 34%. Becker and Murphy argue that consumer incurs a large immediate
to engage in an addictive behaviour this large decrease in consumption loss of utility, but this utility loss is
since this maximizes their lifetime is consistent with rational (i.e. smaller than the utility losses that
utility (Becker and Murphy, 1988). forward-looking) behaviour, and would have been incurred had the
Becker and Murphy distinguish inconsistent with myopic behaviour. If quitting period been extended over a
between myopic and rational addiction. smokers were myopic they would not period of time.
A myopically addicted person’s current respond to information about future Becker and Murphy (1988) point
consumption is determined by his/her consequences of their activities. out some important interactions
past consumption. However, such a The rational addiction framework between time preference and
person does not consider the future provides a theoretical model to addiction. First, people who discount
in determining current consumption. describe three important concepts the future more heavily are more
As such, the future is discounted at in the addiction literature, namely likely to become addicted. Second,
an infinitely high rate. tolerance, reinforcement and addicts with higher discount rates
93
will be relatively more responsive to to “good” things like work, music, underestimate the addictive nature of
changes in the price than those with television, religion, other people, smoking.
lower discount rates. They tend to be etc. Second, the model assumes Auld and Grootendorst (2002)
more affected by short-term shocks, that people recognize the addictive attack the rational addiction model
e.g. current cigarette price increases, nature of the products they choose on a different level. While the rational
than by long-term implications of to consume, but they may still addiction model has been successfully
smoking, e.g. detrimental health make them because the gains of applied to several obviously addictive
consequences in middle and old consuming the product exceed the products (Grossman et al., 1998;; Auld
age. These two theoretical results cost of future addiction. The model and Grootendorst, 2002), presumably
explain why young and poor people does not suggest that if an addict is the strength of the theory should
are more likely to initiate smoking rational, he/she is “happy.” In fact, in lie in the fact that it would find that
than older and richer people and real life addicts are often unhappy people are not addicted to things that
typically have higher price elasticities and depressed. Sometimes the clearly are not addictive. Auld and
of demand, as indicated in Chapters addiction results from anxiety-raising Grootendorst (2002) found that the
6 and 7. Third, the long-run price and depressing events like death or standard methodology is generally
elasticity of demand will be greater, divorce, which lowers their utility. biased in the direction of rational
in absolute terms, than the short-run Becker and Murphy (1988) argue addiction. Using aggregate time
price elasticity. Fourthly, the impact that their model recognizes that series data, they found that milk,
of an expected change in the price people become addicted precisely eggs and oranges were rationally
of the addictive good will be greater because they are unhappy. However, addictive, and, specifically, that milk
than the impact of an unanticipated they would be even unhappier if they was more addictive than cigarettes.
price change. were prevented from consuming the This result implied that the estimable
The Becker-Murphy model has addictive goods. rational addiction model tends to yield
been empirically tested in a variety The most criticized aspect of spurious evidence in favour of the
of contexts, using both aggregate the rational addiction model is the rational addiction hypothesis when
data and cross-sectional data, and assumption of perfect foresight aggregate data are used.
generally the studies find support for (Chaloupka and Warner, 1999). The The rational addiction framework
the hypothesis of rational addiction. model assumes that people rationally assumes that consumers discount
These studies are discussed in later decide that they will maximize the future at an exponentially
in this chapter and in Chapter 5. While their discounted lifetime utility by declining discount factor d(t) =
studies differ in the econometric consuming an addictive product. įW ZKHUH į 7KH GLVFRXQW
detail, the contribution of the Becker- According to Akerlof (Chaloupka et IDFWRU į LV RIWHQ H[SUHVVHG DV
Murphy approach lies in the fact that al., 2000a), the rational addiction (1+r), where r is a discount rate.
forward-looking behaviour is explicitly model does not allow the possibility Exponential discounting implies
modelled into the demand equations. that people regret that they ever that consumers are time-consistent,
However, the rational addiction started smoking, given that they are i.e. have stable preferences. Any
framework has been severely assumed to be fully aware of the discounting is premised on the
criticized from some quarters. For consequences of their consumption idea that people are impatient;; they
many addiction researchers the of a potentially addictive good when place a higher value on a unit of
concept of rational addiction is making these decisions. This is consumption today and demand
oxymoronic. Why would a person unrealistic, because surveys have compensation to defer consumption
pursue an activity that he/she shown that most smokers indicate until tomorrow. In 2001 Gruber and
becomes addicted to? Becker and that they want to quit and regret that Kozsegi added to both the theoretical
Murphy answer this question as they started smoking (Gruber and and empirical debate by arguing
follows. First, the model applies to Köszegi, 2001). Similarly, numerous that consumer preferences may not
a large range of human activities studies find that individuals do not be stable over time. Preferences
and endeavours. People become have sufficient knowledge on which are time-inconsistent if a person,
addicted to different things, some of to base their consumption decisions when asked on different occasions,
which are not necessarily bad. For (Guhl and Hughes, 2006). Chaloupka displays different relative preferences.
instance, Becker and Murphy point and Warner (1999), for example, Camerer and Loewenstein (2002)
out that people can be addicted observe that adolescents often explain hyperbolic discount
94
preferences, one mathematical form welfare, sometimes at the expense differ significantly. Under the rational
of time-inconsistent discounting, as of the future self’s (O’Donoghue and addiction hypothesis, tobacco
follows: Rabin, 2003). Most people exhibit consumption decisions are governed
“Hyperbolic time discounting present-biased preferences;; they by the same rational decision-
implies that people will make have a tendency to pursue immediate making process as any other good
relatively far-sighted decisions gratification in a way that they (Gruber and Mullainathan, 2002).
when planning in advance— themselves may disapprove of in the Under this paradigm the optimal
when all costs and benefits long run. For instance, a smoker might role for government is to correct
will occur in the future—but indicate that he wishes to quit, but for the “external costs” of smoking.
will make relatively short- only in a year’s time. In this scenario Addiction per se does not constitute
sighted decisions when the future self makes the decision. market failure, and the costs smokers
some costs or benefits are However, if one were to ask him in impose on themselves are irrelevant
immediate. The systematic a year’s time whether he has quit for taxation unless they are rooted in
changes in decisions smoking, the typical answer would misperceptions about the harmfulness
produced by hyperbolic time be “no.” Despite the commitment of smoking (Guhl and Hughes, 2006).
discounting create a time- of a year earlier, the current self In comparison, the policy conclusion
inconsistency in intertemporal dominates the decision. The large from the time-inconsistency approach
choice not present in the time delay between exposure and is that “internality costs” should
exponential model. An disease makes smokers particularly be accounted for in the same way
agent who discounts utilities prone to this phenomenon, since the as externality costs when setting
exponentially would, if faced health consequences of their current government policy. Taxation may thus
with the same choice and the actions are only realized at a future be theoretically justified even where
same information, make the date (Jha et al., 2000). This type of no externalities are present (Gruber
same decision prospectively time-inconsistency, which describes and Köszegi, 2001). As a result,
as he would when the time smoking as an outcome of “multiple time-inconsistency models generally
for a decision actually arrives. selves”, strongly accords with common prescribe an optimal tax level which
In contrast, somebody with sense and conventional wisdom is higher than that of the rational
time-inconsistent hyperbolic (O’Donoghue and Rabin, 2003). Many addiction model, since the internal
discounting will wish smokers want to quit smoking, but the costs often dwarf the external costs
prospectively that in the future immediate gratification from nicotine (Gruber and Köszegi, 2002).
he will take far-sighted actions;; dominates the desire to quit. In this
but when the future arrives he framework, cigarette consumption is The empirical strategy
will behave against his earlier more appropriately modelled based
wishes, pursuing immediate on the assumption that consumers Over the past decades there have
gratification rather than long- are time-inconsistent. The existence been vast improvements in the
run well-being.” of an “internality,” arising from the techniques that are available for time
If a person discounts the future psychological phenomenon of series econometricians. Time series
at a hyperbolic rather than an hyperbolic discounting and unstable data are data that are published
exponential rate, time-inconsistent preferences, supports an argument for at regular intervals and that refer
preferences are a likely outcome. a cigarette tax, not only on externality to well-defined time periods, e.g.
Results from laboratory experiments grounds, but on the grounds that years, quarters or months. Most of
and psychological research suggest smoking creates “internal” costs that the studies discussed in this chapter
that consumers are time-inconsistent markets fail to correct. consider the time series data for a
and exhibit self-control problems If consumers exhibit present- particular country, and thus derive
(Gruber and Köszegi, 2002). Self- biased preferences (i.e. the time- appropriate coefficients for that
control problems are introduced inconsistent model), the assumptions country. However, in a limited number
into economic models through the of rational and time-consistent of cases researchers have used a
idea of a competing internal self, behaviour (i.e. the rational addiction pooled set of time-series data. A
where an individual’s preferences model) may be seriously flawed. pooled (or in some cases a panel)
change at different times with a More importantly, the optimal tax data set consists of the time series
view to improving the current self’s rate prescribed by each model will data of several countries (or regions).
95
Pooling data results in many more prices in the demand for imported obvious that past consumption
observations and more often yields cigarettes. determines current consumption.
statistically meaningful results. Early The mathematical form of the An econometric specification that
econometric studies that investigated regression equation is important. includes a lagged dependent variable
the demand for tobacco and other Many studies use a log-log has the property that one can
commodities would be regarded as specification, primarily because differentiate between short-run and
quite out-of-date by many applied this results in a constant and easily ORQJUXQ SULFH HODVWLFLW\ ,I İP is the
time series econometricians today. derived elasticity estimate. The price VKRUWUXQSULFHHODVWLFLW\DQGĮLVWKH
However, since these studies were elasticity is simply the coefficient coefficient on the lagged dependent
important then, and since many on the (logged) price variable, YDULDEOH Į WKHQ WKH ORQJ
recent studies have not employed and similarly the income elasticity UXQ SULFH HODVWLFLW\ LV HTXDO WR İP/(1-
the most up-to-date techniques, can be read off as the coefficient Į 7KH ORQJUXQ HODVWLFLW\ LV DOZD\V
it is worthwhile to discuss the on the (logged) income variable. greater (in absolute terms) than the
methodology of such studies. It is important to note that this short-run elasticity, suggesting that a
The typical starting point for specification assumes a constant current change in the price will have
estimating the price and income elasticity, over time or at different a greater impact on consumption in
elasticities of demand is to specify points on the demand curve. Of the long run than in the short run.
a demand equation. According to course such a specification would To test the rational addiction
standard demand theory, the quantity not allow one to determine whether hypothesis, the focus is on future price
demanded of a product is a function there are changes in the price or future consumption, explaining
of the price of the product, an income elasticity over time, whether the current consumption. Becker et al.
variable, the prices of substitutes elasticity differs for various price (1994) tested their theoretical model
and/or complements, advertising, levels, or whether the elasticity using US state-level time series
and possibly some product specific differs for different magnitudes of data. The forward-looking nature
factors. Within the context of tobacco price changes. of the model made them include
demand studies, price and income Another standard mathematical next-period consumption (i.e. Ct+1)
are the most important and most form is a linear specification. The LQWR WKH GHPDQG HTXDWLRQ ,I Į1 is
commonly used variables. Earlier coefficients cannot be interpreted the coefficient on Ct - 1 DQG Į 2 is the
studies1 often included an advertising as elasticities, but with minor coefficient on Ct+1, then rational
variable in the regression equation effort elasticity estimates can be DGGLFWLRQ ZRXOG LPSO\ WKDW Į 2 is
(either advertising expenditure, some calculated. Usually the elasticity VWDWLVWLFDOO\ VLJQLILFDQW ,I RQO\ Į1 is
measure of the stock of advertising is calculated at the mean quantity significant, this would suggest that
or some rudimentary count of the and price (or income, if one wishes smokers are myopically addicted.
levels of advertising). Relatively to estimate the income elasticity). 2 Also, within this framework, the
few studies included the price of The added advantage of the linear short-run elasticity is derived from
complementary and substitute goods specification is that is allows one to the coefficient on Pt, while the long-
in the regression equation. A notable estimate the elasticity at any point in UXQHODVWLFLW\LVHTXDOWRİPĮ1Į 2).
exception is the seminal study by time (Bardsley and Olekalns, 1999;; If smokers are rationally addicted,
Chapman and Richardson (1990), Van Walbeek, 2002). one would find that the long-run price
who included the price of substitutes In some studies the addictiveness elasticity is greater than if smokers
in their demand equations (leaf of tobacco was modelled by including are myopically addicted.
tobacco prices in the demand for a lagged dependent variable in In the econometric literature the
cigarettes and cigarette prices in the the regression equation. This is a potential problem of identification (i.e.
demand for leaf tobacco). Similarly, standard econometric technique, distinguishing between supply and
Hsieh et al. (1999) included imported based on the concept of “habit demand) has been an important issue.
cigarette prices in the demand for persistence” (Gujarati, 2003). If the The price-quantity combination at any
local cigarettes and local cigarette product is addictive, it is intuitively moment in time is an equilibrium point,
1
Hamilton (1972), McGuinness and Cowling (1975), Fujii (1980), Witt and Pass (1981), Bishop and Yoo (1985), Leeflang and Reuijl (1985), Radfar (1985), Abernethy and Teel (1986),
Baltagi and Levin (1986), Chetwynd et al. (1988), Kao and Tremblay (1988), Duffy (1991), Seldon and Boyd (1991), Tegene (1991), Wilcox and Vacker (1992), Valdés (1993), Duffy
(1995), Tremblay and Tremblay (1995), and Cameron (1997).
2
See Warner (1977), Warner (1981), Leeflang and Reuijl (1985), Warner (1989), Flewelling et al. (1992), Becker et al. (1994), Wilcox et al. (1994), Tremblay and Tremblay (1995), Van
Walbeek (1996), Bardsley and Olekalns (1999) and Bask and Melkersson (2004) for examples of elasticities at the mean.
96
i.e. the intersection of the supply and only;; counterfeit and smuggled price and real income is critical,
the demand curve. Any change in the cigarettes are not considered. If the since the prices of almost all goods
equilibrium position from one period proportion of smuggled cigarettes (both substitutes and complements)
to another is the result of a change remains constant over time, then rise over time. The absolute change
in the demand or the supply curve, or the price elasticity estimates and in nominal cigarette prices is not
both. To identify the demand curve a other coefficients will not be affected. important;; what is important is by
systems approach is recommended. However, if the proportion of smuggled how much cigarette prices change
If the system meets certain criteria, cigarettes increases in response relative to the changes in the prices
one can specify both the supply and to an increase in price, the legal of all other goods. Models using
the demand curves. Very few studies sales of cigarettes will understate only nominal values are also likely
follow this approach, with notable true cigarette consumption. To the to provide spurious results, since all
exceptions being Bishop and Yoo extent that this may occur, the price both price and income are likely to
(1985) and Tremblay and Tremblay elasticity estimate (in absolute terms) trend upward over time. Real income
(1995). Bishop and Yoo (1985) point will be biased upwards. The second can be presented in aggregate or
out that a single equation demand approach calculates cigarette per-capita terms. Some studies
model can provide consistent consumption as cigarette production, regress aggregate consumption on
estimates if the supply curve is either plus imports, less exports. The aggregate income and the real price
perfectly elastic or perfectly inelastic. method of data collection could also while other studies regress per capita
An assumption of a perfectly lead to biases. Local production consumption on per-capita income
inelastic supply curve, given the may be underreported (because of and a price variable (examples).
storability of tobacco, is unrealistic. illicit manufacturing, for example) Most studies use the adult population
On the assumption that the supply or imports may be underreported (aged 15 or 16 and older) as the
curve is perfectly elastic, changes (through large-scale or small-scale appropriate population measure with
in the price will trace out a demand smuggling). If such underreporting which they obtain per-capita values,
curve. Most studies assume that were to increase in sympathy although this is not a strict rule.
the demand curve is stable in price- with increases in the real price of Other variables that have been
quantity space and that movements cigarettes, the absolute value of included in demand equations
in the curve are a result of changes the price elasticity estimates would revolve around policy interventions
in the other demand determinants. be biased upwards. Essentially, (e.g. advertising restrictions, counter
In the absence of data on costs consumption cannot be directly advertising, smoke free policies, age
and other supply determinants, and observed, and what is termed as restrictions, etc). The most common
given that tobacco companies often consumption is proxied by a measure way to account for policy interventions
have significant market power (which that represents tax paid production is by means of dummy variables. For
results in them not having identifiable or shipments. Furthermore, timing example, the counter-advertising that
supply curves), it is little wonder that issues may affect the composition was legislated through the Fairness
so few studies attempt to estimate of the series since retailers or Doctrine in the USA in the late 1960s
supply equations. consumers might increase purchases and early 1970s is typically captured
The dependent variable is in response to an anticipated by a variable that had a value of
almost always the quantity of price increase, thereby artificially one in the period 1968 to 1970, and
cigarettes consumed, although increasing consumption now and zeroes in all other years (Hamilton,
some variations have been used. reducing consumption in some future 1972;; Kao & Tremblay, 1988).
The quantity consumed is typically period. This becomes more of an The most tested interventions
calculated in one of two ways. The issue in higher-frequency data. are advertising bans and
first method is to obtain, or derive, Cigarette prices and income restrictions. Initially the approach
the quantity consumed from the enter the models as independent was to investigate the relationship
fiscal authorities. By dividing the total variables and are presented in between cigarette consumption and
excise tax revenue by the average real terms by dividing the nominal advertising expenditure (or other
excise tax amount per cigarette, variable (or index) by an appropriate measures of advertising including
one can calculate the quantity of deflator (usually the consumer price stocks and counts), controlling
cigarettes consumed. This considers index). The conversion of nominal for other demand determinants
the consumption of legal cigarettes price and nominal income to real (Hamilton, 1972;; Fujii, 1980;; Bishop
97
& Yoo, 1985;; Abernethy & Teel, 1994, Pindyck and Rubinfeld 1998, studies that use microeconomic
1986;; Holak & Reddy, 1986;; Kao & Enders, 2004). The first studies or household cross sectional data.
Tremblay, 1988;; Wilcox & Vacker, that used cointegration techniques Studies that use household data
1992). The argument was that if used the Engle-Granger two- are discussed and reviewed in
a positive relationship between step procedure. Subsequently the Chapters 5 and 6. Studies based
these two variables was found, a Johansen procedure was developed, on aggregate data have some
reduction in advertising expenditure which is theoretically less restrictive significant drawbacks. The first is
would result in a reduction in than the Engle-Granger approach. In the relative paucity of data. Data
tobacco consumption. This would similar vein, econometric techniques sets in excess of 50 time series
be the empirical foundation for a that placed fewer restrictions on the observations are scarce (Prest,
ban on tobacco advertising. No data, like vector autoregression, were 1949;; Tegene, 1991, being important
consistently strong relationships developed. exceptions). In contrast, the number
were found. However, this finding The uptake of the more advanced of observations in household surveys
was subsequently rationalised on econometric techniques has been is limited only by the budget of the
the grounds that, since tobacco relatively slow in tobacco demand survey. Second, the questions one
products were among the most studies based on aggregate data. can answer with aggregate data are
advertised products in the world, and The primary reason seems to be limited. One can estimate price and
advertising expenditure is probably a lack of data. Time series data income elasticities, but one cannot,
subject to decreasing returns, are typically of annual, quarterly for example, estimate the impact
moderate changes in the advertising or monthly frequency, and even of price or income changes on the
expenditure are unlikely to be picked when the series are relatively consumption of different groups (by
up in substantial changes in cigarette long, the data demands of the new gender, race, income, education, etc.),
consumption (Saffer and Chaloupka, techniques are often such that the or determine changes in smoking
2000). Subsequent studies have modern techniques are not applied. prevalence or smoking intensity.
tested the impact of advertising bans, In developing countries this problem Household data sets are much
rather than advertising expenditure, is more acute than in developed richer than aggregate data sets, and
on consumption and have found countries. In many instances the are better suited to answering such
that bans have typically resulted in data are available only at an annual questions. Third, data limitations may
significant declines in consumption frequency, and consists of no more result in biased parameter estimates.
(see the last section of this chapter than 20 or 30 observations. The A high degree of multicollinearity
for a more comprehensive discussion quality and accuracy of data in between the independent variables,
of this literature). developing countries are likely to be a very common problem in time
Since the early 1990s there inferior to that in developed countries. series data, results in large standard
have been significant advances In cases where the data are limited errors on the estimated coefficients.
in time series econometrics. This or of dubious quality, state-of-the- Regressions may also suffer
has been primarily in reaction art techniques will not always be from omitted variable bias. Some
to the long-known but largely possible. However, data limitations variables might be omitted because
suppressed fact that regression should not provide a licence for the data do not exist or because
on non-stationary variables researchers to ignore modern the inclusion of the data worsens
can easily result in statistically developments in econometrics. At the multicollinearity problem to the
significant but economically the very minimum, one should test extent that the researchers decide to
meaningless relationships. Most for the time series properties of exclude variables that they regard as
economic variables are non- the data, and possibly perform the important. Similarly, autocorrelation
stationary, meaning that they tend regression in first differences. The in the residuals, a very common
to be subject to long-run upward or review of the literature suggests that problem in time series data, results
downward trends. To prevent such many studies in the past t wo decades in inefficient coefficients. Having said
spurious relationships, cointegration do not employ these econometric this, studies based on household
techniques were developed. The developments. data are subject to similar and other
technical details of cointegration are The proportion of tobacco estimation and data problems.
available in any standard econometric demand studies that use aggregate A more recent development
textbook (for example, Hamilton, data has been declining relative to has been the estimation of cross-
98
country (or, in the USA, cross-state) Prest, 1949;; Koutsoyiannis, 1963). have on mortality patterns in the 21st
panel models, which are essentially These studies were part of a growing century (Jha and Chaloupka, 1999).
cross sectional models based on literature that aimed to investigate The focus in most of the developing
the aggregated data of a country (or the demand for household goods. country studies was on the price
state). The first study in this genre Price and income elasticities were elasticity of demand.
was by Baltagi and Levin (1986), estimated, but the public health
who exploited the fact that different implications of these estimates were United States of America
US states had different excise not discussed, presumably because
tax regimes, and they could thus the health impact of smoking was not In terms of methodological
estimate price elasticity of demand, well publicized at that time. complexity, the USA-based studies
making appropriate adjustments for The focus of tobacco-related are generally the most advanced,
small-scale smuggling from low- empirical research changed and have addressed issues that have
tax to high-tax states. Models that significantly in the early 1970s. not been addressed by researchers
are based on country data provide During the 1970s–1990s, which in other countries. Researchers
global estimates of price and could be called a second wave, in the US have several significant
income elasticities;; in most cases researchers began to draw policy advantages over their colleagues
these models control for price and conclusions from their results. The in other countries. First, tobacco
income effects when testing the focus in this period shifted away from control is high on the authorities’
effectiveness of advertising bans and the estimation of price and income agendas3 and attracts large financial
restrictions on a global level. Earlier elasticities to the impact of advertising resources from government and
models were elementary and did not and health awareness on the demand private institutions. Consider the
use fixed effects to control for cross- for tobacco products. In estimating following examples. In 2000 the 11th
country differences nor the addictive the impact of advertising and health World Conference on Tobacco or
nature of cigarettes (Laugesen and awareness on cigarette demand, Health, held in Chicago, received
Meads 1991;; Stewart 1993). More price and income were included as a US$10 million sponsorship from
recent additions to the literature control variables in the regression two US health societies and a
have made significant improvements equation, but these were often not private foundation. This is more
in methodology and estimation the focus of the investigation. During than ten times any previous World
techniques. Saffer and Chaloupka this period the lines between “pro- Conference.4 In 2002 the Fogarty
(2000) applied country-specific fixed industry” research and “pro-tobacco International Centre made US$18
effects. Nelson (2003) tested the control” research were drawn, and million available for tobacco control
endogeneity of advertising bans, and the debate between these t wo camps research around the world. One
Blecher (2008) extended the sample was vigorous and often acrimonious. of the requirements was that the
of countries to include a large number The first two waves focused research teams in other countries be
of developing countries. exclusively on developed countries, linked to established research bodies
and the empirical results were in the US. Subsequently much larger
Empirical literature based on time series data. The amounts have been made available
third wave had its origins in 1990, by US aid agencies, primarily for
A survey of tobacco demand studies when price elasticity estimates were tobacco-related research in low-and
reveals that the focus of attention has published for Papua New Guinea, middle-income countries. These
changed several times in the past the first developing country studied studies are often anchored by US
60 years. In fact, several “waves” of (Chapman and Richardson, 1990). researchers. More than 40 working
empirical studies into the demand for During the 1990s the focus gradually papers have been published by
tobacco can be identified. shifted towards developing countries. the prestigious National Bureau for
In what could be termed the first This shift in research focus was Economic Research (NBER) on
wave, a small number of studies in reaction to the large increase in matters relating to the economics
investigated the demand for tobacco smoking in the developing world, of tobacco control. There is a pool
in the 1940s to 1960s (Stone, 1945;; and the likely impact that this would of researchers who specialize in
3
This is generally true for most states and particularly true for states like California and New York State. However, the tobacco-growing states like Kentucky and the Carolinas do not
have a strong tobacco control focus.
4
The World Conference on Tobacco or Health is held every three years and is the premier meeting of tobacco control experts, researchers and lobbyists.
99
the economics of tobacco control repeated year after year, changes in in this chapter, some seminal articles,
research, unmatched in any other smoking behaviour over time can be as well as those that are in some way
country. As an illustration, of the 39 monitored. For example, several waves unique, are briefly discussed here.
authors that contributed to Tobacco of the Monitoring the Future Surveys Warner (1977) estimated the
Control in Developing Countries (Jha have been used to estimate the price impact of various health scares
and Chaloupka, 2000), arguably elasticity of demand for cigarettes and other high-profile anti-smoking
the most comprehensive book on among secondary school pupils, and publicity campaigns on the demand
the economics of tobacco control various demographic groups within for cigarettes. In his study he did
to date, 19 were from academic this sample (e.g. Chaloupka and not estimate the price elasticity of
and/or governmental institutions in Grossman, 1996;; Chaloupka and demand, but included the impact of
the USA, while another eight were Pacula, 1998). These studies are price into the regression model by
from the World Bank, World Health discussed in detail in Chapters 5 and 6. assuming that the demand curve was
Organization or the International As indicated in the discussion linear in price-quantity space and
Monetary Fund. Of the remaining 12 of econometric strategies, the that the price elasticity of demand
authors, ten were from developed demand for cigarettes is typically was the mean price and quantity for
countries (especially the United specified as a function of price, the period (1947–70) being studied.
Kingdom and Australia), and two income and some tobacco control Imposing a price elasticity onto the
were from developing countries. interventions. For most US studies, regression equation, rather than
Second, the existence of more the prime focus of the study is on allowing the data to estimate the
than 50 states and territories, each the impact of cigarette prices on price elasticity, was rationalised
with separate legislative, excise quantity consumed. However, some on the grounds that including price
tax and retail pricing systems, studies focused primarily on the in the regression equation would
often creates a research design effect of a non-price determinant of result in unacceptable levels of
environment that allows researchers the demand for cigarettes (such as multicollinearity. Based on this
to test phenomena that would have advertising, health publicity or other analysis, Warner concluded that the
been virtually impossible to test tobacco control interventions), and various anti-smoking campaigns
otherwise. For example, changes in in such cases the price would enter resulted in a significant decrease in
state-specific excise taxes were used the regression equation as a control smoking. In fact, he suggests that
to obtain quasi-experimental price variable. Table 4.1 is a comprehensive per capita consumption of cigarettes
elasticities of demand (Baltagi and chronological summary of studies would have been 20–30% higher in
Goel, 1987). Also, studies that employ of cigarette demand in the US using 1975 had it not been for the success
individual-level data to determine aggregate data studies. The studies of the anti-smoking campaign.
price elasticities of demand for differed in many respects, including: What is interesting about this study
specific demographic groups can take (1) the frequency of data,6 (2) the use is that changes in the excise tax
advantage of the fact that there is a of econometric or non-econometric on cigarettes were not presented
large degree of variation in individual estimation techniques,7 (3) the use as a particularly effective tobacco
states’ tobacco excise tax regimes and of single equation versus multiple control instrument. The focus was
other tobacco control interventions.5 equation regression techniques, and on anti-tobacco publicity, and excise-
Third, there are some very large (4) the use of national or state-specific induced changes in the retail price
survey data sets that specifically data.8 appear to have entered the analysis
investigate smoking behaviour, While the techniques for estimating as somewhat of an afterthought.
especially among the young. the demand equations are similar
Because these surveys are often enough to not list all the US studies
5
For example, see Chaloupka and Wechsler (1997), Chaloupka and Pacula (1998), Tauras and Chaloupka (1999), and Tauras et al., (2001). These studies are dealt with separately
in Chapters 5 and 6.
6
Most studies made use of annual data. Flewelling et al. (1992), Wilcox and Vacker (1992), Duffy (1995) and Hu et al. (1995a) use quarterly data, while Keeler et al. (1993), Hu et al.
(1994 and 1995b) and Gruber and Köszegi (2001) use monthly data.
7
Studies that have estimated price elasticities without using econometric techniques include Baltagi and Goel (1987) and Peterson et al. (1992). In these studies the researchers
assessed the magnitude of changes in cigarette consumption following state-specific cigarette tax increases.
8
Of all US states, California (closely followed by New York) has the most stringent tobacco control legislation. Published studies that have investigated the impact of tobacco control
interventions on the demand for cigarettes in California include Flewelling et al. (1992), and Hu et al. (1994 and 1995b). Most US studies that investigate the demand for cigarettes
focus on the country as a whole. Some studies take cognizance of differences in taxes between states and try to account for the small- and large-scale smuggling between low-tax and
high-tax states (Baltagi and Levin, 1986, Baltagi and Goel, 1987 and Thursby and Thursby, 2000).
100
Table 4.1. Studies employing aggregate data in the United States
Estimated price Estimated income

Study Country Control variables Estimation method Data Comments
elasticity elasticity
Hamilton (1972) United States -0.51 0.73 Advertising Myopic addiction log Annual time series
expenditures linear and linear OLS data from 1926 to
(positive although 1970
insignificant), health
scares (negative and
generally significant)
Warner (1977) United States -0.5 Not included Time trend, health, Myopic addiction Annual time series
dummies for health data from 1947 to
scares, Surgeon 1970
General’s report and
Fairness Doctrine
Fujii (1980) United States Short-run Short-run Advertising Log linear and linear Annual time series
Log linear Log linear expenditure (positive myopic addiction data from 1929 to
-0.63 0.33 although not OLS 1973
Linear Linear generally significant),
-0.48 0.27 time period
Long-run Long-run dummies (negative
Log linear Log linear but generally
-0.92 0.50 insignificant)
Linear Linear
-0.71 0.33
Warner (1981) United States -0.37 Positive and Many dummies for Static myopic Annual time series
significant various health scares addiction model OLS data from 1947 to
and interventions, 1978
state law population
measure
Bishop and Yoo United States Preferred value Preferred value Advertising Static system Annual time series
(1985) = -0.45;; other = 0.92;; other expenditure, health approach (equations data from 1954 to
specifications yield specifications yield scare dummy for both supply and 1980
elasticities between elasticities from 0.86 variable demand equations)
-0.41 and -0.64 to 1.10 using 2SLS and 3SLS
Abernethy and Teel United States Negative and Not included in model Print advertising Linear OLS and log Annual time series
(1986) significant (unable to (generally positive dependent variable data from 1949 to
calculate elasticity and significant), OLS, myopic 1981
since model is linear) broadcast addiction
advertising (generally
positive although
insignificant),
package warnings
(negative
although generally
insignificant),
broadcast counter
advertising (negative
101
102

Abernethy and Teel and significant), print

(1986) (contd) advertising warning
(negative although
insignificant),
broadcast advertising
ban (insignificant)
Baltagi and Levin 46 US States Short-run Insignificant Price in neighbouring Log linear myopic Pooled annual data Uses the price
(1986) -0.22 to -0.23 states (positive), addiction OLS within, from 1963 to 1980 of cigarettes in
advertising Hausman-Talyor, neighbouring states to

(positive although Zellner-Geisel control for smuggling;;
insignificant), Search Procedure finds that there may
Fairness Doctrine be spill over effects to
dummy (negative neighbouring states
although generally where smuggling may
insignificant) be significant
Holak and Reddy United States Wide variation and Income not controlled None Log linear myopic Annual time series Industry and brand-
(1986) significance for addiction model 1950 to 1979 level regressions
Before advertising (Maximum likelihood)
ban
0.07 to 0.50 (yes,
positive)
After advertising ban
0.70 (yes, positive)
to -0.84
Baltagi and Goel United States -0.17 to -0.56 Not controlled for Small-scale Non-econometric Annual time series Attempts to
(1987) Downward trend smuggling quasi experimental data from 1956 to understand how
in elasticities over computational 1983 elasticities change
time and an upward approach over time and
biased in states how small-scale
where small-scale smuggling impacts
smuggling exists on elasticities
Kao and Tremblay United States Static Static Advertising Log linear myopic Annual time series Improving Bishop
(1988) -0.50 to -0.66 0.77 to 1.30 expenditure (positive), addiction with an IV data from 1953 to and Yoo
Dynamic Myopic dummies for the technique 1980
Short-run Short-run 1964 health scare
-0.50 to -1.02 0.74 to 1.48 (negative), 1968
Long-run Long-run Fairness Doctrine
-0.44 to -1.77 0.59 to 3.03 (insignificant), 1971
broadcast ban
(insignificant)
Seldon and United States Short-run Short-run Health scare Non-linear myopic Annual time series
Doroodian (1989) -0.4 0.27 (negative), media addiction 3SLS data from 1952 to
policy (negative simultaneous 1984
although demand system
Seldon and insignificant),

Doroodian (1989) advertising increase
(contd) competition
Warner (1989) United States -0.2 Significant and Dummies for Linear myopic Annual time series
positive interventions addiction OLS data from 1947 to
1987
Seldon and Boyd United States Short-run Short-run Advertising (positive), Myopic addiction Annual time series Uses a Cooley-
(1991) -0.22 to -0.26. Long- 0.14 to 0.25. Long- policy dummies in Maximum Likelihood data from 1953 to Prescott model,
run not significantly run not significantly 1964, 1968, 1971, 1984 interested in the
different from short- different from short- 1979 (Negative) instability of demand
run run due to continuous
government
intervention
Tegene (1991) United States -0.29 0.53 Advertising Log linear myopic Annual time series Elasticities are found
addiction model with data from 1929 to to decline over time
Kalman filter 1986
Flewelling et al. California, United -0.25 to -0.35 Income not included Seasonal and Linear static OLS Quarterly time series
(1992) States trend dummies and (including ridge data from 1980 to
dummies to control for regression) 1990
consumption spikes
related to previous tax
increases
Peterson et al. (1992) United States Not a demand Not in model None Analytical model Annual time series
model. Tax increases data from 1955 to
were associated 1988
with a decline in
consumption. Larger
the increase, the
larger the decline
Wilcox and Vacker United States Positive and Positive and Advertising, Linear stepwise Quarterly time series Nominal prices rather
(1992) insignificant insignificant (nominal seasonality regression data from 1961 to than real prices are
income) population, inflation, 1990 used
warnings and policies.
Only price and
income in final model
Keeler et al. (1993) California, United Short-run Insignificant Regulation (negative), Non-linear myopic, Monthly time series
States -0.3 to -0.5 Arizona and Oregon rational and data from 1980 to
Long-run tax increase dummies constrained rational 1990
-0.5 to -0.6 (positive) addiction models
Becker et al. (1994) 50+1 US States Short-run Generally positive Index of incentive to Linear myopic and Pooled annual data
-0.36 to -0.48 and significant smuggle, index of rational addiction from 1955 to 1985
Long-run (unable to estimate incentive for short models, 2SLS and
-0.73 to -0.85 elasticies since the distance export OLS with group and
(unable to estimate models were linear smuggling, index of year fixed effects
myopic addiction and the income incentive for short
103
elasticities since the elasticities were not distance export
104

Becker et al. (1994) models were linear) specifically reported) smuggling;; all three
(contd) generally negative
and significant
Hu et al. (1994) California, United No elasticity Not included in model Time trend Analytical time series Monthly time series
States estimated, examining model data from January
the direct impact of a 1984 to December
tax change instead 1991
Sung et al. (1994) 11 Western US states Short-run Positive Many demographic Rational addiction Pooled annual data
-0.40 variables, regulation panel linear from 1967 to 1990

Long-run (negative), small-
-0.48 scale smuggling
(positive)
Thursby and Thursby 39 US states -0.26 0.11 Canada-bordering Log linear static and Pooled annual data Model attempts to
(1994) states (negative myopic addiction from 1972 to 1990 estimate the extent of
although models smuggling
insignificant),
Indian reservations
(positive), military
(negative although
not significant),
tax differentials
(negative although
not significant),
membership of
anti-smuggling
organisations
(positive), felony
(negative),
discount rate paid
to wholesalers
(negative), time
dummies (negative)
Hu et al. (1995a) California, United No elasticity Not included in model Quarterly dummies, Analytical time series Quarterly time series
States estimated, examing tax, media model data from 1980 to
the direct impact of a 1993
tax change instead
Tremblay and United States -0.41 to -0.43 Positive and Advertising Linear static 2SLS Annual time series Includes supply side
Tremblay (1995) significant expenditure, W2SLS data from 1955 to in the equation
dummies for Surgeon 1990
General’s report,
Fairness Doctrine,
time trend
Keeler et al. (1996) United States Not a demand study. N/a N/a Analytical model Pooled annual data
Shows that price from 1960 to 1990
competition can
reduce impact of
regulation. A 1 cent
state tax increase
increases prices by
1.11 cents
Baltagi et al. (2000) 46 US States Traditional models Traditional models State-specific Various dynamic Pooled annual data Tests whether
(OLS, GLS, etc) (OLS, GLS, etc) demographic panel estimators from 1963 to 1992 homogenous
Short-run Short-run variables including including OLS and estimators preferred
-0.09 to -0.30 -0.03 to 0.10 religion, education GLS (with various to heterogeneous
Long-run Long-run and race. Price in effects), and 2SLS, estimators
-1.79 to -2.98 -1.00 to 0.60 neighbouring states EC2SLS and
Instrumental Instrumental (substitutes) FD2SLS for dynamic.
models models
Short-run Short-run
-0.21 to -0.50 -0.02 to 0.19
Long-run Long-run
-0.68 to -1.37 -0.11 to 0.51
Gruber and Köszegi United States Short-run Not reported Unknown Linear rational Annual time series Values in [ ]
(2001) [-0.43 to -0.48] addiction model with data from 1973 to calculated by the
group and year fixed 1996 Working Group
effects
Huang et al. (2004) 42 states and Short-run Short-run Price in neighbouring Log linear myopic Pooled annual data Long-run results
Washington DC, USA -0.21 0.04 states, time trends, addiction, OLS and from 1961 to 2002 calculated from the
Long-run Long-run proportion of 2SLS short-run results and
[-1.88] [3.73] population 65 and the lagged dependent
Instrumental Instrumental above variable by the
Short-run Short-run Working Group
-0.41 0.06
Long-run Long-run
[-2.69] [0.38]
Goel (2009) United States Static Static Advertising, Log linear dynamic Pooled annual data Long-run results
-0.86 to -0.92 0.05 to 0.08 master settlement OLS with fixed from 1975 to 2004 calculated from the
Dynamic Dynamic agreement effects short-run results and
Short-run Short-run the lagged dependent
-0.04 to -0.11 -0.01 to -0.03 variable by the
Long-run Long-run Working Group
[-2 to -3.7] [-0.3 to -1.5]
Values in [ ] calculated by the Working Group using data reported in corresponding studies.
EC2SLS, 2-stage least squares procedure that assumes a one-way error-component model;; FD2SLS, first-difference 2-stage least square estimator;; GLS, generalized least squares estimates/estimator;; OLS, ordinary least squares
estimates/estimator;; W2SLS, weighted least squares estimates/estimator;; 2SLS, 2-stage least squares estimates/estimator;; 3SLS, 3-stage least squares estimates/estimator
105
Fujii’s (1980) study uses ridge elasticity accounting for the small- Given that the monthly data were
regression to alleviate the problems scale smuggling effect and cross too unstable to yield meaningful
associated with multicollinearity border sales is estimated at about results, they were forced to use an
between the independent variables. í unweighted moving average of the
While the focus of his analysis An innovation introduced by previous 12 months’ consumption as
was still largely on the impact of Seldon and Boyd (1991) was to a proxy for current consumption and
the health scare introduced by test and adjust for instability in the an unweighted moving average of
the Surgeon General in 1964 and demand function. Using annual the next 12 months’ consumption as
airing of anti-smoking commercials data for the period 1953–1984 and a proxy for next period consumption.
at the end of the 1960s, he found using a technique developed by They found that the coefficient on
the effects of these interventions Cooley and Prescott (1976), they future consumption is positive and
modest. However, he concluded that found that the demand for cigarettes significant, and thus supportive of
“taxation, however, could have more was in fact quite unstable over this the rational addiction hypothesis.
substantial effects. Estimates of the period. The instability of the demand However, they found that the price
price elasticity of demand (about function is attributed to government elasticities produced by the myopic
í LQGLFDWH WKDW D LQFUHDVH interventions (e.g. the 1964 Surgeon and rational addiction frameworks,
in the price of cigarettes will lower General’s report, the 1965 Cigarette and by a framework which does not
consumption by 4.5%. This effect Labelling and Advertising Act and account for addictive behaviour, were
became more pronounced in the the anti-smoking commercials that remarkably consistent in the range of
long run” (Fujii, 1980). started after 1968). By including íWRí
Baltagi and Levin (1986) appropriate dummy variables for Becker et al. (1994) used a large
estimated the demand for cigarettes these interventions and estimating the aggregate data set of more than
in the USA, based on annual data demand equations with a Maximum 1500 observations (50 states over
for the period 1963–1980 for 46 Likelihood technique derived from the 31 years) to investigate empirically
states. What is interesting about this Cooley-Prescott method, they found whether cigarette smokers are
study is that it accounts for possible that the demand system stabilized, “rational” in the way that rational
small-scale smuggling between and that the inclusion of the dummy addiction is defined. Overall, the
states and cross-border shopping. variables reduces the price elasticity results rejected the myopic model
Given differences in cigarette excise HVWLPDWHLQDEVROXWHWHUPVIURPí of addiction, and provided evidence
taxes and hence prices between WRí that consumers do consider future
states, one can expect a difference The rational addiction framework prices in their current consumption
between sales and consumption of Becker and Murphy spawned a decisions (Becker et al., 1994).
in states (high-price states have substantial literature that aimed to Despite the many differences
greater consumption than sales, and incorporate the principles of forward- in research methodology, there are
vice versa in low-price states). If one looking behaviour into the empirical several generalizations that follow
does not account for the small-scale results. The rational addiction from these studies. First, all empirical
smuggling effect and cross-border framework has been applied to both studies included the cigarette
sales, the price elasticity estimates aggregate and individual-level data. price as a determinant of cigarette
would be biased away from zero. To Studies based on individual-level consumption, and evidence for a
account for small-scale smuggling data, initiated by Chaloupka (1991), strong negative relationship between
and cross-border sales between are covered in Chapter 6. The first these two variables is overwhelming.
high- or low-price states, they studies to test the rational addiction The price elasticity estimates
included the cigarette price in the hypothesis with aggregate data were JHQHUDOO\ YDULHG EHWZHHQ í DQG
lowest-price neighbouring state. by Keeler et al. (1993) and Becker et í EXW VHHP WR EH FRQFHQWUDWHG
They found a neighbouring state al. (1994). LQ WKH í WR í UDQJH 7KLV
price elasticity of 0.08, suggesting Keeler et al. (1993) considered implies that the demand for cigarettes
that if the price in the neighbouring monthly data for California for the is relatively price-inelastic, but
state is 10% lower than the home period 1980–1990. In specifying a certainly not perfectly price-inelastic.
state, the cigarette sales in the demand equation, they regressed From a tobacco control perspective
home state would be expected current consumption on, among this result provides the rationale for
to decrease by 0.8%. The price others, future consumption. using excise tax increases as a tool
106
to reduce cigarette consumption. An consumption. Other studies found a Franke, 1991). This suggests that
increase in the excise tax increases positive relationship, but even when cigarettes are a normal good, which
the retail price of cigarettes, such a relationship was found, its is intuitively reasonable (income
which in turn decreases cigarette impact on cigarette consumption elasticity estimates are explicitly
consumption. Furthermore, since was small (Fujii, 1980;; Bishop and reviewed in this chapter). However,
cigarettes are inelastic, the increase Yoo, 1985;; Abernethy and Teel, the policy impact of this result is not
in price will be greater than the 1986;;9 Holak and Reddy, 1986;; Kao that one should advocate for lower
decrease in consumption. Therefore, and Tremblay, 1988;; Seldon and income as a tobacco control measure,
total tax revenue is likely to rise when Doroodian, 1989;; Seldon and Boyd, but that growth in income should be
taxes rise since the tax increase 1991).10 taken into account when considering
more than compensates for the Saffer (2000) has proposed that tax policy. A separate literature that
declining consumption. advertising per se is an inappropriate considers affordability (price and
Second, studies that investigated measure and that, since many income simultaneously) has recently
“health scares” and anti-smoking countries have experimented with developed and is considered later in
publicity which resulted from the advertising restrictions and bans, we this chapter.
Fairness Doctrine generally found can actually test the impact of policies Baltagi and Goel (1987) and
that they reduced the demand for (we review this literature in the last Tegene (1991) used estimation
cigarettes (Hamilton, 1972;; Baltagi section of this chapter). He argues techniques that allowed the price
and Levin, 1986;; Kao and Tremblay, that the high level of aggregation elasticity to change over time and
1988). According to this Doctrine, of advertising expenditure data found that the demand for cigarettes
companies advertising “controversial used in time series studies leaves in the USA became less elastic over
goods” (which, after the publication of little variation to correlate with time. Andrews and Franke (1991),
the 1964 Surgeon General’s report, consumption data. Generally, since using a meta-analysis of published
included cigarettes) had to pay for the marginal product of advertising studies came to a similar conclusion.
advertisements that presented the is very low (and possibly even zero) Figure 4.1 shows a scatter plot of the
alternative view. This resulted in it is not likely that we would find any estimated price elasticity of demand,
substantial anti-tobacco advertising relationship between advertising against the midpoint of the period
between 1967 and 1970. However, expenditure and consumption. for which the study was performed.
the relative magnitude of the publicity Chapman (1989) also criticized The studies included in this scatter
effect is unclear. Hamilton (1972), the use of these techniques, and plot have been taken primarily from
Warner (1977, 1981 and 1989), Kao in particular noted the inability of Andrews and Franke (1991) and US
and Tremblay (1988), found evidence aggregate studies to examine all Department of Health and Human
of a sizeable long-term effect, while methods of promotion (including Services (2000), supplemented by
Fujii (1980) and Bishop and Yoo non-advertising and price-based more recent research.11 In Figure 4.1
(1985) concluded that the impact was promotion) used by the tobacco the number of years of data used
small and transitory. In more recent industry. Econometric analysis only in each study is represented by the
studies, anti-smoking publicity as a examines the effects of advertising size of each dot. The smallest dots
determinant of consumption seems on aggregate data, while advertising represent the smallest number of
to have received little attention in the also has an influence on smoking years of data, while the studies that
empirical literature. related cognition and beliefs. use the largest number of years
Third, there is no consensus on Fourthly, most studies incorporate of data have the largest dots. The
the impact of advertising expenditure an income variable in the demand smallest number of years of data is
on the demand for cigarettes. Several equation as a control variable. 10 years while the largest is 58 years.
studies (Hamilton, 1972;; Wilcox and There is no consensus on the value Figure 4.1 shows that the majority
Vacker, 1992;; Duffy, 1995), found of the income elasticity of demand of estimates of price elasticity lie
no significant relationship between for cigarettes, other than that it lies EHWZHHQ í DQG í ZLWK PRVW
advertising expenditure and cigarette between zero and one (Andrews and FOXVWHUHG EHWZHHQ í DQG í
9
Abernethy and Teel (1986) found a significant positive relationship for print advertising but not broadcast advertising.
10
See Andrews and Franke (1991) for a more complete list.
11
For studies listed in US Department of Health and Human Services (2000), the price elasticity estimates as published in USDHHS were used in Figure 4.1. For studies listed in
Andrews and Franke (1991), an average price elasticity is calculated if more than one price elasticity estimate was published in any particular study. Long-run price elasticities are not
shown.
107
Figure 4.1. Price elasticity estimates for the USA, based on time series studies
Sources: The studies referred to in the figure can be found in Table 4.1. The table only includes short-run elasticities. The years refer to the middle year under investigation in those
studies, hence the most recent study (Goel, 2009) includes data from 1975 to 2004 and is thus listed in 1989/90.
The small number of studies which the coefficient on price (and thus for, this results in price elasticity
fall outside of this cluster have no reduce the absolute value of the estimates that are biased away from
systematic methodological, data or price elasticity). However, while zero. Accounting for illicit cigarettes
estimation technique differences this is a theoretical possibility, the reduces the bias and hence the
from those within the cluster. On the list of control variables in demand absolute value of the price elasticity
basis of these studies there is no equations based on aggregate data estimates.
strong evidence that the demand for has not changed much over time, Second, as the general level of
cigarettes has become less elastic and thus more complete and more income has increased over the past
over time. More recent studies comprehensive specification of 50 years, cigarettes have generally
suggest that the range of estimated the demand equation is not a likely become more affordable (despite
price elasticities has narrowed reason for the perceived reduction the fact that their real price has
somewhat since the 1970s, with in the price elasticity over time. increased over this time period). Thus
UHGXFHG YDULDELOLW\ EHWZHHQ í However, more recent studies have the proportion of income required to
DQG í 2I FRXUVH WKLV UDLVHV given explicit recognition of illicit purchase a packet of cigarettes has
the question of why this may be the cigarettes (i.e. smuggled, etc.) and decreased over this long time period.
case. Several reasons can be put their impact on the price elasticity However, this is not true for the
forward. First, estimation techniques estimates (Baltagi and Levin, 1986, past two decades, which have seen
(and possibly data quality) may Baltagi and Goel, 1987, Becker et very rapid increases in the price of
have improved over time. For al., 1994, Sung et al., 1994, Thursby cigarettes. Economic theory indicates
example, including variables in the and Thursby, 2000, Baltagi et al., that the price elasticity (in absolute
demand equation that are positively 2000, Huang et al., 2004). If the illicit terms) decreases as the product
related to price will tend to reduce trade in cigarettes is not controlled takes up a smaller proportion of a
108
Figure 4.2. Price elasticity estimates for the United Kingdom and other high-income countries, based on time series
studies. White squares are United Kingdom studies and solid black squares are from other high-income countries.
Sources: Figure calculated by the Working Group.

The studies referred to in the figure can be found in Table 4.2. The table only includes short-run elasticities. Estimates for several countries come from a single study by
Koutsoyiannis (1963) and result in a grouping on the midpoint 1954/1955. The years refer to the middle year under investigation;; hence the most recent study (Pierani and Tiezzi,
2009) includes data from 1960 to 2002 and is thus listed in 1981.
person’s or household’s total income by the solid black squares. Again, is very limited evidence that the
and becomes more affordable. the larger squares represent longer absolute value of the price elasticity
time periods under consideration, of demand may be decreasing in the
Other high-income countries the smallest being 9 years and the USA, there is no such evidence for
largest being 68 years. The price other developed countries. It is part
Table 4.2 is a comprehensive elasticity estimates range between of the conventional wisdom that the
chronological summary of 39 ]HUR LH QRW VLJQLILFDQW DQG í average price elasticity of demand for
published cigarette demand studies with most studies clustered between WREDFFRLVDURXQGíLQGHYHORSHG
in developed countries other than the íDQGí:LWKWZRH[FHSWLRQV countries (Jha and Chaloupka,
US. The list is dominated by studies all estimates in the United Kingdom 1999;; Chaloupka et al., 2000b;; U.S.
based on the United Kingdom (16), OLH EHWZHHQ í DQG í 7KHUH Department of Health and Human
followed by that of New Zealand (four). is greater variation in other high- Service, 2000).
As it turns out, these two countries, income countries, although this might Several of the studies in this
together with Australia, Canada and be expected since the countries group of countries are notable.
the USA, have been at the forefront of vary significantly with respect to Stone (1945), Prest (1949) and
tobacco control policy. many factors (i.e. price, income and Koutsoyiannis (1963) were the first
Figure 4.2 depicts all of the tobacco control measures). Even to estimate demand in any country.
elasticity estimates from Table 4.2. though the methodologies, types of Townsend (1987) was unique in that
The large number of estimates from data and estimation techniques differ it estimated different elasticities for
the United Kingdom are depicted widely between studies, the results different social classes in the United
by the white squares and studies do not differ widely, especially in Kingdom.
from other high-income countries the United Kingdom. Whereas there
109
110
Table 4.2. Studies employing aggregate data in other high-income countries

Stone (1945) United Kingdom and UK UK Time trend (positive), Static log linear OLS Annual time series First known study to
United States between –0.49 and 0.07 (not significant) UK dummy for data from 1920 to attempt to estimate
–0.53 US coupon trading 1938 elasticities
UK 0.33 (1928–33) (positive)
0.24
Prest (1949) United Kingdom Between –0.12 and 0.17 Time trend (positive), Static levels and first Annual time series
–0.31, depending on post-World War I differences log linear data from 1870 to
specification dummy variable model 1938 with 1915 to
(positive) 1919 excluded

Koutsoyiannis 14 high-income US US Population size Static log linear OLS Annual time series Single equation for
(1963) countries -0.94 0.34 (insig.) (positive), data from 1950 to each country
(single country UK UK prices of all other 1959
equations) -0.04 (insig.) 0.07 (insig.) goods and services
France France (negative for Greece,
-0.54 0.83 not included for other
Italy Italy countries),
-0.82 0.48 (insig.) Time trend (generally
The Netherlands The Netherlands positive)
-0.08 (insig.) 0.10 (insig.)
Belgium Belgium
-0.68 Not shown
Sweden Sweden
-0.41 0.26 (insig.)
Norway Norway
Not shown (insig.) 0.18
Finland Finland
-0.41 0.13 (insig.)
Austria Austria
-0.95 0.11
Greece Greece
Not shown (insig.) 0.07 (insig.)
Ireland Ireland
-0.15 (insig.) 0.56 (insig.)
Canada Canada
-0.21 (insig.) 0.09 (insig.)
Australia Australia
-0.36 0.43 (insig.)
Sumner (1971) United Kingdom For annual data Positive Health publicity Annual and quarterly Investigated the
between –0.13 and dummy and trend time series data from impact of the 1962
–0.57, depending on variables (negative) 1951 to 1967 Royal College
specification of Physicians
For quarterly data report on cigarette
between –0.60 and consumption
–0.83, depending on
specification
Atkinson and Skegg United Kingdom All adults Positive Health publicity Annual time series
(1973) between –0.1 and dummy and trend data from 1951 to
–0.4 depending on variables (generally 1970
specification negative)
Males
statistically
insignificant
Females
-0.35
Russell (1973) United Kingdom -0.50 to –0.66 Not explicitly Royal College of Rather than using Annual time series Males only
included, a price- Physicians Reports regression analysis, data from 1946 to
income ratio is (1962 and 1971) the results are based 1971
included (affordablility (negative) on correlations
elasticity) -0.44 to and graphical
-0.58 representations of
the data
Peto (1974) United Kingdom For males 0.14 to 0.49 (generally Health publicity Seemingly a log Annual time series Used to try to address
-0.37 to –0.64 insignificant) dummy and trend linear OLS data from 1951 to the conflicting results
Females not variables (generally 1970 regarding price
investigated negative) elasticity between
Atkinson and Skegg
(1973) and Russell
(1973)
McGuinness and United Kingdom Short run Short run «Stock» od Log linear model Quarterly time series
Cowling (1975) -0.99 0.31 advertising addition (Koyck) data from 1957 (Q2)
Long run Long run expenditure (positive) to 1968 (Q4)
-1.05 0.33
Metra Consulting United Kingdom Short-term Unknown Unknown Unknown Quarterly time series The study was based
Group Ltd (1979) -0.34 to –0.54 data from 1958 to on and aimed to
Long-term 1978 refute McGuinness
-0.42 to –0.54 and Cowling’s
(1975) finding of a
positive relationship
between advertising
and cigarette
consumption.
Using data supplied
by the tobacco
industry, the study
found no significant
relationship
between advertising
and cigarette
consumption. Quoted
in High (1999)
111
112

Witt and Pass United Kingdom -0.3 Income (positive) Advertising Annual time series Investigated whether
(1981) expenditure data from 1955 to the reports by the
(positive), 1975 Royal College of
“Health scare” Physicians (1962
dummy variables and 1971) and the
(negative) US Surgeon General
(1964) had an
impact on cigarette
consumption
Leeflang and Reuijl West Germany Excluded from [0.50 to 0.60] Household Linear OLS myopic Annual, quarterly and The Working Group
(1985) analysis because (insignificant) consumption (proxy addiction monthly time series calculated the
“coefficient of for income) (positive), data from 1961 to elasticities in [ ] at the
variation is extremely Sales quantities 1975 mean
low” of substitutes to
cigarettes (generally
negative), Advertising
expenditure (positive)
Radfar (1985) United Kingdom -0.23 0.12 “Stock” of advertising Myopic addiction log Quarterly time series Replicates
expenditure (positive), linear OLS data from 1965 (Q3) McGuinness and
health publicity to 1980 (Q4) Cowling’s (1975)
(interaction dummy study
variables with
advertising variables)
Johnson (1986) Australia -0.10 at the means of Positive Advertising Annual data from
the sample data expenditure 1961/62 to 1982/83
(insignificant),
ban on electronic
media advertising
dummy variables
(insignificant)
Worgotter and Austria -0.54 Positive Total private Annual time series
Kunze (1986) consumption (proxy data from 1955 to
for income) (positive), 1983
“Stop smoking”
dummy variable for
1974 (negative)
Stavrinos (1987) Greece Linear Linear Health promotion Myopic addiction Annual time series Uses nominal data
Short-run Short-run programme dummy linear and log linear data from 1961 to
-0.01 0.12 variable (negative) OLS 1982
Long-run Long-run
-0.57 0.70
Log linear Log linear
Short-run Short-run
-0.08 0.18
Long-run Long-run
-0.15 0.33
Townsend (1987) United Kingdom Between +0.15 (male Between 2.6 (male Trend (generally Static log linear OLS Annual time series Investigates
professionals) and professionals) and negative, but data from 1961 to differential
–1.26 (for unskilled 0.2 (for unskilled insignificant), 1977 elasticities amongst
male workers), all male workers), all Health publicity demographic groups
insignificant insignificant dummy variables (five social classes);;
(varying, but more educated and
generally negative) well-off people have
lower price elasticity
of demand, but
respond faster to
health information
than less-educated
and poorer people
Chetwynd et al. New Zealand Between –0.13 Between –0.26 Advertising Myopic addiction, log Quarterly and annual
(1988) and –0.73 (but and 0.64 (but expenditure linear OLS time series data from
insignificant) insignificant), only (positive), 1973 to 1985
significant estimates consumption in
are between 0.50 and previous period
0.64 (positive)
Harrison et al. New Zealand Short-run Short-run Lagged consumption, Log linear OLS An extension of
(1989) -0.08 0.50 Seasonal dummy Chetwynd et al.
Long-run Long-run variables (1988) and addresses
[-0.14] [0.78] Jackson and
(but both elasticity The Working Group Ekelund’s (1989)
estimates are calculated the long- criticism that the
statistically run estimates in [ ] original paper suffers
insignificant) from the short-run from a number
estimate and the of econometric
lagged dependent modelling drawbacks.
variable Using some standard
econometric tests,
Harrison et al.
show that “the
original model and
conclusions appear
to be very robust”
Harrison and New Zealand -0.32 Positive Advertising Quarterly time series
Chetwynd (1990) expenditure data from 1973 (Q1)
(positive), to 1989 (Q2)
Anti-smoking
advertising (negative)
Duffy (1991) United Kingdom -0.32 0.89 Advertising Demand system Quarterly time series Demand system
expenditure (not model data from 1963 (Q1) includes alcoholic
significant) to 1987 (Q3) drinks and cigarettes
113
114

Andrews and Meta analysis of Weighted mean Weighted mean Other relationships: Meta-analysis Meta-analysis of 48
Franke (1991) studies performed in elasticity for all elasticity for all Advertising studies
the United Kingdom, 198 regression 190 regression expenditure (positive)
United States, and six equations equations
other countries -0.36 0.36
Weighted mean Weighted mean
elasticity for all 41 elasticity for all 36
studies studies
-0.47 0.40
Valdés (1993) Spain Short-run Short-run Advertising Myopic log linear Annual time series
-0.60 0.17 expenditure OLS data from 1964 to
Long-run Long-run (positive), 1988
-0.69 0.20 dummy variables
for legislative
interventions
(generally negative)
Simester and Brodie Meta-analysis of Average price Positive Advertising (positive) Meta-analysis Meta-analysis of 29
(1994) studies performed elasticity published studies
in United Kingdom, -0.54
United States, New Median price
Zealand and West elasticity
Germany -0.48
Townsend et al. United Kingdom For males (overall) Generally no Health publicity Static log linear OLS, Annual time series Assesses elasticitties
(1994) -0.47 significant income (negative) single equation for data from 1972 to for different
For females effect, generally each group 1990 demographic groups
(overall) positive amongst men
-0.61 with the exception of
Elasticity was older men (50+)
inversely related to
social class. Young
adult males are not
price-responsive, but
young adult females
are
Duffy (1995) United Kingdom -0.35 to -0.47 0.96 to 1.01 Advertising Almost Ideal Demand Quarterly time series System includes
(insignificant) System (AIDS), static data from 1963 (Q1) beer, spirits, wine and
and dynamic to 1992 (Q1) all other commodities
Cameron (1997) Greece Negative and Income (positive Television advertising Linear and log linear Annual time series No end date reported
insignificant price and significant in dummy (positive but myopic and rational data from 1970
elasticity myopic addiction insignificant), Anti addiction OLS
model, positive smoking campaign
and insignificant in dummy (generally
rational addiction negative and
model) significant)
Hondroyiannis and Greece Short-run Short-run Dynamic Error Annual time series
Papapetrou (1997) -0.33 0.35 Correction Model data from 1960 to
Long-run Long-run 1990
-0.60 0.54
Lanoie and Leclair Canada Legal consumption Legal consumption Regulation and time Myopic linear GLS Pooled annual data Looks at smuggling
(1998) Short-run Short-run trend with fixed effects from 1980 to 1995 and its impact on
[-0.34] [0.24] elasticities.
Long-run Long-run Values in [ ]
[-0.61] [0.44] calculated by the
Total consumption Total consumption Working Group
Short-run Short-run
[-0.11] [0.14]
Long-run Long-run
[-0.28] [0.36]
Bardsley and Australia Short-run Short-run Age structure of Linear rational Annual time series Investigates how
Olekalns (1999) Between [-0.2 and Between [0.2 and population (older addiction model, data from 1962/63 elasticity estimates
–0.3] for period 1963 0.4] for period 1963 population ==> GMM to1995/96 have changed over
to early 1980s, but to early 1980s, but more cigarette time.
increases rapidly to increases rapidly to consumption), Elasticity values in
[–1.2] between 1982 [0.7] between 1982 Advertising (positive, [ ] computed at the
and 1996 and 1996 but small), mean by the Working
Long-run Long-run Health warnings Group
Between [-0.5 and Between [0.4 and (negative, but small),
–0.6] for period 1963 0.8] for period 1963 Ban on smoking
to early 1980s, but to early 1980s, but in public places
increases rapidly to increases rapidly to (negative, but small)
[–3.0] between 1982 [1.5] between 1982
and 1996 and 1996
Mindell and Whynes Netherlands -0.45 to -1.03 1.2 Unemployment and Log linear static OLS Annual time series Looks specifically at
(2000) health promotion data from 1970 to substitution between
expenditure 1995 manufactured and
hand-rolled
Reinhardt and Giles Canada Short-run Short-run Includes dummy Log linear OLS Quarterly times
(2001) -0.62 0.19 variables for Myopic model series data from
Long-run Long-run structural breaks 1968:1 to 1990:2
-0.23 no estimated
Yorozu and Zhou Japan -0.64 to -1.23 -0.20 to 0.38 Anti smoking budget, Static linear OLS and Data from two time Study looks at
(2002) time trend WLS periods: 1990 and regional cross
1995 sections at two points
in time
Gallus et al. (2003) Italy -0.43 0.10 Static log linear Annual time series
model data from 1970 to
2001
Gruber, Sen and Canada -0.72 but -0.47 Positive Unemployment Two way fixed effects Annual time series
Stabile (2003) excluding smuggling, linear model data from 1981 to
provinces and years 1991
115
116

Bask and Sweden Long-run Not included in model Price of alcohol Rational addiction Annual time series
Melkersson (2004) Separate models (negative) linear OLS, SUR data from 1955 to
-0.81 (-3.63 when and GMM. First 1999
two lags and leads differences.
included)
Joint models (alcohol
and cigarettes
together)
-0.65 to -1.03
Escario and Molina Spain Virginia tobacco Expenditure Almost Ideal Demand Annual time series Elasticities have
(2004) -0.80 elasticities System data 1964 to 1995 fallen over time in
Black tobacco Virginia tobacco absolute terms
-0.48 0.37
Cigars Black tobacco
-0.93 0.67
Cigars
0.70
Duffy (2006) United Kingdom Static Static Advertising Static, myopic and Quarterly time series
-0.41 to -0.48 0.28 to 0.31 expenditure rational models, data from 1962, 2nd
Myopic Myopic linear, elasticity at quarter to 2002, 3rd
Short-run Short-run the mean. Demand quarter
-0.16 to -0.46 0.08 to 0.28 system.
Long-run Long-run
-0.45 to -0.49 0.20 to 0.29
Rational Rational
Short-run Short-run
-0.01 to -0.13 0.05 to 0.06
Long-run Long-run
-0.26 to -0.41 0.11 to 0.16
Pierani and Tiezzi Italy Short-run Short-run Price of complement Rational addiction Annual time series Studies the
(2009) [-0.09 to -.0.34] [-0.09 to -.034] (i.e. alcohol or first difference GMM data between 1960 relationship between
Long-run Long-run tobacco) in single (independent and and 2002 alcohol and cigarette
[-0.31 to -1.07] [0.19 to 1.56] equation model systems equations) consumption.
Elasticity values in
[ ] computed at the
mean by the Working
Group
For quarterly data, corresponding quarter specified in parenthesis.

In this table, short-run & long-run estimates can also be designated as SR and LR respectively.
Values in [ ] calculated by the Working Group using data reported in corresponding studies.
(insig.) refers to non statistically significant estimates
GLS, generalized least squares estimates/estimator;; GMM, generalized methods of moments estimator;; OLS, ordinary least squares estimates/estimator;; SUR, seemingly unrelated system of equations;; WLS, weighted least squares
estimates/estimator
This is unusual in aggregate cigarettes, beer, wine, spirits and all series techniques is relatively rare
data studies and usually only other commodities. This time Duffy in this literature, and this was the
found in cross-sectional studies. used a dynamic specification of first studies to do so in this context.
Unsurprisingly, the results showed demand although he did not report It found a short-run price elasticity
that unskilled male workers were the cross price elasticities. Escario RI í DQG D ORQJUXQ HODVWLFLW\ RI
more responsive to price changes and Molina (2004) employed a í 2WKHU *UHHN VWXGLHV IRXQG
than were male professionals. dynamic AIDS in Spain. VKRUWUXQ SULFH HODVWLFLWLHV RI í
Townsend et al. (1994), in another However, they did not include DQG í 6WDYULQRV ZKLOH
United Kingdom study again alcohol, rather including three Koutsoyiannis (1963) and Cameron
estimated elasticities for different different types of tobacco products, (1997) found price to be insignificant.
socioeconomic and age groups Virginia tobacco, black tobacco and Lanoie and Leclair (1998) is
using aggregate data. They found cigars. They found that the price a Canadian study that aimed to
that females are more sensitive to elasticities, in absolute terms, fell deal with the problem of cigarette
price changes than were males, over time between 1964 and 1995. smuggling using annual time series
and that elasticities are inversely Price elasticities for all tobacco data from 10 Canadian provinces
related to social class. Young adult products considered were negative, from 1980 to 1995. During this time
males were not found to be price- and Virginia tobacco and cigars period the volume of smuggled
responsive but young adult females were more price sensitive than black cigarettes in Canada grew rapidly.
were (see Chapters 5 and 6, which tobacco. Virginia and black tobacco The existence of a large illegal market
consider cross-sectional studies were found to be substitutes in is likely to result in overstated price
with similar themes). Chetwynd et al. consumption, as were black tobacco elasticities (in absolute terms). Lanoie
(1988) began a series of important and cigars, although Virginia tobacco and Leclair estimated two models,
studies in New Zealand. The study and cigars were complements in one with legal cigarette consumption
was concerned with examining the consumption. as the dependent variable and the
relationship between print advertising Although using a far simpler second using an estimation of total
and consumption but estimated methodology, Mindell and Whynes cigarette consumption by adding
price elasticities ranging between (2000) also estimated product estimates of the volume of illegal
í DQG í DV D FRQWURO LQ substitution although between cigarettes to the volume of legal
the model. Harrison et al. (1989) manufactured cigarettes and hand- cigarettes. The elasticities using only
was an extension to address the rolled cigarettes in the Netherlands. the legal consumption were found to
methodological concerns raised, and Although cross-price elasticities be significantly greater in absolute
improved the paper by using a myopic were negative, this was explained terms (both the short-run and long-
addiction model estimating a smaller by the collinearity between the UXQ í YHUVXV í DQG í
SULFH HODVWLFLW\ RI í +DUULVRQ prices for manufactured and hand YHUVXV í UHVSHFWLYHO\ WKDQ WKH
and Chetwynd (1990) made further rolled cigarettes. However, they model using the total consumption.
improvements using quarterly data found that when the price of hand- Gruber et al. (2003) also tackled the
instead of annual data, and estimated rolled cigarettes increased by a issue of the impact of smuggling on
DSULFHHODVWLFLW\RIí greater proportion than the price elasticities in Canada, and found that
Duffy (1991) estimated cigarette of manufactured cigarettes, the elasticities are larger, in absolute
demand using aggregate data by decline in manufactured cigarette terms, when including provinces and
employing a demand system in the consumption was accompanied by years in which smuggling was known
United Kingdom. The Almost Ideal a decline in hand-rolled cigarette WREHSUHVHQWíYHUVXVí
Demand System (AIDS) yielded consumption. Bardsley and Olekalns (1999)
estimates of own price and cross Hondroyiannis and Papapetrou estimated a rational addiction model
price elasticities for cigarettes as well (1997) is one of several Greek studies. in Australia, and while they found
as alcoholic drinks (beer, wine and It is one of the more thorough time support for rational addiction between
spirits). Duffy found that cigarettes series studies in that it estimates a 1963 and 1996, their interesting
and alcoholic drinks exhibited dynamic specification of demand and innovation was how they specified the
no significantly complementary employs an Error Correction Model model in linear terms to investigate
behaviour. Duffy (1995) conducted to deal with the time series issues in how elasticities changed over time.
another AIDS model including the data. The use of advanced time They found that both short- and long-
117
run elasticities in absolute terms background an empirical literature on predicted by Warner (1990), on the
grew significantly over time, meaning the demand for tobacco in developing grounds that cigarettes are generally
that consumption became more countries developed. less affordable in developing
responsive to changes in price. The A chronological summary of countries, given their much lower
short-run elasticity ranged between studies that investigated the demand per-capita income levels (confirmed
íDQGíEHWZHHQDQGWKH for tobacco is provided in Table 4.3. by Blecher and van Walbeek, 2004
early 1980s, but increased rapidly to The first attempt was by Chapman and 2009). Warner (1990) argued
í EHWZHHQ DQG 7KH and Richardson (1990), who used that, like the lower social classes in
long-run elasticity ranged between annual time series data to estimate the United Kingdom and teenagers in
íDQGíLQWKHILUVWSHULRGDQG the response in tobacco demand to a the USA, tobacco users in developing
LQFUHDVHGUDSLGO\WRíLQWKHODWHU change in tobacco excise taxes. They countries have relatively lower
period. Escario and Molina (2004), used cigarette excise taxes as a proxy incomes, and consequently price
however, reported that elasticities for cigarette prices, because the latter increases for goods in their budgets
have, in absolute terms, fallen over were unavailable. They found that the impinge more significantly on their
time in Spain. ³H[FLVHWD[HODVWLFLW\´ZDVDERXWí ability to purchase other goods and
Studies in high-income countries IRUFLJDUHWWHVDQGíIRURWKHUIRUPV services (Warner, 1990).
have found similar results to those in of tobacco. Subsequent studies, Since the mid-1990s, tobacco
the USA. The variation in countries also based on time series data, control research in developing
considered results in some variation estimated price elasticity estimates countries has received substantial
in results, although the general range for Argentina (Gonzáles-Rozada, financial and institutional support from
is similar. 2006), Bangladesh (Ali et al., 2003), organizations including Research for
Bolivia (Alcaraz, 2006), Brazil (Da International Tobacco Control, the
Low- and middle-income countries Costa e Silva, 1998, Iglesias, 2006), Tobacco-Free Initiative of the World
Chile (Debrott Sanchez, 2006), China Health Organization and the World
As pointed out in the introduction to (Hu and Mao, 2002, Yuanliang & Bank. More recently, support has
this chapter, tobacco use is shifting Zongyi, 2005), Taiwan, China (Hsieh come from the Bloomberg Initiative
from the developed to the developing et al., 1999, Lee, 2007, Lee and Chen, to Reduce Tobacco Use via the
world. Before 1990 the economics 2008), Egypt (Hanafy et al., 2011), Campaign for Tobacco Free Kids
of tobacco control in developing Estonia (Taal et al., 2004), Indonesia and the International Union Against
countries received practically no (Djutaharta et al., 2005), Malaysia Tuberculosis and Lung Disease as
attention from either policymakers (Ross and Al-Sadat, 2007), Morocco well as the Bill and Melinda Gates
or academic researchers. The fall of (Aloui, 2003), Poland (Florkowski Foundation. These organizations
communism and rapid globalization and McNamara, 1992), South Africa realized that there was a need for
created opportunities for multinational (Reekie, 1994, Van Walbeek, 1996, country-specific analytic work with
cigarette companies to diversify Economics of Tobacco Control in a strong policy focus (De Beyer &
their markets into a rapidly growing South Africa Project (ETCSA),1998, Waverley Brigden, 2003). As is to be
developing world, particularly in Boshoff, 2008), Republic of Korea expected, policymakers in developing
eastern Europe and Asia. The USA (Wilcox et al., 1994, Kim and Seldon, countries were unwilling to impose
used the threat of trade sanctions to 2004), Turkey (Tansel, 1993, Onder, tobacco control policies in their
prise open the markets in Thailand, 2002, Yurekli et al., 2010), Ukraine countries solely on the grounds that
Japan, the Republic of Korea and (Peng and Ross, 2009), Uruguay they were successful in developed
Taiwan, China to foreign cigarettes (Ramos, 2006) and Zimbabwe countries. They wanted research that
(Chaloupka and Laixuthai, 1996). (Economics of Tobacco Control in took cognisance of the uniqueness of
Developing countries did not have South Africa Project 1998). Guindon their countries.
effective tobacco control policies et al. (2003) performed a multi-country The research performed under the
in place, and presumably many study in Bangladesh, Indonesia, auspices of these organizations
developing countries did not see the Maldives, Myanmar Nepal, Sri Lanka attempted to address such
need for measures that were imposed and Thailand. policymakers’ concerns. Countries
in rich industrialised countries but The fact that cigarette demand in that were investigated in this research
that were deemed unnecessary in developing countries is more elastic drive and that used aggregate
developing countries. Against this than in developed countries was data in the analysis included
118
Table 4.3. Studies employing aggregate data in low- and middle-income countries

Chapman and Papua New Guinea -0.50 for non- 1.37 for non-cigarette Price of substitutes Log linear static OLS Annual time series First study conducted
Richardson (1990) cigarette tobacco tobacco (positive), from 1973 to 1986 in a low-and-middle-
-0.71 for cigarettes 0.86 for cigarettes Trend (negative) income country.
(Important: these are Excise tax data,
“excise elasticities”: rather than price
see Comments) data, were used
because of data
unavailability;; as
a result the price
elasticities are larger
than the “excise
elasticities” estimated
in the study
Florkowski and Poland -0.11 0.43 Price of substitutes, Linear static SUR Annual time series System included
McNamara (1992) urban/rural system data from 1959 to vodka, wine and beer
population, trend for 1985
addiction
Tansel (1993) Turkey Short-run Short-run Anti-smoking Myopic addiction Annual time series The first study to
-0.21 0.41 campaign dummy model data from 1960 to look at the time
Long-run Long-run variable (negative), 1988 series effects in a
-0.37 0.71 health warning developing country
dummy variable
(negative)
Reekie (1994) South Africa -0.88 0.59 Advertising Log linear static OLS Annual time series The demand
expenditures data from 1970 to equation was used to
(insignificant) 1989 estimate the size of
the consumer surplus
Wilcox et al. (1994) Republic of Korea Significantly negative Significantly positive Advertising Linear static stepwise Monthly time series Regressions on
(but elasticity (but elasticity expenditure (generally OLS data from July 1988 aggregate and brand-
estimate not shown) estimate not shown) insignificant), to April 1992 level data
population
(insignificant),
health warnings
(insignificant)
Van Walbeek (1996) South Africa Short-run Short-run Time trend Linear dynamic OLS Annual time series
–0.32 and –0.99 0.48 to 0.58 data from 1960 to
Long-run 1990
–0.53 and –1.52
Da Costa e Silva Brazil Short-run Short-run Unknown Rational addiction Annual time series Lack of explanation
(1998) between –0.11 and between 0.23 and data from 1983 to precludes evaluation
–0.35 0.31 1994 and unable to
Long-run Long-run replicate
between –0.48 and between 0.76 and
–0.80 0.80
119
120

Economics of South Africa Short-run Positive Advertising Myopic addiction Annual time series Model included a
Tobacco Control in between –0.57 and Long-run income expenditure (positive) Error Correction data from 1970 to supply side equation
South Africa Project –0.59 elasticity Model and long-run 1994
(1998) Long-run 1.70 cointegrating VAR
-0.69
Zimbabwe Short-run Positive Rational addiction Annual time series

-0.52 Long-run income OLS data from 1970 to
Long-run elasticity 1996

-0.85 1.67
Hsieh et al. (1999) Taiwan, China Between –0.6 and Positive Market share of Annual time series Quoted in Chaloupka
–0.7, depending on low tar cigarettes, data from 1966 to et al. (2000b)
specification female labour force 1995
participation rate,
market share of
imported cigarettes
Between -0.56 Between 0.22 and Dependent variable Dynamic OLS, 2SLS Annual times series Domestic and
and -0.73 for local 0.26, but centring on = per capita (adult, and SUR data from 1966 to imported cigarettes
brands;; 0.22 for local brands;; 15+) consumption;; 1995 are considered
Between -1.08 and around 1.42 for Independent substitutes
-1.28 for imported imported brands. variables: Real price,
brands subdivided into
local and imported
brands;; market share
of low-tar brands
as proxy for health
awareness;; dummy
variable (1991-1995
= 1) to account for
warning labels;;
market share of
imported cigarettes;;
female labour force
participation rate
Hu and Mao (2002) China Short-run Positive Time trend Dynamic semi log Annual time series
-0.35 data from 1980 to
Long-run 1997
-0.66
Onder (2002) Turkey Static Static Regulation index (not Log linear static OLS Annual time series
OLS OLS significant), and GLS data from 1960 to
-0.31 to -0.41 0.27 to 0.29 Trend (not 2000
GLS GLS significant),
-0.19 to -0.28 0.23 to 0.25 Price of substitute
Dynamic Dynamic cigarettes (varying)
-0.09 to -0.12 0.10 to 0.17
Ali et al. (2003) Bangladesh -0.27 (but not 0.62 None Log linear static OLS Annual time series
statistically data from 1983 to
significant) 1999
Aloui (2003) Morocco Short-run Short-run Dummy variable to Myopic addiction Annual times series
between –0.51 and between 0.32 and indicate tobacco linear and log linear data from 1965 to
–0.73 0.56 control legislation OLS 2000
Long-run Long-run (insignificant)
–1.54 1.04
Guindon et al. (2003) Bangladesh, Elasticities based Elasticities based Appropriate dummy Various techniques: Annual time series
Indonesia, Nepal, on conventional on conventional variables to account OLS, GLS and 2SLS, data from 1970 to
Sri Lanka, Thailand, demand demand for political crises controling for time- 2000
Maldives and specification specification specific and period-
Myanmar between –0.60 and between 0.28 and specific effects
–0.90 0.99
Elasticities based Elasticities
on myopic addictive based on myopic
model addiction model
Short-run Short-run
–0.65 1.03
Long-run Long-run
–1.40 1.72
Kim and Seldon Republic of Korea Short-run Insignificant Health warnings Myopic addiction log Annual time series
(2004) -0.28 linear OLS data from 1960 to
Long-run 1997
-0.35
Taal et al. (2004) Estonia Short-run Short-run Quarterly dummies, Myopic addiction Monthly time series
-0.32 to -0.34 0.09 to 0.18 (not time trend model Log linear OLS data from 1992 to
significant) 1999 and annual
time series data from
1993 to 2000
Djutaharta et al. Indonesia Annual data Annual data Economic crisis Log linear static OLS Monthly time series Tested for
(2005) -0.35 on preferred 0.47 on preferred dummy (1997-2001 data from January endogeneity of price
model;; for other model;; for other = 1);; trend;; health 1996 to June 2001
specifications specifications warning dummy and annual time
elasticity varies from elasticity varies from (1991-2001 = 1);; for series data from 1971
-0.33 to -0.47 0.14 to 0.51 monthly analysis to 2001
Monthly data Monthly data similar variables were
-0.32 on preferred no specification used
model;; for other yielded significant
specifications results
elasticity varies from
-0.32 to -0.43
121
122

Yuanliang and Zongyi China -0.84 0.90 Education, regional Static OLS with fixed Pooled annual data Cross-sections are
(2005) dummies (fixed effects from 1997 to 2002 provincial and spatial
effects) average municipalities
propensity to
consume
Alcaraz (2006) Bolivia [-0.78 to -2.18] [0.56 to 0.71] Some dummies Static, dynamic and Annual time series Elasticity values in
rational addiction data from 1988 to [ ] computed at the
models. Some IV 2002 mean by the Working
techniques Group
Debrott Sanchez Chile Short-term Short-term Trend variable, Myopic addiction Quarterly time series
(2006) -0.22 0.11 seasonal dummy model data from 1993 (Q1)
Long-term Long-term variable and a dummy to 2003 (Q4)
-0.45 0.22 reflecting tobacco and
health information
campaigns
Iglesias (2006) Brazil Short-run 0.05 to 0.36 Trend variable, Log linear OLS Quarterly time series Tests endogeneity
-0.25 smoking restriction estimation. Dynamic data from 1991 to of price
Long-run index, seasonal myopic addiction 2003
-0.42 dummy
Gonzáles-Rozada Argentina -0.23 to -0.27 0.45 to 0.49 Some dummies Log linear dynamic Monthly time series
(2006) model. Cointegrating data from Jan 1996
VAR to June 2004
Ramos (2006) Uruguay Short-term Short-term Real bilateral Log linear myopic Quarterly time series
-0.34 to -0.49 0.51 to 0.65 exchange rate with addiction model data from 1991 to
Long-term Long-term Argentina, Brazil and using IV, SURE, 2003
-0.39 to -0.55 0.59 to 0.73 Paraguay 2SLS, 3SLS
Lee (2007) Taiwan, China -0.66 Expenditure, Log linear SUR Annual time series
substitutes (alcohol demand system data from 1972 to
and betel nuts) 2002
Ross and Al-Sadat Malaysia Short-run Short-run Tobacco control Dynamic log linear Annual time series
(2007) -0.08 0.03 policies (negative ECM data from 1990 to
Long-run Long-run and significant) 2004
-0.54 to -0.76 1.40 to 1.5
Boshoff (2008) South Africa Short-run elasticities Long-run Market share of light Dynamic VAR Annual times series
not estimated 0.37 to 0.73 cigarettes, smoking data from 1996–2006
Long-run restrictions dummy
-0.16 to -0.66
Lee and Chen (2008) Taiwan, China -0.49 to 0.81 Expenditure, Log linear SUR Annual time series
substitutes (imported, demand system data from 1987 to
domestic cigs and 2000
cigars), smoking risk
information
Turkey, Morocco, China, Egypt and Hsieh et al.’s (1999) Taiwan,
ECM, error correction models estimates;; GLS, generalized least squares estimates/estimator;; GLS-AR, generalized least squares - auto regression process;; IV, instrumental-variable estimation/estimator;; OLS, ordinary least squares
Bangladesh.12 China study was also novel in that it
Some studies in low- and middle- estimated separate price and income
Comments
income countries are of specific elasticities for domestic and imported

interest. South Africa has contributed cigarettes. They found that consumers
estimates;; SUR, seemingly unrelated system of equations;; SURE, Spanish acronym for SUR;; VAR, vector autoregression;; 2SLS, 2-stage least squares estimates;; 3SLS, 3-stage least squares estimates/estimator
the largest number of studies to of imported cigarettes were more
the literature (Reekie, 1994, Van price responsive than consumers of
Monthly time series
Walbeek, 1996, Economics of domestic cigarettes.

Annual time series
Annual time series

data from January
data from 1960 to
data from 1990 to

1997 to May 2006
Tobacco Control in South Africa This should be expected since

Project (ETCSA), 1998, Boshoff, imported cigarettes are more
2008), and most are noteworthy. expensive than domestic cigarettes.
2006
2006
Data
Reekie’s (1994) uniqueness lies in the The estimated price elasticities for
attempt to use the demand estimation imported cigarettes ranged between
Static log linear GLS-
Log linear static GLS
Log linear static OLS
to calculate the consumer’s surplus. í DQG í ZKLOH WKH SULFH
Estimation method
He estimated the price elasticity and elasticities for domestic cigarettes

based on this, together with some UDQJHG EHWZHHQ í DQG í
AR process
fairly strong assumptions about Domestic and imported cigarettes

For quarterly data, corresponding quarter specified in parenthesis.Values in [ ] calculated by the Working Group using data reported in corresponding studies.
the shape of the demand function, were also found to be substitutes in

was able to derive a value for the consumption.
consumer surplus. Van Walbeek Two recent studies in Taiwan,
Economic crisis, time
prohibiting smoking
(1996) used a demand model to China employed a systems approach

Control variables
Dummy for Fatwa
calculate the revenue-maximizing to understand the substitutive

Advertising
excise tax rate of a cigarette- behaviour between cigarettes,

specific Laffer curve. Finding the alcohol and betel nuts (Lee, 2007)
trend
price elasticity to range between and between imported cigarettes,

í DQG í GHSHQGLQJ RQ WKH domestic cigarettes and cigars (Lee
data source), he suggested that the and Chen, 2008). Lee (2007) found
Estimated income
government maximize its tax revenue both alcohol and betel nuts to be
by raising the excise tax so that it complements in consumption to
elasticity
occupied 46% of the retail price. cigarettes. Lee and Chen (2008) found
Boshoff’s (2008) study is the most that imported and domestic cigarettes
0.30
0.56
1.60
methodologically sound of any study are substitutes, as expected. Imported
in a low- or middle-income country, cigarettes and cigars were also found
applying a dynamic specification to be substitutes in consumption,
Estimated price
of demand and estimating this while domestic cigarettes and cigars

-0.33 to -0.44
Insignificant
specification using a cointegrating were found to be complements in

elasticity
Vector Autoregressive (VAR) consumption.

-0.47
model (one of the most advanced The primary aim of most of these
time series techniques available). studies was to estimate the price
However, this technique only yields a elasticity of demand. The elasticity
long-run rather than the conventional estimates varied significantly from
short-run price elasticity, ranging one country to another, but as was
Country
EHWZHHQ í DQG í $OWKRXJK the case with the earlier studies on
Ukraine
Turkey
Egypt
many studies in low-and-middle- the demand for tobacco in developing

income countries now employ countries, they practically all found
dynamic specifications, few consider a relatively inelastic demand for
Hanafy et al. (2011)
Yurekli et al. (2010)
the time series properties of the data cigarettes.

Peng and Ross
and apply appropriate techniques to

account for these properties.
(2009)
Study
These studies were, respectively, by Onder (2002) and Yurekli et al. (2010), Aloui (2003), Hu and Mao (2002), Hanafy
12
et al. (2011) and Ali et al. (2003).
123
Figure 4.3. Price elasticity estimates for low- and middle-income countries, based on time series studies.
White squares are dynamic studies and solid black squares are static studies.
Sources: Figure generated by the Working Group.

Note: The studies referred to in the figure can be found in Table 4.3. The table only includes short-run elasticities. The years refer to the middle year under investigation, hence the two
most recent studies (Hanafy et al., 2011, and Yurekli et al., 2010) include data from 1990 to 2006 and 1960 to 2006 respectively and thus they are listed in 1998 and 1993 respectively.
Figure 4.3 graphs all price estimates we find a narrower range income over time. Almost all of these
elasticity studies in developing RI HVWLPDWHV EHWZHHQ í DQG í studies find that tobacco is a normal
countries. It separates those studies (there is one outlier which is ignored). good in that income has a positive
which assume static specifications First, it is important to again confirm impact on consumption, although
of demand (i.e. which do not control that all estimates indicate that this impact is not always statistically
for the addictiveness of tobacco), tobacco is price-inelastic. Second, significant. The three figures below
represented by the solid black squares the range of elasticities differs indicate the income elasticities
from those which assume dynamic somewhat between the static and estimated in the studies considered
specifications of demand (i.e. which dynamic specifications, with dynamic in the previous section and their
control for the addictiveness of specifications resulting in a tighter evolution over time, Figure 4.4 for
tobacco), represented by the white range than static specifications. those studies in the USA, Figure
squares. Again, the larger data 4.5 for other high-income countries
points represent a longer period Income elasticities and Figure 4.6 for low- and middle-
under consideration in the study. income countries.
The shortest time period is 5 years Most studies discussed have Almost all studies in the United
and the longest 46 years. When estimated income elasticities (the States find that the income elasticity
looking at only the static estimates, responsiveness of consumption as a is positive and thus cigarettes
we find that the price elasticity of result of a change in income). While are a normal good. Prior to 1970
demand varies over a wide range this is generally not the intention of the range of estimates was fairly
EHWZHHQ í DQG í +RZHYHU the studies, it is methodologically wide (between 0 and 1.2) but has
when including only the dynamic important to control for changes in narrowed significantly since. Only
124
Figure 4.4. Income elasticity estimates for the USA, based on time series studies
Sources: Figure calculated by the Working Group. The studies referred to in the figure can be found in Table 4.1. The table only includes short-run elasticities. The years refer to the
middle year under investigation, hence the most recent study (Goel, 2009) includes data from 1975 to 2004 and is thus listed in 1989/90.
four studies found, in some but not From Figure 4.5 one can see that, All estimates of the income
all specifications, that the income for the most part, tobacco is a normal elasticity of demand in developing
elasticity is negative (Schmalensee, good in other developed countries. countries are positive, indicating that
1972, Porter, 1986, Baltagi et al., 2000, The figure separates the estimates for tobacco is a normal good (see Figure
Goel, 2009). Only one study (Porter, the United Kingdom (white squares) 4.6). Since developing countries are
1986) finds that the income elasticity and other developed countries (solid poorer and in an earlier stage of
is negative in all specifications. black squares), and shows that all the tobacco epidemic (Lopez et al.,
Figure 4.4 provides some evidence estimates in the United Kingdom are 1994), one would expect increases
that income elasticities in the USA positive. In other developed countries in income to be associated with
have fallen over time, indicating negative estimates are found in two greater tobacco consumption.
that consumption has become less countries: New Zealand (Chetwynd When one considers estimates that
sensitive to changes in income. et al., 1988, Harrison et al., 1989) and employed a static specification of
This is not an unexpected result Italy (Pierani and Tiezzi, 2009). It is demand (i.e. not accounting for the
since higher incomes are associated not possible to generalize the trends addiction of tobacco, indicated by
with higher education, and higher in elasiticites over time in developed solid black squares) we find a broad
education in itself is associated with countries although one may say range of results between 0.2 and 1.0
lower consumption (presumably with a fair degree of confidence that and two higher outliers. However,
through a better understanding of the tobacco is a normal good in developed when one considers only dynamic
health consequences of smoking) countries and that income elasticities specifications (i.e. accounting for
(Kenkel and Chen, 2000). tend to lie between 0 and 1. addiction through a myopic or rational
addiction model – white squares) we
125
Figure 4.5. Income elasticity estimates for other high-income countries, based on time series studies.
White squares are United Kingdom studies and solid black squares are other high-income countries.
Sources: The studies referred to in the figure can be found in Table 4.2. The table only includes short-run elasticities. Estimates for several countries come from a single study by
Koutsoyiannis (1963) and result in a grouping on the midpoint 1954/1955. The years refer to the middle year under investigation, hence the most recent study (Pierani and Tiezzi, 2009)
includes data from 1960 to 2002 and is thus listed in 1981.
find estimates between 0 and 0.6. variable. Even if one knew by and cigarette price, whereas
However, since many developing how much cigarette consumption conventional studies consider
countries are growing rapidly, large changes in response to a change the effect of price and income in
increases in tobacco consumption in income, few people would argue isolation. One can investigate the
are likely to occur in a relatively against economic growth on the level of affordability (in a cross-
short period of time. For example, grounds that it would increase the sectional context) or inter-temporal
annual GDP growth of 8% may result demand for cigarettes. changes in affordability. Affordability
in an annual increase in tobacco In recent decades some refers to the quantity of resources
consumption of 4.8% (assuming an countries, mainly in Asia, have (not exclusively monetary) required
income elasticity of 0.6). Given the achieved rapid economic growth to purchase tobacco products.
impact of compounding, this could rates. In China, India, Indonesia, A limited number of published
result in a doubling of consumption in Viet Nam and Bangladesh, real studies have explicitly investigated
15 years. per-capita gross domestic product the affordability of cigarettes
(GDP) has grown at annual rates (Scollo, 1996;; Lal and Scollo, 2002;;
Affordability of 6% or more (Blecher and van Guindon et al., 2002;; Blecher and
Walbeek, 2009). The literature van Walbeek, 2004,2009;; Kan,
The studies reviewed typically do not shows that as incomes rise, so 2007). Guindon et al. (2002) and
focus on the impact of changes in does the consumption of tobacco Kan (2007) define affordability as
income on the demand for cigarettes. products. the time worked to purchase a pack
Most studies include income in the Affordability considers the of cigarettes. Guindon et al. (2002)
demand specification as a control simultaneous effect of income found that cigarettes became less
126
Figure 4.6. Income elasticity estimates for low- and middle-income countries, based on time series studies.
White squares are dynamic studies and solid black squares are static studies.
Sources: Figure calculated by the Working Group. The studies referred to in the figure can be found in Table 4.3. The table only includes short-run elasticities. The years refer to the
middle year under investigation, hence the two most recent studies (Hanafy et al., 2011, and Yurekli et al., 2010) include data from 1990–2006 and 1960–2006 respectively;; thus they
are listed in 1998 and 1993 respectively.
affordable in most countries studied countries. Furthermore, they found of their affordability. Fast-growing
(both high-income and low- and that cigarette affordability is inversely countries face greater tobacco control
middle-income countries) between related to consumption, and that the challenges, since rising incomes
1990 and 2000. Kan, focusing on affordability elasticity of demand increase the affordability of cigarettes.
cities rather than countries, and on LV DERXW í 7KLV LV VLPLODU WR WKH The fact that cigarettes have become
lower-income occupations, came to consensus price elasticity estimates increasingly affordable in most low-
a similar result. in the previous sections. In a income and middle-income countries
Blecher and van Walbeek subsequent study, they updated their is a major tobacco control failure.
(2004) considered a larger sample earlier study and found that cigarettes A more recent paper by
of countries, most of which were have become more affordable in low- Blecher (2010) uses the concept
low- and middle-income countries. and middle-income countries at an of affordability to asses the risks
They defined affordability in terms of increasingly rapid rate since 2000. In of aggressive economic growth on
per-capita gross domestic product almost all countries where cigarettes tobacco consumption. Using the case
(GDP), and found that cigarettes became less affordable, the real price study of South Africa he proposes
were generally more affordable in increased;; in most countries where that cigarette taxation be linked to
high-income countries relative to cigarettes became more affordable, changes in affordability rather than
low- and middle-income countries. real price decreased. simply targeting the price or excise
They found that during the 1990s The implication of international tax incidence (the percentage of
cigarettes became less affordable comparisons is that cigarette the retail price that is taken up by
in high-income countries and more prices should not only be viewed in excise). He also points out that
affordable in low- and middle-income monetary terms but also in terms simply maintaining affordability will
127
result in increases in consumption The first cross-country study to consumption had occurred. To test this
and that cigarettes must become analyse the impact of advertising hypothesis, he estimated a two-stage
less affordable to maintain or restrictions was by Laugesen and model treating advertising restrictions
reduce consumption. Furthermore, Meads (1991), who investigated the endogenously. Nelson found the price
this is especially important when impact of advertising restrictions on HODVWLFLW\ WR UDQJH EHWZHHQ í
taking cognisance that tobacco cigarette consumption in 22 OECD DQG í DQG WKH LQFRPH HODVWLFLW\
consumption is more responsive countries (all developed countries). to range between 0.09 and 0.28.
to increases in income in low- and They found that adverting restrictions However, he did not find advertising
middle-income countries relative to significantly reduced tobacco restrictions to be significant although
high-income countries. consumption, and that the impact the consistent negative coefficients
Other recent applications of of these restrictions became more decreased over time, indicating
affordability include the tax reports pronounced after 1970. Furthermore, that advertising restrictions have
commissioned by the Bloomberg they found cigarettes to be price- become less important in determining
Initiative to Reduce Tobacco Use inelastic, and a global price elasticity consumption over time. Nelson was
(see for example Hu et al., 2008 RIíDQGLQFRPHHODVWLFLW\RI unable to reject the null hypothesis
or Guindon et al., 2010), the WHO in line with the single country studies that advertising bans and restrictions
MPOWER package (World Health in developed countries. In response, were exogenous.
Organization, 2008) and the Tobacco Stewart (1993) performed a similar Blecher (2008) advanced the
Atlas (Shafey et al., 2009). study to that of Laugesen and Meads literature further by including 30
(1991) and found that advertising developing countries in addition to
Cross-country studies bans did not have a significant 21 developed countries in the cross-
impact on cigarette consumption. A section of countries. Blecher followed
Some studies have not focused on global price elasticity of demand of the Saffer and Chaloupka (2000)
an individual country but rather on a í ZDV HVWLPDWHG DJDLQ LQ OLQH model of analysing both limited
group of countries. countries (Table with single-country studies in high- and comprehensive advertising
4.4). Such pooled or cross-country income countries) while income was bans, finding that advertising bans
studies have generally, although excluded from the specification. reduce consumption and specifically
not exclusively (Gallus et al., 2006), Economists have noted that that comprehensive bans are
been used to consider the impact partial advertising bans are relatively more effective than limited bans.
of advertising bans on tobacco ineffective in reducing tobacco Furthermore, Blecher found the
consumption. Gallus et al. (2006) consumption, while comprehensive price elasticity of demand to range
provides a cross-sectional view of bans seem to be much more EHWZHHQ í DQG í DQG
demand in Europe (including 52 effective (Jha and Chaloupka, 1999). income elasticities between 0.07 and
countries), finding price to be inelastic Using data from 22 OECD countries, 0.21. The absolute magnitudes of the
ZLWKHVWLPDWHVUDQJLQJIURPíWR Saffer and Chaloupka (2000) found elasticities were significantly smaller
íZLWKKLJKHUDEVROXWHHODVWLFLWLHV that the imposition of comprehensive in the Blecher study compared to the
in the non-European Union countries advertising bans would reduce other studies.
í YHUVXV í 7KLV IDOOV LQ OLQH cigarette consumption while partial All of these studies estimated
with the generalizations of the results advertising bans did not have a a global elasticity rather than
in individual country studies. Even significant impact. A global price individual country elasticities. This
though the study uses aggregate elasticity of demand was estimated may be appropriate in high-income
data, the interpretation of the WROLHEHWZHHQíDQGíZLWK countries but when low-and-middle-
elasticity is not strictly comparable income having a positive impact on income countries are included
with the time series studies. Income consumption. this may not be appropriate due to
is excluded from the analysis rather Nelson (2003) used a cross- larger inter-country differences in
controlling for purchasing power country model to test whether low- and middle-income countries.
parity in the price, and the model is advertising bans are endogenously The use of a model that does not
estimated using a static specification. determined with consumption. Nelson impose homogeneity and a global
Furthermore, annual data are from indicates that the most comprehensive elasticity might result in estimates
different years. advertising bans only came into that fall within the expected range.
being once large-scale decreases in
128
Table 4.4. Cross-country studies employing aggregate data

Study Countries Control variables Estimation method Data Comments
Laugesen and Meads 22 OECD countries -0.20 0.28 Advertising Static pooled log Pooled annual data The advertising
(1991) restrictions score linear GLS from 1960 to 1986 restriction score is
(negative), problematic as well
Female labour as the pooling of the
participation rate data with no fixed
(negative), effects. Thoroughly
Manufactured critiqued by Stewart
cigarettes as fraction (1992 and 1993),
of total tobacco High (1999), Saffer
consumption & Chaloupka (2000),
(positive) Blecher (2008).
Stewart (1993) 22 OECD countries All countries Not included Advertising ban Static pooled non- Pooled annual data In response to
-0.31 for middle year, (insignificant), linear OLS from 1964 to 1990 Laugesen and Meads
but was decreasing Quadratic trend (1991). Thoroughly
(in absolute terms) (varied for different critiqued by Saffer
over time countries) & Chaloupka
Individual countries (2000), and by
Austria Blecher (2008).
(-0.34) The advertising
Australia restrictions are
(insig.) accounted for poorly
Belgium and the method is
(-0.61) unreplicable. Was
Canada paid by the industry
(-0.37) and did not disclose
Denmark funding
(-0.29)
Finland
(-0.45)
France
(-0.23)
Greece
(-0.35)
Iceland
(-0.32)
Ireland
(-0.30)
Italy
(-0.39)
Japan
(-0.18)
129
130
Table 4.4. Cross-country studies employing aggregate data

Study Countries Control variables Estimation method Data Comments
Stewart (1993) Netherlands

