Does the Rise of the Middle Class Lock in Good Government in the Developing World?

April 28, 2015

It is manifest that the best political community is formed by citizens of the middle class, and that those states are likely to be well-administered in which the middle class is large, and stronger if possible than both the other classes [. . .] ; for the addition of the middle class turns the scale, and prevents either of the extremes from being dominant.

—Aristotle, Politics

A version of this essay was published in the European Journal of Development Research.

In his introductory essay in the European Journal of Development research, Juergen Wiemann concludes that the rise of the middle class in the developing world is a cause for optimism. Whether it is seems to me to depend on two factors in each developing country: the proportion of the population that meets a reasonable standard of being not only above some poverty line but sufficiently above that line to constitute a reasonably income-secure “middle class”; and the extent to which the middle class gets richer and bigger as a result of sufficient and sufficiently shared economic growth over the next 15 to 25 years. In other words, the middle class has to be big enough, and it has to continue (as it has in the last 15 years or so) to get richer and bigger still to warrant “optimism” about its role.

The country data I set out below on the two factors, namely the current size of the income-secure middle class and its likely future growth, suggest that optimism is indeed warranted for many of today’s middle-income countries. But it is not warranted for all of them, and especially not for most of the low-income countries of South Asia and sub-Saharan Africa — even if they continue to grow at the relatively healthy rates they have enjoyed in the last decade and more.

Before I unpack and explain the two factors (middle class is big enough and likely to get bigger), it is worth stating about what in particular should the rise of the middle class make us optimistic. My answer is, as Aristotle remarked, good politics: it is about the likelihood that governments in developing countries will become not only more capable in the administrative and organizational sense, but more responsive and accountable to their own citizens (and more “democratic”) in the political sense.[1] The theory, in short, is that a key benefit of the middle class is a greater likelihood of good politics and good government (what is often called in the development community “good governance”) — a benefit that is not inevitable but if the middle class is large enough compared to the poorer and the rich, is more probable than otherwise (Easterly 2001; Loayaza et al., 2012). The middle class wants good government, and where it is large enough, it is willing to pay taxes to finance public provision of collective goods (Moore, 1998; Alesina, Cozzi, and Mantovan, 2012). As Wiemann points out, to the middle class is generally attributed “not only a moderating role vis-à-vis political extremists, but also an interest in political democratization, in good and transparent governance, and respect of civil rights.” I would add that the middle class wants a government that maintains a level playing field in the economic arena, free of insider rents and privileges, capable of regulating effectively natural monopolies, and able to administer and enforce tax systems adequate to provide security, basic infrastructure, and other public and collective goods and services (Birdsall, 2010).

The middle class wants a government that maintains a level playing field in the economic arena.

The $10 identifier of an “income-secure” middle class

What do I mean by an income-secure middle class? I mean people who are not at risk of becoming poor because of some household shock: sudden unemployment; working capital lost to robbery, abusive police or local tax collectors; loss of an uninsured dwelling to fire or floods; a sick child; a failed crop. A simple proxy identifier of middle-class people in the developing world is those living in households in which daily income per capita is $10 or above (purchasing power parity, 2005, in 2010 US$). In several countries of Latin America, households with per capita income of $7 per person per day have a 20 to 25 percent risk of falling into poverty over four years (Lopez-Calva and Ortiz-Juarez, 2011); only at $10 per person per day does that risk fall to just 10 percent.

In the discussion below, I use $10 (household per person per day in 2005 purchasing power parity terms) as a crude cutoff to distinguish between households above the $10 threshold, which are likely to be income-secure and “middle class,” and households below $10, which are likely to be “strugglers” or the absolutely poor (Birdsall et al., 2014). The distinction is one that I emphasized in my keynote speech at the 2014 EADI conference in Bonn.[2] It is a critical one to the extent that many references to the rise of the middle class implicitly include the larger number of households in most developing countries which, though not poor, are not middle class either. They probably help increase growth since they represent growing consumer demand, including for local products and services. But they are not yet a secure middle class that thrives on a competitive economic and political system; they are not as likely to be a political force for better government as is the secure middle class.

