Original research
Tob Control: first published as 10.1136/tc-2022-057787 on 1 December 2023. Downloaded from http://tobaccocontrol.bmj.com/ on December 2, 2023 by guest. Protected by copyright.
► Additional supplemental
material is published online
only. To view, please visit the
journal online (http://​dx.​doi.​
org/​10.​1136/​tc-​2022-​057787).
1
Catholic University of Brasilia,
Brasilia, Brazil
2 Brazil.
Getulio Vargas Foundation
School of Public Policy and
Government, Brasilia, Brazil
Correspondence to
Dr Jose Angelo Divino, Catholic
University of Brasilia, Brasilia
71966-­700, Brazil;
​jadivino@​gmail.​com
Received 6 October 2022
Accepted 15 November 2023
© Author(s) (or their
employer(s)) 2023. Re-­use
permitted under CC BY.
Published by BMJ.
To cite: Divino JA, Ehrl P,
Candido O, et al. Tob Control
Epub ahead of print: [please
include Day Month Year].
doi:10.1136/tc-2022-057787
Effects of the illicit market on the price elasticity of
cigarette consumption in Brazil
Jose Angelo Divino ‍ ‍ ,1 Philipp Ehrl ‍ ‍ ,2 Osvaldo Candido ‍ ‍ ,1
Marcos Aurelio Pereira Valadao ‍ ‍ 2
ABSTRACT
Background An important element to consider in
tobacco tax policy is the illicit market of cigarette sales.
The objective of this paper is to provide estimates of
both conditional and unconditional price elasticities of
cigarette consumption in the licit and illicit markets in
Methodology Microdata from the National Health
Survey in 2013 and 2019 are used to estimate
conditional and unconditional price elasticities of
cigarette consumption in the licit and illicit cigarette
markets by income quartiles and age cohorts. The
identification is based on brand information and
the official minimum cigarette price defined by the
government, as sales below this price are prohibited and
illegal.
Findings The results, robust to potential endogeneity,
indicate that there is joint statistical difference in price
elasticities across age cohorts and income groups
by market type. However, individuals smoking illicit
cigarettes, regardless of age cohort and income quartiles,
are less sensitive to price changes than those consuming
licit brands.
Conclusions The illicit cigarette market prevents the
government from collecting tobacco tax revenues and
weakens the social reach of price-­oriented antismoking
public policies. Fighting the illicit trade should be a major
concern of public policies aiming at reducing cigarette
consumption.
INTRODUCTION
The empirical literature on tobacco economics
is extensive and includes applications to several
countries around the world. Theoretically, it relies
on the studies of addictive goods, as pioneered by
Stigler and Becker,1 Becker and Murphy,2 among
others. Chaloupka,3 for instance, applied the ratio-
nality model of Becker and Murphy.2 A key insight
from a theoretical and empirical standpoint is that
addictive goods, such as cigarettes, are less price
elastic than non-­addictive goods.
The price elasticity is an important tool for
applied researchers and policy makers because it
indicates how effective price-­based tobacco control
measures, such as tax increases, can be. Having
price elasticities for specific groups in the popu-
lation provides additional insights regarding the
distribution of the overall tax burden and whether
tax increases will be progressive or regressive.
Regarding the estimates by age or income groups,
the literature is much less consolidated (Ross and
Chaloupka4). Another important aspect, particu-
larly for low-­income and middle-­income countries,
Divino JA, et al. Tob Control 2023;0:1–6. doi:10.1136/tc-2022-057787
WHAT IS ALREADY KNOWN ON THIS TOPIC
⇒ Price elasticity of cigarette consumption has
been estimated for many countries around the
world. Higher prices decrease consumption and
the sensitivity of consumers to price changes
depends on their level of addiction.
WHAT THIS STUDY ADDS
⇒ First study to estimate price elasticity of
cigarette consumption by age cohort and
income quantiles in both illicit and licit markets.
⇒ Smokers of illicit cigarettes in Brazil, regardless
of age cohort and income quartiles, are
less sensitive to price changes than those
consuming licit brands.
HOW THIS STUDY MIGHT AFFECT RESEARCH,
PRACTICE OR POLICY
⇒ Fighting the illicit trade should be a major
concern of public policies aiming at increasing
tobacco tax collection and the social reach of
price-­oriented antismoking public policies.
that is rarely addressed is how the presence of a
large illicit cigarette market affects the smoker’s
price sensitivity. Furthermore, few studies estimate
price elasticities simultaneously for the licit and
illicit markets as in the present case. Due to varia-
tion in geographical coverage, sampling criteria and
methodological approaches, comparison among
them in review studies tends to be more complex.