(contd) (-0.69)
New Zealand
(-0.25)
Norway
(-0.49)
Portugal
(insig.)
Spain
(-0.16)
Sweden
(-0.45)
Switzerland
(-0.83)
UK
(-0.55)
US
(-0.29)
West Germany
(-0.54)
Saffer and 22 OECD countries -0.41 to 0.55 Positive (significant in Advertising Static log linear static Pooled annual data Categorizes
Chaloupka (2000) some specifications) bans (negative), fixed effects panel from 1970 to 1992 advertising bans
unemployment rate, OLS as weak, limited or
percent of filter comprehensive
cigarettes, time fixed
effects
Nelson (2003) 20 OECD countries -0.26 to -0.44 0.09 to 0.28 Advertising Static log linear fixed Pooled annual data Tested endogeneity
bans, warnings, effects panel OLS from 1970 to 1995 of advertising bans
unemployment rate,
percent filtered
cigarettes
Gallus et al. (2006) 52 European -0.40 to -1.00 (higher Not included Male to female Static log linear Cross sectional data Income is excluded
countries in non-EU member smoking prevalence pooled OLS for 2000 or nearest from the model in
countries) ratio (generally year order to calculate a
insignificant) GDP-PPP adjusted
price (essentially
a measure of
affordability)
Blecher (2008) 51 countries -0.09 to -0.13 0.07 to 0.21 Advertising bans Static log linear fixed Pooled annual data Includes 30 low-
(negative) effects panel OLS from 1990 to 2003 and middle-income
countries
GLS, generalized least square estimates;; OLS, ordinary least square estimates
Country differences include differing The panel models do, but also estimating dynamic panel models
demographics, availability of make assumptions regarding where an instrumental variable
substitutes and different tobacco homogeneity. Baltagi et al. (2000) technique is required.
products, tax structures and tested this homogeneity assumption Advertising bans comprises
affordability. in a panel model using US states, more than just bans on advertising,
The cross country literature is at finding that homogenous estimators but other types of marketing
a relatively early stage;; thus most of are preferred to heterogeneous including sponsorships. Measures of
the studies have varied significantly estimators. However, US states are advertising bans do not capture price-
in methodology. Laugesen and more similar than different countries. based marketing measures including
Meads (1991) and Stewart (1993) Cross-country differences may not price discounts, price promotions,
employed pooled models including be significant in the studies that only multiple pack purchases, etc. The
both cross-sections and time series. include OECD countries, but this is global price and income elasticity
Saffer and Chaloupka (2000), likely to pose a far more significant estimates fall into the same ranges as
Nelson (2003) and Blecher (2008) problem in studies that include those in individual country studies in
employed panel models by including both developing and developed high-income countries, although not
country and/or time period fixed countries, or those that include only in low-and-middle-income countries,
effects. Methodologically, the simple developing countries. Furthermore, likely the result of methodological
pooled models are not appropriate all the cross-country models assume and estimation issues rather than
since they do not control for the static specifications of demand, most fundamental differences.
vast differences between countries. likely as a result of the difficulties of
131
References
Abernethy AM, Teel JE (1986). Advertising Bishop JA, Yoo JH (1985). “Healh scare”, Chaloupka FJ, Pacula RL (1998). An
regulation’s effect upon demand for cigarettes. excise taxes and advertising ban in the examination of gender and race differences
J Advert, 15:51–55. cigarette demand and supply. South Econ J, in youth smoking responsiveness to price and
52:402–411 doi:10.2307/1059626. tobacco control policies. NBER Working Paper
Alcaraz VO (2006). Economia del Control del Series. Working Paper#6541. Cambridge, MA,
Tobaco en los paises del Mercosur y Estados Blecher E (2008). The impact of tobacco National Bureau of Economic Research.
Asociados: Bolivia. Washington, Organizacion advertising bans on consumption in
Panamericana de la Salud. developing countries. J Health Econ, 27:930– Chaloupka FJ, Tauras JA, Grossman M
9 42 .d o i:10.1016 / j. j h e a l e c o. 2 0 0 8 . 0 2 . 010 (2000a). The economics of addiction. In:
Ali Z, Rahman A, Rahman T (2003). Appetite PMID:18440661 Jha P, Chaloupka F, eds., Tobacco control in
for nicotine: an economic analysis of tobacco developing countries. Oxford, U.K., Oxford
control in Bangladesh. HNP Discussion Paper Blecher EH (2010). Targeting the affordability of University Press.
Series, Economics of Tobacco Control Paper cigarettes: a new benchmark for taxation policy
No.16. Washington DC, The World Bank. in low-income and-middle-income countries. Chaloupka FJ, Warner KE (1999). The
Tob Control, 19:325–330.doi:10.1136/ Economics of Smoking. In: Culyer AJ,
Aloui O (2003). Analysis of the economics of tc.2009.030155 PMID:20530141 Newhouse JP, eds., Handbook of Health
tobacco in Morocco. HNP Discussion Paper Economics Volume 1B. Oxford, UK, Elsevier:
Series, Economics of Tobacco Control Paper Blecher EH, van Walbeek CP (2004). An 1539–1627.
No.7. Washington DC, The World Bank. international analysis of cigarette affordability.
Tob Control, 13:339–346.doi:10.1136/ Chaloupka FJ, Wechsler H (1997). Price,
Andrews RL, Franke GR (1991). The tc.2003.006726 PMID:15564616 tobacco control policies and smoking among
determinants of cigarette c onsumption: a meta- young adults. J Health Econ, 16:359–373.
analysis. J Public Policy Mark, 10:81–100. Blecher EH, van Walbeek CP (2009). Cigarette d o i : 1 0 .1 0 1 6 / S 0 1 6 7 - 6 2 9 6 ( 9 6 ) 0 0 5 3 0 - 9
affordability trends: an update and some PMID:10169306
Atkinson AB, Skegg JL (1973). Anti-smoking methodological comments. Tob Control,
publicity and the demand for tobacco in the UK. 18:167–175 .d o i:10.113 6 / tc . 2 0 0 8 .0 2 6 6 8 2 Chapman S (1989). The limitations of
Manchester Sch Econ Soc Stud, 41:282. PMID:19179369 econometric analysis in cigarette advertising
studies. Br J Addict, 8 4:1267–1274.doi:10.1111/j.
Auld MC, Grootendorst P (2002). An empirical Boshoff WH (2008). Cigarette demand in 1360-0443.1989.tb00723.x PMID:2513013
analysis of milk addiction. Working paper South Africa over 1996–2006: the role of price,
2001–17. University of Calgary, Department of income and health awareness. S Afr J Econ, Chapman S, Richardson J (1990). Tobacco
Economics. 76:1–16. excise and declining tobacco consumption: the
case of Papua New Guinea. Am J Public Health,
Baltagi B, Levin D (1986). Estimating dynamic Camerer C, Loewenstein G (2002). Behavioural 8 0:5 37– 5 4 0.doi:10. 210 5 /A J PH . 8 0. 5 . 5 37
demand for cigarettes using panel data: the Economics: Past, Present and Future. Division PMID:2327528
effects of bootlegging, taxation and advertising of Humanities and Social Sciences, Caltech,
reconsidered. Rev Econ Stat, 68:148–155 Pasadena;; Department of Social and Decision Chetwynd J, Coope P, Brodie RJ, Wells E
doi:10.2307/1924938. Sciences, Carnegie-Melton University, (1988). Impact of cigarette advertising on
Pittsburgh aggregate demand for cigarettes in New
Baltagi BH, Goel RK (1987). Quasi- Zealand. Br J Addict, 83:409–414.doi:10.1111/j.
experimental price elasticities of cigarette Cameron S (1997). Are Greek smokers 1360-0443.1988.tb00487.x PMID:3155345
demand and the bootlegging effect. Am J Agric rational addicts? Appl Econ Lett, 4:401–402
Econ, 69:750–754 doi:10.2307/1242184. doi:10.1080/135048597355122. Cooley TF, Prescott EC (1976). Estimation in
the presence of stochastic parameter variation.
Baltagi BH, Griffin JM, Xiong W (2000). To Chaloupka FJ (1991). Rational addictive Econometrica, 44:167–184 doi:10.2307/1911
pool or not to pool: homogenous versus behavior and cigarette smoking. J Polit Econ, 389.
hetergeneous estimators applied to cigarette 99:722–742 doi:10.1086/261776.
demand. Rev Econ Stat, 82:117–126 Da Costa e Silva VL (1998). The Brazilian
doi:10.1162/003465300558551. Chaloupka FJ, Grossman M (1996). Price, cigarette industry: prospects for consumtion
tobacco control policies and youth smoking. reduction. In: The economics of tobacco
Bardsley P, Olekalns M (1999). Cigarette and NBER Working Paper Series. Working Paper control: towards an optimal policy mix. Cape
tobacco consumption: have anti-smoking #5740. Cambridge, MA, National Bureau of Town, Applied Fiscal Research Centre,
policies made a difference? Econ Rec, 75:225– Economic Research. University of Cape Town.
240 doi:10.1111/j.1475-4932.1999.tb02452.x.
Chaloupka FJ, Hu TW, Warner KE et al. De Beyer J, Waverley Brigden L (2003).
Bask M, Melkersson M (2004). Rationally (2000b). Taxation of tobacco products. In: Tobacco control policy: strategies, successes
addicted to drinking and smoking? Appl Econ, Jha P, Chaloupka F, eds., Tobacco control in and setbacks. Washington and Ottawa, World
36:373–381 doi:10.1080/0003684041000167 developing countries. Oxford, U.K., Oxford Bank and Research for International Tobacco
4295. University Press: 237–272. Control.
Becker GS, Grossman M, Murphy KM (1994). Chaloupka FJ, Laixuthai A (1996). US trade Debrott Sanchez D (2006). Economia del
An empirical analysis of cigarette addiction. policy and cigarette smoking in Asia. NBER control del tobaco en los paises del Mercosur
Am Econ Rev, 84:396–418. Working Paper Series. Working Paper#5543. y Estados Asociados: Chile. Washington DC,
Cambridge, MA, National Bureau of Economic Organizacion Panamericana de la Salud.
Becker GS, Murphy KM (1988). A theory of Research.
rational addiction. J Polit Econ, 96:675–700
doi:10.1086/261558.
132
Djutaharta T, Viriya-Surya H, Haidy N et al. Goel RK (2009). Cigarette prices and illicit drug Hanafy K, Saleh ASE, Elmallah MEBE et al.
(2005). Aggregate analaysis of the impact use: is there a c onnection? Appl Econ, 41:1071– (2011). The Economics of Tobacco Taxation
of cigarette tax rate increases on tobacco 1076 doi:10.1080/00036840601019141. in Egypt. Paris: International Union Against
consumption and government revenue: the Tuberculosis and Lung Disease.
case of Indonesia. HNP Discussion Paper Gonzáles-Rozada M (2006). Economia del
Series, Economics of Tobacco Control Paper control del tobaco en los paises del Mercosur Harrison R, Chetwynd J (1990). Determinants
No.25. Washington DC, The World Bank.. y Estados Asociados: Argentina: 1996–2004. of aggregate demand for cigarettes in New
Washington DC, Organizacion Panamericana Zealand. Report No. Discussion Paper 9002.
Duffy M (1991). Advertising and the de la Salud. Canterbury, New Zealand, Department of
consumption of tobacco and alcohol drink: Economics, University of Canterbury.
a system-wide analysis. Scott J Polit Econ, Grossman M, Chaloupka FJ, Sirtalan I
38:369–385 doi:10.1111/j.1467-9485.1991. (1998). An empirical analaysis of alcohol Harrison R, Chetwynd J, Brodie RJ
tb00325.x. addiction: results from the monitoring (1989). The influence of advertising on
the future panels. Econ Inq, 36:39–48 tobacco consumption: a reply to Jackson
Duffy M (1995). Advertising in demand doi:10.1111/j.1465-7295.1998.tb01694.x. and Ekelund. Br J Addict, 84:1251–1254
systems for alcoholic drinks and tobacco: a doi:10.1111/j.1360-0443.1989.tb00720.x.
comparative study. J Policy Model, 17:557–577 Gruber J, Köszegi B (2001). Is addiction “rational”?
doi:10.1016/0161-8938(95)00020-8. Theory and evidence. Q J Econ, 116:1261–1303 High H (1999). Does advertising increase
doi:10.1162/003355301753265570. smoking? Economics, free speech and
Duffy M (2006). Tobacco c onsumption and policy advertising bans. London: The Institute of
in the United Kingdom. Appl Econ, 38:1235– Gruber J, Köszegi B (2002). A theory of Economic Affairs.
1257 doi:10.1080/00036840500392599. government regulation of addictive bads:
optimal tax levels and tax incidence for Holak SL, Reddy SK (1986). Effects of a
Economics of Tobacco Control in South Africa cigarette excise taxation. NBER Working Paper television and radio advertising ban: a study
Project (1998). The economics of tobacco Series. Working Paper #8777. Cambridge, MA, of the cigarette industry. J Mark, 50:219–227
control in South Africa. Report submitted to National Bureau of Economic Research. doi:10.2307/1251297.
the International tobacco Initiative. Cape Town,
Applied Fiscal Research Centre, School of Gruber J, Mullainathan S (2002). Do cigarettes Hondroyiannis G, Papapetrou E (1997).
Economics, University of Cape Town. make smokers happier? NBER Working Paper Cigarette consumption in Greece:
Series. Working Paper #8872 Cambridge, MA, empirical evidence from cointegration
Enders W (2004). Applied econometric time National Bureau of Economic Research. analysis. Appl Econ Lett, 4:571–574
series. John Wiley and Sons, Inc. doi:10.1080/135048597355050.
Gruber J, Sen A, Stabile M (2003). Estimating
Escario J, Molina J (2004). Modeling the price elasticities when there is smuggling: the Hsieh CR, Hu TW, Lin CFJ (1999). Demand for
optimal fiscal policy on tobacco consumption. sensitivity of smoking to price in Canada. J cigarettes in Taiwan: domestic versus imported
J Policy Model, 26:81–93 doi:10.1016/j. Health Econ, 22:821–842.doi:10.1016/S0167- cigarettes. Contemp Econ Policy, 17:223–234
jpolmod.2003.11.003. 6296(03)00058-4 PMID:12946461 doi:10.1111/j.1465-7287.1999.tb00677.x.
Flewelling RL, Kenney E, Elder JP et Guhl N, Hughes D (2006). Cigarette smoking Hu T, Keeler TE, Sung HY, Barnett PG
al. (1992). First-year impact of the 1989 and Market Faikure: a determination of the (1995b). The impact of California anti-smoking
California cigarette tax increase on cigarette economically efficient cigarette tax rate. legislation on cigarette sales, consumption and
consumption. Am J Public Health, 82:867–869. prices. Tob Control, 4 suppl.1;;S34–S38.
doi:10.2105/AJPH.82.6.867 PMID:1585966 Guindon GE, Nguyen TTH, Hoang VK et al.
(2010). Tobacco Taxation in Vietnam. Paris: Hu T, Mao Z (2002). Economics analysis
Florkowski WJ, McNamara KT (1992). International Union Against Tuberculosis and of tobacco and options for tobacco control.
Policy Implications of Alcohol and Tobacco Lung Disease. HNP Discussion Paper Series, Economics of
Demand in Poland+. J Policy Model, 14:93–98 Tobacco Control Paper No.3. Washington DC,
doi:10.1016/0161-8938(92)90025-8. Guindon GE, Perucic A-M, Boisclair D (2003). The World Bank.
Higher tobacco prices and taxes in South-
Fujii ET (1980). The demand for cigarettes: East Asia: an effective tool to reduce tobacco Hu T, Sung HY, Keeler TE (1995a). The state
further empirical evidence and its implications use, save lives and generate revenue. HNP antismoking campaign and the industry
for public policy. Appl Econ, 12:479–489 Discussion Paper Series, Economics of response: the effects of adverising on cigarette
doi:10.1080/00036848000000008. Tobacco Control Paper No.11. Washington DC, consumption in California. AEA Papers and
The World Bank.. Proceedings, 85:85–90.
Gajalakshmi CK, Jha P, Randon K et al. (2000).
Global patterns of smoking and smoking- Guindon GE, Tobin S, Yach D (2002). Trends Hu TW, Bai J, Keeler TE et al. (1994).
attributable mortality. In: Jha P, Chaloupka F, and affordability of cigarette prices: ample The impact of California Proposition 99,
eds., Tobacco control in developing countries. room for tax increases and related health a major anti-smoking law, on cigarette
Oxford, U.K., Oxford University Press: 11–39. gains. Tob Control, 11:35–43.doi:10.1136/ consumption. J Public Health Policy, 15:26–36.
tc.11.1.35 PMID:11891366 doi:10.2307/3342605 PMID:8027359
Gallus S, Fernandez E, Townsend J et al.
(2003). Price and consumption of tobacco in Gujarati DN (2003). Basic Econometrics. New Hu TW, Mao Z, Shi J et al. (2008). Tobacco
Italy over the last three decades. Eur J Cancer York, McGraw-Hill. Taxation and Its Potential Impact in China.
Prev, 12:333–337.doi:10.1097/00008469- Paris, International Union Against Tuberculosis
200308000-00015 PMID:12883388 Hamilton JD (1994). Time series analysis. and Lung Disease.
Princeton, Princeton Univeristy Press.
Gallus S, Schiaffino A, La Vecchia C et al. Huang BN, Yang CW, Hwang MJ (2004). New
(2006). Price and cigarette consumption in Hamilton JL (1972). The demand for cigarettes: evidence on demand for cigarettes: a panel
Europe. Tob Control, 15:114–119.doi:10.1136/ advertising, the health scare, and the cigarette data approach. International Journal of Applied
tc.2005.012468 PMID:16565459 adverising ban. Rev Econ Stat, 54:401–411 Economics, 1:97.
doi:10.2307/1924567.
133
Iglesias R (2006). Economia del control Koutsoyiannis AP (1963). Demand functions Onder Z (2002). The economics of tobacco in
del tobaco en los paises del Mercosur y for tobacco. Manchester Sch Econ Soc Turkey: new evidence and demand estimates.
Estados Asociados: Brasil. Washington DC, Stud, 31:1–19 doi:10.1111/j.1467-9957.1963. HNP Discussion Paper Series, Economics of
Organizacion Panamericana de la Salud. tb01009.x. Tobacco Control Paper No.2. Washington DC,
The World Bank.
Jackson JD, Ekelund RB Jr (1989). The influence Lal A, Scollo M (2002). Big Mac index of
of advertising on tobacco consumption: some cigarette affordability. Tob Control, 11:280– Peng L, Ross H (2009). The impact of cigarette
problems with Chetwynd et al.’s analysis. Br 282.doi:10.1136/tc.11.3.280-b PMID:12198286 taxes and advertising on the demand for
J Addict, 84:1247–1250, discussion 1251– cigarettes in Ukraine. Cent Eur J Public Health,
1254.doi:10.1111/j.1360-0443.1989.tb00719.x Lanoie P, Leclair P (1998). Taxation or 17:93–98. PMID:19662827
PMID:2519560 regulation: looking for a good anti-smoking
policy. Econ Lett, 58:85–89 doi:10.1016/ Peterson DE, Zeger SL, Remington PL,
Jha P, Chaloupka FJ (1999). Curbing the S0165-1765(97)00258-9. Anderson HA (1992). The effect of state
epidemic. Governments and the Economics cigarette tax increases on cigarette sales,
of Tobacco Control. Washington D.C., World Laugesen M, Meads C (1991). Tobacco 1955 to 1988. Am J Public Health, 82:94–96.
Bank. advertising restrictions, price, income and doi:10.2105/AJPH.82.1.94 PMID:1536343
tobacco consumption in OECD countries,
Jha P, Chaloupka FJ (2000). Tobacco control 1960–1986. Br J Addict, 86:1343–1354. Peto J (1974). Price and consumption of
in developing countries. New York, Oxford d o i:10 .1111/ j.13 6 0 - 0 4 4 3 .19 91.t b 01710 . x cigarettes: a case for intervention? Br J Prev
University Press on behalf of WHO and the PMID:1751850 Soc Med, 28:241–245. PMID:4455342
World Bank.
Lee JM (2007). The synergistic effect of Pierani P, Tiezzi P (2009). Addiction and
Jha P, Musgrave P, Chaloupka FJ et al. cigarette taxes on the consumption of interaction between alcohol and tobacco
(2000). Chapter 7: the economic rationale cigarettes, alcohol and betel nuts. BMC Public consumption. Empir Econ, 37:1–23 doi:10.1007/
for intervention in the tobacco market. In: Health, 7:121.doi:10.1186/1471-2458-7-121 s00181-008-0220-3.
Jha P, Chaloupka FJ, eds., Tobacco control PMID:17592627
in developing countries. New York, Oxford Pindyck RS, Rubinfeld DL (1998). Econometric
University Press on behalf of WHO and the Lee JM, Chen HF (2008). The effects of models and economic forecasts, 4th Edition.
World Bank. price and smoking risk information on McGraw-Hill Book Companies Inc, Singapore.
the demand for tobacco in Taiwan: an
Johnson LW (1986). Advertising expenditure empirical study. Appl Econ, 40:1757–1767 Porter R (1986). The impact of government
and aggregate demand for cigarettes in doi:10.1080/00036840600905142. policy on the US cigarette industry. In: Ippolito
Australia. Int J Advert, 5:45–58. P, Scheffman D, eds., Empirical approaches
Leeflang P, Reuijl J (1985). Advertising and to consumer protection economics. US
Kan MY (2007). Investigating cigarette insutry sales: an empirical study of the West Government Printing Office: 447–484.
affordability in 60 cities using the cigarette German cigarette market. J Mark, 49:92–98
price-daily income ratio. Tob Control, doi:10.2307/1251435. Prest AR (1949). Some experiments in
16:42 9 – 4 32 .do i:10.113 6 / tc . 20 07.020 4 87 demand analysis. Rev Econ Stat, 31:33–49
PMID:18048622 Lopez AD, Collishaw NE, Piha TA (1994). A doi:10.2307/1927192.
descriptive model of the cigarette epidemic in
Kao K, Tremblay VJ (1988). Cigarette health developed countries. Tob Control, 3:242–247 Radfar M (1985). The effect of advertising
scare, excise taxes, and advertising ban doi:10.1136/tc.3.3.242. on total consumption of cigarettes in the
[comment]. South Econ J, 54:770–776 UK: A comment. Eur Econ Rev, 29:225–231
doi:10.2307/1059019. McGuinness T, Cowling K (1975). Advertising doi:10.1016/0014-2921(85)90053-4.
and the aggregate demand for cigarettes.
Keeler TE, Hu TW, Barnett PG et al. (1996). Eur Econ Rev, 6:311–328 doi:10.1016/0014- Ramos A (2006). Economia del control del
Do cigarette producers price-discriminate by 2921(75)90015-X. tobaco en los paises del Mercosur y Estados
state? An empirical analysis of local cigarette Asociados: Uruguay. Washington DC,
pricing and taxation. J Health Econ, 15:499– Metra Consulting Group Ltd (1979). The Organizacion Panamericana de la Salud.
512.doi:10.1016/S0167- 629 6(9 6) 0 0 49 8 - 5 relationship between total cigarette advertising
PMID:10164041 and total consumption in the United Kingdom. Reekie WD (1994). Consumers’ surplus and
London. the demand for cigarettes. Managerial and
Keeler TE, Hu TW, Barnett PG, Manning WG Decision Economics, 15:223–234 doi:10.1002/
(1993). Taxation, regulation, and addiction: Mindell J, Whynes D (2000). Cigarette mde.4090150304.
a demand function for cigarettes based consumption in The Netherlands 1970-1995:
on time-series evidence. J Health Econ, does tax policy encourage the use of hand- Reinhardt FS, Giles DE (2001). Are
12:1–18.doi:10.1016/0167-6296(93)90037-F rolling tobacco? Eur J Public Health, 10:214– cigarette bans really good economic
PMID:10126486 219 doi:10.1093/eurpub/10.3.214. policy? Appl Econ, 33:1365–1368
doi:10.1080/00036840010007489.
Kenkel D, Chen L (2000). Consumer information Nelson JP (2003). Cigarette demand, structural
and tobacco use. In: Jha P, Chaloupka FJ, change, and advertising bans: international Ross H, Al-Sadat NA (2007). Demand
eds., Tobacco control in developing countries. evidence, 1970–1995. Contributions to analysis of tobacco consumption in
Washington, World Bank. Economic Analysis and Policy, 2:1–27. Malaysia. Nicotine Tob Res, 9:1163–
116 9 . d o i :10 .10 8 0 /14 6 2 2 2 0 0 7 016 4 8 4 3 3
Kim S, Seldon BJ (2004). the demand for O’Donoghue T, Rabin M (2003). Studying PMID:17978990
cigarettes in the Republic of Korea and optimal paternalism, illustrated by a model
implucations for government policy to lower of sin taxes. Am Econ Rev, 93:186–191 Russell MA (1973). Changes in cigarette price
cigarette consumption. Contemp Econ Policy, doi:10.1257/000282803321947029. and consumption by men in Britain, 1946–71: a
22:299–308 doi:10.1093/cep/byh021. preliminary analysis. Br J Prev Soc Med, 27:1–
Olekalns N, Bardsley P (1996). Rational 7. PMID:4717796
addiction to caffeine: an analysis of coffee
consumption. J Polit Econ, 104:1100–1104
doi:10.1086/262054.
134
Saffer H (2000). Chapter 9: Tobacco advertising Tauras JA, Chaloupka FJ (1999). Price, Van Walbeek CP (2002). Economics of
and promotion. In: Jha P, Chaloupka FJ, eds., clean indoor air laws and cigarette smoking: Tobacco Control Project (Phase II), 2002.
Tobacco control in developing countries. New evidence from longitudinal data for young The economics of tobacco control in South
York, Oxford University Press on behalf of adults. NBER Working Paper Series. Working Africa. Applied Fiscal Research Centre. ISBN
WHO and the World Bank. Paper #6937. Cambridge, MA, National 0–7992–2174–0. Written by CP van Walbeek
Bureau of Economic Research. (145 pp).
Saffer H, Chaloupka F (2000). The effect
of tobacco advertising bans on tobacco Tauras JA, O’Malley PM, Johnston LD (2001). Warner KE (1977). The effects of the anti-
consumption. J Health Econ, 19:1117–1137. Effects of price and access laws teenage smoking campaign on cigarette consumption.
d o i : 1 0 .1 0 1 6 / S 0 1 6 7- 6 2 9 6 ( 0 0 ) 0 0 0 5 4 - 0 smoking intiation: a national longitudinal Am J Public Health, 67:645–650.doi:10.2105/
PMID:11186847 analysis. Research Paper Series Number 2. AJPH.67.7.645 PMID:879393
University of Illinois, Chicago, impacTEEN,
Schmalensee R (1972). The economics of YES! (Youth, Education and Society). Warner KE (1981). Cigarette smoking in the
advertising. Amsterdam, North-Holland. 1970’s: the impact of the antismoking c ampaign
Tegene A (1991). Kalman f ilter and the demand on consumption. Science, 211:729–731.
Scollo M (1996). The Big Mac index of cigarette for cigarettes. Appl Econ, 25:521–529. doi:10.1126/science.7455711 PMID:7455711
affordability. Tob Control, 5:69.doi:10.1136/
tc.5.1.69a PMID:8795863 Thursby JG, Thursby MC (1994). Interstate Warner KE (1989). Effects of the antismoking
cigarette bootlegging;; extent, revenue losses, campaign: an update. Am J Public Health,
Seldon BJ, Boyd RG (1991). The stability of and effects of federal intervention. NBER 7 9:14 4 –151.d o i:10 . 210 5 /A J P H .7 9 . 2 .14 4
cigarette demand. Appl Econ, 23:319–326 Working Paper Series. Working Paper #4763. PMID:2913831
doi:10.1080/00036849100000139. Cambridge, MA, National Bureau of Economic
Research. Warner KE (1990). Tobacco taxation as health
Seldon BJ, Doroodian K (1989). A policy in the Third World. Am J Public Health,
simultaneous model of cigarette advertising: Thursby JG, Thursby MC (2000). Interstate 8 0:52 9 – 5 31.doi:10. 210 5 /A J PH . 8 0. 5 . 52 9
effects on demand and industry response cigarette bootlegging;; extent, revenue losses, PMID:2327526
to public policy. Rev Econ Stat, 71:673–677 and effects of federal intervention. Natl Tax J,
doi:10.2307/1928110. 53:59–77. Waters TM, Sloan FA (1995). Why
do people drink? tests of the rational
Shafey O, Ericksen M, Ross H et al. (2009). Townsend J (1987). Cigarette tax, addiction model. Appl Econ, 27:727–736
The Tobacco Atlas. Third Edition. Atlanta, economic welfare and social class patterns doi:10.1080/00036849500000062.
Georgia, American Cancer Society. of smoking. Appl Econ, 19:355–365
doi:10.1080/00036848700000007. Wilcox WB, Tharp M, Yang K (1994). Cigarette
Simester D, Brodie R (1994). The effects of advertising and consumption in South Korea.
advertising on brand and industry demand Townsend J, Roderick P, Cooper J (1994). Int J Advert, 13:333–346.
for tobacco: a meta-analysis of econometric Cigarette smoking by socioeconomic group,
research. New Zealand Journal of Business, sex, and age: effects of price, income, Wilcox WB, Vacker B (1992). Cigarette
16:21–37. and health publicity. BMJ, 309:923–927. advertising and consumption in the United
PMID:7950662 States, 1961–1990. Int J Advert, 11:269–278.
Stavrinos VG (1987). The effects of an anti-
smoking campaign on cigarette consumption: Tremblay CH, Tremblay VJ (1995). The impact Witt SF, Pass CL (1981). The effects of health
empirical evidence from Greece. Appl Econ, of cigarette advertising on consumer surplus, warnings and advertising on the demand
19:323–329 doi:10.1080/00036848700000004. profit, and social welfare. Contemp Econ Policy, for cigarettes. Scott J Polit Econ, 28:86–91
13:113–124 doi:10.1111/j.1465-7287.1995. doi:10.1111/j.1467-9485.1981.tb00076.x.
Stewart MJ (1992). Tobacco consumption and tb00718.x.
advertising restrictions: a critique of Laugesen Worgotter WG, Kunze M (1986). Cigarette
and Meads (1991). Int J Advert, 11:97–118. U.S. Department of Health Education and prices and cigarette consumption in Austria
Welfare (USDHEW) (1964). Smoking and 1955–83. N Y State J Med, 3:478–479.
Stewart MJ (1993). The effect on tobacco health: report of the advisory committee to the
consumption of advertising bans in OECD surgeon general of the public health service. World Health Organization (2008). WHO report
countries. Int J Advert, 12:155–180. Washington DC, US Department of health, on the global tobacco epidemic 2008: the
Education, and Welfare, Public Health Service. MPOWER package. Geneva, World Health
Stone R (1945). The analysis of the market Organization.
demand. JR Stat Soc, 108:286–346 U.S. Department of Health and Human
doi:10.2307/2981291. Services (USDHHS) (2000). Reducing Yorozu I, Zhou Y (2002). The demand for
Tobacco Use. A Report of the Surgeon cigarettes in Japan: impact of information
Sumner MT (1971). The demand for tobacco in General. Atlanta,GA, U.S. Dept. of Health and dissemination on cigarette consumption.
the UK. Manchester Sch Econ Soc Stud, 39:23– Human Services, Center for Disease Control Contemp Econ Policy, 20:72–82 doi:10.1093/
36 doi:10.1111/j.1467-9957.1971.tb00365.x. and Prevention, National Center for Chronic cep/20.1.72.
Disease Prevention and Health Promotion,
Sung HY, Hu T, Keeler TE (1994). Cigarette Office on Smoking and Health. Yuanliang B, Zongyi Z (2005). Aggregate
taxation and demand: an empirical cigarette demand and regional differences
model. Contemp Econ Policy, 12:91–100 Valdés B (1993). Cigarette consumption in in China. Appl Econ, 37:2523–2528
doi:10.1111/j.1465-7287.1994.tb00437.x. Spain: empirical evidence and implications for doi:10.1080/00036840500358640.
public health policy. Appl Econ, 25:149–156
Taal A, Kiivet R , Hu T W (2004). The economics doi:10.1080/00036849300000019. Yurekli A, Onder Z, Erk N et al. (2010). The
of tobacco in Estonia. HNP Discussion Paper economics of tobacco taxation: challenges and
Series, Economics of Tobacco Control Paper Van Walbeek CP (1996). Excise taxes on opportunities for a tobacco free Turkey. Paris,
No.9. Washington DC, The World Bank. tobacco: how much scope does the government International Union Against Tuberculosis and
have? S Afr J Econ, 64:21–42. Lung Disease.
Tansel A (1993). Cigarette demand, health scares
and education in Turkey. Appl Econ, 25:521–
529 doi:10.1080/00036849300000060.
135
136
Chapter 5
Tax, price and adult tobacco use
Introduction
Two broad types of data have been Survey), while other surveys are more these questions. The most common
used in international studies of generic, and incorporate questions approach, described in more detail
the demand for cigarettes, namely about tobacco use as part of the below, is to separate individual
aggregate data and survey data. greater whole (e.g. National Health tobacco consumption decisions into
In general, it is easier and less and Nutrition Examination Survey in different stages, e.g. the decision
costly to collect aggregate data the USA). In some cases the same to initiate smoking, the decision
than individual or household survey respondents are interviewed again to continue smoking, the decision
data. Both types of data have their in subsequent waves of the survey. about how much to smoke when
advantages and drawbacks. The Such surveys are called longitudinal, being a smoker, and the decision to
previous chapter presents the panel or cohort surveys and are quit smoking. With individual-level
evidence from studies employing particularly useful in seeing how data, and especially with longitudinal
time-series or pooled time-series individuals’ tobacco use behaviours individual data, these questions can
cross-sectional aggregate data. change over time. be answered not only for the whole
This chapter reviews the evidence Analyses that employ survey population, but also for different
from studies using individual-level data to study the demand for tobacco socioeconomic and demographic
and household-level survey data to products are based on the same groups, such as those defined by
examine the effects of tax and price theoretical framework discussed in gender, age, education, income,
on the demand for tobacco products Chapter 4. However, studies that race/ethnicity or other factors.
among adults. Studies that focus have used survey data can answer Studies based on aggregate
on tobacco demand among young more nuanced questions than data often find that the correlation
people are reviewed in Chapter 6 and studies that have used aggregate between independent variables is
tobacco demand among the poor in data. For example, studies based on high. The high degree of correlation
Chapter 7. aggregate data, which are usually complicates the estimation process
Individual-level and household- derived from tax paid sales data, and can result in unstable estimates
level cross-sectional data are cannot determine why tobacco (coefficients) and wide confidence
collected in large nationally consumption falls in response to an intervals. In contrast, the correlation
representative (or sometimes increase in tobacco product prices. between independent variables in a
regionally representative) surveys. Is it because fewer people are using cross-sectional analysis of survey
Depending on the nature of the survey these products (i.e. a decrease in data is generally lower, resulting in
and the size of the questionnaire, smoking prevalence), because those smaller standard errors and more
researchers are able to gather a who use tobacco are consuming stable coefficients.
vast array of information about the less (i.e. a decrease in smoking Studies based on aggregate data
survey respondents. Some surveys intensity), or some combination of often suffer from simultaneity bias,
are designed with a focus on tobacco the two? Studies based on survey as market-level price and quantity
(e.g. the Global Adult Tobacco data are better able to address are jointly determined through the
137
interaction of supply and demand. adult population. In the following multiple countries). They can also
Thus, the price variable may be section, the estimation techniques be a single cross-section over time
endogenous. Standard Hausman applied in studies based on survey with data collected at multiple time
tests can be used to test whether data are discussed in some detail. points, such as studying a sample
price is endogenous. If this is This is followed by a comprehensive from a single country over a period
indeed found to be the case, price review of the evidence on the impact of multiple years (this form of pooled
endogeneity can be corrected using of price on cigarette smoking among cross-sectional data is often referred
instrumental variables methods. adults, including the impact of price to in the literature as repeated cross-
However, when using survey data, on smoking prevalence and intensity. sectional data). Pooling different
the problem of price endogeneity A similarly comprehensive but cross-sections has the benefit of a
is less of a concern because no shorter review of the relatively scarce larger sample size (or a higher number
individual tobacco user or potential evidence on price and non-cigarette of subjects studied) beyond what
user consumes enough to influence tobacco use follows. This is followed would be available in a single cross
the market price. Thus, from the by a discussion of the existing section, thus allowing researchers
perspective of any individual, the empirical literature on smoking to examine different determinants
price variable is exogenously cessation. The chapter concludes of tobacco use over time like prices
determined. This simplifies the with a discussion of the more limited of tobacco products and tobacco
estimation process. research on the relationships of control policies by studying a sample
However, while individual tax and price with other outcomes at multiple time points. Longitudinal
or household survey data have related to tobacco use for adults, data, on the other hand, follows
some clear advantages relative including attitudes and perceptions each subject (either an individual or
to aggregate data, they also have about prices of tobacco products. household) over time. The advantage
several limitations. First, the survey of longitudinal data is that changes in
data are subject to reporting biases, Empirical strategies for estimating individual (or household) tobacco use
in that people tend to underreport demand with individual-level data measured as smoking prevalence
tobacco use and consumption and intensity can be examined, and
(Warner, 1978;; Slattery et al., Data the impact of price or tax on tobacco-
1989;; Gallus et al., 2011). Typically use transitions such as tobacco
researchers assume that people Individual- and household-level data, initiation and cessation can be
underreport their consumption by usually obtained from surveys, rely quantified.
the same percentage of their actual on self-reported measures of tobacco
consumption. This is a strong use. The measures of tobacco use Measuring the price of tobacco
assumption, and to the extent that it typically collected in surveys include products
is violated, it will bias the estimated information on the prevalence
coefficients. Second, many macro- and intensity of tobacco use and In contrast to aggregate databased
level determinants of the demand sometimes on tobacco consumption studies, studies that rely on survey
for tobacco, like advertising histories. Two basic types of data can consider many more
expenditures and/or advertising surveys are typically employed variables in the analysis. For
restrictions, availability of tobacco when examining the impact of price example, individual (or household)
products, and the strength of anti- or tax on adult tobacco use: cross- characteristics such as disposable
tobacco (or pro-tobacco) sentiment, sectional and longitudinal. A cross- income, age, gender, level of
are often not captured in the surveys. sectional survey is used to gather education, employment status,
If information on these determinants information on a population at a marital status, family structure,
are not obtained from other sources, single point in time, and many cross- religiosity and various other
the coefficient estimates may be sectional surveys are repeated socioeconomic and demographic
biased due to omitted variables. regularly. Often, researchers use characteristics can be incorporated
This chapter reviews several pooled cross-sectional data. Pooled in the analysis.
issues related to the impact of taxation cross-sections combine multiple Of particular importance to this
of cigarettes and other tobacco cross-sections at a single point in research is the role of tobacco prices
products, and tobacco products time (such as comparing cross- in affecting adult tobacco demand.
prices, on tobacco use among the sections of the populations of As increases in tobacco taxes are
138
expected to increase tobacco prices, involves several limitations, including goods and services including tobacco
tobacco taxes are a policy mechanism limited geographic coverage, and products. Price is typically derived
with which governments can affect the use of a convenience sample by dividing household consumption
adult consumption of tobacco of stores (i.e. inadequate sample expenditures on tobacco products
(see Chapter 2 for a discussion on frame and hence a sample that may by the total amount/quantity of these
tobacco taxes and prices). As IARC not be representative of all retail products consumed. In addition to
Handbooks of Cancer Prevention: stores). Use of data collected with the endogeneity problems discussed
Tobacco Control, Volume 12 points this approach can also create an above, there are other reasons why
out (IARC, 2008), several sources aggregate price measure that does these price measures should be
of tobacco price data have been not account for the share of sales of used with caution. First, one member
used in prior research. Some studies the brands collected and the share of of the family typically reports total
have used cigarette prices that are sales from different types of outlets. household expenditures on tobacco
collected in retail outlets or reported The final method to collect and quantity purchased, leading
by households/individuals that use tobacco prices is through the use of to potential reporting errors. The
scanner-based technologies that mail and telephone surveys. These extent of reporting errors depends
utilize universal product codes surveys are conducted on tobacco on how well informed the individual is
(UPCs). The major advantages of product vendors, the general about the consumption expenditures
using prices collected using scanner population and the population of and quantities purchased by other
technology is that brand- and smokers. The surveys on tobacco household members. Caution should
package-specific information can be vendors face limitations similar to also be used because sometimes
extracted, and that the final purchase the observational data collection only broad information on tobacco
price will capture the presence of methods described above. expenditures and amount purchased
sale promotions affecting price. Information obtained in population are collected, while only composite
Unfortunately, there is limited surveys is useful in developing prices that combine all tobacco
availability of this technology in aggregate measures of price (such as products can be generated. Finally, it
many countries, particularly in at the national or subnational levels). is very important to account for the
low- and middle-income countries. However, the use of individual’s effects of inflation when evaluating
Even in high-income countries, self-reported price in analyses to the impact of tobacco prices on
these systems typically do not examine the respondent’s smoking tobacco demand and demand-
provide comprehensive geographic behaviour may be problematic due related outcomes when multiple
coverage. Moreover, even in covered to the endogeneity of this price years of data are employed, given
areas, some outlets that sell tobacco variable. That is, holding other things that inflation can significantly affect
products do not participate in scanner constant, heavy smokers may be the price of tobacco products relative
data collection efforts, generating more likely to consume cheaper to the price of other goods and
tobacco price data that may not brands of cigarettes, purchase services.
be completely representative of all cigarettes in greater quantities, look
tobacco sales. for lower-priced retailers, engage Methods
Another approach to obtain in tax-avoiding behaviours, and
tobacco price information is to take advantage of price promotions When using individual-level survey
use observational data collection than individuals who smoke fewer data, economists have traditionally
methods in which trained individuals cigarettes. Treating an individual’s used a two-part model of demand
visit tobacco vendors and collect self-reported price as an exogenous developed by Cragg (1971), in which
price information by observing what variable in an equation that examines tobacco prevalence and tobacco
the posted prices are, asking the his or her tobacco consumption will intensity are modelled separately.
vendor what the price is when the lead to a biased (over-) estimate of The rationale for the two-part
price is not posted, and sometimes the impact of price on consumption. model is that a person faces two
purchasing the product. These In some analyses, efforts have sequential decisions. First, s/he has
prices typically produce the same been made to derive cigarette to decide whether or not to consume
measure of price and account prices from household expenditure tobacco. Then, among those who
for price-related promotions. surveys that collect expenditure have decided to consume tobacco,
Observational data collection information on a wide variety of there is a decision on how much
139
tobacco to be consumed. The first data, numerous other methods as double-hurdle models (Jones,
decision is a dichotomous decision have been employed. Due to limited 1989). A thorough discussion on the
and is typically estimated using a computing power, some early studies variants of the double-hurdle models
probit or logit specification (logistic used a linear probability model (i.e. can be found in Jones and Yen
regression models). For the second an OLS technique) instead of limited (2000).
decision, on how much tobacco to dependent variables techniques for Researchers have also used
be consumed, ordinary least squares the first part of the two-part model. longitudinal data to examine the
(OLS) techniques (linear regression Other researchers have combined impact of price or tax on tobacco-use
models) on a log-transformed both separate models of the two- transitions such as tobacco initiation,
dependent variable are usually part model into a one-part model tobacco use escalation, and tobacco
used. The resulting price elasticity and apply an OLS regression. The cessation. Both discrete-time and
derived from the first step is known limitation of this approach is that continuous-time hazard models have
as the price elasticity of prevalence, researchers cannot disentangle been used to estimate the equations
whereas the resulting price elasticity the effects of changes in prices or specified for those tobacco-use
from the second step is known taxes on tobacco prevalence and transitions. The discrete time model
as the price elasticity of intensity. intensity—a distinction that is often usually takes the form of a probit or
Combining the price elasticity of very important for policy-making logit specification to estimate the
prevalence and the price elasticity purposes. Other researchers have probability of making a transition from
of intensity will yield the total price used sample selection models, one discrete state to another (such
elasticity of tobacco demand. such as Heckman’s (1979) two-step as from non-smoking to smoking)
A potential problem with the correction model. Heckman’s model between waves of data collection.
standard approach of estimating the corrects the second part of the two- Cox’s (1972) partial likelihood method
second part of the two-part model part model for self-selection by using has also been used to estimate
is that the estimates have to be a transformation that is the predicted continuous hazard models. Cox’s
retransformed back from log scale to probability of each individual to use (1972) model is appealing because
original scale. The retransformation tobacco as an additional variable it examines the effects of the
factor that is employed in the in the second equation. Still other covariates on tobacco use without
retransformation process must researchers reformulate the making any assumptions about the
reflect any heteroscedasticity in tobacco demand equation into latent underlying baseline hazard. Other
the conditional smoking equation consumption (i.e. latent demand) researchers have used continuous
if the derived elasticities are to be instead of the actual consumption. time parametric models that have
consistent. An alternative estimator As the latent demand measures assumptions about the shape of the
that Blough, Madden and Hornbrook willingness to consume tobacco, baseline hazard.
(1999), Manning and Mullahy those individuals who hate tobacco
(2001) and Tauras (2006) proposed would likely make negative scores Estimation issues
is the generalized linear model for willingness to consume. Since
(GLM). While OLS models estimate negative latent demand is reflected A central issue to consider when
expectation of the log-transformed by zero actual consumption, the estimating empirically cigarette-
variable E(ln(Y|X)) and then require dependent consumption variable is demand equations for adults is how
retransformation back to ln(E(Y|X)), left censored at zero. Tobit models to account for tobacco sentiment or
GLM models directly estimate have been used by researchers to the social acceptability of smoking.
ln(E(Y|X)) and consequently obtain account for the censored nature Indeed, accounting for this condition
expectation E(Y|X) directly and thus of the tobacco consumption data. is important because it may be
preclude the need to retransform1. Finally, numerous extensions to the sentiment towards tobacco
While the two-part technique Cragg’s (1971) model have been that is driving changes in tobacco
discussed above is the most proposed and employed to examine consumption and in tobacco taxes
frequently used method to estimate the determinants of both prevalence as well as in other tobacco control
tobacco prevalence and intensity and intensity of tobacco use. This policies. If not controlling for tobacco
demand equations using survey class of models has become known sentiment, this overlook may result
1
Y is the dependent variable, X is a vector of explanatory variables, ln is a natural logarithm transformation, and E(Y|X)is the expectation of Y conditional on X.
140
in an omitted variable bias that approach to be viable, researchers simultaneously, it may become
produces a spurious negative must use multiple years of cross- difficult to disentangle the individual
relationship between price or tax sectional data—one year of cross- effects of taxes from those effects
and tobacco demand. This spurious sectional data would result in perfect of other tobacco control policies on
negative relationship results in price multicollinearity between the location tobacco consumption.
elasticity estimates that are biased specific taxes (or prices) and the It is also important to account
away from zero. dichotomous location indicators. for tax avoidance opportunities
Several strategies have been Moreover, even if multiple years of to purchase tobacco products
put forth to account for antismoking data are employed, there must be when estimating tobacco demand
sentiment in adult smoking equations. reasonable variation in tax (or price) equations using individual-level data
One approach that can be applied to over time within locations to avoid (see Chapter 8 for a more complete
cross-sectional data, pooled cross- collinearity issues with the tax (or discussion of tax avoidance and
sectional data and longitudinal price) variable. evasion). Cross-border shopping
data is to include a variable as an The final approach that has opportunities created by substantial
explanatory variable in the cigarette been used is to approximate the differences in tax rates or other
demand equation to capture the magnitude of anti-tobacco sentiment factors that determine prices across
economic importance of tobacco within a location using individual’s political jurisdictions are one form
growing and production in the area attitudes towards smoking and of tax avoidance. If opportunities to
(such as a state, region, province, beliefs about tobacco control policies purchase less expensive tobacco
etc.) where the individual/household from survey data. The derived exist but are not accounted for in
resides. To the extent that residing in tobacco sentiment variable would the specification of the demand
a tobacco producing location proxies then be included as an explanatory equation, biased price elasticity
for positive sentiment towards variable in the tobacco demand estimates may result, as the full price
tobacco, the inclusion of these equation. Some caution should be elasticity estimates in absolute value
variables in the regression model used with this approach because the will tend to be understated. That is,
will mitigate some bias generated derived tobacco sentiment variable individuals living close to a border
by the omitted variables on the price may be endogenous, particularly if of a lower-price location may in fact
elasticity estimates. the same survey data are used both purchase tobacco from the lower-
Another approach that can be to estimate the tobacco demand priced jurisdiction. The price they
applied to pooled cross-sectional equations and to derive the tobacco pay for tobacco bought is smaller
and longitudinal data is to use sentiment variable. than the price that is assigned to
location-specific dummy variables It is also important to control those individuals simply on the basis
as covariates. The use of location- for other tobacco control policies in of their residence.
specific dummies will help to eliminate the tobacco demand equations to Numerous strategies have been
all time-invariant, unobserved and avoid an omitted variable bias. The put forth to address cross-border
location-specific heterogeneity. To omission of other tobacco control purchases using individual-level
the extent that sentiment towards policies will bias the price elasticity data. Several main strategies can
tobacco within a location is time- estimates away from zero if both be named: first, including a covariate
invariant during the period under tobacco prices and tobacco control that represents the lowest price of
investigation, then including those policies affect tobacco consumption tobacco in a neighbouring jurisdiction;;
explanatory dummy variables will and they are correlated to each other. second, including a covariate that
eliminate bias on the price elasticity One potential limitation of including represents the weighted average
estimates due to the omitted tobacco prices and tobacco control price of tobacco (or average price
variables. The use of location- policies simultaneously in the same differential) in all neighbouring
specific dummy variables relies on equation is multicollinearity. That is, jurisdictions where the weights are
within-location variation in cigarette as part of a comprehensive tobacco based on the populations close to the
prices or taxes over time (as opposed control effort, governments may borders;; third, omitting all individuals
to inter-location differences in prices increase tobacco taxes and impose who reside in locations that are
and taxes) to quantify the effect of stronger restriction on tobacco within a certain distance from the
price on consumption. However, for consumption simultaneously. If both border of a lower-priced jurisdiction;;
the location-specific dummy variable the tax and policy levers are pulled fourth, running separate regressions
141
on individuals who reside in locations papers identified by this search on survey data began to appear.
where the tax difference between strategy that reported original data Continued growth in and access to
any neighbouring jurisdiction and on the issue. Moreover, we checked computing power, particularly in the
their own is below and above a the reference lists of the identified past decade, led to the development
certain threshold;; and fifth, including articles, a meta-analysis (Gallet of more sophisticated econometric
a covariate that represents the and List, 2003), a comprehensive methods and software, allowing for
distance to the lowest tax location review of the literature (Chaloupka more sophisticated analyses of larger
in the area. While imperfect, these and Warner, 2000), a review on and more complex survey data and a
efforts have significantly reduced the developing countries from Guindon tremendous increase in the number
biases in price elasticity estimates. et al., (2003), and World Bank of studies based on survey data.
publications on tobacco (such as the As was true for the demand
Summary papers in the World Bank’s Health, studies based on aggregate data,
Nutrition, and Population (NHP) most of the earliest demand
Using individual or household-level working paper series). Many relevant studies based on survey data were
survey data to analyse the impact and appropriate articles were added conducted in the USA, but studies
of tax and price on the demand later based on the references of the based on US data are also the
for tobacco products has several articles read. No study was excluded most prevalent. This is true for
advantages over the use of more a priori because of weaknesses in many reasons, including several
aggregated data, including the ability design and/or data quality. Studies ones described in Chapter 4. First,
to disentangle the impact of tax and reporting estimates only on youth, a variety of high-quality, nationally
price on prevalence, initiation, uptake, adolescents and young adults were representative surveys that collect
cessation and intensity. At the same not considered for the present information on tobacco use have
time, using survey data introduces chapter, as these are reviewed in been conducted for many years in
several empirical challenges that, detail in Chapter 6. Similarly, studies the USA, such as the National Health
if not accounted for, can produce that focus on differences in responses and Nutrition Examination Surveys,
biased estimates of the impact of tax based on income, socioeconomic the National Health Interview Survey,
and price on tobacco use. status, or related factors are not the Tobacco Use Supplement to
discussed in this chapter, as these the Current Population Survey,
Evidence on the impact of tax and are comprehensively reviewed in several state-specific surveys (e.g.
price on the demand for tobacco Chapter 7. the California Tobacco Survey).
products among adults Second, the academic researchers
Systematic review of the who conduct these studies have
Identification of relevant studies scientific literature: Impact of tax access to high-powered computing
and price on the prevalence and technology through their universities
For the review contained in this intensity of adult tobacco use and to the public and private funding
chapter, a systematic search was sources that provide the financial
conducted to identify all publications In contrast to the studies based on research resources needed. Finally
providing evidence on the effects of aggregate data, studies based on and perhaps most importantly, as
price and tax on tobacco consumption survey data did not begin to emerge with the aggregate demand studies
among adults, using individual-level until the early 1980s. Prior to this based on state-level data, the cross-
or household-level data on adults. time, the computing power necessary sectional differences in state and
A MEDLINE search in PubMed up to conduct econometric and other local taxes and prices of tobacco
to February 2010 was performed statistical analyses of large survey products as well as the frequent
using the string “(price[title] OR data was not generally available to changes in national, state and local
prices[title] OR elasticity[title] OR researchers. As survey data and taxes provide considerable variation
elasticities[title] OR tax[title] OR the computing power needed to in taxes and prices needed to most
taxes[title] OR fiscal[title]) (“lung analyse them became more widely effectively conduct this type of
cancer”[title] OR smoking[title] OR accessible to researchers, this research.
cigarette[title] OR cigarettes[title] OR began to change and the earliest As survey data and access to
tobacco[title]) adults.” Two members studies of demand for cigarettes sufficient computing power have
of the Working Group selected the and other tobacco products based become more widely available in
142
several countries, including many the National Health Interview Survey. they found that men, particularly
low- and middle-income countries, They used prices matched to the younger men, were more responsive
demand studies based on individual survey data on the basis respondents’ to changes in cigarette prices than
or household-level survey data have state of residence. Using a two- were women.
been produced for a growing number part model, they first estimated the Mullahy (1985) built on the
of countries. Table 5.1 provides a impact of price on the decision to Lewit and Coate (1982) analyses by
detailed summary of the existing smoke and then the impact of price estimating cigarette demand models
evidence on the demand for tobacco on cigarette consumption among that accounted for the addictive
products from studies that are smokers. In addition to estimating nature of tobacco use. Mullahy
based on survey data. Studies are price elasticities for the overall (1985) assumed myopic behaviour,
organized by country, with those from sample, they also produced separate such that individuals’ current
the USA presented first, followed estimates for subsamples based on cigarette consumption depended
by those from other high-income age groups (20–25 years, 26–35 on their past cigarette consumption
countries, and then by those from years, and older than 35 years) and and that smokers ignored the future
low- and middle-income countries. sex. Lewit and Coate (1982) obtained consequences of their smoking
Within a certain country, studies an overall price elasticity for their full decisions. Applying two-part
are presented chronologically, from VDPSOHRIíZLWKDSUHYDOHQFH methods to data from the 1979 US
oldest to most recent. Table 5.1 also HODVWLFLW\ RI í DQG DQ LQWHQVLW\ Health Interview Survey, Mullahy
includes details on each identified HODVWLFLW\RIí (1985) estimated separate demand
study including the data used in Given the potential for tax equations for men and women. He
the study, basic information on the avoidance through cross-border obtained a total price elasticity of
theoretical and empirical approach, shopping for cigarettes, Lewit and í VLPLODU WR WKDW RI /HZLW DQG
estimates of price elasticities for Coate (1982) restricted their sample Coate (1982), and found that men
overall demand, prevalence (often to respondents for whom the price in were somewhat more price-sensitive
referred to as participation in this the area where they resided was lower than women.
literature) and intensity (often than the price in nearby jurisdictions. A few years later, Chaloupka
referred to as conditional demand). For this restricted sample, they (1990, 1991 and 1992) produced
In addition, for studies that estimate obtained an overall price elasticity a series of papers that applied
demand for various subpopulations, RI FLJDUHWWH GHPDQG RI í ZLWK Becker and Murphy’s (1988) rational
the estimated elasticities for these DSUHYDOHQFHHODVWLFLW\RIíDQG addiction model to examine adult
populations are presented. The DQLQWHQVLW\HODVWLFLW\RIí7KH cigarette demand (ages 17–73
narrative review below follows the differences between the two sets of years). He used individual-level,
organization of the table, providing estimates clearly illustrate how failing cross-sectional survey data from
a discussion of seminal studies or to account for opportunities for tax the Second National Health and
those that are unique in some other avoidance can bias price elasticity Nutrition Examination Survey,
way, rather than a discussion of estimates. In addition, Lewit and conducted from 1976 to 1980. He
each of the studies contained in the Coate (1982) found that smoking used information on smoking at the
table. However, it limitedly discusses was more responsive to price among time of the survey, smoking one
differences in price elasticities based younger persons than among older year before the survey, and past
on age or income or socioeconomic persons, with overall price elasticities smoking. He applied two-stage least
status, as these are discussed more RIííDQGíIRUSHUVRQV squares models that accounted for
fully in subsequent chapters. having ages 20–25, 26–35, and the endogeneity of past and future
over 35, respectively. Similarly, they smoking in the rational addiction
United States of America found that the effect of price on the model. He matched prices to the
decision to smoke among younger survey databased on an individual’s
Lewit and Coate (1982) published persons accounted for more of the area of residence, with the price
the first analysis of cigarette demand overall impact of price on cigarette measure accounting for potential
based on individual-level survey demand than it did among older cross-border shopping by averaging
data from the USA. They analysed persons (prevalence elasticities of prices in the individual’s own state
data on 19 266 respondents ages ííDQGíIRUWKHWKUHH and lower prices in nearby states.
20–74 years from the 1976 wave of age groups, respectively). Finally,
143
144
Table 5.1. Summary of studies on the effects of cigarette price on adults’ demand for tobacco products
Methods (time period,

Publication (author, Total price elasticity Price elasticity of Price elasticity of
study design and Model Comment
yr, country) of demand smoking prevalence smoking intensity
sample size)
USA
Lewitt & Coate (1982) Cross-sectional Health Two models were Entire sample Entire sample Entire sample Restricted sample was
USA Interview Survey (HIS) considered: -0.221 -0.135 0.037 obtained by deleting from
1976 i) an OLS (ordinary Restricted sample Restricted sample Restricted sample the full sample individuals in
N=28 033 (20–74 yrs) least squares);; ii) a Overall Overall Overall Primary Sample Units where
Analysis sample: two-part model: 1) OLS;; -0.416 -0.264 -0.103 the own average price was
Entire: n=19 266 2) OLS By age (in years) By age (in years) By age (in years) greater than the price within
Restricted: n=11 052 Adjusted for family 20–25 20–25 20–25 the 20-mile band
(20–74 yrs) income, education, -0.89 -0.74 -0.20

age, sex, marital status, 26–35 26–35 26–35
health status, race/ -0.47 -0.44 -0.04
ethnicity, family size, >35 >35 >35
employment status, -0.45 - 0.15 -0.15
region and urbanicity By sex and age (in By sex and age (in By sex and age (in
Demand equations years) years) years)
were estimated across Males Males Males
three age groups in the 20–25 20–25 20–25
restricted sample by -1.401 -1.276 -0.171
another two-part model: 26–35 26–35 26–35
1) a logit;; 2) an OLS -0.320 -0.292 + 0.029
Price: composite price >35 >35 >35
in cents per pack of -0.658 -0.246 -0.204
cigarettes a consumer Females Females Females
faces in his immediate 20–25 20–25 20–25
area -0.302 -0.136 -0.025
26–35 26–35 26–35
-0.577 -0.388 -0.134
>35 >35 >35
-0.118 + 0.066 -0.077
Mullahy (1985) National Health A two-part model: 1) a Overall Cigarette demand was
USA Interview Survey, 1979 probit;; 2) an OLS and -0.47 estimated based on a myopic
1 \UV instrumental By sex demand model. The results
variables method Men supported the hypothesis
-0.56 that cigarette smoking is an
Women addictive behaviour
-0.39
Chaloupka (1990) Cross sectional Two-stage least Long-run price Ranges of elasticity estimates
USA NHANES II (1976– squares model for elasticity obtained according to different
1980) current cigarette By sex rates of depreciation on the
N=28 000 (6 months to consumption Men addictive stock
74 yrs) Adjusted for real family -0.55 to -0.69
Analysis sample: 14 income, education, Women
305 (17–73 yrs) age, age2, race, marital +0.37 to 0.63
Men: n=6569 status, employment
Women: n=7736 status
sample size)
Chaloupka (1990) Cigarette price: state-
(contd) specific weighted
USA average price. Cigarette
price and excise tax
adjusted for the national
monthly consumer price
index and a local price
index
Chaloupka (1991) Cross sectional Two-stage least Long-run price Ranges of elasticity estimates
USA NHANES II (1976– squares model for elasticity obtained according to different
1980) current cigarette Overall/Full sample rates of depreciation on the
N=28 000 (6 months to consumption -0.274 to -0.359 addictive stock
74 yrs) Adjusted for real family Ever smokers
Analysis sample: income, education, -0.348 to -0.482
14 005 (17–73 yrs) age, age2, race, marital Current smokers
status, employment -0.296 to -0.890
status By education
Cigarette price: state- <high school
specific weighted -0.587 to -0.618
average price. Cigarette KLJKVFKRRO
price and excise tax +0.135 to +0.268
adjusted for the national By age (in years)
monthly consumer price 17–24
index and a local price -0.103 to +0.050
index 25–64
-0.315 to -0.454
65–73
-0.029 to +0.166
Wasserman et al. Repeated cross Two models were Two-part model Two-part model Two-part model Income elasticity obtained
(1991) sectional NHIS (1970 considered: By survey year By survey year By survey year from the GLM
USA to 1985) 1) Two part model: 1970: +0.072 1970: +0.059 1970: +0.013 By survey year
N=207 647 [seven i) logistic regression 1974: -0.013 1974: +0.002 1974: -0.015 1970: +0.051
surveys] (smoking prevalence);; 1976: -0.057 1976: -0.028 1976: -0.029 1974: +0.01
Analysis sample: 84 ii) simple OLS (smoking 1979: -0.124 1979: -0.074 1979: -0.050 1976: +0.021
301 (either >17 yrs or intensity) 1980: -0.147 1980: -0.098 1980: -0.057 1979: +0.007
>20 yrs, depending on 2) Generalized linear 1983: -0.217 1983: -0.139 1983: -0.078 1980: +0.002
year of survey) model (GLM) with an 1985: -0.263 1985: -0.171 1985: -0.092 1983: 0.013
iterative weighted least 1985: 0.023
squares technique GLM 1988: 0.038
Adjusted for income, By survey year:
education, age, gender, 1970: +0.059 Predicted elasticities for year
family size, marital 1974: -0.017 1988 might not be valid
status, race, , cohort of 1976: -0.055
birth, interactions of age 1979: -0.112
and sex, interactions 1980: -0.131
of birth cohort and sex, 1983: -0.188
and regulation index 1985: -0.226
Price: weighted average 1988: -0.283
145
state-specific price
146

sample size)
Chaloupka (1992) Cross sectional Two-stage and Long-run price Ranges of elasticity estimates
USA NHANES II (1976– restricted two-stage elasticity obtained according to different
1980) least squares model Overall/Full sample rates of depreciation on the
N=28 000 (6 months to for current cigarette -0.268 to -0.322 addictive stock
74 yrs) consumption Ever smokers
Analysis sample: Adjusted for real -0.344 to -0.422
14 005 (17–73 yrs) family income, gender, By sex
education, age, age2, Men
race, marital status, -0.425 to -0.505

labour force status & Women
state clean indoor laws +0.852 to +1.410
Cigarette price: state- Restricted model
specific weighted Overall/Full sample
average price. Cigarette -0.308 to -0.357
price and excise tax: Ever smokers
adjusted for the national -0.326 to -0.433
monthly consumer price
index and a state-
specific price index
Ohsfeldt & Boyle (1994) Cross-sectional Multivariate regression Tax elasticity Ranges of tax elasticities
USA CPS (1985) (for (i) snuff use, (ii) Snuff obtained according to both full
n=100 000 in file chewing tobacco use, -0.41 to -0.61 and restricted models
Analysis sample: only and (iii) any smokeless Chewing tobacco Cross-tax elasticities of
PHQ\UVQQRW tobacco use) -0.56 to -0.63 demand for smokeless
reported) Adjusted for per-capita Smokeless tobacco tobacco with respect to the
income, education, (both snuff and cigarette tax: +0.39 to +0.62
state population chewing tobacco) (see Table 5.3)
characteristics -0.55
(residence, race,
religious denomination,
and divorce rate),
men aged 16–17, and
existing state tobacco
regulation
Price of smokeless
tobacco products:
border-adjusted state
smokeless tobacco
average excise tax
Hu et al. (1995) Cross-sectional (pooled Two-part model: 1) -0.46 (after being -0.54 (overall) -0.22 (overall)
USA surveys) logistic model;; and 2) controlled for “other -0.33 (after being -0.20 (after being
California California BRFSS OLS health behaviours”) controlled for “other controlled for “other
(1985–1991) Adjusted for income, health behaviours”) health behaviours”)
1 \UV sex, age, ethnicity,
education, family
sample size)
Hu et al. (1995) (contd) income, employment
USA status, marital status,
California health status, “other
health behaviours”
(alcohol, exercise, and
BMI)
Price: average price
(monthly price index of
tobacco products for
Los Angeles)
Ohsfeldt et al. (1997) Cross-sectional Logit regression model Cigarettes Cross-tax elasticities of
USA CPS (1985) (for (i) cigarette use, (ii) Overall demand for smokeless
n=100 000 in file snuff use, (iii) chewing -0.05 tobacco with respect to the
Analysis sample: only tobacco use, and (iv) By age (in years) cigarette tax:
men (16–64 yrs) (n not any smokeless tobacco 16-24 Snuff: +0.13
reported) use) -0.07 Chewing tobacco: +0.09
Adjusted for real family Any smokeless tobacco: +0.10
income, age, marital -0.05 See Table 5.3
status, race, ethnicity, Snuff
employment status, Overall
education, occupation, -0.27
metropolitan, religion, By age (in years)
and tobacco control 16-24
policy -0.31
Price of smokeless
tobacco products: -0.13
border-adjusted state Chewing tobacco
average smokeless Overall
tobacco excise tax -0.13
By age (in years)
16–24
-0.07

-0.16
Any smokeless
tobacco
Overall
0.15
By age (in years)
16–24
-0.09

-0.11
Evans & Farrelly (1998) Cross-sectional (pooled Two-part model: Pooled model 1979-87 Tax The models describe
USA surveys) 1) probit (smoking Tax Overall compensating behaviours.
NHIS supplements: prevalence);; 2) OLS Overall -0.330 Taxes reduce cigarette
1979 (smoking) and (cigarette demand) -0.214 For 1987 NHIS consumption, but do not affect
1987 (cancer control) Adjusted for age, Price Tax daily tar and nicotine intake
147
148

sample size)
Evans & Farrelly (1998) 1 \UV age2, income, family Overall By age (in years)
USA (contd) Analysis sample: size, region, sex, race, –0.156 18–24: -0.223
1979 NHIS: n=24 092 marital status, urban Fixed effect model 25–39: -0.325
1987 NHIS: n=22 043 centres and education 1979-87 40+: -0.498
Two models for smoking Tax Price
intensity are considered: Overall Overall
a pooled time-series +0.160 –0.352 (using a two-
cross-sectional Price stage least squares
regression, and a fixed Overall regression)
effect model [controlled +0.344

for state and year]. Tax
Three OLS models for By age (in years) for the
smoking intensity by age 1987 data:
group were estimated 18–24: -0.575
using only the 1987 data 25–39: -0.434
Cigarette prices per 40+: +0.142
pack in constant
1982–1984 cents
Cigarette excise taxes
per pack (state and
federal) in constant
1982–1984 cents
CDC (Farrelly & Bray), Cross-sectional (pooled Two-part model for Overall Overall Overall
(1998) USA six surveys 1976–1993) cigarette use: 1) probit -0.25 -0.15 -0.10
from NHIS (smoking prevalence);; By age (in years) By age (in years) By age (in years)
Analysis sample: 2) OLS (smoking 18–24 18–24 18–24
1 \UV intensity) -0.58 -0.37 -0.21
Adjusted for region of 25–39 25–39 25–39
the country, real family -0.42 -0.25 -0.17
income, education,
sex, age, race/ -0.10 -0.06 -0.04
ethnicity, marital status, By sex By sex By sex
urbanicity and year of Men Men Men
interview -0.26 -0.18 -0.08
Price: state-specific Women Women Women
average price adjusted -0.19 -0.09 -0.10
for inflation By income By income By income
Low Low Low
-0.29 -0.20 -0.09
High High High
-0.17 -0.05 -0.12
Missing Missing Missing
-0.25 -0.23 -0.02
By ethnicity By ethnicity By ethnicity
White White White
-0.14 -0.05 -0.09
Black Black Black
-0.32 -0.36 +0.04
Hispanic Hispanic Hispanic
-1.89 -1.31 -0.58
sample size)
Evans et al. (1999) Cross sectional Two-part model: 1) Real tax Real tax Real tax Total income elasticity
USA (pooled surveys) linear probability model;; Overall Overall Overall estimates
BRFSS (1985–1995) 2) OLS -0.293 -0.144 -0.149 Real tax
Analysis sample: Adjusted for state and Average of tax and Average of tax and Average of tax and Income missing
1 \UV year effects, age, age2, lag tax lag tax lag tax 0.65
sex, annual income, Overall Overall Overall Lowest 50%
marital status, race, -0.406 -0.255 -0.151 –0.32
education, work status, Next group
and race/ethnicity 0.17
Price: real tax in 1997 Average of tax and lag tax
cents per pack Income missing
0.73
Lowest 50%
0.53
Next group
–0.13
Ohsfeldt et al. (1999) Cross-sectional (pooled Multivariate regression Cigarette use Cross-elasticities of demand
USA surveys) Adjusted for price Overall for cigarettes and snuff were
CPS (1992–1993) (tax) of snuff, index of -0.15 also analyzed. See Table 5.3
Analysis sample: smoking regulation, By age (in years) Laws restricting smoking
165 653 males, either family income, age, 16–24 in public places affect both
ZKLWHRUEODFN\UV occupation and -0.22 cigarette and snuff use
Nationally education, ethnicity, 25–44
representative samples marital status, per -0.11
capital income, poverty,
unemployment rate, and -0.07
religion Snuff use
Prices of cigarettes and Overall
snuff were mutually -0.01
adjusted By age (in years)
Price: real federal 16–24
excise tax rate -0.24
25–44
-0.05

+0.003
Hersch (2000) Cross-sectional (pooled Two-part model: 1) Men Men In both sexes, higher elasticity
USA surveys) probit and 2) OLS -0.54 -0.43 in more educated subjects
CPS (1992–1993) Adjusted for real By income By income
Nationally family earning/income, Low Low
representative samples education, age, marital -0.58 -0.60
of households status, race, presence Middle Middle
Analysis sample: 25 of children by age 0.40 0.44
726 men and 28 699 group, labour market High High
women (18–65 yrs) status, and smoking -0.25 -0.13
restriction at work Women Women
Price: average -0.38 -0.57
statewide price per
149
150

sample size)
Hersch (2000) (contd) pack (generic cigarettes By income By income
USA were included in the Low Low
sample of cigarettes, -0.99 -0.72
and the prices reflected Middle Middle
both state and federal 0.06 0.55
taxes) High High
+0.58 -0.30
Among workers Among workers
Men Men
-0.28 -0.46
Women Women
-0.21 -0.38
Farrelly et al. (2001) Cross sectional Two-part model for Overall Overall Overall Data and analysis are very
USA (pooled seven surveys cigarette use: 1) probit -0.28 -0.13 -0.15 similar to those of the previous
1976–1980, 1983, (smoking prevalence);; By age (in years) By age (in years) By age (in years) papers
1985, 1987–1993) from 2) OLS (smoking 18–24 18–24 18–24
NHIS intensity) -0.55 -0.30 -0.25
1 \UV Adjusted for state, 25–39 25–39 25–39
Analysis sample: n=354 real family income, -0.53 -0.25 -0.28
\UV education, sex, age,
race/ethnicity, family 0.00 -0.02 -0.06
size, marital status, By sex By sex By sex
urbanicity, year of Men Men Men
interview, and state- -0.18 -0.03 -0.18
specific effects Women Women Women
Price: state-specific -0.32 -0.19 -0.13
average retail price By income By income By income
adjusted for inflation Low income Low income Low income
-0.43 -0.21 -0.22
High income High income High income
0.00 +0.01 -0.11
By ethnicity By ethnicity By ethnicity
White White White
-0.15 -0.08 -0.15
Black Black Black
-0.35 -0.20 -0.15
Hispanic Hispanic Hispanic
-0.93 -0.62 -0.31
Farrelly et al. (2004) Longitudinal analysis Three linear fixed Price elasticity Smokers respond to higher
USA from COMMIT study effects regression By age (in years) cigarette prices by reducing
(1988 and 1993) models for average 25–34 intensity, but also by switching
N=11 966 smokers cigarette smoked per -0.235 to cigarettes with higher levels
(25–64 yrs) day by age group 35–44 of tar and nicotine
Wave of 1988: 3675 Adjusted for race, -0.115 Tar elasticities:
smokers (25–64 yrs) sex, marital status, 45–64 +0.041 to +0.364
education, gross -0.110
sample size)
Farrelly et al. (2004) Analysis sample household income, Nicotine elasticities:
(contd) (both waves): n=9087 clean air laws index, +0.035 to +0.306
USA smokers (25–64 yrs) age, the year 1993,
nicotine and tar
Price: average price of
all types of cigarettes at
the state level
Sheu et al. (2004) Pooled survey data Zero-inflated negative ZINB ZINB Given the relatively low
BRFSS (1996–1999) binomial (ZINB) Logit NB smoking prevalence, price
N=16 260 regression model1) Overall Overall increases are becoming
Analysis sample: n=11 logit and 2) negative +0.135 0.459 less effective for hard-
180 (Whites, blacks and binomial (NB) Two-part model Two-part model core smokers to quit. Price
Hispanics from selected For comparison Logit Overall increases may not be an
areas with available purpose, also the Overall NB effective strategy to let
information on cigarette two-part model was +0.056 -0.417 occasional smokers quit
price) considered: 1) logit;; 2) OLS
Cigarette price NB or OLS 0.474
information from Adjusted for race, age,
the Bureau of Labor sex, marital status,
Statistics 1996–1999 education, income,
employment status,
health status and year
of interview
Price: average price
(monthly price index of
tobacco products for
Los Angeles, San Diego
and San Francisco)
Sloan & Trogdon (2004) Cross-sectional Probit ( smoking By age (in years) Review authors computed
(13 surveys pooled) prevalence) 18–20 elasticity estimates based on
Behavioural Risk Factor Adjusted for -0.26 the resulting marginal effects
Surveillance System socioeconomic 21–24 for smoking prevalence
BRFSS (1990–2002) and demographic -0.11
1 \UV factors, ethnicity, 25–44
Surveys representative health insurance and -0.10
samples of the US adult behaviour, tobacco 45–64
population control policies and -0.10
border prices
Cigarette price: state- -0.25
specific weighted
average real price
Tauras (2004) National Health A three-part model: Average current Average some-day For each part of the three-
USA Interview Surveys 1) a probit for a current smoking participation smoking conditional part model, three models
(NHIS) (1991, 1993, smoker price elasticity demand elasticity were estimated. Estimated
1994) 2) a probit for a some- -0.265 0.375 elasticities obtained for each
Age: 18–64 years day smoker (conditional part were:
on current smoking) -0.333;; -0.112;; -0.328
151
152

sample size)
Tauras (2004) (contd) 3) a generalized Average price 0.904;; 0.590;; 0.977
USA linear model (GLM) elasticity of some-day -0.479;; -0.202;; -0.444
with log link and smoking participation
gamma distribution conditional on current
for estimating average smoking
number of cigarettes +0.860
consumed monthly
among some-day
smokers
Explanatory variables:
age, sex, real family
income, race/ethnicity,
education, marital
status, metropolitan
area, employment
status, clean indoor
index, and year of
interview, price of a
pack of 20 cigarettes
and survey year
Price: real annual,
sales-taxed weighted
average price for a pack
of 20 cigarettes at state
level
Tauras (2006) Cross-sectional A two-part model: Overall Overall Three different models were
USA (pooled surveys) 1) a probit (smoking -0.120 to -0.129 -0.071 to -0.073 estimated according to
Tobacco Use prevalence);; 2) a whether the smoke-free law in
Supplements to the generalized linear model each state was included
CPS (1992–1999) (GLM) with log-link and More restrictive smoke-free
Nationally Gaussian distribution air laws decrease average
representative surveys (smoking intensity) smoking by adults but have
Analysis sample: Adjusted for age, sex, little impact on prevalence
1 \UV real family income, race/
ethnicity, education,
marital status,
metropolitan area,
employment status,
clean indoor index and
year of interview
Time and state fixed
effects included in
regressions
Real price: average
statewide price
sample size)
Franks et al. (2007) Cross sectional (pooled Fixed-effects logistic By income quartile Using cigarette pack tax
USA surveys) regression (for smoking Period 1984–1996 instead of pack price brought
BRFSS (1984–2004) prevalence) Lowest (95% CI) to greater differences between
N=over 2.6 million Adjusted for sex, -0.45 (-0.67, -0.22) the two periods
\UV age, race/ethnicity, All other quartiles By income quartile:
Nationally education (years of (95% CI) Period 1984–1996:
representative schooling), number of -0.22 (-0.35, -0.10) Lowest: -0.07
telephone survey adults in the household, Period 1997–2004 All other quartiles: 0.06
consumer price index, Lowest (95% CI) Period 1997–2004:
household income, -0.14 (-0.36, +0.08) Lowest: 0.00
survey year and state All other quartiles All other quartiles: 0.04
Pack price at the year (95% CI)
preceding the survey -0.07 (-0.18, +0.05)
year. Price elasticities
were provided before
and after the Master
Settlement Agreement
(MSA)
Price: real state-specific
cigarette tax
Stehr (2007) Cross-sectional (pooled Two-part model for Overall Overall Overall When gender-specific state
USA surveys) cigarette use: 1) probit 0.36 -0.25 -0.10 fixed effects are included,
Behavioral Risk Factor (smoking prevalence);; Males Males Males women were twice as
Surveillance System 2) log-log OLS (smoking -0.26 -0.16 -0.09 responsive to cigarette taxes
BRFSS (1985–2000) intensity) Income (quartile) Income (quartile) Income (quartile)
1 \UV Adjusted for gender- 1st: -0.36 1st: -0.23 1st: -0.13
Men: n=571,631 specific state fixed 2nd: -0.11 2nd: -0.07 2nd: -0.03
Women: n=767,872 effects, income, 3 rd: -0.21 3 rd: -0.09 3 rd: -0.013
education, sex, age, 4th: -0.23 4th: -0.15 4th: -0.07
ethnicity, and year of Females Females Females
interview -0.51 -0.40 -0.12
Price: state-specific Income (quartile) Income (quartile) Income (quartile)
average cigarette tax 1st: -0.59 1st: -0.43 1st: -0.16
2nd: -0.30 2nd: -0.25 2nd: -0.05
3 rd: -0.47 3 rd: -0.30 3 rd: -0.16
4th: -0.53 4th: -0.43 4th: -0.10
DeCicca & McLeod Cross-sectional Two-way fixed effects By age (in years) Sensitivity analyses (for 3
(2008) (5 surveys) models Smoking all days models and two age groups)
USA BRFSS (2000–2005) Adjusted for sex, Overall resulted in almost identical
Telephone interview;; a age, race, education, -0.21 to -0.22 elasticities
representative sample income, marital 45–59 Price elasticities of smoking
of the US population status, health status, -0.29 to -0.31 prevalence by health status:
Analysis sample: unemployment rate, Smoking some days Unhealthy: -0.54
N=543 384 (45–64 yrs);; state ban in workplaces Overall: Healthy: -0.11
435 973 (45–59 yrs) and restaurants, and 0.20 to -0.21
state-level effects 45–59
Price: real monthly -0.24 to -0.28
state-specific cigarette
153
154

sample size)
DeCicca & McLeod excise tax rates per By sex
(2008) (contd) pack in 2001 dollars Men
USA -0.31
Women
-0.27
By education
Low
-0.43
High
-0.12
By income
Low
-0.39
High
-0.09
By health status
Unhealthy
-0.54
Healthy
-0.11
Farrelly & Engelen Cross-sectional (pooled Replication of the Total price elasticity -0.22 (-0.15;; -0.28)
(2008) surveys) previously described Before MSA After MSA (1999-2006)
USA BRFSS (1990–2006) analysis (Franks et (1990-1998) -0.09 (-0.04;; -0.13)
N=over 2.5 million al., 2007) in different -0.22 (-0.15;; -0.28) By income quartile
\HDUV periods After MSA (1999-2006) Period 1990–1998
Pack price at the year of -0.09 (-0.04;; -0.13) Lowest: -0.16
the survey. By income quartile Middle 2: -0.34
Price elasticities Period 1990–1998 Highest: -0.14
provided before and Lowest: -0.16 Period 1999–2006
after the Master Middle 2: -0.34 Lowest: -0.11
Settlement Agreement Highest: -0.14 Middle 2: -0.06
(MSA) Period 1999–2006 Highest: -0.02
Lowest: -0.11
Middle 2: -0.06
Highest: -0.02
Franz (2008) Cross-sectional (8 Two models were Overall Overall Overall Overall price elasticity for
USA surveys from BRFSS considered: -0.374 -0.193 -0.191 cessation: 0.375
(1993–2000) i) A simple OLS By age (in years) By age (in years) By age (in years)
Analysis sample: ii) A two part model: 1) 18–29 18–29 18–29
1 \UV OLS;; 2)OLS -0.518 -0.289 -0.184
Adjusted for sex, health 30–39 30–39 30–39
status, age, race, -0.360 -0.176 -0.192
education, marital 40–64 40–64 40–64
status, income, region -0.327 -0.201 -0.158
and year effects
Price: state-level -0.458 -0.331 -0.154
average real price of a
pack of cigarettes
sample size)
Dinno & Glantz (2009) Cross-sectional survey Two-part model: 1) States with average States with average Prevalence:
USA Tobacco Use fixed effect logistic cigarette price per cigarette price per The association of price with
Supplement (TUS) of regression (current pack <$3.28 pack < $3.17: smoking prevalence did not
the CPS (February smoking prevalence);; OR for an increment of A change of 1.16 change with educational
2002) 2) fixed effect linear 0.10 cents = 0.95 cigarettes/day attainment, household
N=54 024 regression (smoking (95% CI 0.93;; 0.97) (95% CI-0.40;; 2.03) income or race/ethnicity.
A representative sample intensity;; cigarettes/ OR for the highest vs. Effect of clean indoor No statistically significant
of the US population day among current lowest price = 0.83 air legislation: 2.36 interaction between cigarette
(15–80 years) smokers) (95% CI 0.78;; 0.88) cigarettes/day price and clear indoor air
Adjusted for education, States with average States with average coverage
household income, sex, cigarette price per cigarette price per Consumption:
age, race/ethnicity, and SDFN SDFN No statistically significant
clean indoor air law OR not significant No significant change in interaction between price
Price: state-specific (estimates not reported) cigarette consumption and educational attainment,
average cigarette price Price elasticity (estimates not reported) household income or race/
For people living ethnicity was found
in states with price

-0.41
For people living
in states with price

-0.99
Other high-income countries
Australia
Cameron & Williams National Drug Strategy A probit model for Overall Two probit equations for
(2001) Household Surveys cigarette use (smoking -0.436 two other discrete choice
Australia (NDSHS) for 1988, prevalence) dependent variables: (i) using
1991, 1993 and 1995 Adjusted for age, cannabis and (ii) using alcohol
Nationally sex, marital status, were used
representative surveys employment status, Cannabis and cigarettes were
of the Australian education, capital city, found to be complements.
SRSXODWLRQ\UV number of children Similarly, alcohol and
Analysis sample: legal sanctions against cigarettes were complements
Q \UV cannabis use in the
state of residence,
survey years, real
price index of alcohol,
real predicted price
of a gram of head
of marijuana, and
interaction terms of
pairs of real prices
Real price index of
cigarettes
155
156

sample size)
Zhao & Harris (2004) Australian National Multivariate probit for Overall A complementary relationship
Australia Drug Strategy three discrete choice -0.863 between alcohol and tobacco
Household Surveys dependent variables: and the similar relationship
(NDSHS, 1995, 1998, (i) using any tobacco between marijuana and
2001) products (smoking tobacco were found in the
N> 40 000 individuals prevalence), (ii) using participation decisions
\UV marijuana, and (iii)
Nationally using alcohol
representative surveys Adjusted for survey

of the Australian years, age, sex, marital
SRSXODWLRQ\UV status, legal status of
the state of residence,
employment status,
education, capital city,
Aboriginal or Torres
Strait Islander origin,
home language,
number of children,
real household income,
real price of marijuana,
and real price index of
alcohol
Real price index of
tobacco
Harris & Zaho (2007) Australian National A double-hurdle model. Marginal effect of Marginal effect of personal
Australia Drug Strategy The two hurdles were: i) tobacco price on zero income on non-participation
Household Survey decision to participate;; consumption +0.027
(NDSHS, 1995, 1998, ii) level of consumption +0.145 Marginal effect of personal
2001) that also included zero Marginal effect of income on non-participation
N>40 000 individuals consumption tobacco price on non- zero consumption
\UV A zero-inflated ordered zero consumption by -0.017
Nationally model combining a split level of consumption Marginal effect of personal
representative surveys probit model and an 1st level (lowest) income on non-zero
of the Australian ordered probit model +0.005 consumption
SRSXODWLRQ\UV Four discrete variables 2nd level 1st level (lowest)
n=28 813 individuals (0, 1, 2, and 3) -0.070 -0.003
describing different 3 rd level (highest) 2nd level
levels of tobacco -0.081 -0.007
consumption 3 rd level (highest)
Covariates: age, age 0.000
squared, sex, marital
status, education,
region of residence,
employment status,
language spoken
sample size)
Harris & Zaho (2007) at home, real price
(contd) index of alcohol, real
Australia price of marijuana,
real personal annual
before-tax income, and
an indicator of females
younger than 26 years
Real prices and
personal income were
log-transformed
Siahpush et al. (2009) Face-to-face population Dependent variable: By income group The results suggest that
Australia survey conducted in monthly smoking Low increasing the real prices
the 5 largest Australian prevalence -0.32 of cigarettes is an effective
capital cities Poisson regression Medium tobacco control strategy to
(1991-2006) Adjusted for income, -0.04 lower smoking prevalence
Sample size: 515,866 interaction between High and may provide a means of
LQGLYLGXDOV\UV price and income -0.02 reducing social disparities in
category, education, smoking
age, sex and time trend
Cigarette price:
state-specific average
real price of the two
top-selling Australian
brands (Peter Jackson
and Winfield)
Canada
Hamilton et al. (1997) Longitudinal No multivariate Smoking prevalence in Rates of starting cigarette
Canada (retrospective) regression analysis provinces where taxes smoking were higher (ranges:
Survey on Smoking in Simple descriptive were not cut declined 0.9–1.7% vs. 0.7–1.2%) and
Canada (1994/1995) analysis more (from 29.0% to smoking quit rates were
Follow up 4 times No computation of 24.9%) vs. smoking lower (ranges: 2.5–10.3% vs.
during one year elasticities prevalence in provinces 5.7–10.7%) in the provinces
UHVSRQGHQWV without taxes cut (from where taxes had been cut than
15 yrs) 31.0% to 28.3%) in those where taxes had not
Telephone survey been cut
Population in the
provinces with tobacco
tax cut : 5 770 100;;
population in the
provinces without
tobacco tax cuts: 7 019
200
Stephens et al. (1997) Cross-sectional Logistic regression OR of smoking vs. No mention of adjustment for
Canada General Social Survey (current smoking status) non-smoking (never and the analysis using the 1990
(GSS;; 1991) Adjusted for age, former together) HPS data
n=11 652 households sex, marital status, Smoking prevalence Smoking prevalence
LQGLYLGXDOV\UV education, policy associated with associated with infrequent
157
158

sample size)
Stephens et al. (1997) National representative variables (tobacco relatively low prices smoking regulations
(contd) of householders at a price, percent increase (OR=1.26;; (OR=1.21;; 95% CI 1.08-1.36)
Canada provincial level of price over 1 year and 95% CI 1.11-1.43) [according to results from the
Health Promotion 10 years, percentage of No association with analysis using the 1991 GSS
Survey (HPS;; 1990) population covered by price increase data]
n=14 000 no-smoking laws)
No details on interview Price of 200 cigarettes
procedure Percentage increase
in the price over 1 year

and 10 years
Gruber et al. (2003) Canadian Survey of Dependent variables: Overall Overall: -0.02 Overall: -0.41 When smuggling provinces
Canada Family Expenditure (1) total household -0.45 and years were excluded
(FAMEX) spending on cigarettes, By income quartile from the models, the price
8 pooled household (2) any spending on 1st estimates were similar.
surveys (1982–1998) cigarettes (prevalence) -0.99 As the price elasticity of
Analysis sample: 81 479 and (3) level of spending 2nd intensity was large, all of the
observations on cigarettes (intensity) -0.45 response of consumption
Two kinds of models: 3 rd occurred through reductions in
1) an OLS for total -0.31 consumption
spending 4th Using legal sales data, overall
2) a two-part model: (i) -0.36 price elasticity was 0.72,
OLS for prevalence, and By expenditure and declined to -0.47 when
(ii) OLS for intensity quartile excluding smuggling provinces
Controlled for fixed 1st and years.
effects for regions, -0.92
years, region-specific 2nd
linear time trends, -0.73
and household 3 rd
characteristics (after- -0.20
tax income, sex of the 4th
household head, and -0.37
family size)
Cigarette price was
instrumented using the
cigarette excise and
sales tax rate
Gospodinov & Irvine Cross-sectional Two-part model: 1) Overall Another computational method
(2009) Health Canadian Probit;; 2) OLS -0.299 based on the median instead
Canada Tobacco Use Adjusted for health By education of average of the partial effects
Monitoring Survey warnings, year of < High school evaluated at all observations
(CTUMS) (2000-2005) interview, household -0.299 provided similar elasticities
Analysis sample: size, sex, education, High school
\UV age, sex, profession, -0.333
language and region College
Price: retail price of a -0.300
pack of cigarettes
sample size)
Italy
Aristei & Pieroni (2009) Cross-sectional Box-Cox double-hurdle Overall Definition of age and
Italy The 2000 Italian survey model: 1) smoking -0.24 education:
on Health status and prevalence;; 2) smoking By sex Age = age in years
use of health services intensity Men Education = years of formal
(HSHS) Adjusted for age, -0.13 education
Analysis sample: education, sex, marital Women Reported results were
N=47 777 ever smokers status, social and -0.65 obtained from the IV Box-Cox
(current and former) occupation status, double hurdle models in which
Men: n=31 912 health status, smoking peer effects were treated as
Women: n=15 865 habit, BMI, chronic endogenous variables
illness, physical The presence of smokers
activity, children, and (peer measures) significantly
number of earners and increased the probability
other smokers in the of remaining a smoker and
household smoking intensity
Price: weighted average
price of national per
pack prices (including
excise taxes) of both
domestic and imported
cigarettes
Republic of Korea
Chung et al. (2007) Telephone interview Two-part model: 1) Overall Overall Overall Three logistic regressions and
Republic of Korea survey logistic regression -0.66 -0.02 -0.64 four OLS models were run with
Analysis sample: 3000 (smoking prevalence);; different number of covariates
PDOHV\UV 2) OLS (smoking included
intensity) The inclusion of other
Adjusted for age, behavioural risk factors
education, religion, substantially reduced the price
occupation, family elasticity of prevalence but had
income, marital no effect on the price elasticity
status, children, of intensity
area of residence,
lifestyle habits, BMI,
health status, health
perception, age at
starting smoking, and
parental and friend’s
smoking
No experienced smoker
is excluded
Spain
García & Labeaga Spanish Households A double-hurdle model Price elasticities Elasticities of total expenditure:
(1996) Budget Surveys (EPF) Covariates: total Base Base: +0.03
Spain 1980–1981 real expenditure, -0.24 Lowest education: +0.01
n=23 669 households occupation, Highest education: +0.12
159
160

sample size)
García & Labeaga employment status, Lowest education Unemployed: +0.09
(1996) (contd) education, size of the -0.52 Head of household aged
Spain town of residence, age Highest education \HDUVROG
and age squared, and +0.07 Base: a household having four
household size Unemployed members, the head was a non-
Price: retail price index -0.31 manual worked age 50 who
(base year=1976) 7KHKHDGDJHG had secondary education and
years old: 0.70 lived in an Andulusian town of
50 000–500 000 inhabitants

For comparison estimation
results, a Tobit model was also
estimated
Jiménez-Martín et al. Spanish Continuous A double-hurdle Probability of smoking Intensity The Almost Ideal Demand
(1998) Family Expenditure model: (i) probability of WG static model WG static model System was applied. A lagged
Spain Surveys (ECPF) smoking (participation Sample 1 Sample 1 smoking/consumption was
1985–1995 rate), and (ii) intensity -0.16 -0.47 used as an independent
Individual data that (budget share on Sample 2 Sample 2 variable in the dynamic model
were used to construct tobacco per smoking -0.14 -0.40 Elasticity estimates obtained
population averages household) WG dynamic model WG dynamic model from the static OLS models
for smoking prevalence Adjusted for education, SR elasticity SR elasticity using pooled data:
and intensity per employment status, Sample 1 Sample 1 Prevalence
household were from rural/urban area, -0.13 -0.21 Sample 1: -0.28
38 quarterly cross- number of dependents Sample 2: Sample 2 Sample 2: -0.72
sections of the ECPF in the household, an -0.11 -0.22 Intensity
for 1985(1)–1994(2) indicator describing the LR elasticity LR elasticity Both samples: -0.54
Two samples of cohort anti-tobacco legislative Sample 1 Sample 1: Price elasticities obtained from
data were formed measures, and -0.20 -0.56 the models using aggregate
based on (*) the date quarterly dummies Sample 2 Sample 2: individual data and the
of birth of the head of Tobacco expenditure: -0.16 -0.50 heterogenous models (cohort
the household (sample weekly expenditures by cohort estimated) varied
1) and both on (*) and on any kind of tobacco much and were not reported
on his/her education grossed up to the here
level (sample 2). Both quarter
samples contained 456 Reported results here
cell means total were obtained from the
static pooled within-
groups models (WG
static models) and from
the dynamic pooled
within-groups models
accounting for cohort
fixed effects (WG
dynamic models)
sample size)
Labeaga (1999) The Spanish Permanent A double-hurdle rational Price: real price of Model where total Model where total Model where total
Spain Survey of Consumption addiction model with tobacco (price of expenditure was expenditure was expenditure was strictly
(EPC) heterogeneity tobacco was deflated endogenous endogenous exogenous
Unbalance panel of A two-step structural by means of a weighted Short-run price Short-run price Short-run price elasticities
households model using a average of prices of elasticities elasticities Overall
n=34 413 households generalized method of nine goods) Overall By education -0.27
moments (GMM) within- -0.27 No studies: -0.30 Ever smoker
groups estimator Ever smoker +LJKVFKRRO 0.17
Covariates: sex, -0.17 University: -0.21 By education
occupation, Long-run price By age (in years) No studies: -0.30
employment status and elasticities +LJKVFKRRO
education of the head, Overall University: -0.21
number of members -0.36 By income decile By age in years
over 16 yrs, household Ever smoker D1 poorest: -0.50
composition, age and -0.23 D10 richest: -0.17
age squared of the Manual workers: -0.267 By income decile
head and the spouse, Long-run price D1 poorest: -0.50
region of residence, and elasticities D10 richest: -0.17
housing tenure dummy By education Manual workers: -0.267
variables. No studies: -0.40 Long-run price elasticities
Lags and leads of real +LJKVFKRRO Overall
prices and income University: -0.28 -0.36
appeared in several By age in years Ever smoker
models. Income was -0.23
also considered as By education
endogenous in some By income decile No studies: -0.40
models D1 poorest: -0.67 +LJKVFKRRO
D10 richest: -0.23 University: -0.28
Manual workers: -0.36 By age in years