The $10 a day threshold is useful because it is an absolute measure and therefore permits comparison across countries and over time in the size of the middle class. (An education measure would in principle be better as a proxy for an individual’s permanent income, but years of schooling is a poor measure of education, because so much education occurs outside of schooling and because quality of schooling varies so widely.)

At the same time, the $10 threshold is far from precise. Across countries, the minimum necessary daily income to be “income-secure” is likely to vary. It will be higher where publicly funded social insurance is not available; it is probably higher for the “illegal” or unrecognized migrants to urban areas in China who cannot access free public services. In countries outside Latin America, where household surveys measure consumption, not income, the “right” identifier could be as much as 25 percent lower; in India it might be lower as well if the purchasing power parity adjustment compared to countries of Latin America overstates the cost of living for those at or near $10.[3]

Whatever the right threshold, even the “secure” middle class is not immune to a major and prolonged economy-wide downturn. The middle class in Argentina was hit hard by that country’s problems in the 1990s and in the early 2000s, as has been the middle class in Greece, Italy and Spain since 2008 — and some are arguing the middle class in the United States. The distinction is between insecure strugglers living on $10 or less who are vulnerable to household-specific shocks (many subject to the anxieties that Mullainathan and Shafir, 2013, argue reduces their “bandwidth” for rational, self-interested decision-making) and the more secure $10-and-above group who are reasonably insured against such shocks. The latter are likely to have private home and health insurance, for example, and access to publicly financed unemployment insurance, in normal times, but in times of extended economy-wide recessions or the kind of structural changes in the labor market and tax systems that the middle class in the United States has experienced.

#1: The middle class has to be big enough

Table 1 shows the estimated proportion of people in the middle class for selected countries using a minimum of $10 a day as an identifier (and $50 a day or more to identify the “rich”). Overall, the proportion is about 35 percent in Eastern Europe and about 25 percent in East Asia and Latin America. But in Africa and South Asia, the regions where most people live in low-income countries[4], the proportion is much lower: less than 5 percent. Among the BRICS, it is about 35 percent in Brazil and could be under 5 percent in India[5].

Table 1

The $10 a day identifier is meant to separate people in the secure middle class from those in the larger group in the developing world who have escaped poverty defined using the international poverty line of $2 a day (and escaped extreme poverty of $1.25 a day or less), but are not secure from the risk of falling into poverty. Identified (crudely) as those living in households between $4 and $10 per person per day, they are variously referred to as the “strugglers” (by this author), the “vulnerable,” the “fragile middle class.” sometimes the “lower middle class.”[6] In Latin America, they tend to have primary schooling and to work in the informal sector; adults in $10-and-more households tend to have secondary schooling and are more likely to be paystub “employees” (Birdsall, 2012).

A big increase in the numbers of “strugglers” across the developing world (Edward and Sumner, 2014) is widely viewed as a good thing for growth in developing countries. The increase in their numbers is associated with increasing demand for consumer goods (refrigerators, scooters, TVs) and with the potential for developing countries to grow their domestic markets for manufactured goods and reduce their dependence on commodity exports to the advanced economies.

But it is less clear whether the struggler group is good for good government in the manner Aristotle invoked referring to the “middle class.” In the past, economically vulnerable urban populations have been associated in Latin America with populism (consider continuing support for Chavez and his successor in Venezuela) (Sachs, 1989) and with a yearning for the order and stability associated with authoritarian governments.

The distinction between the secure middle class and insecure and vulnerable strugglers seems to me important, as imprecise as it is. It does not reflect a difference in values and aspirations (Lopez-Calva, Rigolini, and Torche, 2012), But it does appear to reflect a difference in political views and demands.[7] The strugglers, who are less secure materially, will be particularly sensitive to policies and political decisions that are costly to them in the short run.[8] The secure middle class will have a longer planning horizon and be better able to focus on public and collective goods: the quality of schooling, health care, and roads.

