See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/228290899

The Role of the Middle Class in Economic Development: What Do Cross-

Country Data Show?

Article in SSRN Electronic Journal · January 2011

DOI: 10.2139/ssrn.1822783

CITATIONS READS

40 3,927

3 authors, including:

Natalie Lea Chun Mehmet A Ulubasoglu

Asian Development Bank Deakin University
21 PUBLICATIONS 388 CITATIONS 65 PUBLICATIONS 2,134 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Mehmet A Ulubasoglu on 11 October 2020.

The user has requested enhancement of the downloaded file.

Review of Development Economics
DOI:10.1111/rode.12265

The Role of Middle Class in Economic

Development: What Do Cross-Country Data Show?

Natalie Chun, Rana Hasan, Muhammad Habibur Rahman, and

Mehmet Ali Ulubasßo
glu*

Abstract
This paper investigates the channels through which the middle class may matter for consumption growth.
Using several different middle-class measures and a panel of 105 developing countries spanning the
period 1985–2013, we find that a larger middle class influences consumption growth primarily through
higher levels of human capital accumulation. There is also a significant direct effect of middle-class size
on consumption growth, which is more pronounced in the latter half of the sample, the 2000–2013 period.

1. Introduction
Do countries with a larger “middle class” grow faster? A number of economists
believe that the answer to this question is in the affirmative. For example, Kharas
and Gertz (2010) compare the growth experience of Brazil and South Korea and
suggest that the differential performance of the two countries can be explained by
differences in the relative sizes of the middle class in the two economies.1 Why
should a larger middle class help foster growth? Banerjee and Duflo (2008) identify
three arguments that are commonly made. The first argument is that the middle
class is where entrepreneurs that foster innovation and growth emerge from. A
second argument stresses middle-class “values” that encourage accumulation of
human capital and savings. A third argument suggests the consumption power of
the middle class leads to diversification and expansion of markets that allow for the
exploitation of economies of scale in production (see Murphy et al., 1989). In
addition, the middle class may play a key role in better governance, because unlike
the poor, the middle class may have the ability and power to demand better public
service delivery and greater accountability from public officials (Birdsall et al.,
2000). These arguments suggest that the presence of a strong middle class in a
country should have a significant positive influence on economic growth.
Nevertheless, while the middle class is often envisioned to have values,
investments and consumption patterns that differ relative to poor and high-income
classes, empirical examinations have been faced with the challenge of defining and
measuring the middle class. A number of relative and absolute measures have been

*Ulubasßo
glu (Corresponding author): Department of Economics, Deakin University, 70 Elgar

Rd, Burwood, VIC, 3125, Australia. E-mail: mehmet.ulubasoglu@deakin.edu.au. Chun and Hasan:
Economic Research and Regional Cooperation Department, Development Economics and Indicators
Division, Asian Development Bank, Manila, The Philippines. Rahman: Department of Economics,
Deakin University, Burwood, VIC, 3125, Australia. This paper represents the views of the authors and
does not necessarily represent those of the Asian Development Bank, its Executive Directors, or the
countries that they represent. The authors would like to thank Editor Andy McKay and two anonymous
reviewers for their detailed comments. Special thanks go to Anil Deolilakar and Jong-Wha Lee. Glennie
Amoranto provided excellent research assistance.

© 2016 John Wiley & Sons Ltd

2 Natalie Chun et al.

put forward based on per capita consumption measures. However, strong

differences in values are often not found between different economic classes and
especially between middle and upper classes (e.g. Amoranto et al., 2010; Lopez-
Calva et al., 2012). This has caused some researchers to suggest alternative cut-offs
such as low probabilities to vulnerability to define class status. This is based on the
finding that preferences and values of poor and vulnerable groups often differ from
middle and upper classes because of concerns about managing basic needs rather
than being able to focus on future investments (e.g. Lopez-Calva and Ortiz-Juarez,
2014). Others have provided evidence that perceived social mobility is a key factor
affecting middle-class values and behavior (Leventoglu, 2014). Moreover, many
debate the role of middle class in actually demanding better governance through
redistribution and greater public services (Rueschemeyer et al., 1992) with some
arguing that this only occurs when a regime transitions to democracy (Przeworski,
1992) or when social mobility is seemingly more restrictive (Piketty, 1995; Benabou
and Ok, 2001).
This paper seeks to examine the role of the middle class in economic growth in a
cross-country context. Using a structural growth model a la Mankiw et al. (1992)
and a panel of 105 developing countries covering the period from 1985 to 2013, we
explore whether the middle class is of direct importance to economic growth and/or
has a more indirect effect on economic growth through its impact on factor inputs,
including human capital, savings and labor force growth.2
Drawing on the debates centered on the definition and measurement of middle
class, our investigation utilizes three different middle-class definitions that are based
on both absolute and relative measures of consumption expenditure. The absolute
middle-class measure, defined as the share of the population living on US$2–$10
per person per day (in 2005 purchasing power parity (PPP) dollars), is similar to
Banerjee and Duflo (2008). It assumes that those living on more than US$2 per day
have a base amount of consumption that can contribute economically to growth.3
We also examine a relative measure, where the middle class is defined as the share
of households that have consumption expenditures between 75% and 125% of the
median expenditure, as in Birdsall et al. (2000). This concept captures the idea that
the middle class may be important not only for its consumption power, but also
through its ability to form a political or economic action group that demands and
implements policies that can contribute to market-oriented growth. It is related to
the theories of the median voter whose support is often sought in the financing of
productive public investments such as health or education. Finally, we examine
another relative measure used by Easterly (2001), which is defined as the
expenditure share held by the middle 60% of the expenditure spectrum. Examining
a diverse set of middle-class measures can provide evidence for the debate over the
type and notions of middle class that are really important to to economic growth in
a cross-country context.
We focus on developing countries for two reasons. First, the definition of middle
class varies across developing and developed countries. Income groups defined as
middle class in developing countries may belong to the poor class in developed
countries. For instance, Ravallion (2010) argues that while people living on US$2–
$13 per day can be treated as middle class in developing countries, they would be
treated as poor by US standards. Second, the growth path and factors that matter
to growth in developed countries are expected to differ fundamentally from the
growth path and factors influencing growth in developing economies. Including
developed countries in the sample makes sense for growth studies which use data

© 2016 John Wiley & Sons Ltd

 ROLE OF MIDDLE CLASS IN DEVELOPMENT 3

starting from the 1960s, or even the 1970s, but given the time frame of the available
middle class data, developed countries are likely to have already reached their
steady states, making our growth model less applicable to them.
Our key finding is that the middle class plays a role in driving economic
development primarily through the channel of human capital accumulation. We find
that, controlling for initial conditions and other factor inputs, there is a robust,
positive and significant relationship between the size of the middle class and
consumption growth through higher levels of secondary school enrolment rates. We
also find a direct, positive and significant relationship between the middle class and
consumption growth, which is more pronounced in the 2000–2013 period compared
with the 1985–1999 period. This finding is consistent with the arguments that the
globalized economy provided the middle class with new opportunities to thrive.
Finally, we document that the relative measure of the expenditure share held by
the middle 60% has a more economically significant and positive association with
consumption growth, schooling, and savings, compared with the other two middle-
class measures. This could be because the middle 60% of the expenditure spectrum
seems to be positioned in the higher ranks of the consumption distribution
compared with the other two segments that are just above the subsistence level of
consumption, which is also close to the median.
An important policy implication of our findings is that policies that factor in the
welfare of the middle class and nurture their growth are likely to be a more
effective long-term strategy for alleviating poverty compared with policies focusing
solely on the poor. One reason such policies may be more effective is that growth
that includes the middle class is likely to be more sustainable, given that more
people across different racial and ethnic groups share in the growth process
(Birdsall, 2010). A politically and economically strong middle class is more likely to
hold a government accountable, which would, in turn, ensure the rule of law,
protection of property rights and continued economic reform.