By income decile
D1 poorest: -0.67
D10 richest: -0.23
Manual workers: -0.36
United Kingdom
Jones (1989) Household survey data A double-hurdle Price elasticity of per Elasticity of Elasticity of intensity Equation of budget share was
United Kingdom 1954–1986 consisted approach capital demand at participation rate at at 1986 values similar to the Almost Ideal
of the annual series A seemingly unrelated 1986 values 1986 values -0.370 Demand System model
published by the regression for (i) -0.562 -0.192
Tobacco Advisory participation rate and
Council 1954–1974 (ii) budget share of
and the bi-annual cigarettes (intensity)
data from the General Covariates in the two
Household Survey equations: price of
(GHS) 1972–1986 cigarettes and total
The data were used to consumers’ expenditure
derive the proportion of per capita at constant
161
162

sample size)
Jones (1989) (contd) cigarette smokers in the prices, and dummies
United Kingdom adult population aged capturing both
over 16 years old permanent effects of
three health events and
returning trends of the
health scares
Low- and middle-income countries
Bulgaria
Sayginsoy et al. (2002) The study is based on a Control variables: Overall HNP study
Bulgaria 1995 household survey Income, Average age -0.80
and encompasses 2259 of household members, Low- and lower-
households highest education of middle- income
a household member, earners
alcohol consumption, -1.33
ratio of adult males in Upper-middle- income
household earners
-1.02
High-income earners
-0.52
China
Mao & Jiang, (1997) Random cluster Two-part model for Overall
China sampling in Sichuan cigarette use: 1) logit -0.69
Province (1995) (smoking prevalence);;
N=2431 subjects and 2) OLS (smoking
\UVLQFOXGLQJ intensity)
men and 921 women.
Mao et al. (2003) National Smoking Two-part model and Overall 3RYHUW\JURXS50%
China Consumption Survey two-stage least square -0.513 household income per month;;
(1998) method By sex /RZLQFRPHJURXS50%
N=24 641 adults Adjusted for gender, Men household income per month;;
\UVLQFOXGLQJ race, education, -0.45 High-income group: >500 RMB
12 854 men occupation, income, Women household income per month;;
and 11 786 women age, price, area, -0.69
perception of risk, By income group
knowledge, propaganda Poverty
of tobacco control -1.906
Price: expenditure of Low-income
tobacco/consumption of -0.774
tobacco (packages) High-income
-0.507
sample size)
Lance et al. (2004) China Health & Nutrition Two-part model for Overall Overall Included in the same study
China Survey cigarette use: 1) logit -0.019 -0.063 with the Russian analysis
(1989–1993 waves) (smoking prevalence);; With fixed effects With fixed effects above
Cross-sectional (from and 2) OLS (smoking (province and (province and
follow-up studies) intensity) urbanicity) urbanicity)
1 PHQ\UV Adjusted for wealth, -0.045 -0.056
age, education, With fixed effects With fixed effects
household size, (community-level) (community-level)
community, price -0.034 +0.027
deflator and survey year
Price: cigarette price
of local market where
participants resided
Lee et al. (2004) Two random samples (1) OLS for smoking Overall In 2002, after Taiwan, China
China (Taiwan) of smokers: 1488 in intensity by smoker By survey year had enacted the new tax
2000–2001 and 1014 in group 2000: -0.31 scheme, there was an increase
2002–2003 (17–69 yrs) Current smokers were 2001: -0.37 in cigarette price elasticity
Annual face-to-face categorized by age, 2002: -0.53 Higher elasticity for men, for
interview sex, education, monthly 2003: -0.31 lower-income men, and for
Analysis sample: 2502 income, and amount Period 2001–2002 smokers of light cigarettes. No
current smokers (4 of cigarettes smoked. Overall specific pattern in strata of age
surveys) Price and income were -0.34 and education
mutually adjusted. No By sex Income elasticity provided
additional covariates Men: -0.31 negligible estimates (ranging
were added into the Women: -0.12 from 0.00 to +0.03)
model Period 2002–2003
New tax scheme in Overall
2002 -0.41
(2) OLS for smoking By sex
intensity by time period Men: -0.39
Adjusted for sex, age, Women: -0.14
education, monthly
income, and smoking
degree
Cigarette price: average
retail price of the top
three most consumed
cigarettes
Tsai et al. (2005) Two-year follow-up Logistic regressions For reduced smoking For brand smoking with
China (Taiwan) cohort (face-to-face (for behavioural with respect to respect to higher retail
survey);; 2001 and 2002 change (i) reduction of increased retailed cigarette price: OR = 1.07
surveys smoking, and (ii) brand cigarette price by 1 NT$: Overall, 17.4% switched brand,
Analysis sample: switching). Odds ratios OR = 1.03 18.8% reduced smoking, 8.4%
n=501 male smokers were obtained both types of behavioural
\UV Adjusted for changes
demographic factors, The study also used
individual income, independent linear regressions
smoking behaviour, and the Zeller’s seemingly
163
164

sample size)
Tsai et al. (2005) living area, expense unrelated regression for
(contd) on smoking, addiction modelling changes of smoking
China (Taiwan) level, advertisement, behaviour
and favourite cigarette
brand
Change in smoking
behaviour before
(2001) and after the
introduction of a new

cigarette tax scheme
(year 2002)
Bishop et al. (2007) 1995 Chinese A two-part model: (i) a Province and county Province and county Province and county The estimated price elasticity
China Household Income probit model (smoking -0.463 -0.213 -0.250 of demand obtained from an
Project (10 provinces) prevalence);; (ii) a linear County only County only County only estimated Tobit model was
Urban adult males regression model -0.511 -0.209 -0.303 1.5. This was not considered,
Province and county: Explanatory variables: as the fat tail of the smoking
N=4051 income, education, age, distribution violated the
County only: N=1963 marital status, number normality assumption of Tobit.
of children in the
household, and mean
price
Price = household
expenditures on
cigarettes/total number
of packs used by a
household
Mao et al. (2007) National Smoking A two-part model for Overall Overall Overall
China Prevalence Survey cigarette use: 1) logit -0.154 -0.064 -0.090
(NSPS;; 2002) (smoking prevalence);; By income group By income group By income group
Face-to-face and 2) OLS (smoking Poor Poor Poor
interview. A national intensity) -0.589 -0.478 -0.111
representative sample Adjusted for sex, age, Low Low Low
1 \UV education, family -0.234 -0.199 -0.035
income, residence area Middle Middle Middle
and smoking initiation -0.018 +0.093 -0.111
age High High High
Cigarette price: average +0.257 +0.340 -0.083
market price (computed
as the average of the
self-reported prices
among smokers)
assigned to the
participants by their
residence location
sample size)
Chang & Chiang (2009) Cross-sectional OLS models (smoking All male smokers The sample was limited to men
China (Taiwan) National Survey on intensity) with a double -0.40 because of low prevalence of
Knowledge, Attitude log specification for: Depressive male smoking among women
and Practice of Health i) depressive smokers smokers
Promotion (2002–2003) ii) non-depressive -0.82
Face-to-face interview smokers Non-depressive male
Representative sample iii) all smokers smokers
Analysis sample: Adjusted for age, -0.41
PHQ\UV education, income,
marital status, and
urbanization
Cigarette price = ratio
of cigarette expenses
to the number of
cigarettes smoked (per
day)
Estonia
Taal et al. (2004) Monthly household Myopic addiction model Price elasticity HNP study
Estonia data from the A log-linear model -0.34
Household Income and for smoking intensity
Expenditure Study, (average monthly
1992 to 1999 cigarette consumption)
Explanatory variables:
Average real income
per household member,
real tobacco product
price index, quarterly
dummies, time trend,
lagged consumption
Egypt
Nassar (2003) The study considered Control variables: National average HNP study
Egypt the 1995–6 and Income quartile (poorer price elasticity for
1999–2000 household households have “tobacco”, not only
expenditure surveys. higher price elasticity), cigarettes
education -0.40
Expenditure elasticities Urban households
were calculated by -0.41
expenditure groups, Rural households
educational level, work -0.39
status and urban/rural
area
India
John (2008) Cross-sectional survey OLS regression Rural (own Own- and cross-price
India (1999–2000) Price of each tobacco elasticities) elasticities considered
N=120 309 households product: average unit Bidis A demand system model
Household with zero values (self-reported -0.91 Spatial variation in prices of
165
166

sample size)
John (2008) (contd) consumption of tobacco expenditure divided by Cigarettes tobacco products were used to
India (37% from rural and quantity) -0.41 estimate own- and cross-price
60% from urban) were Leaf tobacco elasticities
excluded from the -0.87 See also Table 5.3
analyses Urban (own
Unit of analysis: elasticities)
household Bidis
-0.87
Cigarettes
-0.179
Leaf tobacco
-0.877
When symmetry
restrictions were
imposed (for the
unique substitution
complementary
patterns), own price
elasticity estimates:
Rural
Bidis
-0.92
Cigarettes
-0.34
Leaf tobacco
-0.87
Urban
Bidis
-0.86
Cigarettes
-0.20
Leaf tobacco
-0.87
Indonesia
Adioetomo et al. (2005) Study using the 1999 Control variables: Overall price HNP study
Indonesia National Socio- Cigarette price in elasticity
Economic Survey rupiah per pack of 16 -0.61
(Susenas), collected by cigarettes (expenditure By income groups
the Central Bureau of on cigarettes divided (lowest to highest)
Statistics. by the quantity of Q1: -0.67
cigarettes consumed), Q2: -0.33
per capita household Q3: -0.31
income per day in
rupiah, dummy for
urban/rural location,
dummies for education,
profession, age and sex
sample size)
Mexico
Jiménez-Ruiz et al. Cross sectional Two-part model for Overall Overall Overall Income elasticity
(2008) (7 surveys) cigarette use: 1) probit -0.52 -0.06 -0.45 Overall
Mexico National Household (smoking prevalence);; +0.49
Income and 2) log-log OLS (smoking For prevalence
Expenditure Survey intensity) +0.25
(1994–2005) Adjusted for household For smoking intensity
N=109 089 households income, education, sex, +0.24
(individuals aged age of the household
\UV head, alcohol intake,
Nationally number of adults in the
representative samples household, and year of
Unit of analysis: interview
household Price: ratio of total
expenditure on
cigarettes per week
to the number of
cigarettes smoked.
Average price paid by
each household
Myanmar
Kyaing (2003) The survey was The price elasticity of Total price elasticity Price elasticity of Conditional price HNP study
Myanmar performed in 2001 demand was estimated -1.62 smoking participation: elasticity of demand: In Myanmar tobacco is
and includes 9847 using the two-step -1.28 average, but -0.34 average, but consumed primarily in the form
households procedure described in varies from –1.09 for varies from –0.42 for of cheroots
Onder (2002). poorest quintile to -1.41 poorest quintile to –0.24
Control variables: for middle quintile for richest quintile
Income, price, age,
education and literacy,
gender, marital status
and urban/rural
residence
The study considered
the household
expenditure on
cigarettes, cheroots
and phet kyan (tobacco
covered with thenatphet
leaves).
Kyaing et al. (2005) Household survey A log-log OLS model Conditional price HNP study
Myanmar undertaken for the was used for the elasticity for all
study among low- estimation tobacco products
income groups in two Control variables: -0.11
peri-urban communities Income, addiction, Conditional price
and four townships. gender, marital elasticity for cheroots
The data was collected status, education and -0.36
among tobacco users occupation
167
168

sample size)
Kyaing et al. (2005) only (2414 tobacco Conditional price
(contd) users interviewed) elasticity for
Myanmar cigarettes
-0.25
Nepal
Karki et al. (2003) For this study 1400 Based on the cross- Total price elasticity Price elasticity of Conditional price HNP study
Nepal households (about sectional study a (average) smoking participation elasticity of demand
4000 people) were two-step procedure, -0.88 -0.46 -0.42

interviewed. similar to that of Onder Total price elasticity
(2002) was applied to among youth (aged
determine the price 15-24) was much higher
elasticity of smoking than the average (-1.88)
participation and
the conditional price
elasticity of demand.
In this comprehensive
study price elasticities
were estimated to
determine the likely
impact of price and
tax increases on
consumption and
government revenue.
Control variables:
Income, age, gender,
literacy, education level,
occupation, urban/rural,
number of years that
the person has smoked
Poland
Gardes & Starzec Polish Consumption Three different models: Average elasticity
(2004) Panel database 1) Tobit from the 3 models
Poland (1987–1990) 2) Probit (long-run Short-run
3630 households elasticity) -0.4
3) Frisch direct method. Long-run
Adjustment for location, -0.7
education dummies,
logarithmic equivalence
scale, log age,
proportion of children,
quarter dummies
sample size)
Russia
Ogloblin & Brock (2003) Cross-sectional Probit models for men Overall Income elasticity overall:
Russia The Russia Longitudinal and women Men Men: -0.007
Monitoring Survey Computation of price -0.085 Women: +0.047
(RLMS);; two waves of elasticity of smoking Women
data collected in1996 prevalence by using -0.628
and 1998–1999 the resulting marginal For 1996
N=4718 households effects Men
LQGLYLGXDOV\UV Adjustment for age, -0.046
Analysis sample: age2, community Women
Men: N=6015 type, marital status, -0.430
Women: N=8457 education, occupation, For 1998
real income, alcohol Men
drinking and BMI -0.120
Price: geometric Women
average of 4 real prices: -0.919
lowest price of domestic
cigarettes, highest price
of domestic cigarettes,
lowest price of imported
cigarettes and highest
price of imported
cigarettes
Lance et al. (2004) Russian Longitudinal Two-part model: 1) logit Overall Overall Included in the same study
Russia Monitoring Survey (smoking prevalence);; -0.106 -0.026 with the Chinese analysis
(1996, 1998, 2000) and 2) OLS (smoking With fixed effects With fixed effects below
N=10 638 men (>13 yrs) intensity) (province x urbanicity) (province x
Adjusted for wealth, -0.101 urbanicity)
age, education, With fixed effects -0.026
household size, (community-level) With fixed effects
community, price -0. 050 (community-level)
deflator and survey 0.000
round
Price: cigarette price
of local market where
participants resided
South Africa
Berg & Kaempfer Living Standard Censored maximum By ethnicity By ethnicity
(2001) Measures Survey likelihood (ML) and Black Black
South Africa (LSMS), 1993 censored least absolute -0.80 +0.34
Study population: 1131 deviation (LAD) White White
black households and estimation techniques -1.79 +0.09
998 white households Adjusted for prices of
various food items,
number of adults and
children, and income
169
170

sample size)
Van Walbeek (2002) Income and expenditure Cigarette demand Price elasticity of Income elasticity
South Africa surveys (IES) of 1990, (intensity) was smoking intensity for By year and income quartile
1995 and 2000 determined by the real the period 1990–1995 Q
Analysis unit: price of cigarettes and By income quartile Q 1990
household real income Q1 lowest: 1.39 Q1 lowest: +0.27
Analysis sample: Price elasticity for each Q2: 1.13 Q2: +0.48
1990: N=14 332 income quartile was Q3: 1.08 Q3: +0.40
1995: N=16 903 estimated as the net Q4 highest: 0.81 Q4 highest: +0.08
Only households change in cigarette 1995

that lived in urban consumption divided by Q1 lowest: +0.52
areas and had the change in the real Q2: +0.41
(positive) expenditure price of cigarettes Q3: +0.54
for cigarettes were The tax burden with Q4 highest: +0.24
analyzed respect to income, i.e. Cigarette price increases did
the ratio of household’s not fall disproportionately
total annual expenditure heavily on the poor during
on cigarettes to 1990–1995
household income, was
estimated in the first
place
Price: retail price each
year
Sri Lanka
Arunatilake & Opatha Survey data of about Control variables: Overall price HNP study
(2003) 7500 households. Income, occupation, elasticity The aim of the study was
Sri Lanka education, male ratio -0.53 to determine the impact
(males smoke more) By expenditure of price increases on
quintile smoking prevalence and
Q1 (poorest) smoking intensity of different
-0.64 expenditure quintiles. The
Q2 price elasticity of smoking
-0.55 participation was generally
Q3 small and insignificant (and
-0.60 sometimes even positive), but
Q4 the conditional elasticities are
-0.68 as expected
Q5
-0.29
Thailand
Sarntisart (2003) A sub-sample of 11 968 Using a linear Average price elasticity: HNP study
Thailand households that bought expenditure system -0.39;; price elasticity Demand for cigarettes among
cigarettes from 24 747 (LES) approach, the varies between –1.00 rural households is generally
households surveyed in study estimated price, for poorest urban less elastic than among urban
the 2000 household cross-price and income households and –0.04 households.
sample size)
Sarntisart (2003) socioeconomic survey elasticities of 12 for richest urban
(contd) different categories of households
Thailand household goods and
services
Control variables:
Expenditure on
cigarettes and other
tobacco products,
household income,
prices of 12 consumer
goods, age, education,
dummy: urban/rural
Turkey
Onder (2002) This part of the study Using a two-step Total price elasticity Price elasticity of Conditional price HNP study
Turkey considered the 26 166 model, the determinants -0.41 smoking participation: elasticity of demand:
households covered in of smoking participation -0.03 average, but -0.39 average, but
the 1994 Household were estimated first, varies from –0.32 for varies from –0.58 for
Expenditure Survey using a logit model. second poorest quintile second poorest quintile
In the second step, to 0.15 for richest to –0.30 for richest
the conditional price quintile quintile
elasticity of demand
was estimated for those
households that decide
to smoke. The results
were used to determine
the impact of a change
in the tax rate on
cigarette consumption,
government tax revenue
and the regressivity of
the tax.
Control variables:
Income, education, age,
geographic region.
Bilgic et al. (2009) National household A zero-inflated negative Covariates in the Price elasticities of Income elasticities of
Turkey expenditure survey binomial model: 1) equations: age, smoking intensity smoking intensity
(2003) logit (for non-smoking sex and education Family with teenagers Family with teenagers
n=22 208 households prevalence);; 2) parent of the head of the -0.264 +0.075
Families with teenagers: negative binomial (for household, household Family without Family without teenagers
7844 number of cigarette income, households’ teenagers +0.058
Families without packs purchased per socioeconomic -0.221
teenagers: 14 364 week) characteristics, age
groups, special health
insurance status,
number of members
having regular
medication, share of
171
172

sample size)
Bilgic et al. (2009) health care expenditure,
(contd) alcohol intake, number
Turkey of smokers in the family,
and regional variables
Price: price of cigarette
pack in Turkish Liras
in 2003
Ukraine
Krasovsky et al. (2002) The study used A two-part model: Price elasticity of Price elasticity of HNP study
Ukraine data from a survey 1) probit (smoking smoking prevalence smoking intensity
of individuals (both prevalence);; 2) OLS Overall By income group and
smokers and non- (smoking intensity) -0.47 age in years
smokers) conducted in Control variables: By income group High income
all regions of Ukraine For smoking High income 14–17: -0.52
(2700 individuals) prevalence: age, sex, -1.1 18–28: -0.24
education, marital Low-income 28+: -0.15
status, urban/rural -0.27 Middle income
region, smoking status 14–17: -0.7
of a close relative or a 18–28: -0.42
friend, and income 28+: -0.33
For smoking intensity: Low income
age, sex, price, 14–17: -0.65
household income, 18–28: -0.37
strength of addiction 28+: -0.28
(assuming that on
average heavily
addicted smokers
smoke more), region
(assuming that smoking
patterns may differ
across regions of
Ukraine), and dummy
reflecting whether a
smoker has under-age
children
Viet Nam
Van Kinh et al. (2006) Household-level cross- Two-part model: 1) VINATABA price VINATABA price VINATABA price The coefficients of the 555
Viet Nam sectional data from Viet Linear probability model Overall Overall Overall brand price resulting from the
Nam Living Standards (smoking prevalence);; -1.41 -0.94 -0.47 linear probability model were
Survey (VLSS;; 2) OLS (smoking By income By income By income not statistically significant
1997–1998) intensity) Two low quintiles Two low quintiles Two low quintiles at the 10% level, and the
N=6000 households Adjusted for price of -1.77 -1.16 -0.61 estimated price elasticities of
with 28 518 individuals pipe tobacco, annual Two high quintiles Two high quintiles Two high quintiles smoking prevalence were not
\UV per capita -1.17 -0.75 -0.42 provided
sample size)
Van Kinh et al. (2006) income, individual 555 price
(contd) characteristics (sex, Overall
Viet Nam age, education, -0.54
work experience, By income
occupation), household Two low quintiles
characteristics -0.57
(household size, and Two high quintiles
sex, age, education, -0.37
occupation of the Communal average
household head), price:
and geographic Overall
and commune -0.50
characteristics By income
Price: price of Two low quintiles
cigarettes at the -0.85
commune level (price Two high quintiles
of VINATABA brand, -0..35
price of 555 brand, and Average price
average price of the two elasticity
brands) Overall
-0.50
By income
Two low quintiles
-0.59
Two high quintiles
-0.40
NHANES: National Health and Nutrition Examination Survey;; CPS: Current Population Survey;; NHIS: National Health Interview Survey;; BRFSS: Behavioural Risk Factor Surveillance System;; COMMIT: Community Intervention Trial for
Smoking Cessation;; HNP: Health, Nutrition and Population
173
He used alternative specifications conducted between 1970 and 1985 1999) focused their analyses on men.
of the rational addiction model and to examine how the price elasticity Given the lack of data on smokeless
estimated these models for the full of adult cigarette demand was tobacco prices, they employed state-
sample, the sample of those who changing over time. Using two level taxes on smokeless tobacco
ever smoked, and the sample of alternative models (a two-part model products (typically ad valorem taxes
current smokers. For these three that looked separately at prevalence applied at the wholesale level) as their
samples, he obtained overall long- and intensity, and a generalized proxy for price, using the resulting
run price elasticities of cigarette linear model that looked at the estimates and information on the
GHPDQG LQ WKH UDQJH IURP í WR overall impact), Wasserman and share of price accounted for by tax
í í WR í DQG í colleagues (1991) found that cigarette to produce price elasticity estimates
WR í UHVSHFWLYHO\ &KDORXSND demand was becoming increasingly for snuff and chewing tobacco. They
1991). Consistent with the predictions responsive to price over time, going found consistent evidence that higher
of the rational addiction model that from virtually insensitive to price in smokeless tobacco taxes were
individuals with a greater preference the early/mid-1970s to a predicated associated with reduced prevalence
for present would respond more to RYHUDOO HODVWLFLW\ RI í WR í of any smokeless tobacco use among
price, Chaloupka found that smoking by 1985–88. Consistent with Lewit men, with prevalence elasticities in
among less-educated persons was and Coate (1982), Wasserman and WKHUDQJHIURPíWRí
influenced more by price than was colleagues (1991) found that price In the most comprehensive
smoking among more educated had a greater impact on smoking analyses to that point, Farrelly and
persons. In contrast to the predictions SUHYDOHQFH HODVWLFLW\ RI í LQ Bray (Centers for Disease Control
of the rational addiction model and 1985) than on smoking intensity and Prevention, 1998) and Farrelly
to Lewit and Coate’s (1982) finding HODVWLFLW\RIíLQ and his colleagues (2001) pooled
that younger populations were more Hu and his colleagues (1995) multiple waves of the National
sensitive to price than were older were the first to estimate the price Health Interview Survey conducted
populations, Chaloupka (1991) found elasticity of adult cigarette demand between 1976 and 1993 to estimate
that young adults (17–24 years) using state-specific individual-level cigarette demand for US adults, as
were less responsive to price than survey data. They pooled the 1985 well as for multiple subpopulations
middle-aged adults (25–64 years). through 1991 waves of the California defined by gender, race/ethnicity,
Chaloupka (1990) used similar Behavioural Risk Factor Surveillance income and age. They used two-part
models for men and women and System survey. They used two- models that controlled for a variety
found that cigarette demand was part models of cigarette demand of individual characteristics and for
significantly more responsive to price that controlled for other health region. They obtained overall price
among men than among women. behaviours, in addition to the typical elasticities of adult cigarette demand
Finally, Chaloupka (1992) estimated controls included in prior studies. LQ WKH UDQJH IURP í WR í
additional specifications of these Their estimates for California were with similar elasticities estimated
models that included explanatory consistent with those obtained from IRU VPRNLQJ SUHYDOHQFH í WR
variables to control for the presence previous studies for the USA, with an í DQG VPRNLQJ LQWHQVLW\ í
of various state-level restrictions on RYHUDOOSULFHHODVWLFLW\RIí3ULFH WR í &RQVLVWHQW ZLWK /HZLW
smoking in public places. Estimated had a somewhat greater impact on and Coate (1982), they found that
price elasticities obtained from smoking prevalence than on intensity. cigarette demand became more
these models were comparable In a series of papers using inelastic with age and that smoking
to those described above. This different waves of the Tobacco Use was more responsive to price among
result suggests that there were no Supplement to the Current Population men than among women. They also
significant omitted variables biases Survey, Ohsfeldt and his colleagues found relatively greater sensitivity to
in the price elasticity estimates (1994, 1997, 1999) were the first to price of cigarette use among minority
when the smoking restrictions were estimate the impact of smokeless populations (Hispanics and blacks)
excluded from the models. tobacco taxes on smokeless tobacco than among the majority population
At about the same time, use in the USA. Because of the very (whites) and among those in lower-
Wasserman and his colleagues low prevalence of smokeless tobacco income households compared to
(1991) pooled seven waves of the use among women in the USA, those in higher-income households.
National Health Interview Survey Ohsfeldt and colleagues (1994, 1997,
174
Several studies of US adult Consistent with the early estimates price than was cigarette smoking
cigarette demand have been of Lewit and Coate (1982), they among women. Stehr (2007) argues,
published over the past decade. found that higher cigarette prices however, that this is due to an omitted
These more recent studies have significantly reduced smoking variables bias resulting from a
taken advantage of the numerous prevalence among all age groups. negative correlation between gender
and often large local, state and Interestingly, however, the oldest differences in smoking prevalence
federal cigarette tax increases, group they examined (65 years and and cigarette taxes. Using data from
increases in prices due to the pass- older) was as price-responsive as the the 1985 through 2000 Behavioural
through of costs from settlements of youngest group (18–20 years), with Risk Factor Surveillance System,
legal challenges against the tobacco estimated prevalence price elasticities Stehr (2007) developed gender-
industry (most notably the 1998 RIíDQGíUHVSHFWLYHO\7KH specific two-part models that include
Master Settlement Agreement that prevalence elasticity values for the gender-specific state fixed effects to
led to an immediate 45-cent increase age groups in between were less account for this correlation. He found
in prices), and other factors that than half of those values reported in that smoking among women was
have contributed to increasing price the oldest and the youngest groups. significantly more responsive to price
variation over time. At the same time, Over the past two decades, than was smoking among men. For
many of the more recent studies have smoking prevalence among those women, his estimated prevalence and
pooled numerous survey waves, with who smoked less than daily increased WRWDOSULFHHODVWLFLWLHVZHUHíDQG
estimation samples exceeding one considerably, with almost one in five íDQGIRUPHQWKHFRUUHVSRQGLQJ
million respondents in some (e.g. smokers smoking less than daily in HVWLPDWHVZHUHíDQGí
Sloan & Trogdon, 2004;; Franks et 2006–07. Tauras (2004) was the first DeCicca and McLeod (2008)
al., 2007;; Stehr, 2007;; Farrelly & to assess the impact of price on the similarly used multiple waves of the
Engelen, 2008;; Franz, 2008). The decision to smoke daily versus less Behavioural Risk Factor Surveillance
use of multiple survey waves over than daily, using data from the 1991, System data from 2000–2005
a period where taxes and prices 1993 and 1994 US National Health to examine differences in price
have changed considerably in many Interview Surveys. Tauras (2004) elasticity among a variety of US
states has allowed researchers found that higher prices increase population subgroups. Consistent
to better control for a variety of the likelihood that a smoker will with Sloan and Trogdon (2004), they
other tobacco control policies2 smoke less than daily in addition to found that smoking among older
and underlying sentiment towards reducing smoking prevalence and adults was more responsive to price
tobacco. Nevertheless, the findings average cigarette consumption. His than previously thought, suggesting
from these more recent studies are price elasticity estimate of some that higher taxes and prices increase
remarkably similar to those from the day smoking conditional on current the likelihood of smoking cessation
earlier studies. They consistently smoking is 0.86, while his estimates at all ages. Among other subgroups
suggest that higher cigarette taxes for smoke-free air policies were in older adults, they found that higher
and prices lead to reductions in the generally small and statistically taxes have their greatest impact on
prevalence and intensity of smoking insignificant. These findings suggest smoking among those in poor health,
among US adults. A few of these that much of the observed increase in among those on lower incomes
more recent studies are described prevalence of less than daily smoking (explored more fully in Chapter 7),
briefly below. resulted from the impact of tax and and among those with fewer years of
Sloan and Trogdon (2004) price increases rather than from education.
assessed the impact on smoking the impact of other tobacco control
prevalence among different age policies on smoking. Summary
groups of the large price increases in Stehr (2007) focused on gender
the US in the late 1990s that followed differences in the price sensitivity of Over the past three decades,
the settlement of state lawsuits with US cigarette demand. As described numerous studies of adult cigarette
tobacco companies, using data from above, most previous studies found demand based on individual-level
the 1990 through 2002 Behavioural that cigarette smoking among men survey data have been produced
Risk Factor Surveillance System. was somewhat more sensitive to in the USA (see 26 studies in Table
2
Most notably smoke-free air policies targeting various venues, such as workplaces, bars and restaurants, and funding for comprehensive tobacco control programmes.
175
5.1). Over time, the econometric Figure 5.1. Price elasticity for total cigarette consumption (prevalence and
intensity) from 16 studies conducted in the USA
methods used in these studies have
become more sophisticated, and the
amount of data used to for analysis
has increased exponentially. As with
the estimates from the aggregate
demand analyses for the USA
reviewed in Chapter 4, these studies
consistently find that higher cigarette
taxes and prices reduce cigarette
consumption, with most overall price
elasticity estimates in the range from
íWRí&KDSWHU$VVKRZQLQ
Figure 5.1, studies that examine both
smoking prevalence and intensity
generally find that the effects of price
on consumption are about evenly
split between the effect of price on
smoking prevalence and the effect of
price on intensity of smoking among
those who smoke. More recent
studies also find evidence that higher
prices increase the likelihood that
smokers will smoke less than daily. * elasticities estimated as the mean value between the upper and the lower limits of a range of estimates and/or as the
In general, earlier US studies that mean value between men and women’s estimates
have examined gender differences
in price elasticity found that smoking
Figure 5.2. Gender-specific price elasticities for total cigarette consumption
is more sensitive to price among (prevalence and intensity) from 8 studies conducted in the USA
men than among women. However,
studies conducted on more recent
data obtained at a more advanced
stage of the tobacco epidemic (Lopez
et al., 1994), and consequently
displaying a narrower gap between
men and women on smoking
prevalence and consumption, failed
to find differences in elasticities
(Figure 5.2). Moreover, Stehr’s
(2007) more recent study provides a
compelling explanation for why earlier
estimates were likely to be biased,
along with estimates that smoking
among women is roughly twice as
responsive to price as is smoking
among men once potential bias due
to the omitted variables is accounted
for. The current evidence suggests
no major differences between sexes
according to responsiveness to
changes in cigarette prices. * elasticities estimated as the mean value between the upper and the lower limits of a range of estimates
176
Similarly, most US studies that in prices, while substantial increases conducted between 1982 and 1998,
have considered differences in the and reductions in taxes provide with tax rates matched to the survey
price elasticity of cigarette by age considerable intertemporal variation. databased on respondents’ region
find that smoking among younger Stephens and colleagues (1997) and acknowledging the need to
age groups is more price-sensitive used data from two cross-sectional take into account the presence of
(described more fully in Chapter 6). surveys of Canadian adults, the significant smuggling of tobacco
However, price elasticity of demand 1990 Health Promotion Survey and products. They used two different
generally becomes monotonically the 1991 General Social Survey, to sets of data, household-level
more inelastic among older age study the impact of cross-sectional expenditures on smoking from all
groups. Sloan and Trogdon (2004) variability in prices and changes in households and another sample that
find that this is not the case among prices over time on adult smoking excluded provinces and years where
those over 65, perhaps due to the prevalence. They found that cigarette smuggling was thought to
more constrained resources for smoking prevalence was higher for be a significant problem. For both
retirees on fixed incomes. respondents residing in provinces samples, they obtained an overall
where cigarette prices were lower. price elasticity of cigarette demand of
Other high-income countries Hamilton and colleagues (1997) íZKLFKLVTXLWHFRQVLVWHQWZLWK
conducted a similar comparison the estimates described above for the
There have been several studies across provinces, focusing on the USA. These results suggest that the
that examine the effects of price on significant reductions in taxes that availability of smuggled cigarettes
consumption of cigarettes and other were made in some provinces in had little impact on price elasticity in
tobacco products in other high- early 1994. They used cohort data Canada. Additionally, they found that
income countries (see Table 5.1). from the Survey on Smoking in nearly all of the effect of price is on
These have been many fewer than Canada collected over four waves household cigarette consumption,
the number of studies conducted in in 1994 and 1995, with the baseline with little impact on smoking
the USA due to the fact that there survey collecting data on smoking prevalence. Furthermore, smoking
is limited within-country variability status on 1 January 1994, before was more responsive to price among
in tobacco product taxes and prices the tax cuts. To assess the impact of the lowest-income households than
in most of those countries. Many of the tax cuts on smoking prevalence, among higher-income households
these studies use data taken from they compared changes in smoking (see more detail in Chapter 7).
household consumption expenditures prevalence in provinces where taxes Most recently, Gospodinov and
surveys. Information on spending were cut with smoking prevalence in Irvine (2009) explored cigarette
and consumption is used to derive other provinces (in provinces where demand among Canadian adults,
price measures for use in empirical taxes were not cut). Hamilton and using data from the Canadian Tobacco
analyses. Such price measures are colleagues (1997) found that smoking Use Monitoring surveys conducted
likely to be endogenous because they prevalence fell in all Canadian from 2000 through 2005 and applying
at least in part reflect tobacco use provinces between January 1994 and two-part models. Consistent with
behaviours (e.g. heavier consumers February 1995, which was attributed Gruber and colleagues (2003),
purchase cheaper brands), leading to the implementation of various Gospodinov and Irvine (2009) found
to biased estimates of price elasticity. tobacco control efforts throughout little impact of price on smoking
Others take advantage of survey Canada. However, smoking prevalence in Canada, but did find
data on smoking and other tobacco prevalence was reduced more in the that higher prices significantly reduce
use, comparable to those price provinces that did not cut taxes than cigarette consumption, with an
measures used in the US studies, in the provinces that cut taxes (a 4.1 estimated price elasticity of intensity
linked with market-level measures of percentage point reduction versus a LQ WKH UDQJH IURP í WR í0.30.
prices based on respondents’ place 2.7 percentage point reduction). However, they found little difference
of residence. More recently, Gruber and in price elasticity estimates of
Four studies have examined colleagues (2003) conducted a more demand for socioeconomic groups
cigarette demand in Canada. rigorous, econometric analysis of defined by educational attainment
Differences in Canadian provincial cigarette demand. They used data (Gruber et al., 2003).
cigarette excise taxes provide from all eight waves of the Canadian Similarly, Australian researchers
significant cross-sectional variation Survey of Family Expenditure have taken advantage of differences
177
in cigarette prices across states to Figure 5.3. Price elasticity for total cigarette consumption (prevalence and
intensity) from five studies conducted in high-income countries other than
assess the impact of price on smoking
the USA
among adults (see Table 5.1). For
example, Cameron and Williams
(2001) used data from multiple
waves of the National Drug Strategy
Household Surveys conducted in
the late 1980s and 1990s that were
matched with state-level price data.
They obtained a price elasticity of
VPRNLQJSUHYDOHQFHRIí0RUH
recently, Zhao and Harris (2004)
used the 1995, 1998 and 2001
waves of the same surveys, finding
an even greater impact of tobacco
product prices on prevalence of
tobacco use (prevalence elasticity
RI í DV ZHOO DV D VLJQLILFDQW
negative impact of price on the
level of tobacco consumption. Most
recently, Wakefield and colleagues
(2008) used monthly data on
smoking prevalence collected in the
five largest Australian cities between * elasticities estimated as the mean value between the upper and the lower limits of a range of estimates
June 1995 and December 2006 to
examine the impact of changes in studies are generally consistent with price on tobacco use in low- and
cigarette prices relative to income (a those from the USA in concluding middle-income countries. Since
measure of affordability, as described that higher tobacco product taxes then, however, a rapidly growing
in the previous chapter) on smoking and prices lead to reduced smoking body of evidence has emerged from
prevalence during this period. Their among adults. While estimates of studies that use individual-level and/
finding implies that relatively small price elasticity from these studies are or household-level data collected in
changes in affordability would reduce more variable than those from studies low- and middle-income countries,
prevalence. based on US data, they do indicate with studies available for countries
There are relatively few other that tobacco use in high-income in all regions. Much of this research
studies from high-income countries countries is inelastic with respect has been produced with support from
that use individual-level or household- to price, and that price influences the World Bank, the World Health
level data to investigate the impact of both the prevalence and intensity of Organization, and other international
taxes and prices on adult tobacco use tobacco use. However, in contrast organizations;; publications are
(one from Italy, 1 from the Republic of with evidence from the USA, in other available in the Economics of
Korea, three from Spain and one from high-income countries the effect of Tobacco Control Discussion Paper
the United Kingdom). These studies price changes on smoking intensity series produced by the Health,
identified by the search strategy appears to be stronger than that on Nutrition and Population Family
described above are summarized in smoking prevalence (Table 5.3). (HNP) of the World Bank’s Human
Table 5.1 and in Figure 5.3. Development Network. A smaller
Low- and Middle-Income Countries number of these studies have been
Summary published in peer-reviewed journals.
Prior to the publication of the World As with the demand studies
Overall, 14 studies on adult cigarette Bank’s Curbing the Epidemic report based on aggregate data from
demand using individual or household (Jha and Chaloupka, 1999), there low- and middle-income countries
level survey data exist for high-income were almost no survey-based described in the previous chapter,
countries other than the USA. These studies on the impact of tax and there is also considerable variation in
178
the quality of the data and methods within-country variation both in taxes in 2002, with the cigarette price
used in these studies (Chapter 4). and in prices for a particular product/ measure based on the average of
Many studies employ household- brand, matching price data from prices self-reported by respondents
level consumption expenditure other sources to the survey data who resided in the same geographic
data that often reflects all tobacco can introduce its own measurement region (Mao et al., 2007).
consumption, making it difficult to errors that can bias price elasticity Others have produced similar
disentangle the effects of prices on estimates towards zero. Due to estimates for the price elasticity of
use of different tobacco products, these potential problems, much care Chinese cigarette demand. Bishop
on use by different members of the and caution must be taken when and colleagues (2007) used data on
household, and on prevalence and assessing the findings from these adult urban males in ten provinces
intensity of use. In many of these studies. taken from the 1995 Chinese
countries, there is considerable Given the variety of research Household Income Project. Prices
variability not only in the taxes on and available, the discussion below will that were assigned to individuals
prices of different tobacco products, briefly summarize the evidence were based on the average self-
but also in the taxes and prices of by region, highlighting seminal reported prices for respondents from
different brands for the same product. studies and those that are most the same province (which allows
This situation creates significant methodologically sound. Table 5.1 for more observations to be used in
endogeneity problems for measures contains a more comprehensive computing price) or county (which
of price that are derived from self- listing of the studies identified by the reduces the number of observations
reported prices or from self-reported search strategy described above. used to estimate price, but adds to
information on the consumption the variability in the resulting price
expenditures and consumption Asia. Several studies have measure). They estimate two-part
quantities. This typical endogeneity examined the impact of tax and price models using each of the alternative
almost certainly results in biased on tobacco use in various Asian measures of price. Interestingly, the
price elasticity estimates when it is countries, including China, India, Viet estimated elasticities with the two
not accounted for in the estimation. Nam, Thailand, Indonesia, Myanmar price measures are relatively similar,
Many of the studies described below and Nepal. Mao and Jiang (1997) were with prevalence elasticities around
address this problem by trying to the first to estimate the price elasticity íDQGLQWHQVLW\HODVWLFLWLHVLQWKH
assess the endogeneity of self- of cigarette demand using individual- UDQJHIURPíWRí,QFRQWUDVW
reported price using a Hausman or level data for a low- or middle-income Lance and colleagues (2004), using
other test. However, these efforts country, using cross-sectional data from multiple waves of the
are often hampered by appropriate survey data for adults in the Sichuan China Health and Nutrition Surveys
instrument variables for price being province that were augmented with that were supplemented with locally
unavailable in the survey data. price data collected from retailers collected prices from markets in the
Others address this by using a selling cigarettes in the survey communities where respondents
measure of average prices that is respondents’ locations. They used a lived, concluded that cigarette
derived from the self-report data two-part model of cigarette demand, demand in China was much less
and averaged over respondents producing elasticity estimates of elastic. Their prevalence elasticities
in the same location and/or based í IRU VPRNLQJ SUHYDOHQFH UDQJHG IURP í WR í DQG
on other factors (e.g. income). Still DQG í IRU VPRNLQJ LQWHQVLW\ ,Q intensity elasticities from 0.03 to
others apply two-stage least squares follow-up studies using nationally í +RZHYHU DV %LVKRS DQG
models to first predict price (based representative data for China, they colleagues (2007) note, the price
on taxes, location, income, and found that cigarette demand in China measure employed by Lance et al.
other factors), then use the predicted was becoming increasingly inelastic, (2004) ignores the fact that many
price in the demand models. Where ZLWK D SULFH HODVWLFLW\ RI í IRU smokers buy cigarettes from nearby
secondary sources for prices are smoking prevalence and that of wholesalers rather than from local
available, other potential problems í IRU FLJDUHWWH FRQVXPSWLRQ markets, introducing measurement
arise. For example, given the among smokers (Mao et al., 2007). error that can bias price elasticities
extensive variability in availability These recent estimates were based towards zero.
and prices of various tobacco on data from the National Smoking At least four studies from Taiwan,
products/brands and the limited Prevalence Survey conducted China have examined the impact of
179
taxes and prices on tobacco use. They income elasticities generally obtained households. For bidis and leaf
generally take advantage of large in these studies. tobacco, little difference in price
price changes that followed the large John (2008) conducted the only elasticities is found among urban
tax increases implemented in 2002. survey-based study that produces and rural households. For cigarettes,
For example, Lee and Colleagues estimates of price elasticity of adult consumption decisions made by
(2004) used survey data collected tobacco use in India. Using data rural households are significantly
by the Taiwan, China National Health from the 55th round of the National influenced by cigarette prices, but
Research Institutes between 2000 Sample Survey Organization survey, the same is not true for consumption
and 2003 that were augmented with a household expenditure survey decisions made by urban households
monthly data on average prices of conducted from July 1999 through (John, 2008).
leading brands to estimate the price June 2000, he examined the effect Several demand studies
elasticity of cigarette consumption of price on demand for cigarettes, published in the Economics of
among current smokers before and bidis and leaf tobacco separately for Tobacco Control HNP Discussion
after the tax increase for a variety of urban and rural populations. Using Paper series are based on household
population subgroups. Overall, their an empirical approach developed by expenditure or other survey data
estimated price elasticities range Deaton (1988), he constructed unit for various other Asian countries.
IURPíWRíZLWKWKHJUHDWHVW values by dividing the self-reported These studies produce a range of
price elasticity in the year following individual household consumption estimates, generally confirming
the tax increase. In a follow-up expenditure to consumption quantity. that higher taxes and prices will
study using survey data collected The unit values reflect the averages lead to reductions in tobacco use.
by phone in 2004, Lee (2008) for individual households in the Kyaing (2003) and Kyaing and his
used information from responses same geographic areas and account colleagues (2005), for example, used
to a question about a hypothetical for variations in the quality of the two expenditure surveys to estimate
price increase and obtained a price tobacco products consumed (John, price elasticity of adult tobacco use in
elasticity of cigarette consumption 2008). Since this analysis was Myanmar. Kyaing (2003) uses a two-
RI í 7KLV ZDV FRQVLVWHQW ZLWK limited to households that consumed part model to examine the impact
that obtained from survey databased tobacco products, the price elasticity of price on the use of cigarettes,
on actual experiences with tax and estimates reflect the impact of price cheroots, and phet kyan which,
price increases. Related studies on consumption by households that together, account for nearly all of
by Tsai and colleagues (2003, use these products, thus understating tobacco use in Myanmar. This study
2005) concluded that the 2002 tax the full impact of price on tobacco uses a price measure that is derived
increases altered other aspects of demand in India. John’s (2008) from self-reported expenditures and
smoking behaviour in Taiwan, China, analysis produced significantly consumption quantities, treating this
including brand choice and likelihood negative own-price elasticities of price variable as exogenous after
of purchasing smuggled cigarettes. demand for each of the three tobacco having conducted a Hausman test for
In general, and consistent with products considered, with less endogeneity of price. The estimated
the estimates from studies based inelastic estimates for bidis (ranging price elasticities are substantial—a
on aggregate data described in IURP í WR í DFFRUGLQJ WR SUHYDOHQFH HODVWLFLW\ RI í DQG
the previous chapter, it appears rural/urban area) and leaf tobacco DQ LQWHQVLW\ HODVWLFLW\ RI í ,Q
that cigarette demand in China has í WR í DQG D UHODWLYHO\ addition, he finds that price elasticity
become less sensitive to price over inelastic estimates for cigarettes of demand (in absolute values) falls
the past two decades. One likely UDQJLQJ IURP í WR í -RKQ with age, although demand remains
explanation for these increasingly attributes the differences in the price elastic even among the oldest age
inelastic price estimates for cigarette elasticity estimates to the differences groups, and that price elasticity
demand in China is the trend towards in income among those households rises and then falls with income.
increased affordability of cigarettes who consume cigarettes and those In the subsequent study (Kyaing
in the country, as real household who consume other products, with et al., 2005), separately examining
disposable incomes have increased cigarettes generally consumed the effect of price on consumption
much more rapidly than real cigarette by higher-income households, of cigarettes and cheroots among
prices. This trend is consistently while bidis and leaf tobacco are low-income consuming households,
described by the significant, positive typically consumed by low-income only tobacco-consuming households
180
were considered in the analysis. The Sarntisart (2003) takes a somewhat as cigarette prices nearly doubled
price variables that were derived different approach, applying a linear between the two waves of the survey
from self-reported prices were expenditure system model to data they analysed. In their subsequent
determined to be endogenous in from the 2000 Thailand Household analysis, Lance and colleagues
some specifications and exogenous Socioeconomic Survey that were (2004) added the 2000 wave of the
in others, based on Hausman augmented with local price data on survey to assess the impact of price
tests. When the price variable was many goods and services collected on both prevalence and intensity
endogenous, a two-stage least as part of the survey. He obtains an among men. They similarly found
squares model was estimated using RYHUDOOSULFHHODVWLFLW\RIíZLWK little impact of price on prevalence
taxes as instruments for price. demand increasingly inelastic among (prevalence elasticities ranging from
The resulting estimated intensity higher-income households and íWRíDQGDOPRVWQRLPSDFW
HODVWLFLWLHVZHUHíIRUFLJDUHWWHV relatively less inelastic among urban of price on consumption among
DQGíIRUFKHURRWV households. male smokers (intensity elasticities
Karki and colleagues (2003) used UDQJLQJ IURP WR í 2JOREOLQ
Nepalese household consumption Europe. Several studies of the impact and Brock (2003) attributed the low
expenditure data to estimate the of taxes and prices on adult tobacco intensity elasticities obtained for
price elasticity of combined cigarette use based on individual or household male smokers to the fact that they
and bidi demand after having survey data have been conducted mostly consume low-quality, very
concluded that price can be treated for a few former Soviet Republics inexpensive cigarettes (in contrast,
as exogenous based on a Hausman (the Russian Federation, Ukraine, many female smokers use the
test. They obtained an overall price and Estonia) and other central higher-price, higher-quality cigarette
HODVWLFLW\ RI í ZLWK WKH LPSDFW and eastern European countries brands).
of price on combined demand about (Bulgaria and Poland). In general, Krasovsky and his colleagues
evenly divided between its impact these studies find that higher taxes (2002) used data on over 2700
RQ SUHYDOHQFH í DQG XVH LQ and prices reduce tobacco use, with individuals they surveyed throughout
FRQVXPLQJ KRXVHKROGV í widely varying estimates of price Ukraine to explore differences in the
Adioetomo and colleagues (2005) elasticity across countries. price elasticity of cigarette demand
used data from the 1999 National Two peer-reviewed studies used by age and income. Their measure
Socioeconomic Survey to analyse data from the Russian Longitudinal of price was based on a self-reported
cigarette demand in Indonesia. To Monitoring Survey, a nationally price in response to a survey
account for the potential endogeneity representative household survey question that asked respondent
of price, they conducted a two- that collects data on household smokers “What is the price for a
stage least squares model, finding members’ tobacco use and includes pack of the cigarettes you usually
little impact of price on smoking a community module that collects smoke?” This measure of price is
prevalence, while estimating that the information on retail prices for the likely to be endogenous, reflecting
overall price elasticity of consumption highest and lowest domestic and a variety of factors including brand
DPRQJVPRNHUVLVíZLWKGHPDQG imported cigarette brands (Ogloblin choice, quantity purchased and
less inelastic among those smokers and Brock, 2003;; Lance et al., 2004). location of purchase. That is, heavier
on lower incomes. Arunatilake Ogloblin and Brock (2003) used data consumers will likely choose less-
and Opatha (2003) performed a from the 1996 and 1998–99 waves of expensive brands, buy in greater
similar analysis, using household the survey to investigate the impact of quantities, and obtain these products
consumption expenditure and price on smoking prevalence among from lower-priced locations. Their
consumption quantities taken from men and women. They concluded RYHUDOO LQWHQVLW\ HODVWLFLW\ ZDV í
the 1999–2000 Sri Lanka Integrated that smoking prevalence among while their other price elasticity
Survey database. They also find that men was relatively unresponsive estimates indicated that younger or
smoking prevalence is unaffected WR SULFH HODVWLFLW\ RI í ZKLOH lower-income Ukrainian smokers
by price, that higher prices reduce smoking prevalence among women were generally more responsive to
consumption in smoking households was significantly affected by price price.
LQWHQVLW\HODVWLFLW\RIíDQGWKDW HODVWLFLW\RIí,QDGGLWLRQWKH\ Taal and colleagues (2004) used
demand in lower-income households found that smoking prevalence was monthly household survey data on
is most responsive to price. becoming less inelastic over time, cigarette consumption and household
181
composition from 1992 through 1999 Gardes and Starzec, 2004) using that cigarette consumption was
to construct a measure of average the consumption expenditure data, more than twice as responsive to
adult cigarette consumption for use in a Rotterdam system using a Frisch price among whites (total price
their estimation of the price elasticity scheme, and a rational addiction HODVWLFLW\ RI í WKDQ DPRQJ
of cigarette demand in Estonia. The model (Becker and Murphy, 1988). EODFNVWRWDOSULFHHODVWLFLW\RIí
price was the real tobacco product The alternative approaches produced When limiting the analysis to only
price index reported by the Statistical comparable elasticity estimates, with consuming households, however,
Office of Estonia. Using ordinary a short-run price elasticity of overall they found little impact of price on
least squares methods applied to FLJDUHWWHGHPDQGRIíDQGDORQJ consumption for either group. This
a myopic addiction model, they UXQHODVWLFLW\RIí finding suggests that the impact of
obtained an overall price elasticity of Onder (2002) applied a two- higher prices in reducing smoking
FLJDUHWWHGHPDQGRIí part model to examine whether in South Africa is limited to reducing
Sayginsoy and colleagues (2002) price affects smoking prevalence prevalence. However, the estimates
used data from the 1995 Living among households and intensity they obtained may be biased due to
Standards Measurement Survey for among smoking households, using the likely endogeneity of price in their
Bulgaria, a nationally representative 1994 expenditure survey data models.
household expenditures survey, for Turkey. Using two-stage least Van Walbeek (2002) considered
to estimate the impact of price squares methods to account for the smoking prevalence and the
on cigarette demand in Bulgaria. endogeneity of the price measure percentage of total income spent
To estimate price elasticities, that is derived from the self-report on cigarettes by income quartile for
they used a multistep process. data on consumption expenses two periods, 1990 and 1995. The
First, prices were derived from and consumption quantities, Onder data were obtained from Income and
self-reported expenditures and (2002) obtained an overall price Expenditure Surveys that are used
consumption quantities for smoking HODVWLFLW\ RI í EXW ILQGV OLWWOH to determine the weightings of the
households. Second, for non- impact of price on households’ CPI (consumer price index) basket.
smoking households, the average smoking prevalence (elasticity of Van Walbeek (2002) considered the
of the derived prices from smoking í ,Q JHQHUDO VKH IRXQG WKDW expenditure patterns of approximately
households in the same income overall demand is less inelastic 15 000 urban households. In a
quintile was used. Finally, price was among lower-income households. subsequent study, he expanded
modelled as a function of taxes and the period to include the year 2000.
cigarette characteristics. Using a Africa. Few price elasticity estimates He found that the percentage of
two-stage least squares model, they based on survey data exist for African households that bought cigarettes
estimated an overall price elasticity countries. To date, such studies have decreased from 49% in 1990 to 30%
RI FLJDUHWWH GHPDQG RI í ZLWK been conducted in South Africa (Berg in 2000 (Van Walbeek, 2005). This
price elasticity in absolute value and Kaempfer, 2001;; Van Walbeek, was a period in which the real price
being greatest among lower-income 2002 and 2005) and Egypt (Nassar, of cigarettes increased by more than
households and becoming more 2003). 100%. He found that the percentage
inelastic as household income rises. Berg and Kaempfer (2001) of households in the poorest income
Gardes and Starzec (2004) used the 1991 Living Standards quartile that bought cigarettes
applied a variety of methods Measurement Survey to estimate the decreased from 46% to 22% while
to estimate the price elasticity price elasticity of cigarette demand for among the richest income quartile
of cigarette demand in Poland. black and white households in South that percentage decreased from 43%
They used data from the 1987 Africa. Prices appear to be derived to 34%. He concludes that the poor
through 1990 Polish Consumption from the self-reported expenditures are significantly more price sensitive
Panel database—a database that and consumption quantities, while than the rich in consuming cigarettes.
contains information on household the potential endogeneity of price Nassar (2003) estimated the price
consumption expenditures and price does not appear to be addressed elasticity of tobacco use in Egypt,
indices for various goods including in the estimation. Using censored using data from 1995–96 and 1999–
tobacco products. Specifically, they maximum likelihood and censored 2000 household expenditure surveys.
estimated an Almost Ideal Demand least absolute deviations methods In addition to overall price elasticity
System (cited and explained in Berg and Kaempfer (2001) found estimates, she produced similar
182
estimates for a variety of population geographic differences in prices and Systematic review of the
subgroups based on urban/rural the considerable variability in the scientific literature: Impact
location, income and education, availability and prices of different of price on adult cessation
using average unit values that were tobacco products and brands in
derived from the self-reported price these countries, as well as the A growing number of studies c onsider
data. She estimated an overall price reliance on self-reported information the impact of higher tobacco taxes
HODVWLFLW\ RI í IRU WREDFFR XVH on expenditures or/and prices in and prices on cessation of tobacco
in Egyptian households, finding little many of these studies, researchers use among adults. Given the variety
difference in price elasticity for urban conducting these demand analyses of available data and the extensive
and rural households (elasticities of have faced several challenges in cross-sectional and intertemporal
íDQGíUHVSHFWLYHO\ZKLOH estimating price elasticities. As a variation in taxes and prices, much
generally finding that tobacco use result, they have applied a variety of the work on cessation also uses
is less price inelastic among lower- of different approaches in their data from the USA. A few studies
income, less-educated households modelling, overcoming to the extent use data from other high-income
than higher-income, more educated possible the potential measurement countries, such as Canada, the
households. errors and endogeneity biases they United Kingdom, France and
faced. Spain. Only one study to date uses
The Americas. For low- or middle- Despite these challenges and data from a low- or middle-income
income counties in the Americas, only the problems that are likely to remain country, Viet Nam. These studies
one study on the effects of price on in many of the studies, this growing are summarized in Table 5.2 and
adult tobacco use based on survey body of research consistently are described below (see Chapter 6
data was identified. Jiménez-Ruiz and demonstrates that higher taxes for studies that focus on cessation
colleagues (2008) examined cigarette and prices lead to reductions in among young people).
demand in Mexico, using data from tobacco use. There are considerable Most of these studies use one
multiple waves of the National differences across countries in terms of three basic designs: analysis
Household Expenditure Survey of the estimated price elasticities. of cross-sectional data with
conducted from 1994 through 2005 Some studies indicate that price retrospective information on when
and applying two-part models for has limited impact on prevalence an individual quit;; analysis of cross-
demand. To account for the potential of tobacco use, while significantly sectional data with information on
endogeneity of price, they used reducing tobacco consumption recent quit attempts or interests in
average prices derived from self- among users. Others find price to quitting;; or analysis of longitudinal
reported information on consumption have large effects on prevalence but data with information on tobacco use
expenditures and consumption more limited impact on amount used, behaviours over time. Two potential
quantities, stratified by household and still others discover that both problems arise in the analysis of
location (state and rural/non-rural) prevalence and intensity of tobacco retrospective data on cessation:
and income quintile. Jiménez-Ruiz use are reduced by higher tobacco first, there may be measurement
and colleagues (2008) found little prices. While the point estimates vary error in the cessation measure due
impact of price on households’ considerably, they generally imply that to imperfect recall of the age at
smoking prevalence (elasticity of adult demand for tobacco products which an individual quits;; second,
íZLWKDPXFKJUHDWHULPSDFWRQ in low- and middle-income countries there may be measurement error
consumption in smoking households is at least as sensitive to price, and in prices that were matched to the
HODVWLFLW\RIí often more sensitive to price, than it survey data, given the problems with
is in high-income countries. Studies the timing of quitting as well as the
Summary that produce price elasticity estimates possibility that a survey respondent
that are less elastic tend to come has moved since quitting. In
Over the past decade, at least 28 from countries where cigarettes are general, these problems will be
studies have used individual or very inexpensive or affordability has less significant than when using
household survey data to assess increased significantly over time retrospective data on initiation, as
the impact of prices on adult (most notably recent studies for China described more fully in the next
tobacco use in low- and middle- and the Russian Federation). chapter, given that quitting will be
income countries. Given the limited much more recent than initiation.
183
184
Table 5.2. Summary of studies on the effect of cigarette price on smoking cessation
Methods (time period, Sub-populations

Publication (author, yr, Endpoint – (definition
study design and Model Main results (elasticity) (which subpopulations Comment
country) of price)
sample size) are considered)
USA
Douglas (1998) Longitudinal analysis An ordered aprobit split- Weighted average Price elasticity Price elasticity “Quitting hazard”
USA (retrospective) from sample duration model real prices per pack -1.05 1.31 (for future price) increases substantially
reconstructed tobacco Taken into account (including taxes) for each -0.98 (after being 1.07 (after being with future cigarette
histories from the 1987 lagged duration state from 1954 to 1991 controlled for regulation) controlled for regulation) price;; a result that is
cross-sectional survey dependence and time- robust across alternative
(NHIS) varying covariates specifications
N=22 080 With and without a state-
Analysis sample: level regulation

\UV
Franz (2008) Cross-sectional Two models were Dependent variable: Overall By age (in years)
USA (8 surveys from BRFSS considered: i) A simple quitting decision +0.375 18–29
(1993–2000) OLS +0.493
Analysis sample: ii) A two-part model: 1) 30–39
1 \UV OLS;; 2)OLS +0.424
Adjusted for sex, health 40–64
status, age, race, +0.398
education, marital status,
income, region and year +0.202
effects
Price: state-level
average real price of a
pack of cigarettes
Ross et al. (2011) Longitudinal data (three (1) OLS for change in Cessation stage of an Coefficient estimates Cigarette prices
USA and Canada waves of International cessation stage over individual (collapsed into from the OLS models increased the likelihood
Tobacco Control Policy time three stages) Overall: 0.0009 to of actual quitting
Evaluation Survey (ITC) (2) Generalized Price: 0.0015 Only the coefficient
2002, 2003 and 2004) Estimating Equations (i) Change in self Coefficient estimates estimates of the external
Cohorts were based (GEE) models for reported price from the GEE models price obtained from
on the national smoking cessation (ii) Change in external Overall: 0.0007 and the GEE models were
representative samples Different OLS and GEE price 0.0044 statistically significant
of adult smokers models were considered (iii) Change in tax The results lends
annually interviewed in according to the number support to that higher
the three waves (N=4352 of covariates included cigarette prices can
at wave 1) in the models. The first, be used to increase
Study sample: simplest model only cessation and to
N=5304–5973 included demographic motivate smokers to quit
depending on models and socioeconomic
Longitudinal sample: ca. variables. The second
N=2000 model further adjusted
for knowledge of the
health risk of smoking.
The third model
furthermore accounted
for nicotine addiction
country) of price)
Canada
Hamilton et al. (1997) Longitudinal No multivariate No computation of Higher cessation rates Quit rate by different
Canada (retrospective) regression analysis elasticities (10.7% vs. 10.3%) time of interview in
Survey on Smoking in Simple descriptive in provinces where provinces:
Canada (1994) Follow-up analysis taxes were not cut vs. With tax cut:
one year provinces without taxes +2.5 to +10.3
11 119 respondents cut Without tax cut:
\UV +5.7 to +10.7
Telephone survey Changes in taxes across
years not analyzed
France
Peretti-Watel (2004) Life-course perspective Time hazard model Odds ratio (OR) of For subsample of Full sample Risk estimates were
France of retrospective data of smoking cessation: cessation given an individuals aged Cessation by age statistically significant
extracted from the 1999 logistic regression increase in price by 1% 21–50, by sex range for both subsamples
cross-sectional survey Adjusted for time- Men Age 20 or younger S
French Health Barometer constant variables OR=1.007 OR=1.005
(FHB) (A telephone (gender, education Women Age 21–30:
survey representative of and age of smoking OR=1.009 OR=1.017
the French population initiation), and for time- Age older than 30
12–75 yrs) varying variables (age, OR=1.011
n=13 685 age2, parenthood, and
price)
Cessation was assessed
separately for men and
women aged 21–50
Spain
López Nicolás (2002) Longitudinal analysis A log-logistic split Real price of 20 Elasticity of quitting Different price series
Spain (retrospective) from population model with cigarettes with respect to the for black cigarettes and
reconstructed tobacco the Weibull distribution price of black tobacco blond cigarettes, and a
histories from cross- Adjusted for education, By sex weighted average of both
sectional surveys dummies for tobacco Men were used
National Health Interview policies, time trend and -1.32 Different model
Surveys (ENSE) 1993, birth cohort Women specifications were also
1995 and 1995 -1.50 tested
Face-to-face interview Elasticity of starting
$QDO\VLVVDPSOH Men
yrs): +0.07
Starting: Women
Men: N=7092 +0.08
Women: N=9913
Quitting:
Men: N=2305
Women: N=1817
185
186
Table 5.2. Summary of studies on the effect of cigarette price on smoking cessation

country) of price)
United Kingdom
Forster & Jones (2001) Cross-sectional survey Split-population log- Tobacco tax: proxy Full sample By sex Tax elasticities of
United Kingdom (1984) logistic probit duration for the real price of -0.41 to -0.69 Men starting smoking
British Health and model for smoking cigarettes -0.41 to -0.63 By sex
Lifestyle Survey (HALS) initiation Tax elasticity of quitting Women Men
Sample of adults Generalized gamma smoking -0.41 to -0.69 +0.16
\UVUHSUHVHQWDWLYH and Weibull models for Women
of the population living in smoking quitting +0.08
households in England, Adjusted for social class,

Scotland and Wales education, ethnicity,
N=9003 parental smoking, and a
Retrospective smoking time trend
data by using individuals’
self-reporting information
Analysis sample: N=9003
Viet Nam
Laxminarayan & Viet Nam Living Three multinomial logit Given a change in price Cigarette smokers
Deolalikar (2004) Standards Survey, regressions for change of of cigarettes, price appear to be more price-
Viet Nam waves 1992/93 and tobacco use status from elasticities of decisions sensitive than rustic
1997/98 1993 to 1998 for three to quit cigarette smoking: tobacco users
PDOHV\HDUV samples: -0.242
Two forms of tobacco: 1) all smokers in 1993 Given a change in price
cigarettes and rustic 2) all cigarette smokers of rustic tobacco, price
tobacco in 1993 elasticities of decisions
3) all rustic tobacco users to quit rustic tobacco:
Explanatory variables: +0.735
real per capita
expenditure in 1993, age
in 1993, education, price
of cigarettes in 1993,
price of rustic tobacco in
1993;; changes between
1993 and 1998 in real
per capita expenditure, in
price of cigarettes and in
price of rustic tobacco
Price data: collected from
separate, commune-
level surveys. Tobacco
prices for year 1997/98
were imputed from
household rustic tobacco
expenditures and price
of tobacco sold obtained
from households
producing tobacco
BRFSS, behavioral risk factor surveillance system;; NHIS, national health interview survey;; OLS, ordinary least squares
Similarly, the analysis of cross- RI VPRNLQJ EHIRUH TXLWWLQJ RI í who purchased cigarettes from a
sectional data with information on IRU PHQ DQG í IRU ZRPHQ LQ low or untaxed source (e.g. from
recent quit attempts or interest in their baseline models. A variety of duty-free shops, online vendors and
quitting will be useful in understanding sensitivity analyses produce similar various other sources) were less
how tax and price increases motivate HODVWLFLW\ UDQJHV í WR í IRU likely to have made a quit attempt or
cessation, but will be of limited utility PHQDQGíWRíIRUZRPHQ to quit successfully between waves
in understanding their impact on Similar methods were employed by than those who did not purchase
successful, long-term cessation. López Nicolás (2002) for Spain and from these sources. This finding
Multiple waves of longitudinal data by Peretti-Watel (2004) for France. suggests that the availability of
are most useful for addressing this, Both studies similarly concluded that opportunities for tax avoidance and
but few such data are available to higher cigarette prices increased the evasion reduces cessation.
researchers. probability of cessation. More recently, Ross and
Given the available data, the most Franz (2008) applied a related colleagues (2011) used the first
commonly employed design has been approach to look at the impact of three waves of the US and Canadian
the use of retrospective information cigarette prices on quitting in the past ITC surveys to further explore the
on smoking behaviour and cessation year, using cross-sectional data for role of price and related factors on
constructed from cross-sectional the USA from the 1993 through 2000 cessation outcomes. Using a five-
survey data. Douglas (1998) was Behavioural Risk Factor Surveillance level “stages of change” measure of
the first to use this approach, System surveys. He found that quit intentions (pre-contemplation,
constructing smoking histories from cigarette prices were positively contemplation, preparation, action
the retrospective data collected in the associated with this measure of and maintenance), they found that
1987 US National Health Interview cessation, with an estimated elasticity smokers living in areas with higher
Survey, with historical cigarette of cessation of 0.375 (the higher the taxes and higher prices were
prices matched to the survey on price, the higher the probability of significantly more interested in
the basis of each respondent’s having quit the previous year). In quitting. Similarly, they found some
current state of residence. Using addition, the elasticity estimate of evidence that increases in prices
ordered probit, split-sample duration cessation fell with increasing age, resulted in increased motivation to
methods, Douglas (1998) modelled although it remained statistically quit and higher prices increased the
cessation using the rational addiction significant (the elasticity estimate for likelihood of successful quitting. In
framework. He found that higher the youngest group was more than contrast to Hyland and colleagues
prices significantly increased the twice that of the oldest group;; Franz, (2006), they found that the availability
probability of cessation, with the 2008). of cheaper cigarettes did not deter
duration of smoking approximately Two studies have employed cessation, although smokers would
unitarily elastic (e.g. a 10% increase longitudinal data taken from the respond more aggressively in their
in prices would reduce the duration International Tobacco Control (ITC) cessation efforts if lower priced
of smoking by 10%). In addition, he Policy Evaluation Project’s nationally cigarettes were not as readily
found that the probability of quitting representative surveys of adult available.
rose with the duration of smoking, smokers to examine the impact of In the only cessation study
consistent with the predictions of the tax and price on cessation related conducted in a low- or middle-
rational addiction model. outcomes. Hyland and colleagues income country, Laxminarayan
Forster and Jones (2001) applied (2006) used the first (late 2002) and Deolalikar (2004) used data
a similar approach to estimating the and second (mid-2003) waves of for the subset of households
impact of cigarette tax on smoking the US, United Kingdom, Canadian, participating in both the 1992–93
cessation in the United Kingdom, and Australian ITC surveys to look and 1997–98 waves of the Viet Nam
using retrospective information at the effects of cigarette purchase Living Standards Survey3. They
from the 1984 British Health and behaviours in wave one on the considered two forms of tobacco—
Lifestyle Survey. They studied men likelihood of making a quit attempt cigarettes and rustic tobacco—
and women separately, obtaining tax and the likelihood of having quit by and used price data collected from
elasticities for the number of years wave two. They found that smokers separate, commune-level surveys,
3
Laxminarayan and Deolalikar (2004) also assess initiation with these data, as discussed in Chapter 6.
187
with rustic tobacco prices in the Systematic review of the scientific models for cigarettes and smokeless
latter wave imputed because only literature: Impact of relative prices tobacco use, they included state-
cigarette prices were collected at on substitution among tobacco level taxes on both cigarettes and
that wave. They found little evidence product by adult tobacco users smokeless products to examine
that higher cigarette prices lead to cross-price effects on tobacco use.
cessation of cigarette smoking or Even fewer studies have used While the point estimates vary from
that higher prices of rustic tobacco survey data to examine the effects of study to study, they find consistent
increase the probability of quitting changes in the price of one tobacco evidence that higher cigarette taxes
among rustic tobacco users. Their product relative to other tobacco encourage some substitution to
findings indicated the occurrence of products on substitution between smokeless tobacco products, with
product substitution in response to these products by adult tobacco positive cross-tax elasticities for
increases in price;; in this particular users. This is likely due to the very low smokeless tobacco use with respect
instance, smokers switching from prevalence of non-cigarette tobacco to cigarette taxes. In contrast, they
manufactured cigarettes to the use product use in many countries, to find little evidence in the opposite
of rustic tobacco. However, this the lack of detailed data on use of direction: estimates for smokeless
finding may be the result of several multiple tobacco products in the tobacco taxes were generally
factors. The sample sizes used in same survey, and to the correlations insignificant in models of cigarette
their models are relatively small. that exist between taxes and prices smoking.
Correlations in price changes over for various tobacco products which Delnevo and colleagues (2004)
time may reduce the precision of make it difficult to empirically sort find similar evidence of substitution
their estimates. The use of imputed out the differential impact of each on between cigarettes and cigars based
rustic tobacco prices for one of the use. Most of the studies that have on data from the 2001 and 2002 New
two waves included in the analysis examined the impact of changes in Jersey Adult Tobacco Use Surveys.
can also bring about measurement relative prices on substitution have Their analysis takes advantage of a
errors to the estimates. been discussed above where findings significant increase in the New Jersey
regarding the effects of own-price cigarette excise tax (from $0.80 per
Summary on use were reviewed. This section pack to $1.50 per pack) between
briefly reviews the limited existing the two waves of the survey, while
Studies on the impact of tobacco evidence on cross-price effects, with the tax on other tobacco products in
product taxes and prices on Table 5.3 providing a summary of the New Jersey was unchanged. They
cessation are relatively scarce. The relevant studies. found that the prevalence of cigar
majority of these rely on retrospective Most of the evidence on cross- use increased significantly from 2001
data on smoking histories collected price effects also comes from the to 2002 among current cigarette
in cross-sectional surveys that USA, with Ohsfeldt and colleagues smokers, and even more so among
are subject to various sources of responsible for most of those studies those who had recently quit smoking
measurement errors. Nevertheless, (Ohsfeldt and Boyle, 1994;; Ohsfeldt cigarettes. Delnevo and colleagues
the findings from the small but et al., 1997;; Ohsfeldt et al.;; 1999). (2004) concluded that the changes
growing body of research from high- In their various papers, Ohsfeldt in cigar smoking were the result of
income countries are consistent and colleagues used data from some cigarette smokers who had
with the findings described above different waves of the Tobacco Use switched to cigar use following the
for the impact of price on smoking Supplement to the Current Population significant increase in the relative
prevalence. Specifically, these Survey augmented with data on price of cigarettes in New Jersey.
studies consistently find that higher state-level cigarette and smokeless To date, only three survey-based
cigarette prices are associated with tobacco taxes. As described above, studies of cross-price effects for
increased motivation to quit and they consistently obtained negative adult tobacco use exist for low- and
with successful cessation. The only own-tax effects, with higher cigarette middle-income countries. In addition
study from a low- or middle-income taxes associated with reduced to estimating the own-price effects
country, using Vietnamese data, cigarette use and higher smokeless for cigarettes, bidis and leaf tobacco,
does not reach similar conclusions, tobacco product taxes associated John’s (2008) analysis of tobacco
which may be accounted for by data with reduced use of snuff and demand in India estimated cross-
limitations. chewing tobacco. In their demand price effects.
188
Table 5.3. Summary of studies on the effect of cigarette price on substitution among tobacco products

Publication (author, yr,
study design and Model Endpoint Main results Sub-populations Comment
country)
sample size)
USA
Ohsfeldt & Boyle (1994) Cross-sectional Multivariate regression Prevalence of use of Cross-(tax) elasticities Ranges of elasticity
USA CPS (1985) (snuff use, chewing smokeless tobacco of demand for estimates were obtained
n=100 000 in file tobacco use, and any Price of smokeless smokeless with according to two models
Analysis sample: only smokeless tobacco use) tobacco products: respect to the cigarette See also Table 5.1 for
PHQ\UV1QRW Adjusted for per capita border-adjusted state tax price elasticities of
reported) income, education, smokeless tobacco Snuff use demand for smokeless
state population average excise tax +0.3 and +0.62 tobacco use
characteristics Chewing tobacco use
(residence, race, +0.49 and +0.39
religious denomination, Any smokeless
and divorce rate), men tobacco use
aged 16–17, and existing +0.44 and +0.41
state tobacco regulation
Ohsfeldt et al. (1997) Cross-sectional Multivariate Logit Prevalence Cross-tax-elasticities of Cross-tax-elasticities See also Table 5.1 for
USA CPS (1985) regression (for (i) Price of smokeless demand (prevalence) for of demand price elasticities of
n=100 000 in file cigarette use, (ii) snuff tobacco products: smokeless tobacco with (prevalence) for demand for smokeless
Analysis sample: only use, (iii) chewing border-adjusted state respect to the cigarette smokeless tobacco tobacco use
PHQ\UV1QRW tobacco use, and (iv) any average smokeless tax with respect to the
reported) smokeless tobacco use) tobacco excise tax cigarette tax
Adjusted for total real By age (in years)
family income, age, Snuff use
marital status, race, >15: +0.13
ethnicity, employment 16–24: +0.71
status, education,
and occupation, Chewing tobacco use
metropolitan, religion, >15: +0.09
and tobacco control 16–24: +0.19
policy
Any smokeless tobacco
use
>15: +0.10
16–24: +0.23

Ohsfeldt et al. (1999) Cross sectional (pooled Multivariate regression Prevalence Overall cross-tax Cross-tax elasticity of Tobacco models
USA surveys) Adjusted for price (tax) of Price: real federal excise elasticity of demand demand for were estimated with
CPS (1992–1993) snuff, index of smoking tax rate for Cigarettes (given a the cigarette tax and
Analysis sample: 165 regulation, family Cigarette use 1-percent change in smoking regulation
653 males either white or income, age, occupation +0.001 (given a snuff tax rate) variables treated
EODFN\UV and education, ethnicity 1-percent change in By age (in years) as exogenous and
Nationally representative (black/white), marital snuff tax rate) 16–24: +0.002 endogenous variables
samples status, per capital Snuff use 25–44: +0.001 Higher cigarette tax
income, poverty, +0.98 (given a 1-percent rates are associated with
unemployment rate and change in cigarette tax Snuff (given a greater snuff use, but
religion rate) 1-percent change in higher snuff tax rates
189
190

country)
sample size)
Ohsfeldt et al. (1999) cigarette tax rate) are not associated with
(contd) By age (in years) greater cigarette use
USA 16–24: +1.15
25–44: +0.04

Delnevo et al. (2004) Cross-sectional A logistic regression for Comparison between the The odds of being an The adjusted odd Substitution of
USA New Jersey Adult (i) ever cigar use and (ii) odds of being an ever ever and current cigar ratio for current cigar cigarettes by cigars
Tobacco Survey (NJATS) current cigar use and current cigar user smoker substantially use of other tobacco after increases in excise
(2001–2002) Adjusted for sex, race, before (2001) and after increased among recent products tax for cigarettes were
RDD age, education, and (2002) a new cigarette cigarette quitters after Men controlled for (and
Analysis sample: cigarette smoking status excise tax the new tax were taken 2001 reducing that for cigars)
1 \UV No price elasticities into account +13.7
2001 data: N=3090 nor association of 2002
2002 data: N=4004 consumption (OR) with +6.2
price are reported By smoking history
Former smoker
2001: +1.40
2002: +2.68
Recent quitter
2001: +0.52
2002: +4.73
Current smoker
2001: +2.82
2002: +4.51
China
Tsai et al. (2005) Two-year follow-up Logistic regressions (for Change in smoking OR for reduced smoking Overall, 17.4% switched
China (Taiwan) cohort (face-to-face behavioural change (i) behaviour before (year with respect to increased brand, 18.8% reduced
survey) reduction of smoking, 2001) and after the retailed cigarette price smoking, 8.4% both
Analysis sample: N=501 and (ii) brand switching). introduction of a new by 1 NT$=1.03
PDOHVPRNHUV\UV Odds ratios were cigarette tax scheme OR for switching brand
obtained (year 2002) with respect to increased
Adjusted for retailed cigarette price
demographic factors, by 1 NT$=1.07
individual income,
smoking behaviour,
living area, expense
on smoking, addiction
level, advertisement, and
favourite cigarette brand
country)
sample size)
India
John (2008) Cross-sectional survey OLS regression Substitution among Cross-elasticities A demand system model
India (1999–2000) Price of each tobacco tobacco products (consumption) Spatial variation in prices
N=120 309 households product: average unit Rural of tobacco products
Household with zero values (self-reported Demand for cigarettes were used to estimate
consumption of tobacco expenditure divided by given a 1% change in the own and cross-price
(37% from rural and quantity) price of elasticities.
60% from urban) were Bidis: -0.24 When symmetry
excluded from the Leaf tobacco: +0.010 restrictions were
analyses Demand for bidis given a imposed, price elasticity
Unit of analysis: 1% change in the price of estimates did not
household Cigarettes: -0.117 considerably change
Leaf tobacco: -0.010 Authors concluded that
Demand for leaf tobacco there are no substitution
given a 1% change in the effects due to price
price of change
Cigarettes: +0.15 Cross-price elasticities
Bidis: -0.067 reported here were
Urban symmetric-constrained
Demand for cigarettes estimates
given a 1% change in the See also Table 5.1
price of
Bidis: -0.122
Leaf tobacco: +0.002
Demand for bidis given a
1% change in the price of
Cigarettes: +0.011
Leaf tobacco: +0.011
Demand for leaf tobacco
given a 1% change in the
price of
Cigarettes: -0.221
Bidis: +0.252
South Africa
Van Walbeek (2005) Income and Expenditure Two research Percentages of urban The poor are
South Africa surveys (IES), 1990, questions: (1) the households spending significantly more price
1995 and 2000 relative importance money on tobacco sensitive than the rich in
Analysis unit: household of tobacco in South products in 1990, 1995 consuming cigarettes.
Analysis sample: African households’ and 2000 by income The decrease in
1990: N=14 332 (urban expenditure patterns;; quintile (lowest to aggregate
households) and (2) changes in the highest) cigarette consumption
1995: N=16 903 (urban regressivity of cigarette Q1: 46;; 42;; 22 was driven largely by
and rural households) taxes between 1990 and Q2: 54;; 46;; 31 poorer households
2000: N=26 263 (urban 2000 Q3: 51;; 45;; 34 quitting (or not starting)
and rural households) Q4: 43;; 44;; 34 cigarette smoking
Between 1990 and 2000
the percentage of
191
192

country)
sample size)
Van Walbeek (2005) households consuming
(contd) tobacco decreased by
South Africa 14, 19, 16, and 8
percentage points for
income quintiles Q1 to
Q4.
Viet Nam
Laxminarayan & Viet Nam Living Three multinomial logit Given a change in Although higher cigarette
Deolalikar (2004) Standards Survey, regressions for change price of cigarettes, prices discourage
Viet Nam waves 1992/93 and of tobacco use status price elasticities of initiation of cigarette
1997/98 from 1993 to 1998 for decisions to smoking, they may
PDOHV\HDUV three samples: (1) initiate into encourage the use of
Two forms of tobacco: 1) all smokers in 1993 cigarettes: -1.175 rustic tobacco
cigarettes and rustic 2) all cigarette smokers rustic tobacco: +1.375
tobacco in 1993 (2) switching from
3) all rustic tobacco cigarettes to rustic
users tobacco +1.395
Explanatory variables: rustic tobacco to
real per capita cigarettes
expenditure in 1993, age +0.004
in 1993, education, price (3) quitting
of cigarettes in 1993, cigarettes
price of rustic tobacco in -0.242
1993;; changes between rustic tobacco
1993 and 1998 in real +0.243
per capita expenditure, Given a change in
in price of cigarettes and price of rustic tobacco,
in price of rustic tobacco price elasticities of
Price data: collected decisions to
from separate, (1) initiate into
commune- level surveys. cigarettes: 0.477
Tobacco prices for year rustic tobacco: 1.558
1997/98 were imputed (2) switching from
from household rustic cigarettes to rustic
tobacco expenditures tobacco -3.240
and price of tobacco rustic tobacco to
sold obtained from cigarettes
households producing -1.601
tobacco (3) quitting
cigarettes: -1.411
rustic tobacco: +0.735
CPS, current population survey;; OLS, ordinary least squares;; RDD, random digit dialing
In contrast to the evidence from Summary between 800 and 27 000 subjects that
high-income countries, John (2008) in most cases were representative of
found no evidence that changes The limited survey-based research on the corresponding target (state or
in the relative prices of tobacco cross-price effects in adult tobacco national) populations.
products would result in substitution use suggests that, at least in high- The key messages are that a
from higher-priced to lower-priced income countries, tobacco products substantial proportion (i.e. generally
tobacco products. Instead, most of are generally substitutes for one between a third and a half) of
his cross-price elasticity estimates another. An increase in the price the population would support tax
were negative, although few of of one product relative to the price increases, and that such a support
them were statistically significant, of another product will lead some becomes appreciably greater
suggesting complementarity among users of products whose relative (i.e. 60–80% or more) whenever
these tobacco products in India. prices increase to switch to products such tax increases are devoted to
In contrast, Laxminarayan and whose relative prices fall. The even measures for tobacco control. A
Deolalikar (2004) found some scarcer research on this issue from study from New York City (Shelley
evidence of substitution between low- and middle-income countries et al., 2007), based on focus
cigarettes and rustic tobacco use in suggests that cultural differences may groups and qualitative information
their analysis of adult tobacco use be important in reducing cross-price on economically disadvantaged
in Viet Nam. Specifically, they found effects. population’s smoking and purchasing
that higher cigarette prices resulted responses to increases on tobacco
in some substitution from cigarette Systematic review of the scientific taxes, reported that a substantial tax
use to rustic tobacco use, but found literature: Attitudes, perceptions rise (e.g. US$5) increased intention
little impact of higher rustic tobacco and behaviours towards increases to quit but also increased smuggling.
price on cigarette use. Similarly, of cigarette prices In several studies, the support
in his analysis of South African for increased taxation was greater
tobacco prices and tobacco use, Van Tables 5.4 and 5.5 summarize in nonsmokers and in smokers
Walbeek (2005) found evidence of findings on attitudes and perceptions who were more educated, elderly
substitution in response to relative towards increases of cigarette prices or less young. However, a study
price changes, at least in some and willingness of smokers to quit in that concerned African Americans’
populations. Specifically, he found response to increments of cigarette attitudes did not consider a rise in
strong evidence that the poor were prices, respectively. In these tables, taxation unfair for African Americans
switching to pipe and other tobacco papers are sorted by country (USA, (King et al., 2003). At each level of
(presumably to make roll-your-own other high-income countries, and education, responders agreeing
(RYO) cigarettes) much more often low- and medium-income countries) with a tax increase represented the
than the rich as the price of cigarettes and year of publication. most numerous group. A study from
went up relative to these products. Germany (Hanewinkel & Isensee,
In 1990, the poorest quarter of the Attitudes and perceptions towards 2008) also showed growing support
population spent about 5% of their increasing of cigarette prices for increasing taxation over the most
tobacco purchases on pipe and other recent calendar periods (35% in
tobacco;; in 2000, this had increased Table 5.4 gives a summary tabulation 2002 to 42% in 2005).
to 18%. Among the second- of studies providing evidence on the
poorest quarter of the population, attitudes and perceptions of adults Willingness of smokers
there was also an increase in the towards increasing cigarette prices. to quit according to increases
relative consumption share of pipe Studies were available from the USA in cigarette prices
and other tobacco (from 2.4% to (Torabi et al., 1994;; King et al., 2003;;
7.1% of total tobacco expenditure). Hamilton et al., 2005;; Shelley et al., Table 5.5 gives summary findings
However, among the richer half of the 2007), Taiwan, China (Tsai et al., from studies on the willingness of
population, the consumption share 2003), Italy (Gallus et al., 2005;; Gallus smokers to quit according to increases
of pipe and other tobacco remained et al., 2006), Germany (Hanewinkel in cigarette prices and taxation.
unchanged over this period. & Isensee, 2008), and New Zealand
(Wilson et al., 2010). These studies
were based on samples ranging
193
194
Table 5.4. Summary of studies providing data on the attitudes and perceptions of adults towards increasing cigarette prices
Reference, year, country Methods Main results Comments
USA
Torabi et al. (1994) Random sample of 800 adults About 85% in favour of tax rise for health Representative state sample
Indiana, USA education and tobacco research
King et al. (2003) 1000 African American adults from 10 47% of the sample supported increase in Rise in taxation not considered unfair for
USA districts taxation. Positive relation with education African Americans
Hamilton et al. (2005) 14 000 adults from 351 Massachusetts towns Over 80% of support for tax increase if used Greater support by non smokers, more
Massachusetts, USA for tobacco control only, 74% for health or educated, less young individuals and women
tobacco control, 31% for any government

purpose
Shelley et al. (2007) 104 subjects in 14 focus groups The US$5 price increase in New York Qualitative data
New York City, USA City increased interest in quitting, but also
smuggling
Germany
Hanewinkel & Isensee (2008) Ten representative samples of the general Greater support to tax increase over time, The amount of price increase did not affect
Germany SRSXODWLRQSHUVRQVDJHG from 35% in 2002 to 42% in 2005. General this support
support among elderly, more educated
subjects
Italy
Gallus et al. (2005) ,WDOLDQVDJHG 54% of the sample confident that an increase Representative sample of the Italian
Italy of price is effective. population
21% of smokers stated there they would
reduce, and 10% stop smoking following an
1€ increase in price
Gallus et al. (2006) ,WDOLDQVDJHGLQWHUYLHZHGLQ 36% of ever smokers reported that a tax Representative sample of the Italian
Italy increase would have a high (10.5%) or population
intermediate (25.4%) impact on cigarette
consumption in the young
Gallus et al. (2008) $GXOWV\HDUVLQWHUYLHZHGLQ± 25% of 1456 smokers strongly in favour, and Greater support in middle-aged moderate
Italy partly overlapping with Gallus et al.2006 37% moderately favour of an increased of smokers
10 € cents devoted to supporting another
association 39% of smokers indicate that
a minimum price of 5 € per pack, would
decrease smoking
New Zealand
Wilson et al. (2010) 1376 smokers 59% would support a tax increase if the Greater support for smokers with intention
New Zealand revenue if used to promote healthy lifestyle to quit
and quitting

China
Tsai et al. (2003) National sample. Face-to-face interviews of Lower support towards cigarette tax increase Survey conducted before the introduction of
China (Taiwan) 3279 adults aged 18 to 64 years in 2000 among current (OR=0.34) and former a national tax
smokers, and among lower social classes
South Africa
Martin et al. (1992;; 1993) Household survey representative of the South 56% of adults believed that taxes on tobacco
African population products should be increased. Supporters for
Sample size: 2006 adults tax increases were more frequently whites
(60%) and Asians (62%), females (58%),
VXEMHFWVDJHG\HDUVDQGPRUH
educated subjects (63%). Greater support
was evident in non smokers (59%) and
particularly ex-smokers (70%) compared with
current smokers (45%)
195
196
Table 5.5. Summary of studies providing data on the willingness of smokers to quit according to increases of cigarette prices
USA
Biener et al. (1998) Ad hoc survey (1993-1994) Changes in smoking behaviour in front of an Survey after increase of 25¢ and prior to a
1783 adult smokers hypothetical increase in price: statewide tobacco control programme
216 teenager smokers Low income associated to cut cost (switch to Estimates adjusted for age, sex, income, and
Telephone survey cheaper tax or reduce number of cigarettes;; amount smoked
Representative sample of Massachusetts OR=3.30;; 95% CI 1.79–6.36) and consider
quitting (OR=1.98;; 95% CI 1.17–3.34) vs. no
response
Heavy smoking associated to cut cost

(OR=2.08;; 95% CI 1.10–3.93) vs. no response
Teenagers:
Low income associated to cut cost (OR=7.64;;
95% CI 1.37–42.56) vs. no response and
cut cost (OR=13.26;; 95% CI 1.93–91.57) vs.
consider quitting
Republic of Korea
Chung et al. (2007) PHQDJHG Overall price elasticity -0.66 Key impact on amount of cigarette smoked
Telephone interview (-0.64), but very limited on smoking cessation
(-0.02)
Chung et al. (2008) 702 subjects Mean of the willingness to quit for an increase Greater willingness with larger increases
in price of about 40%
Low- and middle- income countries
China
Lee (2008) 483 questionnaires from a population survey Elasticity -0.29 for a 44% increase in price Greater elasticity for women, low income
(Taiwan) smokers, moderately addicted and smokers
who purchase low-price cigarettes
Studies were available from the USA Overall summary from several low- and middle-income
(Biener et al., 1998), the Republic countries generally confirm that
of Korea (Chung et al., 2007, 2008) A large and growing number of various aspects of adult tobacco use
and Taiwan, China (Lee, 2008) and studies have used individual-level are responsive to price, with higher
based on samples from 500 to 3000 or household-level survey data to prices reducing both prevalence
subjects. assess the impact of tobacco product and intensity of use. Several studies
A study conducted in the USA taxes and prices on use of tobacco from high-income countries have
on 1783 adult smokers analysed products among adults. Studies have examined adult smoking cessation,
changes in smoking behaviour in used survey data to examine the generally finding that higher taxes
front of a hypothetical increase differential impact of tax and price and higher prices reduce the duration
of price. Lower-income smokers on tobacco use among population of smoking, raise interest in quitting,
were significantly more likely to subgroups defined by gender, age, boost quit attempts and increase the
cut smoking costs (by switching to socioeconomic status and/or other number of smokers who successfully
cheaper brands or smoking less characteristics, as well as to assess quit smoking. Finally, a few US studies
cigarettes) or seriously consider the separate effects of price on found similar effects of tax and price
quitting than to not adjusting their different aspects of tobacco use, such on the use of other tobacco products,
smoking behaviour in the face of as prevalence, frequency, intensity, such as smokeless tobacco and
a tax increase (“no response”) as initiation, uptake and cessation. The cigars, and produced some evidence
compared to higher-income smokers relatively large literature from the of substitution among tobacco
(Biener et al., 1998). Heavy smoking USA and other high-income countries products in response to changes in
was significantly and positively shows that adult smoking prevalence the relative prices of these products.
associated with the probability of and intensity are negatively related to In contrast, no clear patterns emerge
cutting costs (OR = 2.08;; 95% CI: cigarette taxes and prices, with most from the small number of studies
1.10–3.93) compared to no response. total elasticity estimates falling in the from countries other than the US that
UDQJH IURP í WR í :KLOH WKH consider substitution among tobacco
quality of data and methods varies products in response to changes in
more, estimates obtained in studies the relative prices of these products.
197
References
Adioetomo SM, Djutaharta T Hendratno Chaloupka FJ (1992). Clean indoor air laws, Evans W, Ringel J, Stech D (1999). Tobacco
(2005). Cigarette consumption, taxation and addiction and cigarette smoking. Appl Econ, taxes and public policy to discourage smoking.
household income: Indonesia case study. 24:193–205 doi:10.1080/00036849200000118. In: Poterba JM, ed., Tax Policy and the
HNP Discussion Paper Series, Economics of Economy, Vol.13. Cambridge, MA, MIT Press,
Tobacco Control Paper No. 26. Washington Chaloupka FJ, Warner KE (2000). The 1–56.
DC, The World Bank. Economics of Smoking. In: Culyer AJ,
Newhouse JP, eds., Handbook of Health Evans WN, Farrelly MC (1998). The
Aristei D, Pieroni L (2009). Addiction, social Economics Volume 1B. Handbooks in compensating behavior of smokers: taxes,
interactions and gender difference sin Economics 17. Oxford, UK, Elsevier, Chapter tar, and nicotine. Rand J Econ, 29:578–595.
cigarette consumption. Empirica, 36:245–272 29;; 1539–1627. doi:10.2307/2556105 PMID:11794360
doi:10.1007/s10663-008-9083-2.
Chang HH, Chiang TL (2009). Depressive Farrelly MC, Bray JW, Pechacek TF, Woollery
Arunatilake N, Opatha M (2003). The economics symptoms, smoking, and cigarette price T (2001). Responses by adults to increases
of tobacco in Sri Lanka. HNP Discussion Paper elasticity: results from a population-based in cigarette prices by sociodemographic
Series, Economics of Tobacco Control Paper survey in Taiwan. Int J Public Health, characteristics. Southern Economic Review,
No.12. Washington DC, The World Bank. 54:421–426.doi:10.1007/s00038-009-0080-4 68:156–165 doi:10.2307/1061518.
PMID:19795096
Becker GS, Murphy KM (1988). A theory of Farrelly MC, Engelen M (2008). Cigarette
rational addiction. J Polit Econ, 96:675–700 Chung W, Lee S, Shin K et al. (2008). [Analysis prices, smoking, and the poor, revisited. Am
doi:10.1086/261558. of willingness-to-quit cigarette price among J Public Health, 98:582–583, author reply
Korean male adults]. J Prev Med Public Health, 583 – 58 4.doi:10. 2105/A JPH. 20 07.1326 47
Berg GD, Kaempfer WH (2001). Cigarette 41:136–146.doi:10.3961/jpmph.2008.41.3.136 PMID:18309115
demand and tax policy for race groups in South PMID:18515989
Africa. Appl Econ, 33:1167–1173. Farrelly MC, Nimsch CT, Hyland A, Cummings
Chung W, Lim S, Lee S et al. (2007). [The M (2004). The effects of higher cigarette prices
Biener L, Aseltine RH Jr, Cohen B, Anderka effect of cigarette price on smoking behavior on tar and nicotine consumption in a cohort
M (1998). Reactions of adult and teenaged in Korea]. J Prev Med Public Health, 40:371– of adult smokers. Health Econ, 13:49–58.
smokers to the Massachusetts tobacco tax. Am 3 8 0 . d o i :1 0 . 3 9 6 1/ j p m p h . 2 0 0 7. 4 0 . 5 . 3 71 doi:10.1002/hec.820 PMID:14724893
J Public Health, 88:1389–1391.doi:10.2105/ PMID:17917485
AJPH.88.9.1389 PMID:9736885 Forster M, Jones AM (2001). The role of
Cox D (1972). Regression models and life tobacco taxes in starting and quitting smoking:
Bilgic A, Florkowski W, Abbay C (2009). tables (with discussion). J R Stat Soc [Ser A], duration analysis of British data. J R Stat Soc
Demand for cigarettes in Turkey: an application 34:187–220. Ser A Stat Soc, 164:517–547 doi:10.1111/1467-
of count data models. Empir Econ, 39:733–765 985X.00217.
doi:10.1007/s00181-009-0320-8. Cragg JG (1971). Some statistical models for
limited dependent variables with application to Franks P, Jerant AF, Leigh JP et al.
Bishop JA, Liu H, Meng Q (2007). Are Chinese the demand for durable goods. Econometrica, (2007). Cigarette prices, smoking, and the
smokers sensitive to price? China Econ Rev, 39:829 doi:10.2307/1909582. poor: implications of recent trends. Am J
18:113–121 doi:10.1016/j.chieco.2006.12.001. Public Health, 97:1873–1877.doi:10.2105/
Deaton A (1988). Quality, Quantity, and AJPH.2006.090134 PMID:17761576
Blough DK, Madden CW, Hornbrook MC (1999). Spatial Variation of Price. American Economic
Modeling risk using generalized linear models. Association, 78:418–430. Franz GA (2008). Price effects on the
J Health Econ, 18:153–171.doi:10.1016/S0167- smoking behaviour of adult age groups.
6296(98)00032-0 PMID:10346351 DeCicca P, McLeod L (2008). Cigarette taxes Public Health, 122:1343–1348.doi:10.1016/j.
and older adult smoking: evidence from recent puhe.2008.05.019 PMID:18951594
Cameron L, Williams J (2001). Cannabis, large tax increases. J Health Econ, 27:918–
alcohol and cigarettes: substitutes or 9 2 9 .d o i:10 .1016 / j . j h e a l e c o. 2 0 07.11. 0 0 5 Gallet CA, List JA (2003). Cigarette demand:
complements. Econ Rec, 77:19–34 PMID:18178277 a meta-analysis of elasticities. Health
doi:10.1111/1475-4932.00002. Econ, 12:821–835.doi:10.1002/hec.765
Delnevo CD, Hrywna M, Foulds J, Steinberg PMID:14508867
Centers for Disease Control and Prevention MB (2004). Cigar use before and after a
(CDC) (1998). Response to increases in cigarette excise tax increase in New Jersey. Gallus S, Colombo P, Apolone G et al.
cigarette prices by race/ethnicity, income, Addict Behav, 29:1799–1807.doi:10.1016/j. (2005). A tax to prevent the epidemic of lung
and age groups–United States, 1976–1993. addbeh.2004.04.024 PMID:15530722 cancer. Lancet, 366:288.doi:10.1016/S0140-
MMWR Morb Mortal Wkly Rep, 47:605–609. 6736(05)66974-7 PMID:16039324
PMID:9699809 Dinno A, Glantz S (2009). Tobacco control
policies are egalitarian: a vulnerabilities Gallus S, Fernandez E, Pacifici R et al. (2006).
Chaloupka FJ (1990). Men, women, and perspective on clean indoor air laws, Channels of cigarette distribution, price and
addiction: the c ase of cigarette smoking. NBER cigarette prices, and tobacco use disparities. tobacco consumption in Italy. Prev Med,
Working Paper Series. Working Paper #3267. Soc Sci Med, 68:1439–1447.doi:10.1016/j. 42:132–134.doi:10.1016/j.ypmed.2005.11.003
Cambridge, MA, National Bureau of Economic socscimed.2009.02.003 PMID:19282078 PMID:16375955
Research.
Douglas S (1998). The duration of
Chaloupka FJ (1991). Rational addictive smoking habit. Econ Inq, 36:49–64
behavior and cigarette smoking. J Polit Econ, doi:10.1111/j.1465-7295.1998.tb01695.x.
99:722–742 doi:10.1086/261776.
198
Gallus S, Tramacere I, Boffetta P et al. (2011). Hu TW, Ren QF, Keeler TE, Bartlett J (1995). Kyaing NN, Perucic A-M, Rahman K (2005).
Temporal changes of under-reporting of The demand for cigarettes in California and Study on poverty alleviation and tobacco
cigarette consumption in population-based behavioural risk factors. Health Econ, 4:7–14. control in Myanmar. HNP Discussion Paper
studies. Tob Control, 20:34–39.doi:10.1136/ doi:10.1002/hec.4730040102 PMID:7780529 Series. Economics of Tobacco Control Paper
tc.2009.034132 PMID:20861005 No.31. Washington DC, The World Bank.
Hyland A, Laux FL, Higbee C et al. (2006).
Gallus S, Tramacere I, Zuccaro P et al. (2008). Cigarette purchase patterns in four countries Labeaga JM (1999). A double-hurdle rational
Attitudes and perceptions towards increasing and the relationship with cessation: findings addiction model with heterogeneity: estimating
cigarette price: a population-based survey from the International Tobacco Control (ITC) the demand for tobacco. J Econom, 93:49–72
in Italy. Prev Med, 47:454–455.doi:10.1016/j. Four Country Survey. Tob Control, 15 Suppl doi:10.1016/S0304-4076(99)00003-2.
ypmed.2008.07.002 PMID:18675297 3;;iii 5 9 – iii 6 4.doi:10.113 6 / tc. 20 0 5.012 20 3
PMID:16754948 Lance PM, Akin JS, Dow WH, Loh CP (2004).
García J, Labeaga J (1996). Alternative Is cigarette smoking in poorer nations highly
approaches to modelling zero expenditure: IARC (2008). Measures to assess the sensitive to price? Evidence from Russia and
an application to Spanish demand for effectiveness of tobacco taxation. In: IARC China. J Health Econ, 23:173–189.doi:10.1016/j.
tobacco. Oxf Bull Econ Stat, 58:489–506 Handbooks of Cancer Prevention, Tobacco jhealeco.2003.09.004 PMID:15154693
doi:10.1111/j.1468-0084.1996.mp58003004.x. Control, Volume 12: Methods for Evaluating
Tobacco Control Policies. Lyon, International Laxminarayan R, Deolalikar A (2004). Tobacco
Gardes F, Starzec C (2004). Are tobacco and Agency for Research on Cancer, 189–213. initiation, cessation, and change: evidence
alcohol expenditures price elastic? The case of from Vietnam. Health Econ, 13:1191–1201.
Poland consumption. Paris, Université Paris I Jha P, Chaloupka FJ (1999). Curbing the doi:10.1002/hec.932 PMID:15386650
Panthéon Sorbonne and CNRS-Team, Insee. epidemic. Governments and the Economics
of Tobacco Control. Washington D.C., World Lee JM (2008). Effect of a large increase
Gospodinov N, Irvine I (2009). Tobacco taxes Bank. in cigarette tax on cigarette consumption:
and regressivity. J Health Econ, 28:375– an empirical analysis of cross-sectional
3 8 4 .d o i:10 .1016 / j . j h e a l e c o. 2 0 0 8 .10 . 010 Jiménez-Martín S, Labeaga JM, López A (1998). survey data. Public Health, 122:1061–
PMID:19091432 Participation, heterogeneity and dynamics in 1 0 6 7. d o i : 1 0 .1 0 1 6 / j . p u h e . 2 0 0 7.1 2 . 0 1 3
tobacco consumption: evidence from cohort PMID:18602655
Gruber J, Sen A, Stabile M (2003). Estimating data. Health Econ, 7:401–414.doi:10.1002/
price elasticities when there is smuggling: the 6 , & , $ , ' Lee JM, Hwang TC, Ye CY, Chen SH (2004).
sensitivity of smoking to price in Canada. J HEC361>3.0.CO;;2-2 PMID:9753375 The effect of cigarette price increase on the
Health Econ, 22:821–842.doi:10.1016/S0167- cigarette consumption in Taiwan: evidence
6296(03)00058-4 PMID:12946461 Jiménez-Ruiz JA, Sáenz de Miera B, from the National Health Interview Surveys
Reynales-Shigematsu LM et al. (2008). The on cigarette consumption. BMC Public
Guindon GE, Perucic A-M, Boisclair D (2003). impact of taxation on tobacco consumption in Health, 4:61.doi:10.1186/1471-2458-4-61
Higher tobacco prices and taxes in South- Mexico. Tob Control, 17:105–110.doi:10.1136/ PMID:15598345
East Asia: an effective tool to reduce tobacco tc.2007.021030 PMID:18285383
use, save lives and generate revenue. HNP Lewit EM, Coate D (1982). The potential
Discussion Paper Series, Economics of John RM (2008). Price elasticity estimates for using excise taxes to reduce smoking. J
Tobacco Control Paper No. 11. Washington for tobacco products in India. Health Policy Health Econ, 1:121–145.doi:10.1016/0167-
DC, The World Bank. Plan, 23:200–209.doi:10.1093/heapol/czn007 6296(82)90011-X PMID:10263952
PMID:18424474
Hamilton VH, Levinton C, St-Pierre Y, Grimard Lopez AD, Collishaw NE, Piha TA (1994). A
F (1997). The effect of tobacco tax cuts on Jones A (1989). The UK demand for cigarettes descriptive model of the cigarette epidemic in
cigarette smoking in Canada. CMAJ, 156:187– 1954–1986, a double-hurdle approach. J developed countries. Tob Control, 3:242–247
191. PMID:9012719 Health Econ, 8:133–141.doi:10.1016/0167- doi:10.1136/tc.3.3.242.
6296(89)90012-X PMID:10293368
Hamilton WL, Biener L, Rodger CN (2005). López Nicolás A (2002). How important are
Who supports tobacco excise taxes? Factors Jones A, Yen S (2000). A box-cox double- tobacco prices in the propensity to start and
associated with towns’ and individuals’ support hurdle model. The Manchester School, quit smoking? An analysis of smoking histories
in Massachusetts. J Public Health Manag 68:203–221 doi:10.1111/1467-9957.00190. from the Spanish National Health Survey.
Pract, 11:333–340. PMID:15958933 Health Econ, 11:521–535.doi:10.1002/hec.745
Karki YB, Pant KD, Pande BR (2003). A PMID:12203755
Hanewinkel R, Isensee B (2008). Opinion study on the economics of tobacco in Nepal.
on tobacco tax increase: factors associated HNP Discussion Paper Series, Economics of Manning WG, Mullahy J (2001). Estimating
with individuals’ support in Germany. Tobacco Control Paper No.13. Washington DC, log models: to transform or not to transform? J
Health Policy, 86:234–238.doi:10.1016/j. The World Bank. Health Econ, 20:461–494.doi:10.1016/S0167-
healthpol.2007.10.009 PMID:18054110 6296(01)00086-8 PMID:11469231
King G, Mallett RK, Kozlowski LT, Bendel RB
Harris MN, Zaho X (2007). A zero-inflated (2003). African Americans’ attitudes toward Mao Z, Sung HY, Hu T et al. (2007). Demand
ordered probit model, with an application cigarette excise taxes. Am J Public Health, for cigarette in China. In: Hu TW, ed., Tobacco
to modelling tobacco consumption. J 93:828 – 8 3 4.doi:10. 210 5 /A J PH . 93. 5. 828 control policy analysis in China, Economics
Econom, 141:1073–1099 doi:10.1016/j. PMID:12721152 and Health. World Scientific Publishing Co
jeconom.2007.01.002. Singapore, 129–157.
Krasovsky K, Andreeva T, Krisanov D et
Heckman JJ (1979). Sample selection bias as al. (2002). Economics of tobacco control in Mao Z, Yang GH, Ma JM (2003). Adult’s
a speciation error. Econometrica, 47:153–161 Ukraine from the public health prespective. demand for cigarettes and its determinants in
doi:10.2307/1912352. Kiev, Alcohol and Drug Information Centre. China. Soft Science of Health, 17:19–23.
Hersch J (2000). Gender, income levels, and Kyaing NN (2003). Tobacco economics in Mao ZZ, Jiang J (1997). Demand for cigarettes
the demand for cigarettes. J Risk Uncertain, Myanmar. HNP Discussion Paper Series, and factors affecting the demand: a cross
21:263–282 doi:10.1023/A:1007815524843. Economics of Tobacco Control Paper No.14. sectional survey. Chinese Healthcare industry
Washington DC, The World Bank. Management, 5:227–229.
199
Martin G, Steyn K, Yach D (1992). Beliefs Sarntisart I (2003). An economic analysis of Torabi MR, McAllister L, Kotecki JE (1994).
about smoking and health and attitudes tobacco control in Thailand. HNP Discussion Public opinion on tobacco use, its taxes
toward tobacco control measures. S Afr Med J, Paper Series, Economics of Tobacco Control and public policy. Indiana Med, 87:134–138.
82:241–245. PMID:1411820 Paper No.15. Washington DC, The World Bank. PMID:8176198
Martin G, Yach D, Steyn K (1993). Determinants Sayginsoy O, Yurekli A, De Beyer J (2002). Tsai YW, Yang CL, Chen CS et al. (2005).
of public opinion about tobacco tax increases. Cigarette demand, taxation, and the poor: a The effect of Taiwan’s tax-induced increases
S Afr Med J, 83:359–360. PMID:8211436 case study of Bulgaria. HNP Discussion Paper in cigarette prices on brand-switching and the
Series, Economics of Tobacco Control Paper consumption of cigarettes. Health Econ, 14:627–
Mullahy J (1985). Cigarette smoking: habits, No. 4. Washington DC, The World Bank. 641.doi:10.1002/hec.972 PMID:15791675
health concerns, and heterogeneous
unobservables in an microeconometric Shelley D, Cantrell MJ, Moon-Howard J et al. Tsai YW, Yen LL, Yang CL, Chen PF (2003).
analysis of consumer demand. University of (2007). The $5 man: the underground economic Public opinion regarding earmarked cigarette
Virginia, VA. response to a large cigarette tax increase in tax in Taiwan. BMC Public Health, 3:42.
New York City. Am J Public Health, 97:1483– doi:10.1186/1471-2458-3-42 PMID:14693036
Nassar H (2003). The economics of tobacco 1 4 8 8 . d o i : 1 0 . 2 1 0 5 / A J P H . 2 0 0 5 . 0 7 9 9 2 1
in Egypt: a new analysis of demand. HNP PMID:17600270 Van Kinh H, Ross H, Levy D et al. (2006). The
Discussion Paper Series, Economics of effect of imposing a higher, uniform tobacco
Tobacco Control Paper No.8. Washington DC, Sheu ML, Hu TW, Keeler TE et al. (2004). The tax in Vietnam. Health Res Policy Syst, 4:6.
The World Bank. effect of a major cigarette price change on doi:10.1186/1478-4505-4-6 PMID:16800880
smoking behavior in california: a zero-inflated
Ogloblin C, Brock G (2003). Smoking in Russia: negative binomial model. Health Econ, 13:781– Van Walbeek CP (2002). The distributional
the “Marlboro Man” rides but without “Virginia 791.doi:10.1002/hec.849 PMID:15322990 impact of tobacco excise increases. S Afr J
Slims” for now. Comp Econ Stud, 45:87–103 Econ, 70:560–578.
doi:10.1057/palgrave.ces.8100001. Siahpush M, Wakefield MA, Spittal MJ et al.
(2009). Taxation reduces social disparities in Van Walbeek CP (2005). The economics of
Ohsfeldt RL, Boyle RG (1994). Tobacco excise adult smoking prevalence. Am J Prev Med, tobacco control in South Africa. University of
taxes and rates of smokeless tobacco use in 36:285–291.doi:10.1016/j.amepre.2008.11.013 Cape Town.
the US: an exploratory ecological analysis. Tob PMID:19201146
Control, 3:316–323 doi:10.1136/tc.3.4.316. Wakefield MA, Durkin S, Spittal MJ et al.
Slattery ML, Hunt SC, French TK et al. (1989). (2008). Impact of tobacco control policies
Ohsfeldt RL, Boyle RG, Capilouto E (1997). Validity of cigarette smoking habits in three and mass media campaigns on monthly adult
Effects of tobacco excise taxes on the epidemiologic studies in Utah. Prev Med, smoking prevalence. Am J Public Health,
use of smokeless tobacco products in the 18:11–19.doi:10.1016/0091-7435(89)90050-9 98:1443–1450.doi:10.2105/AJPH.2007.128991
USA. Health Econ, 6:525–531.doi:10.1002/ PMID:2785267 PMID:18556601
6 , & , $ , '
HEC300>3.0.CO;;2-Y PMID:9353656 Sloan FA, Trogdon JG (2004). The impact of Warner KE (1978). Possible increases in the
the Master Settlement Agreement on cigarette underreporting of cigarette consumption. J Am
Ohsfeldt RL, Boyle RG, Capilouto E (1999). consumption. J Policy Anal Manage, 23:843– Stat Assoc, 73:314–318 doi:10.2307/2286658.
Tobacco taxes, smoking, restrictions and 855.doi:10.1002/pam.20050 PMID:15499706
tobacco use. In: Chaloupka FJ, Grossman M, Wasserman J, Manning WG, Newhouse
Bickel WK, et al., eds., The economics analysis Stehr M (2007). The effect of cigarette taxes JP, Winkler JD (1991). The effects of excise
of substance use and abuse: an integration on smoking among men and women. Health taxes and regulations on cigarette smoking.
of econometrics and behavioral economic Econ, 16:1333–1343.doi:10.1002/hec.1223 J Health Econ, 10:43–64.doi:10.1016/0167-
research. Chicago, University of Chicago PMID:17335102 6296(91)90016-G PMID:10112149
Press, 15–29.
Stephens T, Pederson LL, Koval JJ, Kim C Wilson N, Weerasekera D, Edwards R et al.
Onder Z (2002). The economics of tobacco in (1997). The relationship of cigarette prices (2010). Characteristics of smoker support for
Turkey: new evidence and demand estimates. and no-smoking bylaws to the prevalence increasing a dedicated tobacco tax: national
HNP Discussion Paper Series, Economics of of smoking in Canada. Am J Public Health, survey data from New Zealand. Nicotine
Tobacco Control Paper No. 2. Washington DC, 87:1519 –1521.doi:10.2105/AJPH.87.9.1519 Tob Res, 12:168–173.doi:10.1093/ntr/ntp178
The World Bank. PMID:9314807 PMID:20018943
Peretti-Watel P (2004). Pricing policy and Taal A , Kiivet R , Hu T W (2004). The economics Zhao X, Harris MN (2004). Demand for
some other predictors of smoking behaviours: of tobacco in Estonia. HNP Discussion Paper marijuana, alcohol and tobacco: participation,
an analysis of French retrospective data. Int Series, Economics of Tobacco Control Paper levels of consumption and cross-equation.
J Drug Policy, 16:19–26 doi:10.1016/j.drugpo. No. 19. Washington DC, The World Bank. Econ Rec, 80:394–410 doi:10.1111/j.1475-
2004.06.003. 4932.2004.00197.x.
Tauras JA (2004). Public policy and some-day
Ross H, Blecher E, Yan L, Hyland A (2011). smoking among adults. J Appl Econ, 7:137–162.
Do cigarette prices motivate smokers to quit?
New evidence from the ITC survey. Addiction, Tauras JA (2006). Smoke-free air laws,
106:609 – 619.doi:10.1111/j.1360-0443.2010. cigarette prices, and adult cigarette demand.
03192.x PMID:21059183 Econ Inq, 44:333–342 doi:10.1093/ei/cbj028.
200
Chapter 6
Tax, price and tobacco use among young people
Introduction
An overwhelming majority of adult Western Pacific region having the suggest that youth place a relatively
smokers initiate smoking while they lowest prevalence rate at 6.6%. greater importance on the short-
are adolescents or young adults. The term costs of smoking, such as the
World Bank estimates that in high- Youth responsiveness to prices monetary price of cigarettes, than on
income countries, approximately and taxes – Theoretical foundations the long-term costs of smoking, such
eight out of 10 smokers start to as the future health consequences
smoke in their teens, while in low- Consistent with economic theory, of smoking. Lewit et al. (1981) offer
and middle-income countries, most adolescents and young adults are two additional reasons why youth
smokers start to smoke by their early expected to have a larger response are expected to be more price-
twenties (Jha and Chaloupka, 1999) to price changes than older adults elastic than adults. First, established
According to the 1999–2007 (see Chapter 5 for discussion adolescent smokers are likely to
Global Youth Tobacco Survey of price and tax effects on adult have shorter smoking histories than
(GYTS), approximately 9.5% of demand for tobacco). That is, the adults and therefore would likely
students worldwide aged 13–15 demand for tobacco by adolescents respond more to price changes than
currently smoke cigarettes (Centers and young adults is expected to be long-time smokers who are more
for Disease Control and Prevention, more price-elastic than the demand addicted. Second, adolescents are
2008). The smoking prevalence rate by older adults. There are several likely to be more easily influenced by
of these students varies considerably reasons for this prediction. First, as their peers (i.e. friends and siblings)
by region, with the European Region Grossman and Chaloupka (1997) than are adults. That is, relative
having the highest prevalence point out, the fraction of disposable to older adults, young people are
rate at 19.2% and the Eastern income spent on cigarettes by young more likely to smoke if their peers
Mediterranean region having the smokers is likely to be greater than also smoke. This implies that an
lowest smoking prevalence rate at that of adult smokers. Economic increase in cigarette price would not
4.9%. Moreover, estimates from the theory predicts that the greater the only directly reduce youth smoking,
GYTS suggest that 10.1% of students share of disposable income a good but would also indirectly reduce
worldwide aged 13–15 use tobacco takes up, the more responsive youth smoking by decreasing peer
products other than cigarettes, such individuals will be to price changes. smoking. In addition, young people’s
as smokeless tobacco, pipes, water Grossman and Chaloupka (1997) demand for tobacco can also be
pipes and bidis. Again, other tobacco also highlight the point that indirectly influenced, either positively
prevalence rates by students vary adolescents typically have a greater or negatively, through observance
considerably by region, with the propensity to discount the future (i.e. of parental smoking. Finally, young
Eastern Mediterranean region having have higher discount rates) than people’s responsiveness to tobacco
the highest rate at 12.0% and the adults. The higher discount rates price increases is also indirectly
201
influenced through access to cheap Smoking prevalence and intensity was considerably larger than the
or free tobacco from social sources, among young smokers – United consensus estimate for adults. They
such as peers and parents. States found that most of the impact of price
The 1989 U.S. Surgeon General’s on youth smoking came through
report suggests that smoking Numerous studies examining the its impact on smoking prevalence,
behaviour occurs on a continuum, determinants of youth tobacco use with an estimated price elasticity of
from initiation to experimentation have been conducted over the past VPRNLQJ SUHYDOHQFH HTXDO WR í
to regular smoking followed by three decades. With few exceptions, Price played a smaller role in altering
dependence or addiction (U.S. these studies have taken advantage average smoking by young smokers,
Department of Health and Human of natural experiments that result with an estimated price elasticity
Services, 1989). Cigarette prices are from government changes in tobacco of cigarette consumption among
also likely to have a differential effect product taxes and/or prices. Given VPRNHUV HTXDO WR í 7KHVH
on smoking depending on where on the significant spatial (i.e. state estimates imply that a 10% increase
the smoking uptake continuum the and local) and temporal (i.e. over in the price of cigarettes would
individual is. For example, individuals time) changes in tobacco taxes that reduce youth smoking prevalence by
in the experimentation stage of have occurred in the United States 12% and reduce the average number
smoking initiation will likely be less over the past several decades, a of cigarettes consumed by young
affected by cigarette price changes preponderance of these studies have smokers by 2.5%.
than those who are farther along the examined youth tobacco use in the A follow-up study by Lewit and
uptake continuum because those United States. However, a growing Coate (1982) used the 1976 National
individuals in the experimentation number of studies have focused on Health Interview Survey to estimate
stage typically do not yet purchase youth tobacco use in other high- smoking prevalence and intensity
their own cigarettes and are not income countries, and more recently of smoking equations for different
directly exposed to price. On the several studies have focused on subpopulations, including individuals
other hand, individuals who purchase low- and middle-income countries. of different ages (20–25 years,
their own cigarettes are likely to have Table 6.1 summarizes the published 26–35 years, and 36–74 years) and
a larger response to price changes literature providing evidence on the genders. They found that young
than experimenters. effects of cigarette prices and taxes adults (aged 20–25 years) were
As described below, the empirical on smoking prevalence and smoking the most responsive to changes in
evidence on the demand for tobacco intensity among young people. price, with an estimated total price
products among young people is Lewit and colleagues (1981) HODVWLFLW\ RI GHPDQG HTXDO WR í
consistent with economic theory. were the first to examine the effect The authors estimated total price
That is, youth and young adult of cigarette prices on adolescent elasticities of demand for adults
tobacco demand has been found smoking. They employed data on aged 26–35 and adults aged 36–74
to be inversely related to cigarette 6768 youth aged 12 to 17 taken WREHíDQGíUHVSHFWLYHO\
prices, and with few exceptions, the from Cycle III of the nationally The absolute values of the price
absolute value of the price elasticity representative Health Examination elasticities of demand for the
of demand is found to be larger for Survey, conducted from March 1966 adolescents included in this study
young people than for adults. through March 1970. They used a are generally smaller than those
two-part model of cigarette demand found in Lewit et al. (1981), most likely
Youth responsiveness to prices in which smoking prevalence and because individuals in this study
and taxes – Review of evidence intensity of cigarette use among are older than those in Lewit et al.
from demand studies smokers were estimated separately. (1981). With respect to the youngest
In addition to price, they controlled sample, those aged 20–25, males
For the purpose of this review, generally for anti-smoking publicity, cigarette were found to significantly respond
young people refers to individuals advertising and a variety of to price changes with an estimated
under 30 years of age (13–18 youth;; socioeconomic and demographic total price elasticity of demand equal
19–30 young adults);; however, some factors in their equations. Lewit WRíDQGIHPDOHVZHUHIRXQGQRW
studies may provide price elasticity and colleagues estimated that to respond to prices.
estimates for age groups that extend the total price elasticity of youth
above this upper age limit. VPRNLQJZDVíDQHVWLPDWHWKDW
202
Table 6.1. Summary of studies on the effects of cigarette price on young people’s demand for tobacco products
Publication (location, Total price elasticity of Price elasticity of smoking Price elasticity of smoking
Methods and Model Comments
author, year) demand prevalence intensity
High-income countries
USA Methods í –1.20 í
(Lewit et al., 1981) 1966–1970;; Cycle III of the Omitting price differential
nationally representative
Health Examination Survey of
\RXWKDJHGíQ
Model
Ordinary least squares
multivariate regression
controlling for socioeconomic
and demographic factors and
antismoking and tobacco
advertising
USA Methods Overall Overall Overall Restricted sample: individuals
(Lewit and Coate, 1982) 1976;; National Health 20–74 years 20–74 years 20–74 years whose own average price
Interview Survey, nationally í6( í6( í6( was greater than the price
representative sample of 19 Restricted sample Restricted sample Restricted sample within the 20-mile band were
266 individuals aged 20–74 20–74 years 20–74 years 20–74 years removed
from 416 survey sites. í6( í6( í6(
Model By age By age By age
Ordinary least squares 20–25 years 20–25 years 20–25 years
regression and variance í6( í6( í6(
components GLS procedure;; 26–35 years 26–35 years 26–35 years
adjusted for socioeconomic í6( í6( í6(
and demographic factors 36–74 years 36–74 years 36–74 years
and region and city size í6( í6( í6(
characteristics. By age and sex By age and sex By age and sex
Subgroup analysis of demand Females Females Females
by age is estimated by a FIML 20–25 years 20–25 years 20–25 years
logit procedure í6( í6( í6(
26–35 years 26–35 years 26–35 years
í6( í6( í6(
í6( í6( í6(
By age and sex By age and sex By age and sex
Males Males Males
í6( í6( í6(
í6( í6( í6(
í6( í6( í6(
203
204
USA (Chaloupka, 1991) Method By age Long-run price elasticities