See, for instance, Wilson et al.5
Brazil is usually referred to as a successful case
of tobacco control policies, having experienced a
persistent and substantial reduction in the prev-
alence of smoking since 2006. However, an
important element to consider in tobacco tax policy
is the illicit market of cigarette sales. Data from
the Federal Revenue Service (Receita Federal do
Brasil) indicate a decrease close to 50% in the licit
production of manufactured cigarettes from 2006
to 2018. At the same time, the size of the illicit ciga-
rette market in Brazil fluctuates, depending on the
period and method of estimation. Thus, to some
extent, Brazil could be experiencing a substitution
of licit cigarettes by illicit ones, which would require
additional policy measures focused on curbing illicit
trade.
The objective of this study is to estimate both
conditional and unconditional own price elastici-
ties of cigarette consumption in the whole market
as well as in the licit and illicit markets of cigarette
sales in Brazil. Based on the National Health Survey
1
 Original research
(PNS) from 2013 and 2019 (Brazilian Institute of Geography
and Statistics, IBGE6), we estimate price elasticities of cigarette
consumption by income quartile and age cohorts. We apply
a novel way to identify the illicit market based on cigarette
purchases below the official floor price defined by the govern-
ment and brand classification as licit or illicit by the Brazilian
Health Regulatory Agency. The potential endogeneity problem
is accounted for by using the state average prices instead of indi-
vidually reported prices in the regressions. Ross and Chaloupka4
and Sweis and Chaloupka7 used the same approach and argued
that a geographically based measure of individual prices is an
effective way to avoid the endogeneity bias in models of cigarette
demand.
In line with McKelvey,8 we argue that the proposed correction
in Deaton9 is unlikely to improve the results and the required
assumptions are not valid in the current setting. Deaton’s correc-
tion was developed for household survey data, whereas we are
using individual survey with a different questionnaire structure.
First, under the plausible assumptions that smokers did not buy
different brands in their last purchase and that it is represen-
tative for their usual consumption behaviour, the inferred unit
price for cigarettes is more precise than those inferred in house-
hold consumption surveys. Our procedure is an adaptation of
Deaton9 for individual-­level prices. By doing so, we reduce the
risk of reverse causality in the estimated models. The estimated
elasticities suggest that individuals purchasing cigarettes in the
illicit market are less sensitive to price variations than individuals
purchasing in the licit market. Our results also complement the
literature which has produced mixed results regarding the varia-
tion of price elasticities among age and income groups (Ross and
Chaloupka4). Wasserman et al10 and Chaloupka3 find that young
adults are less price sensitive compared with older individuals.
Choi11 observes that lower income smokers in Korea are more
responsive to changes in cigarette prices. However, employing
different methods for other countries has produced opposing
results; see the reviews by Wilson et al,5 Levy et al12 and Bafunno
et al.13
Previous studies estimated price elasticities of cigarette
consumption for Brazil, under different set-­ups, include Carvalho
and Lobão,14 Iglesias,15 Ribeiro and Pinto16 and Lampreia et al.17
However, none of them explored the effects of the illicit market
on the estimates.
MATERIAL AND METHODS
To describe the smoking behaviour of the Brazilian population
and to estimate the sensitivity of cigarette consumption to price
changes, we used two individual-­level surveys: the PNS from
2013 and 2019 (IBGE6). Both surveys are representative of the
population and conducted by IBGE. Their broad questionnaire
is highly comparable and includes a special section on smoking
behaviour. The PNS has a household and an individual compo-
nent. We focused on the latter questionnaire because smoking
behaviour is essentially an individual choice. Using the provided
sample weights, the derived statistics are representative for the
entire Brazilian population. Further details on these surveys are
available in Szwarcwald et al.18
The survey respondents indicate how many cigarettes they
smoke per day and at what age they started smoking. A key
distinction of the PNS data in comparison to commonly
used household survey price data is that the information
regarding the quantity and total price of cigarettes refers
to the last individual purchase. The interviewee reports the
unit of purchase (sticks, packs or cartons) and the total price
2 Divino JA, et al. Tob Control 2023;0:1–6. doi:10.1136/tc-2022-057787
Tob Control: first published as 10.1136/tc-2022-057787 on 1 December 2023. Downloaded from http://tobaccocontrol.bmj.com/ on December 2, 2023 by guest. Protected by copyright.
Table 1 Characterisation of the illicit cigarette market according to
the PNS surveys
Size in terms Size in terms of Share of cigarette
of cigarette total number of purchases below the
Year consumption (%) smokers (%) minimum price (%)
2013 35.8 31.3 100
2019 40.6 35.4 90.6
In the PNS 2013, the definition of illicit market is based on cigarettes sold below
the official minimum price, while in the PNS 2019 the definition is based on the
reported purchased brand and corresponding classification as licit or illicit by the
Brazilian Health Regulatory Agency (Anvisa). The latter is more accurate because
some illicit cigarettes are sold above the minimum price, as illustrated in the fourth
column.