The struggler group was probably at the heart of protests in Brazil in 2013 that were triggered by an increase in bus fares, which would matter for the $4 to $10 group but not for the secure middle class, who aremore likely to commute by car (Ali and Dadush, 2012). Members of the middle class no doubt participated because, along with their struggler counterparts, they were unhappy over corruption in politics and high public spending on World Cup soccer stadiums. (Kees Biekart, 2015, describes the Brazil protests in these two stages). In Chile, it is the struggler group that was probably important in protests over rising student debt being accumulated to pay tuition at inferior private colleges and universities. In Tunisia, the Arab Spring was triggered by the attention to one man’s dramatic protest (immolating himself) in reaction to police harassment that threatened his livelihood (seizure of his vegetable cart). Mohamed Bouazizi was probably a struggler.[9] In all these cases, the protests were a reaction to specific threats to a group’s material well-being — though they got traction and became extended, no doubt, because of broader and deeper citizen concerns of over $10 middle-income households.[10] In Hong Kong, in contrast, recent protests reflect more specific and overt resistance to political decisions in themselves— as in Istanbul in 2013 and in the United States during the 1960s civil rights and antiwar movements. In Hong Kong, close to 90 percent of the population lives on $10 a day or more. There as in the United States decades ago, it is more likely that along with students, it has been the secure middle class (and students who expect to join that class) who have dominated protests and protest movements.

Where the secure middle class is small, as in many countries of sub-Saharan Africa, it is also likely to be more economically dependent on government, either directly in the case of civil servants or indirectly in the case of employees of state-owned banks and other enterprises heavily dependent on government subsidies and political largesse (Birdsall, 2007). In these settings, middle-class households are more likely to be inside the room than out on the streets and to see their interests aligned not with the poor and strugglers but with the rich. As Baud, 2015, suggests in the case of urban India, the (upper) middle class is more likely, when it does act in civil society organizations, to align itself with the post-materialist views of the global “cosmopolitan” class (for example in supporting urban beautification); it does not necessarily focus on “pre-materialist” (my wording) demands for more widespread access to schooling, mass transportation, and public security that are priorities for the poor and the strugglers.

On the other hand, in middle-income countries where median income is higher and the secure middle class ($10 and above) is large as a proportion of a population, the middle class is in a better position to make demands of government for a level playing field such as business-friendly policies that encourage competition and elimination of hidden privileges for the crony capitalist elite. It is less likely to see its interests aligned with the “rich” and more likely to constitute willing taxpayers and a beneficent lobby for public goods that benefit everyone, including the below $10 poor and strugglers.

How big is big enough to have some chance of sustaining good government?

The experience of Brazil and Chile, where the proportion of the population living on $10 a day and more had reached 20 percent and 30 percent respectively by 2005 (Birdsall, 2012) and is now above 30 percent and 40 percent respectively, provides a possible example. In those countries the distributional effect of changes in the budget share dedicated to social programs in the 2000s was distinctly redistributional; in those social democratic regimes (where reductions in inequality were greater even than in populist regimes of Latin America over the last two, three decades) changes in the budget shares for social programs benefited the bottom three quintiles of the income distribution, were neutral for the fourth from the bottom quintile, and were negative for the richest top quintile.[11]

How did that happen? In Brazil during the Lula presidency, the increase in conditional cash transfers under the Bolsa Família program and in other social spending were not resisted by the secure middle class and may have been supported. Presumably that was in part because the Lula regime also hewed to disciplined macroeconomic policies that were good for growth and business and was, in its early years at least, viewed as reasonably effective in targeting the cash transfers to the truly poor and managing well the increase in social spending.