1976–1980;; National Health 17–24 years
and Nutrition Examination íWR
Survey, nationally 25–64 years
representative sample of 28 íWRí
000 individuals aged 6 mos
to 74 years.
Model
Becker-Murphy model of
rational addiction;; two stage,

least squares regression
controlling for socioeconomic
and demographic factors
USA (Wasserman et al., Method Adults No statistically significant
1991) 1970–1985;; National Health 1970: 0.059 (SE 0.076) difference between youth and
Interview Survey data on 207 í6( adult smoking over time
647 individuals from 7 of the í6( Adjustment for state policy
9 supplementary smoking restricting smoking
questionnaires implemented
between 1970 and 1985.
Data on smoking among
12–1 7-year-olds from the
National Health and Nutrition
Examination Survey
(n IRUí
Price data from Tobacco
Institute’s 1986 report The Tax
Burden on Tobacco weighted
average price by state.
Model
Estimated a generalized
linear model using an
iterative weighted least
squares technique, adjusted
for socioeconomic and
demographic factors and
other anti-smoking regulations
USA (Chaloupka and Method Overall Overall Overall Estimates are average of the
Grossman, 1996) 1992, 1993, 1994;; Data from í í –0.638 different models/samples.
8th-, 10th- and 12th-grade Full Sample Full Sample Full Sample Price only model excludes
students (ages 13–18) from Price only model Price only model Price only model other policy related variables,
the Monitoring the Future í –0.799 –0.651 subject to omitted variables
Survey (n = 110 717). Price Full model Full model Full model bias.
data from the Tobacco í –0.376 –0.470 Full model includes other
Institute’s report The Tax policy variables, but likely
Burden on Tobacco subject to collinearity.
USA (Chaloupka and Model Restricted sample Restricted sample Restricted sample Restricted sample excludes
Grossman, 1996) (contd) Two-part model of demand: Price only model Price only model Price only model individuals living within 25
probit methods to estimate í –0.923 –0.779 miles of a state with lower
a smoking prevalence Full model Full model Full model cigarette prices
equation and least squares í –0.602 –0.652
methods to estimate
consumption by smokers,
adjusted for socioeconomic
and other anti-smoking
regulations
USA (Chaloupka and Method Overall Overall Overall Two measures of
Wechsler, 1997) 1993;; Nationally í í í consumption: ordered level
representative sample of 16 Full sample Full sample Full sample consumption (0 = non-
570 students from 140 US í í í smoker, 1 = light smoker
Colleges and Universities Restricted sample etc) and a ‘categorically
from the Harvard College í continuous’ measure 0, 0.5,
Alcohol Study. Site-specific Estimates are averages of the 5, 10, 15, 20, 30). Restricted
cigarette price from the different models/samples sample excludes all students
1993 American Chamber attending a college/university
of Commerce Researchers within 20 miles of a state with
Association Quarterly Inter- lower cigarette excise taxes/
City Cost of Living Index price
Model
Ordered level consumption:
Ordered probit specification.
For ‘categorically continuous’
consumption: two-part
model of demand – probit
methods to estimate smoking
prevalence and least
squares method for the
smoking intensity equation.
Adjusted for socioeconomic
and demographic factors,
regional and institutional
characteristics
USA and Canada Method Price elasticity of smoking
(Lewit et al., 1997) 1990 and 1992;; two cross- prevalence
sectional school-based Full sample
surveys of 9th-grade í
students (aged 13–16) By gender
from 21 North American Boys
communities representing 15 í
432 individuals as part of the Girls
COMMIT Project í
Model Price elasticity of intention
Mutivariate logistic to smoke
205
206
USA and Canada regression models including Full sample

(Lewit et al., 1997) (contd) cigarette price, exposure to í
COMMIT interventions, age, By gender
race, gender, year and state Boys
level policy interventions as – 0.92
independent variables Girls
í
USA Method Overall Overall Overall
(Centers for Disease Control Pooled data from 1976–1980, í í í
and Prevention, 1998) 1983, 1985, 1987–1993 By gender By gender By gender
National Health Interview Males Males Males
Surveys (n = 355 246). í í í
Price data from the Tobacco Females Females Females
Institute. –0.19 í í
Two-part model of demand: 18–24 years 18–24 years 18–24 years
probit model to estimate –0.58 í í
smoking prevalence and 25–39 years 25–39 years 25–39 years
Ordinary Least Squares to –0.42 í í
estimate smoking intensity, 40+ years 40+ years 40+ years
controlled for year, region, –0.10 í í
socioeconomic and
demographic factors
USA Method Tax Tax Tax
(Evans and Farrelly, 1998) 1979, 1987;; nationally 18–24 years 18–24 years 18–24 years
representative cross- í í í
sectional (pooled surveys) 25–39 years 25–39 years 25–39 years
data on adults 18 years í í í
and over from the National
Health Interview Survey
supplements 1979 (Smoking)
and 1987 (Cancer Control)
(n = 48 314).
Model
Two-part model: 1) probit
(smoking prevalence);;
2) simple linear regression
OLS (smoking intensity).
Analyses adjusted for
socioedemographic
variables.
USA (Chaloupka and Pacula, Method Males Statistically significant at
1999) 1992, 1993, 1994;; Data from – 0.928*** ***1%, **5%, and *10%
8th-, 10th- and 12th-grade White
students (ages 13–18) from – 0.861***
USA (Chaloupka and Pacula, the Monitoring the Future Black

1999) (contd) Survey (n = 110 717). Price – 1.646***
data from the Tobacco Females
Institute’s report The Tax – 0.595**
Burden on Tobacco White
Model – 0.451**
Probit maximum likelihood Black
specification – 0.453
Whites
– 0.639***
Blacks
– 1.108*
USA Method Estimates based on pooled Estimates using pooled data
(Dee, 1999) 1977–1992 pooled data from 1977–1992 from 1977–1992 suggest that
representative data on high WRí cigarette taxes appear to
school seniors from 44 US Estimates based on data have implausibly signed or
states from the Monitoring from 1985–1992 statistically imprecise effects
the Future Surveys íWRí on teen smoking prevalence;;
(n = 255 560) richer specifications using
Model data from 1985–1992 provide
Weighted least squares relatively weak statistical
estimates of reduced form significance in some models
equations for smoking
prevalence. Analyses control
for demographic variables,
year and state fixed effects,
socioeconomics and
macroeconomic covariates,
race and gender-specific
year fixed effects and state-
specific time trends.
USA Method By age By age By age
(Harris & Chan, 1999) 1992–1993;; Tobacco Use 15–17 years 15–17 years 15–17 years
Supplements to the Current í6( í6( í6(
Population Survey. 18–20 years 18–20 years 18–20 years
Model í6( í6( í6(
Two-stage model of demand: 21–23 years 21–23 years 21–23 years
probit specification for the í6( í6( í6(
prevalence equation and OLS 24–26 years 24–26 years 24–26 years
equation for the smoking í6( í6( í6(
intensity equation. 27–29 years 27–29 years 27–29 years
Probit equations and í6( í6( í6(
conditional regressions with Separate equations Separate equations
two price variables: average Premium Premium Brands
price of premium and average 15–17 years 15–17 years
price of discount brands. í6( í6(
Model 1 enters each price 18–20 years 18–20 years
variable separately and í6( í6(
207
208
USA Model 2 includes both in a 21–23 years 21–23 years

(Harris & Chan, 1999) (contd) single equation í6( í6(
24–26 years 24–26 years
í6( í6(
í6( í6(
Discount brands Discount Brands
í6( í6(

í6( í6(
í6( í6(
í6( í6(
í6( í6(
Single equation Single equation
Premium Premium Brands
í6( í6(
í6( í6(
í í6(
í6( í6(
í6( í6(
Discount brands Discount Brands
0.041 (SE 0.591) í6(
0.799 (SE 0.376) 0.504 (SE 0.258)
0.563 (SE 0.270) 0.525 (SE 0.216)
0.577 (SE 0.254) í6(
0.575 (SE 0.224 0.346 (SE 0.162)
USA Method í í í Consumption is ‘categorically

(Tauras and Chaloupka, 1976–1993;; Panel data continuous’ assuming the
1999) nationally representative following values: 0, 15, 90,
cross-sectional surveys of 300, 600, 900 and 1200
8th-, 9th- and 10th-grade
USA students (ages 13–18) from

(Tauras and Chaloupka, the Monitoring the Future
1999) (contd) Survey (approximately
50 000 persons). Price data
from the Tobacco Institute’s
report The Tax Burden on
Tobacco
Model
Two-part, individual fixed
effects model of cigarette
demand: linear probability
model used to estimate
smoking prevalence equations
and least squares techniques
to estimate smoking intensity
USA Method Current smokers 14+ years Current smokers 14+ years Current smokers 14+ years
(Emery et al., 2001) 1993;; Nationally í í í
representative data for youths Established smokers 14+ Established smokers 14+ Established smokers 14+
aged 10–22 from the second years years years
wave of the longitudinal í í í
teenage attitudes and Experimenters 10+ years
practices survey (n = 12 952). non-significant
Price data from the Tobacco
Institute.
Model
Two-part model of cigarette
demand. Smoking intensity
modelled separately for
current smokers (smoked in
last 30 days) and established
smokers (smoked in last
30 days and at least 100
cigarettes in lifetime)
separately
USA Method Total sample Total Sample Total sample Only significant estimates of
(Farrelly et al., 2001) Pooled data from 1976–1980, í í –0.15 price elasticity of smoking
1983, 1985, 1987–1993 By gender By gender By gender prevalence and intensity are
National Health Interview Males Males Males included in the total price
Surveys (n = 354 228). í í –0.18 elasticity estimates
Price data from the Tobacco Females Females Females
Institute. –0.32 –0.19 í
Two-part model of demand: 18–24 years 18–24 years 18–24 years
probit model to estimate –0.55 –0.30 –0.25
smoking prevalence and 25–39 years 25–39 years 25–39 years
Ordinary Least Squares to –0.53 –0.25 –0.28
estimate smoking intensity, 40+ years 40+ years 40+ years
controlled for year, region, –0.00 –0.02 –0.06
socioeconomic and
demographic factors, and
209
state-specific effects
210
USA Method MTF data

(Gruber and Zinman 2001) 1991–1997;; 3 data sets: í
Monitoring the Future YRBS data
Surveys of 8th-, 10th-, and í
12th-grade students Natality data
(n = 336 665), Youth Risk í
Behavior Surveys of 9 th –12th
grade students (n = 53 278)
and Vital Statistic Natality
data on teenage mothers

during pregnancy (n = 1985).
State-level price and tax from
the Tobacco Institute
Model
Regression, controlled for
demographic factors, state
clean air laws and youth
access restrictions
USA Method Using average state prices Using average state prices Using average state prices
(Ross and Chaloupka, 2003) 1996;; Cross-sectional í í í
data for 17 287 high school Using perceived prices Using average perceived Using perceived prices
students from 202 schools. í prices í
Three measures of price: í
perceived price from smokers
and non-smokers, Tobacco
Institute price data and
ACCRA cost-of-living index
price data.
Model
demand: probit specification
for smoking prevalence
equation, ordinary least
squares for smoking intensity,
controlling for state-level
smoke-free air laws and
youth access laws and the
possibility of smuggling
USA Method Smokers’ ages Smokers respond to higher
(Farrelly et al., 2004) 1988 and 1993;; Longitudinal 25–34 years cigarette prices by reducing
data for 3675 smokers í aintensity, but also by
aged 25–64 from the 1988 35–44 years switching to cigarettes with
Community Intervention Trial í higher levels of tar and
(COMMIT). 45–64 years nicotine
Model í
USA Linear regression of average

(Farrelly et al., 2004) (contd) daily cigarette consumption
adjusted for age, race, sex,
marital status education
income clean air laws index,
nicotine and tar. State
average cigarette price
USA Method Using average state price Using average state price Using average state price **, * Significant at 5% and
(Ross and Chaloupka, 2004) 1996;; Cross-sectional íWRí –0.393* –0.052 10% level based on two-
data for 17 287 high school Using perceived price Using perceived price Using perceived price tailed test after its standard
students from 202 schools. íWRí í í error was adjusted for
Two measures of price: clustering
perceived price from both
smokers and non-smokers,
and Tobacco Institute price
data
Model
demand: probit specification
for smoking prevalence
equation, generalized linear
model for smoking intensity,
controlling socioeconomic
as well as for state-level
smoke-free air laws and
youth access laws and the
possibility of smuggling
USA Method 18–20 years
(Sloan and Trogdon, 2004) 1990–2002;; repeated cross- í
sectional data on a nationally 21–24 years
representative sample of í
adults (n = 1 762 686) from 25–44 years
the Behavioural Risk Factor í
Surveillance System survey. 45–64 years
Model í
Probit specification to 65+ years
estimate smoking prevalence í
elasticity controlling
demographic factors
USA (DeCicca et al., 2005) Method Model 1
1988–2000;; nationally 2000 tax rate
representative panel micro í
data from the National Model 2
Education Longitudinal 1992 tax rate
Survey of 8th graders í
followed up 2, 4 and 12 years
211
212
USA (DeCicca et al., 2005) later. (n = 11 326) 2000 tax rate
(contd) Model í
Probit model of smoking Full sample
prevalence. Model 1 uses í
contemporaneous smoking Movers
and tax data;; Model 2 í
incorporates past tax data. Stayers
Model 3 compares those who í
move states and those who Corrected for attrition
stay in the same state. All Model 1

three models are estimated 2000 tax rate
with and without correcting í
for attrition. Model 2
1992 tax rate
í
2000 tax rate
í
Model 3
Full sample
í
Non-switchers
í
Switchers
í
USA Method í í í
(Tauras et al., 2005) 1997–2001;; National
Longitudinal Survey of Youth
1997 Cohort. Nationally
representative sample of
9022 youths aged 12–16, first
wave in 1997, annual follow-
up for 4 years. Price data
from Tobacco Institute.
Model
demand using a fixed effects
approach that controls for
unobserved individual level
heterogeneity and individual
invariant year-specific
unobserved heterogeneity
USA (Carpenter and Cook, Method National State and local versions
2008) 1991–2005;; National íí of YRBS only sources of
representative data from the State/local youth smoking information
Youth Risk Behavior Survey íí designed to be representative
USA (Carpenter and Cook, (n = 101 633) and aggregate of the sampled state or
2008) (contd) data from independent state locality
and local versions of the
YRBS representative at the
state/local level (n = 278,
representing over 750 000
youths)
Model
1) For national data: Two-
way fixed effects model,
controlling for demographics
and area and year fixed
effects, using standard logit
regression
2) For state and local data:
separate weighted Ordinary
Least Squares regressions
3) Model including a direct
measure of anti-smoking
sentiment using national data
USA (DeCicca et al., 2008a) Method 1992 to 2000 data 1992 to 2000 data 1992 to 2000 data * Statistically significant at
1992 and 2000;; National íWRí íWRí íWRí 1%.
Education Longitudinal Study, Including direct measure of Including direct measure of Price elasticities estimated
two waves as cross-sections: anti-smoking sentiment anti-smoking sentiment including anti-smoking
1992 (n = 16 730) and 2000 WRí WRí sentiment not statistically
(n = 11 490). State cigarette significant.
data from Tax Burden on ‘Categorical continuous’
Tobacco variable constructed for
Model consumption based on
Two part model of cigarette conditional categorical
demand: probit model of means
smoking prevalence and
ordinary least squares
model of smoking intensity,
controlling for state
anti-smoking sentiment,
youth access restrictions,
socioeconomic and
demographic variables
USA (DeCicca et al., 2008b) Method Model 1 * Statistically significant
1988–2000;; 1988, 1992 and í
2000 waves of the National Model 2
Education Longitudinal Study 0.13
(n = 10 336) including youths
aged 18–26 years.
Model
Myopic addiction model using
probit specifications,
213
214
USA (DeCicca et al., 2008b) controlling for socioeconomic

(contd) and demographic factors
and anti-tobacco sentiment.
Three models estimated:
1) using intrastate variation in
cigarette excise taxes,
2) including a direct measure
of anti-tobacco sentiment
3) using variation in state
excise taxes faced by youths

who moved state between
study waves
USA Method Overall Overall Overall
(Franz, 2008) 1993–2000;; Nationally í í í
representative cross- 18–29 years 18–29 years 18–29 years
sectional data for adults í í í
18+ years from the Behaviour 30–39 years 30–39 years 30–39 years
Risk Factor Surveillance í í í
System Survey (n = 1 million) 40–64 years 40–64 years 40–64 years
Method í í í
Two models were considered: 65+ years 65+ years 65+ years
1) a simple OLS and í í í
2) a two-part model. Analyses
controlled for socioeconomic
and demographic factors.
Price was the state average
real cigarette price
USA (Fletcher et al., 2009) Method Baseline estimates Tax elasticities.
1994–2002;; data from the OLS Study suggests two
first wave of the national í attributes are associated
Longitudinal Study of Poisson with adolescent types: self-
Adolescent Health;; nationally í control and future discount
representative sample of 7th- Negative binomial rates. Adolescents with
to 12th-grade students in 1994 í lower self-control and higher
followed up one and six years Finite Mixture Models future discount rates are
later (about 90 000). State Poisson less responsive to tax than
level excise tax used. Light smokers counterparts with higher self-
Model í control and lower discount
OLS, Poisson, and negative Heavy smokers rates
binomial regression used í
to estimate tax response. Negative binomial
Finite mixture models used to Light smokers
estimate differential response í
by ‘type’ (light and heavy Heavy smokers
smokers). Analyses í
USA (Fletcher et al., 2009) controlled for race, gender,

(contd) age, family income, parental
education and employment,
family structure, religiosity
and rural/urban
United Kingdom (Townsend Method By gender * P < 0.05
et al., 1994) 1972–1990;; nationally Men (95% CI) ** P < 0.01
representative random ííWRí *** P < 0.001
sample of British adults Women
from the Biennial General ííWRí
Household Survey. By age and gender
Model Men
Multiple regression analysis 16–19 years
0.06 (SE 0.32)
20–24 years
0.16 (SE 0.26)
25–34 years
í6(
35–49 years
í6(
50–59 years
í6(
60+ years
í6(
Women
16–19 years
í6(
20–24 years
í6(
25–34 years
í6(
35–49 years
í6(
50–59 years
í6(
60+ years
í6(
Canada Method In 1994 In 1994 In 1994
(Dupont and Ward, 2002) 1994–1998;; National í í í
Population Health Data, In 1998 In 1998 In 1998
nationally representative í 0.434 í
longitudinal data for
individuals aged 14–18
during cycle 1. Price data
from Statistics Canada
Model
demand: smoking propensity
215
216
Canada estimated using a probit

(Dupont and Ward, 2002) function and smoking
(contd) intensity estimated using
linear regression
Canada 1992 to 1999;; Price elasticity of smoking Price elasticity of daily
(Sen and Wirjanto, 2010) pooled data from 1992–1996 participation smoking
Waterloo Smoking Prevention í í
Programme;; 1991 General
Social Survey;; 1994 Youth

Smoking Survey;; 1996/97
and 1998/99 National
Population Health Surveys,
1999 Canadian Tobacco Use
Monitoring Survey
Model
Probit and OLS estimation
controlling for gender and
employing province and year
effects
27 European countries Method Univariate Odds ratio of daily smoking
(Schnohr et al., 2008) 2001/02;; Health Behaviour 0.91 (0.83–1.00) (95% confidence interval).
in School Aged Children: Model 1 No significant relationships
pooled, cross-sectional data 0.86 (0.78–0.95) between price and youth
from 92 217 youths aged Model 2 smoking prevalence,
13 and 15 years from 27 0.91 (0.80–1.04) possibly as a result of
countries. Price and policy Model 3 inability to adequately control
data from 2003 the WHO 0.94 (0.81–1.10) for intercountry factors
Regional Office for European correlated with price
Tobacco Control Database
Model
Regression analysis with
daily smoking as the outcome
and the stepwise inclusion of
national tobacco policy, adult
tobacco epidemiology and
macro-economic variables.
Multilevel modeling to
determine the effects of each
group of variables. Analyses
adjusted for age and stratified
by gender

South Africa Method Not estimated Percentage change Percentage change in Percentage change in real
(van Walbeek, 2002) 1993–2000;; All Media and in estimated smoking aggregate cigarette price of cigarettes 92.7%
Products survey—national prevalence consumption No estimation of price
repeated cross-sectional Total sample í elasticity by age in models
surveys of between í Percentage change in controlling for other
14 000–30 000 individuals By age average consumption determinants of demand
16 years and over. 16–24 years Total sample
Model í í
Regression 25–34 years
í
Ukraine (Krasovsky et al, Method Total sample Self-reported price,
2002) National survey data on í endogenous
individuals 14 years 4–17 years Cigarette consumption
and over (n = 2700) Low income derived from reported
Model í monthly cigarette expenditure
Ordinary least squares Medium income divided by reported pack
regression, adjusted í price.
for socioeconomic and High income No information on statistical
demographic factors í significance of estimates
18–28 years
Low income
í
Medium income
í
High income
í
29+ years
Low income
í
Medium income
í
High income
í
Myanmar (Kyaing, 2003) Method Total sample Total sample Total sample Prices for smoking
1980–2001;; Survey data from í í í households based on
75 townships By age By age By age expenditures and cigarette/
Model 15–24 years 15–24 years 15–24 years cheroots, home rolled
Two part demand model: –2.41 –1.992 –0.418** tobacco consumption and
Logit model to estimate 25–34 years 25–34 years 25–34 years used to predict prices for
smoking probability and –1.596 –1.231 –0.365** non-smoking households –
OLS regression to estimate 35–44 years 35–44 years 35–44 years endogenous
consumption equation –1.398 –1.08 –0.318** **: P
–1.263 –0.943 –0.32**
–1.253 –0.856 –0.397**
65+ years 65+ years 65+ years
– 1.167 –0.887 –0.28**
217
218
Table 6.1 Summary of studies on the effects of cigarette price on young people’s demand for tobacco products
Nepal (Karki et al., 2003) Method Total sample Total sample Total sample Prices for smoking
2000;; Smoking Behaviour í í í households based on
Survey, nationally By age By age By age expenditures and cigarette/
representative data on 15–24 years 15–24 years 15–24 years bidi consumption and
1438 households and 5957 –1.879 –1.32 –0.599* used to predict prices for
individuals 25–34 years 25–34 years 25–34 years non-smoking households—
Model –0.945 –0.449 –0.496** endogenous
Two part demand model: 35–44 years 35–44 years 35–44 years *: P
Logit model to estimate –0.686 –0.306 í **: P
smoking probability and 45–54 years 45–54 years 45–54 years

OLS regression to estimate –0.605 –0.263 –0.342**
consumption equation 55–64 years 55–64 years 55–64 years
–0.682 –0.355 –0.327**
–1.1 –0.459 –0.641**
Thailand (Sarntisart, 2003) Method Total sample Individual smoking

2000 Household í consumption estimated
Socioeconomic Survey data By age based on consumption of
for 11 968 households from 9 8–17 years smoking households with
regions and 4 rural areas –0.3075 individuals of different ages
Model 18–29 years and data on age of initiation
Linear expenditure system í
model using pooled 30–39 years
estimation í
40–49 years
í
50–59 years
í
60 and older
í
China and the Russian Method China China China *P
Federation (Lance et al., 1993–1997;; longitudinal Pooled Pooled Pooled **P
2004) data for respondents over í í í
13 years of age from nine Community fixed effects Fixed effects A Fixed effects A
Chinese provinces from the í í í
Chinese Health and Nutrition Pooled cross-section Fixed effects B Fixed effects B
Survey (n = 8557). Russian \HDUV í í
data from the 1996, 1998 and í Pooled cross-section Pooled cross-section
2000 waves of the Russian 21–24 years 21–24 years 21–24 years
Longitudinal Monitoring –0.053 0.141 0.564
Survey of adults over 13 25–54 years 25–54 years 25–54 years
(n = 10 638 men) í 0.093 0.389
Model 55+ years 55+ years 55+ years
Two-part demand model: logit í í 0.180
China and the Russian specification for prevalence Community fixed effects Fixed effects A Fixed effects A
Federation (Lance et al., equation and linear \HDUV 21–24 years 21–24 years
2004) (contd) regression for the smoking 0.077 0.141 0.571
intensity equation. Pooled 21–24 years 25–54 years 25–54 years
cross-sectional specifications 0.122 0.092 0.404
and including community 25–54 years 55 +years 55+ years
fixed-effects í í 0.187
55+ years Fixed effects B Fixed effects B
í 21–24 years 21–24 years
0.070 í
0.030 í
55+ years 55+ years
í í
The Russian Federation The Russian Federation The Russian Federation
Pooled Pooled Pooled
í í í
Community fixed effects Fixed effects A Fixed effects A
í í í
\HDUV Fixed effects B Fixed effects B
í í í
í í í
í 0.005* 0.025*
í 0.004 0.035*
Community fixed effects Fixed effects A Fixed effects A
\HDUV 21–24 years 21–24 years
í í í
í 0.005* 0.024*
25–54 years 55+ years 55+ years
í 0.004 0.032*
55+ years Fixed effects B Fixed effects B
í 21–24 years 21–24 years
í í
0.005 0.025*
55+ years 55 + years
0.002 0.029*
Russian Federation, Moscow Method íWRí íWRí íWRí
(Ross, 2004b) 1999;; data from the Global
Youth Tobacco Survey on
7th-, 8th-, 9 th-and 10th-grade
students (aged 11–17)
(n = 4074). School-level
measure of average cigarette
price.
219
220
Russian Federation, Moscow Model

(Ross, 2004b) (contd) Two-part model of cigarette
demand: probit model to
estimate smoking prevalence
equation and Ordinary Least
Squares to estimate smoking
intensity equation, controlling
for socioeconomic factors
and school anti-smoking

policies
Ukraine (Kiev), (Ross, 2004a) Method íWRí íWRí Average school price
1999;; Global Youth Tobacco calculated based on reported
Survey data for 4156 price from smoking students
individuals from 100 schools
in Kiev city
Model
Probit model used to estimate
smoking prevalence equation
India (Joseph, 2010) Method í Total sample Total sample
2000–2004;; Global Youth í Insignificant
Tobacco Survey data from By gender By gender
26 of 28 states and two of Women Women
the seven Union Territories í í
on youth aged 13–15 years. Men Men
Average price data from í í
users and non-users to
construct school/state/
territory level prices
Model
demand;; probit model to
estimate smoking prevalence
equation and Ordinary Least
Squares to estimate smoking
intensity equation
20 LMICs (Kostova et al., Method í Local brand prices Local brand prices **P
2010 1999–2006;; Data from í WRí í WRí ***P
the Global Youth Tobacco Foreign Brand Prices Foreign brand prices
Survey, average age í WRí í WRí
14 years (n = 349 930). Price By smoking intensity
data from the Economist Very light smokers
Intelligence Unit World Cost (1–15 cigs/mo)
of Living Survey. 0.396*** to 0.528***
Model
In the early 1990s, two studies

were published in the US (Chaloupka,
1991 and Wasserman et al., 1991) and
one in the United Kingdom (Townsend
et al., 1994) that cast doubt on the
Comments
early studies by Lewit and colleagues

by finding that young people were
not more price-elastic than adults.
Chaloupka (1991) used data from
Price elasticity of smoking
the National Health and Nutrition

Examination Survey conducted from
1976 to 1980 to estimate the cigarette
demand equations. Applying Becker
and Murphy’s theoretical model
intensity
of rational addiction (discussed in

Chapter 4), Chaloupka found young
adults (aged 17–24) to be insensitive
to changes in price, whereas
Price elasticity of smoking
individuals aged 25–64 showed a

Light to medium smokers
í WRí
significant long-run response to a

í WRí
í WRí
(100–300 cigs/mo)
change in price, with an estimated

íFLJVPR
Medium smokers
Heavy smokers
(300+ cigs/mo)
long-run price elasticity of demand in

prevalence
the range of – 0.46 to – 0.32.

Wasserman and his colleagues
(1991) used data on 1891 adolescents
aged 12–17, taken from the Second
National Health and Nutrition
Total price elasticity of
Examination Survey conducted

between 1976–1980, and data on
adults taken from several waves
of the National Health Interview
Surveys conducted in the 1970s
demand
and 1980s to estimate cigarette

demand equations. They focused
on the effects of cigarette prices
and control for the prevalence
demand: smoking prevalence
anti-tobacco media outreach,

local anti-smoking sentiment
estimated with a generalized
and compliance with youth

fixed effects, a measure of
linear model with a normal
while controlling for state policies

distribution and a log link.
Analyses include country
model, smoking intensity
of cigarette advertising,
restricting smoking in public places

estimated using a logit
Methods and Model
access restrictions.
and a host of other socioeconomic

and demographic variables.
Consistent with Chaloupka’s (1991)
results, Wasserman and colleagues
found no statistically significant
difference in the price elasticity of
demand between youth and adult
20 LMICs (Kostova et al.,
smokers. Moreover, there were no

Publication (location,
statistically significant effects of price

on youth smoking. They attribute
much of the difference between
author, year)
2010 (contd)
their findings and those of previous

studies (i.e. Lewit and colleagues) to
the inclusion of a measure reflecting
221
the magnitude of restrictions on equations by demographic elasticities of demand obtained were
smoking in public places and its characteristics including gender ±±DQGíUHVSHFWLYHO\
correlation with cigarette prices. and race/ethnicity. While price Indeed, the estimated total price
When Wasserman et al. excluded was found to have a negative and elasticity of demand for young
the smoking restrictions measure significant impact on both male DGXOWVLVPRUHWKDQGRXEOHWKHí
from their models, they obtained and female smoking prevalence, estimate of the total price elasticity of
similar price elasticities to those the magnitudes of the price effects demand for all respondents.
from the early studies. They argued were very different. Chaloupka and In a follow-up paper, Farrelly
that the price elasticity estimates Pacula found that the prevalence and his colleagues (2001) employed
were subject to an omitted variables price elasticity of demand for young the same data as Farrelly and Bray
bias when the smoking restrictions men was nearly twice as large, in (Centers for Disease Control and
measure was not included. absolute value, as that for young Prevention, 1998), but included state-
A series of papers starting in women. Specifically, they estimated specific effects in their model to
the mid-1990s supported the early that the prevalence price elasticity of control for state-level, time-invariant
findings of Lewit and his colleagues GHPDQGIRUPDOHVZDVíDQGIRU heterogeneity, such as sentiment
that young people are more price- IHPDOHV ZDV í 0RUHRYHU WKH\ towards tobacco. The findings from
responsive than older people. These found that smoking prevalence for these analyses are similar to those
studies were based on data covering African American adolescents was of Farrelly and Bray;; total price
a period with greater variation in PRUH SULFH HODVWLF í WKDQ IRU elasticity of demand was estimated
tax and price, which allowed more ZKLWHDGROHVFHQWVí WR EH í IRU \RXQJ DGXOWV DJHG
precise estimates of the impact of Chaloupka and Wechsler (1997) ± DQG í IRU LQGLYLGXDOV
price. For example, Chaloupka and used data extracted from the 1993 aged 25–39. However, the authors
Grossman (1996) used data on more Harvard Alcohol Study to estimate found no significant effects of price
than 110 000 8th-, 10th-, and 12th- the effects of cigarette prices and on smoking prevalence or smoking
grade students taken from the 1992, restrictions on cigarette smoking intensity among individuals aged 40+.
1993 and 1994 Monitoring the Future among college students in the US. Lewit and his colleagues (1997)
Surveys (MTFS) to estimate smoking The authors estimated a total price used data from cross-sectional
prevalence and smoking intensity elasticity of demand of –1.11 for surveys of 9th-grade students in
demand equations. They calculated a college students. The authors also 1990 and 1992 in the 21 US and
total price elasticity for youth smoking concluded that relatively stringent Canada sites that were a part of
RIí+RZHYHULQFRQWUDVWWR/HZLW restrictions on smoking in public the Community Intervention Trial
and his colleagues’ finding that most places reduce smoking prevalence for Smoking Cessation (COMMIT),
of the impact of a price change was rates among college students, while to examine the impact of cigarette
on smoking prevalence, Chaloupka some restrictions on public smoking prices on youth smoking prevalence
and Grossman (1996) found that the reduce the quantity of cigarettes and intentions to smoke. The authors
effect was about evenly split between smoked. estimated that the prevalence price
prevalence and number of cigarettes A Centers for Disease Control and elasticity of demand for 9th graders
consumed by young smokers. This Prevention (CDC) report authored ZDV í DQG WKDW LQWHQWLRQV WR
study controlled for other smoking- by Farrelly and Bray (Centers for smoke in the future among non-
related interventions, such as Disease Control and Prevention, smoking 9th graders were somewhat
restrictions on smoking in public 1998) used data from the 1976– more price elastic, with an estimated
places and schools, as well as youth 1980, 1983, 1985, and 1987–1993 HODVWLFLW\ RI í 7KH DXWKRUV DOVR
access restrictions, and refuted the National Health Interview Surveys to found strong gender differences,
argument by Wasserman et al. that estimate cigarette demand equations with the estimated prevalence price
previous studies reporting price in the United States for various elasticities of demand for boys and
effects were subject to an omitted subpopulations. In their analyses JLUOV HVWLPDWHG DW í DQG í
variables bias. they estimate separate demand respectively. The estimated price
Using the same MTFS data equations for different age groups, coefficients in the girls equations were
as Chaloupka and Grossman including young adults aged 18–24 not significantly different from zero.
(1996), Chaloupka and Pacula years, adults aged 25–39 years, and Harris and Chan (1999) studied
(1999) estimated separate demand adults aged 40+ years;; the total price the relationship between cigarette
222
smoking and price among individuals of cigarette use among smokers. 9th- to 12th-grade students;; and (3) the
aged 15–29 using data from the 1992– The authors found that increases in Vital Statistics Natality Detail Files
1993 Tobacco Use Supplements to cigarette prices lead to significant (Natality) of mothers during pregnancy.
the Current Population Survey. They reductions in both the number of They found consistent evidence that
partitioned their sample into five age young adults who smoke and the youth smoking responds to changes
groups: 15–17 years, 18–20 years, intensity with which they smoke. in cigarette prices, particularly among
21–23 years, 24–26 years and 27–29 The estimated total average price older adolescents. The prevalence
years. The authors found that the elasticity of demand was found to SULFHHODVWLFLWLHVUDQJHGIURPíLQ
price elasticity of smoking prevalence EHíZLWKSULFHKDYLQJDVPDOOHU WKH1DWDOLW\GDWDWRíLQWKH<5%6
declined in absolute value with age: effect on smoking prevalence than data, with the most reliable estimate
í IRU DJHV ± í IRU on average consumption by smokers. RIíFRPLQJIURPWKH±
DJHV ± í IRU DJHV ± Almost all the research MTFS data. Based on their estimated
íIRUDJHV±íIRUDJHV conducted over the last decade elasticities, Gruber and Zinman
27–29. However, the price elasticity confirms the early findings of conclude that the “Marlboro Friday”
of smoking intensity rose in absolute Lewit and colleagues of an inverse price reductions in 1993 explained 26
YDOXHZLWKDJHXSWRDJHí relationship between price response percent of the rise in youth smoking
IRU DJHV ± í IRU DJHV ± and age. For example, using a single observed in the US during the mid-
í IRU DJHV ± í IRU cross-section collected in 1996 for 1990s. “Marlboro Friday” refers to
DJHV ± í IRU DJHV ± The Study of Smoking and Tobacco Friday, 2 April 1993, when Philip
This is consistent with the evidence Use Among Young People, Ross Morris announced a 20% price
from uptake studies (discussed and Chaloupka (2003) examined the reduction to their Marlboro cigarette
below), which shows that price has effect of cigarette prices on smoking brand to fight off generic competitors.
less of an impact on people at the among high-school students in the Tauras and colleagues (2005)
earliest stages of smoking uptake United States. In their analyses, the investigated the impact of cigarette
and has a more significant effect on authors assessed the use of several prices and tobacco control policies
those at later stages who are more alternative measures of cigarette on youth and young adult smoking
established smokers. Nevertheless, prices, including average state prices prevalence and intensity of cigarette
the total price elasticity of demand and perceived prices among the use among smokers during the
IHOO LQ DEVROXWH YDOXH ZLWK DJH í students, while controlling for both late 1990s through early 2000s,
IRU DJHV ± í IRU DJHV ± state-level smoke-free air laws and which was a period characterized
í IRU DJHV ± í IRU youth-access laws. In their preferred by significant changes in cigarette
DJHV ± í IRU DJHV ± specifications, they estimated total prices and taxes. They employed the
The estimated total price elasticities SULFHHODVWLFLWLHVRIGHPDQGRIí first five waves of data (1997–2001)
of demand in Harris and Chan (1999) when using average state prices and from the National Longitudinal Survey
confirm that an inverse relationship í ZKHQ XVLQJ SHUFHLYHG SULFHV of Youth 1997 Cohort (NLSY97).
exists between age and the absolute among youth. A subsequent analysis Controlling for unobserved year and
value of the total price elasticity of by Ross and Chaloupka (2004) individual characteristics, they found
cigarette demand. using the same data, but explicitly a strong negative impact of cigarette
Tauras and Chaloupka (1999) controlling for the level of compliance prices and taxes on young people’s
employed data on cigarette use from with respect to youth-access laws, smoking prevalence and intensity
the 1976 through 1995 longitudinal resulted in similar price elasticity of cigarette use among smokers.
surveys of high school seniors as estimates. Specifically, they estimated the total
part of the Monitoring the Future Gruber and Zinman (2001) price elasticity of cigarette demand
programme. Individuals from each controlled for both state and year WREHí7KHHVWLPDWHGVPRNLQJ
cohort between 1976 and 1993 were fixed effects in their analyses of prevalence price elasticity of demand
tracked over time, with a maximum youth smoking. They employed three and the elasticity of smoking intensity
number of seven follow-up surveys separate data sets from the 1990s ZHUHíDQGíUHVSHFWLYHO\
conducted on individuals. An in their analyses: (1) Monitoring the Sloan and Trogdon (2004) used
individual fixed effects model was Future Surveys (MTFS) of 8th-, 10th-, data from the Behavioural Risk Factor
used to estimate the determinants and 12th-grade students;; (2) Youth Surveillance System from the 1990s
of smoking prevalence and intensity Risk Behavior surveys (YRBS) of and early 2000s to estimate smoking
223
prevalence equations among young using alternative approaches to cross-sectional and fixed effects
adults (aged 18–20 years) and older dealing with antismoking sentiment. approaches. Importantly, using
adults (aged 21 years+). Employing The strong negative impact of price DeCicca and colleagues’ measure
both state and year fixed effects, on average smoking was robust of antismoking sentiment, they found
the authors concluded that young to all the methods of dealing with a strong negative effect of tax on
adult smoking prevalence was the unobserved state-level sentiment smoking prevalence among youth,
most price-elastic, with an estimated towards tobacco. Moreover, in all alleviating the concerns raised by
smoking prevalence elasticity of of the models except the model DeCicca and colleagues. Using the
GHPDQG RI í )XUWKHUPRUH WKH that included the new measure of tax effects from the national and
authors found evidence that the sentiment, price was found to have state samples, Carpenter and Cook
absolute value of the price elasticity a negative and significant impact on (2008) estimated price elasticities for
of smoking prevalence declined smoking prevalence among youth. \RXWK VPRNLQJ SUHYDOHQFH RI í
monotonically with age, with the Given the findings when the direct LQ WKH QDWLRQDO VDPSOH DQG í LQ
exception of individuals aged 65 measure of antismoking sentiment the state sample.
years and older. was included in the models, DeCicca To assess the impact of price on
DeCicca and colleagues (2008a) and colleagues questioned the youth smoking based on the intensity
developed a direct measure of state- adequacy of other proxies to control of smoking, Liang and Chaloupka
specific smoking sentiment using for antismoking sentiment. However, (2002), used data from the 1992,
a factor analysis procedure that some caution should be used in 1993 and 1994 Monitoring the Future
employed data from the Tobacco interpreting models that include surveys of 8th-, 10th-, and 12th-grade
Use Supplements to the Current direct measure of antismoking students in a Threshold of Change
Population Surveys during the sentiment, as there is likely to be model. The authors grouped the
1990s. They merged this tobacco reverse causality in this type of youth into five categories of smoking
sentiment measure with survey data estimation strategy. That is, the intensity including: no consumption;;
on youth smoking from the 1992 amount of smoking within a state is less than daily smoking;; light daily
and 2000 waves of the National likely to have an impact on the level smoking (defined as one to five
Education Longitudinal Study of antismoking sentiment within a cigarettes per day);; moderate daily
(NELS). The authors found price to state, resulting in simultaneity bias. smoking (defined as one-half pack
have a strong, significant negative Carpenter and Cook (2008) per day);; and heavy daily smoking
impact on smoking prevalence and addressed the concerns of DeCicca (defined as one pack or more per
intensity of cigarette use by young and colleagues (2008a) in a recent day). The authors found that higher
smokers. Between 1992 and 2000, paper that used national, state, prices were associated with lower
the estimated price elasticities of and local Youth Risk Behavior smoking in all cases, but that the
smoking prevalence and intensity Surveillance System data from 1991 greatest impact of the higher prices
of cigarette use among smokers to 2005. The authors tested three was on smoking at the heaviest
UDQJHG IURP í WR í DQG alternative methods of dealing with levels of intensity.
IURP í WR í UHVSHFWLYHO\ antismoking sentiment. First, they
Moreover, even after controlling for estimated a cross-sectional model Smoking prevalence and intensity
the new direct measure of smoking that relied on intrastate variation in among young smokers – Other
sentiment, price was found to have cigarette taxes to identify the impact high-income countries
a strong negative influence on the of price on youth smoking. Second,
intensity of cigarette use by young they estimated a fixed effects model An early study from the United
smokers in the 2000 cross-section. that controlled for area (i.e. state Kingdom by Townsend and colleagues
However, when the direct measure of or local municipality) fixed effects (1994) used data on cigarette smoking
smoking sentiment was included in and year fixed effects. Finally, they from the 1972–1990 British General
the smoking prevalence equations, employed the same direct measure Household Surveys to assess the
the price effects lost statistical of antismoking sentiment used by effects of cigarette prices on smoking
significance. Using the 2000 wave DeCicca and colleagues (2008a). prevalence by gender and age and by
of data, they tested models that used They found consistent evidence of a gender and socioeconomic group in
this newly developed measure of significant negative effect of cigarette the United Kingdom. They concluded
sentiment and compared it to models taxes on smoking prevalence in the that smoking prevalence among
224
young females (aged 16–19 and aged in School-Aged Children (HBSC) greater reductions in the younger age
20–24) was responsive to changes study. In contrast to the US studies, groups (22%;; 16–24 years), largely
in cigarette prices, whereas young the authors found no significant attributable to a 92.7% increase in the
males’ smoking prevalence was not. relationships between price and real price of cigarettes. A reduction in
The estimated price elasticities of youth smoking prevalence in their smoking prevalence explained 40%
smoking prevalence for females aged multilevel analyses of these merged of this decrease in consumption, and
± DQG DJHG ± ZHUH í data. The researchers suggest that the greatest reduction in smoking
DQGíUHVSHFWLYHO\,QWHUHVWLQJO\ the lack of relationship between price prevalence was observed among
the authors concluded that there were and prevalence may result from their the younger cohort (aged 16–34),
no statistically significant differences inability to adequately control for suggesting that youth are more
in the estimated price elasticities of inter-country factors correlated with responsive to price than adults. Karki
demand among women in these two price and youth smoking and their and colleagues (2003) estimated the
age groups and women who were inability to capture the full impact of joint demand for cigarettes and bidis
aged 25–59 years. recent changes in taxes and prices. by age in Nepal. They found that
A Canadian study by Waller young people (aged 15௅24 years)
and colleagues (2003) used data Smoking prevalence and intensity were more than twice as responsive
from the Ontario Student Drug Use among young smokers – Low- to changes in price than the overall
Survey to examine the impact of and middle-income countries: population, and that in general the
decreases in Canadian cigarette cross-sectional studies absolute value of the price elasticity
taxes (and consequently prices) in of demand fell with increasing age.
the early 1990s on youth smoking There are a limited but growing Kyaing (2003) estimated price
prevalence. They found that youth number of studies that investigate elasticities of smoked tobacco
smoking prevalence had been falling the response to price and tax products in Myanmar. The price
steadily in the years leading up to the changes among youth in low- and elasticity estimates for young people
tax decrease, despite the increase in middle-income countries. With were estimated to be approximately
cigarette smuggling into Canada in the few exceptions, the findings from 50% greater than the estimated price
late 1980s and early 1990s. However, these studies are consistent with elasticity for the overall population.
the tax cuts in 1994 led to a significant those from high-income countries: Several studies from low- and
rise in youth smoking prevalence. young people are more price-elastic middle-income countries have
Subsequent price increases led to than the population in general. For made use of Global Youth Tobacco
further reductions in prevalence. example, Krasovsky and colleagues Survey data, which collects uniform
Dupont and Ward (2002), using the (2002) estimated differences in the data from school-aged children
National Population Health Data price elasticity of cigarette demand across several countries. Ross
for the period 1994–1998, reached by age and income in the Ukraine. (2004a) estimated cigarette demand
a similar finding for the effects of The estimated price elasticities equations for students in the Ukraine.
Canadian tax increases in the mid- of average smoking for younger Using a two-part model of cigarette
1990s. They estimated that the price smokers were found to be larger in demand, Ross estimated the price
elasticity of smoking prevalence for absolute value than the estimated elasticities for smoking prevalence
&DQDGLDQ\RXWKZDVí elasticities for older smokers at WREHLQWKHUDQJHRIíWRí
One recent cross-sectional study each income level. van Walbeek with considerably higher estimates
by Schnohr and colleagues (2008) (2002) used regression techniques of smoking intensity ranging from
pooled data from 27 European on repeated cross-sectional surveys í WR í ,Q DGGLWLRQ 5RVV
countries to examine the effects of of between 14 000 and 30 000 (2004b) estimated the price elasticity
prices and tobacco control policies individuals aged 16 and over to of demand for students in Moscow,
on youth daily smoking prevalence. estimate the percentage change in Russian Federation. The average
Data on prices and policies were aggregate cigarette consumption estimated price elasticity of demand
obtained from the 2003 WHO and smoking prevalence by age in for tobacco among Moscow youth in
Regional Office for Europe Tobacco South Africa. Van Walbeek observed WKLVVWXG\ZDVíDQHVWLPDWHZHOO
Control Database and were merged a 26% decrease in aggregate above those produced in the limited
with data on smoking prevalence consumption and a 17% decrease in studies of the impact of price on adult
from the 2001/2002 Health Behaviour estimated smoking prevalence, with smoking in the Russian Federation.
225
Recent work by Kostova et al. of young Chinese men, with a total Smoking initiation
(2010) used data on 349 930 youths SULFH HODVWLFLW\ HVWLPDWH RI í Cross-sectional studies
(average age 14 years) from 20 low- which becomes insignificant when
and middle-income countries taken including community fixed effects. Douglas and Hariharan (1994)
from Global Youth Tobacco Surveys Total elasticity estimates for age were among the first to model
conducted between 1999 and 2006. interactions show the same pattern smoking initiation (see Table 6.2).
They estimated a total price elasticity across the Chinese and Russian They employed cross-sectional
RI í IRU FLJDUHWWH DQG IRXQG WKDW samples: the elasticity falls and then data from the 1978 and 1979
price is an important determinant rises with age. The authors suggested National Health Interview Survey:
of both smoking prevalence and that higher elasticity estimates for Smoking Supplement, and used
smoking intensity, with estimated the younger and older cohorts may a split population duration model
SULFH HODVWLFLWLHV RI í DQG í reflect the effect of prices on initiation to estimate the probability that an
respectively. Finally, Joseph (2010) and cessation decisions. individual would start smoking.
estimated the impact of price on youth To summarize, the small but They concluded that increases
tobacco demand using data from 73 growing number of studies from in cigarette excise taxes had no
356 Indian youths aged 13–15 from low- and middle-income countries influence on individuals’ decisions to
the Global Youth Tobacco Survey is generally consistent with studies start smoking. In a follow-up paper,
conducted between 1999 and 2006. conducted in high-income countries— Douglas (1998) used one year of
This study estimated a total price price has a negative impact on cross- sectional data, the 1987
elasticity of demand for cigarettes consumption among young people, National Health Interview Survey:
RI í ZLWK SULFH LQIOXHQFLQJ WKH and the young are more price-elastic Cancer Risk Factor Supplement, to
decision to smoke more strongly than than the population in general. investigate the determinants of the
the intensity of smoking. Similar to decisions to start and quit smoking
studies from the USA, Joseph (2010) Smoking transitions among young in the context of an economic model
found that price had a greater impact people of addiction. He estimated several
on smoking prevalence among girls alternative parametric duration
íWKDQDPRQJER\Ví Many researchers examining the models in his assessment of smoking
influence of price on adolescent initiation. Douglas concluded that
Longitudinal studies smoking prevalence believe that current, future, and past prices of
much of the effect of price reflects the cigarettes had an insignificant effect
Lance et al. (2004) also estimated impact of price on smoking initiation. on the probability of initiation. The
the price elasticity of demand for Similarly, many researchers believe effects of price on smoking cessation
cigarettes in the Russian Federation;; the effects of price on youth and adult are discussed in detail in Chapter 5.
however, they found elasticity smoking is dominated by the effect Forster and Jones (2001) used
estimates smaller than those of Ross of price on smoking escalation and cross-sectional data to investigate
(2004b). They used three waves of cessation. As discussed in the next the determinants of smoking
longitudinal data on Russian men two sections, several studies have initiation and cessation in the United
aged 13 and over from the Russian attempted to directly quantify the Kingdom. In particular, Forster and
Longitudinal Monitoring Survey effect of price on smoking initiation Jones (2001), using data from the
conducted between 1992 and 2000. among youth (Table 6.2) and the British Health and Lifestyle Survey,
Among young Russian men aged impact of price on smoking escalation relied on retrospective information
13–19, the total price elasticity of and cessation among young people on cigarette smoking to approximate
GHPDQG ZDV LQ WKH UDQJH RI í (Tables 6.3, 6.4). The earliest of the length of time each individual
DQGí,QWKHVDPHVWXG\/DQFH these studies on smoking transitions smoked and abstained from
and colleagues examined three relied on cross-sectional data with smoking. They found cigarette taxes
waves of longitudinal household data retrospective information on smoking. to be a statistically significant but
on Chinese men aged 13 and over Many of the more recent studies have relatively weak determinant of youth
from the China Health and Nutrition used longitudinal data that tracks an smoking initiation. The effects of tax
Survey conducted in 1989, 1993 and individual’s smoking behaviour and on smoking cessation are discussed
1997. Similarly, they found relatively other determinants over time. in detail in Chapter 5.
small price effects for the sample
226
Table 6.2. Summary of studies on the effect of cigarette price on smoking initiation among young people
Methods (location,
Publication (author, time period;; study
Model Endpoint Main results Subpopulation Comments
year) design and sample
size (M/F)
High- income countries
USA 1978/79;; National Split population duration Price elasticity of age Price 18 Includes two price
(Douglas and Hariharan, Health Interview Survey: model of the decision of smoking initiation (% –0.014 (SE 0.229) variables: ‘Price 18’
1994) Smoking Supplement to start smoking, with change in the probability Price 15–18 which is the log real
(n = 10 219). Price ‘failure’ defined as the of ever smoking 0.001 (SE 0.312) cigarette price, when
data from the Tobacco decision of a never associated with a 1% Price 18 the respondent is age 18
Institute (1992) smoker to start smoking change in price variable) 0.009 (SE 0.028) and ‘Price 18–15’ which
Price elasticity of age Price 15–18 is the change in log real
of smoking initiation 0.001 (SE 0.038) cigarette price between
(% change in the age ages 15 and 18.
of starting smoking
associated with a 1%
change in price variable)
USA 1987;; National Health Ordered probit split- Smoking initiation Without controlling for Controlling for state Findings for the impact of
Douglas, 1998) Interview Survey: sample duration model hazard elasticity (% state regulation regulation price on starting hazard
Cancer Risk Factor with lagged duration change in the probability Past price Past price using the split population
Supplement (n = 8754) dependence and of starting smoking in 0 0.60 (SE 0.53) 0.55 (SE 0.51) model are statistically
time-varying covariates a given time period, Present price Present price insignificant.
(cigarette price and conditional on being a –0.30 (SE 0.70) –0.30 (SE 0.75)
regulation) non-smoker until then, Future price Future price
for each % change in –0.58 (SE 0.53) –0.57 (SE 0.54)
price)
USA 1988–1992;; Data from 3 Discrete time hazard Price elasticity of White Youth ** Statistically significant
(DeCicca et al., 2000) waves (1988, 1990 and model of smoking onset smoking initiation 0.000053
1992) of the National Hispanic Youth
Education Longitudinal –0.0086**
Survey (n = 10 893). African-American
Price data from the Youth
Tobacco Institute (1993) –0.0059
USA 1991–1993;; Three Discrete-time hazard Price elasticity of No State-fixed effects With State-fixed * Significant at 1% level
(Tauras et al., 2001) cohorts of nationally model using weighted smoking initiation Any smoking effects
representative dichotomous probit –0.271 Any smoking
longitudinal data on equations to estimate Smoking at least 1–5 –0.111
students in 8th, and 10th the probability that an cigarettes per day Smoking at least 1–5
grades in 1991, 1992, individual starts smoking –0.811* cigarettes per day
1993 with 2–3 waves in a given time period, Smoking at least ½ pack –1.23*
for each cohort from the conditional on being a per day Smoking at least ½ pack
Monitoring the Future non-smoker at the start –0.955* per day
Survey. Price data from of the time period. –1.43*
the Tobacco Institute
227
228
Methods (location,
size (M/F)
USA 1988–1994;; panel micro Ordered probit model Price elasticity of Cross-sectional This study finds weak or
(DeCicca et al., 2002) data from the National using cross-sectional smoking prevalence results non-existent tax effects
Education Longitudinal data Price elasticity of 10 th grade in models of the onset of
Survey of 1988 which Discrete time hazard smoking onset between –1.31 smoking between 8th and
surveys eighth graders model using panel data 8th and 10th grade or 8th 12 th grade 12th grades, models of
with follow-up 2 and 4 Analyses controlled for and 12th grade –0.72 onset of heavy smoking
years later (n = 13 989) socioeconomic status, Onset between 8 th and between 8th and 12th
school and parent 10 th grade grades, and discrete

characteristics, as well From –1.3 to –0.9 time hazard models
as three measures of Onset between 8 th and that include state fixed
smoking-related state 12 th grade effects.
legislation. From –0.72 to –0.46 (not
statistically significant)
USA 1979–1994;; Cross-sectional analysis Tax elasticity of later 1984 (age 19–28) Women * P
(Glied, 2002) representative microdata smoking initiation –0.74* 1984
from the National 1992 (age 27–35) –0.87
Longitudinal Survey 1.10 1992
of Youth. Youth aged 1994 (age 29–37) 1.55
14–23 in 1979 followed 1.34 1994
up annually (n = 2295). 1.03
Price variable represents Men
the mean price of 1984
cigarettes when the –1.22
respondent was 14 1992
0.29
1994
1.23
Low income
1984
–1.35
1992
1.05
1994
1.75
USA 1997–2000;; four Discrete time duration Price elasticity of Males Price was found to be an Uses two measures of
(Cawley et al., 2004) waves of the National method, using a probit smoking initiation Less stringent smoking insignificant determinant smoking initiation: less
Longitudinal Survey specification to estimate initiation of smoking initiation stringent, transition
of Youth: nationally the hazard rate –0.86 among females from non-smoker to
representative, More stringent smoking any positive quantity of
longitudinal data on initiation cigarettes, and more
youths aged 12–16 as of –1.49 stringent, transition from
the first wave (n = 9022). non-smoker to frequent
Price data from the smoker defined as
Tobacco Institute smoking at least 15 of
last 30 days
Methods (location,
size (M/F)
USA 1988–2000;; Children Simple latent variable Price elasticity of Males Price was found to be an
(Cawley et al., 2006) born to female model, controlling for smoking initiation –1.20 insignificant determinant
respondents of the smoke-free air laws, of smoking initiation
National Longitudinal youth access laws, among females
Survey of Youth, 1979 socioeconomic and
Cohort demographic factors
USA 1988–2000;; Pooled data Discrete time hazard Price elasticity of No State-fixed effects *** Significant at 1%
(DeCicca et al., 2008a) from four waves (1988, model, with a dependent smoking initiation –0.0015*** (two-tailed tests)
1990, 1992 and 2000) of variable representing the State-fixed effects
the National Education conditional probability –0.0005
Longitudinal Survey of starting to smoke in a
(n = 37 937) given period, assuming
the individual was a non-
smoker at the start of the
period
USA 1988–2000;; 1988, 1990, Myopic addiction Price elasticity of Model 1 Model 3 * Statistically significant
(DeCicca et al., 2008b) 1992 and 2000 waves of model using probit smoking initiation –0.17* Movers
the National Education specifications, Model 2 1.69
Longitudinal Study controlling for 0.08* Stayers
(n = 10 336) including socioeconomic and –0.7
youths aged 18–26 demographic factors and
years. anti-tobacco sentiment.
Three models estimated:
1) using intrastate
variation in cigarette
excise taxes, 2) including
a direct measure of anti-
tobacco sentiment, and
excise taxes faced by
youths who moved state
between study waves
United Kingdom 1984;; Cross-sectional Split-population duration Tax elasticity of the age Men
(Forster and Jones, data from the British model, using probit of starting smoking 0.164
2001) Health and Lifestyle specification yields to Women
Survey;; representative model the likelihood of 0.081
sample of adults 18+ smoking initiation
years living in England,
Scotland and Wales in
1984 with retrospective
smoking data. (n = 9003)
Spain 1993, 1995, 1997;; Hazard functions for Price elasticity of time to Men Survival and hazard
(López Nicolás, 2002) Pooled, nationally duration to smoking smoking initiation 0.069 (2.68) functions compared to
representative cross- initiation are estimated (% change in time Women those from a Kaplan–
sectional data from the using a log logistic split starting at the mean age 0.076 (2.60) Meier estimate of the
Spanish National Health. population model of smoking initiation for survival function and
229
230
Methods (location,
size (M/F)
Spain Survey with retrospective each% change in price) a standard log logistic
(López Nicolás, 2002) information on smoking. function. Results from
(contd) Price data from National preferred specification
Statistics Office presented here.
t-statistics in
parentheses
France 1999;; Cross-sectional Using a life course Life course smoking Cigarette price has had Life course perspective
(Peretti-Watel, 2005) nationally representative perspective, examined prevalence and no or little impact on cannot separate the
data for individuals smoking cessation by cessation by cohort regular smoking initiation effects of price from
aged 12–75 from distinguishing 6 cohorts in France. other factors that vary
the French Health of individuals: Born across cohorts (tobacco
Barometer Survey with 1969–1978, 1959–1968, control policy, cultural
retrospective smoking 1949–1958. Each cohort trends, etc.)
data (n = 13 685) experienced a different
pattern of price changes
during the critical
period just preceding
and proceeding regular
smoking initiation.
Canada 1994;; nationally Maximum likelihood Point elasticity of Early initiation Early initiation, defined
(Auld, 2005) representative data estimation controlling smoking prevalence –1.268 (SE 0.633) as respondents who
from the Youth Smoking for year of birth, self- with respect to Prevalence in late had smoked one whole
Survey on youths aged reported health status, contemporaneous prices adolescence if early cigarette every day for
15–19 (n = 9139). education, employment, changes initiator seven consecutive days
Province-specific social interactions and 0.012 (SE 0.031) and were 14 or younger
tobacco product price perceived non-pecuniary Prevalence in late when they started such
indexes deflated by CPI costs of smoking adolescence if not smoking behaviour. The
as reported by Statistics early initiator sample includes youths
Canada were used. –0.615 (SE 0.100) who have dropped out of
school
Canada 1994 to 1997;; Multivariate logistic Price elasticity of 3.36 This study examines the
(Zhang et al., 2006) longitudinal data from regression analysis smoking initiation (95% CI: 0.07 to 6.75) impact of decreasing
Statistics Canada’s using bootstrap taxes. Elasticity estimate
National Population weights controlling for should be interpreted
Health Survey for young sociodemographic and as the change in the
adults aged 20–24 who tobacco control variables probability of initiating
were non-smokers at smoking associated
baseline (n = 636) with a 1% decrease in
cigarette tax
Canada 1992 to 1996;; Probit and OLS Tax elasticity of smoking –0.20 to –0.50
(Sen and Wirjanto, 2010) longitudinal data from estimation controlling for initiation and persistence
the Waterloo Smoking gender and employing
Prevention Project province and year effects
Methods (location,
size (M/F)
Canada (WSPP) on students in
(Sen and Wirjanto, 2010) grades 8–12 (n = 591)
(contd)
Australia 1990;; Cross-sectional Split population hazard Price elasticity of 27–37 years 18–26 years **P
(Kidd and Hopkins, data from the National models smoking initiation Men Men *P
2004) Health Survey 1990 and 0.16** 0.11 Price appears to be a
1998 Women Women significant determinat of
aged 27–37 (n = 9402) 0.12* 0.14* smoking for 27–37 year
with retrospective data olds in 1990;; however,
on smoking. Price data sensitivity analysis
from the Australian across age groups
Bureau of Statistics and data source
raise question of the
robustness of these
findings
Viet Nam (Laxminarayan 1992–93 and 1997– Multinomial logit model Price elasticity of Cigarettes While the price elasticity
and Deolalikar, 2004) 98;; data on 4300 to estimate the effect of smoking initiation –1.175* of initiation is estimated
households and 17 780 both individual attributes Rustic tobacco for the entire population
individuals aged 15 and and choice attributes –1.558 in the survey, much of
older from two waves (related to cigarettes and the initiation is likely
of the Viet Nam Living rustic tobacco) on the to have occurred
Standards Survey. Price decisions of individuals among relatively young
data from commune- to initiate use individuals. Estimates
level surveys suggest tat cigarettes
and rustic tobacco are
economic substitutes,
but estimates are not
statistically significant.
*P
231
Several more recent studies and quit smoking. Given that it is a determinants of smoking initiation
from high-income countries have more recent outcome, the recall bias by individuals of different race and
used retrospective data on smoking should be smaller when looking at ethnicities using data extracted from
collected in nationally representative cessation than initiation. Indeed, the 1988, 1990 and 1992 National
surveys of adults to look at the most of the aforementioned studies Educational Longitudinal Survey
impact of prices on smoking find evidence that higher cigarette (NELS). After controlling for state and
initiation. In particular, studies have prices significantly increase the year fixed effects, they found price to
been conducted in Australia (Kidd & probability of smoking cessation (and have a dramatic negative impact on
Hopkins, 2004), France (Grignon & consequently reduce the duration of smoking initiation decisions among
Pierrard, 2002;; Peretti-Watel, 2005), smoking). Second, studies that rely Hispanics and African Americans,
Spain (López-Nicolás, 2002), and on cross-sectional data likely suffer while having no influence on
Germany (Göhlmann, 2007). These from a price-matching problem. The smoking initiation decisions among
studies produced mixed findings on smoking transition studies discussed Whites. The authors estimate that
the effects of price on initiation, while above are based on previous prices/ a price increase of $1.50 would
generally finding that higher taxes taxes that an individual would have decrease smoking initiation rates by
and prices significantly increase paid for cigarettes in the individual’s Hispanics and African Americans
the likelihood that adult smokers current location of residence. If to approximately 1%. However, the
quit (see Chapter 5 for discussion an individual lived in a different authors caution that the African
of cessation results). In addition to location in the past, that individual Americans prediction is based on
the aforementioned measurement would have been matched with an insignificant price coefficient
error problems that are inherent the incorrect cigarette price/tax. estimate. These estimates should
in using retrospective data to look Moreover, a price-matching problem be viewed with caution because the
at smoking transitions, the lack of will exist if there is imperfect recall on use of state fixed effects relies on
variation in price over time in many the timing of initiation or cessation. within-state variation in price over
of these countries adds an additional That is, the assumed prices at the time. During the short time period
challenge to estimating price effects. time of initiation and cessation will under investigation, there may be
Lastly, Madden (2007) used be incorrect due to imperfect recall. insufficient within state variation in
cross-sectional data from the The measurement errors in both the price variable.
Saffron Survey, with retrospective the dependent and independent DeCicca et al. (2008a) used
information on Irish women, to variables will bias estimates of the data from the 1988, 1990, 1992 and
examine the effects of cigarette taxes effect of prices on the smoking 2000 waves of the NELS to examine
on smoking initiation and cessation. transitions towards zero. the influence of cigarette prices on
With regard to smoking initiation, smoking initiation decisions among
Madden (2007) found that an inverse Smoking initiation adolescents and young adults. The
relationship exists between cigarette Longitudinal studies authors found price to have a strong
taxes and smoking initiation, with the and significant negative influence on
strongest effect of taxation on Irish Several recent econometric studies smoking initiation when state fixed
women with intermediate levels of have employed longitudinal data effects were omitted from the model.
education, and weaker tax effects for to examine the impact of economic However, when state fixed effects
those with the least education. factors such as prices and taxes were included in the regressions,
Previous studies on smoking on smoking initiation and uptake. In the price effects failed to reach
transitions have made significant general, these studies do not suffer significance at conventional levels.
contributions to the literature by from the cross-sectional limitations They concluded that unobserved
modelling the decisions to start and discussed above. state-level heterogeneity (possibly
quit smoking;; however, these cross- In a series of papers, DeCicca anti-smoking sentiment), not price,
sectional studies are subject to at and colleagues (2000;; 2008a;; was driving young people’s smoking
least two potential measurement 2008b) investigated the influence initiation decisions. Further, DeCicca
errors. First, the use of retrospective of price and tax on smoking et al. (2008b) used data from the
information on smoking is likely to initiation among adolescents and 1992 and 2000 waves of the NELS
suffer from errors in recall about young adults. Initially, DeCicca and to examine the influence of cigarette
the age at which individuals began colleagues (2000) examined the excise taxes on smoking initiation of
232
young adults (individuals who start smoking initiation using longitudinal (termed “less stringent initiation”).
smoking between the ages of 18 and data was that of Tauras and The second measure (termed “more
26). They used three identification colleagues (2001). They employed stringent initiation”), reflects the
strategies in their equations. First, data from three cohorts of students transition from not smoking in the
they used intra-state variation in enrolled in 8th and 10th grade in immediate previous wave of data
cigarette excise taxes to identify 1991, 1992 and 1993 as part of to being a frequent smoker in the
the impact of price on smoking the longitudinal component of the current wave of data, as measured
initiation. Second, they included Monitoring the Future project. The by having smoked at least 15 days
the direct measure of anti-smoking authors examined three measures of the past 30 days. Given the notion
sentiment developed by DeCicca et of smoking initiation based on that some youth, particularly girls,
al. (2008a) in their smoking initiation alternative smoking thresholds. The may use cigarette smoking as a
equations. Finally, they used measures of initiation included the weight suppressant, which might
variation in cigarette taxes faced transitions from not smoking any make their smoking behaviour less
by young adults who moved across cigarettes in the immediate previous responsive to prices and tobacco
state lines between 1992 and 2000 wave of data to: smoking any positive control policies, they estimated
versus young adults who remained amount in the current wave;; daily separate models for adolescent
in the same state in these two years. smoking, defined as smoking at least boys and girls. While controlling for
Cigarette taxes were found to have 1–5 cigarettes per day on average in smoke-free air laws, youth access
a significant negative impact on the current wave;; and heavy daily laws, and anti-smoking sentiment
young adult smoking initiation using smoking, defined as smoking at least by a dichotomous indicator for
identification strategy 3 for only a half a pack per day on average in tobacco-producing state residence,
those who remained in the same the current wave. After controlling the authors concluded that male
state in both years. Given this, the for youth access laws and region adolescent smoking initiation was
authors concluded that cigarette fixed effects, the authors found very responsive to changes in
prices have little impact on smoking that the average price elasticity of cigarette prices, with the average
initiation. These results should be initiation based on any smoking, price elasticity of less stringent
viewed with some caution. First, the at least 1–5 cigarettes per day on smoking initiation estimated to
study was conducted on a sample average, and at least ½ pack per day EH í DQG WKH DYHUDJH SULFH
of individuals who initiate smoking RQ DYHUDJH ZHUH ± í DQG elasticity of more stringent smoking
later in life (non-smokers in high –0.96, respectively. These estimates LQLWLDWLRQ HVWLPDWHG WR EH í
school but smokers by modal age imply that youth smoking initiation Female smoking initiation was found
26). In the United States, most adult is indeed responsive to changes not to be significantly related to
ever-smokers initiate smoking well in cigarette prices, with the price cigarette prices, but was found to be
before the age range investigated response being positively related very responsive to weight concerns.
by DeCicca and colleagues (2008b), to higher thresholds of smoking A follow-up paper on youth
and the smoking initiation decisions initiation. smoking initiation by Cawley and
of an older cohort may be quite Cawley and colleagues (2004) colleagues (2006) found very similar
different than those of younger used more recent data to investigate findings to Cawley et al. (2004)
cohorts. Second, anti-smoking the determinants of youth smoking despite using a longitudinal data set
sentiment may be an endogenous initiation in the United States. In that spans a much larger time period.
variable being simultaneously particular, they employed the first Specifically, the authors used data
determined with smoking. Moreover, four waves (1997–2000) of the from the 1988–2000 Children of
in the models that rely solely on National Longitudinal Survey of the National Longitudinal Survey of
intrastate variation in taxes, the Youth 1997 Cohort (NLSY97). In Youth, 1979 Cohort (CoNLSY). After
authors found only weak evidence their analysis, the authors employed controlling for smoke-free air laws
of a negative effect of taxes on two alternative measures of smoking and youth access laws, cigarette
smoking prevalence (i.e. price effect initiation. The first measure indicates prices were found to have a negative
fails to reach 5% significance levels a transition from not smoking in the impact on smoking initiation in
of a two-tailed test). immediate previous wave of data all models that were estimated;;
Among the first studies to to smoking any positive quantity of however, the price coefficients were
examine the impact of price on youth cigarettes in the current wave of data significantly different from zero in
233
only the male equations. The price initiation based on daily smoking in neither experimented with smoking
elasticity of male smoking initiation the 30 days before the survey was nor smoked a whole cigarette and
based on any cigarettes consumed í,QDGGLWLRQWKH\HVWLPDWHGWKH definitely will not smoke next year
ZDV HVWLPDWHG WR EH í 7KH tax elasticity of smoking prevalence and will not smoke a cigarette offered
findings from Cawley and colleagues using various data sources, and to them by a friend, to an established
suggest that the gender-specific found tax elasticities ranging from smoker, defined as individuals who
differences in the impact of price on íWRí have consumed more than 100
initiation may account for the mixed cigarettes in their lifetime. They
findings in the effects of price on Smoking uptake found price to have a significant
smoking initiation contained in some Cross-sectional studies impact on progression from each
of the previous studies. stage to the next, and consistent with
Zhang et al. (2006) explored the Emery et al. (2001) used data Liang and Chaloupka (2002), they
effect of decreasing prices on young from the 1993 Teenage Attitudes found price to have a greater impact
adult smoking initiation. They used and Practices Survey to look at on the likelihood of progressing to
longitudinal data from Canada’s the differential impact of price on later stages of uptake. The authors
National Population Health Survey experimentation with cigarettes suggest that the greater impact of
from 1994 to 1997 to examine (defined as having ever smoked price on later stages of smoking
the impact of decreased cigarette or tried a cigarette, but not having uptake reflects the changes in young
prices, resulting from tobacco tax smoked at least 100 cigarettes in smokers’ sources for cigarettes as
reductions, on smoking initiation their lifetime) and current smoking they progress to higher intensities
among young adults in Canada. After (defined by smoking in the past of smoking. Specifically, smokers
controlling for smoke-free air laws, 30 days) and established smoking at earlier stages of uptake smoke
tobacco control expenditures, and (smoking in the past 30 days and relatively few cigarettes and are
socioeconomic and demographic had smoked at least 100 cigarettes more likely to rely on social sources
factors, the authors found that lower in their lifetime) among 10- through for these cigarettes. As they progress
cigarette prices were significantly 13-year-olds and 14- through to more regular smoking, they begin
associated with a higher likelihood 22-year-olds. They found that price to buy their own cigarettes and are
of initiating smoking. Specifically, did not have a statistically significant more directly influenced by prices.
the authors found that the price impact on experimentation in either Finally, Slater et al. (2007)
elasticity of initiation was -3.36, age group. Price was found to have combined data on cigarette prices
suggesting that a 1% decrease in a significant effect on the likelihood and point-of-sale cigarette marketing
price, increases smoking initiation of being a current smoker, with an collected around schools that
among young adults by 3.36% in HODVWLFLW\RIíDQGWKHOLNHOLKRRG participated in the 1999 through 2003
Canada. of being an established smoker, with Monitoring the Future surveys to
In another study examining DQ HODVWLFLW\ í $OVR SULFH ZDV examine the effects of prices, price-
the impact of decreasing prices in found to have a significant effect related promotions, and advertising
Canada, Sen and Wirjanto (2010) on cigarette consumption among on youth smoking uptake. Using a
used longitudinal data from the current and established smokers, generalized ordered logit equation,
Waterloo Smoking Presentation ZLWK HODVWLFLWLHV RI í DQG í Slater and colleagues found that
Project to explain the impact of a respectively (Table 6.3). price had its greatest impact on later
large reduction in cigarette excise Using the survey entitled The stages of smoking uptake, a finding
taxes in Canada on adolescent Study of Smoking and Tobacco Use consistent with previous studies. In
smoking initiation decisions. The Among Young People conducted addition, they found that point-of-
authors found the tax decrease to in 1996, Ross and her colleagues sale promotions related to price (e.g.
have a significant impact on youth (2006) applied a generalized ordered multipack discounts and other retail
smoking initiation in multivariate logit model to examine the differential value added promotions) had little
regression. In particular, the tax impact of price and tobacco control effect on early stages of uptake, but
elasticity of smoking initiation based policies on the five stages of were strongly associated with later
on any cigarettes smoked in the 30 smoking uptake ranging from a stages beyond experimentation.
GD\V EHIRUH WKH VXUYH\ ZDV í low-risk cognition non-smoker,
and the tax elasticity of smoking defined as individuals who have
234
Table 6.3. Summary of studies providing data on the effects of price on youth smoking uptake
Methods (location,
Model Endpoint Main results Subpopulations Comment
size (M/F)
USA 1993;; Nationally Two-part model of Price elasticity of Experimenters 10 – 13
(Emery et al., 2001) representative data for cigarette demand. smoking prevalence years and 14+ years
youths aged 10–22 from Smoking intensity is non-significant
the second wave of the modelled separately
longitudinal teenage for current smokers
attitudes and practices (smoked in last 30
survey (n = 12 952). days) and established
Price data from the smokers (smoked in last
Tobacco Institute. 30 days and at least 100
cigarettes in lifetime)
USA 1992–1994;; Monitoring Threshold of Change Price elasticity of Equal effect odds ratio Varying effects odds Categories of
(Liang and Chaloupka, the Future Surveys model (generalized smoking uptake Medium price ratio smoking intensity: no
2002) (n = 110 717), nationally version of the ordered (the probability of 1.060 (1.017, 1.105)** Medium Price consumption, less than
representative data on logit model), controlled progressing to a higher High Price Threshold 1 daily smoking, light daily
8th-, 10th- and 12th-grade for state level smoke- intensity of smoking 1.146 (1.091, 1.204) 1.057 (1.014, 1.102)** smoking (1–5 cigarettes/
students (aged 13–18) free air laws, youth based on a % increase Threshold 2 day), moderate daily
access restrictions, in the price of cigarettes 1.051 (1.001, 1.104)* consumption (10
price differentials (95% CI) Threshold 3 cigarettes per day), and
within 25 miles and 1.094 (1.027, 1.165)** heavy consumption
socioeconomic and Threshold 4 defined as 20 or
demographic variables. 1.128 (1.035, 1.229)** more per day). Price
High Price categories: high, medium
Threshold 1 and low.
1.132 (1.077, 1.188)*** *P P
Threshold 2 ***P
1.190 (1.124, 1.260)***
Threshold 3
1.255 (1.169, 1.348)***
Threshold 4
1.307 (1.186, 1.439)***
USA 1976–1993;; Nationally Discrete time duration Price elasticity of Daily uptake Three measures of
(Tauras, 2005) representative data on method, using a probit smoking uptake –0.646 smoking progression: 1)
8th-, 10th- and 12th-grade specification to estimate Moderate uptake daily uptake, 2) moderate
students (13–18 years) the hazard rate of –0.576 uptake (transition from
contained in baseline progressing to a higher Heavy uptake light smoking 1–5
surveys from the 1976– smoking intensity. –0.412 cigarettes/day to 10 or
1993 components of the Analyses controlled more cigarettes/day,
Monitoring the Future for socioeconomic and 3) Heavy uptake (from
Survey, with follow up demographic factors moderate smoking to 1+
through 1995. as well as anti-smoking packs/day.
sentiment
235
236
Table 6.3. Summary of studies providing data on the effects of price on youth smoking uptake
Methods (location,
size (M/F)
USA 1996;; Cross-sectional Generalized ordered Regression coefficients State average price ** PEDVHGRQ
(Ross et al., 2006) data from students in logit model reflecting the effect of Stage 2,3,4, or 5 two-tailed test
grades 9–12 attending price on probability of –0.383** (0.143)
202 US high schools in progressing to higher Stage 3,4, or 5
1996 from ‘The Study of stages of smoking –0.387** (0.137)
Smoking and Tobacco uptake, corrected Stage 4 or 5
Use Among Young for within-cluster –0.400** (0.138)
People’ dependence (Standard Stage 5

(n = 17 287). State errors) –0.478** (0.183)
cigarette prices from Average perceived
the Tobacco Institute price
and an average school Stage 2,3,4, or 5
price calculated using –0.336** (0.106)
self-reported price were Stage 3,4, or 5
used. –0.354** (0.100)
Stage 4 or 5
–0.367** (0.107)
Stage 5
–0.457** (0.162)
USA 1999–2003;; Annual, Generalized ordered Price elasticity of Threshold 1 ** P
(Slater et al., 2007) nationally representative logit analyses with smoking uptake, 0.87 (0.74–1.02) * P
cross-sectional data on weighted data reflecting the probability Threshold 2
26 301 8th-, 10th- and that controlled for that an adolescent will 0.76 (0.64–0.89)**
12th-graders, aged 14, demographic and move on to the next level Threshold 3
16 and 18 respectively. socioeconomic factors, of smoking for every 0.77 (0.65–0.92)*
Price data calculated as state level tobacco dollar increase Threshold 4
the average price of a control policies, and (95% confidence 0.81 (0.67–0.98)*
premium brand collected year, and accounted intervals) Threshold 5
through a retail audit for clustering at the 0.79 (0.63–0.98)*
of community tobacco community level
vendors
Smoking uptake To summarize, studies that Future Surveys to examine smoking

Longitudinal studies rely on cross-sectional data and cessation behaviours among 8th-,
retrospective information on smoking 10th- and 12th-grade regular smokers.
One study used longitudinal data find mixed results for the effects of After controlling for demographic
to examine smoking transitions price on smoking initiation, likely due characteristics, smoke-free air laws,
other than initiation or cessation. to measurement error in the timing of and youth access laws, the authors
Tauras (2005) examined the impact initiation and in the price measure. found increases in cigarette price
of cigarette prices on smoking Studies that rely on longitudinal data to be positively associated with
uptake transitions among youth and and do not suffer from cross-sectional smoking cessation behaviours. In
young adults in the United States. limitations generally find that price particular, the authors calculated that
He examined the transition from has a negative and significant impact a $1 per pack increase in the price
non-daily to daily smoking and on youth smoking initiation. Finally, of cigarettes increases: the odds
the transitions from light smoking higher cigarette prices have been of wanting to quit smoking by 30%;;
intensity (defined as 1–5 cigarettes found to decrease the probability of the odds of an ever-regular smoker
per day) and moderate smoking transitioning into higher intensities not smoking in the past 30 days by
intensity (defined as smoking 10 of smoking. Evidence suggests that 20%;; the odds of a high school ever-
cigarettes per day on average) price has greater impact at later regular smoker who has made at
to higher intensities of smoking. stages of the uptake process than at least one quit attempt, not smoking in
Tauras (2005) employed baseline the experimentation stage. the past 30 days by 30%.
surveys from the 1976 through 1993
Monitoring the Future Surveys with Smoking cessation Smoking cessation
follow-up surveys on individuals Cross-sectional studies Longitudinal studies
through 1995 in the analyses. He
controlled for anti-smoking sentiment Using an experimental framework, A few studies have used longitudinal
using a variety of techniques: Ross and colleagues (2005) data to examine the impact of price
including separate indicators for examined the expected smoking on adolescent and young adults’
whether or not the individual resided reaction to a future price increase decisions to quit smoking. Tauras and
in a tobacco-producing state or among smokers in high school in Chaloupka (2001) were the first to
resided in Utah, where there is a the USA. The authors used cross- model smoking cessation decisions
large Mormon population whose sectional data collected in 1996 for using longitudinal data. In particular,
religious beliefs ban tobacco use;; the project The Study of Smoking the authors used data extracted
including census division fixed effects and Tobacco Use Among Young from the 1976–1993 cohorts of
to capture division differences in People that contained information on the Monitoring the Future project
smoking sentiment;; and estimating individual’s current smoking status with follow-up surveys through
the smoking progression equations and expected smoking behaviour 1995. The authors used a semi-
on a subsample of the respondents after hypothetical changes in parametric Cox regression to assess
who did not reside in either a tobacco- cigarette price. Four different price the probability that smokers would
producing state or the state of Utah changes were examined including an make a transition from smoking in
during the time that the surveys were increase by US$0.50, by US$1, by the previous wave of data to non-
being conducted. Tauras (2005) US$2, and by US$4. After controlling smoking in the current wave. After
found cigarette prices to have a strong for smoke-free air laws and youth controlling for smoke-free air laws,
negative impact on all the smoking access laws, the authors found region of residence, socioeconomic
transitions that were estimated. In hypothetical increases in cigarette and demographic characteristics, the
particular, the estimated mean price prices to have a strong positive authors concluded that the likelihood
elasticities of daily uptake, moderate impact on youth smoking cessation of making a smoking cessation
uptake and heavy uptake were – 0.65, decisions. In particular, the estimated attempt among both men and women
–0.58 and –0.41, respectively. These price elasticity of cessation ranged increased significantly as cigarette
findings indicate that increases in from 0.895 to 0.930 (Table 6.4). prices rose.
cigarette prices will prevent many Tworek and colleagues (2010)
young adults from progressing into used cross-sectional data from
higher intensities of smoking. the 1991–2006 Monitoring the
237
238
Table 6.4. Summary of studies on the effect of cigarette price on smoking cessation among young adults
Methods (location,
Publication (author, time period, study
Model Endpoint Main results Subpopulations Comments
size (M/F)
High income countries
USA 1987;; National Health Ordered probit split- Smoking cessation Without controlling for Controlling for state * P
(Douglas, 1998) Interview Survey: sample duration model hazard elasticity (% state regulation regulation **P
Cancer Risk Factor with lagged duration change in the probability Past price Past price
Supplement (n = 8754) dependence and of quitting smoking in –0.07 (SE 0.52) 0.08 (SE 0.51)
time-varying covariates a given time period, Present price Present price
(cigarette price and conditional on being a –1.05 (SE 0.82) –0.98 (SE 0.81)
regulation) smoker at the start of the Future price Future price
time period, for each% 1.31 (SE 0.51)** 1.07 (SE 0.52)*
change in price)
USA 1975–1995;; Panels Semi-parametric Cox Price elasticity of Females Females * Statistically significant
(Tauras and Chaloupka, formed from the duration models to smoking cessation 0.24* to 1.00* Smokers at 18
2001) nationally representative estimate the probability Males 0.57*
cross-sectional surveys that smokers would –0.10 to 1.30 Smokers at 23
of 8th-, 10th- and 12th- make a transition from 0.82*
grade students from the smoking in the previous Smoke at least 1
Monitoring the Future wave to non-smoking cigarette in 30 days
Surveys in the current wave. 0.51*
Analyses controlled Smoke at least 1–5
for socioeconomic and cigarettes/day
demographic factors, 0.88*
year effects, clean air Males
laws and youth access Smokers at 18
restrictions Not significant
Smokers at 23
1.06*
Younger smokers
who have smoked any
cigarettes in 30 days
Not significant
Younger smokers who
smoke 1–5 cigarettes
per day
Not significant
Older smokers who have
smoked any cigarettes in
30 days
0.89*
Older smokers who
smoke at least 1–5
cigarettes per day
1.23*
Methods (location,
size (M/F)
USA 1979–1994;; Cross-sectional analysis Tax elasticity of smoking 1992–94 Women
(Glied, 2002) representative microdata controlled for race, sex, cessation: effect of taxes 0.16 0.02
from the National age, marital status in at age 14 on quitting by Men
Longitudinal Survey year of interview, grade follow up in 1992 or 1994 0.19
of Youth. Youth aged of school completed by Low income
14–23 in 1979 followed interview, AFQT score. –0.04
up annually (n = 5823). Includes only those who Those who moved
Price variable represents reported smoking in State since age 14
the mean price of 1984 or earlier 0.17
cigarettes when the
respondent was 14
USA 1975–1995;; Panels Semi-parametric Cox Price elasticity of Range 0.269 to 0.466
(Tauras, 2004) formed from the duration models to smoking cessation Average 0.350
nationally representative estimate the probability
cross-sectional surveys that smokers would
of 8th-, 10th- and 12th- make a transition from
grade students from the smoking in the previous
Monitoring the Future wave to non-smoking in
Surveys the current wave. Models
allow for the possibility
of multiple quit attempts.
Analyses controlled
demographic factors,
year effects, clean air
laws and youth access
restrictions
USA 1996;; cross-sectional Probit regression model Price elasticity of Model using average
(Ross et al., 2005) data collected for The to estimate an equation smoking cessation future perceived price
Study of Smoking and for the probability of 0.90
Tobacco Use Among future smoking cessation Model using average
Young People containing among current smokers future state price
information on smoking as a function of future 0.93
status and expected cigarette prices.
behaviour after a Analyses controlled
hypothetical change in for socioeconomic and
cigarette price demographic factors
USA 1988–2000;; 1988, 1992 Myopic addiction Price elasticity of Model 1 Model 3 * Statistically significant
(DeCicca et al., 2008b) and 2000 waves of the model using probit smoking cessation 0.93* Movers
National Education specifications, Model 2 1.49
Longitudinal Study controlling for 0.47 Stayers
(n = 10 706) including socioeconomic and 0.82
youths aged 18–26 demographic factors and
years. anti-tobacco sentiment.
239
240
Methods (location,
size (M/F)
USA Three models estimated:
(DeCicca et al., 2008b) 1) using intrastate
(contd) variation in cigarette
excise taxes, 2) including
a direct measure of anti-
tobacco sentiment, and
excise taxes faced by

youths who moved state
between study waves
USA Method Model Price elasticity of Overall
(Franz, 2008) 1993–2000;; Nationally Two models were smoking cessation 0.375
representative cross- considered: 18–29 years
sectional data for adults 1) a simple OLS and 0.493
18+ years from the 2) a two-part model. 30–39 years
Behaviour Risk Factor Analyses controlled 0.424
Surveillance System for socioeconomic and 40–64 years
Survey (n = 1 million) demographic factors. –0.398
Price was the state 65+ years
average real cigarette 0.202
price
USA 1991–2006;; nationally Hierarchical generalized Price elasticity of 1.003 * P
(Tworek et al., 2010) representative cross- linear modeling wanting to quit (CI: 1.001 – 1.005)* ** P
sectional surveys of controlling for student Price elasticity of any 1.001 Non-continuation of
8th-, 10th- and 12th-grade characteristics (age, quit attempt (CI: 1.000 – 1.003) smoking is defined as
students from the gender, race/ethnicity, Price elasticity of non- 1.002 ever-regular smokers
Monitoring the Future parental education, total continuation of smoking (CI: 1.001 – 1.003)** who have not smoked in
Surveys (n = 12 073 to income and work status) Price elasticity of 1.003 the past 30 days
78 584 depending on and state policies and discontinuation of (CI: 1.000 – 1.005)* Discontinuation of
outcome variable) year effects smoking smoking was defined as
ever-regular smokers
who have made at least
one quit attempt and who
have not smoked in the
past 30 days
United Kingdom 1984;; Cross-sectional Split-population duration Tax elasticity of quitting Men
(Forster and Jones, data from the British model, using probit smoking –0.60
2001) Health and Lifestyle specification yields to Women
Survey;; representative model the likelihood of –0.46
sample of adults 18+ smoking initiation
years living in England,
Scotland and Wales in
1984 with retrospective
smoking data (n = 9003)
Methods (location,
size (M/F)
Spain 1993, 1995, 1997;; Hazard functions for Price elasticity of Men Estimates compared
(López Nicolás, 2002) Pooled, nationally duration to smoking smoking duration (% –1.32 (t-value –2.23) with those from the Cox
representative cross- cessation are estimated change in the duration Women proportional hazard
sectional data from the using a log logistic split to smoking cessation for –1.50 (t-value –2.11) model and the Gamma
Spanish National Health population model each % change in price) model;; estimates
Survey with retrospective from the preferred
information on smoking. specification presented
Price data from National here.
Statistics Office
Poland 1999;; cross-sectional Descriptive analysis Percentage of Demand reduced/ Demand reduced/
(Ross & Prezwozniak, data on school-aged respondents reporting altered altered
2004) children from the Global reaction (95% 55.8 (± 3.9) Urban
Youth Tobacco Survey confidence interval) Successfully quit 47.9 (± 5.3)
examining the reported 26.0 (± 3.8) Rural
reaction to price Attempted to quit 66.9 (± 5.1)
increases 10.2 (± 2.2) Women
Smoked fewer 51.6 (± 5.9)
cigarettes Men
15.0 (± 2.8) 59.6 (± 5.7)
Smoked cheaper Successfully quit
cigarettes Urban
4.8 (± 1.7) 16.8 (± 4.5)
Consider quitting Rural
20.4 (± 3.4) 38.8 (± 5.8)
No reaction Women
23.6 (± 3.7) 27.7 (± 6.3)
Men
24.6 (± 5.1)
Attempted to quit
Urban
9.5 (± 2.9)
Rural
11.3 (± 3.4)
Women
8.2 (± 2.7)
Men
12.0 (± 3.3)
Smoke fewer
cigarettes
Urban
16.8 (± 3.6)
Rural
12.4 (± 4.1)
Women
13.6 (± 3.1)
Men
16.2 (± 4.1)
241
242
Methods (location,
size (M/F)
Poland Smoke cheaper
(Ross & Prezwozniak, cigarettes
2004) (contd) Urban
4.6 (± 2.0)
Rural
5.1 (± 3.1)
Women
2.2 (± 1.7)