2. Related Literature
The cross-country literature on economic development has traditionally focused on
the poor and on poverty, while the interest in the middle class and the rich has only
intensified in the past decade (see Piketty and Saez (2006) in relation to the rich). The
inequality between income groups have also been of long interest to researchers, both
theoretically and empirically. For example, a number of papers study inequality’s
impact on economic growth, which yielded mixed empirical evidence.
One of the first cross-country studies focusing exclusively on the middle class is that
of Easterly (2001). Easterly’s work is based on the concept of a “middle-class
consensus,” which is defined as the co-existence of a higher share of income held by
middle-income groups and a lower degree of ethnic conflict in the society. Easterly
argues that the lack of a middle-class consensus leads to polarization in a society
paving the way for a struggle over resource endowments and ultimately resulting in
lower broad-based investments in human capital by the elite who hold the power.
More recently, Banerjee and Duflo (2008) have contributed to the growing
interest in the role of the middle class. Drawing upon the literature at large,
Banerjee and Duflo identify several distinct arguments about why the middle class
is important for growth. First, to the extent that entrepreneurs typically emerge
from the middle class, they generate increases in productivity and employment for
society. Second, “middle-class values” emphasize human capital accumulation and

© 2016 John Wiley & Sons Ltd

4 Natalie Chun et al.

savings, both of which are key inputs into economic growth. Third, the middle class
not only has the resources to consume more than the poor, but they are also willing
to pay a little extra for quality. The demand generated by the middle class, thus,
feeds into investment in production and marketing, and raises economy-wide
income levels, especially by allowing the exploitation of economies of scale in
production (Murphy et al., 1989).
Using household surveys from 13 developing countries to tease out the different
channels through which the middle class may promote growth, Banerjee and
Duflo’s analysis finds little support for the entrepreneurship channel. Instead, they
argue that the single most important characteristic of the middle class is that they
hold a steady job. The investment and consumption behavior supported by steady
jobs is what may enable the middle class to play a key role in spurring growth.
Further arguments can also be advanced for the role of the middle class in
economic growth. There is perceived to be an almost “natural” link between being
middle class and a citizenship trait that supports good governance, elimination of
corruption and rent-seeking activities, public investment in health, education and
infrastructure, trade openness, and modernization. Cross-country evidence suggests
that when there is a large number of people with income above US$10 a day it is
found that governance improves and there are increases in health and education
expenditures (Loayza et al., 2012). The median-voter theory, as it applies to policy
determination in the political economy literature, is most likely to be reliant on the
middle class. A stronger middle class is less likely to be associated with factions,
civil conflict, and political instability. Economic growth that is inclusive of a larger
middle class is also more likely to be sustained, precluding possible collapses that
would reverse the gains made.
Increased focus on the middle class has resulted in suggestions to shift the policy
orientation of some perennial development issues. For instance, middle-class-centric
approaches are suggested to have a greater impact in driving growth and reducing
poverty than policies specifically aimed at the impoverished (Birdsall, 2010). In fact,
using household survey per capita consumption growth models, Ravallion (2010)
finds that the size of the middle class, measured in absolute terms, has an important
connection to growth and poverty reduction. Given that projections suggest a large
rise in middle-class consumers through 2030 (Bussolo et al., 2009; Kharas and
Gertz, 2010; Kharas, 2010), it is expected that the middle class will play an ever
increasing and relevant role in development planning.

3. Middle Class: Definitions and Data

Defining the group of individuals that forms the middle class is far from easy and
encompasses a complex set of characteristics. For example, Merriam Webster
defines the middle class as follows:
a class occupying a position between the upper class and the lower class . . . a fluid
heterogeneous socioeconomic grouping composed principally of business and
professional people, bureaucrats, and some farmers and skilled workers sharing common
social characteristics and values.
The consumption role of the middle class has been highlighted in the literature.
Schor (1999) stated that it is a “new consumerism” that defines the middle class: a
constant, “upscaling of lifestyle norms; the pervasiveness of conspicuous, status
goods and of competition for acquiring them; and the growing disconnect between

© 2016 John Wiley & Sons Ltd

 ROLE OF MIDDLE CLASS IN DEVELOPMENT 5

consumer desires and incomes.” Likewise, Murphy et al. (1989) indicate that the
willingness of the middle class to pay a little extra for consuming quality goods and
services is significantly higher than the lower class. Hence, consumption expenditure
has played a greater role in defining the middle class in the economics literature.
Even restricting our attention to purely economic definitions based on income or
expenditure, there is little consensus on how to measure the middle class. Past
studies have varied in using relative and absolute measures. Moreover, even within
the class of absolute measures, there is little agreement on what the appropriate
thresholds should be for differentiating between different classes. For example,
Ravallion (2010) defines the middle class in developing countries as households
with expenditures between US$2 and $13 per person per day (in 2005 consumption
PPP dollars). The lower cut-off of US$2 represents the median value of the
national poverty line from a sample of 70 developing countries, while the upper
cut-off of US$13 represents the US poverty line in 2005 PPPs. Ravallion argues that
these bounds can be thought of as encompassing those who are not deemed poor
by the standards of developing countries, but are still poor by the standards of rich
countries. In contrast, Kharas and Gertz (2010) use thresholds of US$10 and $100
per person per day to define the middle class, while Milanovic and Yitzhaki (2002)
define the middle class as those living between the mean incomes of Brazil and
Italy, which is roughly between US$10 and $20 per day in 2005 PPPs. This
definition is also used by Bussolo et al. (2009). In contrast, Banerjee and Duflo
(2008) base their analysis on a definition of the middle class as individuals living on
US$2 to $10 per day.
Still others rely on the relative measures of middle class in an attempt to draw a
closer connection between political consensus and inequality. A common approach
is to define the middle class as those falling between 75% and 125% of the median
income, as in Thurow (1987) for the USA and Birdsall et al. (2000) for developing
countries. In contrast, Easterly (2001) uses the expenditure/income share of the
middle 60% as a measure of the middle class. Given there is a correlation of 0.90
between the middle 60% of the income distribution with the Gini coefficient, this
middle-class measure is very similar to a measure of inequality.
There are strong grounds to examine both the absolute and relative measures, as
they relate to different reasons why the middle class is important for economic
growth. Thus, we focus on three sets of middle-class definitions: (i) an absolute
measure, referred to as MC (US$2–$10), representing the share of the population
living on US$2–$10 per day in 2005 PPP dollars; (ii) a relative measure, referred to
as MC middle 60%, representing the share of the total consumption expenditure
accruing to the middle 60% of the expenditure distribution; and (iii) a relative
measure, MC median, representing the share of population that has expenditures at
least above US$2 per day and within 0.75–1.25 of the median expenditure of the
country. When using MC (US$2–$10), we include a measure for the upper class,
UC (US$10+), as a control and discuss the findings related to this class, where
necessary.
The main data source for constructing the middle-class measures at the country
level is the World Bank’s PovcalNet (World Bank, 2015a) dataset, which includes
tabulated per capita consumption distribution and mean per capita consumption
based on an extensive set of approximately 850 household surveys for 130
countries. The survey means are reported in 2005 PPP dollars.4
The correlations between the middle-class variables in our dataset are striking
(unreported in a table). MC (US$2–$10) and the MC median measure are highly