PNS, National Health Survey.
paid. This information is then processed by the authors
to calculate the implied price of a 20-­c igarette pack. The
first advantage of this type of question is that it includes all
kinds of markets where cigarettes are sold. Consequently, it
will be fruitful to exploit the official minimum price for a
20-­c igarette pack, which in 2013 was equal to BRL3.50 and
by 2019 has been raised to BRL5.00. Purchases below this
floor price shall have occurred in the illicit market because
it is forbidden by the law to sell cigarettes below this price
in the legal market. Prices and changes in prices must be
reported to the Federal Revenue Service and any change in
the retail price must also be informed because ad valorem
taxes are based on this price. In addition, cigarette packs are
subject to tax stamps as a control system. No promotional
sales are allowed, and no cigarette can be legally sold at a
price below the official floor price. A remaining concern is
that cigarettes may still be illicit and sold above the minimum
price. For the 2019 PNS data, we can additionally use infor-
mation about the cigarette brands that allowed the IBGE
to classify them as licit or illicit. Overall, in both PNS data
sets, about one-­third of all consumers purchased cigarettes
below the official price floor. For better comparability of the
two data sets, monetary variables were deflated by the wide
consumer price index (IPCA (Índice Nacional de Preços ao
Consumidor Amplo)) calculated and released by the IBGE.
Table 1 provides a brief characterisation of the illicit market
in 2013 and 2019 in terms of size and the share of cigarettes
sold below the minimum price. For a more complete descrip-
tion, see online supplemental appendix table A.1.
The price elasticity of cigarette consumption measures the
per cent change in cigarette consumption due to a per cent
change in the cigarette price. To account for the potentially
different effects of a tobacco price increase over the income
and age distributions, price elasticities are estimated by
income quartiles and age cohorts. We estimate these equa-
tions jointly for the full sample of smokers, which includes
2 years and two sources of cigarette purchase, either the licit
or illicit market. As an extension, we include an indicator
variable for the legal market and interact it with all explan-
atory variables, so that price elasticities in both markets are
identified from a single estimation.
Since the good is highly similar but prices among different
brands vary quite substantially, consumers may adjust to price
changes by switching to a cheaper brand. To deal with a potential
endogeneity bias and the problem of misreporting the price of
the individuals’ last purchase, leading to measurement error and
the well-­known attenuation bias in the coefficients towards zero,
reported unit prices are substituted with average prices within
 Original research

Tob Control: first published as 10.1136/tc-2022-057787 on 1 December 2023. Downloaded from http://tobaccocontrol.bmj.com/ on December 2, 2023 by guest. Protected by copyright.
Figure 1 Average prices per 20-­cigarette pack in the legal market across Brazilian states. Constant values of 2019. Source: National Health Survey
(PNS) 2013 and PNS 2019. BRL, Brazilian real.

each state. Prices differ across states because part of the tobacco
tax burden is given by a state-­specific tax (ICMS (Imposto sobre
Circulação de Mercadorias e Serviços)) that varies across states.
In addition, logistic, distribution and transportation costs to the
countryside are quite high due to poor infrastructure. Figure 1
reports average prices of legal cigarettes across the Brazilian
states. Prices tend to be lower in producer states and higher
in the countryside and in states with larger tax burden. Price
dispersion increased from 2013 to 2019 probably due to a fixed
value of BRL5.00 for the official minimum price since 2016,
product differentiation and more brands entering the market.
The conditional price elasticities, ‍εd ‍, are estimated from the
baseline specification:
∑
ln Qidst = α + εd ln Pidst Idt + βXidst + Ds + Dt + eidst
(1)
‍ d ‍
where Qidst is the number of cigarettes smoked per day by
individual i in income quartile d, federal state s and year t; Pidst
is the price that individual i actually paid per cigarette; Idt is a
binary variable that indicates to which income quartile (or age
cohort) the individual belongs; and the vector Xidst includes
control variables for age, education, years of smoking, income
and gender. Ds represents the state fixed effects, Dt accounts for
year fixed effects and eidst is the compound random error term
of the regression.