Men
6.9 (± 2.7)
Consider quitting
Urban
21.3 (± 3.4)
Rural
19.2 (± 5.5)
Women
20.2 (± 4.7)
Men
20.6 (± 4.4)
No impact
Urban
31.0 (± 5.6)
Rural
13.2 (± 3.6)
Women
19.8 (± 4.3)
Men
28.2 (± 5.3)
France 1999;; Cross-sectional Discrete time hazard Price elasticity of Aged 21–50 Men and Women
(Peretti-Watel, 2005) data nationally model of smoking smoking cessation (the Men Cessation at age 20 or
representative data for cessation, using logistic probability of quitting 3 before
individuals aged 12–75 regression with time smoking for each% Women 1.005 P = 0.174
from the French Health constant variables increase in price) 3 Cessation between 21
Barometer Survey with including gender, and 30 years
retrospective smoking education and age of 1.017 P
data (n = 13 685) initiation and time variant Cessation after age 30
variables including age, 1.011 P
parenthood and price
The estimated price elasticity of lines between 1992 and 2000 versus quitting, increase the probability of
smoking cessation ranged from 0.34 those young adults who remained in making a cessation attempt, and
and 1.00 for women and between the same state in these two years. increase the desire to quit smoking
íDQGIRUPHQLPSO\LQJWKDW In this specification, cigarette taxes among youth and young adults.
a 10-percent increase in cigarette were found to have a positive impact
price raises the probability of making on young adult smoking cessation for Effect of tax and price on other
a cessation attempt by up to ten only those who moved to a different tobacco product use among
percent. state between 1992 and 2000. The young people
Tauras (2004) expanded on the price elasticity of cessation among
original study to look at longer-term movers was relatively large (1.49), While numerous studies on the
cessation effects related to price. and the authors concluded that economic determinants of youth
Once again using the aforementioned despite the lack of significance on cigarette demand have been published
longitudinal component of the price in this specification (due most over the past few decades, there
Monitoring the Future surveys, he likely to the small sample size, n = have been few econometric studies
employed a stratified Cox regression 321), price is likely to play a strong published on the impact of taxes on
to model multiple quit attempts role in the smoking cessation other tobacco products (Table 6.5).
among young adults. After controlling decisions of young adults.
for smoke-free air laws, US Census High-income countries
division indicators (i.e. indicators for Smoking cessation
the region of the country where the Evidence from low- and middle- Chaloupka and his colleagues (1997)
respondent resided), socioeconomic income countries were the first to estimate the effects
and demographic characteristics, of taxes on youth smokeless tobacco
he found a positive and significant Only one study examining the impact demand in the United States. They
effect of cigarette prices on smoking of price on smoking cessation in a used data on the prevalence and
cessation, confirming the earlier low- or middle-income country context intensity of smokeless tobacco
findings. He concluded that a 10% was identified. While data to calculate use taken from the 1992, 1993
increase in the price of cigarettes direct or indirect price elasticity of and 1994 Monitoring the Future
increases successful cessation by demand was unavailable, Ross surveys described above to examine
young adults by approximately 3.5%. and Prezwozniak (2004) used data the impact of smokeless tobacco
DeCicca et al. (2008b) used data from the first Global Youth Tobacco taxes and other tobacco control
from the 1992 and 2000 waves of Surveys on school-aged children policies. They found higher taxes for
the NELS to examine the influence in Poland to examine the reported smokeless tobacco products to be
of cigarette excise taxes on smoking response of youths to a tobacco price associated with reductions in both the
cessation decisions of young adults in increase. They found that 56% of likelihood and intensity of smokeless
the USA. When they used intra-state respondents reported changing their tobacco use among teenage boys.
variation in cigarette excise taxes to smoking behaviour in response to a Using an estimate of the pass-
identify the impact of tax on smoking price change through successfully through of tax to price and the share
cessation, they found young adults quitting (26.0%), attempting to quit of smokeless tobacco price that is
to be very responsive to changes in (10.2%), reducing their consumption accounted for by tax, the authors
cigarette excise taxes. The estimated (15.0%) or smoking cheaper cigarettes estimated a total price elasticity of
price elasticity of cessation in this (4.8%). These data clearly indicate smokeless tobacco demand in the
model was 0.93. In an alternate that higher real cigarette prices can UDQJH IURP í WR í ZLWK WKH
specification, they included the direct reduce cigarette demand among impact on prevalence accounting for
measure of anti-smoking sentiment, teenage students in Poland. most of the total elasticity.
discussed above, and estimated the To summarize, higher cigarette Tauras and colleagues (2007)
price elasticity of cessation to be prices have been found, using both used data extracted from the 1995–
0.47, but this estimate was based on cross-sectional and longitudinal data, 2001 National Youth Risk Behavior
an insignificant parameter estimate to significantly affect youth and young Surveys (YRBS) to examine the
for price. Finally, the authors used adult smoking cessation behaviours. impact of smokeless tobacco taxes
variation in cigarette taxes faced by In particular, higher prices have been on smokeless tobacco use among
young adults who moved across state found to increase the probability of male high school students.
243
244
Table 6.5. Summary of the studies providing evidence on the effects of other tobacco product prices on use among young people
Methods (location,
Model End Point Main results Subpopulations Comment
size (M/F)
High- income countries
USA 1992–1994;; nationally Two-part model of Price/tax elasticity of Price Tax
(Chaloupka et al., 1997) representative data on smokeless tobacco smokeless tobacco Prevalence Prevalence
males in 8th, 10th and demand: probit methods prevalence –0.523 to –0.346 –0.068 to –0.045
12th grades from the to estimate the smoking Price/tax elasticity of Smoking intensity Smoking intensity
Monitoring the Future prevalence equation smoking intensity for –0.233 to –0.092 –0.029 to –0.012
Surveys (n = 19 581) and ordinary least smokeless tobacco Total price elasticity Total tax elasticity
squares to estimate Price/tax elasticity of –0.746 to –0.438 –0.097 to –0.057
the smoking intensity demand for smokeless
for smokeless tobacco. tobacco
Analyses controlled
demographic variables,
cross-border purchasing
effects and youth access
restrictions
USA 1985;; Cross-sectional Multivariate regression Price elasticity of Cigarettes Cross-elasticities
(Ohsfeldt et al., 1997) data on men aged model adjusted for total smoking prevalence Overall analysed
16 years and over from family income, marital –0.05
the Current Population status, race, ethnicity, 16–24 years
Survey (n = 100 000) employment status and –0.07
occupation 25+ years
–0.05
Snuff
Overall
–0.27
16–24 years
–0.31
45+ years
–0.13
Chewing tobacco
Overall
–0.13
USA 1992–1993;; nationally Multivariate regression Price elasticity of Cigarettes Cross-price elasticities
(Ohsfeldt et al., 1999) representative cross- model. Adjusted for smoking prevalence Overall analysed
sectional (pooled index of smoking –0.15* *: P
surveys) data from the regulation, age, 16–24 years
Current Population education, ethnicity and –0.22*
Survey. Analysis marital status. Prices 25–44 years
restricted to 165 653 of cigarettes and snuff –0.11*
male adults either white were mutually adjusted Snuff
or black 16 years and Overall
older –0.01*
Methods (location,
Model End Point Main results Subpopulations Comment
size (M/F)
USA 16–24 years
(Ohsfeldt et al., 1999) –0.24*
(contd) 25–44 years
–0.05*
USA 1999–2000;; nationally Logistic regression Prevalence price –0.336* Males * P
(Ringel et al., 2005) representative data models of the elasticity for cigars –0.349*
on students in grades probability of current Females
6–12 (age 9–17 years) cigar use, defined as –0.240*
from the National Youth having used cigars
Tobacco Survey in the past 30 days.
(n = 33 632) Analyses controlled
demographic factors as
well as smoke-free air
laws and youth access
restrictions
USA 1995–2001;; data from Two part model of Tax elasticity of –0.197 to –0.121
(Tauras et al., 2007) the National School- smokeless tobacco smokeless tobacco
based Youth Risk demand: probit methods prevalence
Behavior Surveys on to examine smokeless Tax elasticity of –0.085 to –0.044
males aged 13–18 years tobacco prevalence and smokeless tobacco
(n = 25 155) generalized linear model Cigarette cross-price –0.715
with log link and Gamma elasticity of prevalence
distribution to examine Cigarette cross-price –0.413
smoking intensity among elasticity of smoking
smokeless tobacco intensity
users
India 2000–2004;; Global Two-part model of Price elasticity of Overall price elasticity Prevalence elasticities Smoking prevalence
(Joseph, 2010) Youth Tobacco Survey cigarette demand;; smoking prevalence and Bidis Women elasticity estimates
data from 26 of 28 states probit model to estimate smoking intensity –1.40 Bidis for gutka use not
and two of the seven smoking prevalence Smoking prevalence –2.91 statistically significant,
Union Territories on equation and Ordinary elasticity Gutka nor are smoking intensity
youth aged 13–15 years. Least Squares to Bidis –1.12 estimates for both gutka
Average price data from estimate smoking –2.70 Men and bidis
users and non-users to intensity equation Gutka Bidis
construct school/state/ –0.58 –2.475
territory level prices Gutka
–0.270
245
The estimates from this study clearly Youth Tobacco Survey to estimate among young people by accounting
indicate that higher smokeless the impact of cigar prices on cigar for peer, parental or sibling influences
tobacco taxes would significantly demand among adolescents in grades or the source of cigarettes (i.e. buying
reduce the number of male students 6 through 12. After controlling for or borrowing) (Table 6.6). Powell et
who use smokeless tobacco and smoke-free air laws and youth access al. (2005) used survey data from The
the number of days those users use laws, the researchers found the price Study of Smoking and Tobacco Use
smokeless tobacco. The estimated of cigars to be inversely related to Among Young People, conducted in
tax elasticities of smokeless tobacco the prevalence of youth cigar use. 1996, to quantify the importance of
SUHYDOHQFH UDQJHG IURP í In particular, the price elasticity of peer effects on smoking among high
WR í ZKHUHDV WKH HVWLPDWHG youth cigar smoking prevalence was school students. Specifically, Powell
tax elasticities with respect to the HVWLPDWHGWREHí and colleagues (2005) allowed
number of days using smokeless cigarette prices to have both a direct
WREDFFRUDQJHGIURPíWRí Low- and middle-income countries effect and an indirect effect, via a
Moreover, the authors found cigarette social multiplier, on youth smoking
prices to have a significant negative Joseph (2010) estimated the impact of prevalence. The price elasticity
impact on both smokeless tobacco price on youth tobacco demand using of youth smoking prevalence was
prevalence and on the number of data from 73 356 Indian youths aged HVWLPDWHGWREHíZLWKWKHSHHU
days male high school students use 13–15 years from the Global Youth effect playing a significant role in the
smokeless tobacco. The estimated Tobacco Survey conducted between smoking decision process. That is,
cross-price elasticity of smokeless 1999 and 2006. This study suggests the aforementioned price elasticity
WREDFFR SUHYDOHQFH ZDV í that price has a stronger effect on comprised a direct prevalence
and the cross-price elasticity of the the decision to smoke than on the SULFH HODVWLFLW\ RI í DQG DQ
number of days of use of smokeless intensity of smoking among Indian indirect prevalence price elasticity
WREDFFRZDVí7KHVHHVWLPDWHV youth, and that the effect holds true (measuring the social multiplier
indicate that a 10% increase in the across three tobacco types: bidis, HIIHFW RI í 7KHVH HVWLPDWHV
price of cigarettes would decrease gutka and cigarettes. In particular, suggest there is a rather large social
smokeless tobacco prevalence bidis have the highest participation multiplier effect with respect to
by more than 7% and would SULFH HODVWLFLW\ í IROORZHG E\ price changes and youth smoking
decrease the number of days using JXWNDíDQGFLJDUHWWHVKDYHWKH prevalence.
smokeless tobacco by more than 4% ORZHVWSDUWLFLSDWLRQHODVWLFLW\DWí Extending the work of Powell
among male high school students. To summarize, evidence from the et al. (2005), Harris and González
Moreover, these estimates indicate USA finds higher smokeless tobacco López-Valcárcel (2008) tested for
that smokeless tobacco products taxes to decrease adolescent and the possibility that peer influences
and cigarettes are economic young adult smokeless tobacco use. have asymmetric effects. They
complements in consumption for Additional evidence from the USA examined cigarette smoking among
young males. In contrast, research finds a negative relationship between individuals aged 15–24 years using
on adult smokeless tobacco use in cigar prices and cigar use among the 1992–1999 Current Population
the United States has found that individuals in grades 6 through 12. Surveys. The authors found that
smokeless tobacco and cigarettes Finally, recent research from India sibling smoking status is a significant
are substitutes among adults (e.g. suggests that youth consumption determinant of own smoking status,
Ohsfeldt et al., 1999). Tauras and of bidis and gutka is responsive but asymmetries in effect size exist.
colleagues (2007) suggested that to changes in the prices of these In particular, they found that each
the complementarity they observed products. additional smoking sibling raises the
is likely the result of young male probability of smoking by 7.6%, while
tobacco users being at relatively Indirect effects of tax and price each non-smoking sibling decreases
early stages of uptake, where they on youth tobacco demand and the probability by 3.5%. Consistent
are still experimenting with different related outcomes – High-income with Powell et al. (2005), Harris and
types of tobacco products. countries González López-Valcárcel (2008)
Finally, Ringel and colleagues found a strong indirect impact of
(2005) used data from the 1999 Some studies have examined the cigarette price on youth smoking
and 2000 waves of the National indirect effect of price on tobacco use prevalence.
246
Table 6.6. Summary of studies providing evidence on indirect effects on youth smoking of an increase in cigarette prices
Methods (location,
Publication (author, yr, time period;; study
country) design and sample
size (M/F)
High income countries
USA 1996;; Nationally Two-stage generalized Total price elasticity of –0.4982
(Powell et al., 2005) representative, cross- least squares estimator smoking prevalence
sectional data on high for a dichotomous Direct prevalence price –0.3152
school students from dependent variable elasticity
The Study of Smoking (youth smoking Indirect (peer effects) –0.1830
and Tobacco Use Among prevalence), controlled prevalence price
Young People for socioeconomic and elasticity measuring the
(n = 12 705) demographic factors and social multiplier effect
state-fixed effects
USA 1996;; Nationally Standard model of Price elasticity of Model accounting for Price elasticity estimates
(Powell and Chaloupka, representative, cross- consumer demand smoking prevalence parental influences based on estimation
2005) sectional data on high theory using probit –0.27 models that do not
school students from specification to estimate Sensitivity analysis account for parental
The Study of Smoking parental influences on without accounting for influences tend to
and Tobacco Use Among demand for cigarettes parental influences capture both the direct
Young People (n = 11 by youth –0.34 price effect and the
237) indirect price effect
that operates through
the effect of price
on parental smoking
behaviour
USA 1996;; cross-sectional Probit regression model Proportion of All future smokers Light future smokers * Difference between
(Ross et al., 2005) data collected for The to estimate an equation respondents who plan to Switch to a cheaper Switch to a cheaper light and heavy smokers
Study of Smoking and for the probability of engage in compensatory brand brand is significant at the 5%
Tobacco Use Among future smoking cessation behaviour in response to 0.29 (0.45) 0.25** (0.43) level
Young People containing among current smokers a hypothetical price rise. Buy fewer cigarettes Buy fewer cigarettes ** Difference between
information on smoking as a function of future Standard deviations are 0.45 (0.50) 0.43* (0.50) light and heavy smokers
status and expected cigarette prices. shown in parentheses Offer less to friends Offer less to friends is significant at the 1%
behaviour after a Analyses controlled 0.60 (0.49) 0.49** (0.50) level
hypothetical change in for socioeconomic and Buy singles instead of Buy singles instead of
cigarette price demographic factors packs packs
0.15 (0.35) 0.20** (0.40)
Buy packs instead of Buy packs instead of
cartons cartons
0.37 (0.48) 0.34** (0.47)
Buy cartons instead of Buy cartons instead of
packs packs
0.26 (0.44) 0.13** (0.33)
Supplement cigarettes Supplement cigarettes
with smokeless tobacco with smokeless tobacco
0.07 (0.25) 0.07 (0.25)
247
248
Methods (location,
size (M/F)
USA Heavy future smokers
(Ross et al., 2005) Switch to a cheaper
(contd) brand
0.32** (0.47)
Buy fewer cigarettes
0.47* (0.50)
Offer less to friends
0.72** (0.45)
Buy singles instead of
packs
0.08** (0.27)
Buy packs instead of
cartons
0.40** (0.49)
Buy cartons instead of
packs
0.39** (0.49)
Supplement cigarettes
with smokeless tobacco
0.06 (0.24)
USA 1995–2001;; nationally Multinomial logit model Regression coefficients All respondents Ever smokers
(Katzman et al., 2007) representative data on to determine the impact and marginal effects of Price (tax) Coefficient Price (tax) Coefficient
high school students in of price on the probability price and tax changes Borrowers Borrowers
grades 9–12 from the of being a non-smoker, for whether a teen is a –0.328–(0.419) –0.342 –(0.444)
biennial National Youth a borrower or a buyer of non-smoker, borrower Buyers Buyers
Risk Behavior Surveys cigarettes. or buyer. ±±í –0.509 –(0.670)
(n = 49 169) Ordinary least squares Price (tax) elasticity of Price (tax) Marginal Price (tax) Marginal
used to estimate the smoking intensity for effects effects
smoking intensity borrowers and buyers Non-smoker Non-smoker
and price elasticity of (demand defined as 0.079 (0.101) 0.107 (0.140)
smoking intensity among number of cigarettes Borrower Borrower
borrowers and buyers of smoked on days –0.020–(0.026) –0.024 –(0.030)
cigarettes smoking) Buyer Buyer
Price (tax) elasticity of –0.059 –(0.075) –0.084–(0.110)
smoking intensity among Price (tax) elasticity of
borrowers and buyers smoking intensity
(smoking intensity Borrowers
defined as number of –0.306 –(0.028)
days smoked) Buyers
–0.504–(0.108)
Price (tax) elasticity of
smoking intensity
Borrowers
–0.833 –(0.155)
Buyers
–0.320 –(0.074)
Methods (location,
size (M/F)
USA 1992–1999;; data on a Estimates a youth Social multiplier for the 1.60
(Harris & González pooled sample of 49 smoking prevalence deterrent effect of price
López-Valcárcel, 2008) 898 households with equation controlling for
1–4 siblings from the sibling smoking status
1992–1999 US Current
Population Surveys
India 2000–2004;; Global Amemiya’s generalized Prevalence price Cigarettes Women * P
(Joseph, 2010) Youth Tobacco Survey least squares estimator. elasticity Model 1 Cigarettes **P
data from 26 of 28 states Model 1 Probit model –0.4056** Model 1 ***P
and two of the seven with no peer effects, Model 2 –0.6019**
Union Territories on Model 2 assumes that –0.1779*** Model 2
youth aged 13–15 years. the peer measure is Model 3 (Direct) –0.3474**
Average price data from exogenous, and Model 3 –0.3017*** Model 3 (Direct):
users and non-users to has the AGLS estimator Model 3 (indirect) –0.4027**
construct school/state/ that accounts for the –0.0564*** Model 4 (Indirect):
territory level prices endogeniety of the peer Bidis –0.1195**
measure and helps Model 1 Bidis
disaggregate the direct –2.727** Model 1
price effect and the Model 2 –2.9325**
indirect price effect –0.8679** Model 2
Model 3 (Direct) –0.8480
–1.3460** Model 3 (Direct):
Model 3 (indirect) –1.3044*
–0.5513*** Model 3 (Indirect):
Gutka –0.5306**
Model 1 Gutka
–0.6915* Model 1
Model 2 –1.1152**
0.0789 Model 2
Model 3 (Direct) –0.3495
0.0800 Model 3 (Direct)
Model 3 (indirect) –0.4853*
–0.1026 Model 3 (Indirect)
0.0103***
Men
Cigarettes
Model 1
–0.2823*
Model 2
–0.0939
Model 3 (Direct)
–0.2408**
Model 3 (Indirect)
–0.0271**
Bidis
Model 1
–2.4860***
Model 2
249
–0.8760**
250
Methods (location,
size (M/F)
India Model 3 (Direct)
(Joseph, 2010) (contd) –1.4297**
Model 3 (Indirect)
–0.5037***
Gutka
Model 1
–0.3118
Model 2
0.2836*
Model 3 (Direct)
–0.1699
Model 3 (Indirect)
–0.1514***
2000–2004;; Global Probit specification to Prevalence price Overall Women * P
Youth Tobacco Survey model the decision to elasticity Cigarettes Cigarettes **P
data from 26 of 28 states smoke. Model 1, Basic Model 1 Model 1 ***P
and two of the seven model;; Model 2, includes –0.1397 –0.1690
Union Territories on household influence Model 2 Model 2
youth aged 13–15 years. –0.1392 –0.1560
Average price data from Bidis Bidis
users and non-users to Model 1 Model 1
construct school/state/ –1.2272*** –0.7631
territory level prices Model 2 Model 2
–1.2497** –0.6010
Gutka Gutka
Model 1 Model 1
–0.3082 –0.6588
Model 2 Model 2
–0.2867 –0.3934
Men
Cigarettes
Model 1
–0.1341
Model 2
–0.1432
Bidis
Model 1
–1.5776**
Model 2
–1.6999**
Gutka
Model 1
–0.2095
Model 2
–0.2628
In particular, the deterrent effect after a price increase. The analysis 7.9 percentage points, reduces the
of an increase in cigarette price further shows that nearly 21% of probability of buying cigarettes by 5.9
on youth smoking prevalence was future smokers who do not expect percentage points, and decreases
approximately 60% greater than the to change their smoking intensity the probability of borrowing cigarettes
sum of the deterrent effect of price on after the price increase also expect by 2.0 percentage points. Their
each individual separately. to buy fewer cigarettes. This implies finding that higher taxes and prices
This study yields a social multiplier that some future smokers believe significantly reduce smoking among
effect of price on smoking prevalence that they can obtain cigarettes from buyers while having less of an impact
of 1.60, which is consistent with alternative sources, although as on borrowers is consistent with
the implied social multiplier of 1.56 discussed above, getting them for studies described in the smoking
estimated by Powell et al. (2005). free from their peers will become uptake section indicating that price
Using the same data as Powell more difficult. In addition, the has a larger impact of price on youth
et al. (2005), Powell and Chaloupka analysis showed that price increases at later stages of the smoking uptake
(2005) examined the indirect effects in the future will also result in other process.
of price on youth smoking that compensatory behaviours such Tauras and Chaloupka (2004)
worked through parental smoking. As as switching to cheaper brands of used data from the 1991–2000
Chaloupka (2003) discussed, higher cigarettes, switching to smokeless nationally representative surveys of
cigarette prices may have an indirect tobacco products, supplementing 8th-, 10th-, and 12th-grade students
influence on youth smoking by both cigarettes with smokeless tobacco as part of the Monitoring the Future
reducing parental modelling of products, and finally switching Surveys to examine the determinants
smoking and by reducing availability the usual quantity purchased (i.e. of youth attitudes and beliefs about
of cigarettes to youth who might singles, packs, or cartons). smoking. After controlling for smoke-
otherwise steal cigarettes from their Other studies have examined free air laws, youth access laws, state
parents. Powell and Chaloupka’s the impact of cigarette prices on spending on tobacco control efforts,
estimates support Chaloupka’s other outcomes such as sharing of and a host of socioeconomic and
(2003) assertion that the indirect tobacco products, risk perceptions, demographic characteristics, higher
influence of price on youth smoking compensatory behaviours and cigarette prices were found to be
that works through parents accounts attitudes towards tobacco products positively and significantly related to
for about one f ifth of the total impact of (Table 6.7). For example, Katzman the prevalence of youths who agree
price on youth smoking prevalence. and colleagues (2007) used data that there is a great risk in smoking
In addition to examining the effect from the National Youth Risk one or more packs a day, and the
of future cigarette price increases Behavior Surveys (YRBS) conducted prevalence of youths who agree
on smoking cessation decisions from 1995 through 2001 to explore that smoking should be banned in
among high school students, the the relationships between cigarette public places. Moreover, Tauras and
aforementioned study by Ross and taxes/prices and youth acquisition of Chaloupka concluded that higher
colleagues (2005) also examined cigarettes. The YRBS data contained cigarette prices were found to be
expected compensatory behaviour information on how cigarettes were negatively and significantly related
after a future price increase. The acquired, including whether young to the prevalence of youth who think
authors estimated that 60% of smokers bought their own cigarettes the harm associated with smoking is
future smokers intend to offer fewer or obtained them from social exaggerated, the prevalence of youth
cigarettes to their friends for free if sources. As expected, Katzman and who think they can smoke one or
price was increased. This implies that colleagues found that higher cigarette more packs a day and quit smoking,
the availability of cigarettes through taxes/prices reduce the prevalence and the prevalence of youth who
social sources would be diminished and intensity of smoking. The authors think that smoking is not dangerous
after a price increase, and that the full also found that acquisition patterns because they can quit.
price of cigarettes would be indirectly were significantly affected by higher One Canadian study examined
increased by reducing the number taxes/prices. In particular, in models cigarette brand preference as a
of distributional channels available based on their full sample, they function of price among youth in
to high school students. The results estimated that a US$1.00 increase Canada. Leatherdale and colleagues
also indicate that 45% of continuing in the price of cigarettes raises the (2009) used nationally representative
smokers plan to buy fewer cigarettes probability of being a non-smoker by data from youths in 5th through 12th
251
grade in 10 Canadian provinces Indirect effects of tax and price To summarize, higher tobacco
to identify factors associated with on youth tobacco demand and product prices have been found to
smoking premium brands (any related outcomes – Low- and have a direct negative impact on
cigarettes sold for the traditional middle-income countries youth tobacco use and to have an
price per carton), newly emerging indirect negative impact through peer
discount brands (cigarettes sold for Similar to the study by Powell et al. and parental effects. These findings
CAD$10–$12 less than premium (2005), Joseph (2010) estimated suggest that there is a sizeable
brands), and native brands (tax- the direct and indirect effects of social multiplier effect with respect
exempt cigarettes sold to aboriginals, price through peers on young to price changes and youth tobacco
or counterfeit cigarettes). While adult tobacco use prevalence and use. Moreover, higher tobacco
almost half of the young people in intensity. Using data on 73 356 Indian prices have been found to change
this study preferred the premium adolescents aged 13–15 years taken the attitudes and perceptions about
brands, half were accessing cheaper from the Global Youth Tobacco tobacco among youths, alter the
cigarettes. The study suggests that Survey, Joseph found a significant availability of tobacco through social
youth with less spending money impact of price on youth demand for sources, and increase the likelihood
and those who smoke more were cigarettes, bidis and gutka, as well as that young individuals engage in
more likely to access these cheaper a significant indirect effect through compensatory behaviours.
sources of cigarettes. peers on demand for all three
tobacco products (Table 6.6).
252
Table 6.7. Summary of studies providing evidence on other outcomes related to an increase in cigarette prices
Methods (location, time period;;

Publication (author, year) study design and sample size Outcomes Main results Comments
(M/F)
USA 1993–1994;; Reported reactions Possible reactions to price were % (95% CI) Among young smokers (aged
(Biener et al., 1998) to a price increase were assessed typified as: Adults 12–17), those from lower-income
through telephone interviews 1) Cut costs associated with Cut costs groups were significantly more
with a representative sample continued smoking (switching to a 19.0 (15.1, 23.7) likely than their more affluent
of Massachusetts adults and cheaper brand or reducing number Consider quitting smoking counterparts to cut the costs
teenagers (aged 12–17). Only smoked) 35.0 (29.6, 39.6) associated with continued smoking
smoking respondents included in 2) Quit smoking No response in response to a tax/price increase
the analysis (n = 1999). Descriptive 3) No response to price increase. 46.0 (40.9, 51.1) rather than doing nothing or
statistics indicate response to Regression coefficients for all Teenagers considering quitting smoking
price increase. Multinomial logistic possible pairs of above outcomes in Cut costs
regression used to test hypothesis a multinomial set 26.0 (10.4, 42.0)
that lower-income smokers would Consider quitting smoking
be more responsive to price than 21.0 (9.3, 31.9)
higher income smokers and that No response
heavier smokers would be more 53.0 (36.8, 69.6)
likely to cut costs rather than
attempt to quit when compared to
lighter smokers
USA Probit methods to estimate attitudes Twelve dependent variables on Cigarette prices have a positive and
(Tauras and Chaloupka, 2004) and beliefs equations. Analyses attitudes and beliefs about smoking, significant impact on the ‘great risk’
controlled for socioeconomic and 6 expected to be positively affected and ‘public smoking ban’ outcomes
demographic factors as well as year and 6 negatively affected by and a negative and significant
effects tobacco control policies. impact on the ‘harm exaggerated’,
‘smoke 1+ packs daily and quit’,
‘smoking is not dangerous – you
can quit’ and ‘don’t mind being
near smokers’ outcomes. Price
was found to be an insignificant
determinant of ‘smoking reflects bad
judgement’, ‘dirty habit’, ‘disapprove
of adults smoking’, ‘prefer to date
non-smoker’, ‘smokers enjoy
life’ and ‘cigarettes are easily
accessible’ outcomes
Canada 2006–2007;; Nationally Odds of smoking discount Factors associated with smoking * P
(Leatherdale et al., 2009) representative data for 5th – 12th cigarettes versus premium discount versus premium ** P
grade students from the Canadian cigarettes (95% CI) cigarettes *** P
Youth Smoking Survey. Two logistic Odds of smoking native cigarettes a. Smoking status
regression models were used: versus premium cigarettes (95% CI) Occasional smoker
Model 1 estimates the odds of 1.00
smoking discount cigarettes versus Daily smoker
premium cigarettes and Model 2 1.55 (1.14 to 2.10)**
253
254
Table 6.7. Summary of studies providing evidence on other outcomes related to an increase in cigarette prices
Methods (location, time period;;

Publication (author, year) study design and sample size Outcomes Main results Comments
(M/F)
Canada estimates the odds of smoking b. Aboriginal status

(Leatherdale et al., 2009) (contd) native cigarettes versus premium Non-aboriginal
cigarettes 1.00
Aboriginal
0.028 (0.19 to 0.43)***
c. Weekly spending money
$C0
1.00
$C1 to $C10
0.39 (0.19 to 0.79)**
$C11 to more
0.84 (0.48 to 1.48)
d. Average cigarettes/day
Few puffs to 1 cigarette
1.00
2–3 cigs
2.11 (1.01 to 4.00)*
4–10 cigs
4.02 (2.00 to 8.10)***
11 or more cigs
2.87 (1.34 to 6.12)*
Factors associated with smoking
native versus premium brands
a. Sex
Girl
1.00
Boy
1.72 (1.24 to 2.40)**
b. Smoking Status
Occasional smoker
1.00
Daily smoker
1.45 (1.02 to 2.06)**
c. Aboriginal status
Non-aboriginal
1.00
Aboriginal
0.56 (0.37 to 0.84)**
d. Weekly spending money
$C0
1.00
$C1 – $C10
0.72 (0.39 to 1.32)
$C11 or more
0.34 (0.20 to 0.56)***
References
Auld MC (2005). Causal effect of early Chaloupka FJ, Wechsler H (1997). Price, Farrelly MC, Bray JW, Pechacek TF, Woollery
initiation on adolescent smoking patterns. tobacco control policies and smoking among T (2001). Responses by adults to increases
Can J Econ, 38:709–734 doi:10.1111/j.0008- young adults. J Health Econ, 16:359–373. in cigarette prices by sociodemographic
4085.2005.00299.x. d o i : 1 0 .1 0 1 6 / S 0 1 6 7 - 6 2 9 6 ( 9 6 ) 0 0 5 3 0 - 9 characteristics. Southern Economic Review,
PMID:10169306 68:156–165 doi:10.2307/1061518.
Biener L, Aseltine RH Jr, Cohen B, Anderka
M (1998). Reactions of adult and teenaged DeCicca P, Kenkel D, Mathios A (2000). Farrelly MC, Nimsch CT, Hyland A, Cummings
smokers to the Massachusetts tobacco tax. Am Racial difference in the determinants of M (2004). The effects of higher cigarette prices
J Public Health, 88:1389–1391.doi:10.2105/ smoking onset. J Risk Uncertain, 21:311–340 on tar and nicotine consumption in a cohort
AJPH.88.9.1389 PMID:9736885 doi:10.1023/A:1007819625751. of adult smokers. Health Econ, 13:49–58.
doi:10.1002/hec.820 PMID:14724893
Carpenter C, Cook PJ (2008). Cigarette taxes DeCicca P, Kenkel D, Mathios A (2002). Putting
and youth smoking: new evidence from national, out the fires: will higher taxes reduce the onset Fletcher JM, Deb PS et al. (2009). Tobacco use,
state, and local Youth Risk Behavior Surveys. of youth smoking? J Polit Econ, 110:144–169 taxation and self c ontrol in adolescence. NBER
J Health Econ, 27:287–299.doi:10.1016/j. doi:10.1086/324386. Working Paper Series #15130. Cambridge,
jhealeco.2007.05.008 PMID:18242745 MA, National Bureau of Economic Research.
DeCicca P, Kenkel D, Mathios A (2005). The
Cawley J, Markowitz S, Tauras J (2004). fires are not out yet: higher taxes and young Forster M, Jones AM (2001). The role of
Lighting up and slimming down: the effects of adult smoking. Adv Health Econ Health tobacco taxes in starting and quitting smoking:
body weight and cigarette prices on adolescent Serv Res, 16:293–312.doi:10.1016/S0731- duration analysis of British data. J R Stat Soc
smoking initiation. J Health Econ, 23:293– 2199(05)16014-2 PMID:17867245 Ser A Stat Soc, 164:517–547 doi:10.1111/1467-
311.d o i:10 .1016 / j . j h e a l e c o. 2 0 0 3 .12 . 0 0 3 985X.00217.
PMID:15019756 DeCicca P, Kenkel D, Mathios A (2008b).
Cigarette taxes and the transition from youth Franz GA (2008). Price effects on the
Cawley J, Markowitz S, Tauras J (2006). to adult smoking: smoking initiation, cessation, smoking behaviour of adult age groups.
Obesity, cigarette prices, youth access laws and participation. J Health Econ, 27:904– Public Health, 122:1343–1348.doi:10.1016/j.
and adolescent smoking intiation. East Econ 917.d o i:10 .1016 / j . j h e a l e c o. 2 0 0 8 . 0 2 . 0 0 8 puhe.2008.05.019 PMID:18951594
J, 32:149. PMID:18513811
Glied S (2002). Youth tobacco control:
Centers for Disease Control and Prevention Decicca P, Kenkel D, Mathios A et al. (2008a). reconciling theory and empirical evidence. J
(2008). Global Youth Tobacco Surveillance, Youth smoking, cigarette prices, and anti- Health Econ, 21:117–135.doi:10.1016/S0167-
2000–2007. MMWR Morb Mortal Wkly Rep, smoking sentiment. Health Econ, 17:733–749. 6296(01)00118-7 PMID:11845920
57:1–28. PMID:18185492 doi:10.1002/hec.1293 PMID:17935201
Göhlmann S (2007). The determinants of
Centers for Disease Control and Prevention Dee TS (1999). The complementarity of teen smoking initiation - empirical evidence for
(CDC) (1998). Response to increases in smoking and drinking. J Health Econ, 18:769– Germany. Rheinisch-Westfalisches Institut
cigarette prices by race/ethnicity, income, 793.doi:10.1016/S0167- 629 6(9 9) 0 0 018 -1 fur Wirtschaftsforschung, Ruhr-Universitat
and age groups–United States, 1976–1993. PMID:10847934 Bochum, Universitat Dortmund, Universitat
MMWR Morb Mortal Wkly Rep, 47:605–609. Duisburg-Essen.
PMID:9699809 Douglas S (1998). The duration of
smoking habit. Econ Inq, 36:49–64 Grignon M, Pierrard B (2002). Youth tobacco
Chaloupka FJ (1991). Rational addictive doi:10.1111/j.1465-7295.1998.tb01695.x. initiation and the effect of tobacco price:
behavior and cigarette smoking. J Polit Econ, evidence from France. Paper presented
99:722–742 doi:10.1086/261776. Douglas S, Hariharan G (1994). The hazard at the 4th European Conference on Health
of starting smoking: estimates from a split Economics.
Chaloupka FJ (2003). Contextual factors and population duration model. J Health Econ,
youth tobacco use: policy linkages. Addiction, 9 8 13:213–230.doi:10.1016/0167-6296(94)90024- Grossman M, Chaloupka FJ (1997). Cigarettes
Suppl 1;;147–149.doi:10.1046/j.1360-0443.98. 8 PMID:10138026 taxes, the straw to break the camel’s back.
s1.10.x PMID:12752367 Public Health Rep, 112:291–297.
Dupont SD, Ward AJ (2002). The economic
Chaloupka FJ, Grossman M (1996). Price, impacts of cigarette tax reductions on youth Gruber J, Zinman J (2001). Youth smoking in
tobacco control policies and youth smoking. smoking in Canada. Brock University. the United States: evidence and implications.
NBER Working Paper Series. Working Paper In Risky behaviour among youths: an economic
#5740. Cambridge, MA, National Bureau of Emery S, White MM, Pierce JP (2001). analysis. Chicago: University of Chicago Press.
Economic Research. Does cigarette price influence adolescent p. 69–120.
experimentation? J Health Econ, 20:261–
Chaloupka FJ, Pacula RL (1999). Sex and race 270.doi:10.1016/S0167- 629 6(0 0) 0 0 0 81- 3 Harris JE, Chan SW (1999). The continuum-
differences in young people’s responsiveness PMID:11252373 of-addiction: cigarette smoking in relation
to price and tobacco control policies. Tob to price among Americans aged 15–
Control, 8:373–377.doi:10.1136/tc.8.4.373 Evans WN, Farrelly MC (1998). The 29. Health Econ, 8:81–86.doi:10.1002/
PMID:10629242 compensating behavior of smokers: taxes, 6 , & , $ , '
tar, and nicotine. Rand J Econ, 29:578–595. HEC401>3.0.CO;;2-D PMID:10082146
Chaloupka FJ, Tauras JA, Grossman M (1997). doi:10.2307/2556105 PMID:11794360
Public policy and youth smokeless tobacco use.
South Econ J, 64:503 doi:10.2307/1060863.
255
Harris JE, González López-Valcárcel B (2008). Lewit EM, Hyland A, Kerrebrock N, Cummings Ross H (2004b). Russia (Moscow) 1999 Global
Asymmetric peer effects in the analysis of KM (1997). Price, public policy, and smoking in Youth Tobacco Survey: Economic aspects.
cigarette smoking among young people in young people. Tob Control, 6 Suppl 2;;S17–S24. HNP Discussion Paper Series, Economics of
the United States, 1992–1999. J Health Econ, doi:10.1136/tc.6.suppl_2.S17 PMID:9583648 Tobacco Control Paper No. 23. Washington
27:249–264. PMID:18179836 DC, The World Bank.
Lewitt EM, Coate D (1982). The potential
Jha P, Chaloupka FJ (1999). Curbing the for using excise taxes to reduce smoking. J Ross H, Chaloupka FJ (2003). The effect of
epidemic. Governments and the Economics Health Econ, 1:121–145.doi:10.1016/0167- cigarette prices on youth smoking. Health
of Tobacco Control. Washington D.C., World 6296(82)90011-X PMID:10263952 E c o n ,12 : 2 17– 2 3 0 . d o i :10 .10 0 2 / h e c . 7 0 9
Bank. PMID:12605466
Liang L, Chaloupka FJ (2002). Differential
Joseph RA (2010). The Economics of Youth effects of cigarette price on youth smoking Ross H, Chaloupka FJ (2004). The effect of
Tobacco Use in India. Chicago: University of intensity. Nicotine Tob Res, 4:109–114. public policies and prices on youth smoking.
Illinois. d o i : 1 0 . 1 0 8 0 / 1 4 6 2 2 2 0 0 1 1 0 1 0 3 1 8 8 South Econ J, 70:796 doi:10.2307/4135273.
PMID:11906687
Karki YB, Pant KD, Pande BR (2003). A Ross H, Chaloupka FJ, Wakefield M (2006).
study on the economics of tobacco in Nepal. López Nicolás A (2002). How important are Youth smoking uptake progress:price and
HNP Discussion Paper Series, Economics of tobacco prices in the propensity to start and public policy effects. East Econ J, 32:355.
Tobacco Control Paper No.13. Washington DC, quit smoking? An analysis of smoking histories
The World Bank. from the Spanish National Health Survey. Ross H, Powell LM, Tauras JA, Chaloupka
Health Econ, 11:521–535.doi:10.1002/hec.745 FJ (2005). New evidence on youth smoking
Katzman B, Markowitz S, McGeary KA (2007). PMID:12203755 behaviour based on experimental price
An empirical investigation of the social market increases. Contemp Econ Policy, 23:195–210
for cigarettes. Health Econ, 16:1025–1039. Madden D (2007). Tobacco taxes and starting doi:10.1093/cep/byi015.
doi:10.1002/hec.1215 PMID:17266167 and quitting smoking: does the effect differ
by education? Appl Econ, 39:613–627 Ross H, Prezwozniak K (2004). Poland 1999
Kidd MP, Hopkins S (2004). The hazards of doi:10.1080/00036840500447898. Global Youth Tobacco Survey: Economic
starting and quitting smoking: some Australian Aspects. HNP Discussion Paper Series,
evidence. Econ Rec, 80:177–192 doi:10.1111/ Ohsfeldt RL, Boyle RG, Capilouto E (1997). Economics of Tobacco Control Paper No. 22.
j.1475-4932.2004.00171.x. Effects of tobacco excise taxes on the Washington DC, The World Bank.
use of smokeless tobacco products in the
Kostova D, Ross H, Blecher E et al. (2010). USA. Health Econ, 6:525–531.doi:10.1002/ Sarntisart I (2003). An economic analysis of
Prices and cigarette demand: evidence from 6 , & , $ , ' tobacco control in Thailand. HNP Discussion
youth tobacco use in developing countries. HEC300>3.0.CO;;2-Y PMID:9353656 Paper Series, Economics of Tobacco Control
NBER Working Paper Series. Working Paper Paper No.15. Washington DC, The World Bank.
#15781. Cambridge, MA, National Bureau of Ohsfeldt RL, Boyle RG, Capilouto E (1999).
Economic Research. Tobacco taxes, smoking, restrictions and Schnohr CW, Kreiner S, Rasmussen M et al.
tobacco use. In: Chaloupka FJ, Grossman (2008). The role of national policies intended
Krasovsky K, Andreeva T, Krisanov D et M, Bickel WK, et al., editors. The economics to regulate adolescent smoking in explaining
al. (2002). Economics of tobacco control in analysis of substance use and abuse: an the prevalence of daily smoking: a study of
Ukraine from the public health prespective. integration of econometrics and behavioral adolescents from 27 European countries.
Kiev: Alcohol and Drug Information Centre. economic research. Chicago: University of Addiction, 103:824–831.doi:10.1111/j.1360-
Chicago Press. p. 15–29. 0443.2008.02161.x PMID:18412761
Kyaing NN (2003). Tobacco economics in
Myanmar. HNP Discussion Paper Series, Peretti-Watel P (2005). Pricing policy and Sen A, Wirjanto T (2010). Estimating the
Economics of Tobacco Control Paper No.14. some other predictors of smoking behaviours: impacts of cigarette taxes on youth smoking
Washington DC, The World Bank. an analysis of French retrospective data. participation, initiation, and persistence:
Int J Drug Policy, 16:19–26 doi:10.1016/j. empirical evidence from Canada. Health
Lance PM, Akin JS, Dow WH, Loh CP (2004). drugpo.2004.06.003. Econ, 19:1264–1280.doi:10.1002/hec.1548
Is cigarette smoking in poorer nations highly PMID:19816884
sensitive to price? Evidence from Russia and Powell LM, Chaloupka FJ (2005). Parents,
China. J Health Econ, 23:173–189.doi:10.1016/j. public policy, and youth smoking. J Policy Anal Slater SJ, Chaloupka FJ, Wakefield M et al.
jhealeco.2003.09.004 PMID:15154693 Manage, 24:93–112 doi:10.1002/pam.20071. (2007). The impact of retail cigarette marketing
practices on youth smoking uptake. Arch Pediatr
Laxminarayan R, Deolalikar A (2004). Tobacco Powell LM, Tauras JA, Ross H (2005). Adolesc Med, 161:440–445.doi:10.1001/
initiation, cessation, and change: evidence The importance of peer effects, cigarette archpedi.161.5.440 PMID:17485618
from Vietnam. Health Econ, 13:1191–1201. prices and tobacco control policies for youth
doi:10.1002/hec.932 PMID:15386650 smoking behavior. J Health Econ, 24:950– Sloan FA, Trogdon JG (2004). The impact of
9 6 8 .d o i:10.1016 / j.j h e a l e c o. 2 0 0 5 .0 2 .0 0 2 the Master Settlement Agreement on cigarette
Leatherdale ST, Ahmed R, Barisic A et al. PMID:15990184 consumption. J Policy Anal Manage, 23:843–
(2009). Cigarette brand preference as a 855.doi:10.1002/pam.20050 PMID:15499706
function of price among smoking youths in Ringel JS, Wasserman J, Andreyeva T (2005).
Canada: are they smoking premium, discount Effects of public policy on adolescents’ Tauras J, Powell L, Chaloupka F, Ross H (2007).
or native brands? Tob Control, 18:466–473. cigar use: evidence from the National The demand for smokeless tobacco among male
doi:10.1136/tc.2009.029736 PMID:19797534 Youth Tobacco Survey. Am J Public Health, high school students in the United States: the
95:995–998.doi:10.2105/AJPH.2003.030411 impact of taxes, prices and policies. Appl Econ,
Lewit EM, Coate D, Grossman M (1981). PMID:15914822 39:31–41 doi:10.1080/00036840500427940.
The effects of government regulation on
teenage smoking. J Law Econ, 24:545–549 Ross H (2004a). Ukraine (Kiev) 1999 Global Tauras JA (2004). Public policy and
doi:10.1086/466999. Youth Tobacco Survey: Economic issued. smoking cessation among young adults in
HNP Discussion Paper Series, Economics of the United States. Health Policy, 68:321–
Tobacco Control Paper No. 20. Washington 3 32.doi:10.1016 / j.healthpol. 20 0 3.10.0 07
DC, The World Bank. PMID:15113643
256
Tauras JA (2005). Can public policy deter Tauras JA, Markowitz S, Cawley J (2005). U.S.Department of Health and Human
smoking escalation among young adults? J Tobacco control policies and youth smoking: Services (USDHHS) (1989).Reducing the
Policy Anal Manage, 24:771–784.doi:10.1002/ evidence from a new era. In: Lindgren B, health consequences of smoking: 25 years
pam.20137 PMID:16201059 Grossman M, editors. Substance use: individual of progress: a report of the Surgeon General:
behaviour, social interaction, markets and 1989 Executive Summary. Centers for Disease
Tauras JA, Chaloupka FJ (1999). Price, clean politics. Amsterdam: JAI, an imprint of Elsevier Prevention and Health Promotion. Office on
indoor air laws and cigarette smoking: evidence Ltd. p. 227–91. Smoking and Health.
from longitudinal data for young adults. NBER
Working Paper Series. Working Paper #6937. Tauras JA, O’Malley PM, Johnston LD (2001). van Walbeek C (2002). Recent trends in
Cambridge, MA, National Bureau of Economic Effects of price and access laws teenage smoking prevalence in South Africa–some
Research. smoking intiation: a national longitudinal evidence from AMPS data. S Afr Med J,
analysis. Research Paper Series No. 2. 92:468–472. PMID:12146134
Tauras JA, Chaloupka FJ (2001). Determinants University of Illinois, Chicago, impacTEEN,
of smoking cessation: an analysis of young YES! (Youth, Education and Society). Waller BJ, Cohen JE, Ferrence R et al. (2003).
adult men and women. In: Grossman M, Hsieh The early 1990s cigarette price decrease and
CR, editors. Economics of substance abuse: Townsend J, Roderick P, Cooper J (1994). trends in youth smoking in Ontario. Can J
the experience of developed countries and Cigarette smoking by socioeconomic group, Public Health, 94:31–35. PMID:12583668
lessons for developing countries. Cheltenham, sex, and age: effects of price, income,
United Kingdom: Edward Elgar Publishing and health publicity. BMJ, 309:923–927. Wasserman J, Manning WG, Newhouse
Limited. p. 35. PMID:7950662 JP, Winkler JD (1991). The effects of excise
taxes and regulations on cigarette smoking.
Tauras JA, Chaloupka FJ (2004). The impact of Tworek C, Yamaguchi R, Kloska DD et al. J Health Econ, 10:43–64.doi:10.1016/0167-
tobacco control spending and tobacco control (2010). State-level tobacco control policies 6296(91)90016-G PMID:10112149
policies on adolescents’ attitudes and beliefs and youth smoking cessation measures.
about cigarette smoking. Evid Based Prev Health Policy, 97:136–144.doi:10.1016/j. Zhang B, Cohen J, Ferrence R, Rehm J (2006).
Med, 1:111–120. healthpol.2010.04.009 PMID:20483500 The impact of tobacco tax cuts on smoking
initiation among Canadian young adults.
Am J Prev Med, 30:474–479.doi:10.1016/j.
amepre.2006.02.001 PMID:16704940
257
258
Chapter 7
Tax, price and tobacco use among the poor
In general, the prevalence of tobacco consumption in response status are presented. The opportunity
tobacco use is higher among the to increased tobacco taxation. Also, cost of spending on tobacco is
poor.1 Several reasons are given to by reducing consumption or quitting, discussed in the second section.
explain why the poor smoke more the poor benefit most by avoiding A third explores the relationship
(Bobak et al., 2000). First, people at the health and economic cost from between tobacco use and poverty.
lower socioeconomic status tend to tobacco use. Thus, increasing taxes The possible regressivity of tobacco
be less aware of the adverse health on tobacco is considered a win-win taxes is discussed and the existing
consequences of smoking and thus situation, not only because it serves evidence from several countries
more likely to be smokers. Second, the dual goal of curbing the tobacco is then presented, followed by the
smoking is an outlet for releasing epidemic and increasing government existing evidence on price elasticity
stress resulting from material revenue, but also due to the fact that of tobacco use for different income
deprivation of the poor (Graham, it works more effectively for the poor or socioeconomic status in several
1987). Third, a smoker sees his/her than for the rich. counties. The empirical evidence
smoking habit as a way of rewarding Raising tobacco excise taxes found on price sensitivity of different
himself/herself, an attitude more as a tobacco control policy is often socioeconomic groups is then
likely to be observed among the poor criticized because of the regressivity explained, and the chapter concludes
(Graham, 1994). Fourth, the loss of of tobacco taxes. It is argued that the with a discussion on the implications
productivity and income due to ill increase in taxes on tobacco may of the empirical findings.
health caused by smoking-induced increase the burden on low-income
diseases is lower for the low-income people by reducing the overall Patterns of tobacco use within
people. Thus, the poor seem to have ability to purchase and consume countries by socioeconomic
more reasons to adopt the practice of other goods and services and thus status
smoking as a part of their life. increasing the inequality in the post-
The high level of smoking tax distribution of income. The relationship between smoking
prevalence among the poor This chapter concentrates on behaviour and socioeconomic
imposes a heavier disease burden smoking prevalence, price sensitivity status is well documented in the
on this disadvantaged group of and tobacco consumption by literature. In general, the studies
the population and widens social socioeconomic status, and their examining this relationship measure
and economic disparity between implications on potential regressivity socioeconomic status with respect
the rich and the poor. However, of tobacco taxes. It is organized as to a wide range of variables such as
the poor are also generally more follows: First, the patterns of tobacco income level, education, profession,
responsive to price changes, which use within countries with different expenditure levels and other relevant
makes them more inclined to cut income levels and socioeconomic socioeconomic characteristics.
Throughout the chapter, the status of poverty has been identified with low socioeconomic status of individuals or households in terms of household income, expenditure, level of
1
education of individuals. Depending on how different empirical studies conceptualized poverty and defined low socioeconomic status of population, we have used the t wo concepts (i.e.
low socioeconomic status or poverty) interchangeably without further specification of how poverty is measured in general. In most cases, the stratification of the poor and the non-poor
is based on the classification of households by total household income or expenditure.
259
The effect of socioeconomic Figure 7.1. Prevalence of daily tobacco smoking by income group and income
status or education or income on quintile
smoking differs from one country
to another and depends on the
diffusion of smoking within the
country (Graham, 1996). If the
country is in the later stage of the
tobacco epidemic, when the smoking
prevalence rate is declining, smoking
is typically observed more in the
lower socioeconomic sector of the
population (Pierce et al. 1989;; Lopez
et al. 1994;; Graham, 1996). Bobak
et al. (2000) reported the ratio of
smoking prevalence rate of the lowest
socioeconomic groups to the highest
socioeconomic groups, calculated Notes: 1. Q1 to Q5 indicate income quintiles. Q1 represents the lowest income group and Q5 is the highest income
from 74 studies which examined group. 2. The graph was made using average prevalence figures from 44 countries. Prevalence of China and India were
removed from these averages to avoid skewed results from their large population weights. 3. High-income countries are
this relationship in 41 countries. In not included in the graph.
many countries this ratio is found to Source: Adapted from David A , Esson K, Perucic AM, Fitzpatrick C (2010).Tobacco use: equity and social determinants.
In: Blas E, Sivasankara Kurup A, editors. Equity, social determinants and public health programmes. Geneva: World
be greater than 1, which indicates Health Organization.
that people in the lowest group
have a higher smoking prevalence classification of countries, reveals (aged 50–79) Italian males, smoking
rate than those in the highest more information about the economic prevalence did not change with their
socioeconomic group. However, gradient of tobacco use. Evidence education level – it is around 25%,
there are huge differences between from the World Health Survey 2003 whereas among old Italian women
the countries, even within the same showed that tobacco smoking is there was a positive relationship
income category. For example, mostly strongly related to household between education level and
among low-income countries, the permanent income or wealth (WHO, smoking prevalence rates (Federico
ratio of smoking prevalence in the 2007). As shown in Figure 7.1 et al., 2004). Similarly, in Portugal
lowest socioeconomic group relative reproduced from WHO (2010), poorer and Greece, women with higher
to the highest socioeconomic group people tend to have greater smoking education and in higher income
varied between 9 in India and 1 in prevalence than richer people in low- groups are more likely to smoke. The
China. Among lower-middle-income income- and lower-middle-income differences among the EU countries
countries the ratio varied between as well as upper-middle-income were explained by the stage of the
3 in the Russian Federation to less countries, and the disparity in terms country in the tobacco epidemic. If
than 1 in Thailand and Bolivia. Thus, of smoking prevalence rate is more the country is at the earlier stages,
in the latter two countries people in apparent among the lower-income then smoking prevalence might be
the lowest socioeconomic group groups of countries. positively associated with education,
have lower smoking prevalence than The general pattern observed i.e. more educated men and women
people in the highest socioeconomic in high-income countries is that as were more likely to smoke than less-
group. Among high-income countries the level of income or education educated individuals. More recent
the ratio is typically greater than 1, of individuals increases, smoking studies reported that the gaps in
indicating that smoking prevalence is prevalence rate declines. Those at the prevalence rates between low-
higher among the poor than the rich. the lowest income group or those with income and high-income groups has
An exception is Japan, which has a the lowest level of education have a widened in recent years because
ratio of less than 1. higher smoking prevalence rate than even though the prevalence rates
The examination of the those at the highest income group or are generally declining, the decrease
relationship of smoking prevalence those with highest level of education. is greater among those with higher
with socioeconomic status of However, there are some exceptions levels of education or those with low
people, using the World Bank’s as well. For example, among old income (for example, Townsend et al.
260
(1994), Federico et al. (2004), Franks Since in high-income, lower-middle- Thomson et al. (2002) reported that
et al. (2007), Khang et al. (2009)). income and low-income countries the low-income smoking households in
Compared to high-income poor smoke more than the rich and New Zealand spent almost 14% of
counties, the evidence on the their income level is lower, it means non-housing spending on tobacco.
relationship between income level and that, on average, they spend a higher In contrast to other studies, John
prevalence rate is different in upper- share of their income on tobacco than (2008) reported that in India, the
middle income countries. According the rich. It is reasonable to expect share of tobacco expenditure in poor
to the results of the limited number that tobacco expenditures have a households’ total budget was slightly
of studies, as education level or the higher opportunity cost among the less than that of richer households
income level of household increases, poor. Spending on tobacco would be even though the poor had a higher
smoking prevalence and tobacco expected to reduce the consumption prevalence rate of tobacco use in
consumption increase as well (Onder of food and non-food items, thus India. These results are summarized
(2002) for Turkey, Sayginsoy et al. reducing the quality of life (Wang et in Table 7.2.
(2002) for Bulgaria and van Walbeek al., 2006).2 Several studies estimated the
(2002a) for South Africa). The percentage of income spent share of income or expenditures
The relationship between on tobacco products is generally spent on tobacco and what individuals
prevalence rate and socioeconomic found to be higher when the level of could purchase with that money if they
status in lower-middle-income and household income or expenditure were to divert it away from tobacco.
low-income countries is similar to that is lower. For example, in Myanmar, For example, a study conducted by
observed in high-income countries. in 2000, the share of tobacco Efroymson et al. (2001) in Bangladesh
In general, those with higher income expenditures constituted 4.04% of showed that during 1992–1996, the
level or those with more education total household expenditures of all average male smoker spent more
have lower smoking prevalence households in the lowest income than twice as much on cigarettes as
rates than those with lower income quintile, compared to 1.58% in the he spent on clothing, housing, health,
level or those with less education. highest income quintile (Kyaing, and education combined. Moreover,
Although this is the general trend, 2003). In Indonesia, cigarette on average male and female smokers
there are some differences in rural expenditure as a percentage of would be able to buy 1402 calories
or urban parts of the country as well income was 7.24% for the low-income and 770 calories of rice, respectively,
as the types of tobacco consumed. group and 3.02% for the high-income with the money saved by quitting.
Table 7.1 presents the prevalence group (Adioetomo et al., 2005). On This study further showed that in
rates of individuals or households average, poor households in Morocco Bangladesh in 1995/96, the lowest
by socioeconomic status in different spent the same amount on tobacco household expenditure (below $18)
countries grouped according to products and education, while their group spent almost 10 times as much
World Bank income classification. cigarette expenditure was half of the on tobacco as on education. If the
average spending on health (Aloui, poor had reallocated 69% of their
Opportunity cost of tobacco use 2003). Similarly, in Turkey in 1994, tobacco expenditure to food, which
the households in the lowest income a typical lowest-income household
Opportunity cost is defined as the cost quintile spent 3.08% of their monthly without smokers would do, then
of an alternative that must be foregone income on cigarettes, whereas this over 10.5 million fewer people would
to purchase a particular product, in this percentage was only 1.65% for the have been malnourished and about
case tobacco. Spending on tobacco households in the highest income half of them would be from the
often constitutes a significant part quintile (Onder, 2002). Siahpush hard core poor group (consuming
of smoking households’ budgets. Its (2003) reported that average tobacco below 1805 calories/day). More
share in smoking households’ budgets expenditure as a percentage of total strikingly, if parents were to use their
varies between close to 1 percent in household expenditures varied from tobacco expenditure for feeding
areas such as Mexico and China’s 7.7% in the lowest income quintile to their children, it would save 127 750
Hong Kong Special Administrative 2.4% in the highest income quintile children under 5 years from death
Region to around 10 percent in in Australia. Using the 1998–99 from malnourishment each year in
Zimbabwe and China (John, 2008). Household Spending Surveys, Bangladesh alone.
2
Only selected papers are reviewed in this section.
261
Table 7.1. Smoking prevalence rates by socioeconomic status
Publication (location, author, year) Data sets and year Prevalence rate (%)
Low-income and lower-middle-income countries

China Household survey from rural areas in 3 By Education
(Hu and Tsai, 2000) Chinese provinces: Henan, Fujian and Shanxi, Illiterate
conducted during 1996–1998 24.1
Primary school
32.8
Junior high school
32.5
Senior high school or higher
37.1
By Occupation
Farmer
32.2
Worker
34.1
Other
26.2
Student or housekeeper
12.6
Bangladesh The National Health and Demographic Survey, By monthly household income (currency:
(Efroymson et al., 2001) 1995 US$)
<24
58.2 (M)
24–30
56.7 (M)
30–35
54.4 (M)
35–47
53.7 (M)
47–59
45.6 (M)
59–71
46.1 (M)
71–94
38.4 (M)
94–118
36.3 (M)
118+
32.3 (M)
Bangladesh Bangladesh Bureau of Statistics Survey, 1995 By monthly household income (currency:
(Ali et al., 2003) Taka)
<1000
58.2 (M)
1000–1499
56.7 (M)
1500–1999
53.7 (M)
2000–2499
45.6 (M)
2500–2999
46.1 (M)
3000–3999
38.4 (M)
4000–4999
36.3 (M)
5000+
32.3 (M)
By Education
No education
61.4 (M)
262
Bangladesh Class I-V

(Ali et al., 2003) (contd) 40.7 (M)
Class VI-IX
31.3 (M)
At least secondary school
25.0 (M)
By monthly household income (currency:
Taka)
<1000
53.0(M)
1000–1499
48.1 (M)
1500–1999
47.2 (M)
2000–2499
44.2 (M)
2500–2999
42.4 (M)
3000–3999
38.7 (M)
4000–4999
34.4 (M)
5000+
31.7 (M)
By Education
No education
59.5 (M)
Class I-V
34.9 (M)
Class VI-IX
28.5 (M)
At least secondary school
31.6 (M)
India National Family Health Survey-2 (1998–1999) By residence
(Rani et al., 2003) Urban
21.4 (M);; 0.8 (F)
Rural
32.5 (M);; 3.0 (F)
By Household wealth
Richest 20%:
16.0 (M);; 0.5 (F)
2nd richest
25.7 (M);; 1.3 (F)
Middle
31.6 (M);; 2.4 (F)
2nd poorest
36.4 (M);; 3.3 (F)
Poorest 20%
39.8 (M);; 4.9 (F)
By years of schooling
11+ years
12.8 (M);; 0.1 (F)
6–10 years
22.3 (M);; 0.3 (F)
1–5 years
36.9 (M);; 0.9 (F)
No education
45.3 (M);; 4.1 (F)
India The 50th and 55th Round National Sample By household income
(John, 2005) Survey, July 1993–June 1994 and July 1999– Smoke tobacco
June 2000 /RZHVWth percentile)
32.5 (R);; 28.7 (U)
Middle (30–70)
44.4 (R);; 30.5 (U)
263
India Higher (>70th percentile)

(John, 2005) (contd) 45.6 (R);; 27.3 (U)
Smokeless tobacco
/RZHVWth percentile):
32.7 (R);; 17.0 (U)
Middle (30–70)
29.5 (R);; 14.7 (U)
Higher (>70th percentile)
24.7 (R);; 10.9 (U)
Total
/RZHVWth percentile)
59.1 (R);; 42.6 (U)
Middle (30–70)
65.7 (R);; 41.5 (U)
Higher (>70th percentile)
60.8 (R);; 34.5 (U)
Nepal Nepal Smoking Behaviour Survey, November– By illiteracy
(Karki et al., 2003) December 2000 Literate
31.9 (M);; 7.5 (F)
Illiterate
59.9 (M);; 38.6 (F)
By length of education
Illiterate
50.6
<6 years
45.8
6–12 years
32.7
>12 years
32.9
Upper-middle-income countries
Bulgaria Living Standards Measurement Study By household income group
(Sayginsoy et al., 2002) Household Survey, 1995 Lower and lower middle
32.3
Upper middle
43.5
High
51.4
South Africa All Media and Products Survey, 1993 and 2000 By education
(van Walbeek, 2002a) Year 1993
No education
26.8
Primary education
33.6
Secondary education
31.4
Tertiary education
29.8
Year 2000
No education
23.6
Primary education
29.5
Secondary education
27.0
Tertiary education
25.7
By household income (currency: Rand (R))
Year 1993
R1–R499
29.4
264
South Africa R500–R899

(van Walbeek, 2002a) (contd) 30.6
R900–R1399
31.8
R1400–R2499
31.6
R2500–R3999
34.2
R4000–R6999
35.2
R7000–R11999
33.0
R12000+
28.1
Year 2000
R1–R499
23.5
R500–R899
23.2
R900–R1399
25.7
R1400–R2499
29.8
R2500–R3999
29.3
R4000–R6999
30.7
R7000–R11999
35.1
R12000+
31.0
Turkey Household Expenditure and Consumption By household income
(Onder, 2002) Survey, 1994 Lowest
52.0
2nd quintile
62.6
3 rd quintile
65.1
4th quintile
68.2
Highest
70.3
United Kingdom Central Statistical Office, 1982 By social class
(Townsend, 1987) Professional
20 (M);; 21 (F)
Employers and managers
29 (F)
Skilled manual
40 (F)
Semi-skilled manual
40 (F)
Unskilled working
49 (M);; 40 (F)
USA National Health Interview Survey, 1976–1980, By real family income
(Farrelly et al., 2001) 1983, 1985, 1987–1993 0HGLDQLQFRPH
31.7
0HGLDQLQFRPH
27.5
265
USA National Health Interview Survey, 2000 By education level

(Barbeau et al., 2004) 0–12 grade, no diploma < 24 y
30.9
0–12 grade, no diploma > 24 y
36.7
12 th grade, diploma
31.9
GED diploma
53.1
Some college/assoc degree
24.2
\HDUFROOHJHGHJUHH
12.5
By income/poverty
Poor
34.7
Near poor
34.2
Middle income
31.4
Higher income
20.7
Unreported income
25.5
By occupational class
White collar
20.3
Service workers
31.1
Farm workers
24.2
Blue collar
35.4
Not in labour force
27.2
Australia Household Expenditure Survey, 1998–1999 By household income
(Siahpush, 2003) Lowest
33.0
2nd quintile
31.6
3 rd quintile
33.6
4th quintile
35.0
Highest
32.8
By education
No qualification
36.3
Diploma
34.0
Degree
19.8
Unknown
46.5
By occupation
Blue collar
44.0
White collar
38.4
Professional
27.6
266
Australia Triennial cross-sectional national studies of By socioeconomic status, aged 12–15 years
(White et al., 2008) representative random samples of secondary Lowest
students aged 12–17 years, 1987–2005 15 (1987);; 8 (2005)
Second
16 (1987);; 7 (2005)
Third
16 (1987);; 6 (2005)
Highest
16 (1987);; 5 (2005)
By socioeconomic status, aged 16–17 years
Lowest
25 (1987);; 16 (2005)
Second
25 (1987);; 17 (2005)
Third
30 (1987);; 17 (2005)
Highest
30 (1987);; 15 (2005)
Italy National Health Interview Surveys, 1980 By education
(Federico et al., 2004) Aged 25–49 years
Elementary
64.7 (M);; 17.2 (F)
Lower secondary
65.2 (M);; 30.9 (F)
Upper secondary
61.6 (M);; 37.7 (F)
Post-secondary
53.6 (M);; 40.5 (F)
Aged 50–79 years
Elementary
53.6 (M);; 6.2 (F)
Lower secondary
48.3 (M);; 19.4 (F)
Upper secondary
49.8 (M);; 22.8 (F)
Post-secondary
42.9 (M);; 25.3 (F)
National Health Interview Surveys, 1999–2000 By Education
Aged 25–49 years
Elementary
50.2 (M);; 24.6 (F)
Lower secondary
45.0 (M);; 27.1 (F)
Upper secondary
34.2 (M);; 24.6 (F)
Post-secondary
26.9 (M);; 21.7 (F)
Aged 50–79 years
Elementary
25.6 (M);; 9.2 (F)
Lower secondary
26.0 (M);; 16.5 (F)
Upper secondary
24.4 (M);; 19.2 (F)
Post-secondary
24.4 (M);; 20.0 (F)
EU countries (Austria, Belgium, Denmark, The fifth wave of the European Community By education
Germany, Greece, Finland, Ireland, Italy, Household Survey, 1998 Highest
Portugal, Spain, United Kingdom) 22 (M);; 19 (F)
(Huisman et al., 2005) Middle
33 (M);; 22 (F)
Lowest
40 (M);; 28 (F)
267
EU countries (Austria, Belgium, Denmark, By household income

Germany, Greece, Finland, Ireland, Italy, Lowest
Portugal, Spain, United Kingdom) 39 (M);; 25 (F)
(Huisman et al., 2005) (contd) 2nd quintile
34 (M);; 22 (F)
3 rd quintile
32 (M);; 21 (F)
4th quintile
30 (M);; 20 (F)
Highest
27 (M);; 19 (F)
Japan Comprehensive Survey of the Living By household income
(Fukuda et al., 2005) Conditions of People on Health and Welfare, Lowest
2001 (aged 18–54) 58.8 (M);; 23.6 (F)
2nd quintile
61.0 (M);; 20.1 (F)
3 rd quintile
57.5 (M);; 16.9 (F)
4th quintile
57.3 (M);; 13.7 (F)
Highest
52.4 (M);; 11.6 (F)
By Employment status
Unemployed
33.2 (M);; 14.6 (F)
Employed
59.2 (M);; 17.5 (F)
By residential area
Non-urban
57.8 (M);; 15.8 (F)
Urban
52.6 (M);; 21.6 (F)
Republic of Korea Social Statistics Survey, 1999 and 2006 Year 1999
(Khang et al., 2009) By education
College or higher
61.7 (M);; 1.9 (F)
High school
73.1 (M);; 2.8 (F)
Middle school or less
74.3 (M);; 3.7 (F)
By Income quartile
I Highest
67.1 (M);; 2.8 (F)
II
70.0 (M);; 1.9 (F)
III
71.4 (M);; 3.1 (F)
IV Lowest
72.7 (M);; 4.5 (F)
Non-manual
62.3 (M);; 2.0 (F)
Manual
74.1 (M);; 4.1 (F)
Others
70.7 (M);; 2.6 (F)
Year 2006
By education
College or higher
49.1 (M);; 1.6 (F)
High school
62.3 (M);; 4.0 (F)
268
Republic of Korea Middle school or less

(Khang et al., 2009) (contd) 64.7 (M);; 7.9 (F)
By income quartile
I Highest
51.4 (M);; 2.2 (F)
II
55.3 (M);; 2.4 (F)
III
57.7 (M);; 3.5 (F)
IV Lowest
61.5 (M);; 4.8 (F)
Non-manual
48.3 (M);; 1.5 (F)
Manual
61.6 (M);; 4.7 (F)
Others
59.1 (M);; 2.9 (F)
HH, M and F represent household, male and female prevalence rates respectively. Similarly, R and U stand for rural and urban areas.
269
Table 7.2. Tobacco expenditure as a percentage of total household expenditure/income by socioeconomic status of
households
Publication (location, author, year) Tobacco expenditure as % of household expenditure/income
Low and middle-income countries

Turkey % of income by income quintile
(Onder, 2002) (1) 4.25
(2) 2.87
(3) 2.52
(4) 2.17
(5) 1.65
Egypt % of household expenditure by expenditure level (Currency: LE)
(Nassar, 2003) í/(
3.60 (R);; 1.11 (U)
3200- LE
1.69 (R);; 1.04 (U)
5600- LE
0.95 (R);; 0.48 (U)
10 000- LE
0.62 (R);; 0.26 (U)
14 000- LE
0.50 (R);; 0.15 (U)
Myanmar % of household expenditure by income quintile
(Kyaing, 2003) (1) 3.77 (R);; 5.12 (U)
(2) 3.31 (R);; 4.03 (U)
(3) 1.83 (R);; 2.52 (U)
(4) 1.68 (R);; 2.64 (U)
(5) 1.35 (R);; 2.09 (U)
Nepal % of household expenditure by income quintile
(Karki et al., 2003) (1) 4.4
(2) 5.7
(3) 3.5
(4) 3.3
(5) 2.3
Sri Lanka % of household income by income quintile
(Arunatilake and Opatha, 2003) Overall: 1.5
(1) 3.2
(2) 3.2
(3) 2.8
(4) 2.2
(5) 0.6
China % of household expenditure
(Hu et al., 2005) Poor
11.3 (R);; 6.6 (U)
Near-poor
9.9 (R);; 6.7 (U)
Non-poor
8.4 (R);; 9.1 (U)
% of household income
Poor
7.1 (R);; 5.8 (U)
Near-poor
8.9 (R);; 5.9 (U)
Non-poor
5.7 (R);; 4.6 (U)
Indonesia % of household expenditure
(Adieotomo et al., 2005) Low-income
7.24
Middle-income
5.53
High-income
3.02
270
Indonesia % of total monthly expenditure by income quintile for households with smokers
(Barber et al., 2008) (1) 11.9
(2) 12.3
(3) 12.4
(4) 11.7
(5) 9.2
South Africa % of household income by income quartile for urban smoking households
(van Walbeek, 2005) Cigarette
1990
(1) 1.71
(2) 1.54
(3) 0.96
(4) 0.49
1995
(1) 1.79
(2) 1.29
(3) 1.06
(4) 0.66
2000
(1) 3.17
(2) 2.84
(3) 2.61
(4) 1.53
All tobacco products
1990
(1) 1.74
(2) 1.57
(3) 0.99
(4) 0.51
1995
(1) 1.68
(2) 1.29
(3) 1.06
(4) 0.66
2000
(1) 2.87
(2) 2.71
(3) 2.57
(4) 1.55
Viet Nam % of household expenditure by income quintile
(Kinh et al., 2006) (1) 5.29
(2) 4.58
(3) 4.30
(4) 4.06
(5) 3.60
Mexico % of total household expenditure by income quintile
(de Miera Juarez et al., 2007) (1) 5.5
(2) 4.0
(3) 3.7
(4) 4.7
(5) 2.5
USA % of income by income quartile
(Gruber and Koszegi, 2002) (1) 0.032
(2) 0.014
(3) 0.009
(4) 0.004
% of expenditure by consumption quartile
(1) 0.015
(2) 0.013
(3) 0.010
(4) 0.004
271
Table 7.2. Tobacco expenditure as a percentage of total household expenditure/income by socioeconomic status of
households
USA % of income by education groups

(Gruber and Koszegi, 2002) (contd) High school dropouts
0.014
High school graduates
0.013
Some college education
0.008
College graduates
0.003
USA % of income among smokers by income group
(Colman and Remler, 2008) Low
7.7%
Medium
3.1%
High
1.4%
Canada % of income by income quartile
(Gruber et al., 2003) (1) 4.14
(2) 2.16
(3) 1.72
(4) 1.01
% of expenditure by expenditure quartile
(1) 2.28
(2) 1.82
(3) 1.43
(4) 0.93
(R) and (U) stand for rural and urban respectively
Wang et al. (2006) examined the entertainment and a rise in the the cost of smoking. Cutler et al.
opportunity cost of smoking in China budget share of health care in both (2002) estimated that on average
by estimating the impact of tobacco rural and urban India. He reported smokers lived six years less than
spending on expenditure on other that the change in budget shares was comparable non-smokers because
goods, controlling for demographic similar in both low-income and high- of the increase in heart and lung
characteristics of households. They income households. diseases.
found that tobacco spending was In addition to losses due to
negatively and significantly related Tobacco use and poverty premature deaths and increases
to spending on food, education, in healthcare cost, smoking may
farming, medical care and durable The poor smoke more than the rich reduce the earnings and wages
goods. John (2008) compared the in many countries, and they are of individuals. Smokers may earn
expenditures on several categories of observed to have more reasons to less than non-smokers for several
food and non-food items by tobacco- adopt the practice of smoking as a reasons: smoking may reduce
consuming and non-consuming part of their life (Bobak et al., 2000). productivity, and smokers may
households in India. He found that However, smoking can make people be costly to their employers with
tobacco-consuming households had poor—it adversely affects the income increased absenteeism, higher
a lower share of their income spent on level of individuals or households health and insurance premiums,
the consumption of milk, education, in several ways in addition to the higher maintenance cost and
clean fuels and entertainment, but a decline in the availability of income negative effects on morale. Levine
higher share devoted to health care, for other goods and services and its et al. (1997) compared the wages
clothing and fuels. He estimated that opportunity cost. Evidence shows of continuous smokers and those
an increase in tobacco expenditure that increased premature deaths of the workers who quit smoking
leads to a fall in the budget share and healthcare cost associated with in the USA, and estimated that
devoted to food, education and smoking-related illnesses increase the former group had about 4–8%
272
lower wages than the latter group, urban areas and 6.3 million from Another concern for any tax policy
after controlling for observable rural areas) fall below the poverty is the principle of horizontal equity,
and unobservable time-invariant line. Moreover, they predicted that i.e. the principle that individuals who
characteristics. Similarly, Heineck smoking-related expenses pushed have similar ability to pay should pay
and Schwarze (2003) reported that 41.8 million low-income people into the same or similar amount of taxes.
smoking workers have 2–8% lower poverty (24.7 million urban residents Tobacco taxation seems to violate
earnings compared to non-smoking and 17.1 million rural residents). this principle as well, since otherwise-
workers aged between 25–55 All of these findings indicate that a identical people who consume
years in Germany. The evidence reduction in tobacco use would not different quantities of tobacco
from Netherlands and Canada also only improve public health but also products will be taxed differently.
indicated lower earnings because of would reduce global poverty. As is observed in high- and
smoking. Van Ours (2003) estimated low-income countries (but not
that the wages of male smokers were Regressivity of tobacco taxes higher middle-income countries),
10% lower than those of identical the prevalence of tobacco use
non-smokers in the Netherlands. Since those individuals with low is inversely related to income
Auld (2005) found that the difference income are more likely to smoke and level. Hence, the regressivity of
was 24%. Lokshin and Beegle spend a higher share of their income tobacco taxes is exacerbated. In
(2006) report that male smokers on tobacco compared to individuals the few countries where tobacco
earn 20% less than similar non- with high income, when increasing use increases with income levels,
smokers in Albania, after controlling tobacco taxes are discussed, the tobacco taxes can be considered
for observed characteristics and dispute focuses on equity. The to be less regressive, even though
unobserved heterogeneity in debate considers whether taxes tobacco taxes as a share of income
personal characteristics. make the income distribution more declines as income level increases.
All of these studies indicate that or less equal. If the tax is paid only This definition can be considered as
smoking can reduce the income by the wealthy, it will reduce income an accounting method to assess the
level of individuals. Moreover, inequality. On the other hand, if the tax burden (Remler, 2004).
smoking-attributable deaths and tax is paid by the poor, the post-tax As a second definition of tax
smoking-related diseases reduce income disparity widens, increasing burden, Fullerton and Rogers (1993)
the productivity of individuals as well. the inequality. If a tax raises income considered the welfare effect. They
Furthermore, smoking-attributable inequality, it is regressive, whereas if argued that the utility or the welfare
medical expenses reduce the income it reduces inequality, it is progressive. of the individual will decline because
available for other expenditures. For If the taxes paid are an equal share of the decline in the consumption
example, Liu et al. (2006) reported of income for all income groups, tax of the cigarettes with the increase
that on average, smokers spend 45% is said to be neutral and it does not in price. A third method to assess
and 28% more on medical services influence income redistribution. the tax burden was developed by
than non-smokers in urban and rural The notion of vertical equity is one behavioural economists (Gruber &
China, respectively, in 1998. approach to measure the regressivity Koszegi, 2004). According to this
Liu et al. (2006) further showed of taxes. The basic tenet of vertical definition, smokers do want to quit
that even though high-income equity is as follows: Individuals smoking but have difficulty doing
people had spent more on smoking- with the greater ability to pay taxes so because they have a conflict
attributable medical expenditures should be taxed more. The example between their desire for nicotine
than low-income individuals in is an income tax system which has and their desire for good health. The
absolute terms, the percentage of the characteristic that the tax rate model is described in Chapter 4. The
income spent on these types of increases with the amount of income tax increase helps them to quit so
expenditures was 6.5% for the lowest- earned. However, taxes on tobacco they are doing what they are willing to
income quintile group and only 1.5% and tobacco products seem to do in the long run. This is called time-
for individuals in the highest-income violate this notion, because in many inconsistent preferences (Gruber
quintile in urban China. They also countries the tobacco use prevalence and Koszegi, 2004). If the poor are
estimated that because of medical rate and expenditure on tobacco as willing to quit but have difficulty, the
expenditures attributable to smoking, a percentage of income are higher tax increase will help them to quit,
12.1 million people (5.8 million from among low-income individuals. thus reducing the tax burden.
273
Remler (2004) summarized the
eventual gains of quitting outweigh costs

Time-inconsistent welfare-based tax
Worse off owing to higher expenditures

Worse off owing to cigarette consumption Better off owing to commitment device:
commitment device;; eventual gains of

effects of higher cigarette taxes
on three types of smokers using
quitting outweigh cost of quitting

these three methods to assess the
Somewhat better off owing to

tax burden (Table 7.3). The first
smoker (Smoker A) responds to the
Table 7.3. Effects of higher cigarette taxes on three types of smokers, measured under three alternative methods of assessing tax burden
increase in taxes by quitting;; the
Source: Used with permission of DK Remler, Poor smokers, poor quitters, and cigarette tax regressivity. American Journal of Public Health, 94:225–229, 2004. Copyright American Public Health Association.
on cigarette second smoker (Smoker B) does
not change his/her consumption of
of quitting
burden
cigarettes after the tax increase;;

the third smoker (Smoker C)
reduces the cigarette consumption
Worse off owing to both higher price paid
Overall worse off since not compensated
Willingness-to-pay welfare-based tax
to keep his/her tax expenditure the

Worse off owing to higher expenditures
per cigarette and cutting back cigarette
same. Smoker A is better off with

quitting because of the decline in
Better off owing to lower tax bill
the share of tobacco expenditure

in total expenditures according to
for being forced to quit
the accounting definition of tax

burden, but worse off because the
decline in cigarette consumption
consumption
on cigarette
might reduce the utility according

decrease
burden
to willingness-to-pay welfare-based
tax burden, and better off since s/
he accomplishes quitting according
Accounting (income share) tax burden
to time-inconsistent definition. The

smoker who does not change his/
her consumption pattern (smoker
B) will be worse off under three
measures of regressivity. Smoker
C, who reduces consumption so
Same as before
that there will be no change in tax

share, will be worse off based on
Worse off
Better off
willingness-to-pay welfare-based
tax burden and better off according
to the time-inconsistent welfare-
based tax burden measure.
Cuts back to keep tax expenditures
Average versus marginal

progressivity/regressivity
Response to tax increase
of increases in tobacco tax

No change in smoking
To determine whether a tax is

progressive or regressive using the
accounting measure, the share of
constant
taxes paid from a person’s income

Quits
must be calculated. Let ut represent

the unit excise tax on tobacco and
Q and Y be the quantity of tobacco
consumed and income levels
Smoker C
Smoker A
Smoker B
respectively. Then, an individual

pays (ut*Q)/Y fraction of his/her
274
income as tax. If this ratio does not has an income of Y (low-income) Colman and Remler (2008) examined
change with income, then tax is and the other one earns three times the regressivity of increases in excise
said to be neutral. If it increases as higher income (3Y). Suppose that taxes on cigarette products in the
income level declines, then taxes the low-income smoker has a higher USA using both tax-expenditure
are considered to be regressive, SULFH UHVSRQVLYHQHVV í WKDQ based (share of tax paid in income)
meaning that the poor pay a larger WKHKLJKLQFRPHVPRNHUKDVí and welfare (using consumer
proportion of their income in the and 50% of price is tobacco tax. If surplus estimates) measures of
form of tobacco taxes. On the other the price of cigarettes (p) increases regressivity. They claimed that
hand, if this ratio increases as the as result of doubling taxes (assume cigarette tax increases are not close
income level increases, taxes are that it increases from P = 1 to P = 1.5 to being progressive, regardless
defined as progressive, since the rich and all of the increase is transferred of the measures used. Moreover,
pay a greater share of their income to consumers), then total tax paid according to the results of the recent
as tobacco taxes. This measure will rise from 0.5x/Y to 0.6x/Y (=(1– study, Gospodinov and Irvine (2009)
of progressivity (or regressivity) is 0.80*0.50) x/y) for the low-income concluded that in Canada “there is
defined as average progressivity (or smoker and from 0.167x/Y (= 0.5x/3Y) little reason to overturn the traditional
regressivity) because it examines to 0.3x/Y (=(1–0.20*0.50) x/3y) for concerns about regressivity.”
the fairness of total taxes. Because the high-income smoker at the new Studies from low- and middle-
the poor have lower incomes and consumption levels. The increase will income countries also observed that
consume more tobacco in general, be 0.1x/Y for the low-income smoker the burden of current excise taxes
we can say that tobacco taxes are and 0.133x/Y for the high-income falls heavily on poor smokers. In
regressive. smoker. Hence, although taxes are these countries, the share of cigarette
Although tobacco taxes may be regressive at the beginning, the expenditure in household income or
regressive on average, they may not tax increase is less regressive and expenditure varies between 4.25%
be marginally regressive, as pointed total regressivity of tobacco taxes is and 7.2%. The tax burden decreases
out by Warner (2000), who used an reduced with the increase in tax. by income, with the share in rich
alternative approach to assess the households ranging from 1.65% to
regressivity of tobacco taxes. He Evidence about regressivity 3% (Onder, 2002;; Arunatilake and
argued that tobacco taxes may not of tobacco taxes Opatha, 2003;; Karki et al., 2003;;
be regressive because if the poor are Adioetomo et al., 2005). Excise taxes
more sensitive to price increases than Incidence analyses of tobacco were found to be more regressive
the rich, then an increase in cigarette excises have been extensive in high- mainly due to the high smoking
taxes and a resulting increase in income countries, notably in the USA prevalence rate or low income level
cigarette prices make the poor quit and United Kingdom (Browning, among the poor. However, results are
smoking or reduce their consumption 1978;; Townsend, 1987;; Congress mixed in terms of the regressivity of
more than the rich would do. This of the United States, Congressional excise tax increases.
could shift the burden of tax from the Budget Office, 1990, 2001;; Borren Studies using a budgetary
poor to the rich. Hence, an increase in and Sutton, 1992;; Fullerton and approach in estimating the
cigarette taxes may turn out to reduce Rogers, 1993;; Townsend et al., 1994;; progressivity of excise tax increases
the regressivity of the tax, depending Lyon and Schwab, 1995;; Evans et have supported the claim that an
on whether the poor are more al., 1999;; Viscusi, 2002;; Colman increase in excise taxes may not be
sensitive than the rich to the price. and Remler, 2004). Overall, these regressive. For example, Arunatilake
From the policy perspective, the studies found that current excises and Opatha (2003) found that current
marginal regressivity of tobacco on tobacco are regressive. Viscusi excises are regressive in Sri Lanka.
taxes is important. It is calculated (2002) argued that cigarette taxes However, this study projected that
as a ratio of the change in tobacco in the USA fell predominantly on after a 100% tax increase, the pre-
taxes paid because of the change in the very poor, so that in 1990, low- tax regressivity gap (the difference
taxes, or tax rates, to income. It can income people who earned less than between the share of household
be explained by using the example US$10 000 a year paid almost twice income spent on tobacco) would be
specified in Chaloupka et al. (2000). as much in cigarette taxes as higher- reduced between the poorest and the
Assume that there are two smokers, income people who earned US$50 richest groups, from a pre-tax level of
consuming x units of cigarettes;; one 000 and more a year. Recently, 3.2% to a post-tax level of 2.3% for
275
the poorest group and from a pre-tax by socioeconomic status of smokers was thus driven by their greater
level of 0.6% to a post-tax level of in different countries classified smoking prevalence elasticity.
0.9% for the richest group. Similarly, according to the World Bank income The CDC (Centers for Disease
Sayginsoy et al. (2002) found that classification. Control and Prevention, 1998) further
after a 72% tax increase in Bulgaria, observed that Hispanics were more
the share of cigarette expenditure in Evidence from high-income price responsive than were the
poor households was reduced from countries black and white populations, both
a pre-tax level of 4.9% to 3.96%. In with respect to smoking prevalence
fact, the budget burden of cigarette Chaloupka (1991) was the first to and smoking intensity. Blacks
tax was observed to be more equal investigate the price responsiveness were more price responsive than
among groups: 3.96%, 3.45% of different socioeconomic groups’ whites with respect to smoking
and 3.43% from the lowest to the tobacco consumption. Using prevalence, and less so with respect
highest income group respectively. individual-level data from the Second to smoking intensity. To the extent
In another study, Rajemison et al., National Health and Nutrition that race and ethnicity are proxies for
(2003) showed that excise taxes Examination Survey conducted in socioeconomic status of population
on cigarettes were progressive, the late 1970s, Chaloupka estimated groups in the USA, the greater
but that this result was driven by that the price elasticity of cigarette price responsiveness of the minority
the peculiarities of the Madagascar demand was greater for individuals (Hispanic and Black) is attributable
market. Cigarettes, which are taxed, with less than high school education to differences in their socioeconomic
are consumed by the better-off;; while UDQJLQJ IURP í WR í WKDQ status. Subsequent studies by Biener
another tobacco product (parakay), those with at least high school et al. (1998), Evans et al. (1999),
which is not taxed, is consumed education, who were found to be Hersch (2000), Farrelly et al. (2001),
widely by the poor. unresponsive to price changes. Gruber and Koszegi (2002;; 2004),
Hence, to determine whether Separate estimates of price Stehr (2007), and DeCicca and
tax increases increase the burden elasticity of smoking prevalence and McLeod (2008) confirm the negative
on the poor, it is important to know smoking intensity were not available relationship between socioeconomic
how the poor and the rich will change until 1998. Pooling data for 14 years status and price responsiveness of
their consumption because of the (1976–1980, 1983, 1985 and 1987– tobacco demand.
increase in prices. In other words, 1993) from the US National Health In a recent study using data from
the regressivity of tax increases Interview Survey, CDC (Centers for the 1984–2004 Behavioural Risk
depends on the price sensitivity of Disease Control and Prevention, Factor Surveillance System surveys
the poor and the rich. 1998) showed that smoking for the US, Franks et al. (2007)
prevalence among people below found that the elasticity of smoking
Differences in price sensitivity the median income level was more prevalence with respect to cigarette
by socioeconomic status price responsive than among people price was larger among the lowest
above the median income level. LQFRPHJURXSíFRPSDUHGWRWKH
Studies based on aggregate data The converse was true for smoking KLJKHU LQFRPH JURXSV í LQ WKH
are limited by the fact that the intensity—the number of cigarettes pre-Master Settlement Agreement
estimates are not obtainable by smoked per day by smokers above (MSA) period (1984–1996), whereas
income, education, or any type of the median income level was found in the post-MSA period (1997–2004),
indicator of socioeconomic status of to be more price elastic than that for smoking prevalence became price
individuals who consume tobacco smokers below the median income insensitive for all income groups.
products. Hence, studies estimating level. However, the overall price The study concluded from these
price elasticity of tobacco demand by elasticity was greater for people in results that in the USA, “increasing
socioeconomic status are based on WKH ORZHU HFRQRPLF FODVV í cigarette prices may no longer be an
individual or household-level data.3 than for people in the upper economic effective policy tool and may impose
Table 7.4 reports the estimates of FODVV í 7KH JUHDWHU SULFH a disproportionate burden on poor
price elasticity of cigarette demand responsiveness of low-income people smokers.”
3
One exception is the study on UK by Townsend et al. (1994) that used aggregate level data to obtain estimates of price elasticity by socioeconomic groups.
276
Table 7.4. Price elasticity of tobacco use by income/SES/education
Methods and model Comment
Low-income countries
Myanmar Myanmar Study on Tobacco Overall Overall Overall Dependant variable:
(Kyaing, 2003) Economics 2001 í í í Monthly consumption of
- OLS estimation of the price By income quintile By income quintile By income quintile cigarettes, cheroots and phet
equation í í í kyan (per stick)
- Logit model (with í í í Independent variables:
independent variables in í í í price of cigarettes, cheroots
log) to estimate probability í í í or phet kyan;; per capita
of smoking in the household í í í monthly income;; tax on
(extract from their price cigarette/cheroots;; dummies
elasticity of smoking for age, sex, literacy, marital
prevalence) status, addiction, education
- OLS estimation method for level, place of residence
the consumption equation.
Total price elasticity
= (1-probability of
smoking)*(price elasticity of
smoking prevalence) + price
elasticity of smoking intensity
Myanmar Survey of low-income Price elasticity of demand Dependent variable:
(Kyaing et al., 2005) consumers By income (currency unit: consumption of cheroots
Log-log OLS estimation of Kyat) * P
conditional cigarette demand Group 1 ** P
function
Group 2
24 000–45 000
Group 3
45 000 – 75 000
Group 4

Cheroots
2YHUDOOí
í
í
í
í
Cigarettes
2YHUDOOí
í
í
í
í
Cigarettes and cheroots
2YHUDOOí
í
í
í
277
í
278
Nepal Smoking Behaviour Survey, Overall Overall Overall Dependant variable

(Karki et al. 2003) November–December 2000 í í í - Monthly consumption of
- OLS estimation of the price By income quintile By income quintile By income quintile cigarettes and bidis (per
equation í í í stick)
- Logit model (with í í í Independent variables
independent variables in log) í í í price of cigarettes or bidis,
to estimate the probability í í í per capita monthly income,
of smoking of the household í í í cigarette and bidi tax,
(extract from there price dummies for: age, sex,

elasticity of smoking occupation, marital status,
prevalence). addiction, education level,
- OLS estimation method for place of residence
the consumption equation.
Total price elasticity
= (1-probability of
smoking)*(price elasticity
of smoking prevalence) +
conditional price elasticity of
smokers
Viet Nam 1997–1998 Overall Overall Dependent variable: dummy
(Kinh et al. 2006) Household level cross- í í variable for smoking status
sectional data from Viet Nam By income By income Independent variables:
Living Standards Survey Two low quintiles Two low quintiles Price of cigarettes, price of
-Linear probability model of í í piped tobacco, annual per
smoking status Two high quintiles Two high quintiles capita income, individual,
-Double log model of í í household, geographic and
cigarette consumption commune characteristics
conditional of smoking
participation
Bangladesh International Tobacco Control Cigarette demand Cigarette demand Cigarette demand Dependent variable: Smoking
(Nargis et al. 2010) Policy Evaluation Bangladesh Overall Overall Overall participation, daily cigarette
Survey, 2009 í í í (bidi) consumption
Two-step method: By socioeconomic status By socioeconomic status By socioeconomic status Independent variables:
(1) Logit estimation of Low Low Low Price per pack of cigarette
smoking prevalence í í í (bidi), level of education,
(2) OLS estimation of Medium Medium Medium marital status, number of
cigarette (bidi) consumption í í í years of smoking since
of smokers High High High initiation, village dummy,
í í í urban/rural residence,
Bidi demand Bidi demand Bidi demand socioeconomic status
Overall Overall Overall represented by CASHPOR
í í í Housing Index
By socioeconomic status By socioeconomic status By socioeconomic status * P
Low Low Low ** P
í í í *** P
Bangladesh Medium Medium Medium:

(Nargis et al. 2010) (contd) í í í
High High High
í í í
Lower-middle-income countries
Sri Lanka Household level data Overall Dependent variable: per
(Arunatilake, 2002) 1999/2000 í capita monthly tobacco
First step: logit model By expenditure quintiles consumption Independent
estimating the smoking í variables: tobacco price per
prevalence í unit of tobacco, monthly per
Second step: OLS estimation í capita income, male ratio,
of the tobacco consumption (4) 0.03 occupation, education,
Total price elasticity derived (5) 0.12 age category, and location
from the elasticity estimates (equations divided by
of the smoking prevalence expenditure groups, quintiles)
and conditional demand *P
equations
Sri Lanka Srilanka Integrated Survey Overall Overall Overall Dependent variable:
(Arunatilake and Opatha, data of households, í í í -household decision to smoke
2003) 1999–2000 By expenditure quintile By expenditure quintile By expenditure quintile - per capita monthly
Two-part model: í í í consumption of cigarettes
1 – logit model of household í (2) 0.17 í Independent variables: real
decision to smoke í (3) 0.21* í price of cigarettes, real per
2 – OLS estimation of í (4) 0.01 í capita income, ratio of adult
conditional demand equation í (5) 0.34** í males to adult females,
for participating households occupation, education, age,
location
*P
**P
***P
Ukraine (Krasovsky et al., June 2001 national Overall Dependent variable: number
2002) representative survey í of cigarettes consumed
1 – OLS estimation of the 1 – For adults aged 18 to 28 (cigarette expenditure divided
demand equation by income group by cigarette prices)
2 – Binary regression using Low Independent variables: real
the ML method to estimate í price of cigarettes, household
the prevalence equation Medium income, age, sex, strength of
model í addiction, region dummies,
High dummy reflecting whether
í a smoker had underage
2 – Aged 29+ years children
Low
í
Medium
í
High
í
279
280
China 1998 National tobacco survey Overall

(Mao et al., 2003) in 8 cities and 16 counties í
Low-income
í
Middle-income
í
High-income
0.51
China Overall
(Mao et al., 2007) í
Poor
í
Low-income
í
Middle-income
–0.018
High-income
0.257
Taiwan, China Cross-sectional data from By education Dependent variable:
(Lee, 2008) 483 respondents of a College and above willingness of current
telephone survey of current í smokers to quit smoking or
smokers (15+ years) from all High school reduce cigarette consumption
23 major cities and counties, í when faced with a tax
collected in April–July 2004 Junior high school increase of NT$22 per pack,
Tobit regression model í which would raise the price of
By monthly income cigarettes by 44%
(currency unit: NT$) Independent variables:
cigarette prices, monthly
í income
10 000–29 999 *P
í
30 000–49 999
–0.48*

0.18
Egypt 1995/1996 and 1999/2000 Price elasticity of tobacco Dependant variable:
(Nassar, 2003) Household expenditure By expenditure quartiles - tobacco expenditure
surveys 1995/96 Urban Rural Independent variables:
Log-log estimation of OLS on íí aggregate prices on different
the pooled data íí types of tobacco, total
íí household expenditure,
íí education, work status,
1999/00 Urban Rural occupation, urban/rural
íí regions
íí
Egypt íí
(Nassar, 2003) (contd) íí
By education status
1995/96 Urban Rural
íí
íí
íí
íí
1999/00 Urban Rural
íí
íí
íí
íí
Thailand Household socioeconomic Overall Variables included: tobacco
(Isra, 2003) survey 2000 (Consumer price í consumption, expenditure on
index from the Department By income class cigarettes and other tobacco
of Business Economics, Urban products, cigarette price,
Ministry of Commerce) í household income, prices
Linear expenditure system í of 12 consumer goods, age,
model í education Dummy: urban/
Estimation by 5 income í rural
classes in urban and rural í
areas separately Rural
í
í
í
í
í
Indonesia 1999 Social and Economic Overall Overall Overall Dependant variable: Monthly
(Adioetomo et al. 2005) Survey (SUSENAS) í í í cigarette consumption
household data By income By income By income divided by the number of
First step: logit model Low Low Low household members
estimating the smoking í í í Independent variables:
prevalence. Medium Medium Medium Cigarette price in rupiah
Second step: OLS estimation í 0.09 í per pack of 16 cigarettes
of the tobacco consumption High High High (expenditure on cigarettes
conditional on smoking í 0.20 í divided by the quantity of
participation. cigarettes consumed), per
Total price elasticity derived capita household income
from the elasticity estimates per day in rupiah, dummy for
of the smoking prevalence urban/rural location, dummies
and conditional demand for education, profession, age
equations and sex
*P
281
282
Upper-middle-income countries
Bulgaria Living Standards Overall Dependent variable:
(Sayginsoy et al., 2002) Measurement Study í Number of packs of 20
Household survey, 1995 By income smoked by the household per
2SLS log-log estimation for Low and low-middle month per capita
overall sample and by income í Independent variables:
group. High-middle average price paid for a pack
í of cigarette, total household
High income, mean age of all

í members of household, years
of education received by the
most educated household
member, litres of alcohol
consumed per capita in each
household, ratio of number
of adult males in each
household to the size of the
household, Dummy = 1 if the
household has at least one
member who is a widow(er),
divorced, living separate from
husband/wife or is older than
35 and not married
South Africa The Income and Expenditure By income quartile Dependent variable:
(van Walbeek, 2002b) household surveys of 1990 í Household’s annual
and 1995 í expenditure on cigarettes
- Log-log estimation of í Independent variables:
income elasticity for each í real household income
income quartile for 1990 and subdivided into four income
1995 (log of the cigarette quartiles, real price of
expenditure divided by the cigarettes
income and function of the
income quartiles).
- Calculation of net change
of cigarette consumption
between 1990 and 1995 (%
change in average cigarette
consumption minus the
effect of income change on
consumption)
- Estimation of price elasticity
(net change of cigarette
consumption divided by real
price change)
Turkey Household Expenditure and Overall Overall Overall Dependent variables:

(Onder, 2002) Consumption í í í - number of cigarettes
Survey, 1994 By household income By household income By household income smoked by the smoker
First step: Estimation of í í í households
the smoking participation í í í - smoking participation
decision of the households í í í decision (probability of
with a logit model for all í í í smoking)
households and by income í (5) 0.15 í - cigarette price
quintiles;; - cigarette excise tax.
estimation of the cigarette tax Independent variables:
as a function of household cigarette price per package
income and estimation of of 20, per capita household
the price for non smoker income per month, education,
households gender, age, number of adults
- Second step: Log-log in the household, location,
estimation of the conditional region, dummy: employed
demand for cigarettes by the head, dummy: white collar
smoker head
Two-stage least squares
estimation to correct for
endogeneity of the price
variable
USA (Chaloupka 1991) 1976–1980 individual- Long-run price elasticity of Dependent variable: Average
level data from the second cigarette consumption daily cigarette consumption
National Health and Nutrition Overall Independent variable:
Examination Survey íWRí Current price, past price,
Two-stage least-squares Less than a high school future price, addictive stock,
estimation education lagged consumption, future
íWRí consumption
At least a high school
education
unresponsive
USA (Centers for Disease Pooled data from 14 years Overall Overall Overall -Smoking prevalence
Control and Prevention 1998) (1976–1980, 1983, 1985, and í í í Dependent variable:
1987–1993) from the National By income By income By income prevalence of smoking
Health interview Survey for PHGLDQ PHGLDQ PHGLDQ -Smoking intensity
DJHJURXS í í í Dependent variable:
-Probit model of smoking > median > median > median cigarettes per day for
prevalence í í í smokers
-OLS estimation of By race By race By race Independent variables: age,
conditional demand model of White White White age squared, real family
cigarette consumption í í í income, family size, state of
Black Black Black residence, year, city size,
í í 0.04 race or ethnicity, educational
Hispanic Hispanic Hispanic level, marital status, gender,
í í í price of cigarettes
283
284
USA (Hersch 2000) 1992/93 Current Population Male Male Dependent variable: smoking
Surveys: Tobacco Use Overall Overall participation, cigarette
Supplement í í consumption of smokers
Participation equation By income By income Independent variables: price,
estimated by probit model Low Low family earning, education,
and smokers’ consumption í í age, age squared, marital
equation by OLS regression Medium Medium status, race, age of youngest
for men and women í í child, occupational status
High High *P
í í **P

Female Female
Overall Overall
í í
By income By income
Low Low
í í
Medium Medium
í í
High High
0.58* í
USA (Farrelly et al. 2001) Pooled data from 14 years Overall Overall Overall -Smoking prevalence
(1976–1980, 1983, 1985, and í í í Dependent variable:
1987–1993) from the National By income By income By income prevalence of smoking
Health interview Survey for PHGLDQ PHGLDQ PHGLDQ -Smoking intensity
> = 18 age group í í í Dependent variable:
-Probit model of smoking > median > median > median cigarettes per day for
prevalence í 0.01 í smokers
-OLS estimation of By race By race By race Independent variables: age,
conditional demand model of African-American African-American African-American age squared, real family
cigarette consumption í í í income, family size, state of
Hispanic Hispanic Hispanic residence, year, city size,
í í í race or ethnicity, educational
White White White level, marital status, gender,
í í í price of cigarettes
USA Cross-sectional By education Dependent variable: smoking
(Ringel and Evans, 2001) Nataly Detail Files College participation
(1989–1995) í Independent variables:
n = 20 025 000 pregnant Some college monthly state excise rates
women (14–21 years) í (that near the beginning of
Probit model High school the pregnancy) in real 1997
í prices, age, race/ethnicity,
< High school education, marital status,
í parity of birth, adequacy
prenatal care, sex of child,
state and month of conception
USA By race
(Ringel and Evans, 2001) Black
(contd) í
White
í
Hispanic
í
Others
í
USA Consumer Expenditure Price sensitivity of Dependent variable: cigarette
(Gruber and Koszegi, 2004) Survey, 1980–1998 cigarette expenditure expenditure
Instrumental variable By income quartile Independent variables: price
estimation of cigarette í instrumented by excise tax,
expenditure function í age, education, sex, race
í of the household head,
í dummies for number of
By consumption quartile persons in the household,
í state dummies, year
í dummies, calendar month
í dummies
í
By education groups
High school dropouts
í
High school graduates
í
Some college education
í
College graduates
í
USA 1984–2004 Behavioral Risk 1984–1996 Dependent variable: Smoking
(Franks et al., 2007) Factor Surveillance Lowest income quartile prevalence
System surveys í Independent variables:
Logistic regression All other income quartiles Cigarette pack price, age,
í age squared, gender, race/
1997–2004 ethnicity, years of schooling,
Lowest income quartile number of adults in the
í household, consumer price
All other income quartiles index, household income,
í dummy variables for each
survey year and each state,
Gini coefficient for each year
* P
USA Cross sectional (pooled Male Male Male Dependent variables:
(Stehr, 2007) surveys) Overall Overall Overall smoking participation,
Behavioral Risk Factor í í í cigarette consumption
Surveillance System (1985– By income quartile By income quartile By income quartile Independent variables:
2000);; sample size over 1.3 í í í gender, state fixed effects,
285
286
USA PLOOLRQDJH\HDUV í í í income, education, sex,

(Stehr, 2007) (contd) Two-part model: í í í age, ethnicity, and year of
1) probit (smoking í í í interview
participation) Female Female Female
2) log-log OLS (conditional Overall Overall Overall
demand for cigarettes) í í í
By income quartile By income quartile By income quartile
í í í
í í í
í í í

í í í
USA Current Population Survey Low income Low income Low income -Smoking prevalence
(Colman and Remler, 2008) (CPS) Tobacco Use í í í Dependent variable:
Supplements (TUS) and Medium income Medium income Medium income prevalence of smoking
March Income Supplements, í í í -Smoking intensity
six pooled cross- High income High income High income Dependent variable:
sections—1993, 1996, 1999, í í í cigarettes per day for
2001, 2002, and 2003 smokers
Two-part model: Independent variables:
First stage—Linear cigarette price, family
probability model (OLS) income, year, state indicators,
Second stage—OLS dummies for the education
regression of cigarette levels of high school, some
consumption among those college, college graduate,
who smoke and more than college (high-
school dropout is the omitted
category);;
Hispanic, non-Hispanic black,
and non-Hispanic other
race (non-Hispanic white
is the omitted category);;
female;; divorced, separated,
widowed, and never-married
(married is the omitted
category);; unemployed
and not-in-the-labour-force
(employed is the omitted
category);; the occupational
categories blue-collar,
service, farm, military, and
never-employed (white-collar
is the omitted category);;
and several continuous
variables, including age and
age squared, thenumber of
children under 6 years old
USA in the family, interactions

(Colman and Remler, 2008) between year and age and
(contd) age squared to allow time
trends to differ by age, and
an index measure of legal
restrictions (“clean air index”)
on indoor smoking
USA (DeCicca and McLeod, Cross-sectional (5 surveys) By education Dependent variable: smoking
2008) Behavioral Risk Factor Low participation
Surveillance System í Independent variables:
(2000–2005) High real monthly state-specific
435 973 adults (45–59 years í cigarette excise tax rates per
old) By income pack of 20 cig in 2001 dollars,
Telephone interview Low gender, age, race, education,
Representative sample US í income, marital status, health
population High status, unemployment rate,
Two-way fixed effects models í state ban in workplaces and
(multilevel model) restaurants
USA (Farrelly and Engelen, Pre and post- Master By income quartile Dependent variable: Smoking
2008) Settlement Agreement (MSA) 1990–1998 (pre-MSA) prevalence
estimates as reported by Overall Independent variables:
Nielsen Co. í Cigarette pack price, age,
1990–2006 Lowest age squared, gender, race/
Sample size: í ethnicity, years of schooling,
1990–1998: 782 403 Middle 2 quartiles number of adults in the
1998–2006: 1 763 952 í households, consumer price
Logistic regression Highest index, household income,
í dummy variables for each
1998–2006 (post-MSA) survey year and each state,
Overall Gini coefficient for each year
–0.09
Lowest
í
Middle 2 quartiles
í
Highest
í
United Kingdom Biennial data from British Men Men Dependent variable:
(Townsend et al., 1994) general household survey, All All prevalence of smoking,
1972–90 í í defined by the proportion of
Model of smoking prevalence Socioeconomic group (5) By socioeconomic group adults currently smoking
and cigarette consumption by –0.61* (1) 0.03 one or more cigarettes a day,
socioeconomic groups Women í and numbers of cigarettes
All 1RQPDQXDOí smoked per smoker
í 0DQXDOí *P
Socioeconomic group (5) í **P
í í
287
288
United Kingdom Women