© 2016 John Wiley & Sons Ltd

6 Natalie Chun et al.

and positively correlated with each other (0.92). In contrast, the correlation
between MC (US$2–$10) and the MC middle 60% is only 0.11, while the
correlation between the MC median and MC middle 60% is 0.32. All these point to
a highly skewed consumption distribution where the median income is closer to
those who live on US$2–$10 in many economies than to those who comprise the
middle 60% of the consumption distribution.

4. Econometric Specification and Methodology

Our analysis of the relationship between consumption growth and the size of the
middle class is based on a general empirical formulation of the augmented Solow
growth model, where economic growth is determined by investment in human
capital, savings and labor force growth. This model is derived from a Cobb–Douglas
production function Y = AF (K, H, N) where Y is output, K is physical capital
stock, H is human capital, N is labor force and A captures total factor productivity
and other factors such as investment climate and institutional characteristics. We
follow the approach of Mankiw et al. (1992) and Islam (1995), who provide an
appropriate basis for estimating specifications according to the Solow model using a
panel of cross-country data to examine whether after controlling for the key factor
inputs, the middle class has an added effect in the model.5 We further examine
whether the middle class has an impact on factor inputs, that is, human capital,
savings and labor force growth.
We first start with the growth model. Specifically, we estimate consumption
growth models of the following form:
D ln Ci;tþr
¼a þ b ln Ci;t þ k ln Hi;t þ d ln Si;t þ w lnðni;t þ g þ dÞ
r (1)
þ u ln MCi;t þ li þ ei;t

where i denotes country, t denotes year, r represents the number of years till the
next household survey following the household survey in year t in our (unbalanced)
panel. C represents consumption per capita, Δln C/r is the average yearly growth rate
of consumption per capita between time t and t + r, H is human capital, S is savings
rate, n is labor force growth, g is technology growth and d represents the depreciation
rate of capital. MC denotes the middle-class measure, l country fixed effects and e the
error term. Our panel dataset is unbalanced owing to survey availability for each
country in the PovCal database.6
We measure growth with the change in the logarithm of consumption per capita.
Consumption is more tightly linked to the welfare of the populations, compared with
output or income. Our analysis of consumption also follows from the emphasis in the
middle-class literature on “new consumerism” and the demand for higher varieties for
consumption as well as better services arising from the middle classes. A further
reason for focusing on consumption is the comparability of the key middle-class
variables and consumption growth, as they are drawn from the same survey.7 To
proxy H, we use the secondary school gross enrolment rate of the total population
above 15 years of age.8 We measure S with the gross savings as percentage of
gross domestic product (GDP). As per the measurement of ln(n+g+d), we follow
the standard growth literature and assume that technology growth and the
depreciation rate have a joint value of 0.07. Therefore, ln(n+g+d) refers to labor force
growth rate plus 0.07.

© 2016 John Wiley & Sons Ltd

 ROLE OF MIDDLE CLASS IN DEVELOPMENT 7

The logged specification, which follows from Mankiw et al. (1992), captures the
country’s march towards the steady state. Note that all of the countries in our
sample are developing countries. Unlike developed countries that might have
already reached their steady-state equilibria developing countries still are
progressing on their trajectory towards their predicted steady-state equilibria. Thus,
our logged specification represents the notion that, conditional on initial per capita
consumption level and factor accumulation rates, countries with a larger middle
class are likely to approach their steady-state levels of consumption faster.
Our parameter of interest is φ, representing the direct effect of the middle class
on growth. In the specifications where we use an absolute measure for middle class,
we also include as a further control the size of the upper class. This allows us to
better differentiate between a story where the total magnitude of consumption
expenditures leads to greater economic growth versus one where the density or size
of a particular consumption expenditure bracket (for example, middle class vs
upper class) is what matters for economic growth.
To examine whether the middle class has an indirect effect on economic growth
through factor inputs I, we estimate human capital, savings, or labor force growth
equations using the following general form:
ln Ii;t ¼ g þ c ln Ci;t þ f ln MCi;t þ nXi;t þ li þ ti;t (2)

These regressions include, in addition to the middle-class measures, log per capita
consumption and a vector of controls, Xi,t. Specifically, we use the controls
commonly used in our factor input models, including urbanization, trade share in
GDP, democracy, the share of services relative to agriculture in GDP and the share
of industry relative to agriculture in GDP. These variables capture important
dimensions of the development process, as well as the factors that are important to
factor accumulation within a country. For example, it is generally accepted that
urbanization, trade openness, political freedom and sectoral growth are associated
with higher factor rewards, leading to higher factor accumulation rates. In other
words, economic activities spurred by higher levels of urbanization, trade openness,
political freedom and growth of the industry and services sectors are likely to lead
to the generation of better employment opportunities that result in higher wages,
profits and rents. We also include additional controls that may cause significant
variability in the observed levels of factor inputs. In particular, for human capital,
we control for the share of public spending on education as a share of GDP. For
savings, we include the number of assassinations and riots as measures of political
instability, log population (to proxy for the size of the domestic market), the share
of population above 65 and the log price of investment. For labor force input, we
include gender-specific measures of primary and secondary school enrolment rate
(see Barro and Sala-i-Martin, 2004).
We estimate equations (1) and (2) with ordinary least squares. There might be
endogeneity bias owing to omitted variables or reverse causation. Ideally, an
effective way of addressing all sources of endogeneity is a randomized experiment,
or an instrumental variable that approximates such an experiment. However, this is
a difficult task in cross-country data. Nevertheless, controlling for country-fixed
effects in equations (1) and (2) enables us to capture a myriad of time-invariant
effects, such as ethnic fractionalization, geography and religion, which might play
important roles in shaping consumption growth, factor accumulation and middle-
class development. Coupled also with an array of time-varying controls, our panel