To estimate price elasticities for the licit and illicit cigarette
markets, we include a binary variable for these markets and

Table 2 Conditional price elasticities by income quartile and market type

(1) (2a) (2b) (3) (4a) (4b)
Market Both Legal Illegal Both Legal Illegal
Price measure Individual State average
1st quartile −0.23*** −0.39*** −0.27*** −0.41*** −0.90*** −0.31***
(0.02) (0.04) (0.04) (0.07) (0.17) (0.10)
2nd quartile −0.21*** −0.37*** −0.26*** −0.42*** −0.90*** −0.35***
(0.02) (0.04) (0.04) (0.07) (0.17) (0.10)
3rd quartile −0.23*** −0.42*** −0.25*** −0.39*** −0.87*** −0.32*
(0.02) (0.04) (0.04) (0.07) (0.17) (0.10)
4th quartile −0.30*** −0.44*** −0.35*** −0.36*** −0.87*** −0.20**
(0.03) (0.05) (0.05) (0.07) (0.17) (0.10)
Observations 10 326 10 326 10 326 10 326
R2 0.09 0.10 0.06 0.07
F-­quartiles 5.15*** 1.39 5.08*** 3.06** 1.85 4.62***
F-­markets 2.84** 4.83***
The table reports coefficients of price elasticities by income quartile according to the estimation of equation (1). The White-­Huber heteroscedasticity robust SEs are in parenthesis.
Columns (1) and (3) indicate purchases of both licit and illicit cigarettes, while columns (2a) and (4a) refer only to licit cigarettes and (2b) and (4b) remit only to licit cigarettes.
Columns (1), (2a) and (2b) refer to individually reported prices while columns (3), (4a) and (4b) indicate state average prices. The last two rows report the F-­statistics for testing
if the estimated elasticities are equal. When both markets are considered separately in columns (2a), (2b), (4a) and (4b), the equation is estimated jointly with all variables being
interacted with a market indicator variable. The estimations are based on pooled data from 2013 and 2019. All regressions include controls for gender, age group, education
group, year and log income. Federal state fixed effects were only included in the estimations in columns (1), (2a) and (2b). Estimations in columns (3), (4a) and (4b) include
region fixed effects instead.
*, ** and *** indicate statistical significance at the 10%, 5% and 1% levels, respectively.

Divino JA, et al. Tob Control 2023;0:1–6. doi:10.1136/tc-2022-057787 3

 Original research

Tob Control: first published as 10.1136/tc-2022-057787 on 1 December 2023. Downloaded from http://tobaccocontrol.bmj.com/ on December 2, 2023 by guest. Protected by copyright.
Table 3 Conditional price elasticities by age cohorts and market type
(1) (2a) (2b) (3) (4a) (4b)
Market Both Legal Illegal Both Legal Illegal
Price measure Individual State average
Aged 15–29 −0.17*** −0.30*** −0.23*** −0.47*** −0.99*** −0.38***
(0.02) (0.05) (0.04) (0.07) (0.17) (0.10)
Aged 30–39 −0.24*** −0.40*** −0.28*** −0.41*** −0.92*** −0.30***
(0.02) (0.04) (0.04) (0.07) (0.17) (0.10)
Aged 40–49 −0.26*** −0.45*** −0.28*** −0.38*** −0.89*** −0.28***
(0.02) (0.04) (0.04) (0.07) (0.17) (0.10)
Aged 50–59 −0.28*** −0.50*** −0.28*** −0.35*** −0.85*** −0.27***
(0.02) (0.04) (0.04) (0.07) (0.17) (0.10)
Aged 60+ −0.25*** −0.48*** −0.24*** −0.37*** −0.86*** −0.34***
(0.02) (0.04) (0.04) (0.07) (0.17) (0.10)
Observations 10 362 10 362 10 362 10 362
R2 0.08 0.09 0.06 0.06
F-­age groups 10.27*** 13.78*** 2.35* 20.49*** 3.85*** 20.44***
F-­markets 7.09** 4.11***
Columns (1) and (3) indicate purchases of both licit and illicit cigarettes, while columns (2a) and (4a) refer only to licit cigarettes and (2b) and (4b) remit only to licit cigarettes.
Columns (1), (2a) and (2b) refer to individually reported prices while columns (3), (4a) and (4b) indicate state average prices. The last two rows report the F-­statistics for testing
if the estimated elasticities are equal. When both markets are considered separately in columns (2a), (2b), (4a) and (4b), the equation is estimated jointly with all variables being
interacted with a market indicator variable. The estimations are based on pooled data from 2013 and 2019. All regressions include controls for gender, age group, education
group, year and log income. Fixed effects for federal states were only included in the estimations in columns (1), (2a) and (2b). Estimations in columns (3), (4a) and (4b) include
region fixed effects instead.
*, ** and *** indicate statistical significance at the 10%, 5% and 1% levels, respectively.

interact it with all explanatory variables, as reported in equation availability, not all studies are able to deal with the problem.