(Townsend et al., 1994) All
(contd) í
By socioeconomic group
(1) 0.50
í
1RQPDQXDOí
0DQXDOí
í
í
Canada (Gruber et al., 2003) 1982–1998 Canadian Survey By income quartile Dependent variable: cigarette
of Family Expenditure í expenditure
Instrumental variable í Independent variables:
estimation of cigarette í price of cigarettes, region,
expenditure function í year, gender, income, income
By expenditure quartile squared, regional time trends
í
í
í
í
Canada (Gospodinov and Statistics Canada/Health Overall Dependent variable: smoking
Irvine 2009) Canada Canadian Tobacco í participation, cigarette
Use Monitoring Survey < High school consumption
(CTUMS), 2000–2005 í Independent variables: price,
Two-part model of estimation High school province of residence, level of
í education
College
í
Australia (Siahpush et al., Australian population survey By income Dependent variable: smoking
2009) data collected monthly from Low-income prevalence
January 1991 to December í Independent variables: price,
2006 Medium-income income, interaction of price
Sample size: 515 866 í with income, education, age,
LQGLYLGXDOVDJHG\HDUV High-income gender, time
Poisson regression to í
estimate monthly smoking
prevalence as a function of
covariates
The above conclusion sparked women among the white population í LQ WKH ORZHVW VRFLRHFRQRPLF
an empirical debate regarding the are more conscious of the adverse JURXS FRPSDUHG ZLWK í IRU
role of socioeconomic status in the health consequences of smoking overall female population. Similarly,
welfare consequences of tobacco compared to the minorities and are the elasticity of smoking intensity was
tax increase. Farrelly and Engelen more responsive to price increases. HVWLPDWHG WR EH DW í DQG í
(2008), for example, reanalysed the However, this result applies to the for men and women in the bottom
data used in Franks et al. (2007), pregnant women in the USA, not socioeconomic group respectively,
adding two years of data after the necessarily to the general population. FRPSDUHG ZLWK í DQG í IRU
MSA and limiting the pre-MSA Estimates of price elasticity the total male and female population.
period from 1990 to October 1998. of smoking intensity reported in The inverse relationship of price
They dropped the earlier years Gospodinov and Irvine (2009) responsiveness of tobacco demand
because only 15 US states were for Canada show greater price to the socioeconomic status of people
included in the data for 1984–1989. responsiveness among high school in the United Kingdom was further
Then replicating the same model and JUDGXDWHVíWKDQDPRQJWKRVH confirmed in Townsend (1996).
analytic approach, they found that with less than a high school education The relationship between
the size of the price effect did decline í RU DPRQJ FROOHJH JUDGXDWHV price responsiveness of smoking
in the post-MSA period, but only í VXJJHVWLQJ DQ LQYHUVH8 prevalence and socioeconomic status
among the middle- and high-income relationship between price sensitivity is, however, not monotonic. Borren
people. Thus the lowest income and level of education. However, and Sutton (1992) found evidence of
group was still found to be price- Gruber et al. (2003) found a more an ‘inverse-U’ relationship of price
sensitive, in contrast to what Franks systematic relationship between responsiveness with income status in
and his colleagues observed. This income and cigarette expenditure the United Kingdom—middle-income
disagreement reflects that the results elasticity in Canada—the estimates men were found to be more price
of estimation by socioeconomic were considerably greater in the elastic than lower- and higher-income
status were sensitive to the period lowest t wo quartiles than in the upper men. Their evidence for women,
of the data they used. Franks et al. two quartiles. however, corresponds to the more
(2008) immediately responded by In a seminal study in the United common finding that price elasticity
pointing out several inconsistencies Kingdom, using biennial data from declines as income increases.
in Farrelly and Engelen’s analysis, British general household survey Schaap et al. (2008) studied
especially the loss of sample size (1972–90), Townsend et al. (1994) the effectiveness of comprehensive
by over one million persons due to found that both smoking prevalence tobacco control policies including
dropping two years of observations. and intensity among men and women price and taxation in 18 European
A contemporary study by Colman and in lower socioeconomic groups were countries.4 They found the strongest
Remler (2008) reinforced the greater more responsive to changes in association of quit ratio with price
price elasticity for lower-income cigarette prices and less responsive policy among all the tobacco control
JURXSV í IRU WKH ORZLQFRPH to information about the adverse policies implemented in these
í IRU WKH PHGLXPLQFRPH DQG health consequences of smoking. countries, including taxes on tobacco
íIRUWKHKLJKLQFRPHJURXS The uniqueness of this study lies in products, bans or restrictions on
An earlier study by Ringel and the fact that it uses aggregated data smoking in public places, advertising
Evans (2001) found that among by socioeconomic group rather than bans, public information campaign
pregnant women in the USA, price cross-sectional level data, which spending, health warnings, and
responsiveness of cigarette demand form the basis for all other studies treatment. However, they observed
was highest among the most- in this genre. The price elasticity no significant difference in the
educated cohort and was lower the of smoking prevalence of men in impact of an increase in tobacco
lower the level of education. They also the lowest socioeconomic group price on quitting between high and
found lower price responsiveness among five groups was estimated to low education groups. Similarly, Lee
among black and Hispanic pregnant EH í ZKLOH LW ZDV í IRU WKH (2008) found no significant difference
women than among whites. Their overall male population. For women, in price responsiveness across
findings suggest that the pregnant smoking prevalence elasticity was different education groups in Taiwan,
Belgium, Czech Republic, Denmark, Estonia, Finland, France, Germany, Hungary, Ireland, Italy, Latvia, Lithuania, Netherlands, Portugal, Slovakia, Spain, Sweden and United Kingdom.
4
289
China. However, this study found Among low-income countries, not differ much with respect to the
the highest price sensitivity among evidence from Myanmar suggests price responsiveness of smoking
the smokers with the lowest monthly an inverse-U relationship of intensity of cigarette consumption,
income. price responsiveness of smoking although for bidi consumption it was
In Australia, on the other hand, prevalence to income group—the greater for high-income people.
Siahpush et al. (2009) found prevalence elasticity for the third The total price elasticity, however,
consistently greater price sensitivity TXLQWLOH í ZDV HVWLPDWHG WR EH reflected the inverse relationship
of smoking prevalence among lower- greater than those for the bottom found between smoking prevalence
income respondents. The difference í DQG WRS TXLQWLOHV í elasticity and socioeconomic status.
of price elasticity of smoking (Kyaing, 2003). The price elasticity These results indicate that price has
prevalence estimated for low-income of smoking intensity was, however, strong influence on the decision to
SHRSOHíIURPWKRVHHVWLPDWHG highest for the lowest income smoke in Bangladesh, but that the
IRUPHGLXPLQFRPHíDQGKLJK quintile, although the price elasticity effect is less obvious on the average
LQFRPHíJURXSVLVVWDUN7KLV did not decrease monotonically consumption of smokers.
study is unique in using monthly with higher economic status. A Among the lower-middle-income
data on smoking prevalence, and later study by Kyaing et al. (2005) countries, in China, low-income
presents a significant improvement found the highest price elasticity people’s tobacco consumption is
over the previous studies that among the lowest income group for more price responsive than that of
were limited by their use of annual cheroots and among the second higher-income people (Mao et al.,
prevalence data. The authors argue lowest income group for cigarettes 2003, 2007). The evidence is again
that annual data are likely to provide in Myanmar. For Nepal, as shown in mixed for Egypt;; using Household
inaccurate estimates of the effect Karki et al. (2003), the price elasticity Expenditure Survey data for two
of price on smoking behaviour, as it of smoking intensity was greatest years (1995–96 and 1999–2000),
may miss any fast-acting change in for the second-poorest income Nassar (2003) found no systematic
smoking behaviour in response to a quintile, and it decreased with higher relationship between price elasticity
price increase or stabilization within income quintiles. This pattern was and expenditure quartiles in either
a given year. also reflected in total price elasticity. urban or rural areas in Egypt in
Smoking prevalence did not show 1995–96;; whereas the price elasticity
Evidence from low- any systematic relationship with was greater for higher expenditure
and middle-income countries income status in Nepal. On the other quartiles in 1999–2000 in both urban
hand, Kinh et al. (2006) estimated and rural areas. This finding stands
The variation of price responsiveness larger elasticities of both smoking in contrast with the conventional
of smoking prevalence and intensity prevalence and smoking intensity for pattern. However, when price
across socioeconomic groups is the lowest two quintiles compared to elasticity estimates were ranked by
evident in low and middle income the top two quintiles for Viet Nam. educational status of individuals,
countries as well. The relevant The study on Bangladesh people with lower levels of education
literature proliferated during 2002– conducted by the International were consistently found to be more
2003 and thereafter with the Tobacco Control Policy Evaluation responsive to price changes than
launching of the HNP Discussion Project in 2009 found systematic those with higher levels of education.
Paper series by the Health, Nutrition evidence of greater price elasticity A study on Indonesia by
and Population Family (HNP) of the of smoking prevalence among Adioetomo et al. (2005) found
World Bank’s Human Development lower income households for that smoking intensity in lower-
Network on the Economics of both cigarette and bidi smoking, income groups was consistently
Tobacco Control, in collaboration and mixed evidence on the price more sensitive to price changes
with the Tobacco Free Initiative of the responsiveness of smoking intensity. than in higher-income groups. They
World Health Organization. These The daily consumption of smokers estimated the elasticity of smoking
studies are based on relatively recent belonging to lower socioeconomic SUHYDOHQFHDWíIRUORZLQFRPH
data sets and are uniform in method status was found to be insensitive to households. For medium- and high-
of estimation of price elasticities for price changes in both cigarette and income households, the elasticities
smoking prevalence and smoking bidi consumption. The moderate- were positive. None of these
intensity. and high-income households did estimates were, however, statistically
290
significant. This study, which factors prevalence in general. The existing preference, which pertains to
in the number of cigarettes smoked literature rarely differentiates the premium a person places on
by the smokers, therefore suggests between smoking initiation, quitting enjoyment in the immediate period
that in Indonesia price is not a or relapse, primarily due to the fact over delayed gratification (Becker
significant factor in people’s decision that this type of analysis requires and Murphy, 1988;; Becker et al.,
to smoke, particularly for the well-off. longitudinal or retrospective data 1991). An individual with high time
Among other lower-middle- on individual smoking behaviour. preference is focused more on
income countries, such as Sri Lanka, Madden (2007), for example, used his well-being in the present, and
Thailand and Ukraine, the findings retrospective data on a sample of therefore discounts the payoffs in
are not systematic enough to reach Irish women aged 48 and under to the future more heavily. Individuals
a conclusion about the variation conduct a duration analysis of the with high time preference not only
in price responsiveness among factors influencing smoking initiation discounts future payoffs more than
different socioeconomic groups and quitting. The study found mixed an average person would do, but
(Table 7.4). This is also the case evidence on the effect of price on they also tend to discount future loss
for upper-middle-income countries, smoking initiation and quitting by of health or earnings or any harmful
such as Bulgaria and Turkey. In education status of individuals. For effects of current consumption at
Bulgaria, for example, the average starting smoking, the strongest effect a higher rate. Smokers from lower
price elasticity of smoking propensity of tax was observed on those with socioeconomic status often fall into
is highest among the low-income intermediate levels of education, this category. Relative to people with
group, but the high-income group and weaker effects for those with higher socioeconomic status, they
showed greater price sensitivity than least education or higher levels of assign greater value to the current
the middle-income group (Sayginsoy education, evidence suggesting an gratification from smoking, so much
et al., 2002). In Turkey, price elasticity inverse-U effect. The finding that so that it more than compensates
was highest for the second income the price sensitivity for intermediate for the future loss of health and
quintile, in line with an inverse-U levels of education was greater than productivity they expect to incur for
relationship (Onder, 2002). for higher levels of education was, their current smoking habit. In the
In South Africa, on the other however, not statistically significant. same vein, a given increase in the
hand, price sensitivity was larger at On the other hand, tax seemed price of tobacco in the present period
lower income quartiles (van Walbeek, to be most effective in inducing imposes greater cost on the addicts
2002b, 2005). The percentage of quitting among those with the least belonging to lower socioeconomic
households that bought cigarettes education, and there was little status, and induces them to cut
decreased from 49% in 1990 to evidence of any significant difference tobacco consumption more than
30% in 2000, while the real price of in the effectiveness of taxation well-off addicts would do.
cigarettes increased by more than between groups with higher levels of The share of tobacco expenditure
100%. The percentage of households education. in total household expenditure
in the poorest income quartile that partly accounts for the greater price
bought cigarettes decreased from Explaining differences in price responsiveness of lower-income
46% to 22%, while among the richest sensitivity by socioeconomic people. It can be shown that a 1%
income quartile that percentage status increase in the price of cigarettes
decreased from 43% to 34%. This translates into equal percentage
implies that the poor are significantly The greater price sensitivity of people increase in total expenditure on
more price-sensitive than the rich belonging to lower socioeconomic tobacco, given income and the
(van Walbeek, 2005). status, as observed consistently in quantity demanded of tobacco. This
higher-income countries, has a clear is the case when the price elasticity
Price responsiveness of smoking theoretical basis for explanation. It of smoking intensity for continued
initiation, quitting and relapse follows from the theory of rational smokers is zero. Tobacco expenditure
addiction that the variation in the can increase in response to a price
The evidence available on the price price sensitivity of smokers across increase even when the quantity
sensitivity of smoking prevalence different socioeconomic groups demanded of tobacco decreases,
by socioeconomic characteristics in an economy is attributable to provided that the price elasticity is
mostly pertains to smoking the variation in their rate of time less than one, which is typically the
291
case for an addictive commodity such response to price increases. To the products to compensate for the rising
as tobacco. When tobacco demand extent that this happens, smokers costs of smoking, and expressed
is inelastic, tobacco demand falls maintain their habit and the impact the view that smugglers provide
less than proportionately in response of price increases is reduced. The a valuable service to society
to a price increase raising the total existing literature provides ample (Wiltshire et al., 2001). Evidence
expenditure on it. Thus as long as evidence in support of the substitution from Taiwan, China also indicated
demand for tobacco is price-inelastic, to cheaper or alternative tobacco that low-income and poorly educated
a given increase in its price results products in response to cigarette tax smokers are more likely to purchase
in greater proportion of household and price increase (e.g. Thompson smuggled cigarettes—smokers with
expenditure needed to maintain the and McLeod, 1976;; Pekurinen, personal monthly income lower than
habit of tobacco use, other things 1989;; Ohsfeldt and Boyle, 1994;; NT$10 000 in 2004 and the lowest
remaining the same. The lower the Ohsfeldt et al., 1997, 1999;; Evans level of education were 54% more
level of total household expenditure, and Farrelly, 1998;; Laxminarayan likely to smoke smuggled cigarettes
the greater is the burden of increased and Deolalikar, 2004;; Young et al., than the smokers who had either of
expenditure. Given the scarcity of 2006). These studies, however, do these two characteristics or none
household resources to meet the not report whether the likelihood (Lee et al., 2009).
basic necessities in low-income of tax avoidance through product These findings imply that in
households, they incur relatively substitution varies by socioeconomic the presence of widespread tax
higher opportunity cost of tobacco status of smokers;; the evidence base avoidance and evasion behaviour
use, as discussed above. The greater to address this issue is very limited. among tobacco users, the price
opportunity cost of an additional unit Using the ITC data set for four responsiveness would be lower
of money spent on tobacco is greater countries, USA, Canada, the United than intended, and it may vary
for lower-income households, which Kingdom and Australia, Hyland et across the socioeconomic strata
leads the smokers at the lower end al. (2006) found evidence of tax of the population. Since lower-
of household income distribution to avoidance, particularly among income people are more prone to tax
cut down tobacco consumption more younger, non-white, male and higher- avoidance and evasion than others,
in response to price increases than income smokers who smoke more price elasticity estimates for this
higher-income smokers would do. cigarettes per day than the older, group are likely biased upward. Thus
The effect of the increase in white, female and lower-income the greater price sensitivity of people
tobacco price on the share of smokers respectively. Smokers from lower socioeconomic status
tobacco expenditure and consequent who tend to purchase more from found in most studies in high-income
opportunity cost of tobacco use is low-taxed and untaxed sources of countries may prove to be illusive if
accentuated when smokers from cigarettes or switch to roll-your-own this compensatory behaviour pattern
lower-income households spend cigarettes are expected to be less of lower-income people is taken into
a greater proportion of household likely to quit smoking, and their price account for the estimation of price
expenditure on tobacco to begin sensitivity would be lower than the elasticity of tobacco demand.
with. The relatively greater burden rest of the population. The mixed evidence of the
of tobacco expenditure on poorer Lower-income smokers would relationship of price responsiveness
households is widely evident in the likely also seek tobacco products to socioeconomic status in low- and
low- and middle-income countries made available through illicit trade middle-income countries, however,
(see Table 7.2 for the evidence at prices cheaper than the market cannot be explained adequately with
on varying expenditure pattern by rates. To the extent that they succeed time preference, share of tobacco
socioeconomic status of households). in maintaining their consumption expenditure or product substitution
The studies cited above do not level and expenditure on tobacco behaviour, as explained above. The
take into account the possibility by using contraband tobacco poorest people are not necessarily
of substitution from higher-priced products, their responsiveness to the most sensitive to tobacco price
to lower-priced smoked tobacco tobacco tax increases is reduced. changes in these countries. Although
products, nor from smoked to In the United Kingdom, for example, circumstances can differ between
smokeless tobacco products, or any smokers from socioeconomically countries, in some countries this
other form of compensatory change disadvantaged areas of Edinburgh can be attributed to the availability
in tobacco consumption pattern in reported purchases of contraband of low-cost or cheaper tobacco
292
products that are predominantly responsiveness by socioeconomic with disproportionately larger share
consumed by lower-income people status of tobacco users. As of tobacco expenditure and disease
and are often untaxed or subject to discussed in Chapter 2, most low- burden borne by the poor. In these
preferential tax treatment. Examples income countries have ad valorem circumstances, increasing tobacco
of such products are as kreteks in taxes, while high-income countries taxes can yield greater public health
Indonesia;; bidi and chewing tobacco such as the USA, Canada, Australia, gain to them than to the rich.
in Bangladesh, India and other New Zealand, Japan and Singapore The price responsiveness of
southeast Asian regions;; waterpipe rely on a specific tax system, and tobacco demand also varies by
tobacco in southeast Asia and the European countries mostly rely the socioeconomic status (such as
the eastern Mediterranean, “Eco on a mixed structure with both ad income, expenditure, education, race
cigarillos” and fine-cut tobacco in valorem and specific taxes. It is or ethnicity) of the population groups
the EU;; and “small cigars” in the well recognized that the burden of of a country. It is consistently higher
USA. (Chapter 2 of this volume increased tax is greater on low- among the poor than the rich in high-
describes these tobacco products priced products under the specific income countries. This finding can
in detail). van Walbeek (2005) found tax system than under the ad valorem be explained by the present-oriented
strong evidence that the poor were system. Thus it is expected that nature of poorer people, who tend to
switching to pipe and other tobacco consumers of low-priced tobacco discount at a higher rate the future
(presumably to make roll-your-own products in countries using specific health cost and loss of earning owing
cigarettes) much more than the rich. tax system would be more sensitive to tobacco use. Poor people also
In 1990 the poorest quarter of the to tax increases than in countries incur increasingly higher opportunity
population spent about 5% of their using an ad valorem system. cost of tobacco use when tobacco
tobacco purchases on pipe and other However, no study has been price increases, and thus tend to
tobacco;; in 2000 this had increased undertaken to examine the effect of reduce tobacco consumption more
to 18.7%. Among the second poorest either the differential tax treatment than the rich would do. As a result,
quarter of the population there was of tobacco products or of different the share of the total quantum of
also an increase in the relative share tobacco tax structure across tobacco tax paid by the poor is
of pipe and other tobacco (from 2.4% countries that can potentially explain expected to decrease when tobacco
to 7.1% of total tobacco expenditure), the variation in the price sensitivity tax and prices are increased.
but the share of pipe and other of tobacco users across countries Evidence on the variation of
tobacco among the richer half of or across socioeconomic groups price responsiveness of tobacco
the population remained unchanged within countries to a great extent. demand by socioeconomic status of
over this period. The dichotomy in the evidence base population groups is mixed in low-
Taxes on this type of product are between high- and lower-income and middle-income countries. The
kept low to make them affordable countries clearly calls for more poorest people are not necessarily
to lower-income consumers. The rigorous evaluation of the data and the most sensitive to tobacco price
differential tax rates applied to a methods used for these analyses changes in these countries. Although
wide range of prices across types of that are likely to affect the results of circumstances can differ between
tobacco products or even within the estimation, and of country-specific countries, in some countries this
same tobacco product encourage socioeconomic environments and can be attributed to the availability
substitution of lower-taxed products tobacco tax structures that can of untaxed and cheaper tobacco
when taxes are increased, as potentially influence the mechanisms products, which are consumed
opposed to quitting or reduction through which tax increase affects more by the poor. This explains
in consumption. It in turn lowers tobacco demand. why the existing empirical literature
the effectiveness of increased tax, is not conclusive about the equity
particularly on the lower-income Conclusion implication of the increase in tobacco
people who primarily consume lower- taxes in low- and middle-income
taxed tobacco products. The prevalence of tobacco use countries.
The difference in the tax structure is in general higher among the
between low- and high-income poor. The greater prevalence of
countries is also likely to contribute tobacco use among the poor adds
to the observed pattern of price to economic and social disparity,
293
References
Adioetomo SM, Djutaharta T, Hendratno Browning EK (1978). The burden of taxation. J DeCicca P, McLeod L (2008). Cigarette taxes
(2005). Cigarette consumption, taxation and Polit Econ, 86:649–671 doi:10.1086/260703. and older adult smoking: evidence from recent
household income: Indonesia case study. large tax increases. J Health Econ, 27:918–
HNP Discussion Paper Series, Economics of Centers for Disease Control and Prevention 9 2 9 .d o i:10 .1016 / j . j h e a l e c o. 2 0 07.11. 0 0 5
Tobacco Control Paper No. 26. Washington DC, (CDC) (1998). Response to increases in PMID:18178277
The World Bank. cigarette prices by race/ethnicity, income,
and age groups–United States, 1976–1993. Efroymson D, Ahmed S, Townsend J et al.
Ali Z, Rahman A, Rahman T (2003). Appetite MMWR Morb Mortal Wkly Rep, 47:605–609. (2001). Hungry for tobacco: an analysis of the
for nicotine: an economic analysis of tobacco PMID:9699809 economic impact of tobacco consumption on
control in Bangladesh. HNP Discussion Paper the poor in Bangladesh. Tob Control, 10:212–
Series, Economics of Tobacco Control Paper Chaloupka F, Hu TW, Warner KE et al. 217.doi:10.1136/tc.10.3.212 PMID:11544383
No.16. Washington DC, The World Bank. (2000). Taxation of tobacco products. In: Jha
P, Chaloupka F, editors. Tobacco control in Evans W, Ringel J, Stech D (1999). Tobacco
Aloui O (2003). Analysis of the economics of developing countries. Oxford, U.K.: Oxford taxes and public policy to discourage smoking.
tobacco in Morocco. HNP Discussion Paper University Press;; p. 237–72. In: Poterba JM, editor. Tax Policy and the
Series, Economics of Tobacco Control Paper Economy, Vol.13. Cambridge, MA: MIT Press;;
No. 7. Washington DC, The World Bank. Chaloupka FJ (1991). Rational addictive p. 1–56.
behavior and cigarette smoking. J Polit Econ,
Arunatilake N (2002). An economic analysis 99:722–742 doi:10.1086/261776. Evans WN, Farrelly MC (1998). The
of tobacco demand in Sri Lanka. Sri Lanka compensating behavior of smokers: taxes,
Economic Journal, 3:96–120. Colman G, Remler DK (2004). Vertical equity tar, and nicotine. Rand J Econ, 29:578–595.
consequences of very high cigarette tax doi:10.2307/2556105 PMID:11794360
Arunatilake N, Opatha M (2003). The economics increases: if the poor are the ones smoking, how
of tobacco in Sri Lanka. HNP Discussion Paper could cigarette tax increases be progressive. Farrelly MC, Bray JW, Pechacek TF, Woollery
Series, Economics of Tobacco Control Paper NBER Working Paper Series. Working Paper T (2001). Responses by adults to increases
No.12. Washington DC, The World Bank. #10906. Cambridge, MA, National Bureau of in cigarette prices by sociodemographic
Economic Research. characteristics. Southern Economic Review,
Auld M (2005). Smoking, drinking and income. J 68:156–165 doi:10.2307/1061518.
Hum Resour, 40:505–518. Colman G, Remler DK (2008). Vertical equity
consequences of very high cigarette tax Farrelly MC, Engelen M (2008). Cigarette
Barbeau EM, Krieger N, Soobader MJ (2004). increases: if the poor are the ones smoking, how prices, smoking, and the poor, revisited. Am
Working class matters: socioeconomic could cigarette tax increases be progressive? J J Public Health, 98:582–583, author reply
disadvantage, race/ethnicity, gender, and Policy Anal Manage, 27:376–400 doi:10.1002/ 583 – 58 4.doi:10. 2105/A JPH. 20 07.1326 47
smoking in NHIS 2000. Am J Public Health, pam.20329. PMID:18309115
9 4:2 6 9 – 278 .doi:10. 210 5 /A J PH .9 4. 2. 26 9
PMID:14759942 Congress of the United States, Congressional Federico B, Kunst AE, Vannoni F et al. (2004).
Budget Office (1990). Federal taxation of Trends in educational inequalities in smoking
Barber S, Adioetomo SM, Ahsan S et al. tobacco, alcoholic beverages and motor fuels. in northern, mid and southern Italy, 1980–
(2008). Tobacco Economics in Indonesia. Paris, Washington DC: US Government Printing 2000. Prev Med, 39:919–926.doi:10.1016/j.
International Union Against Tuberculosis and Office, Congress of the United States ypmed.2004.03.029 PMID:15475024
Lung Disease.
Congress of the United States, Congressional Franks P, Jerant AF, Leigh JP (2008). Franks et
Becker GS, Grossman M, Murphy KM (1991). Budget Office (2001). Historic effective al. respond. Am J Public Health, 98:583–584
Rational addiction and the effect of price on tax rates, 1979–1997. Washington DC: US doi:10.2105/AJPH.2007.132654.
consumption. Am Econ Rev, 84:396–418. Government Printing Office, Congress of the
United States. Franks P, Jerant AF, Leigh JP et al.
Becker GS, Murphy KM (1988). A theory of (2007). Cigarette prices, smoking, and the
rational addiction. J Polit Econ, 96:675–700 Cutler DM, Gruber J, Hartman RS et al. poor: implications of recent trends. Am J
doi:10.1086/261558. (2002). The economic impacts of the tobacco Public Health, 97:1873–1877.doi:10.2105/
settlement. J Policy Anal Manage, 21:1–19. AJPH.2006.090134 PMID:17761576
Biener L, Aseltine RH Jr, Cohen B, Anderka doi:10.1002/pam.1037 PMID:11887906
M (1998). Reactions of adult and teenaged Fukuda Y, Nakamura K, Takano T (2005).
smokers to the Massachusetts tobacco tax. David A, Esson K, Perucic AM, Fitzpatrick Socioeconomic pattern of smoking in Japan:
Am J Public Health, 88:1389–1391.doi:10.2105/ C (2010).Tobacco use: equity and social income inequality and gender and age
AJPH.88.9.1389 PMID:9736885 determinants. In: Blas E, Sivasankara Kurup A, differences. Ann Epidemiol, 15:365–372.
editors. Equity, social determinants and public d o i :1 0 .1 0 1 6 / j . a n n e p i d e m . 2 0 0 4 . 0 9 . 0 0 3
Bobak M, Jha P, Nguyen SN, Jarvis MJ (2000). health programmes. Geneva: World Health PMID:15840550
Poverty and smoking. In: Jha P, Chaloupka F, Organization.
editors. Tobacco control in developing countries. Fullerton D, Rogers DI (1993). Who bears
Oxford, U.K.: Oxford University Press;; p. 41–61. de Miera Juarez BS, Jimenez-Ruiz JA, Reynales- the lifetime tax burden? Washington DC,
Shigematsu LM et al. (2007). El consumo de Borookings Institute.
Borren P, Sutton M (1992). Are increases in tabaco en los hogares mexicanos, 1994–2005.
cigarette taxation regressive? Health Econ, Salud Publica Mex, 49 suppl.2;;263–269.
1:24 5 – 25 3.doi:10.10 02 / hec.473 0 010 4 0 6
PMID:1341940
294
Gospodinov N, Irvine I (2009). Tobacco taxes Isra S (2003). The economics of tobacco in Liu Y, Rao K, Hu TW et al. (2006).
and regressivity. J Health Econ, 28:375– Thailand. HNP Discussion Paper, Economics Cigarette smoking and poverty in China.
3 8 4 .d o i:10 .1016 / j . j h e a l e c o. 2 0 0 8 .10 . 010 of Tobacco Control Paper No.15. Washington Soc Sci Med, 63:2784–2790.doi:10.1016/j.
PMID:19091432 DC, World Bank. socscimed.2006.06.019 PMID:16959391
Graham H (1987). Women’s smoking and John RM (2005). Tobacco consumption Lokshin M, Beegle K (2006). Foregone earnings
family health. Soc Sci Med, 25:47–56. patterns and its health implications in India. from smoking: evidence from a developing
d o i : 1 0 .1 0 1 6 / 0 2 7 7 - 9 5 3 6 ( 8 7 ) 9 0 2 0 6 - 1 Health Policy, 71:213–222.doi:10.1016/j. country. World Bank Policy Research Working
PMID:3616697 healthpol.2004.08.008 PMID:15607383 Paper No.4018. Washington DC, The World
Bank.
Graham H (1994). Gender and class as John RM (2008). Crowding out effect of
dimensions of smoking behaviour in Britain: tobacco expenditure and its implications Lopez AD, Collishaw NE, Piha TA (1994). A
insights from a survey of mothers. Soc on household resource allocation in India. descriptive model of the cigarette epidemic in
Sci Med, 38:691–698.doi:10.1016/0277- Soc Sci Med, 66:1356–1367.doi:10.1016/j. developed countries. Tob Control, 3:242–247
9536(94)90459-6 PMID:8171347 socscimed.2007.11.020 PMID:18187245 doi:10.1136/tc.3.3.242.
Graham H (1996). Smoking prevalence among Karki YB, Pant KD, Pande BR (2003). A Lyon AB, Schwab RM (1995). Consumption
women in the European c ommunity 1950–1990. study on the economics of tobacco in Nepal. taxes in a life-cycle framework: are sin taxes
Soc Sci Med, 43:243–254.doi:10.1016/0277- HNP Discussion Paper Series, Economics of regressive? Rev Econ Stat, 77:389–406
9536(95)00369-X PMID:8844928 Tobacco Control Paper No.13. Washington DC, doi:10.2307/2109902.
The World Bank.
Gruber J, Koszegi B (2002). A theory of Madden D (2007). Tobacco taxes and starting
government regulation of addictive bads: Khang YH, Yun SC, Cho HJ, Jung-Choi and quitting smoking: does the effect differ
optimal tax levels and tax incidence for K (2009). The impact of governmental by education? Appl Econ, 39:613–627
cigarette excise taxation. NBER Working Paper antismoking policy on socioeconomic doi:10.1080/00036840500447898.
Series. Working Paper #8777. Cambridge, MA, disparities in cigarette smoking in South Korea.
National Bureau of Economic Research. Nicotine Tob Res, 11:262–269.doi:10.1093/ntr/ Mao Z, Sung HY, Hu T et al. (2007). Demand
ntn036 PMID:19251768 for cigarette in China. In: Hu TW, ed., Tobacco
Gruber J, Koszegi B (2004). Tax incidence when control policy analysis in China, Economics
individuals are time-inconsistent: the case of Kinh HV, Ross H, Levy D et al. (2006). The and Health. World Scientific Publishing Co
cigarette excise taxes. J Public Econ, 8 8:1959– effect of imposing a higher, uniform tobacco Singapore, 129–157.
1987 doi:10.1016/j.jpubeco.2003.06.001. tax in Vietnam. Health Res Policy Syst, 26:4–6
doi:10.1186/1478-4505-4-1. PMID:16800880 Mao Z, Yang GH, Ma JM (2003). Adult’s
Gruber J, Sen A, Stabile M (2003). Estimating demand for cigarettes and its determinants in
price elasticities when there is smuggling: the Krasovsky K, Andreeva T, Krisanov D et China. Soft Science of Health, 17:19–23.
sensitivity of smoking to price in Canada. J al. (2002). Economics of tobacco control in
Health Econ, 22:821–842.doi:10.1016/S0167- Ukraine from the public health prespective. Nargis N, Yurekli AA, Bettcher DM (2010).
6296(03)00058-4 PMID:12946461 Kiev, Alcohol and Drug Information Centre. Poverty and development impact. In:
Chaloupka FJ, Yurekli AA, Fong GT, eds.,
Heineck G, Schwarze J (2003). Substance use Kyaing NN (2003). Tobacco economics in Tobacco control in developing countries. NCI
and earnings: the case of smokers in Germany. Myanmar. HNP Discussion Paper Series, Monograph 21., 18
IZA Discussion Papers, No.743. Economics of Tobacco Control Paper No.14.
Washington DC, The World Bank. Nassar H (2003). The economics of tobacco
Hersch J (2000). Gender, income levels, and in Egypt: a new analysis of demand. HNP
the demand for cigarettes. J Risk Uncertain, Kyaing NN, Perucic A-M, Rahman K (2005). Discussion Paper Series, Economics of
21:263–282 doi:10.1023/A:1007815524843. Study on poverty alleviation and tobacco Tobacco Control Paper No.8. Washington DC,
control in Myanmar. HNP Discussion Paper The World Bank.
Hu TW, Mao Z, Liu Y et al. (2005). Smoking, Series. Economics of Tobacco Control Paper
standard of living, and poverty in China. No.31. Washington DC, The World Bank. Ohsfeldt RL, Boyle RG (1994). Tobacco excise
Tob Control, 14:247–250.doi:10.1136/ taxes and rates of smokeless tobacco use in
tc.2004.010777 PMID:16046687 Laxminarayan R, Deolalikar A (2004). Tobacco the US: an exploratory ecological analysis. Tob
initiation, cessation, and change: evidence Control, 3:316–323 doi:10.1136/tc.3.4.316.
Hu T W, Tsai Y W (2000). Cigarette c onsumption from Vietnam. Health Econ, 13:1191–1201.
in rural China: survey results from 3 provinces. doi:10.1002/hec.932 PMID:15386650 Ohsfeldt RL, Boyle RG, Capilouto E (1997).
Am J Public Health, 9 0:1785–1787.doi:10.2105/ Effects of tobacco excise taxes on the
AJPH.90.11.1785 PMID:11076252 Lee JM (2008). Effect of a large increase use of smokeless tobacco products in the
in cigarette tax on cigarette consumption: USA. Health Econ, 6:525–531.doi:10.1002/
Huisman M, Kunst AE, Mackenbach JP (2005). an empirical analysis of cross-sectional 6 , & , $ , '
Inequalities in the prevalence of smoking in the survey data. Public Health, 122:1061–1067. HEC300>3.0.CO;;2-Y PMID:9353656
European Union: comparing education and doi:10.1016/j.puhe.2007.12.013 PMID:18602655
income. Prev Med, 40:756–764.doi:10.1016/j. Ohsfeldt RL, Boyle RG, Capilouto E (1999).
ypmed.2004.09.022 PMID:15850876 Lee JM, Chen SH, Chen HF, Jeng HY (2009). Tobacco taxes, smoking, restrictions and tobacco
Price sensitivity and smoking smuggled use. In: Chaloupka FJ, Grossman M, Bickel WK, et
Hyland A, Laux FL, Higbee C et al. (2006). cigarettes. Eur J Public Health, 19:23–27. al., editors. The economics analysis of substance
Cigarette purchase patterns in four countries doi:10.1093/eurpub/ckn115 PMID:19039020 use and abuse: an integration of econometrics
and the relationship with cessation: findings and behavioral economic research. Chicago:
from the International Tobacco Control (ITC) Levine P, Gustafson A, Valenchik A (1997). University of Chicago Press;; p. 15–29.
Four Country Survey. Tob Control, 15 Suppl More bad news for smokers?The effect of
3;;iii 5 9 – iii 6 4.doi:10.113 6 / tc. 20 0 5.012 20 3 cigarette smoking on wages. Ind Labor Relat Onder Z (2002). The economics of tobacco in
PMID:16754948 Rev, 50:493–509 doi:10.2307/2525187. Turkey: new evidence and demand estimates.
HNP Discussion Paper Series, Economics of
Tobacco Control Paper No. 2. Washington DC,
The World Bank.
295
Pekurinen M (1989). The demand for tobacco Siahpush M, Wakefield MA, Spittal MJ et al. Viscusi WK (2002). The new cigarette
products in Finland. Br J Addict, 84:1183– (2009). Taxation reduces social disparities in paternalism. Regulation, 25:58–64.
1192.doi:10.1111/j.1360-0443.1989.tb00714.x adult smoking prevalence. Am J Prev Med,
PMID:2819275 36:285–291.doi:10.1016/j.amepre.2008.11.013 Wang H, Sindelar JL, Busch SH (2006). The
PMID:19201146 impact of tobacco expenditure on household
Pierce JP, Fiore MC, Novotny TE et al. (1989). consumption patterns in rural China. Soc Sci
Trends in cigarette smoking in the United Stehr M (2007). The effect of cigarette taxes Med, 62:1414–1426.doi:10.1016/j.socscimed.
States. Educational differences are increasing. on smoking among men and women. Health 2005.07.032 PMID:16137812
JAMA, 261:56–60.doi:10.1001/jama.261.1.56 Econ, 16:1333–1343.doi:10.1002/hec.1223
PMID:2908995 PMID:17335102 Warner KE (2000). The economics of tobacco:
myths and realities. Tob Control, 9:78–89.
Rajemison HS, Haggblade S, Younger SD Thompson ME, McLeod I (1976). The effects doi:10.1136/tc.9.1.78 PMID:10691761
(2003). Indirect tax incidence in Madagascar: of economic variables upon the demand for
updated estimates using the input-output table. cigarettes in Canada. Mathl Scientist, 1:121– White VM, Hayman J, Hill DJ (2008). Can
Working paper #147.Cornell University, Food 132. population-based tobacco-control policies
and Nutrition Policy Program. change smoking behaviors of adolescents
Thomson GW, Wilson NA, O’Dea D et al. from all socio-economic groups? Findings from
Rani M, Bonu S, Jha P et al. (2003). Tobacco (2002). Tobacco spending and children in low Australia: 1987–2005. Cancer Causes Control,
use in India: prevalence and predictors of income households. Tob Control, 11:372–375. 19:631– 640.doi:10.1007/s10552-008-9127-8
smoking and chewing in a national cross doi:10.1136/tc.11.4.372 PMID:12432164 PMID:18264783
sectional household survey. Tob Control, 12:4.
doi:10.1136/tc.12.4.e4 PMID:14660785 Townsend J (1987). Cigarette tax, Wiltshire S, Bancroft A, Amos A, Parry O
economic welfare and social class patterns (2001). “They’re doing people a service”-
Remler DK (2004). Poor smokers, poor of smoking. Appl Econ, 19:355–365 qualitative study of smoking, smuggling,
quitters, and cigarette tax regressivity. Am doi:10.1080/00036848700000007. and social deprivation. BMJ, 323:203–207.
J Public Health, 94:225–229.doi:10.2105/ doi:10.1136/bmj.323.7306.203 PMID:11473911
AJPH.94.2.225 PMID:14759931 Townsend J (1996). Price and consumption
of tobacco. Br Med Bull, 52:132–142. World Health Organization (2007). World
Ringel JS, Evans WN (2001). Cigarette PMID:8746302 Health Statistics 2007. Geneva, World Health
taxes and smoking during pregnancy. Am J Organization.
Public Health, 91:1851–1856.doi:10.2105/ Townsend J, Roderick P, Cooper J (1994).
AJPH.91.11.1851 PMID:11684615 Cigarette smoking by socioeconomic group, World Health Organization (2010).Equity, social
sex, and age: effects of price, income, determinants and public health programmes.
Sayginsoy O, Yurekli A, De Beyer J (2002). and health publicity. BMJ, 309:923–927. Geneva, World Health Organization.
Cigarette demand, taxation, and the poor: a PMID:7950662
case study of Bulgaria. HNP Discussion Paper Young D, Borland R, Hammond D et al.;; ITC
Series, Economics of Tobacco Control Paper Van Ours J (2003). A pint day raises a man’s Collaboration (2006). Prevalence and attributes
No. 4. Washington DC, The World Bank. pay, but smoking blows that gain away. CEPR of roll-your-own smokers in the International
Discussion papers, 3308. Tobacco Control (ITC) Four Country Survey.
Schaap MM, Kunst AE, Leinsalu M et al. Tob Control, 15 Suppl 3;;iii76–iii82.doi:10.1136/
(2008). Effect of nationwide tobacco control van Walbeek C (2002a). Recent trends in tc.2005.013268 PMID:16754951
policies on smoking cessation in high and low smoking prevalence in South Africa–some
educated groups in 18 European c ountries. Tob evidence from AMPS data. S Afr Med J,
Control, 17:248–255.doi:10.1136/tc.2007.024265 92:468–472. PMID:12146134
PMID:18483129
van Walbeek CP (2002b). The distributional
Siahpush M (2003). Socioeconomic status impact of tobacco excise increases. S Afr J
and tobacco expenditure among Australian Econ, 70:560–578.
households: results from the 1998–99
Household Expenditure Survey. J Epidemiol van Walbeek CP (2005). Do tax increases on
Community Health, 57:798–801.doi:10.1136/ tobacco hurt the poor?
jech.57.10.798 PMID:14573585
296
Chapter 8
Tax avoidance and tax evasion
Introduction
Tax avoidance and tax evasion can next section explains the motivation to products illegally traded across
decrease the economic welfare for tax avoidance and tax evasion, borders. Illicit trade is defined in
by making tobacco products more and categorizes these motives based Article 1 of the WHO Framework
affordable and available, thus whether they are related to profit Convention of Tobacco Control
exacerbating the negative health generation, costs of supplying illicit (World Health Organization, 2005) as
consequences associated with products, deterrence or an overall any practice or conduct prohibited by
tobacco use and secondhand state of the economy. The third section law and which relates to production,
smoking. provides the most recent estimates shipment, receipt, possession,
Furthermore, tax avoidance of the extent of tax evasion globally, distribution, sale or purchase,
and tax evasion can undermine the regionally and also in some selected including any practice or conduct
impact of tobacco control measures, countries. The following section intended to facilitate such activity.
primarily tobacco tax policies. The reviews the literature on the impact Illicit tobacco trade covers more
existence of illicit tobacco trade of tax avoidance and tax evasion activities than the circumvention of
has been used to increase political on public health measures such as taxes, but includes all illegal activities
pressure on governments and smoking rate, smoking intensity and related to the tobacco trade.
discourage them from adopting and heath disparities. The final section Economists mostly refer to the
implementing effective tobacco tax reviews the impact of policies circumvention of taxes, and prefer to
strategies. Moreover, illicit tobacco attempting to curb illicit tobacco trade use the terms tax avoidance (legal
trade can channel sales proceeds to and summarizes the lessons learned methods of circumventing tobacco
organized crime and lead to a loss in from the implementation of those taxes) and tax evasion (illegal
government tax revenues. policies. methods for circumventing tobacco
This chapter reviews and taxes).
summarizes the research evidence Defining and measuring tax This section defines the activities
related to tobacco tax avoidance and evasion and avoidance of tax avoidance and tax evasion that
tobacco tax evasion from published fall into each category and briefly
literature and empirical evidence. This Among those working on tobacco describes approaches to measuring
body of information is organized in five tax issues, a variety of circumventing the extent of both, drawing heavily
sections. The first section explains the activities for not paying all tobacco from the classification scheme
difference between tax avoidance and taxes are often grouped together proposed by Joossens and colleagues
tax evasion, defines the activities that and referred to as “smuggling” or (2000) and the methods described
fall into each category, and describes “illicit trade” in tobacco products. in IARC’s Handbooks of Cancer
methods used in measuring the A clarification of the terms used is Prevention Volume 12, Methods for
extend to which these activities supply necessary as those terms cover Evaluating Tobacco Control Policies
tobacco products to the market. The different activities. Smuggling refers (IARC, 2008), the World Bank’s tool
297
on tobacco smuggling (Merriman, US states). However, enforcement of through direct channels (Chaloupka
2001), and the book chapter by these provisions is difficult. et al., in press). Differences in legal
Merriman and colleagues (2000). Tourist shopping. This type of obligations for paying taxes in the
tax avoidance is similar to the cross- home jurisdiction can make this
Tax avoidance border shopping described above, type of activity more a form of tax
but involves the purchase of tobacco evasion than tax avoidance (e.g. EU
Tax avoidance includes legal activities products in more distant jurisdictions, countries).
and purchases in accordance with again subject to the constraints Industry reformulation or
customs and tax regulations, most imposed by customs laws and/or repositioning. Tobacco companies
of which include the payment of other policies (e.g. consumption can reduce the tax imposed on
some tobacco taxes, and are done of tax-paid cigarettes by the large their products by reformulating or
mostly by individual tobacco users, tourist population in Thailand). This repositioning their products. For
including cross-border shopping, is a more limited phenomenon, but example, in countries with multitiered
tourist shopping, duty-free shopping, can account for a significant share tax structures where higher taxes
Internet and other direct purchases, of overall tobacco product sales in are levied on higher-priced brands,
industry reformulation and/or popular tourist destinations. a company can lower the price of
repositioning. These include: Duty-free shopping. This type of its product so that it moves from a
Cross-border shopping. This tax avoidance is similar to the others, higher tax tier to a lower tax tier. As
type of tax avoidance involves but in this case involves the purchase occurred recently in Germany, where
individual tobacco users residing in of tax-free tobacco products cigarette taxes are based on quantity,
higher tax jurisdictions purchasing purchased in airports, on airlines, companies produced long cigarettes
tobacco products in nearby lower- and in other travel-related venues. that were readily cut into smaller,
tax jurisdictions for their own Again, most governments impose standard size cigarettes, effectively
consumption within the customs limits on how much an individual can reducing the tax per cigarette.
constraints. This can involve crossing purchase and bring home from duty-
national borders, particularly where free sources. Tax Evasion
such border crossing is freely or Internet and other direct
relatively easily done (as between purchases. This type of tax avoidance The activities included under tax
the European Union Member States) involves individual tobacco users evasion are the illegal methods
or can take place within a given buying tobacco products online, of circumventing tobacco taxes,
country where there are significant through the mail, or over the phone such as the purchase of smuggled
differences in subnational taxes (as from establishments based in low- and illicit manufactured tobacco
in Canada where provincial taxes tax jurisdictions for consumption in products. Those activities include
differ, or in the United States where their own higher tax jurisdiction. This both small and large quantities and
state and local taxes can vary has attracted the most attention in often, but not always, involve efforts
considerably across jurisdictions). the USA, given relatively widespread to avoid paying any taxes. Many of
Within some countries, this also access to the Internet and significant these activities are done by larger
involves purchases from shops differences in subnational taxes. criminal networks or other large-
located in tax-exempt areas, such For US consumers, for example, scale operations.
as Aboriginal reserves in Canada these may include purchasing from Small-scale smuggling. This type
and Native American reservations vendors based on Native American of tax evasion involves the purchase,
in the USA. In some cases, there reservations, in low-tax states or by individuals or small groups,
are limits on how much can be in low-tax countries;; as above, of tobacco products in low tax
purchased outside of the jurisdiction however, purchasers are obligated jurisdictions in amounts that exceed
in which the individual resides (e.g. to pay taxes to their home state on the limits set by customs regulations,
the European Union), while in others these types of purchases. Over the for smuggling or resale in high-tax
the individual is supposed to pay the past several years, US states have jurisdictions.
difference between the tax in their taken steps to curb direct purchases This type of tax evasion is
home jurisdiction and the tax they through a variety of policy and illegal in that the quantities involved
have paid on the products purchased enforcement actions and Internet exceed the allowable limits and that
in other jurisdictions (e.g. in various and few smokers purchase cigarettes the purchase does not include the
298
taxes that are supposed to apply in Illicit manufacturing. This type of Over the past few decades, multiple
the jurisdiction where they are used. tax evasion refers to the production approaches have been developed
As with cross-border shopping, of tobacco products contrary to law. and applied, each of which captures
this is more likely to occur when The laws in question may be taxation some part of the full picture. The
tax differentials among nearby laws or other laws (such as licensing various approaches used to estimate
jurisdictions are large and where or monopoly-related laws) that the scope of tax avoidance and
border crossing is relatively easy. For restrict the manufacture of tobacco. evasion are briefly described below;;
example, small-scale smuggling is This type of tax evasion includes estimates based on these methods for
one form of tax evasion that occurs underreporting of actual production various countries are presented later
between US states, most notably quantities with the difference between in this chapter. Given the different
low-tax states with no tax stamps reported and actual production dimensions of tax avoidance and
(such as South Carolina) where diverted through illegal channels or, evasion captured using the different
small-scale smugglers buy tax-paid in some cases, complete avoiding of methods and the inherent limitations
cigarettes and then resell in high-tax reporting with all production diverted in each method, a combination of
states. to black markets. This type of illegal multiple methods will be most likely
Large-scale smuggling. This type manufacturing is more likely to occur to produce a good measure of the
of tax evasion involves the illegal in countries without effective tax extent of overall tax avoidance and
transportation, distribution, and administration that includes monitoring evasion (IARC, 2008).
sale of large quantities of tobacco of actual production, and in regions Expert opinion. One widely used
products, conducted by criminal where distribution of the illegally approach to assess the extent of
networks, that generally avoid all produced cigarettes is relatively easy. overall tax avoidance and evasion is
taxes. As Joossens and colleagues Counterfeit tax stamps are often to ask “experts” for their estimates
(2000) describe, this typically applied to illegally manufactured of these activities, where the
involves: international brands that products when these products are experts may be customs and other
are sold by multinational tobacco distributed in countries that require law enforcement officials, industry
companies and which are easily sold;; tax stamps. The destination of the representatives, researchers,
transportation over longer distances illegally manufactured cigarettes tobacco control advocates or others
and often involving “in-transit” can be the domestic or a foreign with a particular interest in the issue
regimes and tax-free zones;; the market. Illicit manufacturing includes (Merriman, 2001). Such estimates
passing of tobacco products through counterfeiting. are subjective and can be biased
a wide range of owners;; large Counterfeiting. Counterfeiting based on the individual expert’s
organized-crime networks;; and a involves the production and position and interests. Tobacco
sophisticated system for distributing distribution of products bearing a industry informants, for example,
smuggled cigarettes locally. To avoid trademark without the approval of the may have an incentive to report high
detection, counterfeit tax stamps are trademark owner. These products levels of tax avoidance and evasion
often applied to smuggled cigarettes are produced illegally, often bear as a way of averting tax increases,
that are being sold in jurisdictions counterfeit tax stamps (depending while tobacco control advocates
that require such stamps. Large- on where they are being sold), and may understate the extent of the
scale smuggling has also been used are distributed through the networks problem in their efforts supporting
for large consignments of counterfeit established by large-scale smuggling tobacco tax increases. Estimates
cigarettes or for legally manufactured operations. China has been a based on expert opinion are most
cigarettes which are targeting the major manufacturer of counterfeit prominent in trade and government
illicit markets in other countries. One cigarettes (Shen et al., 2010) publications. Overall, however,
example is the cigarette brand Jin measures based on expert opinion
Ling, which is legally manufactured in Measuring tax avoidance are generally consistent with those
the Russian Federation, but destined and evasion derived from other approaches,
for the illegal market in Germany and suggesting that such measures are
other European countries and which Given the illicit nature of tax valid (IARC, 2008).
was one of the most seized cigarette avoidance and evasion, developing Comparison of export and
brands in Europe in 2008 (World accurate measures of the extent import statistics. One approach to
Customs Organization, 2009). of these activities is challenging. estimating the extent of large scale
299
smuggling is to compare export of a measure based on this approach Hyland and colleagues (2006), for
and import statistics. The difference (IARC, 2008). example, reported data from the
between recorded exports and Modelling of tobacco product International Tobacco Control Policy
recorded imports is likely to reflect demand. A relatively widely used Evaluation Study’s (ITC) surveys of
product diverted to illegal markets approach among tobacco control representative samples of smokers
while in transit. This approach is researchers is the econometric in Australia, Canada, the United
likely to produce a valid measure modeling of tobacco product Kingdom and the United States which
for the global level of large-scale demand using data from multiple include questions on cross-border,
smuggling of legally manufactured neighbouring jurisdictions (e.g. US duty-free, native reserve, Internet
cigarettes, but is unlikely to produce states, European Union Member and other direct purchases, and
the same for small-scale smuggling States). Researchers using this other options that potentially reflect
and illicit manufactured cigarettes approach have included variables untaxed or lower taxed purchases.
or at the country level (IARC, 2008). measuring the opportunities for They found relatively low rates of
Merriman and colleagues (2000), tax avoidance and evasion based individual tax avoidance in Australia
for example, used this approach in on differences in prices across (0.7–1.1%), Canada (3.1–3.7%), and
their efforts to assess the extent jurisdictions, population distributions the United States (4.8–6.1%), but
of illicit trade in cigarettes, finding near borders, extent of cross-border high rates in the United Kingdom
that about one third of recorded or tourist traffic, Internet penetration, (15.3–19.7%), with rates increasing
exported cigarettes in the mid-1990s and other factors reflecting access in each country over the two waves
did not appear in recorded imports, to lower tax/price jurisdictions. of their surveys.
accounting for about 6% of global Coefficient estimates from the Observational data collection.
cigarette consumption. resulting models can be used to A relatively untested but potentially
Comparison of tax-paid sales and predict what tax-paid sales would promising approach to assessing
individually reported consumption have been if the variables reflecting multiple dimensions of tax avoidance
measures. If there are no reporting the incentives/opportunities were set and evasion is direct observation
biases in measures of tax-paid to zero, with the difference between of tobacco product vendors or
sales and measures of average predicted sales and actual sales collection of packs/containers from
consumption and prevalence reflecting the extent of tax avoidance tobacco product users or other
obtained from representative and evasion. Depending on what sources. Products can be examined
population surveys, then the opportunities are being modelled, for tax stamps, local warning labels,
difference between the t wo will reflect this approach can be used to assess other pack markings, and product
the extent of overall tax avoidance individual cross-border shopping and constituents to identify products
and evasion (IARC, 2008). However, direct purchases, as well as small- that do not bear the appropriate
it is likely that there will be some scale smuggling. This approach has stamps/labels/markings or that
temporal biases in tax-paid sales been used widely in the USA (see, include constituents that differ from
measures, as these generally reflect for example: Becker et al., 1994;; those contained in locally produced
shipments at the factory or wholesale Yurekli and Zhang, 2000;; Farrelly et products. As part of the ITC survey
level rather than actual consumption. al., 2003) and, to a limited extent, in in Poland, for example, interviewers
More importantly, there is likely to be European countries (Merriman et al., were trained to recognize Polish tax
some degree of underreporting of 2000). stamps, warning labels and other
tobacco use in population surveys. Survey of tobacco users’ purchase pack markings, as well as the same
To the extent that the bias in each behaviours. Representative surveys for Ukraine, Belarus and the Russian
is constant over time, changes in of tobacco product users that collect Federation in an effort to assess the
the difference between the two can information on various aspects extent of tax avoidance/evasion in the
indicate whether tax avoidance and of purchase behaviour, including Polish cigarette market (IARC, 2008).
evasion are increasing or decreasing purchase location and price, can Merriman (2010) applied a novel t wist
over time (Merriman, 2001). However, be helpful in assessing the extent on this approach by collecting littered
as social norms against tobacco use of various forms of individual tax cigarette packs around Chicago
strengthen over time, the extent of avoidance, including cross-border in an effort to assess the extent of
underreporting in population surveys shopping, direct purchases, and avoidance/evasion of the local Cook
is likely to grow, reducing the validity duty-free purchases (IARC, 2008). County and Chicago cigarette taxes,
300
finding that three fourths of the packs In a competitive market, profits extensively in the USA and in Europe
collected in Chicago did not bear the from illicit tobacco trade will be driven where sufficient data exist and where
Chicago tax stamp. In an ongoing to zero. This implies that, in the long a sizable population lives in a close
effort, the ITC project is collecting run, the cigarette tax revenue the proximity to another state/country
cigarette packs from survey government loses to smugglers is (Vedder, 1997;; Cnossen, 2005).
respondents in a variety of countries entirely consumed by excess travel Several US studies measured the
that will be examined for relevant costs and costs to avoid detection. strength of the relationship between
pack markings, with sophisticated This is a wasteful use of scarce the price/tax differences and cross-
product testing methods used to test societal resources (Bhagwati and border shopping. Baltagi and Levin
for various product constituents that Hansen, 1973). (1986) studied cigarette small-
appear at higher levels in counterfeit scale smuggling in the USA using a
cigarettes. These approaches are Price/tax differences and their dynamic demand model and a panel
limited by observers’ abilities to magnitude as determinants of profit data from 46 states on per-capita
distinguish licit and illicit (particularly sales over the period 1963–1980.
counterfeit) products and by the ability Cigarette prices are high relative to They found the cigarette demand
to differentiate licit and illicit products their production costs, in part due to be sensitive to price differences
based on product constituents, but to higher level of taxation. Price of between the state of residence and
do appear promising for capturing at tobacco products and the total tax the neighbouring states: a 10% price
least some aspects of tax avoidance levied on them are related to the increase in a neighbouring state
and evasion. amount of profit for those involved in caused a 0.8% increase in taxed
illicit trade. sales in the home state in the short
Determinants of tax avoidance/ The absolute price level and run and a 0.21% increase in the
evasion the relative price differences can long run. Baltagi and Levin (1992)
affect the method of delivery of illicit updated their previous study by
The determinants of tax avoidance/ cigarettes to the market. adding data for 8 additional years
evasion are related to tax/price Small-scale smuggling and extending their panel data to 1963–
differences, tobacco products’ legal cross-border shopping are 1988 for these 46 US states. Using
affordability, corruption, weak tax primarily motivated by the relative various model specifications, they
administration and/or customs, and price differences between adjacent found similar results with respect to
informal distribution networks as well geographical areas. These price the price sensitivity of the cigarette
as to the involvement of global and differences may be driven by tobacco market to the price difference with
new firms. The determinants of the taxes or tobacco industry pricing the neighbouring states.
supply of illicit tobacco products are strategy (Baltagi & Levin, 1986, Goel (2008) used more recent
related to the profit from the sale of 1992). The difference in price or/and cross-section state-level cigarette
these illegal products. The greater tax rate represents the upper limit sales data from the United States
the reward and the lower the costs on the incentive to bootleg or shop from 2002 to conclude that price
of supplying these illicit products, the across the border due to transaction differences provide the main
greater the probability an individual costs involved in this form of supply motivation for cigarette smuggling as
will engage in it. The reward depends (Licari & Meier, 1997). compared to non-price factors related
on the difference in profits from Wholesale/large scale smuggling to the probability of apprehension.
legally sold versus illegally supplied is likely to be correlated with the According to his study, a 10%
cigarettes. country’s absolute retail price of increase in the lowest cigarette price
The costs are related to the cigarettes (Merriman et al., 2000), in adjacent US states increases a
probability of detection, the magnitude because the larger the retail price, state’s cigarette sale by about 10%.
of punishment, the opportunity costs the larger the profit for large-scale DeCicca et al. (2010) found
like foregone salaries from other smugglers who pay wholesale relatively larger sensitivity of tax
employment and the cost of the international price for their supply (FIA avoidance and tax evasion to the
capital employed in smuggling. Other International Research Ltd, 1999a). tax/price difference: 1% increase
costs may include the cost of bribery The motivation for small- in home-state price increased the
(Merriman et al., 2000). scale smuggling and legal cross- likelihood of purchasing cigarettes
border shopping has been studied in a neighbouring state by 3.1%
301
while controlling for the distance to A review of 24 peer-reviewed represent a fairly small part of the
the border. However, the study used studies and studies by reputable total cigarette market. However, legal
individual-level survey data that do researchers using US data provided cigarette purchases by tourists and
not capture small-scale smuggling, evidence that tax avoidance/evasion foreign visitors (that are not part of
and self-reported prices that can be via Internet was related to the level of local consumption) were significant—
endogenous. Their result that the cigarette taxes: US smokers living in up to 50% of legal sales. This
probability of cross-border shopping states and cities with high cigarette has been confirmed by the Finish
approaches zero at a distance to the excise taxes were more likely to authorities that reported that legal
border of about 300 miles does not purchase non-taxed cigarettes cross-border cigarette shopping by
seem realistic and indicates problems online than smokers living in low-tax Finish travellers amounted to 12% of
with the model specification. The jurisdictions (Ribisl et al., 2006). total national sales in 1996 (Lipponen
author suggested that there is a In Europe, Merriman et al. et al., 1998).
surge in cigarette tax avoidance (2000) used cigarette sales data Buck, Godfrey and Richardson
and evasion immediately after a for 1989–95, cigarette prices and (1994) showed that in the early to
tax increase, which then quickly frequency of travel from 18 countries mid-1990s there was little incentive
subsides within several months. A to estimate the incentives for small- for cigarette cross-border shopping
similar observation was made by scale smuggling and cross-border between France and Britain, because
Farrelly, Nimsch, and James (2003). shopping. Controlling for the level the savings on 800 cigarettes bought
As noted above, Merriman (2010) of corruption and income, he found in France and taken back to Britain
employed a novel empirical approach that policies that raise incentives for would be outweighed by the cost
to estimate tax avoidance/evasion in small-scale smuggling and cross- of the trip. Cross-border shopping
the city of Chicago and its sensitivity border shopping, such as a cigarette existed at that time, but only when
to tax differences in the neighbouring tax increase, will significantly reduce smokers were already across the
jurisdictions. He collected a random domestic tax-paid cigarette sales. border for other reasons. Estimates
sample of 2391 littered cigarette If these incentives to bring cheaper of United Kingdom Department of
packs in the city in 2007 and studied cigarettes from abroad were reduced HM Customs & Excise confirmed that
their probability of having the correct to zero, the official domestic sale legitimate cross-border shopping
local tax stamp. The difference would increase by 3%. If foreigners was a minor problem in 1997 when
between the tax in Chicago and would not have incentives to buy the legitimate personal imports of
surrounding counties equal to $2.68 cheaper cigarettes in a country, this tobacco products was less than 0.5%
decreased the probability that a country’s domestic sale would fall by of total cigarette sales (HM Customs
littered pack has a local stamp by 1%. For example, a unilateral 10% & Excise, 1998).
almost 60 percent. On the other cigarette price increase in Germany The situation changed in the
hand, increasing the distance to would reduce yearly cigarette sale second part of 1990s when cigarette
the lower-tax state border by a one by 6 packs per capita, but would prices in the United Kingdom
mile increased the probability a increase yearly cigarette purchases increased by about 25% from 1997 to
pack has a local stamp by about one abroad by 3 packs per capita, 2000. The United Kingdom Treasury
percent. That means that compliance resulting in a 3-pack per capita estimates that the market share of
increases rapidly with the distance reduction in consumption. illicit cigarettes in the United Kingdom
to lower-taxed border and that the Taal et al. (2004) analysed sales rose from about 3% in 1996–97
distance provides a significant and survey data from Estonia, to about 18% by 1999–2000 (HM
barrier to tax avoidance. However, a European country with high Customs & Excise, 2000).
the results of Merriman (2010) are not incentives for small-scale smuggling In France, similarly, the substantial
directly comparable to econometric and cross-border shopping from tobacco tax increases in 2003 and
studies that report tax compliance 1993 to 2000, when cigarette prices 2004 that led to higher cigarette
at the national or state level due to there were up to four times lower than prices were blamed for an increase
the potential selection bias. This bias in neighbouring Finland and Sweden in cross-border purchases of tobacco
arises due to the possibility that those (but considerably higher than in products (both legal and illegal)
who litter are also less law obedient, another neighbour, the Russian IURPDQHJOLJLEOHDPRXQWWRí
therefore more likely to engage in tax Federation). They found that illegal of total sales in 2005 and 2006
avoidance/evasion. purchases of cigarettes by Estonians (Lakhdar, 2008).
302
The evidence on tax avoidance/ They found that the countries with Islamic Republic of Iran (World
evasion in Asia is scarce. Tsai, Sung, the lowest income had the lowest Health Organization, 2003) have
Yang and Shih (2003) used survey average prices for legally sold prohibited or severely limited the
data collected in 2000 among 437 cigarettes, but also the highest illicit importation of foreign cigarettes.
smokers of imported cigarettes and cigarette market share. This suggests The tax on imported cigarettes in
found that higher cigarette prices a negative relationship between the China was 244% in 1997 (reduced
were the primary motivation for illicit cigarette market share and the to 217% in 1999), but foreign
purchasing of smuggled cigarettes cigarette price level. cigarettes with prices lower than
in Taiwan, China. A 1% increase in Joossens and Raw (1998) found this tax were easy to find in urban
cigarette price raised the likelihood the same situation in EU countries China (Hu & Mao, 2002). Viet Nam
of purchasing smuggled cigarettes in 1995: many European countries banned cigarette imports from 1990
at least 2.60 times (95% confidence with high tobacco taxes and prices till 2007. Yet, foreign cigarettes were
interval: 1.08–6.26). Smokers who had low estimates of cigarette sold at a premium in the street of
spent more money on cigarettes were smuggling, while illicit cigarette trade Hanoi (Joossens, 2003). Even after
more likely to purchase smuggled was prevalent in southern European lifting the import ban, cigarettes
cigarettes, but personal income countries where taxes and prices and cigars are subject to an import
was not significantly associated were low. The authors suggest that duty of up to 225%. The illegally
with smuggled cigarettes purchases the size of illegal market in a country imported brands (primarily Winston
(Tsai et al., 2003). is determined by fraud and illegal and Magna) represented about 50%
Nelson (2002) suggested that trade, and not by the level of tobacco of the total cigarette market in the
the size of the potential cross-border taxes/prices. Islamic Republic of Iran in 1994, and
market plays an important role in the Data from southeastern Asia the industry estimated that the illegal
formulation of tax policy for cigarettes. also point to no relationship between or tax-free products had 68% market
US states with a large potential countries’ tax rates, cigarette prices share in the country (World Health
cross-border market are more likely and the level of illicit cigarette trade. Organization, 2003).
to set tobacco tax rates below that For example, in the early 2000s Two older studies suggest that
of neighbouring states to attract non- cigarette tax in Singapore was smuggling in low- and middle-income
residents to purchase cigarettes in among the highest in the region countries increases with the level
their state, therefore exporting their (about 51% of retail price), but the of tariffs and taxes (Bhagwati and
tax burden. This seems to be the estimated share of illegal cigarettes Hansen, 1974;; Simkin, 1974).
case of Luxembourg, a country with on the market was only 2% of Differential pricing for cigarettes
a high density of foreign population domestic sales. On the other hand, intended for export and for the
living near its borders that is setting its illicit cigarettes accounted for about domestic market can motivate re-
cigarette taxes at a level considerably 37% of domestic sales in Cambodia, import of products designated for
lower compared to its neighbours. yet the tax represented only 20% export. To limit this behaviour, in
Using data from 1993, the estimates of retail price, one of the lowest tax 2000 the US federal government
indicate that 85% of cigarette sales rates in the region (Ministry of Health banned imports of cigarettes that
in Luxembourg were due to cross- Republic of Indonesia, 2004). are intended for export only (FIA
border sales (Joossens & Raw, 1995). International Research Ltd, 1999b).
There is less evidence to support Differential treatment of domestic
the theory that wholesale/large-scale versus foreign products Cost of supplying illicit tobacco to the
smuggling is likely to be correlated market determines net profit
with the country’s absolute retail price. Restrictions on the supply of imported
Merriman et al. (2000) found tobacco (by imposing quotas, tariffs, The costs of supplying illicit products
no significant correlation between and other non-tariff barriers such as a include the cost of manufacturing
experts’ estimates of large-scale prohibition on sale, which is similar to and/or obtaining tax-free cigarettes,
smuggling from 38 countries and the an infinite tax) can lead to significant the cost of access to capital,
average legal cigarette price. price differences and motivate illicit transportation, distribution and
Joossens et al. (2010) employed trade (Taylor et al., 2000). countering the government’s effort
a larger data set of low-, middle- and China (Hu & Mao, 2002), Viet to control tobacco illicit trade as
high-income countries (84 in total). Nam (Joossens, 2003) and the this relates to the possibility that
303
the smuggler\counterfeiters will The transportation costs for small- Insufficient pack markings reduce
be apprehended and penalized if scale smuggling will be lower if the probability of illicit cigarettes being
caught. In addition, these costs a large share of the population confiscated, therefore reducing the
depend on the opportunity cost of lives near the border with another costs of involvement in illicit cigarette
time (as this influences the cost jurisdiction (Saba et al., 1995). This trade. Press articles have reported
of labour) and social norms with is particularly relevant for Europe that introducing a tax-stamp by the
respect to supplying illicit cigarettes and the USA, where a sizable state of Michigan in 1998 helped to
(people who believe that engaging population lives in a close proximity reduce the sale of illicit cigarettes
in cigarette smuggling is immoral to another state/country (Vedder, (Hyde, 1998).
will need to be compensated more 1997;; Cnossen, 2005). The relatively weak government
to convince them to get involved in Joossens (1998) argued that the controls and the slowness of
this activity). ease of evasion of legal authorities the judicial process reduce the
The costs of obtaining legally might be more important than price opportunity costs of illegal cigarette
produced cigarettes are related to when it comes to the proclivity of manufacturing in Brazil (Ramos,
export/wholesale prices and can individuals to engage in tobacco 2009). The majority of cases
vary by the country of destination smuggling. involving illicit cigarettes do not
(Yurekli & Sayginsoy, 2010). The perceived probability of fall under criminal jurisdiction, and
New technology such as interdiction depends on the actual contraband cases are mostly treated
sophisticated and less-expensive level of enforcement. Anthony (2004) as low-priority misdemeanours that
computer scanners reduce the suggests that drug smugglers ignore end with the release of the prisoner
production costs of counterfeit the risk of being caught up to a and sometimes even include the
cigarettes and boost their supply on point, and that there is a threshold restitution of goods and/or vehicles
the market, such that in 2007 these interception rate beyond which the involved. Inadequate legislation
cigarettes made up 15% of illicit smuggling rapidly declines. However, compounded by a lack or absence
cigarette trade globally (Euromonitor some smugglers are undeterred of specialized prosecutors and
International, 2008). even when there is 100% probability judges does not enable authorities to
Obtaining tax-free tobacco of detection. This could be due to confront associated drug trafficking
can require capital investment the practice of receiving advanced and money-laundering activities. In
into transportation equipment. payments for taking the risk, which addition, handling illicit merchandise
Smugglers’ transport costs are likely could be greater than the lost earnings is socially accepted in Brazil. As
to be greater than those engaged due to jail time. a result, the production of illegal
in transporting a similar weight and In the United States, small- cigarettes in Brazil reached some 9
volume of legal goods, because of scale smugglers often use the billion cigarettes in 2007, about 6.4%
the need to avoid detection (Bhagwati superhighways, where the likelihood of the total cigarette market (Ramos,
and Hansen, 1974). Therefore, this of detection is minimal (Walsh & 2009). The size of the illegal market
activity represents inefficient use Ottaway, 2000). Compared to large- in Brazil is unclear, but experts
of limited resources (Bhagwati and scale smugglers, they face lower estimate that it is about 30% of total
Hansen, 1974). In Malaysia, for costs of supplying illegal products to sales (Ramos, 2009).
example, smugglers use speedboats the market, because purchasing tax- The opportunity costs of time
to increase their chances of escaping paid tobacco in a low-price jurisdiction can determine the willingness of
when chased by police and/or border and transporting it to a high-priced population to engage in illegal
patrol (Unknown, 1995). one reduces the legal risk for those activities. When Australian farmers’
On the other hand, Jensen involved in the business (Licari & opportunity costs of time dropped
and colleagues (1991) argued that Meier, 1997). due to declining official market price
transport costs in the USA account Goel (2008) found that the of raw tobacco and the diminishing
for only 0.5 percent of the value of probability of apprehension, legal outlet for tobacco, some of
tobacco products, and that their measured by the number of police per them began to divert their tobacco
effect on smuggling is negligible. 1000 inhabitants, plays a relatively to illicit markets (Geis et al., 2003;;
Transportation costs may wary minor role in cross-border cigarette Geis, 2005).
by geographical regions and the smuggling in the USA. The costs of distribution of illicit
type of tax avoidance/evasion. cigarettes are important motivators.
304
These are discussed in more details Germany. Taal et al. (2004) reported US$ 2 068 668 revenue loss for the
below. that Finns made approximately 3 state of Washington (The Financial
million visits to Estonia in 1998, Action Task Force, 2010). Duty-
The supply chain for illicit tobacco and that travellers from Finland and free shops facilitated small-scale
(distribution) reduces cost of illicit Sweden consume up to 50% of smuggling from Estonia to Finland and
transactions total legal cigarette sales in Estonia. Sweden (Taal et al., 2004).
Higher prices in Sweden motivated The Akwesasne Indian reservation,
Increase in international/interstate airline passengers from the Russian located on the US–Canadian border,
trade. Large-scale organized Federation to bring with them was used by the tobacco industry for
cigarette smuggling generally cigarettes when they travelled to re-importation of cigarettes exported
involves cigarettes that have passed the country (Unknown, 1994). There from Canada to avoid Canadian taxes
through a wide range of international is anecdotal evidence that youth (Cunningham, 1996;; Segal, 1999).
traders (European Parliament, in Scotland have been given free Native American reservations in
1997). A loose export-import vacations in Europe in exchange for the USA are also the primary source
recording system combined with smuggling tobacco back to Scotland of tax-free cigarettes sold over the
large quantities of traded cigarettes (Unknown, 1994). Internet. In 2005, nearly two thirds of
reduces the probability of detection. Presence of tax free zones. Tax- the US web sites selling cigarettes
Pitt (1981) concluded that the greater free zones or free-trade zones play a were affiliated with a reservation. In
the volume of legal trade, the lower significant role in illicit trade in both 2003, 95% of Seneca tribe cigarette
the costs of smuggling. In Hong genuine and counterfeit cigarettes sales were conducted over the web or
Kong Special Administrative Region, (World Customs Organization, 2007). phone (Chen, 2008).
trucks with smuggled cigarettes The Financial Action Task Force Informal market/retail networks,
have been disguised as containing (FATF) of the OECD estimated that street-selling, existing smuggling routes
duty-free components for assembly there were about 3000 free-trade and black markets. The opportunity
at factories within mainland China zones in 135 countries around the cost of labour could determine the
(Unknown, 1999). The opening of the world in 2009 and that their numbers existence of the informal distribution
market in Taiwan, China in 1987 led are growing (The Financial Action networks. If the potential rewards
to an increase in cigarette smuggling Task Force, 2010). from legal occupations are very small
measured by an increase in cigarette The International Consortium of compared to engagements in shadow/
seizures and reports by the industry Investigative Journalists discovered underground economy, people should
(Wen et al., 2006). that free-trade zones on the island be more willing to provide an informal
Liberalization and frequency of Aruba and in Colombia are being distribution network for illicit products
of international travel. The ease of used for cigarette smuggling into including cigarettes (Merriman et
small-scale smuggling is proportional other areas of Colombia, such as the al., 2000). However, Merriman et al.
to the total number of cross-border city of Bogota (Beelman, 2000). (2000) found no association between
travellers taking into account Free-trade zones have been experts’ estimates of smuggling in 38
the stringency of border controls misused for money laundering and countries and these countries’ income
(Merriman et al., 2000). terrorist financing. A US company measured by per-capita GDP.
Travellers can also take used a free-trade zone to repackage In some countries, illicit cigarettes
advantage of duty-free sales. In 1996, tobacco and ship it to another free- are distributed via separate channels,
approximately 45 billion cigarettes trade zone, ultimately smuggling it but in some countries illicit tobacco
were sold through duty-free outlets. into US market. The profit was used products are distributed alongside
This represents 0.8% of all cigarettes to finance the Abu Sayyaf Group, a the legal products via regular retail
sold in the world (Market Tracking terrorist organization based in the channels. The existence of smuggling
International Ltd, 1997). Philippines (The Financial Action routes and black markets for other
Merriman et al. (2000) partially Task Force, 2010). Large-scale products as well as the presence of
attributed a higher share of the cigarette smuggling from China to the unlicensed distributors reduce the
German cigarette market being United States (state of Washington) transportation and the distribution
supplied via cross-border shopping has been facilitated by a free-trade costs of illicit tobacco supply (Thursby
and small-scale smuggling to high zone in Hawaii. The cigarettes seized & Thursby, 2000;; Campaign for
frequency of travel from and to in this free-trade zone represented a Tobacco-Free Kids, 2008).
305
Joossens and Raw (1998) and time and the lower the probability of These resources are related, for
Joossens et al. (2000) pointed to the buying illicit products. In California, example, to travel costs or to costs of
association between the tradition of most smokers purchased their getting internet access (Hyland et al.,
street-selling and higher share of cigarettes from the most convenient 2005;; DeCicca et al., 2010).
illicit cigarettes in markets in Spain (but more expensive) sources Social norms with respect to the
and Italy. despite relatively large price use of illicit products and government
Social networks often serve as differences between the legal and interventions can be an important
important distribution channels for illegal cigarettes (Emery et al., 2002). determinant of the demand for non-
illicit products (Merriman et al., 2000). Very few (5.1%) of California smokers taxed products. In Taiwan, China,
In the United Kingdom, between purchased cigarettes from non- or smokers who opposed cigarette
2.5% and 3.3% of all cigarettes sold lower-taxed sources, such as out-of- taxation policy were 1.69 times more
in 2002–2003 were supplied via an state outlets, military commissaries or likely to buy smuggled cigarettes
informal network of independent the internet (Emery et al., 2002). (Tsai et al., 2003). United Kingdom
sellers (Hyland et al., 2006). Coleman The opportunity costs of time smokers living in socioeconomically
(1998) found that the majority of illegal seemed to be a major deterrent of deprived areas were quite supportive
tobacco products consumed in the cross-border shopping between of smuggling, as they perceive it
United Kingdom were sold in pubs. California and Mexico, as the wait helpful in dealing with the rising
Technology allowing virtual to enter Mexico was well over 30 financial costs of smoking (Wiltshire
transactions (online purchases minutes and travellers were also et al., 2001). DeCicca et al. (2010)
and mail-orders). In the USA, the subject to a customs check on return reports that anti-smoking sentiment
number of web sites selling cigarettes to the USA (Emery et al., 2002). reduces the likelihood of cross-
increased from 88 in 2000 to 772 in Using the 2003 CPS Tobacco border shopping.
2006 (Chen, 2008). Smokers in the Use Supplement (TUS) on individual
US who purchase cigarettes online purchase quantities and locations, Ineffective tobacco tax
are primarily motivated by lower Chiou and Muehlegger (2008) found administration, insufficient tax
prices, because Internet vendors that the impact of a state’s tax change enforcement resources, and lack of
generally sell cigarettes without on cigarette sales depends upon the control over the movement of tax-
paying tobacco excise taxes for the tradeoffs between the cigarette price free cigarettes
destination state (Kim et al., 2006). difference and distance to the state
Almost all online cigarette sales are with lower cigarette prices. They The lack of clarity regarding the
illegal due to failure to report the calculated that a consumer is willing regulations covering the control of
transaction and/or verify the age to travel 2.7 miles to save $1 on a free-trade zones is a major reason for
of the buyer (Chen, 2009). With the pack of cigarettes. The willingness their misuse for illicit trade according
recent crackdown on this form of sale to travel depends on the number to the World Customs Organization.
in the USA, the overseas internet of cigarette consumed: smokers As the government and the customs
vendors expanded their role in the US who report smoking every day (or authorities dispute who has the
market. In 2003, 10% of all internet individuals who smoke more than 14 jurisdiction over free-trade zones,
vendors were based outside the cigarettes a day) have a significantly enforcement loopholes allow these
USA, but in 2006 over 45% of them lower marginal cost of travelling than zones to play a facilitating role in
were based overseas (Chen, 2009). smokers who only report smoking illicit cigarette trade (World Customs
These overseas vendors are beyond some days (or those who smoke less Organization, 2007).
the reach of the US law enforcement. than 14 cigarettes a day), and are Examining trade transaction
therefore more likely to travel across between the USA and North Africa
Transaction cost associated the border to purchase cigarettes. and between Hong Kong Special
with illicit tobacco products Everyday smokers purchase Administrative Region and the rest
approximately 3 times as many of SE Asia showed that inadequate
The convenience of a transaction that cigarettes when crossing a border controls over the in-transit cigarettes
reflects the opportunity cost of time is than do some-days smokers. results in a substantial leakage,
related to costs of illicit tobacco use. A minimum set of resources seems with many of the cigarettes failing to
The less convenient the transaction, to be necessary for a person to have arrive at their intended destination
the higher the opportunity cost of access to low/untaxed cigarettes. (Joossens et al., 2000).
306
In the USA, a change in the associated with a two-percentage- and fostered the development of a
balance of enforcement activities point decrease in experts’ estimate of regional net of smuggling channels
between the US state and US cigarette smuggling. Since corruption affecting neighbouring countries
federal authorities after passing the is more pervasive in low-income like Bulgaria, Romania and Albania.
Contraband Cigarette Act (CCA) in and middle-income countries, these These illegal activities became an
1978 generated a loophole in the countries are at greater risk of large- important source of income during
tax audit (Thursby and Thursby, scale smuggling activities. the war for people of all social groups
2000). This led to an increase in Cambodia is an example of (Hajdinjak, 2002).
illicit cigarette trade as well as to a country where a high level of The US Government
a change in the preferred method corruption facilitates cigarette Accountability Office reported that
of tax evasion. Commercial/large smuggling. It is estimated that 79% terrorist organizations including
scale smuggling began to focus on of cigarettes imported into Cambodia Hezbollah made money through
underreporting of the amount of are re-exported or smuggled across the tobacco black market (United
cigarettes released to the distribution Cambodia’s borders (Ministry of States General Accounting Office,
(“diversion”) that could have been Health Republic of Indonesia, 2004). 2003). The profit from large-scale
discovered by the tax audit. The The European Commission cigarette smuggling via free-trade
study concluded that effective (1998) identified over 50 criminal zones was used to finance the
enforcement requires participation of networks potentially engaged in terrorist organization Abu Sayyaf
both law enforcement agencies and large-scale tobacco smuggling. Group based in the Philippines (The
tax administrators. In 2005, the investigation by the Financial Action Task Force, 2010).
A law that would allow US states European Council concluded that Corruption and organized crime
to recover taxes from cigarettes sold the threat imposed by cigarette does not seem to be a predictor of
over the Internet (the Jenkins Act) is smuggling within the European Union small-scale smuggling and cross-
enforced by the US federal authorities, is primarily related to the involvement border shopping. Goel (2008) found
who are much less motivated than of organized crime (Council of the that the average number of public
individual states to enforce the law European Union, 2005). corruption convictions is not a
(Goolsbee et al., 2007). However, Cigarette smuggling is claimed to statistically significant predictor of
in October 2002 Washington State be the third largest illegal business cross-border cigarette smuggling in
successfully applied the Jenkins in Germany, behind drug trafficking the USA.
Act to internet sales, leading to and illegal gambling, with extensive
similar efforts in many other states involvement of organized crime (Von Tobacco companies’ profit motivation
(Chaloupka et al., in press). Lampe, 1999 as cited in Merriman, translated to their marketing strategy
2001). In southern Italy, Calabrian and political agendas reduce the cost
Corruption, war and organized crime gangsters have been involved in the of supplying illicit cigarettes
facilitate law circumvention, reducing tobacco smuggling (Unknown, 1997).
the cost of supplying illicit products The presence of organized crime The tobacco industry itself often
networks can increase pressure promotes the smuggling of their
Criminal networks specializing in on legitimate distribution networks products, because smuggling does
cigarette smuggling operate more by reducing their profitability or by not impact their profit margins,
easily in countries where corruption forcing them to join in the black reduces the impact of higher tax/
is high, the control of the authorities market (Joossens et al., 2000). price on cigarette consumption
is lax and commodities other than Smuggling of various goods, and can be used to advocate for
tobacco are also being smuggled and particularly cigarettes, was an excise tax reduction as a way to
(Joossens, 1999). Using data from a prosperous “business” in the reduce smuggling and associated
38 countries, Merriman et al. (2000) Yugoslav republics during the Balkan criminal activities (Leverett et al.,
found that the level of corruption War in 1991–1995, helping to finance 2002). In fact, cigarettes confiscated
(measured by the transparency weaponry and otherwise supporting and destroyed by law enforcement
index) is positively related to the war operations (Hajdinjak, 2002). authorities increase demand for
size of the illicit cigarette market. Smuggling was done with the close replacement cigarettes and therefore
A one-point improvement in a cooperation of politicians, their the profitability of the industry
country’s transparency index was security forces and organized crime, (Leverett et al., 2002).
307
Delipalla (2009) developed a Canada’s significant increases of tobacco growers and cigarette
decision-making model analysing in cigarette tax in 1980s and early manufacturers, has long argued
the factors motivating the industry to 1990s motivated the tobacco that high taxes relative to those
get involved in illicit tobacco trade. industry to orchestrate smuggling of neighbouring countries are
The model assumes that the industry of cigarettes to Canada to exercise responsible for the growth in illicitly
operates as an oligopoly, and that pressure on Canadian government traded cigarettes. They claim that illicit
the market does not differentiate to reduce these taxes (Canadian cigarettes represent 20% of the total
between legal and illegal products Cancer Society, Non-smokers’ market (Tobacco Institute of South
(their prices are similar). To maximize Rights Association, Physicians for Africa, 2008) without providing solid
its profit, each firm decides the a Smoke-free Canada, & Quebec evidence to support this claim. A peer-
total quantity sold on the market Coalition for Tobacco Control, reviewed academic study contradicts
and the fraction of that total it will 1999). The industry developed the industry estimates, estimating
attempt to sell illegally (without tax) a tax-dodging scheme allowing the maximum penetration of illicit
taking into account the probability of the company to smuggle billions FLJDUHWWHVRIíRIWKHWRWDO
successful tax evasion, the level of of cigarettes into Canada using market in 2007 (Blecher, 2010).
enforcement, the cost of smuggling, a Native American reservation in Illicit cigarette trade can provide
and fines paid if caught. The paper the USA (Segal, 1999;; Sugarman, the industry a market entry for
then examines the relative effects 2002). In response, the government prohibited brands (brands that
of specific and ad valorem taxes cut cigarette taxes in 1994. Imperial are not legally sold in the market).
on the firm’s evasion decision given Tobacco Canada Limited and Producers then use customer
that it is optimal for a firm to engage Rothmans Benson & Hedges, the loyalty as a political wedge to lobby
in tax evasion. It concludes that two tobacco companies involved for legal access to the market. The
the optimal fraction of tax-not-paid in this smuggling scheme, pleaded international trade journal World
products is increasing in sales under guilty in 2008 and admitted “aiding Tobacco reported in 1996 that
purely specific taxation and when persons to sell or be in possession “smuggling has helped to promote
the ad valorem tax is imposed as a of tobacco products manufactured some of the world’s leading brands
percentage of the fixed wholesale in Canada that were not packaged in markets which had remained
price. When the ad valorem tax and were not stamped in conformity closed to foreign imports” (Market
is imposed as a percentage of the with the Excise Act” between 1989 Tracking International Ltd, 1997).
retail price, the optimal fraction of and 1994. The companies agreed to Since the 1980s, the British
tax-not-paid products is decreasing. pay CA$1.15 billion, the largest fine American Tobacco (BAT) relied on
The analysis suggests that (for ever levied in Canada (Joossens illegal channels to supply markets in
given tax rates and enforcement) and Raw, 2008). R. J. Reynolds, the Africa (LeGresley et al., 2008). Illicit
if demand falls (for example due to third tobacco company involved in cigarette trade has been an important
effective tobacco control policies), the smuggling scheme, has agreed component of BAT’s market
smuggling will also be reduced to pay CA$325 million ($325 million) entry strategy to gain leverage in
under a specific tax regime, but it will to the Canadian governments in negotiating with governments for tax
increase under an ad valorem one. April 2010 (Schneider, 2010). concessions, compete with other
The impact of a mixed tax structure There is evidence that British transnational tobacco companies,
on optimal illicit trade is ambiguous. American Tobacco (BAT) in circumvent local import restrictions,
The analysis concludes that higher Bangladesh had some degree and to gain a market presence. BAT
tax rates do encourage smuggling of control over the proportion exploited weak government capacity
under purely or predominantly ad of contraband products on the to combat illicit trade and gained
valorem taxation, but their effect market. BAT has altered the flow substantial market share in major
is ambiguous under purely or of illicit cigarettes while making the countries (LeGresley et al., 2008).
predominantly specific taxation. argument that tax increases lead to Smuggling has been an important
The industry has used the higher consumption of illegal tobacco component of BAT’s market entry
occasion of a tax increase to products (Collin et al., 2004). strategy in Lebanon, a country with a
promote illicit cigarette trade to put In South Africa, the Tobacco state tobacco monopoly and chronic
pressure on the governments to Institute of South Africa (TISA), a political instability. After the end of the
reverse their decision. body which represents the majority civil war in the early 1990s, BAT and
308
other transnational tobacco companies Kingdom (HM Customs & Excise, of global cigarette consumption is
sought a legal presence in the country, 2001a, 2001b;; House of Commons illicit, including 17% in low-income
but continued to achieve substantial & Treasure Committee, 2005). countries, 12% in middle-income
sales through illicit cigarette trade The volume of seized counterfeit countries, and 10% in high-income
(Nakkash & Lee, 2008). cigarettes increased threefold countries.
Smuggling has been used as between 2000 and 2004. In 2003–
a market entry in Argentina, Italy, 2004, 5 4% of seized c igarettes were Europe
Islamic Republic of Iran, Bulgaria, counterfeit (House of Commons,
the former Soviet Union and China, Treasure Committee, 2005). European Union. Using cigarette
for example (Joossens & Raw, 1998;; sales data 1989–95 from 18
World Health Organization, 2003;; Extent of tax evasion and tax European countries, Merriman et al.
Gilmore and McKee, 2004). avoidance globally, regionally (2000) estimated that, in a typical
After the market opening in and in key countries European country, the share of
Taiwan, China in 1987, Japan cigarettes acquired by small-scale
Tobacco International (JTI) set up a Joossens, Merriman, Ross and smuggling and/or cross-border
Swiss plant as a legal cover for its Raw (2010) collected data on shopping accounted for about 3% of
smuggling operation to the country. illicit cigarette trade from different domestic consumption.
This allowed JTI to overcome the sources. Their estimates of illicit Based on in-depth analysis of
existing legal quota of exports to market share (Table 8.1) are based data collected by the professional
Taiwan, China (Wen et al., 2006). on academic articles, official services company Klynveld Peat
The involvement of the industry government publications, estimates Marwick Goerdeler (KPMG, 2005),
in smuggling operations in Europe from market research companies a study commissioned by the
became clear when the level of illicit (whose clients include the tobacco European Commission estimated
trade on the continent substantially industry and governmental that in 2004 total market penetration
declined following the 2004 and organizations), tobacco trade of illicit cigarette trade represented
later agreements between major journal articles, newspaper articles approximately 8–9% of European
manufacturers (e.g. PMI, JTI) and the and sometimes estimates from Union (which had 25 Member
EU authorities without admission of personal contacts in customs States at the time, designated
liability. This agreement financially organizations. The estimates vary EU25) cigarette sales. The above-
motivated the industry to control greatly in rigour and in the measure cited report noted also that the illicit
the illegal supply of their products of illicit trade used. Some for market share in the new EU Member
to the EU market. Since counterfeit example express the size of the illicit States (Estonia, Hungary, Lithuania,
cigarettes are excluded from the market as a percentage but without Poland and Slovakia) were far above
agreement, companies have an saying what it is a percentage of, the EU25 average (KPMG, 2005).
incentive to have seized products and some do not even state what The KPMG report has limitations,
declared as counterfeit (World measure was used. Nor is there a as it is based on cigarette seizures
Health Organization, 2007). clearly defined methodology for in the EU and on studies provided
The improved collaboration assessing if an estimate is accurate. by the tobacco industry and
between the industry and the United Joossens and colleagues (2010) governments, however as its
Kingdom government reduced have included in their estimates estimate falls between the higher
the availability of genuine brand what seem reasonable in terms of estimates from the United Kingdom
cigarettes on the black market. the country’s population, smoking and eastern and central European
This can be regarded as evidence prevalence, legal infrastructure and countries and the lower estimates
that cigarette manufacturers are other relevant parameters. Thus a from southern European countries
indeed in a position to control combination of methods, including like Spain and Italy, the overall f igure
the supply of illicit cigarettes to informed expert judgement, is of 8–9% is a reasonable estimate.
the market. However, counterfeit often necessary to cross-validate In 2008, the European
products began to play a larger estimates. Bearing in mind these Commission estimated that around
role in the supply of illicit cigarettes methodological c hallenges, the data 13% of total consumption in the 27
as evidenced by the type of illegal on illicit and legal market estimates EU Member States had not been
cigarettes seized in the United from 2007 show that almost 12% taxed in the country of consumption.
309
Table 8.1. Estimates of the illicit cigarette market around the world
Country % illicit market Measure used Year Illicit % total market
Hong Kong Special 42 Percentage of legal sales 2005 30

Administrative Region
UAE 30 Percentage of legal cigarette sales 2005 23
Singapore 18 Percentage of legal cigarette sales 2005 15
Percentage of total cigarette market: estimate

Canada 15–20 2007 15–20
based on multiple sources and surveys
Percentage of consumers that purchased
USA 13–25 1992–2002 13–25
lower-priced cigarettes
Percentage of total cigarette consumption (not
United Kingdom 13 2006–2007 13
including hand rolled tobacco)
Taiwan, China 11 Percentage of legal cigarette sales 2005 10
Australia 6 Percentage of legal cigarette sales 2007 6
Israel 5 Percentage of legal cigarette sales 2005 5
Saudi Arabia 4 Percentage of legal cigarette sales 2006 4
Italy 2 Percentage of total cigarette market 2006 2
Japan 2 Percentage of legal cigarette sales 2006 2
New Zealand 1 Percentage of legal cigarette sales 2003 1
Spain 1 Percentage of total cigarette market 2006 1
Upper-middle-, lower-middle- and low-income countries
Georgia 49 Percentage of total cigarette market 2005 49
Bolivia 46 Percentage of legal cigarette sales 2005 32
Albania 40–50 Not stated Not stated 40–50
Bosnia & Herzegovina 35–45 Not stated Not stated 35–45
Smuggling as a percentage of total cigarette

Uzbekistan 40 2006 40
consumption
Ethiopia 38 Percentage of total cigarette market 2006 38
Brazil 35 Percentage of legal cigarette sales 2006 26
Lao People’s Democratic

35 Not stated 2005 35
Republic
Iraq 35 Percentage of total cigarette market 2006 35
310
Former Yugoslav Republic

30–35 Not stated Not stated 30–35
of Macedonia
Cameroon 26 Percentage of legal cigarette sales 2005 21
Syrian Arab Republic 26 Percentage of total cigarette market 2007 26
Estonia 19–32 Percentage of total cigarette market 2003 19–32
Sudan 25 Percentage of legal cigarette sales Not stated 20
Zambia 25 Not stated 2003 25
Croatia 25 Percentage of legal cigarette sales 2005 20
Malaysia 24 Percentage of total cigarette market 2008 24
Venezuela 23 Percentage of legal cigarette sales 2005 19
Russian Federation 23 Percentage of legal cigarette sales 2004 19
Peru 23 Percentage of total cigarette consumption 2006 23
Lebanon 23 Not stated 2000–2006 23
Morocco 23 Percentage of total cigarette market 2006 23
Algeria 20 Percentage of total cigarette market 2007 20
Philippines 19 Percentage of legal cigarette sales 2006 16
Nigeria 18 Percentage of total cigarette consumption 2006 18
Ghana 18 Percentage of legal cigarette sales 2005 15
Pakistan 17 Percentage of total cigarette market 2005 17
Armenia 16 Percentage of total cigarette consumption 2004 16
Côte d’Ivoire 15 Percentage of legal cigarette sales 2005 13
India 14 Percentage of total cigarette consumption 2004 14
Columbia 14 Percentage of total cigarette consumption 2004 14
Islamic Republic of Iran 14 Percentage of total cigarette market 2007 14
Ecuador 12 Percentage of total cigarette consumption 2006 12
Uruguay 12 Percentage of total cigarette market 2006 12
Guatemala 12 Percentage of total cigarette market 2006 12
Jordan 10–12 Percentage of total cigarette market 2007 10–12
311
Table 8.1. Estimates of the illicit cigarette market around the world
Thailand 11 Not stated Not stated 11
Yemen 11 Percentage of legal cigarette sales Not stated 10
Turkey 11 Percentage of total cigarette market 2006–2007 11
Nicaragua 10 Not stated 2001–2002 10
Panama 10 Percentage of legal cigarette sales 2000 9
Tunisia 10 Percentage of total cigarette consumption 2007 10
El Salvador 10 Not stated Not stated 10
Argentina 10 Percentage of total cigarette market 2006 10

Viet Nam 10 2004 10
market
Percentage of total cigarette market:
China 8–10 Multiple 8–10
extrapolated from multiple sources
Kazakhstan 9 Early 2000s 9
consumption
South Africa 9 Percentage of total cigarette consumption 2007 9
Percentage of total cigarette market: multiple