© 2016 John Wiley & Sons Ltd

8 Natalie Chun et al.

fixed effects model is likely to limit any endogeneity bias arising from omitted
variables. To alleviate reverse causation, we used the lags of all explanatory
variables, but this approach delivered qualitatively similar results as those without
such lags, so we opt to use the contemporaneous values of explanatory variables.
Taken together, our regressions capture whether the variations in middle-class sizes
over time significantly correlate with the variations in the dependent variable of
interest, conditional on a wide range of time-varying and time-invariant controls.
This means that our reported effects are partial correlations rather than causation.
A final econometric consideration is that there could be a parameter instability
problem associated with the middle-class effects. One may argue that the late 1980s
and early 1990s was a period of structural adjustment for many low-income
countries when economies and the middle class collapsed. The time frame is
theoretically important because, depending on the severity of collapse, the impact
of the middle class on economic growth could be linear or nonlinear.9 To account
for this concern, in alternate regressions we revise equations (1) and (2) to include
an interaction term between MC and a time dummy that takes 1 for the 2000–2013
period and 0 for the 1985–1999 period.10 These regressions suggest that the
estimated middle-class effects may indeed differ across the two halves of our
sample period.
Data sources for each of the variables used in our empirical specifications are
provided in Table 1, while summary statistics of the key variables are provided in
Table 2.

5. Empirical Results

Does the Middle Class Have a Direct Effect on Per Capita Consumption Growth?
Table 3 presents the results of our consumption growth regressions. Consistent with
the convergence hypothesis that countries with higher inital level of development
grow more slowly, we find the expected negative sign for initial per capita
consumption, which is highly significant across almost all specifications. Columns
(1)–(3) introduce the three MC measures one at a time. While MC (US$2–$10) and
MC middle 60% are estimated to be statistically insignificant, MC median is
significant at the 5% level with a positive sign.
For a fuller judgment of the role of middle class in consumption growth, we next
control for factor inputs in columns (4)–(6). Focusing first on factor inputs, schooling
is estimated to be positive and highly significant. Similarly, the savings variable is also
positively related to consumption growth and is statistically significant in all models.
The coefficient on ln(n + g + d), has a negative sign, as predicted by the Solow model,
but is insignificant in all specifications. The negative coefficient estimate captures that
in the steady state a rapidly growing labor force tends to slow down per capita
consumption growth owing to the need for a greater amount of resources to be
dedicated to support the new entrants into the labor force. Given the amount of
physical capital and human capital available in the economy, higher labor force
growth results in less resources per person. However, the insignificant coefficient
estimate in all specifications suggests that, in developing countries, labor force growth
is not a significant factor constraining per capita consumption growth.
Considering the estimates for middle-class measures in the presence of factor
inputs in the model, we find that the coefficient estimate on the absolute
measure of MC (US$2–$10) is insignificant. This indicates that there is no added

© 2016 John Wiley & Sons Ltd

 ROLE OF MIDDLE CLASS IN DEVELOPMENT 9

Table 1. Data and Sources

Variable Description Source

C Mean of household consumption or income PovcalNet, World Bank

for the survey year are expressed in (2015a)
the prices of the International Comparison
Program (ICP) base year, and then
converted to PPP US dollars
H Gross secondary enrollment ratio (%); the PovcalNet, World Bank
total enrollment in secondary education, (2015a)
regardless of age, expressed as a
percentage of the population of official
secondary education age
S Gross savings (% of GDP) calculated based World Bank (2015b)
on gross national income less total
consumption, plus net transfers
n Labor force growth measured as an annual World Bank (2015b)
change in the proportion of the
population aged 15 and older
pi The price level of investment http://www.pwt.sas.upenn.edu
Democracy Institutionalized democracy, which is an Polity IV data base,
additive eleven-point scale ranged [0, 10] Marshall and Jaggers
(2005)
Assassinations Total number of assassinations in a given Banks and Wilson (2014)
year
Riots Total number of riots in a given year Banks and Wilson (2014)
Trade in GDP The sum of exports and imports of goods World Bank (2015b)
and services measured as a share of GDP
Urban % share of population living in urban areas World Bank (2015b)
Population as defined by national statistical offices
Industry/ The ratio of industry to agriculture value World Bank (2015b)
Agriculture added (constant 2005 US dollars)
Services/ The ratio of services to agriculture value World Bank (2015b)
Agriculture added (constant 2005 US dollars)
Public Total general (local, regional and central) World Bank (2015b)
Education government expenditure on education
Expenditure (current, capital, and transfers),
(% of GDP) expressed as a percentage of GDP
Population Total population World Bank (2015b)

effect on growth from a larger portion of the population having more than some
threshold amount of consumption (column 4). We estimate the direct effect of
MC middle 60% on growth to be positive. The coefficient estimate suggests that
1% higher MC middle 60% is associated with 0.12% higher consumption growth
(column 5), but this effect is only marginally significant with a t-statistic of 1.61.
Finally, MC median retains its significance at 5% with a positive sign when
factor inputs are controlled (column 6). The estimated coefficient indicates that a
1% higher MC median is related to 0.04% higher consumption growth,
suggesting a relatively low economic significance for those who are around the
median of the comsumption distribution.

© 2016 John Wiley & Sons Ltd

10 Natalie Chun et al.

Table 2. Summary Statistics of Key Variables

Variables Mean Max Min SD Obs.

Unbalanced panel
HH survey monthly per cap. 162.209 521.890 13.920 103.075 647
consumption mean (US$)
MC: HH survey US$2–$10 per 53.062 98.910 1.120 21.975 647
day 2005 PPPs (% of total)
UC: HH survey US$10–$20 per 13.215 85.960 0.100 14.660 647
day 2005 PPPs (% of total)
MC: Cons. share 20–60% of cons. 45.982 56.976 25.140 5.837 647
distribution (% of total)
MC: HH survey 75–125% of 22.558 67.220 1.020 11.944 647
median income (% of total)
School enrollment, secondary 65.735 110.484 5.357 25.668 465
(% gross)
School enrollment, secondary, 66.478 112.099 6.051 24.660 428
male (% gross)
School enrollment, secondary, 65.109 108.773 3.852 27.494 428
female (% gross)
Gross savings in GDP (%) 19.815 206.816 –22.109 12.644 589
Labor force growth rate –0.00003 0.048 –0.055 0.009 598
Democracy (higher values = better 5.067 10 0 3.530 612
democracy)
Government share in GDP (%) 10.337 66.901 1.903 6.124 545

Notes: SD, standard deviation. HH, Household; MC, Middle Class; UC, Upper Class.

Does the Middle Class Have an Effect on Factor Inputs?

Human capital Table 4 displays the results from the model of schooling and
middle class. We estimate all the three middle-class measures to be robustly
significant and positive across specifications with or without controls (columns 1–6).
Focusing on the results with controls for discussion (columns 4–6), the absolute
measure of MC (U$2–$10) has a positive effect on schooling. This indicates that
having a base amount of expenditure above subsistence is substantially important
for accumulating human capital. The coefficient estimate indicates that a 1%
increase in MC (US$2–$10) is associated with a 0.18% higher secondary schooling
enrolment rate. MC middle 60% has a significant positive effect on schooling with a
much higher coefficient estimate, suggesting that countries with 1% higher MC
middle 60% are associated with greater school enrolment rates in the order of
0.5%. Finally, MC median has a positive relationship with secondary school
enrolment rates; the coefficient estimate points to 0.23% higher schooling
enrolment rate for an additional 1% higher MC median. The significance of all the
middle-class measures lends credence to arguments that the middle class is
important to human capital accumulation, as they may have had policies
implemented that support greater investment in schooling.
In terms of control variables, a higher per capita consumption mean is positively
and generally significantly associated with schooling. Further, greater urbanization
and higher levels of democracy are also positively and significantly related to
schooling (unreported).