(1′): Following the exposition in WHO,19 this study estimates the
∑ ∑
α+ εd ln Pidst Idt + ξd ln Pidst Idt Iimt + βXidst + unconditional price elasticity related to the number of smokers
ln Qidst = d d
(1′) from the following probit model.
‍ ‍
δXidst Iimt + Ds + Dt + Ds Iimt + Dt Iimt + eidst ( ) ( )
‍ Pr Sist = ϕ α + γPst + δXist + Dr + Dt ‍ (2)
where ‍Iimt ‍ is a dummy variable that assumes value of 1 for
cigarettes purchased by individual i in the licit market in time t The dependent variable in equation (2) is an indicator of
and 0 for purchases from the illicit market. As explained above, whether individual i is a smoker or not, and Pst is the average
to account for potential endogeneity, unit prices are replaced by price in federal state s and year t of the individual reported price.
state average prices, in which case the variable ‍Pidst ‍becomes ‍Pdst ‍ The explanatory variables X are the same as in the conditional
because there is price variation only across Brazilian states in price elasticity estimation by equation (1), except that we use the
this case. The interaction dummy is also adjusted accordingly to state average cigarette prices without interaction terms, because
capture the average price of licit cigarettes across the states in the once the coefficients are estimated, the non-­linear probit model
joint estimation of the price elasticities. allows us to calculate the price elasticities for different income
To calculate how tax adjustments affect cigarette consump- or age levels. For this reason, we also use region fixed effects
tion, one needs to be aware that a price change may induce indi- (Dr) instead of the more disaggregated federal state fixed effects.
viduals either to start smoking or to quit. This issue is also well The price elasticity from the smoking decision model for an indi-
1 ∑N
known in the literature on tobacco, although, mainly due to data vidual with average characteristics ‍X = N i=1 Xi ‍ is calculated
as follows:
∂E(S|X̄) Pst
εu ≡ (3)
Table 4 Unconditional price elasticities from the smoking probability ‍ ∂Pst E(S|X̄) ‍

model by subgroups where ‘E(.)’ represents the expected value of the expression in
Age cohorts 15–29 30–39 40–49 50–59 60+ parenthesis. It is important to note that this elasticity indicates
−0.51*** −0.50*** −0.48*** −0.44*** −0.50*** the per cent change in smoking prevalence due to a per cent
(0.06) (0.06) (0.06) (0.06) (0.06) change in the cigarette prices.
Income quartiles Q1 Q2 Q3 Q4
−0.48*** −0.48*** −0.48*** −0.51***
RESULTS
(0.06) (0.06) (0.06) (0.06) Conditional price elasticity of cigarette consumption
The table shows the estimated price elasticities and their White-­Huber The estimations of the conditional price elasticity of cigarette
heteroscedasticity robust SEs in parenthesis. All elasticities are calculated from a consumption reported in table 2 show that the differences
single probit estimation according to equation (2), where the smoking indicator
is regressed on the observed mean price of cigarettes in each federal state and
between the licit and illicit markets are substantial. Consumers
year, and we control for year, gender, education group, region and log income. The in the illicit market are less sensitive to price changes.
estimations are based on the entire sample of smokers and non-­smokers pooled for Plausible explanations are that the smaller number of brands,
the years 2013 and 2019 that contains 122 947 observations. the limited number of sales points and other characteristics of an
*** indicates significance at the 1% level. illegal market limit the smokers’ choices and flexibility to adjust
4 Divino JA, et al. Tob Control 2023;0:1–6. doi:10.1136/tc-2022-057787

their consumption behaviour due to price changes. A recent
report on illicit cigarette market found that in five representative
cities of the country there were only a small handful of illicit
brands (Drope et al20). Additionally, the observations may stem
from the fact that smokers of illegal cigarettes show a higher
degree of addiction in line with Chaloupka3 and Becker and
Murphy.2 The differences in price elasticities may also be due to
other intrinsic characteristics of the illegal cigarette consumers.
Online supplemental appendix table A.1 shows that smokers of
illegal products consume more cigarettes and have been smoking
for a longer time, in line with the previous literature in Brazil
(Ribeiro and Pinto16). Note that switching between the legal
and illegal cigarette markets in response to price changes is very
limited, as recent evidence from Divino et al21 indicates.
Based on the individually reported price of cigarettes, price
elasticities in the illegal market vary between −0.25 and −0.35,
whereas in the legal market they range from −0.37 to −0.44.
Notice that the estimated elasticities by income quartile are not
statistically different from each other within the legal market,
but they are in the illegal market, as illustrated by the F-­test for
quartiles at the end of columns (2a) and (2b), respectively. They
are also statistically different when considering the entire ciga-
rette market in column (1).