In contrast, the secure middle class is probably too small still in India, Bangladesh, Kenya, and Senegal to discover and organize its collective interest in a good government in which pro-growth and pro-poor policies are combined (table 1). Where the middle class is very small and highly dependent on government directly or indirectly, it is more likely to see its interests as aligned with rich insiders than with poor outsiders.

I am, in short, optimistic about development prospects in those countries where the $10-and-above middle class has already reached 20 to 25 percent of the population. That is a rough guess at a critical threshold, based on Brazil and Chile’s experience since the turn of the century, above which a country has a reasonable probability of locking in institutions of good-enough government. (It is not a guarantee. Consider Russia, for example, though even in Russia there is considerable resistance to the current political regime.) In those countries, good government (and all that implies about adequate “institutions”) has a better chance of being sustained once a large-enough and secure-enough middle class is in place.[12]

Good government has a better chance of being sustained once a large-enough and secure-enough middle class is in place.

Table 1 highlights selected large countries in which the middle class is at least 20 percent of the population.[13]

Still, that optimism has to be tempered. A large proportion of the developing world’s population (about 50 percent and growing as population growth is greater in low-income countries, particularly Africa) now lives in countries where less than 20 percent of the population is middle class. Many of these of course are what are labeled “fragile states” in the development community. The middle class is probably too small to be a factor in the demand for good government in those countries for several decades. China, which has a relatively large middle class in urban areas, but a tiny one in rural areas, is likely to see a continued increase in the proportion of middle-class households; even if growth slows dramatically from the 10 percent average over the last 30 years, it is still likely to be on the order of 5 percent or more barring an unforeseen and dramatic disruption, and the middle class is likely to benefit at least proportionately.[14] But India and most other countries of South Asia and sub-Saharan Africa still have relatively few middle-class people; in those countries those few are in the top 5 to 10 percent of their country distributions (in India 5 percent of the population is over 60 million people); many are likely to benefit from the current economic and political rules of the game, even where those rules are not friendly to competition and growth. Continued and shared growth in these countries depends heavily on one or another contingency: positive external conditions, a well-managed natural-resource windfall, or a period of good political leadership without Aristotle’s large middle class.

#2. The middle class has to keep getting richer and bigger

Will economic growth persist, and be sufficiently shared within countries, to ensure that the middle class gets richer and bigger in developing countries as a proportion of countries’ populations? That requires that countries continue to enjoy economic growth and that the growth they enjoy is sufficiently shared so that it continues to move people out of the struggler group into the middle class over the next several decades.

That is what has happened to visible effect in the last two decades. A now well-known figure constructed by Branko Milanovic (figure 1 — see Milanovic, 2012) shows the incidence of growth for each percentile of the global distribution of income (in which all people in the world are lined up by their income, independent of their country of residence) for the period 1998–2008. I have imposed on the figure the approximate per person daily income at key points in the curve. What the graph illustrates is that the biggest beneficiaries of growth in that decade, other than the 5 percent or so of the richest people, were the strugglers, namely those in the $4–$10 group. This is the group that moved out of poverty in that decade, as a result of high growth rates especially but not only in China and India. A smaller but significant proportion of the world’s population has moved above the $10 line into the income-secure middle class.

Figure 1

Table 2 shows projections of the proportion of countries’ (growing) populations that will be in the middle class (using the proxy of $10–$50) in 2030 and in 2050, assuming continued growth and continued sharing of the benefits of growth along the lines of the recent past.[15] The projections reflect country-specific growth rates over the next 35 years that are reasonably conservative, and they assume for every country that there is no change in the future in their current distributions of income or consumption.[16] They suggest that by 2030 all of today’s upper-middle income countries will have 35 percent or more of their populations in the middle class — all comparable to Brazil today. The increases slow in percentage terms in the subsequent two decades.

Table 2

See also footnotes [20] and [21]

In short, continued steady growth of economies over the next several decades would move people up the income ranks from poverty to struggler and from struggler to middle class in many countries, as long as that growth is shared sufficiently to avoid worsening inequality in any country. By 2050 most of today’s upper-middle-income countries could have well over 50 percent of their households in the middle class.