Ukraine 9 Multiple 9
sources
Costa Rica 9 Percentage of legal cigarette sales 2006 8
Indonesia 5–6 Percentage of total cigarette market 2005 5–6
Mexico 3 Percentage of total cigarette sales 2006 3
Chile 3 Percentage of legal cigarette sales 2006 3
Adapted from Joossens L, Merriman D, Ross H, Raw M. The impact of eliminating the global illicit cigarette trade on Health and Revenue. Addiction. 2010;; 105(9):1640–1649 with
permission from John Wiley and Sons. Notes: The% illicit market has been rounded to nearest integer;; UAE = United Arab Emirates;; Russian Fed. = Russian Federation;; the EU-25
average illicit market share is 8.5%. In this table we list only countries for which we have country-specific data. The EU-25 countries, which are included in the model calculations using
8.5% illicit market share, are: Austria, Belgium, Cyprus, Czech Republic, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Ireland, Italy, Latvia, Lithuania, Luxembourg,
Malta, Netherlands, Poland, Portugal, Slovenia, Slovakia, Spain, Sweden, United Kingdom. EU-27 includes in addition to the EU-25 Bulgaria and Romania. Estimates contained in
Table 8.1 uses standardized terminology determined by the authors to facilitate cross-study comparison;; verbatim study terminology are reported in the online version. The third
column, ‘Measure used’, reports the measure used in each source document, as described in the original publication, including where the measure was not specified. There is no
standard measure thus the table reflects the varied data sources.
About 8–9 percentage points had commissioned by the Commission in products being sold in the past six
been attributed to tax evasion and 2005 (KPMG, 2005). months which they think might have
around 4 percentage points to legal According to a survey of 26 500 been smuggled into the country. The
cross-border shopping (European Europeans (EU27 plus Norway;; EU27 proportion of respondents who had
Commission, 2009). These is EU25 plus Bulgaria and Romania;; seen potentially smuggled tobacco
estimates are in line with the findings see Table 8.1 footnotes) conducted in products being sold in the past six
of the KPMG study cited above on December 2008, just over one tenth of months was the highest in Lithuania
the EU25 tobacco market in 2004 EU citizens (12%) had seen tobacco (36%), followed by Greece (25%),
312
then Poland, Hungary and Latvia (22– suspected that some of the USA. Cigarette taxes in the
24%). In Belgium, the Netherlands, cigarettes that they had purchased US vary at the different levels of
Italy, Portugal, Luxembourg and in the last year were illicit. Half of government. Saba, Beard, Ekelund
Denmark, on the other hand, only 5% these respondents cited a missing and Ressler (1995) used data from
of respondents had seen potentially tax stamp, and/or a missing Albanian the 48 continental US states and
smuggled tobacco products in the health warning and/or not having the District of Columbia from 1960
past six months. In Norway, where in nicotine/tar information in Albanian to 1986 to estimate that small-
January 2008 a packet of Marlboro as features of an illicit cigarette pack. scale smuggling accounts for a
cost $12 (the country with the highest Another 29% cited tax stamps and small portion of market, usually not
cigarette prices in the world), only health warnings written in a foreign exceeding 2% of legal sales despite
6% of survey respondents had seen language, and 27% cited the taste the price/tax differences.
tobacco products during the last 12 of cigarettes. It is important to note In a more recent study using data
months which they believed were that is not illegal to purchase illicit from the Behavioural Risk Factor
smuggled. In Lithuania, where in cigarettes in Albania, and smokers Surveillance System (BRFSS), Stehr
January 2008 a packet of Marlboro would not be subject to legal (2005) found that the tax avoidance
cost $2 (the country with the lowest sanction by providing this information represented up to 9.6% of sales
cigarette prices in the EU) the (Zaloshnja et al., 2010). in the US between 1985 and 2001.
percentage was 36% (European Russian Federation. The However, according to Stehr’s
Commission, 2009). European Regional office of the findings the level of legal border
United Kingdom. According to World Health Organization estimates crossing in the USA was low relative
United Kingdom customs officials the that in the Russian Federation 20– to other forms of tax avoidance.
illicit market share (of consumption) 30% of cigarettes are smuggled, Combined federal, state and city
in 2006–07 was 13% for cigarettes and concludes that the Russian taxes are highest in New York City.
and 53% for hand-rolled tobacco Federation remains the biggest illicit In 2004, 57% of smokers in New
in the United Kingdom. The United European market in terms of volume York state purchased cigarettes at
Kingdom is one of the few countries (World Health Organization, 2007). least once from a low-tax or untaxed
to produce reliable yearly estimates Independent research estimated source, while 37% purchased low-tax
of illicit trade, with a methodology that 23% of legal sales were illicit in or untaxed cigarettes regularly (New
based on the discrepancy between the Russian Federation in 2004 (70 York State Department of Health,
trends in legal sales and household billion cigarettes) (Ross et al., 2008;; 2006). The large tax differentials
survey smoking habits (HM Revenue Joossen et al., 2009). between Chicago and neighbouring
& Customs, 2008). jurisdictions provide an incentive for
Finland. The Finish authorities Americas cigarette tax avoidance. Data from
reported that legal cross-border a random sample of cigarette packs
cigarette shopping by Finish travellers Canada. The Canadian tobacco littered in Chicago in 2007 reveals a
amounted to 12% of total national industry contracted a research high degree of tax avoidance: three
sales in 1996 (Lipponen et al., 1998). company, GfK Group, to assess fourths did not display a Chicago tax
Poland. A 2004 survey of the smoking trends in Canada. Their stamp (Merriman, 2010). Based on a
Cancer Epidemiology & Prevention research reported that 16.5% of comparison between cigarette sales
Division of the city of Warsaw also smokers said in 2006 that they had data and cigarette consumption data
suggested that only 11% of smokers purchased illicit tobacco products from surveys, a researcher from the
could have bought cigarettes on the within the previous seven days, the Department of Economics of Drexel
illicit market in Poland (Gumkowsky figure rising to 22% in 2007. The major University estimated that in 1985 in
et al., 2006). Studies based on source of the Canadian illegal trade the USA, 7.2% of cigarettes were
six surveys in the period 2004–6 is cigarettes illicitly manufactured purchased without payment of state
concluded that 11% of cigarettes in aboriginal native reserves on the taxes and that this had risen to 12.7%
sold in Poland were illicit (Ciercierski, border between Canada and the USA, in 2001 (Stehr, 2005). A researcher
2007). which are smuggled into Canada from the Stanford University Institute
Albania. In 2009, approximately (mainly the provinces of Ontario and for Economic Policy Research
19% of smokers in urban areas Quebec) (Royal Canadian Mounted estimated that between 13% and 25%
and 27% in rural areas in Albania Police, 2008) (average 17.5%) of US consumers
313
purchased cigarettes in a lower-price Middle East and Africa as 80% of the cigarette market in
bordering state or Native American Guinea Bissau and the Libyan Arab
reservation over the period 1992– Islamic Republic of Iran. Based Jamahiriya is illicit. In Mali the illicit
2002 (Lovenheim, 2007). on a report of the Iranian tobacco market share is 40%;; estimates
companies and the Central are lower for other listed countries
Asia and Australasia Headquarters of the Fight against (United Nations Office of Drugs and
Smuggling, which is a department Crime, 2009).
China. There are varying and of the Presidential office, the illicit
contradictory estimates of the level cigarette market share in the Islamic Impact of tax avoidance/evasion
of illicit cigarette trade in China. Republic of Iran was 74% of the total on public health
China is by far the biggest producer market in 2001 (40 billion cigarettes)
in the world of counterfeit cigarettes, and 14% of the total market in 2007 The impact of tax avoidance/
which are destined for domestic (8.3 billion cigarettes) in the Islamic evasion on tobacco products on
and foreign markets. A 2005 Republic of Iran (Joossens et al., public health can be classified as
national survey conducted by the 2009). The extremely high level the impact on average tobacco
China National Bureau of Statistics of smuggling in 2001 is probably price, on health disparity, on other
on behalf of the China National because there was insufficient tobacco control policies, and
Tobacco Company (CNTC) found domestic production to meet demand, generally on public safety.
that about 10% of cigarettes on the and imported cigarettes were subject The impact of higher taxes/prices
market were counterfeit (Joossens to high import duties, so the Islamic on tobacco use is not diminished by
et al., 2009). China’s State Tobacco Republic of Iran was a target for smuggling if the market equilibrium
Monopoly Administration announced internationally smuggled cigarette quantity of the taxed products in the
in January 2008 that it had seized brands. However, between 2001 and absence of tax avoidance/evasion is
9.28 billion counterfeit cigarettes in 2007 the market was liberalised and equal to the quantity of their supply
2007 (Globalink news service, 2008). national production was increased, in the presence of tax avoidance/
Thus the production of counterfeit leading to a dramatic fall in smuggling. evasion. This equilibrium depends
cigarettes can be estimated at 93– Studies on illicit cigarettes undertaken on the degree of substitution
186 billion cigarettes if we assume by the Tobacco Prevention and between legal and illegal cigarettes
that the seized cigarettes represent Control Research Center and relying and whether the availability of
about 5–10% of total illicit counterfeit on smoker self-reports, found that illegal substitutes increases total
production, a plausible assumption. 22.5% of the cigarettes consumed in consumption. To the extent that
Japan. Most observers in the f ield Teheran in 2006 were illegal (Heydari legal and illegal tobacco products
agree that illicit cigarette trade is low et al., 2009). A similar study was are not perfect substitutes, an
in Japan. A possible explanation for undertaken during a randomized increase in cigarette taxes translated
the low level of smuggling is the strict cross-sectional survey in Tehran in to increased cigarette prices will
control of the distribution network in 2008–2009, finding that 21% of the reduce their consumption even when
this country. All retailers of tobacco cigarettes were smuggled (Heydari smuggling is possible (Merriman et
products have to be approved, et al., 2010). al., 2000;; Merriman, 2002).
and are licensed by the Ministry of South Africa. An independent The degree of substation
Finance. researcher estimates the size of the between legal and illegal products is
Australia. The large difference illicit market in South Africa to have based on the notion that consumer
between the price received by grown substantially from 1997 until incurs both “transaction” and
tobacco farmers in Australia, and peaking in 2000 between 9.4% and “inconvenience” costs (Merriman,
the price of raw tobacco purchased 11.5% of the total market. The most 2002). The transaction price/cost
on the market has motivated tobacco recent estimate for 2007 suggests is the amount of money paid for the
growers to sell their products to illegal that the illicit market occupied product at the point of sale. The
markets to increase their profit from between 7.0% and 11.2% of the total inconvenience price/cost is the
tobacco growing (Geis et al., 2003). market (Blecher, 2010). value of the additional time it takes
North and West Africa. According to obtain a product illegally as well
to a report of the United Nations as the value of discomfort to engage
Office on Drugs and Crime, as much in an illegal activity. Although the
314
price of inconvenience does not break the law, and only some of them manufactured in the United Kingdom,
require a monetary transaction, it are willing to use illegal products. was higher than the locally produced
represents real costs since black Therefore, the observed reduction 555, because the smuggled
market purchasers may face in the legal consumption after a tax cigarettes were perceived as being
potential legal sanctions and other increase is only partially driven by of higher quality (Joossens, 2003).
risks. For example, the location of the substitution towards contraband. Several studies used the gap
street sellers can be undependable, Even though the theory suggests between the legal and illegal
there can be uncertainty about that the impact of illicit cigarette trade products and calculated the impact
the authenticity of the brand, or on cigarette prices is likely to be of eliminating this price difference
consumers may fear embarrassment small, the competition between the on tobacco use and on public health.
or legal penalties if caught buying legal and the illicit cigarettes could Joossens and colleagues (2009)
smuggled cigarettes. The sum of the result in lower average cigarette used data from 84 countries and
transaction price and inconvenience prices and therefore in higher estimated that the global average
price represent the “effective” price, consumption (Joossens et al., 2000). cigarette price in 2007 was about
the price that consumers consider The empirical evidence on prices 3.75% lower due to the presence of
when deciding whether to make of illicit tobacco products is mixed, illegal trade in cigarettes. If the global
a purchase or not. In general, the as the price differences between illicit market were eliminated in 2002,
higher the effective price, the lower legal and illegal cigarettes vary by 164 000 premature deaths would be
the quantity of cigarettes demanded. country, the location of the selling averted a year from the year 2030 on.
Illicit products often have a lower point, and the brand name and That means that by 2036, close to a
transaction (or sales) price than legal the perceived quality of the illicit million tobacco-related premature
products, because consumers who cigarettes. In general, the price of deaths would not occur.
purchase smuggled cigarettes pay the illicit cigarettes is lower than A similar study has been
higher inconvenience price. Given the retail price of legal products. In conducted in the United Kingdom,
that the general theoretical model of the United Kingdom, for example, where in the presence of smuggling
smuggling is applicable to a broad smuggled cigarettes in 2005 were the average tobacco product price
range of societies, it is reasonable sold at 50% of the duty-paid products was found to be about 11.6%
to expect that conclusions based on (West et al., 2008). In Germany, the under the legal market equilibrium.
empirical evidence from high-income price of the smuggled Chinese brand Eliminating the illicit cigarette trade
countries will be applicable to certain Jin Ling was 40% of the retail price of in the United Kingdom would reduce
degree to middle- and low-income a premium brand in 2009 (Candea et the cigarette consumption by 5.0–
countries (Merriman et al., 2000). al., 2009). The price of illegal “chop- 8.2% and lower the tobacco death
Therefore, in theory, smuggling chop” tobacco in Australia ranged toll in the United Kingdom by 4000–
does not completely reduce the from AUS$45 to AUS$100 per 6500 premature deaths a year (West
public health benefits of cigarette kilogram in 2001, while a kilogram et al., 2008)
taxes (Merriman, 2002). of legal roll-your-own tobacco was Experience from several
There is some empirical evidence AUS$320 in the same year (Geis et European countries suggests that
that consumers are aware of the al., 2003). an increase in cigarette taxes can
inconvenience price. A study in the Internet sites often sell cigarettes result in increased smuggling, but
United Kingdom found that 17% of tax-free or with taxes from low-tax also in the decline in total cigarette
adult smokers prefer to buy cigarettes jurisdictions (Ribisl et al., 2006). consumption.
from recognized outlets rather than The average online site selling Two sizeable tobacco tax
individuals even if the transaction cigarettes in the USA passes about increases in Sweden (December
price of the cigarettes sold by the 90% of the tax savings through to 1996 and August 1997) led to a 43%
individuals is £1.00 per pack lower the consumer (Goolsbee et al., 2007) increase in average cigarette prices,
(DTZ Pieda Consulting, 2000). making cigarettes available at prices but also an increase in the estimated
Using United Kingdom data, Duffy substantially lower than in stores. amount of cigarette smuggling (from
(2006) also suggests that smuggled In some markets, however, 200 million cigarettes in 1996 to 500
and non-smuggled tobacco products the price of illicit cigarettes can be million in 1998). However, smoking
are imperfect substitutes. The higher. In Viet Nam, for example, the prevalence has also declined (from
majority of smokers are not inclined to price of the smuggled brand 555, 1996 to 1997 there were 19.1%
315
and 4.4% declines among men and an increase in the consumption of free Canada, & Quebec Coalition
women, respectively), particularly products evading taxes. for Tobacco Control, 1999). As in
among youth and young adults (in Merriman et al. (2000) used data Sweden, the Canadian government
the age group 16–24 there were from 23 European countries during responded to the political pressure
25% and 17.4% declines among í DQG SUHGLFWHG WKDW D WD[ to reduce cigarette smuggling and
men and women from 1996 to 1997). increase in an individual country in 1994 reduced cigarette taxes.
In addition, tobacco tax revenue will increase small-scale cigarette This led to a sharp increase in per-
rose by 9% in 1997 compared to smuggling, but coordinating these capita consumption (a 27% increase
1996 (Wendleby & Nordgren, 1998;; increases with neighbouring countries between 1993 and 1998) and higher
Joossens, 1999). would reduce the incentives for this smoking prevalence among youth
After the Swedish government type of tax avoidance. For example, and adults as well as loss of tax
responded to pressure to reduce a unilateral 10% price increase revenue (Canadian Cancer Society,
cigarette smuggling by reducing in Germany would reduce annual Non-smokers’ Rights Association,
cigarette tax in August 1998, per- sales by 6 packs per capita, but total Physicians for a Smoke-free Canada,
capita tax paid cigarette sales consumption only by 3 cigarette & Quebec Coalition for Tobacco
increased, but tax revenue went packs per capita per year, with 3 Control, 1999).
down (Joossens et al., 2000). packs per capita being smuggled Several empirical studies used
France nearly doubled its nominal to Germany from other countries. Canadian data to study the response
cigarette retail price between With a multilateral price increase, to these cigarette tax changes.
September 1991 and December the consumption in Germany would Galbraith and Kaiserman (1997) used
1996 (74% increase in real terms) by still drop by 3 packs per capita, but monthly sales data from 1980–1994
increasing tobacco taxes. During the domestic sales would drop by only and found that the total consumption
same time, cigarette sales dropped 4 packs (with one pack still being (taxed plus smuggled sales) was
from 97 billion cigarettes in 1991 supplied from other countries). This less responsive to price increases
to 83 billion in 1997, adult smoking means that the health impact of a (short-run elasticity of –0.40) than
prevalence decreased from 40% in tax increase will be independent of taxed consumption (short-run
1991 to 34% in 1997 (Baudier, 1997), the coordination of tax increases, elasticity of –1.01), meaning that 1%
and youth smoking prevalence (12– but the revenue impact will depend increase in price due to tax increased
18 years old) dropped from 30% in on this coordination. If incentives illicit cigarette trade by 0.5–0.6%.
1991 to 25% in 1997 (Arènes et al., for small-scale smuggling in Europe Gruber, Sen and Stabile (2002) and
1998). Tobacco tax revenue rose disappeared, legal cigarette sales Goel (2004) confirmed that taxed
from 32 billion FF in 1991 to 57 billion would increase by 3%. cigarette consumption in Canada is
FF in 1996 while illicit cigarettes kept The European experience more sensitive to price than is total
occupying a relatively unimportant with tobacco tax increases and cigarette consumption, meaning that
share of the market—around 2% their impact on tobacco use in the there is a substitution between the
(Baudier, 1997). The relatively low presence of illicit cigarette trade has illicit and legal cigarettes, but a tax
level of illicit cigarette trade in France been very similar to what happened increase is still an effective measure
has been explained by its efficiently on the American continent. for reducing total cigarette use.
controlled retail environment in which Canada’s significant increase A similar conclusion has been
all tobacco retailers must be licensed. in cigarette prices due to its tax reached in the USA. Baltagi and Levin
A United Kingdom study found policy in 1980s and early 1990s (1986) showed that price elasticities
that higher taxes increased prices resulted in 43% decline in per-capita of cigarette demand in the USA were
of both legal and illegal tobacco cigarette consumption from 1979 to lower after controlling for small-scale
products and led to an overall 1993 despite the presence of illicit smuggling. Licari and Meier (1997)
decline in tobacco consumption cigarettes on the market. Smoking used a pooled time series of cigarette
(Duffy, 2006). However, the price prevalence fell sharply, particularly sales in all 50 US states from 1951
elasticity of duty-paid tobacco has among youth (15–19 years old): to 1994 and found that higher taxes
also increased since 1995 when from 43% in 1981 to 23% in 1991 reduce consumption, but their effect
cigarette smuggling began to grow, for this age group (Canadian Cancer is lower when controlling for cross-
meaning that some of the drop in Society, Non-smokers’ Rights state small-scale smuggling.
legal consumption was replaced by Association, Physicians for a Smoke-
316
A study from California using data in the sensitivity of taxable cigarette real GDP growth. This reduction
from the California Tobacco Surveys sales that is correlated with the rise would be even larger if enforcement
found that a significant cigarette tax of internet use within states. The of anti-smuggling laws would improve
increase in 1999 that resulted in growth of internet penetration in the at the same time.
relatively large price difference with USA has induced an elevation in The studies above estimate the
all its bordering states (including the taxable sales elasticity of over impact of tax avoidance and tax
Mexico) motivated only 5.1% of 60%. Cigarette tax evasion over the evasion on the general public. There
all smokers to purchase tax-free internet has substantially reduced are also studies that have examined
cigarettes. The study concluded that the revenue-generating potential the impact of this behaviour on health
higher cigarette taxes did not pose a of cigarette tax increases: states disparities, since the availability of
threat to the public health objective have collected 8% less cigarette tax lower-priced smuggled cigarettes
of reducing smoking despite the revenue between 2001 and 2005 could have a disproportionate impact
presence of tax avoidance/evasion due to tax-free cigarette internet on smoking and health among
(Emery et al., 2002). sales. This has serious implications children and the poor given their
One US study using data from for funding of state programmes, greater price sensitivity (Townsend
four waves of the CPS Tobacco including tobacco control and et al., 1994;; DTZ Pieda Consulting,
Supplement 1992–2002 found that other public health programmes. In 2000;; Joossens et al., 2000). In
the possibility to obtain cheaper addition, unregulated web sites offer addition, illicit cigarettes are primarily
cigarettes from another state or Native various price promotions or gifts products of the multinational tobacco
American reservation could reduce with purchases, and many require companies, because these are
the average price responsiveness a minimum purchasing quantity easy marketable and have price
of consumers to zero, but the price thus further encouraging cigarette advantage over less-known brands
sensitivity also varied with the consumption (Ribisl et al., 2001). Of (Joossens & Raw, 1998). Marlboro,
distance of residence to a lower-price 88 internet sites selling cigarettes in for example, represented 66% of all
border: a 1% increase in distance the USA in 2000, one third featured seized cigarettes worldwide in 2005
resulted in a decrease in the home promotional programmes (Ribisl et (World Customs Organization, 2007).
state price elasticity by about 0.2 al., 2001). International brands, and particularly
percent (Lovenheim 2007).The study The evidence on the impact Marlboro, are favoured by young
found that cross-border purchases of higher prices/taxes on the people in low-income and middle-
increase consumption by 4.0–8.2% substitution between legal and illegal income countries where Western
and the smoking participation by cigarettes outside Europe and North products are especially attractive
2.0–4.3%. The study concluded that America is limited. (Joossens & Raw, 1998).
cigarette price increases would be Yurekli and Sayginsoy (2010) This evidence of the impact of
effective in reducing cigarette use if used the market shares of illicit illicit trade on health disparities is
cigarette smuggling was eliminated. cigarettes in 110 countries to much more scarce compared to the
Making cheaper cigarettes estimate the impact of a tax increase evidence on the general population,
available via the internet can on tax avoidance/evasion. Assuming and most of it comes again from
undermine the public health gain a perfect substitution between North America and Europe.
from imposing higher tobacco legal and illegal cigarettes in their A study from Canada using data
taxes. Kim et al. (2006) found that econometric and simulation models, from 2006/2007 Youth Smoking
New Jersey adult smokers who they found that a global tax-induced Survey reported that over 13% of
purchased cheaper cigarettes via the increase in real cigarette prices daily teenage smokers reported
internet significantly increased their would lead to higher smuggling if it is usually consuming contraband
consumption over time, compared not accompanied by an improvement cigarette brands (Callaghan et al.,
to smokers who reported paying full in law enforcement. Despite the tax 2009). These smokers consumed
price at retail stores. avoidance, the overall cigarette use significantly more cigarettes
Goolsbee, Lovenheim and would be reduced: a 10% increase compared to those who smoke
Slemrod (2007) used US sales data, in total tax (excise and VAT) would legally circulating cigarette brands.
smoking prevalence and tax rates reduce total consumption by 1.97% Contraband cigarettes represented
from 1980 to 2005 to find that there with no change in Gross Domestic 17.5% of all youth daily-smokers
has been a considerable increase Product (GDP) or by 0.2% with 5% cigarette consumption in Canada,
317
but in some regions such as Ontario smoking and perceived smugglers as al., 2007). In addition, the presence
and Quebec, the share of illegal “providing a valuable service” to the of smuggled cigarettes can result
cigarettes were much higher—over community (Wiltshire et al., 2001). in a competitive disadvantage for
25% of the youth cigarette market The availability of cheaper illicit legitimate retailers, increasing
consisted of illicit cigarettes. cigarettes undermined the desire of their motivation not to comply with
Another study analysing cigarette many smokers to quit, thus reducing tobacco-control laws (Joossens et
butts collected in 2007 at smoking the impact tobacco tax policy can al., 2000).
locations near high schools in two have on consumption (Wiltshire et There is some evidence that illicit
Canadian provinces found that 26% al., 2001). cigarette trade undermines efforts
and 36% of them were contraband in One study from Asia, in Taiwan, to limit youth access to tobacco
Ontario and in Quebec, respectively. China, indicated that low-income products, since vendors of smuggled
The comparison with 2006 results and poorly-educated smokers are cigarettes are less likely to comply
when the share of illicit butts was more likely to purchase smuggled with these restrictions (Joossens
24% and 35%, respectively, suggests cigarettes (Lee et al., 2009). Smokers et al., 2000). Data from the USA
an increasing trend in illicit cigarette who had a personal monthly income suggest that retailers selling tax-free
use among the teenage population of less than NT$10 000 in 2004 cigarettes via internet, phone or by
(Canadian Convenience Stores (US$287) and had the least amount mail often fail to check the age of the
Association, 2007;; Unknown, 2008). of education were 54% more likely buyer. Of the 88 internet sites selling
Gruber, Sen and Stabile (2002) to smoke smuggled cigarettes than cigarettes in the USA in 2000, 18.2%
used data from the Canadian those with just one or none of these did not feature a minimum age of
Survey of Family Expenditure characteristics. sale warning (Ribisl et al., 2001). The
1982 –1998 and concluded that Using data from 84 countries, issue of youth access to cigarettes via
cigarette smuggling disproportionally Joossens et al. (2009) found that Internet may become more important
affects low-income groups: illicit the burden of illicit trade falls with better retailers’ compliance
trade reduces the average price of disproportionately on lower-income with youth access laws, as it may
tobacco products, leading to the countries, where the illicit cigarettes drive youth to use the internet
disproportionally higher consumption in 2007 made up on average 16.8% more frequently as their source of
of tobacco among the poor, who are of the market;; The average market cigarettes (Ribisl et al., 2006).
more price sensitive compared to share of illicit cigarettes in high- A US study found that only 28.4%
the general population. Therefore, income countries was about 9.8% of internet sites selling cigarettes
cigarette smuggling increases in that year. The tax loss associated featured the legally required health
smoking-related disparities. with illicit cigarette trade was warning (Ribisl et al., 2001). Only
Cantreill et al. (2008) interviewed estimated at US$18.3 and US$13 3.5% (5 sites) of internet cigar
614 Chinese-American smokers billion in low- and high-income vendors had health warning on their
living in New York City between countries, respectively. If the problem web sites (Malone & Bero, 2000).
September 2002 and February 2003 of illicit cigarettes were solved, low- Internet sites selling cigarettes often
and found that younger smokers and middle-income countries would failed to comply with advertising
were most likely to engage in tax experience 132 000 fewer premature ban and, instead promote specific
avoidance behaviour and were also tobacco-related deaths per year from tobacco brands and use of other
less likely to change their smoking 2030 on. In high-income countries, tobacco products such as cigars
behaviour in response to the tax the toll attributable to illicit cigarette and loose tobacco, and feature links
increase. Close to 55% of male trade is lower, about 32 000 per year. to pro-smoking or smoker’s rights
Chinese smokers engaged in at least In addition to the impact on the organizations (Ribisl et al., 2001).
one low- or no-tax strategy after a average tobacco price, the presence Illicit tobacco trade can affect
substantial increase of tobacco tax of illegal cigarettes on the market public safety by attracting organized
in 2002. can interfere with other tobacco crime and by increasing the general
Smokers in socioeconomically control regulations, such as those level of corruption (FIA International
deprived areas of Edinburgh in the related to youth access to tobacco Research Ltd, 1999a). In New York
United Kingdom admitted buying and requirements of tobacco product City, cigarette smuggling has been
contraband products as a way to contents and labelling of products associated with organized crime as
deal with the rising financial costs of (Stephens et al., 2005;; Pappas et well as activities of small-time crooks,
318
and has led to murders, kidnappings governments and international neighbouring or overseas jurisdictions.
and armed robberies to earn and organizations. Key informants Inter-agency cooperation (both
protect illicit profits. Such crime has were chosen for their overall domestic and international) emerges
exposed average citizens, such as knowledge of tobacco smuggling as a vital component of all successful
truck drivers and retail store clerks, or their intimate knowledge of a anti-contraband strategies. The
to violence (Fleenor, 2003). specific anti-contraband policy dynamic nature of contraband supply
Money gained from illicit tobacco measure. requires a comprehensive approach
trade is used for other serious criminal 3. Four expert focus panels that focuses on both immediate and
enterprises, including terrorist were convened to validate, future threats. Policies designed
operations (United States General enrich understanding of, and to ensure that contraband tobacco
Accounting Office, 2003). In 2002, further assess the feasibility of products do not appear in the
a US federal court found Mohamad implementing the various policy legitimate retail sector (such as
Hammoud guilty of providing material measures. tax-paid markings, licensing and
support for terrorists in his role as In its conclusions, the report record-keeping) and measures to
leader of a Charlotte, North Carolina– identifies and defines the different ensure that counterfeit products are
based cell that raised money for the forms of contraband tobacco, easily identified (such as enhanced
Lebanese terrorist group Hezbollah including casual small-scale taxation stamps) are vital resources.
by smuggling cigarettes within smuggling, organized international Adequate investment in enforcement
the United States (Fleenor, 2003). smuggling, illicit manufacturing, tax is also essential to the success of
Thus, illicit tobacco trade can have avoidance from duty-free sources, anti-contraband measures. Given
implications for the overall welfare of and counterfeit cigarettes. The the global scope of the phenomenon,
the country (Bhagwati and Hansen, effectiveness of ten anti-contraband greater international cooperation and
1973;; Ray, 1978) including the overall policy measures are explored: information sharing is paramount.
public health. licensing, tax-markings/stamping, A central theme in the research
tracking and tracing, record-keeping/ findings is the multifaceted nature
Policies to curb tax avoidance/ control measures, enhanced of successful anti-contraband
evasion;; new policies/ enforcement, export taxation, tax tobacco policies, which require
technologies harmonization, tax agreements/ combinations of regulation, fiscal/
compacts, legally binding taxation policy, enforcement, and
Sweeting, Johnson and Schwartz agreements with the tobacco industry, public awareness campaigns.
made an extensive review of the memoranda of understanding and The literature on the effectiveness
effectiveness of policy measures to public awareness campaigns. of anti-contraband policies is rather
combat cigarette contraband in 2009. According to the authors, the limited, and even less evidence is
While focusing on the Canadian analysis suggests that both the available for measures to reduce
situation, the report provides updated type of contraband and means tax avoidance. The methodology
information on anti-contraband of distribution influence the of the illicit trade estimates is often
policies in different parts of the world. effectiveness of different policies poorly described, and the research
Case studies in Brazil, Canada, and the unintended consequences of is underfunded for an area which is
Australia, United Kingdom, Spain and action. For example, policy measures described as difficult. For evident
the EU are discussed in detail in the that were effective in the 1990s for reasons, smugglers do not keep
report (Sweeting et al., 2009). legally manufactured cigarettes records and are not interested
Their research methods consisted smuggled across borders are less in research on their activities.
of three elements: effective for the illicitly manufactured Enforcement agencies are often
1. A systematic literature review and counterfeit cigarettes that reluctant to make their findings public
was undertaken, collecting both dominate contraband activity today for confidentiality reasons. One of the
academic and non-academic in many countries. major recommendations of Sweeting
publications on contraband Case studies indicate that and colleagues (2009) is to make
tobacco and relevant policies. while contraband sources often statistics and information regarding
2. Interviews were conducted with emerge domestically, given the the tobacco trade and contraband
representatives from academia, ease of transport and manufacture, tobacco much more available to the
nongovernmental organizations, sources can be easily displaced to public, to assist with research and
319
debate on this subject. For example, Under the new system, retailers Case study: Washington State
many jurisdictions (including Canada) and distributors can easily detect
do not provide official estimates counterfeit cigarettes by using Cigarette tax evasion from Internet
of the size of the illicit market or specific hand-held scanners. Law sales motivated the state of
comprehensive data on contraband enforcement field inspectors are Washington to apply the Jenkins Act
tobacco seizures made by federal equipped with more sophisticated (15 USC 376a), a US federal statute
and provincial agencies, making it scanners which give them access to originally intended to curb tax evasion
extremely difficult to evaluate and a whole range of data for compliance resulting from the interstate sale of
assess approaches in this area verification. The legislation mail-order cigarettes (Chaloupka
(Sweeting et al., 2009). imposes fines of up to US $25 000 et al., in press). The Act requires
There are many forms of illicit for possessing, selling or buying vendors who market or ship cigarettes
trade, and illicit activities may change counterfeit cigarettes or fraudulent across state lines to register with
over time as illicit traders adapt their cigarette tax stamps. the tobacco tax administrator of the
business as a result of the measures The results of the Californian destination states, and to send that
taken by governments. Some system have been evaluated office monthly statements or copies
examples of the changing nature of positively. The costs have been of invoices documenting the names
the illicit trade of tobacco products calculated to be US$9 million per year and addresses of recipients and
and the effectiveness of the anti-illicit in return for significant additional tax the quantities of cigarettes shipped.
trade policies are described below revenue on cigarettes—an additional This information should aid states
in the case studies of California, US$75 million was collected between with collecting excise taxes from
Washington state, Brazil and Europe. January 2004 and March 2006 as a recipients of cross-state shipments.
result of the licensing act and the tax However, the effort to apply the
Case study: California stamps. Investigators have tracked Jenkins Act met resistance from
retailers’ tax compliance since the internet cigarette vendors and
The Californian Board of Equalization provisions of the Act commenced received little support from the US
estimated in 2001–2002 that 25% operation. Their reports suggest federal government. However, in
of the state’s retailers were selling that after implementation, the 2002 the state of Washington filed a
counterfeit cigarettes (California seizures of counterfeit products at complaint against an online vendor
State Auditor, 2006). To reduce the retail locations and the percentage who failed to provide information
losses in revenue, the authorities of retailers carrying counterfeit on Internet sales to Washington
introduced licensing obligations, high- products decreased (California State residents. A federal court decided
tech tax stamps and investigative Auditor, 2006). that the state has the right to enforce
authority to better control the In combating illicit trade, no one the Jenkins Act to crack down on
distribution chain. In January 2004, measure is likely to be effective when cigarette tax avoidance, and the
the Cigarette and Tobacco Products implemented in isolation. Tax stamps online vendor agreed to provide the
Licensing Act was introduced, which and coded information should be information required by the Jenkins
required all entities engaged in the implemented in combination with Act (Banthin, 2004). Washington’s
sale of tobacco products within the other measures such as licensing. success in applying the Jenkins
state to be licensed. In January California had a problem with the Act led to similar efforts in many
2005, the attachment of tax stamps retail selling of counterfeit cigarettes;; other US states. Comprehensive
to tobacco products containing coded the solution was easy detection of information on the extent of these
information that was more difficult to the counterfeit cigarettes combined efforts, their costs, and the revenues
counterfeit than the stamps previously with the introduction of high-tech generated from them is not publicly
used was required. The stamping tax stamps, better control over the available, but newspaper articles
machines applied a new generation distribution chain by the enforcement suggest that these efforts are cost-
of tax stamps using invisible ink, officers and the withdrawal of the effective. For example, Michigan
featuring a unique covert code licence for those retailers who sell reported collecting US$5.9 million
containing product data relating to illicit cigarettes. from about 9000 residents based on
each cigarette pack and uploaded to lists provided by 13 online vendors
a central Data Management System. (Christoff 2006).
320
Case study: Brazil manufacturers on the cigarette and enforcement) that led to positive
packs (Fisch, 2006, 2009). results.
The illicit tobacco trade has been a The high-tech tax stamps contain
concern for the Brazilian authorities a unique code using invisible ink Case Study: United Kingdom
since the mid-1990s. In 1998, for each cigarette pack. The codes
Brazilian manufacturers were on the tax stamps include product In the United Kingdom, the main
exporting 34 billion cigarettes to data relating to each cigarette problem was genuine cigarettes
neighbouring countries, many of pack, which is uploaded to a Data produced in the United Kingdom that
which returned illegally to Brazil Manager Server under the control of were exported in large numbers to
as contraband. To deal with this the Ministry of Finance. Besides the dubious export markets, then illegally
problem, the government imposed possibility of verifying whether the imported though smuggling networks
an export tax of 150% on cigarettes products are authentic or counterfeit, back into the United Kingdom
to neighbouring countries. After the the stamps are encrypted with the (Joossens and Raw, 2008).
export tax was introduced, exports following information, such as name Over the last decade, illicit
of cigarettes declined rapidly, but of the manufacturing site, the date cigarette trade fell from about 21%
cigarette smuggling continued, as the stamp was validated, and the tax to 13% in the United Kingdom (HM
newly established factories in a category of the stamp. Revenue & Customs, 2008). Anti-
neighbouring country were fuelling If a manufacturer uses tax stamps smuggling measures in the United
the contraband market. In addition to whose electronic codes are not Kingdom included scanners for
the smuggling problem, 14 domestic detected, not allocated to that specific container detection, prominent
cigarette companies were not paying manufacturer, or not in accordance fiscal marks on packs, increased
the taxes on cigarettes. According with the fiscal category of the pack, punishment for offenders, more
to the Brazilian Ministry of Finance, the Data Manager Server will issue customs officers, and parliamentary
the illicit cigarette trade represented an alert to the Secretariat of Federal hearings, which exposed tobacco
35% of the market in Brazil in 2006: Revenues to start an investigation. industry export practices. The United
20% smuggling from neighbouring Inspectors, retailers and Kingdom strategy to tackle illicit
countries and 15% illicit domestic distributors can easily detect trade was continuously updated
manufacturing (Fisch, 2006, 2009). counterfeit cigarettes by using and included a strong cooperation
To tackle illicit domestic specific hand-held scanners. Law between different agencies. The
manufacturing, the licensing enforcement field inspectors can approach includes improved
of manufacturers was made have access online to package intelligence, risk profiling, tasking and
mandatory. Noncompliance with related data available on the Data coordination to detect and disrupt the
the legislation or failure to pay taxes Manager Server by scanning the supply of illicit tobacco products (HM
could lead to the withdrawal of the code. Revenue & Customs, 2008).
manufacturer’s license and the According to the Brazilian Ministry
closure of the factory. In addition to of Finance, the implementation of Case study: Italy and Spain
the licensing obligations, a national the programme led to the closure
integrated control and monitoring of several companies that did not In the 1990s, cigarette smuggling
system for cigarette production comply with the licensing rules and was a significant problem in the EU.
was fully implemented in 2008. The to US$100 million less tax evasion at In 1996, US cigarette companies
system included the installation of the domestic market in 2008 (Fisch, were exporting billions of cigarettes
automatic production counters at 2009). to Europe, under the transit regime,
each production line and mandated The new policy was not intended many of which disappeared during
the implementation of a digital to have an impact on the smuggling transport and ended up in the illegal
tax stamp, with capabilities for of cigarettes from the neighbouring markets of Italy, Spain, Germany and
identifying each individual pack. The countries, and its impact on this other EU markets. However, over the
purpose of the new legislation was type of illicit trade is probably minor last decade illicit cigarette trade fell
to ensure that all due taxes were or non-existent. As in the case of from about 15% to 1–2% in Italy and
collected on cigarettes produced in California, it was only a combination Spain (Joosens & Raw, 2008).
Brazil and to verify the authenticity of policies (control and monitoring In Italy, following Italian and
of the tax stamps applied by the system, licensing of manufacturers European investigations which led
321
to legal action against the tobacco dramatic fall in US exports to Europe 1989–95 sales data suggests that
industry, and subsequently to a legally (Joossens and Raw, 2008). coordinated multilateral increases
binding agreement with Phillip Morris, The European examples have in cigarette taxes would result in
there was a dramatic fall in customs common factors. Smuggling was significantly less tax avoidance
seizures and corresponding rise in reduced by interrupting the supply and tax evasion than unilateral tax
legal sales. The supply of smuggled chain from the manufacturers to increases.
cigarettes into Spain was reduced by the illicit market, and the evidence Combating tax evasion remains
a combination of measures, including suggests that the supply chain is difficult, but empirical evidence from
intelligence, customs activity in to a great extent controlled by the around the world suggests that a
border areas, and international tobacco industry. International combination of measures such as
cooperation, both within Europe and cooperation was also crucial. more investment in enforcement and
with US authorities over the supply Enforceable measures to control dissuasive penalties, international
of seized US brands. The European the supply chain, and international cooperation, including information
Fraud Office (OLAF) investigation cooperative measures including sharing and cooperation in the
of the tobacco companies in the US information sharing and cooperation investigation and prosecution of
in 1998 and the Spanish and Italian in the investigation and prosecution offences, legislative measures to
customs activities and ensuing of offences are essential to deal with control the supply chain and the legal
lawsuit against American tobacco the cross border illicit cigarette trade. business can lead to positive results.
companies also appear to have had One study (Merriman et al., 2000)
a significant impact. Over the period on small-scale smuggling and cross-
covered by these actions there was a border shopping in Europe using
322
References
Anthony R (2004). A calibrated model of the Campaign for tobacco-free kids (2008). Illicit Coleman T (1998). Stuck in the middle.
psychology of deterrence. Bull Narc, 56:49-64. tobacco trade: illegal profits and public peril. Tobacco International, March 26-28.
Washington DC, Campaign for Tobacco-Free
Arènes J, Arwidson P, Baudier F et al. (1998). Kids. Collin J, Legresley E, MacKenzie R et al.
Baromètre Santé Jeunes 97/98. Paris, Comité (2004). Complicity in contraband: British
Français d’Education pour la Santé. Canadian Cancer Society, Non-smokers’ Rights American Tobacco and cigarette smuggling
Association, Physicians for a Smoke-free in Asia. Tob Control, 13 Suppl 2;;ii104–ii111.
Baltagi B, Levin D (1986). Estimating dynamic Canada, et al. (1999). Surveying the damage: doi:10.1136/tc.2004.009357 PMID:15564212
demand for cigarettes using panel data: the cut-rate tobacco products and public health in
effects of bootlegging, taxation and advertising the 1990s. Ottawa, Ontario, Canadian Cancer Council of the European Union (2005). Report
reconsidered. Rev Econ Stat, 68:148–155 Society;; Non-smokers’ Rights Association;; on cigarette smuggling in the European Union.
doi:10.2307/1924938. Physicians for a Smoke-free Canada;; Quebec Brussels, Council of the European Union.
Coalition for Tobacco Control.
Baltagi BH, Levin D (1992). Cigarette taxation: Cunningham R (1996). Taxation and smuggling
raising revenue and reducing consumption. Canadian Convenience Stores Association smoke and mirrors: the Canadian tobacco war.
Struct Change Econ Dyn, 3:321–335 (2007). Cigarette butt study. Available at: http:// Ottawa, International Development Research
doi:10.1016/0954-349X(92)90010-4. www conveniencestores ca. Centre.
Banthin C (2004). Cheap smokes: state and Candea S, Campbell D, Lavrov V et al. Made to DeCicca P, Kenkel D, Liu F (2010). Excise Tax
federal responses to tobacco tax evasion over be smuggled, Russian Contraband cigarettes Avoidance: the case state of cigarette taxes.
the internet. Health Matrix Clevel, 14:325–356. ‘flooding’ EU. Kyiv Post. 7 April 2009 NBER Working Paper Series. Working Paper
PMID:15503692 #15941. Cambridge, MA, National Bureau of
Cantreill J, Hung D, Fahs MC, Shelley D Economic Research.
Baudier F (1997). Le tabagisme en 1997: état (2008). Purchasing patterns and smoking
des lieux. La santé de l’homme, 331(14). behaviors after a large tobacco tax increase: Delipalla S (2009). Tobacco tax structure and
a study of Chinese Americans living in New smuggling. Finanzarchiv, 65:93–104.
Becker GS, Grossman M, Murphy KM (1994). York City. Public Health Rep, 123:135–146.
An empirical analysis of cigarette addiction. PMID:18457066 DTZ Pieda Consulting (2000). The black
Am Econ Rev, 84:396–418. market in tobacco products. London, Tobacco
Chaloupka FJ, Cook PJ, Peck RM et al. Manufacturer’s Association
Beelman M, Campbell D, Ronderos M et al. Enhancing compliance with tobacco control
(2000). Major tobacco multinational implicated policies. In: Bearman P, Neckerman K, Wright Duffy M (2006). Tobacco c onsumption and policy
in cigarette smuggling, tax evasion, documents L, eds., Social and Economic Consequences of in the United Kingdom. Appl Econ, 38:1235–
show. Washington DC, The Center for Public Tobacco Control Policy. New York: Columbia 1257 doi:10.1080/00036840500392599.
Integrity. University Press. (In press)
Emery S, White MM, Gilpin EA, Pierce JP
Bhagwati JN, Hansen B (1973). A theoretical Chen T. Smoke2U. Tobacco Sales Tax Take Off (2002). Was there significant tax evasion
analysis of smuggling. Q J Econ, 87:172–187 in Cyberspace. The Center for Public Integrity. after the 1999 50 cent per pack cigarette tax
doi:10.2307/1882182. 19 December 2008. http://www publicintegrity increase in California? Tob Control, 11:130–
org/investigations/tobacco/articles/entry/1085/. 134.doi:10.1136/tc.11.2.130 PMID:12035006
Blecher E (2010). A mountain on a molehill:
is the illicit trade in cigarettes undermining Chiou L, Muehlegger E (2008). Crossing the Euromonitor International (2008). London:
tobacco control policy in South Africa? Trends line: direct estimation of cross-border cigarette Euromonitor (updated 2008 Jun 5). Illicit
Organ Crim, 13:299–315 doi:10.1007/s12117- sales and the effect on tax revenue. Regulatory Trade. In: Global Report: Tobacco – World.
010-9092-y. Policy Program Working Paper RPP-2008-02. Available from Euromonitor International, w ww.
Cambridge, M A, Mossavar-Rahmani Center for euromonitor.com.
Buck D, Godfrey C, Richardson G (1994). Business and Government, John F. Kennedy
Should cross border shopping affect tax policy. School of Government, Harvard University. European Commission (1998). Fight against
Yartic Occasional Paper Series No.6. University fraud. COM(98)276 Final. Brussels, CEC.
of York, Centre for Health Economics. Christoff C (2006). State busts nearly 9,000
who dodged cigarette tax. Detroit Free Press, European Commission (2009). Flash
California State Auditor, Bureau of State February 28. Eurobameter number 253, Survey on tobacco.
Audits (2006). Board of Equalization: its Analytical Report. Brussels, European
implementation of the cigarette and tobacco Ciercierski C (2007). The market for legal and Commission.
products licensing act of 2003 has helped illegal cigarettes in Poland : a closer look at
stem the decline in cigarette tax revenues, but demand and supply-side characteristics. IDRC European Parliament (1997). Report on the
it should update its estimate of cigarette tax working paper series/ITEN working paper Community Transit System, Committee of
evasion. California State Auditor report No. series No.1. Chicago, IDRC. inquiry into the community transit system.
2005-034. Sacramento, CA, The California A4-0053/97. Strasbourg, France, European
State Auditor. Cnossen S (2005). Economics and Politics of Parliament.
Excise Taxation. In: Theory and Practice of
Callaghan RC, Veldhuizen S, Leatherdale S et Excise Taxation: Smoking, Drinking, Gambling,
al. (2009). Use of contraband cigarettes among Polluting and Driving. Oxford, UK, Oxford
adolescent daily smokers in Canada. CMAJ, University Press, 1-19.
181:384–386. PMID:19737829
323
Farrelly MC, Nimsch CT, James J (2003). Goel RK (2008). Cigarette smuggling: price Hu T, Mao Z (2002). Economic analysis of
State cigarette excise taxes: implications for vs.nonprice incentives. Appl Econ Lett, tobacco and options for tobacco control: China
revenue and tax evasion. Final report prepared 15:587–592 doi:10.1080/13504850600721981. case study. HNP Discussion Paper Series,
for Tobacco Technical Assistance Consortium, Economics of Tobacco Control Paper No. 3.
Emory University, Rollins School of Public Goolsbee A, Lovenheim M, Slemrod J Washington DC, The World Bank.
Health. North Carolina, RTI International, (2007). Playing with fire: cigarettes, taxes and
Health, Social and Economics Research, competition from the Internet. Working Paper Hyde J. It’s just a speck of color. Associated
Research Triangle Park. No. 2007-5. Ann Arbor, Michigan, Office of Press. 5 May 1998
Tax Policy Research, Michigan Ross School of
FIA International Research Ltd (1999a). Business. Hyland A, Bauer JE, Li Q et al. (2005). Higher
Organized crime and the smuggling of cigarette prices influence cigarette purchase
cigarettes in the United States - The 1999 Gregg B, Cain C (1999). Cigarette tax revenues patterns. Tob Control, 14:86–92.doi:10.1136/
update. Washington, FIA International soars: jump linked to stamp on packs to fight tc.2004.008730 PMID:15791017
Research Ltd. smuggling. The Detroit News.
Hyland A, Laux FL, Higbee C et al. (2006).
FIA International Research Ltd (1999b). The Gruber J, Sen A, Stabile M (2002). Estimating Cigarette purchase patterns in four countries
Gray market in cigarettes in the United States: price elasticities when there is smuggling: and the relationship with cessation: findings
A primer. Washington, FIA International the sensitivity of smoking to price in Canada. from the International Tobacco Control (ITC)
Research Ltd. NBER Working Paper Series. Working Paper Four Country Survey. Tob Control, 15 Suppl
#8962. Cambridge, MA, National Bureau of 3;;iii 5 9 – iii 6 4.doi:10.113 6 / tc. 20 0 5.012 20 3
Fisch M (2006). The illegal cigarette Economic Research. PMID:16754948
market in Brazil. A case study. A non-paper
commissioned by the WHO TFI for the Gumkowski J, Prezwozniak K, Zatonski W IARC (2008). Measures to aseess the
technical briefing during the first session of (2006). Cigarette smuggling in Poland: tobacco effectiveness of tobacco taxation. In: Methods
cop of the WHO-FCTC, 6-17 February 2006, industry views and smokers behaviours. 13th for evaluating tobacco control policies. Lyon,
Geneva, Switzerland. World Conference on Tobacco or Health, International Agency for Research on Cancer,
Washington. http://2006.confex.com/uicc/ 189-213.
Fisch M (2009). Medidas a considerar para wctoh/techprogram/P8144.HTM
limitar el commercio ilicito de cigarillos. La Jensen R, Thursby JG, Thursby MC (1991).
experiência de Brasil [Possible measures to Hajdinjak M (2002). Smuggling in Southeast Smuggling, c amouflaging and market structure.
limit the illicit cigarette trade. The Brazilian Europe. The Yugoslav wars and the NBER Working Paper Series, Working Paper
experience]. Presentation at a technical development of regional criminal networks no.2630. Cambridge, MA, National Bureau of
meeting on illicit cigarette trade, Sâo Paulo, in the Balkans. Sofia, Center for the Study of Economic Research.
May 2009. Democracy.
Joossens L (1998). Tobacco smuggling: an
Fleenor P (2003). Cigarette taxes, black Heydari G, Sharafi H, Masjedi M et al. (2009). optimal policy approach. In: Abedian A, van
markets, and crime. Lessons from New York’s What kind of cigarettes do smokers use in der Merwe R, Wilkins N, et al., eds., The
50-year losing battle. Policy Analysis Report Tehran in 2006? Tanaffos, 8:54–58. Economics of Tobacco Control. Cape Town,
N°.468. Washington DC, The Cato Institute. Applied Fiscal Research Center
Heydari G, Tafti SF, Telischi F et al. (2010).
Galbraith JW, Kaiserman M (1997). Taxation, Prevalence of smuggled and foreign cigarette Joossens L (1999). Smuggling and cross-
smuggling and demand for cigarettes in use in Tehran, 2009. Tob Control, 19:380–382. border shopping of tobacco products in
Canada: evidence from time-series data. J doi:10.1136/tc.2009.033191 PMID:20876076 the European Union. London, The Health
Health Econ, 16:287–301.doi:10.1016/S0167- Education Authority.
6296(96)00525-5 PMID:10169302 HM Customs & Excise (1998). Estimates of
cross-border shopping and smuggling of Joossens L (2003). Vietnam: smuggling adds
Geis G (2005). Chop-chop: illegal cigarette alcohol and tobacco. London, The Stationery value. Tob Control, 12:119–120.doi:10.1136/
market in Australia. Working Paper No.48. Office Limited, HM Treasury. tc.12.2.119 PMID:12773713
Canberra, Australia, Center for Tax System
Integrity, Research School of Social Sciences, HM Customs & Excise, (2000). Tackling Joossens L, Chaloupka F, Merriman D et al.
Australia National University. tobacco smuggling. London, The Stationery (2000). Issues in the smuggling of tobacco
Office Limited, HM Treasury. products. In: Jha P, Chaloupka F, eds., Tobacco
Geis G, Cartwright S, Houston J (2003). Public Control in Developing countries. London,
wealth, public health, and private stealth: HM Customs & Excise (2001a). Estimates Oxford University Press, 393-406.
Australia’s black market in cigarettes. Aust J of cross-border shopping and smuggling of
Soc Issues, 38:363–378. alcohol and tobacco. London, The Stationery Joossens L, Merriman D, Ross H et al. (2009).
Office Limited, HM Treasury. How eliminating the global illicit cigarette trade
Gilmore AB, McKee M (2004). Tobacco and would increase tax revenue and save lives.
transition: an overview of industry investments, HM Customs & Excise (2001b). Tackling indirect Paris, International Union against Tuberculosis
impact and influence in the former Soviet tax fraud. London, The Stationery Office and Lung Disease.
Union. Tob Control, 13:136–142.doi:10.1136/ Limited, HM Treasury.
tc.2002.002667 PMID:15175530 Joossens L, Merriman D, Ross H, Raw M
HM Revenue & Customs (2008). Departmental (2010). The impact of eliminating the global
Globalink news service. China seizes 9 billion autumn performance report. London, The illicit cigarette trade on health and revenue.
counterfeit cigarettes in 2007, China View. Stationery Office Limited, HM Revenue and Addiction, 105:1640–1649.doi:10.1111/j.1360-
http://tobacco cleartheair org hk/, Last Update: Customs. 0443.2010.03018.x PMID:20626371
January 2008.
House of Commons Treasury Committee Joossens L, Raw M (1995). Smuggling and
Goel RK (2004). Cigarette demand in Canada (2005). Excise duty fraud: fourth report of cross border shopping of tobacco in Europe.
and the US-Canadian cigarette smuggling. session 2004-05. London, The Stationery BMJ, 310:1393–1397. PMID:7787549
Appl Econ Lett, 11:537–540 doi:10.1080/1350 Office Limited, House of Commons, Treasury
485042000263043. Committee.
324
Joossens L, Raw M (1998). Cigarette smuggling Merriman D (2010). The Micro-Geography Saba RR, Beard R, Ekelund RB Jr,
in Europe: who really benefits? Tob Control, of Tax Avoidance: Evidence from Littered Ressler RW (1995). The demand for
7:66–71.doi:10.1136/tc.7.1.66 PMID:9706757 Cigarette Packs in Chicago. Am Econ J - Econ cigarette smuggling. Econ Inq, 33:189–202
Policy, 2:61–84 doi:10.1257/pol.2.2.61. doi:10.1111/j.1465-7295.1995.tb01856.x.
Joossens L, Raw M (2008). Progress in
combating cigarette smuggling: controlling Merriman D, Yurekli A, Chaloupka F (2000). Schneider J (2010). R.J. Reynolds Agrees
the supply chain. Tob Control, 17:399–404. How big is the worldwide cigarette smuggling to Pay Canada C$325 Million. Bloomberg
doi:10.1136/tc.2008.026567 PMID:18784154 problem? In: Prabhat J, Chaloupka F, eds., Businessweek.
Tobacco control in developing countries.
Kim AE, Ribisl KM, Delnevo CD, Hrywna M Oxford, UK, Oxford University Press, 1-464. Segal D (1999). Canada sues tobacco giant.
(2006). Smokers’ beliefs and attitudes about Washington Post, A7.
purchasing cigarettes on the Internet. Public Ministry of Health Republic of Indonesia (2004).
Health Rep, 121:594–602. PMID:16972513 The Tobacco Source Book: Data to support a Shen A, Antonopoulos G, Von Lampe K
National Tobacco Control Strategy. Republic of (2010). ‘The Dragon breathes smoke’, cigarette
KPMG (2005). Study on the collection and Indonesia, Ministry of Health. counterfeiting in the People’s Republic of
interpretation of data concerning the release China. Br J Criminol, 50:239–258 doi:10.1093/
for consumption of cigarettes and fine-cut Nakkash R, Lee K (2008). Smuggling as the bjc/azp069.
tobacco for rolling of cigarettes. Brussels, “key to a combined market”: British American
European Commission. Tobacco in Lebanon. Tob Control, 17:324–331. Simkin C (1974). Indonesia’s unrecorded trade.
doi:10.1136/tc.2008.025254 PMID:18818226 In: Bhagwati J, ed., Illegal transactions in
Lakhdar CB (2008). Quantitative and qualitative international trade. Amsterdam, North Holand
estimates of cross-border tobacco shopping Nelson MA (2002). Using excise taxes to Publishing Company, 151-171.
and tobacco smuggling in France. Tob finance state government: do neighbouring
Control, 17:12–16.doi:10.1136/tc.2007.020891 state taxation policy and cross-border Stehr M (2005). Cigarette tax avoidance
PMID:18218801 markets matter? J Reg Sci, 42:731–752 and evasion. J Health Econ, 24:277–297.
doi:10.1111/1467-9787.00279. d o i : 1 0 .1 0 1 6 / j . j h e a l e c o . 2 0 0 4 . 0 8 . 0 0 5
Lee JM, Chen SH, Chen HF, Jeng HY (2009). PMID:15721046
Price sensitivity and smoking smuggled New York State Department of Health (2006).
cigarettes. Eur J Public Health, 19:23–27. Cigarette purchasing patterns among New Stephens WE, Calder A, Newton J (2005).
doi:10.1093/eurpub/ckn115 PMID:19039020 York smokers: implications for health, price and Source and health implications of high toxic
revenue. New York, New York City Department metal concentrations in illicit tobacco products.
LeGresley E, Lee K, Muggli ME et al. (2008). of Health. Environ Sci Technol, 39:479–488.doi:10.1021/
British American Tobacco and the “insidious es049038s PMID:15707047
impact of illicit trade” in cigarettes across Pappas RS, Polzin GM, Watson CH, Ashley
Africa. Tob Control, 17:339–346.doi:10.1136/ DL (2007). Cadmium, lead, and thallium in Sugarman SD (2002). International aspects of
tc.2008.025999 PMID:18617598 smoke particulate from counterfeit cigarettes tobacco control and the proposed WHO treaty.
compared to authentic US brands. Food In: Rabin L, Sugarman D, eds., Regulating
Leverett M, Ashe M, Gerard S et al. (2002). Chem Toxicol, 45:202–209.doi:10.1016/j. tobacco. New York, Oxford University Press,
Tobacco use: the impact of prices. J Law Med fct.2006.08.001 PMID:17011104 245-284.
Ethics, 30 Suppl;;88–95. PMID:12508509
Pitt MM (1981). Smuggling and price disparity. Sweeting J, Johnson T, Schwartz R (2009).
Licari MJ, Meier KJ (1997). Regulatory policy J Int Econ, 11:447–458 doi:10.1016/0022- Anti-contraband policy measures: evidence
when behavior is addictive: smoking, cigarettes 1996(81)90026-X. for better practice - summary report. Special
taxes and bootlegging. Polit Res Q, 50:5–24. report series. Toronto, The Ontario Tobacco
Ramos A (2009). Illegal trade in tobacco in Research Unit.
Lipponen S, Hara M, Waller M et al. (1998). MERCOSUR countries. Trends Organ Crim,
Non-taxable imports of tobacco. Helsinki, 12:267–306. Taal A, Kiivet R, Hu TW (2004). The economics
Ministry of Social Affairs and Health. of tobacco in Estonia. HNP Discussion Paper
Ray A (1978). Smuggling, import objectives, Series, Economics of Tobacco Control Paper
Lovenheim M (2007). How far to the border? and optimum tax structure. Q J Econ, 92:509– No. 19. Washington DC, The World Bank.
The extent and impact of cross-border casual 514 doi:10.2307/1883156.
cigarette smuggling. Discussion Paper N° Taylor A, Chaloupka F, Guindon E et al.
06-40. Stanford, CA, Stanford Institute for Ribisl KM, Kim AE, Williams RS (2001). Web (2000). The impact of trade liberalization on
Economic Policy Research (SIEPR). sites selling cigarettes: how many are there in tobacco consumption. In: Jha P, Chaloupka
the USA and what are their sales practices? Tob F, eds., Tobacco control policies in developing
Malone RE, Bero LA (2000). Cigars, youth, Control, 10:352–359.doi:10.1136/tc.10.4.352 countries. London, Oxford University Press,
and the Internet link. Am J Public Health, PMID:11740027 343-364.
9 0:79 0 –79 2 .do i:10. 210 5 /A J PH . 9 0. 5 .79 0
PMID:10800432 Ribisl K M, Williams RS, Kim AE (2006). Internet The Financial ACtion Task Force (FATF) (2010).
cigarette sales. Briefing to the Congressional Money laundering vulnerabilities of Free Trade
Market Tracking International Ltd (1997). World task force on tobacco and health. Washington Zones. Paris, France, FATF/OECD.
Tobacco file 1996: emerging markets in Central DC, The Robert Wood Johnson Foundation.
and Eastern Europe. London, Argus Business. Thursby JG, Thursby MC (2000). Interstate
Ross HZ, Shariff S, Gilmore A (2008). cigarette bootlegging;; extent, revenue losses,
Merriman D (2001). Tool 7. Smuggling. Economics of Tobacco Taxation in Russia. and effects of federal intervention. Natl Tax J,
Understand, measure and combat tobacco Paris, International Union Against Tuberculosis 53:59–77.
smuggling. Economics of tobacco toolkit. and Lung Disease.
Washington DC, The World Bank. Tobacco Institute of South Africa. (2008).
Royal Canadian Mounted Police (2008). Presentation to Botswana Unified Revenue
Merriman D (2002). Cigarette smuggling Contraband tobacco enforcement strategy. Service.
does not reduce the public health benefits of Ottawa, Royal Canadian Mounted Police.
cigarette taxes. Appl Econ Lett, 9:493–496
doi:10.1080/13504850110095468.
325
Townsend J, Roderick P, Cooper J (1994). Vedder R (1997). Bordering on chaos: fiscal World Customs Organization (2009). Customs
Cigarette smoking by socioeconomic group, federalism and excise taxes. In: Shughart and Tobacco Report 2009. Brussels, World
sex, and age: effects of price, income, WF, ed., Taxing choice: the predatory politics Customs Organization.
and health publicity. BMJ, 309:923–927. of fiscal discrimination. New Brunswick,
PMID:7950662 London, The Independent Institue, Transaction World Health Organization (2003). The cigarette
Publishers «transit» road to the Islamic Republic of Iran
Tsai YW, Sung HY, Yang CL, Shih SF (2003). an Iraq: illicit tobacco trade in Middle East.
The behaviour of purchasing smuggled Von Lampe K (1999). The nicotine racket, Report no. WHO/EM/TFI/011/E/G. Cairo, WHO,
cigarettes in Taiwan. Tob Control, 12:28–33. trafficking in untaxed cigarettes: A case study Regional Office for the Eastern Mediterranean.
doi:10.1136/tc.12.1.28 PMID:12612358 of organized crime in Germany. Lecture at
the Institute of Criminology, University of World Health Organization (2005). WHO
United Nations Office on Drugs and Crime Oslo, Norway, May 6 1999. Available From: Framework Convention on Tobacco
(2009). Transnational trafficking and the rule Organized Crime Research. Control. Geneva, Switzerland, World Health
of law in West Africa: a threat assessment. Organization.
Vienna, United Nations Office on Drugs and Walsh S, Ottaway DB (2000). NY Ethnic
Crime. Groups sell close to home - Smugglers Find World Health Organization (2007). The
Loyal Clientele Among Neighbors. Washington European tobacco control report 2007.
United States General Accounting Office Post. Copenhagen, WHO, Regional Office for
(2003). Terrorist Financing. US Agencies Europe.
should systematically assess terrorists’ Wen CP, Peterson RA, Cheng TY et al. (2006).
use of alternative financing mechanisms. Paradoxical increase in cigarette smuggling Yurekli A, Sayginsoy O (2008). Worldwide
Report to congressional requesters. Report after the market opening in Taiwan. Tob Control, organized cigarette smuggling: an
no.GAO-04-163. Washington DC, United 15:16 0 –16 5 .d o i:10.113 6 / tc . 2 0 0 5 .0119 4 0 empirical analysis. Appl Econ, 42:545–561
States General Accounting Office. PMID:16728745 doi:10.1080/00036840701720721.
Unknown. Ex-Soviet mob grips. Pittsburgh Wendleby M, Nordgren P (1998). Balancing the Yurekli AA, Zhang P (2000). The impact of
Post-Gazette, A6 . 4 November 1994 price: Sweden. Presentation at tourist imports clean indoor-air laws and cigarette smuggling
and smuggling of cigarettes. Helsinki, Ministry on demand for cigarettes: an empirical
Unknown. Kids are paid to smuggle. Scottish of Social Affairs and Health. model. Health Econ, 9:159–170.doi:10.1002/
Daily Record. 16 October 1994 6 , & , $ , '
West R, Townsend J, Joossens L, Arnott HEC499>3.0.CO;;2-T PMID:10721017.
Unknown. Boys now hired to smuggle D, Lewis S (2008). Why combating tobacco
cigarettes. New Strait Times. 13 November smuggling is a priority. BMJ, 337:1028–1029. Zaloshnja E, Ross H, Levy DT (2010). The impact
1995 of tobacco control policies in Albania. Tob Control,
Wiltshire S, Bancroft A, Amos A, Parry O 19:4 6 3 – 4 6 8.doi:10.113 6/ tc. 20 0 9.03 4 6 52
Unknown. Wild art. Toronto Sun. 28 December (2001). “They’re doing people a service”- PMID:20679417.
1997 qualitative study of smoking, smuggling,
and social deprivation. BMJ, 323:203–207.
Unknown. Nemesis looms for smuggling trade. doi:10.1136/bmj.323.7306.203 PMID:11473911
Hong Kong Standard. 11 July 1999
World Customs Organization (2007). Customs
Unknown. Study suggests more teens smoking and Tobacco Report 2006. Brussels, World
contraband tobacco. Canadian Press. 24 Customs Organization.
September 2008
326
Chapter 9
Health and economic impact of tobacco taxation
Introduction
According to evidence from around revenue. However, empirical studies government revenue, followed by a
the world, raising the tax on tobacco (Chaloupka et al., 2000) have shown discussion of the pros and cons of
products is a highly effective that since the demand for cigarettes earmarked tobacco taxation. The
control policy instrument to improve is inelastic (that is, the percentage of finally section reviews the impact
population health and reduce reduction in cigarette consumption of tobacco taxation increases on
smoking-related health risks (Jha & is less than the percentage of inflation and the consumer price
Chaloupka, 1999). A tax increase increasing price), governments will index.
leads to an increase in cigarette in fact realize increased cigarette tax
prices, which in turn will cause some revenue as a result of an increase in Impact on population health
smokers to quit smoking or not to the tobacco tax. Some governments and healthcare cost savings
initiate or relapse to the smoking allocate part of the additional tax
habit, while others may continue revenue to tobacco control activities Direct impact of smoking
smoking but will smoke less. The act or tobacco-related medical care. on population health
of quitting smoking itself and less Fiscal experts may argue against
secondhand smoke (SHS) exposure such an earmarked tax. There are Smoking is a major cause of
will reduce premature deaths and other potential effects of tobacco premature death and morbidity.
improve the health of the population, taxation on the consumer price index. Smoking has been directly linked
thereby reducing healthcare costs. This chapter reviews and to cancer, cardiovascular disease
On the other hand, the cigarette summarizes the population health and respiratory disease, among
industry is concerned with the possible and economic impact of tobacco others (US. Department of Health
negative impact on employment taxation from past published literature and Human Services, 2000). A vast
resulting from a reduction in cigarette and empirical evidence. The overall literature exists on the relationship
consumption. Tobacco farming also conceptual framework for the impact between smoking and mortality (Peto
may be affected by a loss in earnings. analysis on this topic is displayed in et al., 1992;; Peto et al., 1994;; Peto,
In addition to smuggling and the Figure 9.1. 1994). These studies estimate about
regressive effect on low-income This review is organized as 4 million deaths per year worldwide
smokers, these are key concerns follows. The first section reviews attributable to smoking. Half of these
that many governments or tobacco the impact of tobacco taxation on deaths are in low- and middle-income
industries have used to argue against population health, healthcare cost countries. By 2030, the annual death
an increase in tobacco taxes. savings, and productivity gain. This toll could reach 10 million if no tobacco
Some officials also are is followed by a presentation of the control measures are taken. More
concerned that with an increase impact of tobacco taxation on industry recently, a country-specific study
in the tobacco tax, the demand for employment and tobacco farming. for China (Gu et al., 2009) estimated
cigarette consumption will decrease, The next section summarizes the that in 2005 a total of 673 000 deaths
and with it government cigarette tax impact of tobacco taxation on were attributable to smoking. The
327
Figure 9.1. The overall conceptual framework for the population health and economic impact of tobacco taxation

Adapted from Hu T-w, Mao Z, Shi J, Chen W. Tobacco Taxation and Its Potential Impact in China. Paris, France: The International Union Against
Tuberculosis and Lung Disease, 2008, page 26.
leading causes were cancer (268000 to smoking, suggesting a reduction by 5% (Chaloupka & Warner, 1999).
deaths), cardiovascular disease in economic productivity due to lost Empirical literature indicates that
(146 200 deaths) and respiratory work years. about one third to one half of the 5%
disease (66 800 deaths). Increasing tobacco taxes reduction in cigarette consumption
Studies have also been conducted can reduce smoking-attributable is the result of people quitting or not
for the USA (Burns et al., 1997), India mortality and morbidity. The impact initiating smoking (US Department of
(Gupta et al., 2005;; Jha et al., 2008) of tobacco taxation on the reduction Health and Human Services, 2000).
and Germany (Neubauer et al., 2006) of mortality depends on (1) the The literature suggests that
that have provided a quantitative magnitude of the price increase about one quarter to one half of
relationship between the length of resulting from a tobacco tax increase, those who quit smoking will avoid a
smoking, smoking cessation, and (2) the magnitude of the price smoking-related premature death
benefits of quitting smoking for elasticity (the negative relationship (Jha & Chaloupka, 1999;; Shafey
mortality. The risk of death declines between price and consumption), et al., 2009). The same analytical
with the length of time since an which includes smoking behaviour methods have been used in several
individual quits smoking, with the (quitting or initiating) in relation studies supported by the International
reductions in relative risk on heart to the price increase, and (3) the Union Against Tuberculosis and Lung
disease and stroke more immediate relationship between mortality and Disease (Barber et al., 2008;; Hu et
than the effects on respiratory quitting smoking. For example, if al., 2008;; John et al., 2009a;; Ross et
disease and cancer. Studies indicate the price elasticity of demand for al., 2008). In China, it was estimated
that a large portion of heart disease FLJDUHWWHV LV í D SULFH that an increase in the excise tax of 1
deaths of those aged 40–64 are due increase will reduce consumption RMB (US$0.15) per pack could save
328
one million lives if the price elasticity of these estimated lives saved are 2005) using time series data between
RI GHPDQG LV í PLOOLRQ OLYHV only for those people who are already 1965 and 1994 found that taxes on
could be saved if the price elasticity smoking before the tax increase. tobacco were inversely related to the
RIWKHGHPDQGIRUFLJDUHWWHVLVí These examples show the impressive male standardized suicide rate. They
(Hu et al., 2008). The India study population health ramifications of concluded that a policy designed to
indicates that with a price elasticity of raising the tobacco tax. reduce tobacco consumption has
GHPDQGIRUFLJDUHWWHVRIíD Several studies have empirically the benefit of reducing suicides,
price increase would save 2.7 million examined the direct relationship particularly for men.
smokers lives (John et al., 2009a). In taxes/price and deaths. A study Kaplan et al. (2001) generalized
the Russian Federation, with a price (Moore, 1996) used a national data the analysis of health impacts to
elasticity of the demand for cigarettes for the USA to examine the effects consider overall health benefits. They
DWíDSULFHLQFUHDVHZRXOG of tobacco excise tax on changes of simulated the impact of a US$0.50
mean that 2.7 million tobacco-related mortality. Controlling for state, year, per pack cigarette tax increase in
deaths could be averted (Ross et socioeconomic variables and alcohol California. Assuming a price elasticity
al., 2008). A study in another major tax, he found that variations in tobacco RIíLWZRXOGUHVXOWLQDERXW
cigarette smoking country, Indonesia taxes between 1954 and 1988 were quality-adjusted life years (QALYs)
(Barber et al., 2008) indicates that inversely related to smoking-related saved during the first year. Higher
ZLWK D SULFH HODVWLFLW\ RI í DQG mortality. He concluded that a 10% taxes would produce even greater
an increase in the tax rate from increase in the tobacco tax would health benefits in the long-term.
37% to 70% of the sales price, 2.5 save over 6000 lives a year. A study Table 9.1 summarizes the impact of
million deaths could be averted. All for Switzerland (Yamasaki et al., tobacco taxation on lives saved.
Table 9.1. Impact of tobacco taxation on lives saved
References Country Amount of tax Assumptions Lives saved
Asia
Levy et al. (2006) Viet Nam From 36% to 72% of price íWRíE\DJH 5921 lives per year after 20 years,
9490 lives per year after 30 years
Hu et al. (2008) China Additional 1 RMB (US$0.15 Price elasticity 1 million lives of current smokers
or from 40% to 51% of retail í 3.4 million lives of current
price) í smokers
Barber et al. (2008) Indonesia Raised from current level to í 2.5 million lives of current
70% of retail price smokers
John et al. (2009a) India Raised from current level to í 2.7 million lives of current
Europe
Levy et al. (2008b) Albania Increase in price from 125 lek íWRíE\DJH 462 lives saved per year after 20
to 250 lek years
Ross et al. (2008) Russian Federation Raised from current level to í 2.7 million lives of current
North America
Moore (1996) USA 10% tax increase N/A 6000 lives per year
Kaplan et al. (2001) USA (California) Raised US$ 0.50 per pack í 8389 QALYs first year
Levy et al. (2008c) USA (Kentucky) Tax increase from 0.30 to íWRíE\DJH 401 lives per year after 20 years
$2.00
South America
Ferrante et al. (2007) Argentina From 68% of price to 85% íWRíE\DJH 5283 lives per year after 20 years,
of price 7851 lives per year after 30 years
World
Ranson et al. (2000) Worldwide 10% tax increase 10 million lives of current smokers
329
Several studies have examined deaths avoided increases to over saved per year after 20 years and
the impact of tobacco taxation on 5223 in the year 2010 alone. Over 9490 lives would be saved per year
disease-specific mortality, such as the years 1989 to 2010, the model after 30 years. Despite a smaller
heart disease or stroke, based on a estimates that about 55 444 lives population and using lower price
pre-post tobacco control intervention were saved as a result of tobacco elasticities, Ferrante et al. (2007)
programme. Studies for California control policies in the state of show that increasing the tax as a
statistically have identified a clear California. Further, the model shows percentage of price from 68% to 85%
association between taxation and that about 60% of the deaths avoided in Argentina would yield 5283 lives
mortality. In 1989, California can be attributed to cigarette price saved per year after 20 years and
increased tobacco taxation and increases. For Arizona (Levy et al., 7851 lives saved per year. Based on
earmarked part of the revenue 2007b), which like California had studies of relative risk, a smaller total
for a tobacco control programme. mounted a major tobacco campaign, mortality relative risk was used for
Lightwood & Glantz (1997), using data the SimSmoke model shows that Viet Nam (1.35, based on studies for
collected between 1980 and 1995, by price increases lead to 60% of the China) than Argentina (2.1, based on
comparing the pre- and post-tobacco premature deaths avoided. studies for the USA).
control programme (1989–1995), The SimSmoke model has also The SimSmoke models show
showed via a regression model been applied to several nations. that the effects of a price increase
that the California tobacco control While the Arizona and California on mortality are small at first but
programme was able to reduce models applied price elasticities grow over time, because 1) the
deaths by about 13 000 per year by based on studies for the USA, reduction in the relative mortality
preventing cardiovascular and stroke the models for other nations use risks increases with the time since
deaths. Another study (Fichtenberg elasticity estimates based on studies quitting, 2) the effects on reduced
& Glantz, 2000) compared per-capita conducted for the respective nation. A initiation are delayed until about age
cigarette consumption and the death SimSmoke model for Thailand (Levy 40 when deaths due to tobacco begin
rate from heart disease for California et al., 2008a) estimates that between to appear, and 3) the effects of a price
to that of the other 49 US states. 1991 (shortly after tobacco control increase grow over time because its
Regression results indicated that policies began) and 2026 about 320 effects on initiation and cessation
between 1989 and 1992, California 000 premature deaths would be continue into the future, and because
had 33 000 fewer deaths from heart avoided due to tobacco control, of the larger effects of price on youth
disease than would have been which approximately 60% would be (see previous chapters on the higher
expected if California had had no again due to price increases. The price elasticity for youth) spread over
tobacco control programme. One Thailand estimate uses a relatively time through to the population (Levy
major limitation of these studies is conservative estimate of the relative et al., 2000).
that tobacco control programmes mortality risks of smoking. If relative In general, studies indicate that
in California include not only risks increase to those for the USA, reductions in smoking rates due to
taxation, but also non-price tobacco the number of premature deaths increased taxes lead to a reduction
control programmes. Although it is avoided could rise to about 540 000 in premature deaths due to smoking,
difficult to sort out the net effect of by the year 2026. Price increases also including deaths during the ages
tobacco taxation from other tobacco played an important role in Korea’s of 40–64 when smokers would
control programmes that were also reduction in smoking prevalence otherwise be working. Because the
implemented, tobacco taxation is the (Levy et al., 2010). risks of death decrease with the years
key feature of the California tobacco The SimSmoke model has been since quitting and the effects occur
control programme, and thus much used to project the lives saved from largely after the age of 40, the full
of the effect of tobacco control in imposing a hypothetical tax increase, effects of a tax increase will extend
California can be attributed to taxes. and shows how the number of lives far into the future. Two limitations
The effects of taxes and tobacco saved generally increases steadily should be noted on previous work.
control in California have also been per year. For example, using the First, studies for low-income nations,
estimated using the SimSmoke SimSmoke model for Viet Nam, such as China, use relative risks of
tobacco control simulation model Levy et al. (2006) show that with the smoking much lower than those for
(Levy et al., 2007a). Based on that tax increasing from 36% of price to high-income nations, such as the
model, the number of premature 72% of price, 5921 lives would be USA. The relative risks of smoking
330
may be expected to increase in these the healthcare cost of current 1995) and US$5.0 billion in 2000
low-income nations as the duration smokers, former smokers and never (Sung et al., 2006). Using a similar
and intensity of smoking increase, smokers, mainly for three disease method, the medical cost of smoking
thus suggesting that future deaths categories: cancer, vascular disease in India was $907 million in 2004
due to smoking may be understated. and respiratory diseases. Using (John et al., 2009b). Comparing each
In addition, while some studies healthcare expenditures surveys country’s healthcare cost of smoking
have distinguished the relative risks and associating with these disease to their respective total healthcare
of cigarettes from bidis (Jha et al., categories, the SAF is calculated expenditures, the United States
2008), none of the studies of tax by disease category and relevant spends about 5% to 7% of its total
effects consider the effects on health socio-demographic distinctions healthcare costs for smoking-related
of smokeless tobacco both alone (rural/urban, gender, age). The costs, while China and Germany
and synergistically when used with estimated SAF is then multiplied spend about 3% of their total
smoked tobacco. by each cost estimate of interest to healthcare costs in this area.
obtain smoking attributable cost. One approach for assessing
Impact of smoking on health cost For example, the product of SAF the impact of tobacco taxation on
savings and total inpatient hospitalization healthcare cost savings is to first
expenditure is smoking-attributable estimate the amount of healthcare
As shown in Figure 9.1, one of the hospitalization costs;; the product services (inpatient, outpatient,
immediate impacts of raising the of SAF and total morbidity cost is medication) prevented or averted
tobacco tax is that some smokers quit smoking-attributable morbidity costs. due to quitting smoking, and then
smoking. Quitting smoking reduces Most cost-of-smoking estimates for to estimate the unit cost of these
the incidence of smoking-related the United States (Rice et al., 1986), healthcare services. The relationship
illness, thus saving healthcare China (Jin et al., 1995;; Sung et al., between tobacco taxation and
costs. The potential impact of health 2006), Viet Nam (Ross et al., 2007), healthcare cost savings can be seen
cost savings can be documented and India (John et al., 2009b) have in a Chinese study (Hu et al., 2008).
by numerous studies of the cost of used this SAF method. With a tax increase of one RMB
smoking around the world. An alternative method of cost (US$0.15) per pack (tax rate at retail
There are two approaches to estimation is to use a regression price from 40% to 51%) and a price
estimating the cost of smoking: one model comparing the overall HODVWLFLW\RIíPLOOLRQVPRNHUV
is based on the incidence rate, and healthcare expenditures between in China would quit. According to a
the other is based on the prevalence smokers (former and current) and cost-of-smoking analysis (Sung et
rate. The incidence-based approach never smokers, as employed in al., 2006), per-smoker medical costs
estimates the lifetime cost of smoking studies for the United States (Bartlett attributable to smoking were about
(Hodgson, 1992;; Manning et al., et al., 1994;; Miller et al., 1998;; Miller US$25.4 in China in the year 2000.
1989) while the prevalence-based et al., 1999). This method requires Thus, 4.1 million fewer smokers would
studies estimate the cost of smoking extensive data to compare various result in savings of US$100 million
in a given year for the USA (Bartlett et smoking states as well as a set of in medical care costs. Table 9.2
al., 1994;; Miller et al., 1998;; Miller et socio-demographic variables. summarizes studies of the impact of
al., 1999), for Viet Nam (Ross et al., The magnitude of smoking costs taxation on healthcare cost savings.
2007), for China (Sung et al., 2006) is influenced by method of estimation, A Netherland study (Barendregt et
and for India (John et al., 2009b). cost data source, assumption of SAF, al., 1997) compared healthcare costs
The incidence approach requires and time period of the study. Even for among smokers and nonsmokers
extensive sources and numerous the same country and the same year, to estimate the effect of smoking
lifetime epidemiological and economic costs could vary by these factors. cessation on healthcare costs over
assumptions. Thus, most cost of For instance, the estimated cost time. They found that the healthcare
smoking studies are estimated by the of smoking in the United States for costs for smokers at any given age
prevalence-based approach. 1993 varied between $53.4 billion are as much as 40% higher than
The key assumption used in (Miller et al., 1999) and $72.7 billion those for nonsmokers, but only in
the prevalence approach is to (Miller et al., 1998);; in China, the the short run. In the long run, the
estimate the smoking-attributable direct medical cost of smoking was study compared the healthcare costs
fraction (SAF) for comparing US$0.8 billion in 1989 (Jin et al., incurred by smokers and lifetime
331
Table 9.2. Impact of tobacco taxation on healthcare cost savings
References Country Amount of tax increase Assumptions Healthcare cost savings
Asia
Hu et al. (2008) China 1 RMB (US$ 0.15) 3ULFHHODVWLFLW\ í US$ 100 million
Europe
Reed (2010) United Kingdom 5% tax increase 3ULFHHODVWLFLW\ í £ 10.2 billion
North America
Ahmad (2005) USA (California) 20% tax increase – US$12 billion for 75 years paid
Lightwood et al. (2008) USA (California) US$ 0.050 increase per 3ULFHHODVWLFLW\ í± US$ 86 billion (95% CI = 28, 151
pack í billion) between 1989–2004
healthcare cost incurred by non- A California simulation study Cost-effectiveness analyses

smokers;; with the additional years (Ahmad, 2005) projected the impact
gained, the total lifetime healthcare of a tobacco tax increase over a 75- Several cost–effectiveness studies
costs for non-smokers are higher year period. The study found that a have been conducted to consider
than smokers. If these costs are 20% cigarette tax increase would how the benefits of a tax compare to
converted into present value (with a increase tax revenue by almost $10 its costs. Ranson et al. (2000) used
discount rate of 3% to 5%), smoking billion, and smoking-related medical 1995 data and estimated that if the
cessation would not lead to increased costs would decrease by $188 billion inflation-adjusted price of cigarettes
healthcare cost. during the period. The smoking were raised by 10% worldwide
The California 1989 tobacco prevalence level was projected to through tax increases, 42 million
control legislation has had a decrease from 17% to 12% over smokers would be induced to quit.
significant impact in reducing heart the 75-year period. Indeed, current This price increase would prevent
disease and lung cancer. Between California smoking prevalence rate is a minimum of 10 million tobacco-
1989 and 1995, in California the cost almost approaching this target due to related deaths. In comparison,
savings from fewer cardiovascular active tobacco control programmes non-profit measures, such as a
hospitalizations and other medical implemented in California. comprehensive ban on advertising,
services resulted in US$390 million While some studies have bans on smoking in public places,
in savings of 1997 direct medical examined the healthcare costs of prominent warning labels, and mass
costs (Lightwood & Glantz, 1997). smoking for countries other than the information programmes would
A recent study on the effect of USA (see Table 9.2), more research prevent about 5 million deaths. Thus
the California tobacco control is needed on these costs, since it is highly cost-effective to initiate an
programme on personal healthcare these costs are likely to vary from increase in tobacco tax. A study of the
expenditures (Lightwood et al., 2008) country to country. In addition, the United Kingdom prepared for Action
showed that between 1989 and specific effects of a tax increase have on Smoking and Health (Reed, 2010)
2004, the California programme was only been estimated for California provided an economic analysis of the
associated with $86 billion in reduced in the USA. A potentially important impact of increasing tobacco tax on
medical spending. It should be noted healthcare costs savings from higher net benefits to the United Kingdom
again that this estimated healthcare tobacco taxes is through reduced economy as a whole. Assuming
cost savings is attributable not only maternal smoking and reduced í DV SUHYDOHQFH HODVWLFLW\ WKH\
to the increase in the tobacco tax secondhand smoke. These effects of estimate that an increase of 5% in
but also to other non-price tobacco taxes are discussed below. tobacco prices would cause 190
control programmes implemented in 000 smokers to quit smoking, a total
California during this same period. cost savings of 10.2 billion in 2010
332
prices. These cost savings consist improve health, prevent tobacco- who are exposed to their parents’
of £1.97 billion for National Health related illness, and thus save on smoke. Thus, excise tobacco taxes
Services, £1.36 billion for reduced healthcare costs. Cost savings in may have an important health effect
absenteeism, £1.15 billion for output health care is thus an added benefit through reducing the exposures of
from extra working life, and £5.74 from implementation of tobacco children living in families where at
billion for values of extra life years. taxation policy. least one parent smokes.
A cost–effectiveness study Prenatal smoking and postnatal
of tobacco tax increase from a Secondary effects SHS exposure have been identified
healthcare perspective was published of a tax increase as a strong risk factor for SIDS, a
in the Netherlands (van Baal et al., leading cause of infant mortality. A
2007). It was concluded that a tobacco While most of the literature on health US study by Markowitz (2008) found
tax increase is a cost-effective effects focuses on the direct effects higher cigarette prices/taxes and
intervention for public health, even of smoking on the smoker, some smoke-free air laws are associated
when considering additional medical studies examine the relationship with reduced cases of SIDS.
costs from life years gained. between cigarette taxes and the Besides the effects of passive
In a recent study to estimate the health of others or its relationship to smoking, maternal smoking has
health efficacies and financial costs other unhealthy life styles. been associated with low-birth-
of strategies to reduce salt intake Not directly estimated in the weight babies, which leads to
and control tobacco use (Asaria above reviewed studies is the effect substantial added costs over the life
et al., 2007), the authors used the of tobacco tax increases on deaths of the individual, as well as other
WHO Comparative Risk Assessment due to SHS. Substantial evidence health risks to the child (Adams &
project to estimate the effects of now exists that SHS increases the Young, 1999;; US Department of
successful implementation of price risks associated with heart disease, Health and Human Services, 2004).
tobacco control and non-price control stroke, respiratory disease, lung A growing literature examines the
strategies. With the assumption of cancer and sudden infant death relationship of cigarette taxes to birth
price elasticities ranging between syndrome (SIDS) (US Department of outcomes. Using the US Natality
í DQG í IRU ORZLQFRPH Health and Human Services, 2006). Detail File data from 1989–1992,
and middle-income countries, an As tobacco tax increases cause Evans and Ringel (1999) estimated
increased real price of cigarettes smokers to quit, exposure to SHS is that the maternal smoking rate
to reduce smoking prevalence by likely to decrease in the home and at falls by 5% for a 10% increase in
10% in combination with mid-range work, thereby reducing the mortality cigarette prices. In addition, among
estimates of non-price interventions risks associated with secondhand those women who quit smoking in
would reduce the smoking smoke and thereby saving lives. response to a tax increase, average
prevalence rate by 20%, and 2.19 A recent study examined the birth weights rose by approximately
million deaths from cardiovascular effect of taxes and smoke-free air 400 g. In their follow-up study, Ringel
disease would be averted, as well as laws on SHS (Adda & Cornaglia, and Evans (2001) provide more
2.12 million from respiratory disease, 2010). They used the nationally detailed information about the impact
and 1.20 million from cancer, for a representative sample from the of a tobacco tax on the propensity of
total of 5.5 million deaths averted. US National Health and Nutrition pregnant women to smoke, and found
The cost of implementation of this Examination Survey from 1988– that a 10% increase in cigarette price
tobacco control programme would 1994 and from 1999–2002, and would reduce maternal smoking rate
range from US$.04 to US$.032 per quantified the relationship between by 7%. Colman et al. (2003) examined
person for the countries analysed. As the concentration of cotinine in body the effect of cigarette excise taxes
the authors point out, implementation fluids of non-smokers over time. They on smoking before, during and after
costs associated with an increase in found that tobacco excise taxes have pregnancy, and found a statistical
tobacco taxes would be largely if not a significant effect on SHS exposure. association between taxes and
completely offset by the generated A 10% increase in the state excise tax prenatal smoking through both
tax revenues. reduced the cotinine concentration quitting and relapse during and after
Either from a conceptual model or in nonsmokers by about 3.6% for pregnancy. The price elasticity of
empirical analysis, evidence shows children 8 –12 years of age. The effect prenatal quitting and postpartum
that increasing the tobacco tax will was particularly large for children relapse was close to one in absolute
333
value. Finally, Lien and Evans Several other studies have used and marijuana are closely related,
(2005) estimated the impact of large survey data at the level of individual with marijuana being a substitute
cigarette tax hikes in four US states consumers. Jimenez and Labeaga for alcohol and a complement of
on maternal smoking and infant birth (1994) estimated demand equations tobacco, and alcohol and tobacco
weight. Smoking during pregnancy in a demand system using individual being complements. Zhao and Harris
doubled the chance that an infant expenditures taken from the Spanish (2004) investigated marijuana, alcohol
was born with a low birth weight. Family Expenditure Survey (SFES). and tobacco consumption using
Several studies have examined The resulting cross-price elasticity micro-unit data from the Australian
the interrelationship between tobacco of tobacco with respect to alcohol National Drug Strategy Household
prices and other unhealthy lifestyles. SULFH DYHUDJHG í VXJJHVWLQJ Surveys. Consistent with findings of
To date, studies have considered the complementarity between the two complementarity, the results indicate
effect on alcohol, marijuana use and commodities. Decker and Schwartz a positive relationship across all
obesity. (2000) considered two separate three substances, with the correlation
Most econometric analyses demand equations for alcohol and between marijuana and tobacco use
addressing tobacco and alcohol use cigarettes where each included the being the highest.
are based on estimating demand price of both goods among other Studies have also focused
functions and calculating cross- factors. The overall cross-price specifically on the relationships
price effects from estimated price elasticity of alcohol with respect to between cigarette taxes and prices
coefficients. Some studies rely on cigarettes was +0.50, suggesting and other substance use among
aggregate data at regional or national that the two goods are substitutes, youth. Chaloupka et al. (1999) found
level. Using aggregate quarterly while that of cigarettes with respect that higher cigarette prices were
expenditure data for the United WR WKH SULFH RI DOFRKRO ZDV í associated with a reduced frequency
Kingdom over the period 1964– Fanelli and Mazzocchi (2004) of youth marijuana use. Farrelly et al.
1983, Jones (1989) estimated budget developed a dynamic modelling (2001) reached similar conclusions,
shares equations which included four approach, which is consistent with finding that higher cigarette taxes
categories of alcoholic beverages the rational addiction theory and with reduced the probability that young
and tobacco. He found tobacco to be the hypothesis of adjustment costs. males would use marijuana as well
a complement to all four categories A strong complementarity between as the intensity of youth marijuana
of alcohol. Goel and Morey (1995) alcohol and tobacco consumption use. Pacula (1998a, 1998b) provides
used a pooled set of data organized was found. Picone et al. (2004) additional support, finding that youth
by year and state on US cigarette examined the increases in the costs marijuana use was lower in US states
and liquor consumption for the period and barriers to smoking in the USA with higher cigarette taxes, although
1959–1982, and found the t wo goods to study the relationships between her estimates were not statistically
to be substitutes. However, the cross- smoking and drinking behaviours. significant. Extending this work to
price effects differed markedly: from They found that past consumption allow for habit formation, Pacula found
+0.33 for liquor to +0.10 for cigarettes, of cigarettes was positively related that higher past and current cigarette
suggesting that there may be some to current alcohol consumption, prices significantly reduced current
asymmetry in the number of people that increasing the cost of smoking youth marijuana use, but that higher
who smoke and drink liquor and (through the introduction of smoking past cigarette prices were associated
those who only smoke or only drink bans) was associated with reduced with greater current alcohol use,
liquor. Bask and Melkersson (2004) alcohol consumption, and, finally, suggesting that cigarettes and alcohol
modelled the demand for alcohol and that higher cigarette prices were are economic substitutes for youth. In
cigarettes as two separate equations associated with increased alcohol contrast, based on an analysis of teen
and then as a simultaneous system, consumption (suggesting a substitute smoking and drinking using state-
taking into account habit formation. effect). level data, Dee (1999) concluded
They used aggregate annual time Several studies of the adult that stronger alcohol control policies
series on sales volumes for the population considered the use of reducing youth smoking prevalence
period 1955–1999 in Sweden, and marijuana along with tobacco and and higher cigarette taxes reduced
found that alcohol and cigarettes are alcohol. Cameron and Williams the prevalence of youth drinking.
complements in consumption. (2001) showed that decisions of Finally, using longitudinal data,
participation in tobacco, alcohol Markowitz and Tauras (2006) found
334
negative (though generally non- amount of tax revenue designated allocation of general revenue among
significant) effects of beer prices for spending on specific government competing uses. Earmarking may
on youth smoking prevalence, and or public services (Buchanan, 1963). be an arbitrary fiscal policy that
their proxy for marijuana prices In other words, earmarking calls for does not lead to an optimal social
generally indicated negative and a simultaneous choice both on the welfare budget allocation principle
significant effects on youth smoking level of taxation and expenditures on (Musgrave & Musgrave, 1980).
prevalence, consistent with most of an item-by-item basis. Furthermore, relying on a particular
the rest of the literature implying that Earmarking of tax revenue has type of tax revenue may generate
cigarettes, marijuana and alcohol been adopted by countries worldwide uncertainty for future tax revenue
are economic complements (i.e. and has a long history. Justification is streams. However, researchers in
used together) for youth. often based on the benefit principle. the tobacco control field have shown
Most of the studies find negative For example, gasoline or automobile that since earmarked tobacco taxes
cross-price effects and therefore tax proceeds are used for highway are used to fund health promotion
conclude that alcohol and tobacco financing, property tax revenue is and disease prevention programmes,
are complements. Two notable used by the local government for this practice is consistent with the
exceptions, however, are Goel residential services and local public “benefit principle” of taxation and can
and Morey (1995) and Decker and school education, and social security reduce the welfare losses resulting
Schwartz (2000). taxes are used for retirement income. in higher tobacco taxes (Hu et al.,
Studies have also examined impact Here, the benefits to the smokers in 1998). Furthermore, given that many
of tobacco taxes on obesity, since light of the effects of secondhand publicly provided health insurance
quitting smoking may be associated smoke can be considered a programmes target lower-income
with an increase in appetite. Using US justification for earmarking. populations, this type of earmarking
deaths between 1984–1999, Chou Earmarking the tobacco tax for is consistent with the objective of a
et al. (2004) found a positive effect health-related activities is often tax and transfer system.
of cigarette prices on body weight. justified not only on the benefit
Since tobacco taxes are associated principle but also as a “users’ fee” Examples of an earmarked
with higher cigarette prices, this study or a “sin tax.” It is argued that the tobacco tax and its impact
suggests that quitting smoking would tobacco tax covers the external on tobacco control
lead to weight gain. However, a follow- costs of smoking (Manning et al.,
up study (Gruber & Frakes, 2006) 1989;; Manning et al., 1991) and In the 1980s, earmarking tobacco
used the same data set and found no discourages smokers from using taxes was adopted in Australia
evidence that reduced smoking leads cigarettes, thus leading to a reduction and New Zealand via the “Vic-
to weight gain. Chou et al. (2004) used in tobacco-related healthcare costs Health” model that used tobacco tax
the cigarette price as the key variable and secondhand smoke exposure, revenue to fund sports and artistic
in the model, while Gruber and Frakes by reducing the morbidity and events previously funded by the
(2006) used the cigarette tax as a key mortality associated with cigarette tobacco industry. Also, some of the
variable, suggesting that the results smoking. Since the 1980s, countries tax revenue was dedicated to help
are sensitive to model specification. such as Australia, Canada, New tobacco farmers and those employed
Future research is needed to address Zealand, Finland, Egypt, Thailand, in the tobacco industry to move to
the issue between tobacco taxation and some states in the United States other crops and industries.
and the effect of smoking cessation (California, Massachusetts, Oregon Perhaps the first most
on weight gain. and others) have designated part of comprehensive earmarked tobacco
their tobacco tax revenue to finance tax revenue programme was initiated
Impact on earmarked tobacco tax tobacco control activities, health in California. The California Tobacco
revenue promotion, disease prevention, and Tax and Health Promotion Act was
healthcare insurance. passed in 1988 by a popular vote.
Pros and cons of earmarking Some public finance experts The cigarette tax was raised from
tobacco tax revenue have argued that earmarking 10 cents per pack to 35 cents per
may not be a good tax budgeting pack. The goal was to achieve a 75%
As discussed in Chapter 2, an procedure because it introduces reduction in the smoking prevalence
earmarked tax is defined as the rigidities and does not permit proper rate by the year 2000. The Act
335
created the Tobacco Product Suntax taxes played an important role in regarding the earmarked cigarette
Fund, which allocated the revenue reducing smoking rates in the states tax in Taiwan, China (Tsai et al.,
into six accounts (Bal et al., 1990): of Arizona and California. A national 2003) indicated that more than half
(1) 20% of the funds for health and study (Chaloupka & Grossman, 1996) of respondents were in favour of
anti-smoking media campaigns;; further confirmed that the states that earmarking tobacco taxes for health
(2) 35% for indigenous hospital have earmarked a portion of their programmes;; this was especially
services;; tobacco tax revenue for anti-smoking true among nonsmokers with higher
(3) 10% for indigenous physician campaigns have all experienced a income and higher education.
services;; significant impact on both reducing A World Bank report on financing
(4) 5% for research on tobacco- the probability that a youth will initiate healthcare services in China
related diseases;; smoking and on reducing cigarette (Saxenian & McGreevey, 1996) noted
(5) 5% for the environment;; and consumption among young smokers. that, since China is the world’s largest
(6) 25% for the government Besides the United States, cigarette consumption country and a
general account. Thailand is another example of large portion of its rural low-income
The basic philosophy behind success with earmarking. In 2001, population lacks health insurance
this earmarked tobacco tax is that the Thailand government passed coverage, raising the tobacco tax in
population-based tobacco control the Health Promotion Foundation China could double health benefits.
programmes offered through media, Act, which allocates 2% of the A 10% increase in the cigarette tax
schools, websites and public areas total national tobacco tax revenue would reduce cigarette consumption
will both reduce smoking prevalence (about US$93 million per year) to by 5%, and the increased 5% in
and protect nonsmokers. Between establish the Thai Health Promotion government revenue could have
1989 and 1995, about US$1.5 billion Foundation. This Foundation serves been used for one third of China’s
in revenue was appropriated for as a key organization working on poorest 100 million uninsured
these earmarked accounts. public health issues, including population. While the World Bank in
Between 1990 and 1993, the tobacco control programmes. general does not support the practice
State of California spent $26 million Levy and collaborators show the of earmarked taxes, the World Bank
on a state-wide media campaign importance of these programmes in Report noted that, in the case of
designed to change tobacco-related Thailand (Levy et al., 2008a). The tobacco taxation, the benefits of
attitudes and behaviours of the Thai Health Promotion Foundation earmarking tax revenue for health
adult population. To evaluate the not only been actively working with and tobacco control exceed its cost
possible impact of this earmarked the Thai government and parliament with respect to fiscal rigidities.
media campaign together with the to enact tobacco control legislation Several studies have examined
impact of the additional tobacco tax and conduct tobacco control policy the impact of tobacco control
on cigarette consumption, Hu et research, but it has also provided programme expenditures on
al. (1995) employed a time-series tobacco control technical assistance reducing prevalence of smoking.
regression model. They found that to neighbouring countries, such as For the USA, Farrelly et al. (2003)
during the study period, cigarette Viet Nam, Cambodia, and the Lao analysed the impact of funds used in
sales were reduced by 819 million People’s Democratic Republic. those states from 1981 and through
packs attributable to the state tax Taiwan, China also was inspired 2000 on per-capita consumption.
increase, and by 232 million packs by the California Tobacco Tax and Controlling for the effect of price
attributable to the anti-smoking media Health Promotion Act in passing and other factors, they found that
campaign. This study suggests that the Tobacco and Alcohol Tax Act in campaigns meeting the CDC-
earmarking the tobacco tax for a non- 1999 and implementing it in 2002. recommended minimum per-capita
price tobacco control programme This act specified that about 70% of expenditures reduced per-capita
(i.e. anti-smoking campaign, bans on the additional tobacco tax revenue cigarette consumption by 8.5%, with
advertising and promotion of tobacco would go towards the national health ranges depending on the model
products, health-warning labels, and insurance, 10% to anti-smoking specification between 3.4% and
others) had an additional impact on activities, 10% to health promotion 21%. Tauras et al. (2005) examined
tobacco consumption. Levy et al. and disease prevention, and 10% the relationship between state-
(2007 a, 2007b) found that the media to social welfare programmes. A level tobacco control expenditures
campaigns funded by the earmarked follow-up public opinion analysis and youth smoking prevalence and
336
quantity smoked using 1991 through comprehensive strategy of smoke- to other tax bases, such as income
2000 national representative youth free air laws, media campaigns, tax or property tax.
survey. They found that real per- cessation treatment programmes,
capita expenditures on tobacco and advertising restrictions can most Tobacco tax revenue and price
control had a negative and significant effectively reduce smoking and the elasticity of the demand
impact on youth smoking prevalence harms associated with smoking (US for cigarettes
and on the average number of Department of Health and Human
cigarette smoked by smokers (Tauras Services, 2000). Chapter 3, however, The magnitude of tobacco tax
et al., 2005). A similar conclusion discusses some of the complexities revenue that a government can
was reached by another US study involved in the implementation of generate largely depends on (1) the
(Farrelly et al., 2008) for reducing earmarked taxes. level of taxation (either a percent of
adult smoking prevalence. The effect the price or absolute amount of tax
of state tobacco control programme Impact on government revenue per pack) and (2) the price elasticity
expenditures, if all states had of the demand for cigarettes. The
invested at the CDC-recommended As discussed in Chapter 2, two more inelastic the demand, the
minimum ($9.19) or optimum ($22.18) major purposes of tobacco taxation more government revenue can be
levels from 1995 to 2003, would be are (1) to generate government tax generated. When the amount of
relative declines in prevalence of revenue and (2) to reduce smoking to additional tax increases with an
5.4% and 17.4%, respectively, and promote public health. As discussed inelastic demand, the proportional
translate into an estimated 2.2 million in an earlier section of this chapter, reduction in cigarettes purchased
to 7.1 million fewer adult smokers in research evidence shows that raising by the consumer is smaller than
the United States (these numbers the tobacco tax improves population the proportional increase in taxes
were calculated using 25% discount health and reduces healthcare costs. generated. In a simple example
for expenditures). Thus, these This section will address the impact provided by Sunley (2009), if the
studies provide confirmation of the of tobacco taxation on government SULFH HODVWLFLW\ RIGHPDQG LV íD
potential benefits from earmarking revenue. 10% increase in price would reduce
revenues from taxes to tobacco When tobacco taxes are the percentage of consumption
control programmes. increased, the tax is generally by 6%;; however, the tax revenue
Thus, the earmarking of taxes shifted to consumers (see Chapter 3) would increase by 25%. As the
to tobacco control has been used to through an increase in retail prices. price elasticity of demand increases
further reduce smoking rates beyond Consumers reduce their cigarette in absolute value, consumption is
the effects of a tax and to pay for purchases as price increases. It has reduced and the smaller the increase
the healthcare costs associated been shown in previous chapters in government revenue.
with smoking. As described in that the percentage change in the Many examples can be found
earlier sections in this chapter, the demand for cigarettes is negatively from around the world of how
reduction in smoking rates from related to the percentage change in raising the tobacco tax has led to
funding non-price programmes price (i.e. price elasticity of demand). an increase in government revenue
can also further reduce healthcare Therefore, the tobacco tax is a very and, at the same time, a reduction in
expenditures for smokers and non- important policy measure for tobacco cigarette consumption. For instance,
smokers. In particular, funding for control. in 1998 California (USA) passed a
cessation treatment programmes Taxing tobacco products is law (Proposition 99) to increase the
and media campaigns can help those considered an efficient tax for the tobacco tax beginning in 1989 by
quit who have not been induced by following reasons: (1) the demand for US$0.25 (20% of price) per pack
the tax to quit, and those smokers cigarettes is less elastic (somewhere of cigarettes. Between 1989 and
thereby avoid the higher personal EHWZHHQíWRíVHH&KDSWHU 1992, the per-capita annual sales
expenditures associated with the tax. 4) such that a significant amount of cigarettes in California dropped
The promotions and enforcement of government revenue can be by about 10.8 packs, while the state
of smoke-free air laws can help to generated through a small percentage tobacco tax revenue increased
reduce secondhand smoke exposure reduction in cigarette consumption, by $2.0 billion. The estimated tax
and thereby the effects on non- and (2) the administrative cost of tax elasticity with respect to reduction of
smokers. Generally, studies find that collection is relatively low compared FLJDUHWWHVDOHVZDVDERXWíLH
337
a 10% increase in tax resulted in a 6% (2009) show that in each state that an additional US$8.3 billion in
reduction in cigarette sales (Sung et experienced a tax increase showed revenue, even though the number
al., 2005). To further achieve the goal an increase in tax revenues, with of smokers would be reduced by 7.3
of reducing the smoking prevalence the exception of New Jersey, million (Barber et al., 2008). In India,
rate in California, in 1999 the state where a small increase in tax was with a tax rate increase from 38% to
increased the cigarette tax by an accompanied with a stronger smoke- 50%, the Indian government would
additional $0.50 per pack and at the free air law that reduced smoking gain Rs.11 billion, and consumption
same time the industry increased rates. of cigarettes would be reduced by
the price another $0.45 following During a similar time period, the 8 billion sticks (John et al., 2009a).
the Master Settlement Agreement South African government initiated Finally, the Russian study also
(MSA). From 1999–2002, this a tobacco tax increase as a means indicated that a tax of 70% of the retail
additional increase in price resulted of tobacco control. The government price of cigarettes would increase tax
in a further reduction of annual per- gradually raised the tobacco tax revenue by US$5.98 billion at a price
capita cigarette sales by 9.6 packs. rate (Van Walbeek, 2002). Other HODVWLFLW\RIí5RVVet al., 2008).
At the same time, the state cigarette examples from around the world of In sum, all of the post-tax
tax revenue increased by an how an increase in the tobacco tax implementation examples and
additional $2.1 billion during the four- rate is associated with an increase simulation studies uniformly show
year period. The impact of this tax in government revenue include that the tobacco tax is an effective
increase on cigarette consumption the United Kingdom (Townsend, policy instrument to raise government
ZDV í LQ WKH VKRUW UXQ DQG 1998) and Thailand (World Health revenue. Several qualifications should
í LQ WKH ORQJ UXQ 6KHX et al., Organization, 2009). be mentioned. First, other tobacco
2004;; Sung et al., 2005). Again, this Additional studies have been control policies that reduce smoking
example shows that higher taxes will funded by the International Union rates may offset some of the effects
generate more government revenue Against Tuberculosis and Lung of the tax increase, as they reduce
and reduce cigarette consumption. Disease to estimate the impact of a smoking rates independent of the tax
For the United States as a tobacco tax increase on government increase. Second, tax revenues will
whole, tax increases have led to revenue for major smoking countries fluctuate with the general state of the
revenue increases. Giovino et al. such as China, India, Indonesia, economy, e.g. a recession may lead to
(2009) shows the pattern of average and the Russian Federation, who temporary reductions in the quantity
state taxes rates as adjusted for have yet to implement tobacco of cigarettes sold independent of
inflation relative to average state taxation as a measure to control the tax. Third, with additional life
tax revenues. In a graphical display tobacco use. The China study (Hu years gained attributable to tobacco
they show that as average state et al., 2008) used a demand model tax, government may need to incur
taxes have increased, tax revenues to estimate the effect change tax additional pension funds. Fourth,
have increased and the tax revenues on the quantity demand. The study cross-border shopping and smuggling
declined as average state taxes found that an increase of one RMB may offset some of the increase
went down (corresponding to figure (or US$0.15), an increase from a or in exceptional cases reduce tax
37 in page 69 in the above cited tax rate 40% of retail price to 51% revenues, as discussed in Chapter
report). The large price increases of retail price per pack of cigarettes, 8. Finally, at some point, even in the
following the MSA and other tobacco would raise US$31.2 billion in tax absence of the factors described
control policies implemented during UHYHQXHDWDSULFHHODVWLFLW\RIí above, tax increases may reduce
the 1990s and early 2000s have or an additional US$28.7 billion in tax tax revenues when there are so few
somewhat dampened the magnitude UHYHQXHDWDSULFHHODVWLFLW\RIí smokers left.
of the effect of tax increases on If the tax rate increased to 68% of
tax revenues, because these the retail price, even at a low price Impact of a tax increase on overall
policies have reduced smoking HODVWLFLW\ RI í WKH JRYHUQPHQW employment
rates independent of the state would gain an additional US$54.5
tax increases. Nevertheless, tax billion in tax revenue. In Indonesia, at One of the arguments that has been
revenues still increased over that DSULFHHODVWLFLW\RIíDWD[UDWH raised against increasing tobacco
time period. In addition, data in the increase from 37% to 70% of the sale taxes is that the reduction in tobacco
report by Giovino and colleagues price would garner the government consumed either through quitting
338
or reducing the quantity smoked will transfer to other industries. As government sector. If the tax change
will translate into a reduction in consumers spend less on tobacco, is not too large, revenue from taxes
employment, or more generally a they spend more on other goods or will probably increase at least in
shrinking economy. The impact has increase their savings. The increased the short run, because demand is
been estimated as a reduction in the spending in other goods and service inelastic. While more money will be
gross number of tobacco-related directly creates new jobs in the drawn out of the economy by the
jobs in tobacco industry-sponsored economy. Likewise, additional tax government, the increased revenue
reports (Agro-Economic Services revenues generated from raising may be used to replace other taxes
Ltd & Tabacosmos Ltd, 1987;; Arthur tobacco product taxes do not that would otherwise be raised.
Andersen Economic Consulting, 1993;; disappear from the economy, but Alternatively, government could
PEIDA, 1991;; Price Waterhouse, are redirected into other uses by the pay off the national debt (which
1992;; Tobacco Merchant Association, government, creating employment would reduce the interest rate, but
1995). These effects are compounded and other benefits in those sectors. probably also reduce consumption)
as multipliers are applied to the In general, reallocation generates or increase government expenditure
tobacco-related job losses to account alternative business opportunities in (which could generate additional
for the reverberations in other parts other sectors of an economy, along employment). As taxes increase and
of the economy from lost consumer with the associated employment and with sufficient time for adjustment,
expenditures (Jacobs et al., 2000). In income that counteract the loss in the tax revenues will eventually
addition, tobacco industry-sponsored tobacco sectors. decline. Commonly, studies of the
studies overestimate the number of Studies taking into account the employment effects assume that
tobacco-related jobs, since many jobs change in production and spending government will then react either by
are part-time and seasonal, and few patterns have assumed that money decreasing government expenditure,
farms grow only tobacco. not spent on tobacco is spent on which would result in government job
The estimates from the tobacco- other goods and services according losses, or by increasing taxes from
sponsored studies present only part to consumers’ existing (average) other goods and services. However,
of the picture. They should not be expenditure patterns, or, if the data is some increase in taxes would occur
interpreted, as has been suggested, available, they have considered how naturally as consumer expenditure is
as the number of people who would expenditures are modified by those switched from tobacco to other goods
be unemployed if the tobacco industry of smokers who have recently quit and more taxes are collected on these
is reduced. Unlike tobacco industry- (which may further vary depending items. Thus, even a long-run demise
sponsored studies, economic-based on whether the individual quit or of the tobacco industry would cause
studies consider the employment cut back the quantity smoked). For a governmental revenue shortfall
impact of shifting consumer spending those who became former smokers, only if the tobacco tax revenue were
from tobacco products to other goods Buck et al. (1995) found that the not replaced with an equal-yield
and services for a nation or a region. marginal increase in income in the alternative revenue source.
In a dynamic economy, resources— short term was spent more on luxury
including workers—are constantly items, recreational goods, transport, Empirical studies on the effect
shifting between industries due to communication and educational on overall employment of reduced
changes in prices. Thus, economic services (Buck et al., 1995). tobacco use
studies estimate the contribution of the Instead of spending the money
tobacco industry to net employment, saved from not purchasing cigarettes, Most studies that have attempted
the change in employment in a nation some portion of the additional to investigate the impact of falling
or a region after considering the money in Indonesia (Barber et al., tobacco consumption on employment
redistribution of the same resources 2008) and in China (Hu et al., 2008) were done between 10 and 15 years
to alternative uses. could be saved. This would mean a ago, and have been reviewed by
reduction in consumption, but would Jacobs et al. (2000). Since then,
The effect on net employment also mean lower interest rates which studies have been conducted for
would translate into an increase in China, Indonesia and Bulgaria and
In general, as tobacco consumption investment and thus future growth. have been added to the paper. Table
in a country is reduced, e.g. due to A higher cigarette tax may also 9.3 summarizes the results of the
taxes, resources for the most part be associated with changes in the empirical studies.
339
Table 9.3. Studies of the effect on employment of a fall in cigarette consumption
Country (reference) Assumptions Finding
Net exporters
Canada (Allen 1993 as cited in Jacobs et al., Domestic cigarette sales are 0 in 1989, $ Little change in jobs
2000) switched to “average” expenditure patterns
Canada (Irvine & Sims, 1997) 20% decline in tobacco product expenditure 1HWORVVRIMREVíHPSOR\PHQW
Expenditure allocated by average expenditure
pattern.
Government spending reduced.
Indonesia (Ahsan & Wiyono, 2007) Tax increases of 25, 50 and 100% from 26% of Net gain in jobs of 281 135 (0.3% of
price allocated by average expenditure pattern. employment), 140 567 and 84 340 for a 100%,
50% and 30% increase in the cigarette tax
Government spending reduced
United Kingdom (Buck et al., 1995) Tobacco sales fall by 40%, switch to “recent Net gains of 155 542 jobs (+0.5%)
stopper” expenditure patterns
United States (Warner et al., 1996) Eliminate all domestic consumption, switch to Net job gains of 47 in first year, and 133 000
“average” expenditure patterns over 8 years (0%)
Zimbabwe (van der Merwe, 1998b) All domestic production and consumption Net loss of 88 000 jobs in first year, agriculture
eliminated, agriculture shifts to alternative SURGXFWLRQVKLIWVUHVWRUHDERXWKDOIí
crops, expenditures shift to average input-
output pattern
Balanced tobacco economies
Bulgaria (Petkova et al., 2001) 10% consumption fall, expenditures shift to 1HWORVVRIMREVíHPSOR\PHQW
average input-output pattern
China (Hu et al., 2008) A linear production relationship between :LWKHODVWLFLW\RIíORVVRIMREV
production volume and employment loss a RIMREV$WDSULFHHODVWLFLW\RIí
SULFHHODVWLFLW\RIíDQGí loss of 5 49 jobs
An additional US$0.15 (1 RMB) per pack tax