Columns (3), (4a) and (4b) in table 2 report the results when
individually reported cigarette prices are replaced by average
price within each federal state. The estimation of the price
elasticities now stems exclusively from variations between state
average prices and income quartiles. Consequently, the elas-
ticities unambiguously increase in magnitude. This change is
expected because the regional prices eliminate the endogeneity
bias caused by possible adjustments to cheaper brands under
a price increase. Again, the price elasticities are statistically
different between the two markets, legal and illegal, and they
are different across income quartiles in the aggregated and the
illegal market, as indicated by the F-­statistics reported in the last
row of table 2.
Table 3 provides price elasticities for five different age cohorts.
Based on the previous findings, the distinction between licit and
illicit markets is maintained. Regardless of individually reported
price or state average price as a regressor, the five price elastic-
ities are statistically different across markets and across the age
groups in all of the six estimations, as illustrated by the calcu-
lated F-­statistics in the last row of the table. As in table 2, the
price-­elasticities in the illicit market are lower than in the licit
market while the endogeneity of individual prices downwards
bias the estimates. Columns (3), (4a) and (4b) in table 3 report
the results for the price elasticities by age cohort using the state
average cigarette prices in the pooled sample, the licit and illicit
markets, respectively. In all three cases, younger smokers seem to
be more price sensitive. The estimates range from –0.99 to –0.85
in the licit market and from –0.38 to –0.27 in the illicit market
in columns (4a) and (4b).
Unconditional price elasticity of smoking probability
Table 4 reports the unconditional price elasticities derived from a
single probit estimation according to equation (2). The observed
pattern across age and income groups is in line with the condi-
tional price elasticity estimates. That is, younger and wealthier
individuals show a slightly higher, not statistically significant,
propensity to quit smoking in response to a cigarette price
increase. This finding is in accordance with the level of addic-
tion for these groups observed in the literature and in online
supplemental appendix table A.1. Yet, the differences between
Divino JA, et al. Tob Control 2023;0:1–6. doi:10.1136/tc-2022-057787
Original research
Tob Control: first published as 10.1136/tc-2022-057787 on 1 December 2023. Downloaded from http://tobaccocontrol.bmj.com/ on December 2, 2023 by guest. Protected by copyright.
the estimates across groups are much less pronounced than the
ones from table 3. That is, for most Brazilians, a 10% cigarette
price increase should reduce the number of smoking individuals
by about 5%.
DISCUSSION
We used two detailed individual surveys—the PNS from
2013 and 2019 (IBGE6)—to estimate cigarette price elastici-
ties by income quartile and age groups and to account for the
effects of the illicit cigarette market in Brazil. The information
regarding registered brands and the official minimum price
for a 20-­cigarette pack allowed us to identify those consumers
engaged in the illicit consumption. That is, purchases below this
floor price are considered illicit. According to the results, there
is a significant difference in the price elasticity estimations from
the licit and illicit cigarette markets. Specifically, the smokers are
less sensitive to price changes when consuming cigarettes from
the illicit market (about –0.3) compared with elasticities about
–0.9 in the licit market.
In addition, to the best of our knowledge, this is the first study
to estimate price elasticities of cigarette demand by income and
age groups. We find that younger and poorer individuals seem to
be more price sensitive; however, the differences across income
quartiles are less pronounced as compared with the age groups.
Tax increases would thus particularly induce young adults and
poorer households to smoke less.
Another important implication of tax increases is that they
should have a smaller effect, if any, on prices in the illicit market.
Obviously, a tax does not directly affect illicit cigarette prices
but previous evidence from Brazil that compares the evolution
of prices in the licit and illicit cigarette markets points out that
illegal cigarette prices increase when taxes are raised because
there is a strong positive correlation between licit and illicit prices
(Brazilian Health Ministry22 23). Therefore, even if we suppose
that there is a correlation between the two prices, the lower
price sensitivity of smokers in the illicit market suggests that they
would be less affected by tax changes that eventually ended up
raising both prices. However, smokers of illicit cigarettes are still
demanding medical treatment for tobacco-­related diseases and
losing future years of productive life because of cigarette-­related
diseases. The existence of an illicit cigarette market prevents the
government from collecting tobacco tax revenues and weakens
the social reach of price-­oriented antismoking public policies by
blunting the effectiveness of the tax increases. Public policies that
target cigarette smuggling should always be a priority because
they reduce the illicit trade itself and they increase effectiveness
and efficiency of other tobacco control policies.