At the same time rapid growth (over 5 percent annually) for another 35 years in low-income countries like Tanzania and India is not enough to build a middle class that constitutes 20 percent a year even by 2050: only a few, including Egypt, Ethiopia, and urban India and urban Indonesia, will have even 20 percent of their population in the income-secure middle class, a proportion reached in about 2000, almost 50 years earlier, in Brazil, Chile, and Mexico. Figure 2 shows projections for India and Mexico.

Figure 2a

Figure 2b

The projections are cause for optimism that government in the broad sense will improve, at least in today’s middle-income countries, to the extent that two assumptions are correct: 1) a larger middle class has a greater interest in a responsive and accountable government and greater probability of having its collective interests reflected in political life, and 2) a larger middle class is more likely to support a social contract in which the taxes it pays are largely channeled to collective and public goods from which all, including the strugglers and the poor, benefit.

The question of course is whether the current set of middle-class countries and the emerging set including China, will continue to enjoy the (relatively rapid) growth that most have enjoyed in the last 10 to 15 years (and China and India for at least a decade before 2000).[17] With, at the time of writing, commodity prices falling as demand from China declines, and Europe (another big market) on the edge of deflation, the growth outlook for many developing countries is not good. However that is the short run. A second question is whether whatever growth there is will be distributed at least as well as it has been in recent years in most developing countries, benefiting the strugglers sufficiently to move them into the middle class. On this issue there is also cause for optimism; if reasonable rates of growth persist, inequality is unlikely to rise and could decline as it has in many countries of Latin America in the last decade (Lopez-Calva and Lustig, 2013), so that the trend of the poor moving into the struggler group and the strugglers moving into the middle class will continue.

Of course these are only projections. On the downside is the possibility that the commodity-driven growth of the last decade in Latin America, for example, benefited most those with fewer skills, reducing inequality but discouraging the upgrading of skills that will be needed to fuel continued growth in the future[18]; or the possibility that the second machine age[19], sometimes associated with robots and their use not only in industry but in personal services, will hurt the middle class in developing as well as in the most advanced economies; or that low-income countries will make poor use of the natural-resource windfalls many are expecting (Ghana, Uganda, Mongolia, Tanzania, East Timor), leading to deteriorating governance; or that unabated climate change will disrupt growth especially in those countries too far from whatever threshold proportion of middle-class households matters for good government.

A concluding reflection

The projections of the future size of the middle class in developing countries rely on country-specific growth rates (fixed starting in 2010) over 20 and 40 years, and on the assumption that there will be no change in each country’s current distribution of income/consumption over that entire period.

On the key argument of this paper — that the middle class is good for good government — the reader must come along with me on three other key assumptions. The first is that in the developing world there is, at about $10 of household income per person, a “security” line above which people can be called middle class, but below which they cannot; there is at least indicative evidence of such a threshold from several countries in Latin America, but the exact threshold will clearly vary by country. (The $10 threshold implies that in most developing countries the “middle class” is not anywhere near the middle of the distribution; it is currently crowded mostly in the top quintile, and in lower-income countries, including India and most countries of Africa, into the top decile or even the top ventile.)

The second assumption is more fundamental. It is that Aristotle’s observation was right, namely that it is the materially secure middle class that benefits from good government and thus demands it and helps sustain it by financing public goods (and the regulatory regime and contract enforcement and rule of law and so forth through willing payment of taxes).

The third assumption has to do with what constitutes a middle class “society.” I have roughly assumed that it is only when some minimum proportion — at least 20 percent and perhaps closer to 30 percent — of the population is rich enough and materially secure enough to be “middle class” that Aristotle’s connection between the middle class and good government becomes more likely and more sustained.