© 2016 John Wiley & Sons Ltd

 ROLE OF MIDDLE CLASS IN DEVELOPMENT 11

Table 3. Consumption Growth Regressions

Dependent variable: Change in Log Mean Consumption Per Capita

(1) (2) (3) (4) (5) (6)

Initial mean 0.140*** 0.130*** 0.165*** 0.166*** 0.157*** 0.186***

consumption (0.033) (0.017) (0.022) (0.034) (0.017) (0.022)
Middle class 0.019 0.011
US$2–$10 2005 (0.016) (0.015)
PPP (SM)
Upper class 0.0002 0.001
US$10+ 2005 (0.011) (0.011)
PPP (SM)
Middle 60% 0.106 0.123
income share (0.075) (0.076)
(Easterly)
Middle class 0.048** 0.040**
75–125% of (0.020) (0.019)
the median
School 0.060*** 0.058*** 0.059***
enrollment, (0.015) (0.015) (0.015)
secondary
(% gross)
Gross savings 0.0307*** 0.029*** 0.028***
(% of GDP) (0.0107) (0.011) (0.011)
ln(n + g + d) 0.004 0.009 0.003
(0.030) (0.030) (0.030)
Country fixed Yes Yes Yes Yes Yes Yes
effects
Observations 540 540 540 540 540 540
Number of 95 95 95 95 95 95
countries

Notes: Standard errors are reported in parentheses. All variables are in natural logs. Labor force growth
takes the growth rate of technology and rate of depreciation (i.e. g + d) into account at the rate of 7%.
***p < 0.01; **p < 0.05; *p < 0.1.

Savings Table 5 presents the savings regressions. We find that the absolute
measure of MC (US$2–$10) has a significant and positive relationship with
savings, but this effect, while significant without controls, becomes insignificant
when controls are included (columns 1 and 4). The MC middle 60% measure
exhibits a positive relationship with savings, however, this measure is also
insignificant at conventional levels (columns 2 and 5). The MC median is
positively associated with savings. The effect is significant at 5% without controls
but is only marginally significant with a t-statistic of 1.60 after adding the controls
(columns 3 and 6).
Among the control variables, log initial consumption mean is not robustly related
to savings. The only robustly significant controls are log population and the share of
population above 65 years of age, both with positive signs, and urbanization rate,
which has a negative sign (unreported).

© 2016 John Wiley & Sons Ltd

12 Natalie Chun et al.

Table 4. Schooling regressions

Dependent variable: Log of

Secondary School Enrollment Rate (% gross)

(1) (2) (3) (4) (5) (6)

Initial mean 0.710*** 0.506*** 0.341*** 0.395*** 0.236*** 0.072

consumption (0.072) (0.039) (0.046) (0.064) (0.040) (0.044)
Middle class 0.212*** 0.175***
US$2–$10 2005 (0.032) (0.027)
PPP (SM)
Upper class 0.123*** 0.102***
US$10+ (0.024) (0.020)
2005
PPP (SM)
Middle 60% 0.329* 0.504***
income (0.170) (0.137)
share
(Easterly)
Middle class 0.248*** 0.231***
75–125% (0.044) (0.035)
of the
median
Controls No No No Yes Yes Yes
Country fixed Yes Yes Yes Yes Yes Yes
effects
Observations 465 465 465 465 465 465
Number of 92 92 92 92 92 92
countries

Notes: Standard errors are reported in parentheses. ***p < 0.01; **p < 0.05; *p < 0.1. All variables are in
natural logs. Control variables include share of trade in GDP, urbanization, democracy, ratio of industry
to agricultural value added, ratio of services to agricultural value added and share of public education
expenditure in GDP.

Labor force growth Table 6 reports the ln(n + g + d) regressions. While MC (US$2–
$10) and MC median are negatively associated with labor force growth, the MC
middle 60% has a positive association (columns 1–6). Of these, the MC middle 60% is
significant at 10% with and without controls. The coefficient estimate suggests that a
1% higher MC middle 60% is associated with 0.20% higher ln(n + g + d). The
positive association implies based on the Solow model that the costs required to
endow the labor force with more resources are lower for those who hold the middle
60% of the consumption distribution, facilitating higher growth in labor force
participation. The factor underlying this sign could be the skill levels associated with
this group. That is, the characteristics of those in the middle 60% of the consumption
distribution have skills which give them more flexibility to move within the labor
force, compared with those who are just above the subsistence level or those around
the median.
Among the controls variables, initial mean consumption per capita is positive and
generally significant, while no other control variables are estimated to be significant
(unreported).

© 2016 John Wiley & Sons Ltd

 ROLE OF MIDDLE CLASS IN DEVELOPMENT 13

Table 5. Savings Regressions

Dependent variable: Log of Gross Savings (% of GDP)

(1) (2) (3) (4) (5) (6)

Initial mean 0.431*** 0.312*** 0.160* 0.224 0.096 0.006

consumption (0.144) (0.072) (0.092) (0.159) (0.102) (0.117)
Middle class 0.113* 0.036
US$2–$10 2005 (0.066) (0.071)
PPP (SM)
Upper class 0.077 0.065
US$10+ 2005 (0.054) (0.054)
PPP (SM)
Middle 60% 0.347 0.057
income share (0.319) (0.324)
(Easterly)
Middle class 0.210** 0.140
75–125% of (0.084) (0.088)
the median
Controls No No No Yes Yes Yes
Country fixed Yes Yes Yes Yes Yes Yes
effects
Observations 575 575 575 575 575 575
Number of 100 100 100 100 100 100
countries

Notes: Standard errors are reported in parentheses. ***p < 0.01; **p < 0.05; *p < 0.1. All variables are in
natural logs. Controls variables include share of trade in GDP, urbanization, democracy, ratio of industry
to agricultural value added, ratio of services to agricultural value added, the price level of investment,
number of assassinations, number of riots, log population, share of population above 65 and log price of
investment.

To summarize the results so far, we find some evidence that the middle-class
influences directly per capita consumption growth; however, we find even stronger
evidence that a larger middle class is associated with higher levels of schooling.
Given that human capital is a significant factor input that has a robust, positive and
significant relationship with consumption growth, we find that it is mainly through
this channel that the middle class has an effect on consumption growth. We also
find that the middle-class groups defined as those holding the middle 60% of the
consumption distribution have an economically stronger association with
consumption growth and factor accumulation, compared with those who are just
above the subsistence level of consumption or those around the median of the
consumption distribution.