A potential limitation regarding the identification of the illegal
market is that we cannot account for high-­price illegal cigarettes
when using the official minimum price as a threshold to define
illicit cigarettes in the 2013 data. Moreover, if an individual
purchased both illegal and legal cigarettes at the same time, it is
not possible to distinguish these products because we can only
classify them by the mean price, leading us to consider them
as either illegal or legal cigarettes. There might be some vari-
ability in general levels of smoking by state due to state-­specific
tobacco control measures that are captured by the state average
prices leading to biased coefficients. A final drawback is that the
Deaton’s average price methodology was designed for relatively
small geographical units, while our dataset is aggregate for the
Brazilian states. Due to the low number of smokers in the data
and the representativity of the survey at the federal state level,
further disaggregation is unfeasible.
5
 Original research
Acknowledgements We are grateful to Frank Chaloupka, German Rodriguez
Iglesias, Michal Stoklosa, Jeffrey Drope, Erika Siu, Carlos Manuel Guerrero López,
Alan Fuchs and the seminar participants at the UIC and LAC partners meeting in
Mexico City for their comments and suggestions.
Contributors JAD and MAPV collaborated on planning the study, defining
the objectives and the main variables to analyse. JAD, PE and OC conceived and
designed the study, conducted the document search, and read and analysed the
documents. PE and OC analysed the data, conducted the estimations and reported
the results. JAD, PE, OC and MAPV reported the work, revised the manuscript and
edited the article. JAD is responsible for the overall content as the guarantor.
Funding This research is funded by the University of Illinois at Chicago’s (UIC)
Institute for Health Research and Policy (grant number 17409-­02) to conduct
economic research on tobacco taxation in Brazil. UIC is a partner of the Bloomberg
Initiative to Reduce Tobacco Use.
Disclaimer The views expressed in this document cannot be attributed to, nor do
they represent, the views of UCB, FGV, UIC, the Institute for Health Research and
Policy or Bloomberg Philanthropies.
Map disclaimer The depiction of boundaries on this map does not imply the
expression of any opinion whatsoever on the part of BMJ (or any member of its
group) concerning the legal status of any country, territory, jurisdiction or area or of
its authorities. This map is provided without any warranty of any kind, either express
or implied.
Competing interests None declared.
Patient consent for publication Not applicable.
Provenance and peer review Not commissioned; externally peer reviewed.
Data availability statement Data are available in a public, open access
repository.
Supplemental material This content has been supplied by the author(s). It
has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have
been peer-­reviewed. Any opinions or recommendations discussed are solely those
of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and
responsibility arising from any reliance placed on the content. Where the content
includes any translated material, BMJ does not warrant the accuracy and reliability
of the translations (including but not limited to local regulations, clinical guidelines,
terminology, drug names and drug dosages), and is not responsible for any error
and/or omissions arising from translation and adaptation or otherwise.
Open access This is an open access article distributed in accordance with the
Creative Commons Attribution 4.0 Unported (CC BY 4.0) license, which permits
others to copy, redistribute, remix, transform and build upon this work for any
purpose, provided the original work is properly cited, a link to the licence is given,
and indication of whether changes were made. See: https://creativecommons.org/​
licenses/by/4.0/.
ORCID iDs
Jose Angelo Divino http://orcid.org/0000-0001-7359-7539
Philipp Ehrl http://orcid.org/0000-0002-6661-8976
Osvaldo Candido http://orcid.org/0000-0002-4850-6989
Marcos Aurelio Pereira Valadao http://orcid.org/0000-0002-9301-3366
REFERENCES
1 Stigler GJ, Becker GS. De Gustibus non EST disputandum. Am Econ Rev
1977;67:76–90.
2 Becker GS, Murphy KM. A theory of rational addiction. J Polit Econ 1988;96:675–700.
3 Chaloupka F. Rational addictive behavior and cigarette smoking. J Polit Econ
1991;99:722–42.
6 Divino JA, et al. Tob Control 2023;0:1–6. doi:10.1136/tc-2022-057787
Tob Control: first published as 10.1136/tc-2022-057787 on 1 December 2023. Downloaded from http://tobaccocontrol.bmj.com/ on December 2, 2023 by guest. Protected by copyright.
4 Ross H, Chaloupka FJ. The effect of cigarette prices on youth smoking. Health Econ
2003;12:217–30.
5 Wilson LM, Avila Tang E, Chander G, et al. Impact of tobacco control interventions on
smoking initiation, cessation, and prevalence: a systematic review. J Environ Public
Health 2012;2012:961724.
6 Instituto Brasileiro de Geografia e Estatística (IBGE). Pesquisa Nacional de Saúde -
Microdados. Brasília Instituto Brasileiro de Geografia e Estatística; 2022. Available:
https://www.ibge.gov.br/estatisticas/sociais/saude/9160-pesquisa-nacional-de-saude.​
html?=&t=microdados [Accessed 30 May 2022].