increase
Scotland (McNicoll & Boyle, 1992) Eliminate all domestic consumption in 1989, Net gain of 7869 jobs (+0.3% employment).
switch to “average” expenditure patterns
South Africa (van der Merwe & Abedian, 1999) Eliminate all domestic consumption, switch to Net gain of 50 236 jobs (+0.4% employment)
“recent stopper” expenditure patterns
Net Importers
Bangladesh (van der Merwe, 1998a) Eliminate all domestic consumption and all Net gain of 1 098 919 (+2.0%) jobs.
domestic production of cigarettes and bidis in
1994, switch to “average” expenditure patterns
United States;; Michigan State (Warner & Eliminate all domestic consumption in 1992, Net gain of 7100 jobs over time (+0.1%
Fulton, 1994) switch to “average” expenditure patterns employment)
340
Most of the studies have 1992;; Jacobs et al., 2000). A study by that a 10% decrease in consumption
compared the actual level of Buck et al. (1995) used a static model with an expenditures shift to average
employment in their different of the United Kingdom comparing input-output pattern yielded a
economies with the predicted the economy with one where 40% of QHW ORVV RI MREV í RI
level of employment when tobacco tobacco expenditure was switched employment).
expenditure is reduced. All studies to other forms of spending. Where A recent study by Ahsan and
have assumed that consumers other studies had all assumed that Wiyono (2007) for Indonesia used
who stop smoking reallocate their consumers would spend their money price elasticities to examine tax
tobacco expenditure to other goods according to average consumption effects. They found that new jobs
and services in the economy. Falling expenditure patterns, this study would increase by 281 135 (0.3% of
employment in the tobacco industry considered four patterns of changed employment), 140 567 and 84 340
will thus be offset by increases in consumer spending: (1) average for a 26% (100%), 13% (50%) or
employment in other industries, consumers (the usual assumption), 8% (25%) increase in the cigarette
depending on how labour intensive (2) all nonsmokers, (3) all former price (tax), respectively (Ahsan &
these other industries are relative smokers, and (4) recent stoppers (the Wiyono, 2007). Other non-tobacco
to the tobacco industry. This is the more realistic case). All categories crops showed the largest gain. In
first logical underpinning of all the showed net increases in jobs, except Indonesia, 13% of expenditures are
research on this topic. if released expenditure followed for tobacco in households with at
To simulate the change in nonsmokers’ spending patterns, least one smoker.
employment from a reduction or which was considered unrealistic. To estimate the potential impact
elimination of tobacco consumption, If ex-smokers spent freed money of cigarette tax increase on cigarette
the amount of expenditure like recent stoppers, a net increase industry employment in China,
released from tobacco spending of 115 688 full-time equivalent jobs a linear production relationship
distributed according to an assumed was predicted for the UK (+0.5% of between production volume and
expenditure pattern is then applied employment). employment loss was used (Hu
to either a static input-output model Static input-output models were et al., 2008) to show that under a
or a dynamic regional econometric conducted for four low- and middle- SULFH HODVWLFLW\ RI í DQG ZLWK
model. Both types of models income nations. For South Africa, van an additional US$0.15 (1 RMB)
contain the interdependencies or der Merwe and Abedian (1999) found per pack tax increase, only 1656
relationships between industry a net gain of 50 236 jobs (+0.4% individuals would be unemployed.
sectors or subsectors in the economy employment) in 1995 with tobacco $W D SULFH HODVWLFLW\ RI í ZLWK
and can be used to simulate the elimination, using expenditures of the same amount of tax increase,
impact of an external policy change recent quitters, and assuming the approximately 5549 individuals
on outputs and employment of each same government spending. For would lose their jobs, as compared
sector of the economy. The static Bangladesh, van der Merwe (1998a) to the loss of 59 000 employees due
approach usually compares two found a net gain of 1 098 919 jobs to company merging in 2006. The
alternative situations in a given year, (+2.0% employment) in 1994 with amount of employment loss from an
one with and one without (or with domestic consumption expenditure increase in taxes would be minimal.
reduced) tobacco expenditure taking and all tobacco production for Furthermore, the effect of the
place. The dynamic model allows tobacco products and bidis in 1994 reduction in cigarette consumption
one to simulate trade flows and eliminated, and with previous average would lead the cigarette industry to
feedback effects over time. expenditures and no change in diversify into other products.
McNicoll and Boyle (1992) government spending by other taxes. In contrast to the static models,
estimated the impact on the Scottish For the tobacco-growing nation of Warner and Fulton (1994) developed
economy of a reduction in spending Zimbabwe, van der Merwe (1998b) a dynamic regional economic model
on cigarettes in Glasgow in 1989 using calculated 87 798 jobs lost in 1980 to examine falling consumption in the
a static input-output model. Their í HPSOR\PHQW DQG state of Michigan, a non-tobacco-
results indicated a net gain of nearly jobs when all domestic consumption producing state. They examined
8000 jobs (+0.1% of employment) if expenditure is eliminated and farming two scenarios: a complete and
everyone in Glasgow had stopped moves to alternative agriculture. For instantaneous cessation of tobacco
smoking in 1989 (McNicoll & Boyle, Bulgaria, Petkova et al. (2001) found expenditure, and a more realistic
341
gradual acceleration in the rate of effect on the total output and for males and US$32.6 billion for
decline in tobacco consumption. employment of the national economy, females) from 2000–2004 (Centers
Their results indicate that Michigan except in countries and some states for Disease Control and Prevention,
would experience a net gain of that are heavily dependent on tobacco 2008). These costs did not include
5600 jobs in 1992, and 1500 more production. A reduction in smoking other potentially important costs,
jobs (+0.1% employment) by the does lead to job losses in those sectors such as the value of lost work time
year 2005. The Michigan study was immediately associated with tobacco from smoking-related disability,
extended to include the entire US to product production, such as tobacco absenteeism, excess work breaks,
assess whether declining tobacco manufacturing and farming, and and secondhand smoke-related
product sales significantly reduced supplier sectors indirectly associated disease morbidity and mortality
employment across all tobacco- with tobacco such as the fertilizer (Centers for Disease Control and
producing and non-tobacco- and paper industries. But these Prevention, 2005). Nevertheless,
producing states from 1993 to the losses are generally outweighed there are costs associated with the
year 2000 (Warner et al., 1996). The by increases in employment in all transition from the current tobacco
results showed that the tobacco- other industries. For any country or economy to a smaller or tobacco-free
producing region would have region, the estimated net change of economy.
lost jobs, but collectively all other employment depends on the specific
regions would have gained enough assumptions used in the studies (e.g., Transition costs
employment to completely offset the reactions by government and
the losses. Eliminating all tobacco assumed spending patterns of those As suggested by Warner (2000), an
spending was predicted to result in a who reduce consumption) and the economic presence of tobacco does
net gain of 133 000 jobs (~0.0% of structure of the domestic economy. not imply an economic dependence.
employment) in the US economy by The effects of a tax increase on The extent and speed at which
the year 2000, while a gradual decline employment will also depend on resources are shifted from the
in tobacco consumption would result whether a country is a net importer tobacco industry to other industries
in a net gain of 19 719 jobs. or exporter of tobacco. In particular, will depend on the opportunity cost
A study for Canada (Allen, 1993) worldwide demand for tobacco of resources used in the production
argued that, first, with consumption products has been declining in most of tobacco and tobacco products.
dropping slowly, those jobs that were regions of the world and is likely to For example, agricultural resources,
lost would mostly be met by attrition. continue to decline as rapidly, and such as land and capital, currently
Second, technological changes more rapidly if tobacco control is used in tobacco leaf production
caused many of the job losses in successful. Therefore, the value of can be used for other agricultural
the industry, and third, the most investments made in these industries purposes, and tobacco farmers can
potent cause of falling consumption is likely to decline in those exporting produce other crops or engage in
was increased excise taxes and countries regardless of tax increases other economic activities if tobacco
the government revenues that were in the country in which they are leaf production ceases. Similarly
raised through these taxes were located. for the resources used in tobacco
spent to maintain public services, Two other savings may be product manufacturing, wholesaling,
which created many new jobs. realized by economies that reduce and retailing.
However, Irvine and Sims (1997) tobacco consumption—the savings The transition costs of switching
conducted a static analysis for a from reduced medical costs (some resources in response to a tax
20% decline in tobacco expenditure, of which will be offset by higher increase will depend on the extent to
and found a net loss of 6120 jobs medical costs as the individual which resources currently involved
RI HPSOR\PHQW LQ &DQDGD lives longer) and productivity gains. in tobacco-related industries are
However, they assumed government For example, the annual economic specific to those industries. Many of
spending was reduced, and Jacobs cost associated with reduced the jobs involve tobacco in a limited
et al. (2000) note that Canada is a productivity for the United States as way (e.g. retailers who sell tobacco
tobacco-growing region. a results of deaths due to smoking products, jobs in the heavy industry
Economic studies generally and secondhand smoke exposure sector where farming equipment is
conclude that reducing tobacco was estimated to be approximately produced), and are not dependent
consumption has a small positive US$96.8 billion (US$64.2 billion on tobacco. Primarily jobs in tobacco
342
farming (which are often part-time of tobacco farmers. For example, develop alternatives in the 1980s
and for which other crops besides Price Waterhouse (1992) estimated (Pan American Health Organization,
tobacco are grown), tobacco leaf that increases in the federal tax on 1992). Many farmers have ceased
drying and warehousing (which cigarettes from 24 to 48 cents per tobacco production through this
generally involves few jobs) and pack in the United States would lead programme, but many participants
tobacco product manufacturing are to a loss of farm revenues by US$50 acknowledge that they would have
partially dependent on tobacco. million and farm jobs by 8140. In quit tobacco farming without the
Labour and land currently used contrast, Sumner and Alston (1985) programme. In addition, 24% of
for tobacco leaf production are and Brown (1998) claimed that the the participants continue to work
sometimes considerably more impact of tobacco measures such in tobacco farming as employees,
productive in tobacco than in other as increasing cigarette taxes on rather than as entrepreneurs (Pan
uses, since tobacco grows well on farmers’ revenues would depend American Health Organization,
some lands where most other crops on the response by the federal 1992). Tobacco farmers are unlikely
do not. The ability to transfer skills government to the lower demand to reduce their tobacco production
and experience specific to tobacco for tobacco due to the governmental as long as tobacco remains more
growing to other crops may also be intervention in tobacco farming. They profitable than the other crops.
incomplete. Similarly, the equipment, concluded that adoption of a policy to Several studies have evaluated
technology and labour skills used to fix the tobacco quota while allowing the alternative crops for tobacco in
produce tobacco products may have the support price to fall would lead to developing counties. Those crops
limited value when released from a loss of revenue for quota owners include cassava in Brazil, sugar cane
tobacco product manufacturing. and no change in revenue for the in Kenya and chillies, soybeans,
While there may be some costs tobacco farmers who rent quotas cotton and mustard in India (Jacobs
due to dislocation, these costs for growing tobacco, while adopting et al., 2000). A study on potential
will be lessened if the reduction in the converse policy would lead to a crops in Bangladesh has found that
demand for tobacco is gradual, as is revenue gain for the quota owner and several vegetable species could be
typically the case. Tobacco farmers a revenue loss for the quota renter. more profitable than tobacco (Naher
and manufacturers can adjust by not While reducing cigarette & Efroymson, 2007). Rose blooms
replacing equipment as it depreciates, consumption can create economic have been identified as profitable
not hiring additional workers, and hardship for those whose livelihood alternatives to tobacco in Zimbabwe
making the land available for other depends on tobacco, programmes (Maravanyika, 1998). Yach (1996)
purposes. In addition, for a country can be implemented to reduce reported that more than fifty
that imports tobacco, these effects these hardships. In high-income alternative crops and land use have
will be minimal, and for a country nations, efforts have been made to been identified.
that exports tobacco, the exported diversify the economic activities for Crop diversification programmes
shares will depend on the tobacco tobacco farmers and reduce their and retraining for workers in
control efforts of other countries. dependence on tobacco farming. In tobacco product manufacturing
Furthermore, independent of tobacco the United States, the efforts have could be funded by some of the new
control policy-induced changes in had limited success. Although the revenues that result from the tax
domestic tobacco use, employment high return from growing tobacco increase. In Turkey, for example, the
in tobacco dependent sectors has has generally limited the impact of government-sponsored alternative
been falling over time as farming efforts to encourage the production crop programme was implemented
techniques and manufacturing of alternatives, many tobacco- in anticipation of privatization of
process have improved (Capehart, growing households are already the country’s monopoly (Yurekli et
2000;; Van Liemt, 2001). quite diversified. For example, in al. 2010). For Brazil, a World Bank
Various studies for the United the United States, some flue-cured report (Vargas & Campos, 2005)
States (e.g. Chase Econometrics, tobacco farms also grow soybeans, found that diversification strategies,
1985;; Price Waterhouse, 1992;; corn, cotton and wheat (Jacobs et especially those part of broader rural
Sumner & Alston, 1985;; Brown, al., 2000). In Canada, the Tobacco development programmes, were
1998) have concluded that increased Diversification Plan provided successful despite considerable
cigarette taxes would reduce demand tobacco farmers with incentives barriers.
for tobacco, and reduce the incomes to stop growing tobacco and to
343
Impact of tobacco tax increases example, if taxes account for 50% of tax increases will be overstated as
on inflation and the consumer tobacco product prices, a doubling of consumption of these products falls
price index the tax (100% increase) will increase in response to tax increases.
prices by 50%. If the weight given to While cigarette tax increases are
Consumer price indices reflect tobacco products in the price index not likely to have a major impact on
the average prices paid by the is 3%, the index will rise by 1.5% in the price index, some governments
typical consumer in a country. The response to the tax increase. have excluded tobacco from the
percentage increase in the consumer For estimating the impact of index. The European Union, for
price index measures the inflation cigarette expenditures on the example, has recommended that its
rate, which directly affects domestic consumer price index, the total member countries exclude tobacco
interest rates and foreign exchange percentage of tobacco expenditures products (Baille, 1998). France since
rates. These rates are key economic from the entire population (smoking 1992 and Belgium since 1994 have
indicators for most countries and and non-smoking) is used. Indonesia excluded tobacco products from
often a key determinant of monetary and China households with smokers the price index used for adjusting
policy. In many countries, changes in spend about 6% to 12.4% of their wages (Guindon et al., 2002). Since
wages, social security benefits, and total household expenditures on 1991, Luxembourg has excluded
other payments are tied to a price tobacco (Barber et al., 2008;; Hu et tobacco products, hard liquor, and
index to keep pace with inflation. In al., 2008). However, slightly more other certain other products from
addition, some countries link taxes to than 1% of all household expenditure its consumer price index because
price indices in order reflect current (smoking and non-smoking) in China they are considered unnecessary
prices. is for cigarettes (Djibuti et al., 2007). or inappropriate. To date, however,
An argument made against The US cigarette expenditures are most countries include tobacco
raising cigarette taxes is that tobacco less than 1% (Busch et al., 2004) and products in their most widely-used
taxes have an inflationary impact, in the Russian Federation tobacco indices—especially those used for
especially in countries where wages accounts for between 1 and 3% of indexing wages, pension payments
are indexed to inflation or where household expenditures (Djibuti et and other outlays. Excluding tobacco
government policy is to keep inflation al., 2007). products from price indices would
low. The effect of tobacco product Thus, tobacco is likely to play increase the public health impact of
tax increases on inflation depends a small role in the price index. tobacco tax increases by providing
on the percentage of the tobacco Furthermore, since the weights used less of a cushion for users whose
price increase due to the tax, and to compute price indices in many wages or benefit payments are
the weight that tobacco prices are countries change infrequently, the indexed (Alchin, 1995).
given in computing a price index. For inflationary impact of tobacco product
344
References
Adams EK, Young TL (1999). Costs of Barendregt JJ, Bonneux L, van der Maas PJ Centers for Disease Control and Prevention
smoking: a focus on maternal, childhood, and (1997). The health care costs of smoking. (CDC) (2008). Smoking-attributable mortality,
other short-run costs. Med Care Res Rev, N Engl J Med, 337:1052–1057.doi:10.1056/ years of potential life lost, and productivity
56:3 –29.doi:10.1177/107755879905600101 NEJM199710093371506 PMID:9321534 losses–United States, 2000–2004. MMWR
PMID:10189774 Morb Mortal Wkly Rep, 57:1226–1228.
Bartlett J, Miller LS, Rice DP, Max W;; Centers PMID:19008791
Adda J, Cornaglia F (2010). The effects of for Disease Control and Prevention (CDC)
taxes and bans on passive smoking. Appl (1994). Medical-care expenditures attributable Chaloupka F, Pacula RL, Farrelly MC et al.
Econ, 2:1–32. to cigarette smoking–United States, 1993. (1999). Do higher cigarette prices encourage
MMWR Morb Mortal Wkly Rep, 43:469–472. youth to use marijuana? NBER Working Paper
Agro-Economic Services Ltd Tabacosmos PMID:8007931 Series. Working Paper # 6939. Cambridge, M A,
Ltd. (1987). The employment, tax revenue National Bureau of Economic Research.
and wealth that the tobacco industry creates. Bask M, Melkersson M (2004). Rationally
London, Agro-Economic Services. addicted to drinking and smoking? Appl Econ, Chaloupka FJ, Grossman M (1996). Price,
36:373–381 doi:10.1080/0003684041000167 tobacco control policies and youth smoking.
Ahmad S (2005). Increasing excise taxes on 4295. NBER Working Paper Series. Working Paper
cigarettes in California: a dynamic simulation #5740. Cambridge, MA, National Bureau of
of health and economic impacts. Prev Med, Brown AB (1998). Farm level effects an Economic Research.
41:276–283.doi:10.1016/j.ypmed.2004.10.024 increase in federal cigarette taxes under two
PMID:15917022 scenarios: Keep vs. Eliminate the Tobacco Chaloupka FJ, Hu TW, Warner KE et al.
Program. Washington D.C., USDA Outlook (2000). Taxation of tobacco products. In:
Ahsan A , Wiyono N (2007). The impact analysis Forum. Jha P, Chaloupka F, eds., Tobacco control in
of higher cigarette price to employment in developing countries. Oxford, U.K., Oxford
Indonesia.SEATCA (South East Asia Tobacco Buchanan JM (1963). The economics of University Press, Chapter 10;; 237-272.
Control Alliance), Demographic Institute, earmarked taxes. J Polit Econ, 71:457–469
Faculty of Economics, University of Indonesia. doi:10.1086/258794. Chaloupka FJ, Warner KE (1999). The
Economics of Smoking. In: Culyer AJ, Newhouse
Alchin TM (1995). A note on tobacco product Buck D, Godfrey C, Raw M et al. (1995). JP, eds., Handbook of Health Economics
prices in the Australian CPI. Appl Econ Lett, Tobacco and jobs, Society for the Study of Volume 1B. Handbooks in Economics 17.
2:273–277 doi:10.1080/135048595356907. Addiction and Centre for Health Ecconomics. Oxford, UK, Elsevier, Chapter 29;; 1539-1627.
York, University of York.
Allen RC (1993). The false dilemma: the impact Chase Econometrics (1985). The economic
of tobacco control policies on Employment in Burns D, Lee L, Shen L et al. (1997). Cigarette impact of the tobacco industry in the United
Canada. Ottawa, Ontario, National Campaign smoking behavior in the United States. In: States Economy in 1983. Bala Cynwyd, PA,
for Action on Tobacco. Burns D, Garfinkel L, Samet J, eds., Changes Chase Econometrics.
in Cigarette-Related Disease Risks and Their
Arthur Andersen Economic Consulting (1993). Implications for Prevention and Control. Chou SY, Grossman M, Saffer H (2004).
Tobacco industry employment: a review of Smoking and Tobacco Control Monograph An economic analysis of adult obesity:
the price waterhouse economic impact report N°2. Bethesda, MD, U.S. Department of Health results from the Behavioral Risk Factor
and tobacco intsitute estimates of economic and Human Services, National Institutes of Surveillance System. J Health Econ, 23:565–
losses from increasing the federal excise tax. Health, National Cancer Institute, NIH Pub No. 5 87.d o i:10 .1016 / j . j h e a l e c o. 2 0 0 3 .10 . 0 0 3
Los Angeles, California, Arthur Andersen 97-4213;; 13-112. PMID:15120471
Economic Consulting.
Busch SH, Joffre-Bonet M, Falba T et al. Colman G, Grossman M, Joyce T (2003).
Asaria P, Chisholm D, Mathers C et al. (2007). (2004). Tobacco spending and its crowd-out The effect of cigarette excise taxes on
Chronic disease prevention: health effects and of other goods. NBER Working Paper Series. smoking before, during and after pregnancy.
financial c osts of strategies to reduce salt intake Working Paper #10974. Cambridge, MA, J Health Econ, 22:1053–1072.doi:10.1016/j.
and control tobacco use. Lancet, 370:2044– National Bureau of Economic Research. jhealeco.2003.06.003 PMID:14604560
2053.doi:10.1016/S0140 - 6736(07)61698 -5
PMID:18063027 Cameron W, Williams J (2001). Cannabis, Decker SR, Schwartz R (2000). Cigarettes and
alcohol and cigarettes: substitutes or alcohol: substitutes or complements? NBER
Baille JC (1998). L’Union Européenne et le complements. Econ Rec, 77:19–34 Working Paper Series. Working Paper #7535.
tabac. Faculté de Médecine de Marseille, Aix- doi:10.1111/1475-4932.00002. Cambridge, MA, National Bureau of Economic
Marseille II. Research.
Capehart T (2000). U.S. Tobacco Farming
Bal DG, Kizer KW, Felten PG et al. (1990). trends. Dee TS (1999). The complementarity of teen
Reducing tobacco consumption in California. smoking and drinking. J Health Econ, 18:769–
Development of a statewide anti-tobacco use Centers for Disease Control and Prevention 793.doi:10.1016/S0167- 629 6(9 9) 0 0 018 -1
campaign. JAMA, 264:1570–1574.doi:10.1001/ (CDC) (2005). Annual smoking-attributable PMID:10847934
jama.264.12.1570 PMID:2395199 mortality, years of potential life lost, and
productivity losses–United States, 1997–2001. Djibuti M, Gotsadze G, Mataradze G,
Barber S, Adioetomo SM, Ahsan S et al. MMWR Morb Mortal Wkly Rep, 54:625–628. Zoidze A (2007). Influence of household
(2008). Tobacco Economics in Indonesia. PMID:15988406 demographic and socio-economic factors on
Paris, International Union Against Tuberculosis household expenditure on tobacco in six New
and Lung Disease. Independent States. B MC Public Health, 7:222.
doi:10.1186/1471-2458-7-222 PMID:17760965
345
Evans WN, Ringel JS (1999). Can higher Gupta PC, Pednekar MS, Parkin DM, Jones AM (1989). A system approach to the
cigarettes taxes improve birth outcomes? J Sankaranarayanan R (2005). Tobacco demand for alcohol and tobacco. Bull Econ
Public Econ, 72:135–154 doi:10.1016/S0047- associated mortality in Mumbai (Bombay) Res, 41:85–106 doi:10.1111/j.1467-8586.1989.
2727(98)00090-5. India. Results of the Bombay Cohort Study. tb00282.x.
Int J Epidemiol, 34:1395–1402.doi:10.1093/ije/
Fanelli L, Mazzocchi M (2004). Back to the dyi196 PMID:16249218 Kaplan RM, Ake CF, Emery SL, Navarro
future? habits and forward-looking behaviour AM (2001). Simulated effect of tobacco tax
for UK alcohol and tobacco demand. Mimeo, Hodgson TA (1992). Cigarette smoking and variation on population health in California.
Università degli Studi di Bologna. lifetime medical expenditures. Milbank Q, 70:81– Am J Public Health, 91:239–244.doi:10.2105/
125.doi:10.2307/3350086 PMID:1588892 AJPH.91.2.239 PMID:11211632
Farrelly MC, Bray JW, Zarkin GA, Wendling
BW (2001). The joint demand for cigarettes Hu TW, Mao Z, Shi J et al. (2008). Tobacco Levy DT, Bales S, Lam NT, Nikolayev L (2006).
and marijuana: evidence from the National Taxation and Its Potential Impact in China. The role of public policies in reducing smoking
Household Surveys on Drug Abuse. J Paris, International Union Against Tuberculosis and deaths caused by smoking in Vietnam:
Health Econ, 20:51–68.doi:10.1016/S0167- and Lung Disease. results from the Vietnam tobacco policy
6296(00)00067-9 PMID:11148871 simulation model. Soc Sci Med, 62:1819–
Hu TW, Sung HY, Keeler TE (1995). Reducing 1830.doi:10.1016/j.socscimed.2005.08.043
Farrelly MC, Pechacek TF, Chaloupka cigarette consumption in California: tobacco PMID:16182422
FJ (2003). The impact of tobacco control taxes vs an anti-smoking media campaign. Am
program expenditures on aggregate cigarette J Public Health, 85:1218–1222.doi:10.2105/ Levy DT, Benjakul S, Ross H, Ritthiphakdee
sales: 1981–2000. J Health Econ, 22:843– AJPH.85.9.1218 PMID:7661228 B (2008a). The role of tobacco control
859.doi:10.1016/S0167- 629 6(03) 0 0 057-2 policies in reducing smoking and deaths
PMID:12946462 Hu TW, Xiao-peng X, Keeler TE (1998). in a middle income nation: results from the
Earmarked tobacco taxes: lessons learned. Thailand SimSmoke simulation model. Tob
Farrelly MC, Pechacek TF, Thomas KY, Nelson In: Abedian I, van der Merwe R, Wilkins N, et Control, 17:53–59.doi:10.1136/tc.2007.022319
D (2008). The impact of tobacco control al., eds., The Economics of Tobacco Control- PMID:18218810
programs on adult smoking. Am J Public Health, Towards an Optimal Policy Mix. Rondebosch,
98:304–309.doi:10.2105/AJPH.2006.106377 South Africa, University of Cape Town, Applied Levy DT, Cho SI, Kim YM et al. (2010).
PMID:18172148 Fiscal Research Centre, Chapter 9 SimSmoke model evaluation of the effect of
tobacco control policies in Korea: the unknown
Ferrante D, Levy DT, Peruga A et al. (2007). Irvine IJ, Sims WA (1997). Tobacco Control success story. Am J Public Health, 100:1267–
The role of public policies in reducing smoking Legislation and Ressource Allocation 1 2 7 3 . d o i : 1 0 . 2 1 0 5 / A J P H . 2 0 0 9 .1 6 6 9 0 0
prevalence and deaths: the Argentina Tobacco Effects. Can Public Policy, 23:259–273 PMID:20466968
Policy Simulation Model. Rev Panam Salud doi:10.2307/3551571.
Publica, 21:37–49.doi:10.1590/S1020- Levy DT, Cummings KM, Hyland A (2000).
49892007000100005 PMID:17439692 Jacobs R, Gale F, Capehart T et al. (2000). Increasing taxes as a strategy to reduce
The supply-side of tobacco control policies. cigarette use and deaths: results of a simulation
Fichtenberg CM, Glantz SA (2000). Association In: Jha P, Chaloupka F, eds., Tobacco control model. Prev Med, 31:279–286.doi:10.1006/
of the California Tobacco Control Program with in developing countries. Oxford, UK, Oxford pmed.2000.0696 PMID:10964642
declines in cigarette c onsumption and mortality University Press, 13;; 311-341.
from heart disease. N Engl J Med, 343:1772– Levy DT, Hyland A, Higbee C et al. (2007b).
1777.doi:10.1056/NEJM200012143432406 Jha P, Chaloupka FJ (1999). Curbing the The role of public policies in reducing smoking
PMID:11114317 epidemic. Governments and the Economics prevalence in California: results from the
of Tobacco Control. Washington D.C., World California tobacco policy simulation model.
Giovino GA, Chaloupka FJ, Hartman AM et Bank. Health Policy, 82:167–185.doi:10.1016/j.
al. (2009). Cigarette Smoking Prevalence and healthpol.2006.09.008 PMID:17055104
Policies in the 50 States: A n Era of Change-The Jha P, Jacob B, Gajalakshmi V et al.;; RGI-
Robert Wood Johnson ImpacTeen Tobacco CGHR Investigators (2008). A nationally Levy DT, Ross H, Powell L et al. (2007a). The
Chart Book. Buffalo, NY, University at Buffalo, representative case-control study of role of public policies in reducing smoking
State University of New York. smoking and death in India. N Engl J Med, prevalence and deaths caused by smoking
358:1137–1147.doi:10.1056/NEJMsa0707719 in Arizona: results from the Arizona tobacco
Goel RK, Morey MJ (1995). The PMID:18272886 policy simulation model. J Public Health Manag
interdependence of cigarette and liquor Pract, 13:59–67. PMID:17149101
demand. South Econ J, 62:451–459 Jimenez S, Labeaga JM (1994). Is it possible
doi:10.2307/1060696. to reduce tobacco consumption via alcohol Levy DT, Ross H, Zaloshnja E et al. (2008b).
taxation? Health Econ, 3:231–241.doi:10.1002/ The role of tobacco control policies in reducing
Gruber J, Frakes M (2006). Does falling hec.4730030405 PMID:7994323 smoking and deaths caused by smoking in
smoking lead to rising obesity? J Health Econ, an Eastern European nation: results from the
25:183–197, discussion 389–393.doi:10.1016/j. Jin SG, Lu BY, Yan DY et al. (1995). An Albania SimSmoke simulation model. Cent Eur
jhealeco.2005.07.005 PMID:16165234 evaluation on smoking-induced health costs J Public Health, 16:189–198. PMID:19256288
in China (1988–1989). Biomed Environ Sci,
Gu D, Kelly TN, Wu X et al. (2009). Mortality 8:342–349. PMID:8719176 Levy DT, Tworek C, Hahn EJ, Davis RE
attributable to smoking in China. N Engl J Med, (2008c). The Kentucky SimSmoke tobacco
360:150 –159.doi:10.1056/NEJMsa0802902 John R, Rao K, Rao G (2009a). Economics of policy simulation model: reaching Healthy
PMID:19129528 tobacco and tobacco taxation in India. Paris, People 2010 goals through policy change.
International Union Against Tuberculosis and South Med J, 101:503–507. PMID:18414175
Guindon GE, Tobin S, Yach D (2002). Trends Lung Disease.
and affordability of cigarette prices: ample Lien D, Evans W (2005). Estimating the impact
room for tax increases and related health John RM, Sung HY, Max W (2009b). Economic of large cigarette tax hikes: the c ase of maternal
gains. Tob Control, 11:35–43.doi:10.1136/ cost of tobacco use in India, 2004. Tob Control, smoking and infant birth weight. J Hum Resour,
tc.11.1.35 PMID:11891366 18:13 8 –14 3 .d o i:10.113 6 / tc . 2 0 0 8 .0 274 6 6 XL:373–392.
PMID:19131453
346
Lightwood JM, Dinno A, Glantz SA (2008). Pacula RL (1998a). Adolescent alcohol Ross H, Trung DV, Phu VX (2007). The costs
Effect of the California tobacco control and marijuana consumption: is there really of smoking in Vietnam: the case of inpatient
program on personal health care expenditures. a gateway effect? Working Paper 6348. care. Tob Control, 16:405–409.doi:10.1136/
PLoS Med, 5:e178.doi:10.1371/journal. Cambridge, MA, National Bureau of Economic tc.2007.020396 PMID:18048618
pmed.0050178 PMID:18752344 Research.
Ross HZ, Shariff S, Gilmore A (2008).
Lightwood JM, Glantz SA (1997). Short-term Pacula RL (1998b). Does increasing the Economics of Tobacco Taxation in Russia.
economic and health benefits of smoking beer tax reduce marijuana consumption? J Paris, International Union Against Tuberculosis
cessation: myocardial infarction and stroke. Health Econ, 17:557–585.doi:10.1016/S0167- and Lung Disease.
Circulation, 96:1089–1096. PMID:9286934 6296(97)00039-8 PMID:10185512
Saxenian H, McGreevey B (1996). China:
Manning WG, Keeler EB, Newhouse JP et Pan american Health Organization (PAHO) Issues and Options in Health Financing. Report
al. (1989). The taxes of sin. Do smokers and (1992). Tobacco or health: status in the 15278-CHA. Washington DC, The World Bank.
drinkers pay their way? JAMA, 261:1604–1609. Americas. A report of the Pan American
doi:10.1001/jama.261.11.1604 PMID:2918654 Health Organization. Pan American Health Shafey O, Ericksen M, Ross H et al. (2009).
Organization. The Tobacco Atlas. Third Edition. Atlanta,
Manning WG, Keeler EB, Newhouse JP et Georgia, American Cancer Society.
al. (1991). The costs of poor health habits. PEIDA (1991). The economic significance of
Cambridge, MA, Harvard University Press. the UK tobacco industry. Manchester, PEIDA. Sheu ML, Hu TW, Keeler TE et al. (2004). The
effect of a major cigarette price change on
Maravanyika E (1998). The economics of Petkova R, Ileva R, Rusenov A, et al. (2001). smoking behavior in california: a zero-inflated
tobacco in Zimbabwe. In: The Economics of Evaluating the impact of tobacco control negative binomial model. Health Econ, 13:781–
tobacco control in South Africa. Cape Town, policies on employment, Bulgarian case 791.doi:10.1002/hec.849 PMID:15322990
School of Economics, University of Cape (Unpublished, commissioned by WHO Kobe
Town, Chapter 12;; 162-172. Centre for Health Development). Sumner DA, Alston JM (1985). Removal of
price supports and supply controls for the
Markowitz S (2008). The effectiveness of Peto R (1994). Smoking and death: the past supply-side effects of tobacco-control policies
cigarette regulations in reducing cases 40 years and the next 40. BMJ, 309:937–939. in US Tobacco: An economic analysis of the
of Sudden Infant Death Syndrome. J PMID:7950669 impact. Washington D.C., National Planning
Health Econ, 27:106–133.doi:10.1016/j. Association.
jhealeco.2007.03.006 PMID:17498829 Peto R, Lopez AD, Boreham J et al. (1992).
Mortality from tobacco in developed countries: Sung HY, Hu TW, Ong M et al. (2005).
Markowitz S, Tauras JA (2006). Even for indirect estimation from national vital statistics. A major state tobacco tax increase, the
teenagers, money does not grow on trees: Lancet, 339:1268–1278.doi:10.1016/0140- master settlement agreement, and cigarette
teenage substance use and budget contraints. 6736(92)91600-D PMID:1349675 consumption: the California experience. Am
Working Paper 12300. Cambridge, MA, J Public Health, 95:1030–1035.doi:10.2105/
National Bureau of Economic Research. Peto R, Lopez AD, Boreham J et al. (1994). AJPH.2004.042697 PMID:15914829
Mortality from tobacco in developed countries,
McNicoll IH, Boyle S (1992). Regional ecnomic 1950-2000. Oxford, Oxford University Press. Sung HY, Wang L, Jin S et al. (2006). Economic
impact of a reduction of resident expenditure on burden of smoking in China, 2000. Tob Control,
cigarettes: a c ase study of Glasgow. Appl Econ, Picone GA, Sloan F, Trogdon JG (2004). 15 Suppl 1;;i5–i11.doi:10.1136/tc.2005.015412
24:291–296 doi:10.1080/00036849200000141. The effect of the tobacco settlement and PMID:16723677
smoking bans on alcohol consumption. Health
Miller LS, Zhang X, Rice DP, Max W (1998). Econ, 13:1063–1080.doi:10.1002/hec.930 Sunley EM (2009). Taxation of cigarettes in
State estimates of total medical expenditures PMID:15386690 the Bloomberg initiative countries: overview of
attributable to cigarette smoking, 1993. Public policy issues and proposals for reform. Paris,
Health Rep, 113:447–458. PMID:9769770 Price Waterhouse (1992). The economic impact International Union Against Tuberculosis and
of the tobacco industry on the United States Lung Disease.
Miller VP, Ernst C, Collin F (1999). Smoking- Economy. Arlington, VA, Price Waterhouse.
attributable medical care costs in the USA. Tauras JA, Chaloupka FJ, Farrelly MC et
Soc Sci Med, 48:375–391.doi:10.1016/S0277- Ranson MK, Jha P, Chaloupka F et al. (2000). al. (2005). State tobacco control spending
9536(98)00344-X PMID:10077285 The effectiveness and cost-effectiveness of and youth smoking. Am J Public Health,
price increases and other tobacco control 95:338–344.doi:10.2105/AJPH.2004.039727
Moore MJ (1996). Death and tobacco taxes. policies. In: Jha P, Chaloupka F, eds., Tobacco PMID:15671473
Rand J Econ, 27:415–428 doi:10.2307/2555934. control in developing countries. Oxford, UK,
Oxford University Press, Chapter 18 Tobacco Merchant Association (1995).
Musgrave R , Musgrave P (1980). Public f inance Tobacco’s c ontribution to the national economy.
in theory and practice. New York, McGraw-Hill. Reed H (2010). The effects of increasing Princeton, NJ, Tobacco Merchant Association.
tobacco taxation. A cost benefit and public
Naher F, Efroymson D (2007). Tobacco finances analysis. London, UK, ASH (Action Townsend J (1998). The role of taxation policy
cultivation and poverty in Bangladesh. Issues on Smoking and Health), Landman Economics. in tobacco control. In: Abedian I, van der
and potential future directions. Case study Merwe R, eds., The Economics of Tobacco
prepared for WHO. Rice DP, Hodgson TA, Sinsheimer P et al. Control: toward an optimal policy mix. Applied
(1986). The economic costs of the health Fiscal Research Center, University of Cape
Neubauer S, Welte R, Beiche A et al. (2006). effects of smoking, 1984. Milbank Q, 64:489– Town, South africa, 8-101.
Mortality, morbidity and costs attributable 547.doi:10.2307/3349924 PMID:3102916
to smoking in Germany: update and a 10- Tsai YW, Yen LL, Yang CL, Chen PF (2003).
year comparison. Tob Control, 15:464–471. Ringel JS, Evans WN (2001). Cigarette Public opinion regarding earmarked cigarette
doi:10.1136/tc.2006.016030 PMID:17130376 taxes and smoking during pregnancy. Am J tax in Taiwan. BMC Public Health, 3:42.
Public Health, 91:1851–1856.doi:10.2105/ doi:10.1186/1471-2458-3-42 PMID:14693036
AJPH.91.11.1851 PMID:11684615
347
U.S.Department of Health and Human Services van der Merwe R (1998b). Employment and Warner KE, Fulton GA, Nicolas P, Grimes DR
(2000). Reducing Tobacco Use. A Report of the output effects for Zimbabwe following a decline (1996). Employment implications of declining
Surgeon General. Atlanta, GA, U.S. Dept. of in tobacco consumption. HNP Discussion tobacco product sales for the regional
Health and Human Services, Center for Disease Paper Series, Economics of Tobacco economies of the United States. JAMA,
Control and Prevention, National Center Discussion Paper. Washington DC, The World 275:1241–1246.doi:10.1001/jama.275.16.1241
for Chronic Disease Prevention and Health Bank. PMID:8601955
Promotion, Office on Smoking and Health.
van der Merwe R, Abedian A (1999). Reduction World Health Organization (2009). WHO
U.S.Department of Health and Human in consumer expenditure on cigarette and its report on the Global Tobacco Epidemic 2009:
Services (2004). The Health Consequences effects on employment: a case study of South implementing smoke-free environments.
of Smoking: A Report of the Surgeon General. Africa. Contemp Econ Policy, 17:412–422 Geneva, World Health Organization.
Atlanta, GA, U.S. Department of Health and doi:10.1111/j.1465-7287.1999.tb00693.x.
Human Services, Center for Disease Control Yach D (1996). Tobacco in Africa. World Health
and Prevention, National Center for Chronic Van Liemt G (2001). The world tobacco Forum, 17:29–36. PMID:8820139
Disease Prevention and Health Promotion, industry: trends and prospects.International
Office on Smoking and Health. Labour Office, Sectoral activities programme Yamasaki A , Chinami M, Suzuki M et al. (2005).
working paper. Tobacco and alcohol tax relationships with
U.S.Department of Health and Human suicide in Switzerland. Psychol Rep, 97:213–
Services (2006). The Health Consequences Van Walbeek CP (2002). The distributional 216. PMID:16279328
of Involuntary Exposure to Tobacco Smoke: impact of tobacco excise increases. S Afr J
A Report of the Surgeon General. Atlanta, Econ, 70:560–578. Yurekli A, Onder Z, Erk N et al. (2010). The
GA, U.S. Department of Health and Human economics of tobacco taxation: challenges and
Services, Center for Disease Control and Vargas MA, Campos RA (2005). Crop opportunities for a tobacco free Turkey. Paris,
Prevention, National Center for Chronic substitutionand diversification strategies: International Union Against Tuberculosis and
Disease Prevention and Health Promotion, empirical evidence from selected Brazilian Lung Disease.
Office on Smoking and Health. Municipalities. Washington DC, The
International Bank for Reconstruction Zhao X, Harris MN (2004). Demand for
van Baal PH, Brouwer WB, Hoogenveen Development, The World Bank. marijuana, alcohol and tobacco: participation,
RT, Feenstra TL (2007). Increasing tobacco levels of consumption and cross-equation.
taxes: a cheap tool to increase public health. Warner KE (2000). The economics of tobacco: Econ Rec, 80:394–410 doi:10.1111/j.1475-
Health Policy, 82:142–152.doi:10.1016/j. myths and realities. Tob Control, 9:78–89. 4932.2004.00197.x.
healthpol.2006.09.004 PMID:17050031 doi:10.1136/tc.9.1.78 PMID:10691761
van der Merwe R (1998a). Employment Warner KE, Fulton GA (1994). The economic
and output effects for Bangladesh following implications of tobacco product sales in
a decline in tobacco consumption. HNP a nontobacco state. JAMA, 271:771–776.
Discussion Paper Series, Economics of doi:10.1001/jama.271.10.771 PMID:8114214
Tobacco Discussion Paper. Washington DC,
The World Bank.
348
Chapter 10
Summary
Overview of tobacco taxation Industry pricing strategies

and other pricing policies
Worldwide, different types of taxes Specific excise taxes generally
apply to tobacco products, with result in higher tobacco product This chapter examines industry
different tax levels (rates) contributing prices. Specific excises can increase pricing strategies, pricing and price-
to significant price differentials. tobacco companies’ pricing power, related marketing strategies, as well
Historically, revenue generation has raise profits and increase market as industry efforts to influence all
been the primary aim of tobacco concentration. dimensions of tobacco tax policy,
taxation. However, the retention About 75% of the world’s tobacco including tax structure, tax levels, and
and the increase of excise taxes on product market is accounted for earmarking of tobacco tax revenues.
tobacco products increasingly aim by cigarettes. However, in some The impact of tobacco taxes on
to improve public health by reducing countries tobacco products other prices will vary according to how the
tobacco consumption and accounting than cigarettes have an important, industry reacts to the tax increase.
for the external cost of smoking. sometimes significant, market share. Early empirical evidence was mixed
Nonetheless, when determining Tax levels are often much lower in the extent to which taxes were
their taxation policy, governments on these products as compared passed through in prices. More recent
will take into account other, at to cigarettes. Differential rates are studies in the USA, South Africa and
times competing, political, social or sometimes applied within the same Jamaica found that tax increases led
economic considerations. These product category. These differences to price increases larger than the tax
considerations are often reflected result in price gaps and opportunities increase (overshifting).
in the applicable tax structure and for product substitution to lower- Tobacco companies employ a
rates. taxed products and brands. Applying variety of marketing techniques that
Moreover, tax levels are generally a similar tax level would reduce reduce prices on some tobacco
directly related to income levels by the incentive for substitution and products or brands. These efforts
country, with high-income countries increase the effectiveness of taxation are targeted at some specific
having high taxes and vice versa. policy in reducing tobacco use. populations, including young people,
Simultaneously high-income countries Given their important revenue- and have been used to soften the
tend to favour specific (per-unit) excise generating potential, part or all of impact of tobacco tax increases and
tax structures, while low- and middle- tobacco tax revenues have been other tobacco control interventions.
income countries rely more on ad used to fund health or tobacco Tobacco companies have lobbied
valorem (value-based) excise taxes. control activities in some countries. aggressively to influence tobacco
Some countries have designed more This practice can be adopted in tax levels and the earmarking of
complex taxation structures in an low-resource countries as a way to tobacco tax revenues. Evidence,
attempt to find a balance between strengthen weak health systems, almost entirely from North America,
budgetary, health and competition as proposed by the Taskforce on shows that industry lobbying has
objectives. Innovative International Financing for been successful, particularly at the
Health Systems. federal level. At the subnational level,
349
adequately funded tobacco control 7REDFFR FRPSDQLHV OREE\ are positive, with greater elasticity
efforts have been more successful in against tobacco tax increases and observed in low- and middle-income
overcoming industry opposition. The earmarking of tobacco tax revenues. countries. With the exception of the
industry has been more successful at 7REDFFR FRPSDQLHV WU\ WR United States, where income elasticity
the subnational level combating the influence the structure of tobacco has declined, there is no evidence
earmarking of tax revenues, arguing taxes to work in their interests and that income elasticity has changed
that the revenues will be diverted or against those of their competitors. in other countries. This implies that
misused. for a growing economy, the price of
The industry has also attempted Tax, price and aggregated cigarettes would have to increase at
to influence tobacco excise tax demand for tobacco the same rate as income to maintain
structures. Studies from the USA, the the affordability of cigarettes.
former Soviet Union, Hungary, China A large and growing body of empirical There is significant variation in
and Lebanon show that different literature, dominated by studies from the theoretical models employed
companies support different tax the USA and to a lesser extent the and the econometric methods
structures, favouring those that will United Kingdom, has found that used to estimate them. Given the
benefit their brands at the expense of tobacco consumption decreases development in economic models of
their competitors. when the price of tobacco increases. addiction, estimated specifications
Governments can influence The highly-influential World Bank should account for addiction where
tobacco product prices through publication Curbing the Epidemic the data allow. While it has become
tobacco taxes, price regulations, and states that the price elasticity of the norm to control for addiction in the
limits on price-related marketing. demand (the percentage change in model specifications, the use of state-
Some have implemented minimum consumption of a product that results of-the-art time series techniques
pricing policies where such policies from a 1% price increase) is around has not become prevalent. However,
are allowed under competition law;; í IRU GHYHORSHG FRXQWULHV DQG despite the variety of techniques
higher specific taxes can have similar EHWZHHQíDQGíIRUGHYHORSLQJ that have been used, estimates from
effects where minimum pricing countries. studies using different techniques are
policies are not allowed. Others have The evidence that has emerged similar.
included bans on price-reducing since the World Bank’s publication
marketing as part of a comprehensive suggests that, at least based on Conclusions
ban on industry marketing. WHO aggregate demand studies, the
FCTC Article 5.3, which aims to consensus price elasticity of around 6WXGLHV WKDW DUH EDVHG
limit tobacco industry influence in í LV VWLOO YDOLG IRU KLJKLQFRPH on aggregate data are useful
tobacco control policy-making, may countries, although the price in estimating price and income
provide a tool for governments to elasticity estimates for high-income elasticities of overall demand for
constrain tobacco company lobbying countries other than the USA and tobacco products.
efforts focusing on tax policy or other United Kingdom are somewhat more 7KHUH LV D QHJDWLYH
tobacco control policies. dispersed. The majority of price relationship between cigarette
elasticity estimates from aggregate prices and cigarette consumption
Conclusions demand studies for low- and middle- in countries at all levels of income.
LQFRPH FRXQWULHV OLH EHWZHHQ í While fewer studies have examined
0RVW WREDFFR SURGXFW DQG í +RZHYHU PDQ\ OLH RXWVLGH the demand for other tobacco
markets are highly concentrated. this range (especially on the inelastic products, those that have done so
In these markets, recent empirical side). have reported a similar negative
evidence indicates that tobacco Income is a significant determinant relationship.
taxes are generally overshifted. of tobacco product demand;; as 3ULFHHODVWLFLWLHVRIFLJDUHWWH
7REDFFR FRPSDQLHV XVH aggregate income increases, the demand in high-income countries
price-reducing marketing techniques aggregate demand for tobacco DYHUDJH í 0RVW HVWLPDWHV IURP
to counteract the impact of tobacco increases. Most estimated income the USA and the United Kingdom fall
excise tax increases and other elasticities (the percentage change in a relatively narrow range between
tobacco control policies. in consumption of a product that íDQGíZKLOHWKHUHLVJUHDWHU
results from a 1% income increase) variation in high-income countries
350
Summary
other than the USA and the United Empirical methods for estimating done in Canada;; Taiwan, China;; Italy;;
Kingdom. tobacco demand using individual or the Republic of Korea;; Spain;; and
7KHUHLVJUHDWHUYDULDWLRQLQ household data have become more the United Kingdom. These studies
price elasticity estimates in studies sophisticated over time, and greater generally find that cigarette smoking
from low- and middle-income computing power has allowed the prevalence and intensity are inversely
countries, likely due to differences application of these methods to associated with cigarette prices, with
in affordability and availability of increasingly large survey data sets, the size of the association varying
different tobacco products across including repeated cross-sectional across countries. Estimates from
and within countries. Most estimates and longitudinal survey data sets. these studies are generally in the
indicate that demand in low- and Most studies based on survey range produced by the aggregate
middle-income countries is less data come from the United States, demand studies from high-income
price-inelastic than in high-income because of its extensive subnational countries other than the USA, as
countries, as most estimates lie and temporal variation in taxes and discussed in Chapter 4. Studies that
EHWZHHQíDQGí prices. These studies find that adult consider gender differences produce
0RVW LQFRPH HODVWLFLWHV smoking prevalence, frequency and contrasting conclusions about the
are positive and lie between 0 and intensity are negatively related to relative price elasticities of men and
1 in both high-income and low-and- cigarette taxes and prices. As with women, depending on the country
middle-income countries. There the elasticity estimates obtained for which the study was done. A few
is evidence of declining income from aggregate data for the USA, studies from high-income countres
elasticities over time in the USA. most of the total elasticity estimates have concluded that higher prices
$IIRUGDELOLW\ KDV EHFRPH from the studies based on US survey increase cessation.
an increasingly recognized measure, GDWDIDOOZLWKLQWKHUDQJHIURPíWR There exists a larger number
especially in rapidly growing low- and íZLWKURXJKO\KDOIRIWKHLPSDFW of survey-based studies of adult
middle-income countries where the of price on adult prevalence and the tobacco demand in low- and
impact of price increases might be remainder on the number of cigarettes middle-income countries—including
offset by growth in income. consumed by adult smokers. Several Bulgaria, China, Estonia, Egypt,
studies have examined adult smoking India, Indonesia, Mexico, Myanmar,
Tax, price and adult tobacco use cessation, generally finding that Nepal, the Russian Federation,
higher taxes and prices reduce the South Africa, Sri Lanka, Thailand,
A large and growing number of duration of smoking, raise interest Turkey, and Viet Nam—with many of
studies have used individual-level in quitting, boost quit attempts, and these studies based on household
or household-level survey data to increase the number of smokers who expenditure survey data. The
assess the impact of tobacco product successfully quit smoking. Finally, a sophistication of the methodological
taxes and prices on adult tobacco few US studies found similar effects approach used in these studies varies
use. Studies have used survey data of tax and price on the use of other considerably. Several studies use
to examine the differential impact of tobacco products such as smokeless self-reported prices, which almost
tax and price on tobacco use among tobacco and cigars, and produced certainly produce biased estimates
population subgroups defined by some evidence of substitution that suggest greater price elasticity.
gender, age, socioeconomic status among tobacco products in response Some studies use composite tobacco
and/or other characteristics, as well to changes in the relative prices of price measures rather than product-
as to assess the separate effects these products. specific measures of price. The range
of price on different aspects of While several studies employing of price elasticities for adult tobacco
tobacco use, including prevalence, survey data on adult tobacco use use produced by these studies
frequency, intensity, initiation, uptake have been conducted in other high- varies considerably. Despite these
and cessation. This chapter reviews income countries, relatively few of limitations, these studies generally
studies that examine the effects these have estimated the effects confirm that various aspects of adult
of price on adult tobacco use, with of tax and price on adult tobacco tobacco use are responsive to price,
subsequent chapters (Chapters 6 use in these countries, largely due with higher prices reducing adult
and 7) reviewing studies on tobacco to the limited price variation within prevalence and intensity of use.
use among young people and most countries. Studies on the price No clear patterns emerge from
different socioeconomic groups. elasticity of adult smoking have been the small number of studies from
351
countries other than the USA that estimates of price elasticity consistent cross-sectional or longitudinal data,
consider substitution among tobacco with those found in aggregate consistently find that higher prices
products in response to changes in demand studies for these countries, lead to smoking cessation among
the relative prices of these products. with total elasticity estimates in the young people. Similar studies find
The same is true for studies that UDQJHIURPíWRí that higher prices deter progression
examine differences in price elasticity Studies from a growing number into later stages of smoking uptake.
by gender. of low- and middle-income countries
indicate that adult tobacco use Conclusions
Conclusions in these countries is inversely
related to tobacco product prices. $V WREDFFR SULFH DQG WD[
A relatively large body of literature on Methodological differences and other increases, smoking prevalence
adult tobacco demand based on US factors contribute to the considerable among young people falls, and vice
survey data shows that: variation in the price elasticity versa.
$GXOW VPRNLQJ SUHYDOHQFH estimates produced by these studies. $V WREDFFR SULFH DQG WD[
is inversely related to price, with increases, smoking intensity among
prevalence falling as prices rise and Tax, price and tobacco use young people falls, and vice versa.
vice versa;; among young people 0RVW ORQJLWXGLQDO VWXGLHV
,QWHQVLW\RIVPRNLQJDPRQJ from high-income countries find that
adults is inversely related to price, Economic theory predicts that smoking initiation is inversely related
with cigarette consumption among tobacco use among young people to price.
smokers falling as prices rise and would be more responsive to price (VWLPDWHV IURP VWXGLHV RQ
vice versa;; than would adult tobacco use. To a tobacco use among young people
7KH SULFH HODVWLFLW\ RI DGXOW great extent, the empirical evidence indicate that their tobacco use is
VPRNLQJLVLQWKHUDQJHIURPíWR supports this. Many econometric more responsive to price than is
íZLWKDERXWKDOIRIWKHHIIHFWRI studies from countries at all income tobacco use among adults.
price on smoking prevalence and the levels find that smoking prevalence The same conclusions
remainder on intensity of smoking and intensity among young people generally apply in low- and middle-
among adult smokers;; and decreases as cigarette price income countries but are based on
&LJDUHWWH SULFHV DIIHFW increases. The estimated overall fewer studies.
cessation-related outcomes, with price elasticity of demand for young $ IHZ VWXGLHV QHDUO\ DOO
higher prices increasing interest in people in most high-income country from the USA, find:
quitting, quit attempts, and successful VWXGLHV UDQJHV EHWZHHQ í DQG - Cigarette prices influence different
cessation. í KRZHYHU WKH GLVSHUVLRQ RI stages of uptake of cigarette use,
A few studies from the USA price elasticity estimates for young with a relatively larger impact at later
indicate that: people in low- and middle-income stages of uptake.
+LJKHU FLJDUHWWH SULFHV countries is greater. While the studies - Price has both a direct effect and
increase the likelihood that smokers examining the impact of price and tax an indirect effect, through peer and
will smoke only on some days rather on other tobacco products are fewer family influences, on tobacco use
than every day;; in number than those for cigarettes, among young people.
&KDQJHV LQ RWKHU WREDFFR a similar relationship exists for other - As cigarette prices increase,
product taxes and prices inversely tobacco products. smoking cessation among young
affect the use of these products;; and Studies of smoking initiation that people increases.
7KHUH LV VXEVWLWXWLRQ use cross-sectional data produce
between cigarettes and smokeless mixed results primarily because of the Tax, price and tobacco use
tobacco products in response to measurement error inherent in using among the poor
changes in the relative prices of retrospective measures of initiation.
these products. Most studies that use longitudinal In most countries, tobacco use is
A small number of studies from data find that higher prices of inversely related to indicators of
other high-income countries show cigarettes significantly decrease socioeconomic status, with the
that adult smoking in these countries youth smoking initiation and vice possible exception of some upper
responds to changes in price, with versa. Studies from the USA, using middle-income countries. Spending
352
Summary
on tobacco diverts spending from Tax avoidance and tax evasion challenges that researchers have
other goods and services, and this attempted to overcome via innovative
tends to be greater for lower-income Tax avoidance refers to legal methodologies. No single method
households. There is a two-way methods of circumventing tobacco addresses all of these difficulties,
relationship between tobacco use taxes, and tax evasion refers to illegal and each method is inherently
and poverty. methods of circumventing tobacco limited. A combination of methods is
Current excise taxes on tobacco taxes. Illicit trade includes both needed to fully assess the extent of
are generally regressive in high- legally produced products illegally these measurement challenges.
income countries;; that is, the amount traded across borders (smuggling) Studies show that tax avoidance
of tax as a percentage of income and illegally manufactured products. and tax evasion reduce but do
increases as income decreases. Most activities that comprise tax not eliminate the effectiveness of
Few studies have examined the avoidance include the payment tobacco tax increases in reducing
regressivity of tobacco taxes in low- of some tobacco taxes. In most tobacco use and raising revenues.
and middle-income countries, finding cases, tax avoidance is practiced Tobacco tax avoidance and evasion
mixed evidence of regressivity of by individuals. Tax evasion includes undermine other tobacco control
tobacco taxes, in part because of the both small and large quantities of measures. Tax evasion adds to health
structure of these taxes. tobacco products, and usually no disparities between socioeconomic
The price responsiveness of taxes are paid. These activities may groups.
tobacco demand is generally found to involve criminal networks, tobacco Research on the effectiveness
be higher among the poor than among companies, or other large-scale of strategies aimed at reducing tax
the rich in high-income countries, operations. avoidance and evasion is limited
based largely on evidence from the Several studies show that tax and and difficult. Tax evasion can be
USA and the United Kingdom. Given price differentials play an important reduced via a combination of various
this evidence, increases in tobacco role in individual tax avoidance and strategies such as international
taxes can reduce the regressivity of small-scale smuggling operations. cooperation, legislative measures to
these taxes. Other important factors include the control the supply chain, increased
Evidence from other countries extent of tourist travel, access to enforcement and strong penalties, as
does not suggest an obvious duty-free and other tax-free outlets, shown by evidence largely from case
relationship between price and technology allowing virtual studies. The modus operandi for the
responsiveness and socioeconomic transactions such as internet sales. supply of illegal tobacco products
status. This may be due to differences The determinants for large-scale may change over time as illicit traders
in the extent of opportunities for illicit trade are far more complex. adapt their practises in response to
tax avoidance and evasion across The greater the reward and the lower government actions.
countries, tax structures and the the costs of supplying these illicit To reduce cross-border shopping,
resulting price gaps, and other products, the greater the probability small-scale smuggling, and some
factors. Given this mixed evidence, that this activity will occur. Research types of tax avoidance, effective
the equity implications of tobacco tax shows that these determinants strategies include coordination of
increases in low- and middle-income include corruption, informal tobacco tax levels and increases
countries is unclear. distribution networks, the presence of across jurisdictions, and elimination
organized crime, the extent of cross- of duty-free shopping.
Conclusions border trade and strength of border
controls, insufficient penalties, weak Conclusions
The price responsiveness of tobacco tax administration and enforcement,
demand is generally higher among differential treatment of domestically- 7KH SULPDU\ GHWHUPLQDQW
the poor than the rich in high-income produced and imported products, the of tax avoidance and small-scale
countries, based largely on evidence pre-tax price of tobacco products, smuggling is differentials in taxes
from the USA and the United Kingdom. and the strategies of tobacco and prices across jurisdictions.
Evidence from other countries companies, as well as price and tax 7KHUH DUH PXOWLSOH
on the relationship between price levels. determinants of tax evasion. These
responsiveness and socioeconomic Comprehensively measuring tax include the level of corruption,
status is mixed. avoidance and tax evasion presents informal distribution networks,
353
the presence of organized crime, 7R UHGXFH FURVVERUGHU Tobacco tax increases generally
the extent of cross-border trade shopping, small-scale smuggling, increase tax revenues. Some
and strength of border controls, and some types of tax avoidance, governments have earmarked
insufficient penalties, weak tax effective strategies include tax revenues for tobacco control
administration and enforcement, coordination of tobacco tax levels and programmes and/or other health
differential treatment of domestically- increases across jurisdictions, and promotion efforts. Tax-funded
produced and imported products, the elimination of duty-free shopping. tobacco control programmes further
pre-tax price of tobacco products, reduce tobacco use and its health
and the strategies of tobacco Economics and health impact of and economic consequences.
companies, as well as price and tax tobacco taxation Most economic studies on the
levels. employment impact of reduced
7KH PDMRULW\ RI H[LVWLQJ Tobacco taxes can be expected tobacco consumption have found
studies capture both tax evasion and to have a large impact on society that reductions in tobacco-dependent
tax avoidance to some extent. The through several different channels. employment are more than offset by
limited empirical evidence indicates A large body of literature examines increases in employment elsewhere.
that tax evasion is much larger in the effect of smoking on health Finally, increased tobacco taxes
scale than tax avoidance. care expenditures and health have been shown to have minimal if
&LJDUHWWH WD[ HYDVLRQ LV outcomes. These studies find any effect on general price inflation.
higher in countries that have lower that reducing smoking through
cigarette prices and taxes. increased taxes increases the life Conclusions
'DWDLQGLFDWHWKDWWKHVL]HRI expectancy of smokers by reducing
this illicit market is inversely related cancers (especially lung cancer), ,QFUHDVLQJ WREDFFR H[FLVH
to a country’s income. respiratory diseases and cerebral taxes is a cost-effective way to
7KH WKUHDW RI VPXJJOLQJ and cardiovascular diseases. The improve population health and
discourages governments from reduction in smoking that results reduce medical costs.
raising tobacco taxes or motivates from higher taxes also reduces (PSLULFDO HYLGHQFH IURP
tobacco tax reductions. the medical costs attributable to around the world shows that
6WXGLHVGHPRQVWUDWHWKDWDQ smoking. Further health gains can increasing tobacco taxes will
increase in cigarette taxes will reduce be expected through reductions generally increase government
cigarette consumption and increase in secondhand smoke exposure revenue.
cigarette tax revenues even in the and reduced maternal smoking. In 0RVW VWXGLHV RQ WKH
presence of illicit cigarette trade. addition, reductions in tobacco use employment impact of reductions
(OLPLQDWLQJ WKH VXSSO\ RI through price increases have been in tobacco use find that reductions
illicit tobacco would result in larger associated with reduced alcohol and in tobacco-dependent employment
than average reductions in smoking illicit drug use, although the evidence are more than offset by increases in
among young people and the poor. is mixed. The effects on obesity employment elsewhere.
,OOLFLW WUDGH LQ WREDFFR of reduced smoking resulting from (DUPDUNLQJRIWREDFFRWD[HV
products undermines the impact of higher taxes are less clear. Given for tobacco control programmes
other tobacco control policies such the low costs of administration, tax has been shown to be an effective
as youth access laws and health increases are a highly cost-effective policy instrument to achieve tobacco
warnings. intervention for improving health. control.
7R UHGXFH WREDFFR WD[ ,QFUHDVHG WREDFFR WD[HV
evasion, a multifaceted and flexible have been shown to have minimal, if
approach is most effective. any, effect on general price inflation.
354
Chapter 11
Evaluation
The Working Group evaluated the or confounding have not been ruled and a given effect. Alternatively, this
strength of the evidence for drawing out with reasonable confidence. category is used when no studies are
the conclusions shown in the However, explanations other than available.
accompanying table using categories causality are unlikely.
defined as follows: (YLGHQFH RI QR HIIHFW
/LPLWHG HYLGHQFH There Methodologically sound studies
6XIILFLHQW HYLGHQFH An is some evidence of association consistently demonstrate the lack
association has been observed between the intervention under of an association between the
between the intervention under consideration and a given effect, but intervention under consideration and
consideration and a given effect alternative explanations are possible. a given effect.
in studies in which chance, bias
and confounding can be ruled out ,QDGHTXDWHQR HYLGHQFH
with reasonable confidence. The There are no available
association is highly likely to be methodologically sound studies
causal. showing an association;; the available
studies are of insufficient quality,
6WURQJ HYLGHQFH There consistency or statistical power to
is consistent evidence of an permit a conclusion regarding the
association, but evidence of causality presence or absence of a causal
is limited by the fact that chance, bias association between the intervention
355
Sufficient Strong Limited Evidence of Inadequate /

Concluding Statements
Evidence Evidence Evidence No Effect No Evidence
Increases in tobacco excise taxes that increase

1 X
prices result in a decline in overall tobacco use.
2 prices reduce the prevalence of adult tobacco X
use.
3 X
prices induce current tobacco users to quit.
4 prices lower the consumption of tobacco products X
among continuing users.
5 prices reduce the prevalence of tobacco use X
among young people.
prices reduce the initiation and uptake of tobacco
6 X
use among young people, with a greater impact
on the transition to regular use.
Tobacco use among young people responds more
7 to changes in tobacco product taxes and prices X
than does tobacco use among adults.
The demand for tobacco products in lower-
income countries is more responsive to price than
8 X
is the demand for tobacco products in higher-
income countries.
In high-income countries, tobacco use among
lower-income populations is more responsive
9 X
to tax and price increases than is tobacco use
among higher-income populations.
In low- and middle-income countries, tobacco
use among lower-income populations is more
10 X
responsive to tax and price increases than is
tobacco use among higher-income populations.
Changes in the relative prices of tobacco products
11 lead to some substitution to the products for X
which the relative prices have fallen.
Tobacco tax increases increase tobacco tax
12 X
revenues.
Tobacco tax increases that increase prices
13 X
improve population health.
Tobacco tax increases do not increase
14 X
unemployment.
Tax avoidance and tax evasion reduce, but do not
15 eliminate, the public health and revenue impact of X
tobacco tax increases.
A coordinated set of interventions that includes
international collaborations, strengthened tax
16 administration, increased enforcement, and swift, X
severe penalties reduces illicit trade in tobacco
products.
Higher and more uniform specific tobacco excise
taxes result in higher tobacco product prices and
17 X
increase the effectiveness of taxation policies in
reducing tobacco use.
Tobacco industry price discounting strategies,
price-reducing marketing activities, and lobbying
18 efforts mitigate the impact of tobacco excise tax X
increases.
356
Chapter 12
Recommendations
Based on the body of evidence Including measures of these *LYHQ DGYDQFHV LQ WLPH
examined in this volume, the Working techniques in surveillance systems is series research and economic
Group made the following research necessary for such research. modelling of addiction, future time
recommendations: 0RVW UHVHDUFK RQ LQGXVWU\ series analyses should apply state-
efforts to prevent large tax increases of-the-art methods as the data allow.
Overview of tobacco taxation or earmarking of taxes and to reduce )XWXUH UHVHDUFK VKRXOG
current taxes focuses on North study how price elasticities vary
5HVHDUFK LV QHHGHG WR America. Further research on these over time and at different price
assess the impact of complex tax issues in other countries is required. levels, for different magnitudes of
structures on prices, profitability, 5HVHDUFKRQLQGXVWU\HIIRUWV price increases and under different
market concentration and public to influence tax structures is limited tax structures to provide a better
health. to a small number of countries. More understanding of the shape of the
5HVHDUFK LV QHHGHG WR research on these efforts in other demand curve and hence the design
identify a comparable taxable base countries is needed. of taxation policies.
for the wide range of smoking and $OPRVW QR HPSLULFDO 6WXGLHVWKDWSRRODJJUHJDWH
smokeless tobacco products. evidence exists on the impact of data from multiple countries over time
minimum tobacco pricing policies or should investigate the homogeneity
Industry pricing strategies and on bans on tobacco company price- of the countries and consider the
other pricing policies related marketing efforts. Research is use of heterogeneous estimation
needed to evaluate the effectiveness techniques.
(PSLULFDO VWXGLHV RQ of these policies.
industry pricing are limited in number &RQWH[WVSHFLILF HYLGHQFH Tax, price and adult tobacco use
and geographic coverage, and have documenting the effective
focused exclusively on cigarettes. implementation of Article 5.3 of the 5HVHDUFK RQ KRZ SULFH
Research is needed to assess the WHO FCTC and research evidence elasticity of adult tobacco use
impact of tobacco tax structure on addressing industry arguments changes over time, varies with price
prices and industry short- and long- aimed at influencing tobacco tax levels, and differs for large versus
term pricing strategies, particularly policy are needed. small price changes is needed in all
for products other than cigarettes. countries.
Surveillance systems that include Tax, price and aggregated 5HVHDUFK LV QHHGHG RQ WKH
brand-specific data on tobacco demand for tobacco effects of tax and price on various
product prices and price-reducing aspects of adult tobacco use, including
marketing are needed for such ,Q FRXQWULHV ZLWK OLPLWHG prevalence, frequency, intensity, and
research. research on price elasticity, a well cessation, as well as on differences in
0RUH UHVHDUFK LV QHHGHG conducted time series study can the impact of tax and price on male
on tobacco companies’ use of provide a major contribution. and female tobacco use.
price-related marketing techniques.
357
0RUH UHVHDUFK LV QHHGHG 5HVHDUFK LV DOVR QHHGHG 7KHUH LV D QHHG IRU IXUWKHU
on the effect of tax structures and to understand the choices between study on the impact of taxation of
differences in relative prices on licit and illicit products, particularly other tobacco products on health
substitution among tobacco products. among those of lower socioeconomic outcomes and costs.
*OREDO DGXOW WREDFFR status. 7KHUH LV D QHHG IRU IXUWKHU
surveillance systems such as those study on the impact of taxation
described in IARC Handbooks of Tax avoidance and tax evasion of smoked tobacco products on
Tobacco Control, Volume 12 should secondhand smoke exposure (SHS)
regularly collect data on adult 5HVHDUFKRQWD[DYRLGDQFH and its related health outcomes and
tobacco use in more countries. and tax evasion must apply costs.
clear definitions for the type of 7KHUH LV D QHHG IRU IXUWKHU
Tax, price and tobacco use circumvention being addressed. studies on the impact of tobacco
among young people $OO VWXGLHV PHDVXULQJ taxation on other unhealthy
tobacco tax evasion and tax behaviours.
0RUH VWXGLHV IURP ORZ avoidance should clearly describe
and middle-income countries are the type of data and methods used, Proposed policy
necessary to explore price effects as well as their limitations. recommendations
on tobacco demand, initiation, ,QIRUPDWLRQ VWDWLVWLFV
cessation, uptake and substitution reports and research findings 1. To improve public health by
between tobacco products. on tobacco tax evasion and tax reducing tobacco use, governments
0RUHUHVHDUFKLVQHHGHGRQ avoidance from government should adopt relatively simple
how price influences the availability agencies should be made publicly tobacco excise tax structures that
of tobacco through peers and family, available. emphasize specific taxes and that
and how this affects tobacco use 0RUHUHVHDUFKLVQHHGHGWR include regular tax increases that
among young people. estimate the size and the scale of the outpace growth in general price
illicit tobacco trade and to examine levels and incomes.
Tax, price and tobacco use the effectiveness of the measures 2. To further reduce tobacco
among the poor taken to address this problem. use, tobacco tax revenues should be
Particular attention needs to be paid used to fund comprehensive tobacco
0RUH UHVHDUFK LV QHHGHG to low- and middle-income countries. control programs and other health
to examine the price elasticity of 5HVHDUFKLVQHHGHGLQWRWKH promotion activities.
tobacco use among individuals from extent of tax avoidance, its impact 3. A multinational surveillance
different socioeconomic groups, on tobacco use and measures to and monitoring system should be
and the implications of this for the address this issue. implemented that regularly collects
impact of tax increases on equity. data on tobacco use among adults
These studies should examine the Economics and health impact and young people, tobacco product
price elasticity of demand for the full of tobacco taxation taxes and prices, price-reducing
range of tobacco products, as well as marketing and lobbying efforts of
across price tiers for a given product, 7KHUH LV D QHHG IRU VWXGLHV tobacco companies, tax avoidance
particularly for those used by people on the impact of cigarette taxation and evasion, and tax administration
in low-income strata. on health care spending attributable and enforcement activities.
to smoking, particularly for countries
other than the USA.
358
Working procedures
Starting in 2006, the series of given hazard and disease outcome, The Handbooks on tobacco
International Agency for Research are based on the meaning of the term control will evaluate the strength of
on Cancer (IARC) Handbooks of “exposure” implying increased risk to the available evidence on the effects
Cancer Prevention added tobacco an undesired health effect. However, of interventions intended to prevent
control as a new area of prevention in this series of Handbooks, dedicated or reduce tobacco use, tobacco
for their reviews. When appropriate, to the evaluation of the preventive supply, and, when possible, tobacco-
in addition to cancer, other health effects of compounds, biological associated morbidity and mortality.
outcomes preventable by avoiding or pharmaceutical products, The aim of the Handbook series is
tobacco use or exposure to behaviours, programmes, and policy to provide the scientific community,
secondhand smoke (SHS) may interventions, the traditional meaning policymakers, and governing bodies
be included when evaluating the of the term “exposure” is unfitting. of IARC member states, as well as
effectiveness of interventions in a Therefore in several instances the other countries with evidence-based
Handbook. term “intervention,” which lacks a assessments of these interventions at
The Working Procedures hazardous connotation, is preferred. the individual and population levels,
described herein are largely Examples of interventions with with the ultimate goal of assisting in
taken from the Handbooks of expected benefits in the area of the global implementation of tobacco
Cancer Prevention devoted to tobacco control are tobacco use control provisions within national and
Chemoprevention and Screening, cessation, banning of smoking international programmes aimed at
and from the IARC Monograph in public places, and taxation on reducing tobacco-related morbidity
Preamble (updated in January 2006). tobacco products. and mortality.
The text that follows is
organised in two principal parts. Part one: General principles Objectives
The first addresses the general
scope, objectives, and structure General scope The objective is to prepare and
of the Handbooks with emphasis publish, in the form of Handbooks,
on tobacco control. The second The prevention and control of cancer critical reviews and consensus
describes the scientific procedures are the strategic objectives of IARC. evaluations of evidence on the
for evaluating cancer-preventing Cancer prevention may be achieved effects of interventions focusing
agents and tobacco control policy at the individual level by avoiding on tobacco control, with the
interventions. cancer-causing agents (e.g. not help of an internationally formed
The term “exposure” appears using tobacco products), and at Working Group (WG) of experts.
repeatedly in these procedures, the population level by adopting The Handbooks may also indicate
borrowed from the IARC Monographs programmes or legislation to reduce where additional research efforts
devoted to the evaluation of or eliminate exposure to cancer- are needed, specifically when data
carcinogenicity. Epidemiological causing agents (e.g. removing immediately relevant to an evaluation
studies conducted to assess the exposure to SHS through banning are not available. The evaluations
association between exposure to a smoking in public and workplaces). in the Handbooks are scientific
359
and qualitative judgments of peer- reviews and evaluations that are participate in the evaluations.
reviewed, published data, conducted developed during the meeting. 4. Observers with relevant
during a six-day meeting of peer WG members are selected based scientific credentials may be
review and discussions by the WG. on knowledge and experience admitted to a meeting by IARC
pertinent to the topic evaluated in limited numbers. Priority will
Topic for the Handbook and absence of real or apparent be given to achieving a balance
conflicts of interest. The tasks of of Observers from constituencies
The topic to be evaluated in a the WG are: (i) to ascertain that with differing perspectives. They
Handbook is selected approximately all appropriate data have been are invited to observe the meeting
12 months prior to the meeting by the collected;; (ii) to select the data and should not attempt to influence
head of the Lifestyle, Environment, relevant for the evaluation on it. Observers serve as sources
and Cancer Group, after consultation the basis of scientific merit;; (iii) of first-hand information from
with IARC scientists involved in to prepare accurate summaries the meeting to their sponsoring
tobacco research. A Handbook may of the data to enable the reader organisations. They can play
cover a single topic or a group of to follow the reasoning of the a valuable role in ensuring
related topics in the area of tobacco WG;; (iv) to critically evaluate that all published information
control. the results of epidemiological, and scientific perspectives are
clinical, and other type of studies;; considered. Observers will
Meeting participants (v) to prepare recommendations not serve as meeting chair or
for research and for public health subgroup chair, draft any part of
Soon after the topic of a Handbook is action;; and (vi) if the topic being a Handbook, or participate in the
chosen, international scientists with reviewed so permits, to make an evaluations. At the meeting, the
relevant expertise are identified by overall evaluation of the evidence chair and subgroup chairs may
IARC staff (usually through literature of a protective effect or reduced grant Observers the opportunity
searches), in consultation with other risk associated with the exposure to speak, generally after they have
experts. Each participant serves as or intervention focus of the heard a discussion.
an independent scientist and not as evaluation. 5. The IARC Secretariat consists
a representative of any organisation, 2. Invited Specialists are experts of scientists who have relevant
government, or industry. Every who also have critical knowledge expertise and are designated
effort is made to achieve a balanced and experience, but have a real by IARC to attend a meeting.
group of experts in terms of gender, or apparent conflict of interest. They serve as rapporteurs and
geographic origin, expertise, and These experts are invited, when participate in all discussions.
diversity of scientific opinion. necessary, to assist in the WG When requested by the meeting
Five categories of participants by contributing their unique chair or subgroup chair, they may
may attend Handbook meetings: knowledge and experience also draft text or prepare tables
WG members, Invited Specialists, during subgroup and plenary and analyses.
Representatives of national and discussions. They may also The WHO Declaration of Interest
international health agencies, contribute text on the intervention form is sent to each prospective
Observers, and the IARC Secretariat. being evaluated. Invited participant at the first contact, with
Participants in the first two groups Specialists do not serve as the preliminary letter presenting the
generally have published significant meeting chair or subgroup chair, Handbook meeting. Before an official
research related to the topic being redact summaries, or participate invitation is extended, each potential
reviewed or in tobacco control in in the evaluations. participant, including the IARC
particular. All participants are listed, 3. Representatives of national Secretariat, completes the WHO
with their addresses and principal and international health agencies Declaration of Interest form to report
affiliations, at the beginning of each may attend meetings because financial interests, employment
Handbook volume. A description their agencies are interested and consulting, and individual and
of each participant type, and their in the topic of a Handbook. institutional research support related
responsibilities, is listed below. Representatives do not serve as to the topic of the meeting. IARC
1. The Working Group is meeting chair or subgroup chair, assesses the declared interests to
responsible for the critical draft any part of a Handbook, or determine whether there is a conflict
360
Working procedures
that warrants some limitation on and interpretation of the results, and request by any of the members of the
participation. WG members are limitations identified by the WG are WG, pertinent studies are identified
selected based on the absence of clearly outlined in square brackets by IARC from recognised sources
real or apparent conflicts of interest. (i.e. [ ]). The reasons for not giving of information, such as PubMed,
If a real or apparent conflict of interest further consideration to an individual and made available to WG members
is identified, then the expert is asked study are also indicated in square and Invited Specialists to prepare
to attend as an Invited Specialist. brackets. Important aspects of the working papers for the meeting.
The declarations are updated and a study, directly impinging on its Paper and/or electronic version
reviewed again at the opening of interpretation, are brought to the of the studies cited in the working
the meeting, approximately eight attention of the reader. In general, papers are available at the time of the
months later. Interests related to the numerical findings are indicated as meeting for consultation by the WG
subject of the meeting are disclosed they appear in the original report;; and all other participants including
to the meeting participants and in the units are converted when necessary the IARC Secretariat.
published volume (Cogliano et al., for easier comparison. The WG
2004). may conduct additional analyses of Working papers
the published data and use them in
Data for the Handbooks their assessment of the evidence. Working papers are due about six to
These analyses and their results eight months after original contact of
The Handbooks review all pertinent are outlined in square brackets or in invited experts. The first version of
studies on the intervention to be italics in the Handbook. the working papers is compiled and
evaluated. Only those data considered formatted by IARC staff about two
relevant to evaluate the evidence are Working procedures months prior to the meeting, or as
included and summarised. Those soon as they are received, and made
judged inadequate or irrelevant to Chair of the meeting available ahead of time through
the evaluation may be cited but not IARC’s internet to all WG members,
summarised. If a group of similar The chair of the Handbook meeting Invited Specialists, and the IARC
studies is not reviewed, the reasons is identified among leading Secretariat. Reception of working
are indicated. international experts soon after the papers ahead of the established
With regard to reports of basic topic of a Handbook is chosen. The deadline is encouraged, as it allows
scientific research, epidemiological chair will help develop an outline review of their content, facilitating
studies, and clinical trials, only for the Handbook early on and aid identification of information gaps
studies that have been published in identifying prospective experts to from the start. When possible, or
or accepted for publication in the form the WG. The chair participates when deemed necessary, working
openly available scientific literature on conference calls with WG papers may be discussed early
are reviewed. In certain instances, members and Invited Specialists in on among experts to expedite the
government agency reports that preparing for the meeting, provides review process to be accomplished
have undergone peer review and are early feedback on working papers, during the meeting. Conference calls
widely available can be considered. directs the meeting, and helps will be scheduled after reception of
Exceptions may be made ad hoc resolve queries emerging on the all working papers and prior to the
to include unpublished reports that working papers once the meeting is meeting, with the aim of identifying
are in their final form and publicly over. areas deserving additional work by
available, if their inclusion is experts before the meeting.
considered pertinent to making an Literature to be reviewed Acknowledgement of significant
evaluation. Abstracts from scientific contributions to the chapters by
meetings, and other reports that do After the topic of the Handbook is colleagues of the invited experts,
not provide sufficient detail upon chosen and invited experts have either at their home institution
which to base an assessment of their accepted to take part in the Handbook or elsewhere, can be included
quality are generally not considered. meeting, participants are expected in the Handbook under an
Inclusion of a study does not to conduct their own literature acknowledgement paragraph to be
imply acceptance of the adequacy searches in support of their working shown following the listing of the
of the study design or of the analysis assignments. Only upon expressed meeting participants.
361
Meeting (including adequate description of defined by the authors. Cases of

the intervention and follow-up), data disease in the study population should
The meeting participants convene analysis, and ultimately, deciding if have been identified independently
at IARC for six days to discuss and chance, bias, confounding, or lack of the intervention of interest, and the
finalise the texts of the working papers of statistical power may account for intervention assessed in a way that
that will constitute the Handbook and the observed results. The experts was not related to disease status.
to formulate the evaluations. The WG will ascertain how the limitations of Second, in the study design and
members and Invited Specialists are the studies affect the results and analysis, the authors should have
grouped into subgroups according to conclusions reported. The criteria taken into account other variables that
their area of expertise. Subgroups that follow apply to epidemiological can influence the risk of disease or
meet during the first three to four and clinical studies, and therefore impact of an intervention and that may
days to review in detail the first may not be as relevant to studies have been related to the intervention
versions of their working papers, where other quality criteria would of interest. Potential confounding by
develop a joint subgroup draft, and be indicated (e.g. those assessing such variables should have been
write summaries. During the last few the impact of economic policies dealt with either in the design of the
days the participants meet in plenary or when health outcomes are not study, such as by matching, or in the
session to review the subgroup contemplated). analysis, by statistical adjustment.
working papers and summaries In cohort studies, comparisons with
and to develop the consensus Quality of studies considered local rates of the disease may or
evaluations. Scheduling of plenary may not be more appropriate than
and subgroup time may change from It is necessary to take into account the those with national rates. Internal
one Handbook meeting to another. possible roles of bias, confounding, comparisons of disease frequency
and chance in the interpretation among individuals at different levels
Post-meeting of epidemiological studies. Bias of the intervention are also desirable
is the operation of factors in the in cohort studies, since they minimise
After the meeting, the draft Handbook study design or execution that the potential for confounding related
is verified by consulting the original leads erroneously to a stronger or to difference in risk factors between
literature, edited, and prepared for weaker association than in fact exists an external reference group and the
publication by IARC staff. The aim between the exposure/intervention study population.
is to publish Handbooks within 12-15 being evaluated and the outcome. Third, the authors should have
months of the meeting. If applicable, Confounding is a form of bias that reported the basic data on which
summaries reporting the results of occurs when the association with the the conclusions are founded, even
the evaluation may be available on disease is made to appear stronger if sophisticated statistical analyses
the IARC website (http://www.iarc. or weaker than it truly is, as a result of were employed. The numbers of
fr) soon after the meeting, and a an association between the apparent exposed and unexposed cases and
short report may be published in the causal factor and another factor that controls in a case-control study,
international literature. is associated with either an increase and the numbers of cases observed
or decrease in the incidence of and expected in a cohort study
Part two: Scientific review the disease. The role of chance is should have been provided. Further
of the evidence and evaluation related to biological variability and tabulations by time since exposure
the influence of sample size on the began, and other temporal factors,
Scientific review precision of estimates of effect. are also important. In a cohort study,
In evaluating the extent to which data on all cancer sites and all causes
The evidence forming the foundation these factors have been taken into of death should have been given to
of the evaluation results from the account in an individual study, the reveal the possibility of reporting
studies reviewed. The validity of Handbook considers a number of bias. In a case-control study, the
these studies will be examined aspects of design and analysis as effects of investigated factors other
critically to determine the weight they described in the report of the study. than the exposure of interest should
contribute to the assessment. This First, the study population, have been reported.
entails judging the appropriateness disease (or diseases), and exposure/ Finally, the statistical methods
of study design, data collection intervention should have been well- used to obtain estimates of relative
362
Working procedures
risk, absolute rates of cancer, explain heterogeneity among studies 0DJQLWXGH RI WKH UHSRUWHG
confidence intervals, and significance in more detail. A disadvantage of association;;
tests, and to adjust for confounding combined analyses is the possible 7HPSRUDO UHODWLRQVKLS EHWZHHQ
should have been clearly stated by lack of compatibility of data from exposure/intervention and
the authors. various studies due to differences in change in disease;;
Aspects that are particularly subject recruitment, data collection ([SRVXUHUHVSRQVH ELRORJLF
important in evaluating experimental procedures, measurement methods, gradient;;
studies are: the selection of and effects of unmeasured covariates %LRORJLFDOSODXVLELOLW\
participants, the nature and adequacy that may differ between studies. &RKHUHQFH RI UHVXOWV DFURVV
of the randomisation procedure, Meta-analyses may be conducted other lines of evidence;; and
evidence that randomisation achieved by the WG during the course of $QDORJ\ SUHVHQW LQ UHODWHG
an adequate balance between groups, preparing a Handbook and are exposures and their effects on
the exclusion criteria used before and identified as original calculations by health.
after randomisation, compliance with placement of the results in square If the results are inconsistent
the intervention in the intervention brackets or in italics. These may among investigations, possible
group, and ‘contamination’ with the be de novo analyses or updates explanations are sought (e.g.
intervention in the control group. of previously conducted analyses differences in level of exposure/
Other considerations are the that incorporate the results from intervention). Results of studies
means by which the endpoint was new studies. Whenever possible, judged to be of high quality are given
determined and validated, the length however, such analyses are more weight than those of studies
and completeness of follow-up of preferably conducted preceding the judged to be less methodologically
the groups, and the adequacy of the Handbook meeting. Publication of sound. When several studies show
analysis. the results of such meta-analyses little or no indication of an association
Detailed analyses of both relative prior to, or concurrently with, the between an intervention and cancer
and absolute risks in relation to Handbook meeting is encouraged for prevention, the judgment may be
temporal variables, such as age purposes of peer review. The same made that, in the aggregate, they
at first exposure, time since first criteria for data quality that would be show evidence of lack of effect. The
exposure, duration of exposure, applied to individual studies must be possibility that bias, confounding,
cumulative exposure, peak exposure applied to combined analyses, and or misclassification of exposure or
(when appropriate), and time since such analyses must take into account outcome could explain the observed
exposure ceased, will be reviewed heterogeneity between studies. results should be considered and
and summarised when available. excluded when reasonable certainty
Independent, population-based Criteria for causality exists.
studies of the same exposure or
intervention may lead to ambiguous After the quality of each study has Assessing studies reporting
results. Combined analyses of data been summarised and assessed, the impact of tobacco control
from multiple studies may be a means a judgment is made concerning policy interventions not necessarily
of resolving this ambiguity. There are the strength of evidence that the contemplating health outcomes
two types of combined analysis: the exposure or intervention in question
first combines summary statistics, reduces the risk of disease or is Evaluating the outcomes of
such as relative risks, from individual protective for humans. Hill (1965) population level tobacco control
studies (meta-analysis);; the second lists areas for evaluating the strength policy involves three interrelated
involves a pooled analysis of the of epidemiological associations used questions: (1) Does the policy have
raw data from the individual studies in the review of human data when an impact? (causality);; if so, (2) Under
(pooled analysis). assessing carcinogenesis. These what conditions? (moderation);; and
Advantages of combined criteria, in many instances, will apply (3) How (mediation)?
analyses include better precision to the assessment included in a The choice of design elements
due to increased sample size, as Handbook: will depend on which questions
well as the opportunity to explore &RQVLVWHQF\ RI REVHUYHG are considered to be a part of the
potential confounders, interactions, associations across studies and evaluation effort. It is important to
and modifying effects that may populations;; ensure that the appropriate concepts
363
are chosen, and, that for each, have not changed;; such variables that demonstrate the effect of price
measures are identified that are act as another form of control. on consumption is a good example of
suitable to answer the evaluation Recommendations for measures the power of multiple measurements.
question. pertinent to the evaluation of each Both repeated cross-sectional
In the absence of a randomised WHO Framework Convention on and longitudinal (cohort) designs
trial, there are two study design Tobacco Control policy domain are are useful for assessing the impact
strategies that can be employed provided in Handbook Volume 12 of a given policy. The use of cohort
for the rigorous evaluation of the (IARC, 2008). designs provides an additional
effects of policies. First is the use Combining the two design and capability for tracking the impact of
of measurements both before and two study feature strategies, along policies within individuals, allowing
after the policy’s implementation. with the inclusion of other explanatory stronger tests of mediational
These measurements can be taken variables (covariates) that might pathways.
from either units (usually, but not help explain differences between Addition of samples from
limited to, individuals;; the same logic two jurisdictions, creates a powerful other populations to either or both
would apply if the measures were of research design, allowing more intervention and control arms, also
households, schools, or other venues) confident inferences to be made adds strength to the evaluation
that are either the same (as in a about the causal effects of policies design, as does having varying levels
cohort design) or different, but drawn and/or combinations of policies. of intensity of the intervention.
from the same sampling process (as Evaluation efforts should be Similarly, parallel assessment of
in a repeat cross-sectional design). informed by knowledge of the nature alternative explanations for observed
The second design strategy is the of the policy being evaluated, and the changes in outcomes (e.g. possibly
use of a quasi-experimental design, goals of the evaluation study should being due to other policies or industry
in which one group that is exposed be clearly stated. Evaluation planning counter-actions) adds strength over
to a policy is compared to a similar should be guided by understanding assessing these effects in separate
unexposed group, as discussed what threats to internal validity may studies.
above. Combining these two be present in the study of a given The existence of studies with
strategies in a single study yields a policy situation, and then adding complementary strengths and
two-group, pre-post design, which design elements and other measures weaknesses is particularly useful in
offers a higher degree of internal to reduce or eliminate those threats. triangulating the results of a corpus
validity than either feature alone. Knowledge of the mediational of evaluation studies to see if a
The utility of longitudinal designs is pathways that are theorised to explain consistent pattern emerges.
strengthened if there are multiple how policy affects behaviour and The use of probability sampling
data collections before and/or after environment (or environmental risk) in an evaluation study increases its
policy implementation, allowing should lead to an appropriate study external validity - the extent to which
more precise specification of effects design, the inclusion of appropriate the findings of a policy evaluation
(e.g. taking into account temporal constructs and measures, and the study can be generalised to draw
trends that were occurring before the selection of analytic tools that are conclusions about the impact of the
implementation of the policy). well-suited to estimating the causal policy on the larger population.
A distinction between study impact of policies by providing an At a broader level, the design of
designs and study features is worth explanatory pathway and helping to an evaluation study should be guided
noting. In addition to the two design eliminate alternative explanations. by knowledge of how prior evaluation
considerations stated above, there The utility of longitudinal designs studies in the same policy domain
are two study feature strategies that is strengthened if there are multiple have been conducted. An analysis of
contribute to increasing an evaluation data collections before and/or after the similarity or differences in policy
study’s internal validity. The first is policy implementation, as this allows impact across similar studies can
the measurement of policy-specific more precise specification of effects yield powerful conclusions about the
variables that are theorised to be (e.g. taking into account temporal overall impact of a policy.
affected initially after the policy is trends that were occurring before
implemented. A second strategy the implementation of the policy).
is the measurement of policy- The role of time series analysis on
specific variables for policies that aggregate sales/consumption data
364
Working procedures
Summary of the data reviewed intervention under consideration determine the diversity of scientific
(evidence) and a given effect, but evidence opinion on issues where consensus
of causality is limited by the fact is not readily apparent.
This section summarises in a concise that chance, bias, or confounding
manner the results of the evidence have not been ruled out with Recommendations
presented in the preceding sections in reasonable confidence. However,
a Handbook. Traditionally, this section explanations other than causality After reviewing the data and
does not include citation of literature, are unlikely. deliberating on them, the WG may
as do preceding sections presenting Limited evidence: There is formulate recommendations, where
and discussing the evidence. some evidence of association applicable, for further research and
between the intervention under public health action.
Evaluation of the evidence consideration and a given effect,
but alternative explanations are
An evaluation of the strength of the possible.
evidence for disease prevention or Inadequate/no evidence: There
reduction in morbidity and mortality are no available methodologically
References
is made using standard terms sound studies showing an
described in previous volumes of the association. The available
Handbooks of Cancer Prevention studies are of insufficient quality, Cogliano VJ, Baan RA, Straif K et al. (2004).
(e.g. Volume 11). In evaluating the consistency, or statistical power The science and practice of carcinogen
identification and evaluation. Environ Health
strength of the evidence on the effects to permit a conclusion regarding Perspect, 112:1269–1274.doi:10.1289/ehp.6950
of tobacco control interventions the presence or absence of a PMID:15345338
directed at the population, disease causal association between the Hill AB (1965). The environment and disease:
prevention or health outcomes intervention and a given effect. association or causation? Proc R Soc Med,
58:295–300. PMID:14283879
may not always be a measurable Alternatively, this category is
endpoint. Also, it is conceivable used when no data are available. IARC (2008). IARC Handbook of Cancer
that not every exposure/intervention Evidence suggesting lack Prevention, Tobacco Control, Vol.12: Methods
for Evaluating Tobacco Control Policies.
reviewed in a Handbook of tobacco of effect: There are several
control will permit a formal evaluation methodologically adequate
of the evidence, as traditionally studies that are mutually
done in other Handbooks of Cancer consistent in not showing
Prevention and in the Monographs. an association between the
The following criteria are intervention and a given effect.
proposed when evaluating the weight
of the evidence on the effects of Overall evaluation
tobacco control interventions:
Sufficient evidence: The WG The overall evaluation, usually in
considers that an association has the form of a narrative, will include
been observed in studies in which a summary of the body of evidence
chance, bias, and confounding considered as a whole, and summary
can be ruled out with reasonable statements made about the strength
confidence. The association of the evidence for policy effects,
is highly likely to be causal. A including changes in tobacco
statement that there is sufficient use, changes in health risks, and
evidence should be followed by a incidental effects.
separate sentence that identifies IARC WGs make every effort to
the nature and magnitude of the achieve a consensus evaluation.
observed effect. Consensus reflects broad agreement
Strong evidence: There is among WG members, but not
consistent evidence of an necessarily unanimity. The chair
association between the may elect to poll WG members to
365
366

Effectiveness of Tax and Price

Uploaded by

Copyright:

Available Formats

You might also like

Effectiveness of Tax and Price

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Effectiveness of Tax and Price

Uploaded by

Copyright:

Available Formats

IARC

HANDBOOKS OF CANCER PREVENTION

Effectiveness of Tax and Price

Effectiveness of Tax and Price

Published by the International Agency for Research on Cancer,

© International Agency for Research on Cancer, 2011

Format for Bibliographic Citation

IARC Library Cataloguing-­in-­Publication Data

(IARC Handbooks of Cancer Prevention ;; 14)

ISSN 1027-­5622 (NLM Classification QZ 39)

List of participants......................................................................................................................................................................iv

Frank Chaloupka (Chair) Silvano Gallus Carlo La Vecchia

Luke Clancy Fiona Godfrey Nigar Nargis

Sofia Delipalla Roberto Iglesias (Unable to attend) Zeynep Onder

Anne-­Marie Perucic Frank van Driessche1 IARC Secretariat

publication, WHO Technical Manual References

tobacco is predominantly consumed Smokeless tobacco products a variety of dissolvable tobacco

excise duties on tobacco products Austria 2.0%

Tobacco use is the leading cause of Hungary 3.2%

preventable death, and is estimated Estonia 3.4%

single form of avoidable death and Romania 5.8%

However, a policy aimed at Employment rates to tobacco products depending

falls below a minimum f loor, a specific Purely specific system 55

WHO African Region (AFRO)

WHO Region of the Americas (AMRO)

WHO Eastern Mediterranean Region (EMRO)

WHO South-­East Asia Region (SEARO)

WHO Western Pacific Region (WPRO)

WHO European Region (EURO)

Ad valorem with a specific Mixed with a specific

Consumers: Ad valorem with a specific Mixed with a specific

Ad valorem with a specific Mixed with a specific

price reductions price reductions price reductions

Ad valorem with a specific Mixed with a specific

Ad valorem with a specific Mixed with a specific

WAP: Weighted Average Price;; MPPC: Most Popular Price Category

Imperial Tobacco Group Plc 25.6 2.9 9.2 43.9 4.0

China National Tobacco Corp (CNTC) 98

ITC Group* 58.3

Golden Tobacco Ltd 10.9

VST Industries Ltd 9.2

Godfrey Phillips India Ltd** 0.4

Gudang Garam Tbk Ltd** 28.3

Djarum PT 13.8

Bentoel Internasional investama Tbk PT 5.9

Nojorono Tobacco Indonesia PT 5.5

Philip Morris USA Inc 48.4

Reynolds American Inc* 26.5

Lorillard Inc 10.1

Liggett Vector Brands Inc 1.8

Societé industrielle des Tabacs du Cameroun 1.4

Company 2004 2005 2006 2007 2008* 2009* 2010* 2011*

Cadbury 15.6 15.9 14.4 13.5 12.0 13.0 13.8 14.9

Danone 12.7 13.1 13.3 12.1 14.4 16.9 15.7 15.9

Nestle 12.7 12.9 13.5 14.0 14.3 14.4 13.0 13.2

Consumer products companies

Henkel 9.4 9.7 10.2 10.5 10.3 9.0 10.6 11.6

L’Oreal 15.3 15.6 16.4 16.6 15.5 14.3 14.9 15.5

SABMiller 18.1 20.2 16.9 16.8 16.6 16.8 17.3 18.5

IARC Library Cataloguing-in-Publication Data

ISSN 1027-5622 (NLM Classification QZ 39)

Anne-Marie Perucic Frank van Driessche1 IARC Secretariat

WHO South-East Asia Region (SEARO)

Moon et al., (1993) - Multiple mailings were

Goldman and Glantz, 39 and against Measure - Philip Morris concluded

Givel and Glantz, (2001) - Forming relationships

- The LMC lobbied and Hence, “While regularly

Raebeck et al., (2009) study claimed “An African-

- an ad valorem domestic

One study found the industry framed 7REDFFR,QVWLWXWH¶V/DERU0DQDJHPHQW&RPPLWWHHPDGHXSRIXQLRQJURXSV

smokers to pay for services for others, (QRXJKLV(QRXJK%LDORXVDQG*ODQW]

advertising Hausman-Talyor, neighbouring states to

-0.40 variables, regulation panel linear from 1967 to 1990

Zimbabwe Short-run Positive Rational addiction Annual time series

Long-run elasticity 1996