Do the Relationships Differ between 1985–1999 and 2000–2013?

Our inference so far has been predicated on the full sample that covers the years
1985–2013. However, there are compelling reasons to think that the role of middle
class in consumption growth and factor accumulation might have differed across the
sample period given that many developing economies underwent structural changes
in the 1990s. Table 7 presents the results on consumption growth, schooling, savings

© 2016 John Wiley & Sons Ltd

14 Natalie Chun et al.

Table 6. ln(n + g + d) Regressions

Dependent variable: Log of Labor Force Growth

(1) (2) (3) (4) (5) (6)

Initial mean 0.011 0.054** 0.068** 0.017 0.057* 0.071*

consumption (0.050) (0.025) (0.033) (0.057) (0.032) (0.040)
Middle class 0.008 0.014
US$2–$10 (0.023) (0.025)
2005 PPP (SM)
Upper class 0.020 0.018
US$10+ 2005 (0.018) (0.019)
PPP (SM)
Middle 60% 0.192* 0.206*
income share (0.108) (0.112)
(Easterly)
Middle class 0.021 0.025
75–125% of (0.029) (0.030)
the median
Controls No No No Yes Yes Yes
Country fixed Yes Yes Yes Yes Yes Yes
effects
Observations 598 598 598 598 598 598
Number of 105 105 105 105 105 105
countries

Notes: Standard errors are reported in parentheses. Labor force growth takes the growth rate of
technology and rate of depreciation (i.e. g + d) into account at the rate of 7%. ***p < 0.01; **p < 0.05;
*p < 0.1. All variables are in natural logs. Control variables include share of trade in GDP, urbanization,
democracy, ratio of industry to agricultural value added, ratio of services to agricultural value added,
female primary schooling, female secondary schooling, male primary schooling and male secondary
schooling.

and ln(n + g + d) by interacting the middle-class measures with a time dummy that
takes 1 for the 2000–2013 period and 0 otherwise. All the regressions include
the controls in the respective regressions discussed above. The results are striking.
The direct relationship between all the three middle-class measures and
consumption growth is positive and strongly significant at 1% to 5% in the 2000–
2013 period. The coefficient magnitudes for MC (US$2–$10), MC middle 60% and
MC median are, respectively, 0.04, 0.26 and 0.05 (columns 1–3). The significance of
all three MC measures in the 2000–2013 period lends strong credence to the
viewpoints that argue for the role of middle class in demanding more product
varieties and driving the consumption growth in globalized economies.
Another important finding in Table 7 is in regard to the role of middle class in
schooling. We find that all the three MC measures are positively and significantly
related to secondary school enrolment rates in the first half of the sample.
However, the magnitude of the estimated relationships, while still positive, are
uniformly lower in the 2000–2013 period. Specifically, the estimated coefficients for
MC (US$2–$10), MC middle 60% and MC median for 1985–1999 are, respectively,
0.16, 0.55 and 0.19, while, for 2000–2013, they are 0.09, 0.20 and 0.09,
respectively (columns 4–6). The time dummy for 2000–2013 itself is positive and

© 2016 John Wiley & Sons Ltd

 Table 7. Disentangling the Time Effect: Periods of 1985–1999 and 2000–2012

Log of Log of Log of

Change in Change in Change in secondary secondary secondary Log of Log of Log of
log mean log mean log mean school school school gross gross gross
consumption consumption consumption enrol. enrol. enrol. savings savings savings ln ln ln
per capita per capita per capita rate rate rate rate rate rate (n+g+d) (n+g+d) (n+g+d)
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12)

Initial mean 0.224*** 0.201*** 0.236*** 0.337*** 0.202*** 0.079* 0.203 0.081 0.0171 0.018 0.059* 0.071*
consumption (0.033) (0.018) (0.021) (0.063) (0.039) (0.043) (0.160) (0.102) (0.118) (0.058) (0.032) (0.040)
Middle class 0.004 0.159*** 0.020 0.012
US$2–$10 (0.015) (0.026) (0.074) (0.026)
2005 PPP (SM)
Upper class 0.008 0.088*** 0.061 0.018
US$10+ (0.011) (0.019) (0.054) (0.019)
2005 PPP (SM)
Middle 60% 0.019 0.546*** 0.402 0.183
income share (0.077) (0.146) (0.352) (0.126)
(Easterly)
Middle class 0.034* 0.191*** 0.102 0.025
75–125% of (0.018) (0.03) (0.090) (0.031)
the median
Time dummy 0.067 0.927*** 0.076** 0.369*** 1.469*** 0.422*** 0.154 3.597*** 0.267 0.015 0.196 0.017
(1 if year (0.057) (0.253) (0.038) (0.119) (0.562) (0.084) (0.291) (1.214) (0.217) (0.094) (0.397) (0.064)
> 1999,
0 otherwise)
Middle class 0.035** 0.066* 0.058 0.005
US$2–$10 (0.015) (0.030) (0.074) (0.024)
2005
PPP(SM) 9
time dummy
ROLE OF MIDDLE CLASS IN DEVELOPMENT
15

© 2016 John Wiley & Sons Ltd

 Table 7. Continued
16

Log of Log of Log of

Change in Change in Change in secondary secondary secondary Log of Log of Log of
log mean log mean log mean school school school gross gross gross
consumption consumption consumption enrol. enrol. enrol. savings savings savings ln ln ln
per capita per capita per capita rate rate rate rate rate rate (n+g+d) (n+g+d) (n+g+d)
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12)

© 2016 John Wiley & Sons Ltd

Natalie Chun et al.

Middle 60% 0.261*** 0.349** 0.965*** 0.049

income (0.067) (0.146) (0.319) (0.104)
share
(Easterly) 9
time dummy
Middle class 0.048*** 0.102*** 0.115 0.005
75–125% (0.013) (0.027) (0.071) (0.022)
of the
median 9
time dummy
School 0.044*** 0.043*** 0.042***
enrollment, (0.015) (0.015) (0.015)
secondary
(% gross)
Gross savings 0.017 0.013 0.012
(% of GDP) (0.010) (0.010) (0.010)
ln(n+g+d) 0.009 0.016 0.012
(0.028) (0.028) (0.028)
Controls na na na Yes Yes Yes Yes Yes Yes Yes Yes Yes
Country fixed Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes
effects
Observations 540 540 540 465 465 465 575 575 575 598 598 598
Number of 95 95 95 92 92 92 100 100 100 105 105 105
countries