7 Sweis NJ, Chaloupka FJ. The economics of tobacco use in Jordan. Nicotine Tob Res
2014;16 Suppl 1:S30–6.
8 McKelvey C. Price, unit value, and quality demanded. J Dev Econ 2011;95:157–69.
9 Deaton A. Quality, quantity, and spatial variation of price. Am Econ Rev
1988;78:418–30.
10 Wasserman J, Manning WG, Newhouse JP, et al. The effects of excise taxes and
regulations on cigarette smoking. J Health Econ 1991;10:43–64.
11 Choi SE. Are lower income Smokers more price sensitive?: the evidence from Korean
cigarette tax increases. Tob Control 2016;25:141–6.
12 Levy DT, Tam J, Kuo C, et al. The impact of implementing tobacco control policies: the
2017 tobacco control policy scorecard. J Public Health Manag Pract 2018;24:448–57.
13 Bafunno D, Catino A, Lamorgese V, et al. Impact of tobacco control interventions
on smoking initiation, cessation, and prevalence: a systematic review. J Thorac Dis
2020;12:3844–56.
14 Carvalho JL, Lobão W. Vício Privado E Políticas Públicas: a Demanda Por Cigarros no
Brasil. Revista Brasileira de Economia 1998;52(Esp):67–104.
15 Iglesias R. A economia do controle do tabaco nos países do Mercosul e associados.
Brasil, Washington: Opas, 2006.
16 Ribeiro L, Pinto V. Accelerating effective tobacco taxes in Brazil: trends and
perspectives. In: Country Study N° 3/2019. Red Sur Country Studies Series Tobacco
Taxes in Latin America. Montevideo, Uruguay, 2019. Available: https://tobacconomics.​
org/files/research/555/EP3_FUNCEX_2019_comp_0.pdf [accessed 12 Feb 2022].
17 Lampreia S, Costa R, Biz A, et al. Tabagismo no Brasil: Estimação Das Elasticidades
Preço E Renda NA Participação E NA Demanda Por Cigarros Industrializados. Pesquisa
e Planejamento Econômico 2015;45:245–70.
18 Szwarcwald CL, Malta DC, Pereira CA, et al. Pesquisa Nacional de Saúde no Brasil:
Concepção E Metodologia de Aplicação. Ciênc Saúde Coletiva 2014;19:333–42.
19 WHO. Economics of tobacco Toolkit: economic analysis of demand using data from
the global adult tobacco survey (GATS). Geneva World Health Organization; 2010.
Available: https://www.who.int/publications/i/item/economics-of-tobacco-toolkit-​
economic-analysis-of-demand-using-data-from-the-global-adult-tobacco-survey
[Accessed 19 Feb 2022].
20 Drope J, Iglesias R, Figueiredo VC, et al. Consumo de Cigarros Ilegais em Cinco
Cidades Brasileiras. Centro de Estudos Sobre Tabaco E Saúde [Centro de Estudos
sobre Tabaco e Saúde. Escola Nacional de Saúde Pública, Fundação Oswaldo Cruz].
2022. Available: https://tabaco.ensp.fiocruz.br/pt-br/consumo-de-cigarros-ilegais-em-​
cinco-cidades-brasileiras [Accessed 23 May 2023].
21 Divino JA, Ehrl P, Candido O, et al. Tobacco tax reform and demand-­switching effects
between the Licit and illicit markets in Brazil [Working Paper UCB]. 2022. Available:
https://www.tobacconomics.org/tobacco-tax-reform-and-demand-switchingeffects-​
between-the-licit-and-illicit-markets-in-brazil-working-paper-series
22 Brazilian Health Ministry (BHM). Vigitel Brasil 2018: Vigilância de fatores de risco e
proteção para doenças crônicas por inquérito telefônico: estimativas sobre frequência
e distribuição sociodemográfica de fatores de risco e proteção para doenças crônicas
nas capitais dos 26 estados br. Brasilia Ministério da Saúde Brasilia; 2019. Available:
https://portalarquivos2.saude.gov.br/images/pdf/2019/julho/25/vigitel-brasil-2018.pdf
[Accessed 17 May 2021].
23 Brazilian Health Ministry (BHM). Vigitel Brasil 2019: Vigilância de fatores de risco e
proteção para doenças crônicas por inquérito telefônico: estimativas sobre frequência
e distribuição sociodemográfica de fatores de risco e proteção para doenças crônicas
nas capitais dos 26 estados br. Brasilia Ministério da Saúde; 2020. Available: https://​
bvsms.saude.gov.br/bvs/publicacoes/vigitel_brasil_2019_vigilancia_fatores_risc o.pdf
[Accessed 17 May 2021].