Notes: Standard errors are reported in parentheses. Labor force growth takes the growth rate of technology and rate of depreciation (i.e. g+d) into account at the rate
of 7%. ***p < 0.01; **p < 0.05; *p < 0.1. All variables are in natural logs. Controls for each regression are those reported in Tables 4–6.
 ROLE OF MIDDLE CLASS IN DEVELOPMENT 17

strongly significant, indicating that secondary school enrolments were higher

globally in that period. The question at this point is: why did the middle class
contribute to secondary schooling to a lesser extent in 2000–2013 than in 1985–
1999? To shed light on this question, we estimated models where the dependent
variables are tertiary school enrolment rate, the ratio of industrial to agricultural
value added, the ratio of services to agricultural value added and trade openness
(unreported). We find that MC middle 60% is positively and significantly related to
tertiary schooling, as well as the ratio of the industrial to agricultural value added
and the ratio of services to agricultural value added in the 2000–2013 period. The
MC (US$2–$10) and MC median are found to be significantly related only to the
ratio of services to agricultural value added in the same period. These results seem
to suggest that those who hold the middle 60% of the consumption distribution
went on to attain higher levels of schooling and/or obtained opportunities in the
industry and services sectors, while the other two groups sought opportunities in
presumably low-skill services sectors.
As per the role of middle class in savings, we estimate that only the MC middle
60% is positively and significantly associated with gross savings rate in the 2000–
2013 period (columns 7–9). This suggests that the population segments that hold the
middle 60% of the consumption distribution had the power to save more such that
the economies could invest in more in product varieties or perhaps in housing.
Finally, the role of middle class in ln(n + g + d) is found to be insignificant in both
halves of the sample period (columns 10–12).11

6. Conclusions
The definition and measurement of middle class and the mechanisms through which
the population segments that make up the middle class contribute to higher levels
of welfare have been hotly debated. To contribute to this debate, we investigate the
relationship between the middle-class size and consumption growth, secondary
school enrolment rates, savings rate and labor force growth, using a sample of 105
developing countries over the period 1985–2013. Drawing on diverse arguments in
the literature concerning the measurement of middle class, we utilize three different
middle-class indicators: an absolute measure defined as the share of the population
living on US$2–$10 per person per day (in 2005 PPP dollars), and two relative
measures, including the expenditure share held by the middle 60% of the
expenditure spectrum and the share of households that have consumption
expenditures between 75% and 125% of the median expenditure.
Our analysis documents four key conclusions. First, the middle class contributes
to economic development primarily through its contribution to factor inputs, chiefly
human capital. The relationship between all the three middle-class measures and
secondary school enrolment rates is found to be robust, positive and highly
significant. Schooling, in turn, is a robust factor input that is associated with higher
consumption growth, indicating higher welfare levels. Second, there is evidence that
a larger middle class measured by the expenditure share held by the middle 60% of
the expenditure spectrum is associated with higher levels of savings during 2000–
2013 period. Thus, savings constitute another mechanism for higher welfare levels
for this particular middle-class measure in the latter part of our sample. Third, we
find evidence that all three middle-class measures have a direct relationship with
consumption growth in the 2000–2013 period outside the factor inputs channel. This
result is consistent with the role attributed to middle classes in “new consumerism”

© 2016 John Wiley & Sons Ltd

18 Natalie Chun et al.

that suggests that the middle classes demand a higher quality and more diverse set
of products and are willing pay extra for better services. Finally, the economic
significance of the expenditure share held by the middle 60% of the expenditure
spectrum is consistently higher than the other two measures of middle class. This
could be because the middle 60% of the expenditure spectrum seem to be lying on
the higher ranks in the consumption distribution than the other two segments, who
are just above the subsistence level of consumption, which also seems to be in
proximity to the median of the consumption distribution.
Overall, our results are consistent with Banerjee and Duflo’s (2008) conclusion
that the middle class may matter for growth on account of their investments in
human capital that is facilitated by steady employment. These results fit well with
the view that a strong middle class is likely to lead to long-run development by
positively affecting the proximate causes of growth (Acemoglu et al., 2005). The
robust impact on economic growth through schooling is also consistent with the
increased demand for human capital during the industrialization process as
suggested by Galor’s (2005) unified growth theory, indicating where the demand
may originate from. Further, our results support the view that specifically targeting
the middle class may help in the fight against poverty, compared with policies that
solely aim to help the impoverished through investments specific to the poor. The
extra kick to consumption growth provided by a large middle class can translate
into higher economic growth rates, which in turn could reduce poverty through the
well-known regularity on economic growth and poverty reduction in the literature
(Ravallion, 2010; Birdsall, 2010).
What can governments do to spur middle-class growth? Policies that enhance the
business climate to facilitate labor productivity and the generation and growth of
stable, sufficiently well-paying jobs could strengthen the middle class (Haltiwanger
et al., 2014). In addition, social protection policies that help minimize the downside
risks associated with retirement savings can lead to more productive investments by
the middle-class groups. Moreover, fiscal policy that sets budget priorities by
weighing equity–efficiency trade-offs, assists in access to land and credit, and invests
in providing more equal access to quality health and education services can play an
important role in strengthening middle class (Estache and Leipziger, 2009; Roemer,
1998). Future research that identifies measurable and comparable indicators of
policies that can enhance the middle class across countries and time, can assist with
identifying key policies that are essential to supporting the middle class and
weighing equity–efficiency trade-offs.

References
Acemoglu, D., S. Johnson, and J. Robinson, “Institutions as a Fundamental Cause of Long-
run Growth,” in P. Aghion and S. Durlauf (eds), Handbook of Economic Growth. vol. 1A,
Amsterdam: North Holland (2005).
Amoranto, G., N. Chun, and A. Deolalikar, “Who are the Middle Classes and What Values Do
They hold?” Asian Development Bank Economics working paper series 229, Manila (2010).
Banerjee, A. and E. Duflo, “What is Middle-class About the Middle-classes Around the
World?,” Journal of Economic Perspectives 22, no. 2 (2008):3–28.
Banks, A. S. and K. A. Wilson, “Cross-national Time-series Data Archive,” Databanks
International, Jerusalem, Israel, available at: http://www.databanksinternational.com (2014).
Barro, R. and X. Sala-i-Martin, Economic Growth, Cambridge, MA, The MIT Press (2004).

© 2016 John Wiley & Sons Ltd

 ROLE OF MIDDLE CLASS IN DEVELOPMENT 19

Benabou, R. and E.A. Ok, “Social Mobility and the Demand for Redistribution: The POUM
Hypothesis,” The Quarterly Journal of Economics, 116 (2001):447–87.
Birdsall, N., “The (Indispensable) Middle-class in Developing Countries; or The Rich and the
Rest, Not the Poor and the Rest,” Center for Global Development working paper 207,
Washington, DC (2010).
Birdsall, N., C. Graham, and S. Pettinato, “Stuck in a Tunnel: Is Globalization Muddling the
Middle?” Brookings Institution Center working paper 14, Washington, DC (2000).
Bussolo, M., R. De Hoyos, and D. Medvedev, “The Future of Global Income Inequality,” in
A. Estache and D. Leipziger (eds), Stuck in the Middle: Is Fiscal Policy Failing the Middle-
Class? Washington, DC: Brookings Institution (2009).
Chen, S. and M. Ravallion, “The Developing World is Poorer Than We Thought, But No
Less Successful in the Fight Against Poverty,” Quarterly Journal of Economics 125
(2010):1577–1625.
Chun, N., “The Middle Class in Developing Asia,” Journal of Asian Business 25 (2012):27–47.
Datt, G., “Computational Tools for Poverty Measurement and Analysis,” Food Consumption
and Nutrition Division (FCND) discussion paper 50, International Food Policy Research
Institute, Washington, DC (1998).
Easterly, W., “Middle-class Consensus and Economic Development,” Journal of Economic
Growth 6 (2001):317–35.
Estache, A. and D. Leipziger, “Overview: Fiscal Policy, Distribution, and the Middle Class,”
in A. Estache and D. Leipziger (eds), Stuck in the Middle: Is Fiscal Policy Failing the
Middle Class? ch. 1, Washington DC: Brookings Institution Press (2009): 1–22.
Galor, O., “From Stagnation to Growth: Unified Growth Theory,” in P. Aghion and S.
Durlauf (eds), Handbook of Economic Growth. vol. 1A, Amsterdam: North Holland
(2005).
Haltiwanger, J., S. Scarpetta, and H. Schweiger, “Cross Country Differences in Job
Reallocation: The Role of Industry, Firm Size and Regulations,” Labour Economics 26
(2014):11–25.
Islam, N., “Growth Empirics: A Panel Data Approach,” Quarterly Journal of Economics 110
(1995):1127–70.
Kharas, H., “The Emerging Middle-class in Developing Countries,” OECD Development
Centre working paper 285, Paris (2010).
Kharas, H. and G. Gertz, “The New Global Middle-class: A Cross-over from West to East,”
in C. Li. (ed.), China’s Emerging Middle Class: Beyond Economic Transformation, ch. 2,
Washington, DC: Brookings Institution Press (2010): 32–52.
Leventoglu, B., “Social Mobility, Middle Class, and Political Transitions,” Journal of Conflict
Resolution 58 (2014):825–64.
Loayza, N., J. Rigolini, and G. Llorente, “Do Middle Classes Bring Institutional Reforms?,”
Economic Letters 116 (2012):440–44.
Lopez-Calva, L. and E. Ortiz-Juarez, “A Vulnerability Apporach to the Definition of the
Middle Class,” Journal of Economic Inequality 12 (2014):23–47.
Lopez-Calva, L., J. Rigolini, and F. Torche, “Is There Such as Thing as Middle Class Values?
Class Differences, Values, and Political Orientations in Latin America,” Center for Global
Development working paper 286, Washington, DC (2012).
Mankiw, G., D. Romer, and D. Weil, “A Contribution to the Empirics of Economic
Growth,” The Quarterly Journal of Economics 107 (1992):407–37.
Marshall, M. and K. Jaggers, “Polity IV Project: Dataset Users’ Manual,” Center for Global
Policy, George Mason University, Fairfax, VA, available at www.systemicpeace.org/
inscrdata.html (2005).
Milanovic, B. and S. Yitzhaki, “Decomposing World Income Distribution: Does the World
Have a Middle-Class?,” Review of Income and Wealth 48 (2002):155–78.
Murphy, K., A. Shleifer, and R. Vishny, “Industrialization and the Big Push,” Journal of
Political Economy 97 (1989):1003–26.

© 2016 John Wiley & Sons Ltd

20 Natalie Chun et al.

Piketty, T., “Social Mobility and Redistributive Politics,” The Quarterly Journal of
Economics 110 (1995):551–84.
Piketty, T. and E. Saez, “The Evolution of Top Incomes: A Historical and International
Pespective,” American Economic Review, Papers and Proceedings 96 (2006):200–05.
Przeworski, A., Democracy and the Market, Cambridge: Cambridge University Press (1992).
Ravallion, M., “The Developing World’s Bulging (but Vulnerable) Middle-class,” World
Development 38 (2010):445–54.
Roemer, J., Equality of Opportunity, Cambridge, MA: Harvard University Press (1998).
Rueschemeyer, D., E. Stephens, and J. Stephens, Capitalist Development and Democracy,
Chicago, IL: University of Chicago Press (1992).
Schor, J., “The New Politics of Consumption: Why Americans Want So Much More Than
They Need”. New Democracy Forum, Boston: Review (Summer 1999).
Thurow, L., “A Surge in Inequality,” Scientific American 256 (1987):30–37.
World Bank, PovcalNet, available at: http://iresearch.worldbank.org/PovcalNet/jsp/index.jsp
Washington, DC (2015a).
World Bank, World Development Indicators (WDI) Online, available at: http://
devdata.worldbank.org/dataonline/ Washington, DC (2015b).

Notes
1. Kharas and Gertz (2010) note that Brazil grew very rapidly from 1965 until 1980. By 1980,
only around 29% of its population was middle class and, between 1980 and 1996, its economy
stalled. In comparison, South Korea had 53% of its population considered as middle class in
the early 1980s and its economy continued to grow rapidly over the next decade until it had
reached high-income status.
2. Acemoglu et al. (2005, p. 388) argue that “factor inputs are not the determinants of
growth, they are growth” (italics in original).
3. Chen and Ravallion (2010) find that those who are consuming US$1.25 per person per day
in 2005 PPPs are on the developing country poverty line and represent the set of people who
are truly poor.
4. This allows constructing the entire distribution based on Lorenz curve parameterizations,
as detailed in Datt (1998).
5. One modification that we make to the Solow growth model is to use per capita
consumption growth instead of output growth. Likewise, savings replaces physical capital
investment in the model.
6. We utilize the period after 1985 because surveys for years prior to this date are sparse.
7. Clearly, an alternative welfare measure is output/income. The difference between income
and consumption in a simple two-sector economy is savings. We prefer to investigate the
relationship between middle class and savings and consumption separately, as opposed to
combining consumption and savings into income. Note, however, that the PovCal database
sometimes use income from the surveys when the data on consumption is unavailable (see
Chen and Ravallion, 2010). As indicated by Chun (2012), there is a high correlation between
income and consumption especially at lower levels, which thus make the measurement error
negligible.
8. We also used average years of schooling to measure H, which could arguably measure
better the stock of human capital owned by the workforce that is needed for economic
growth. Our results remained very similar with this measure compared with when we use
gross secondary school enrolment rate.
9. We thank an anonymous referee for having pointed this out to us.
10. Of the 647 survey data points obtained from PovCal, 231 (35.7%) belong to the 1985–
1999 period and 416 (64.3%) belong to the 2000–2013 period.
11. An alternative consideration in regard to sample composition could be that low-income
developing countries (e.g. Uganda, Bangladesh, Malawi) may feature different relationships
between the middle class and the four dependent variables of interest, compared with higher-

© 2016 John Wiley & Sons Ltd

 ROLE OF MIDDLE CLASS IN DEVELOPMENT 21

income developing countries (e.g. Chile, Turkey, Brazil). We estimated models similar to
those in Table 7 whereby the time dummy is replaced with a dummy on low- vs higher-
income developing countries. We do not find any significant difference between these two
sets of countries in the consumption growth, savings and ln(n + g + d) models. We find some
difference in schooling regressions using MC (US$2–$10) and MC median, which indicate
higher coefficient magnitudes in low-income countries, but this difference is significant only
at the 10% level. The results are available upon request.

© 2016 John Wiley & Sons Ltd

View publication stats