What are the implications? One implication is that today’s middle-income countries, assuming reasonable continuing growth and no increase in current inequality, are likely to become more and more middle-class societies in the next two decades, and to enjoy the political and economic stability associated with more democratic, participatory and accountable government. About these countries we can therefore afford to be reasonably optimistic.

But a second implication is that today’s low-income countries, even those that could enjoy another two decades of healthy growth, will not enjoy that benefit. They will continue to need better than average luck and leadership if they are to maintain or build more democratic, accountable, and participatory governments.

The lottery of birth will still matter greatly for individuals. There will just be more good-luck countries.

A third is that in 2050, there may still be an important divide in the world, between middle-class countries and societies with reasonably sound institutions and good government on the one hand, and countries where institutions are still fragile and most people are, in a material sense, poor on the other. The lottery of birth — the country in which individuals are born — will still matter greatly for individuals (barring dramatic easing of migration barriers). There will just be more good-luck countries from individuals’ point of view.

CGD is grateful for contributions from the Australian Department of Foreign Affairs and Trade, the William and Flora Hewlett Foundation, and the Lakeshore Foundation in support of this work.


Acemoglu, Daron, and James Robinson. 2013. Why Nations Fail: The Origins of Power, Prosperity, and Poverty. Reprint edition. New York: Crown Business.

Alesina, Alberto, Guido Cozzi, and Noemi Mantovan. 2012. “The Evolution of Ideology, Fairness and Redistribution.” Economic Journal 122 (565): 1244–61.

Ali, Shimelse, and Uri Dadush. 2012. “The Global Middle Class Is Bigger Than We Thought.” Foreign Policy, May 16.

Birdsall, Nancy. 2007. “Do No Harm: Aid, Weak Institutions, and the Missing Middle in Africa.” CGD Working Paper 113. Washington, DC: Center for Global Development. /publication/do-no-harm-aid-weak-institutions-and-missing-middle-africa-working-paper-113.

———. 2010. “The Rich and the Rest, Not the Poor and the Rest.” CGD Working Paper 207. Washington, D.C.: Center for Global Development. /publication/indispensable-middle-class-developing-countries-or-rich-and-rest-not-poor-and-rest.

———. 2012. “A Note on the Middle Class in Latin America.” CGD Working Paper 303. Washington, DC: Center for Global Development. /publication/note-middle-class-latin-america-working-paper-303.

Birdsall, Nancy, Nora Lustig, and Darryl McLeod. 2011. “Declining Inequality in Latin America: Some Economics, Some Politics.” CGD Working Paper 251. Washington, DC: Center for Global Development. /publication/declining-inequality-latin-america-some-economics-some-politics-working-paper-251.

Birdsall, Nancy, Nora Lustig, and Christian Johannes Meyer. 2014. “The Strugglers: The New Poor in Latin America?” World Development 60 (August): 132–46.

Brynjolfsson, Erik and Andrew McAfee. 2014. The Second Machine Age. New York: W.W. Norton & Company.

de la Torre, Augusto, and Julian Messina. 2013. “The trend reversal in income inequality and returns to education: How bad is this good news for Latin America?”,

Eduardo Levy Yeyati, Guillermo Beylis, Tatiana Didier, Carlos Rodríguez Castelán, and Sergio Schmukler. 2014. “Inequality in a Lower Growth Latin America.” LAC Semmiannual Report (October). Washington, D.C.: The World Bank.

Deaton, Angus. 2003. “Measuring Poverty in a Growing World (Or Measuring Growth in a Poor World).” NBER Working Paper 9822. Cambridge, MA: National Bureau of Economic Research.

Easterly, William. 2001. “The Middle Class Consensus and Economic Development.” World Bank.

Edward, Peter, and Andy Sumner. 2014. “The Geography of Inequality: Where and by How Much Has Income Distribution Changed since 1990?” CGD Working Paper 341. Washington, DC: Center for Global Development. /publication/geography-inequality-where-and-how-much-has-income-distribution-changed-1990-working.

Fukuyama, Francis. 2012. The Origins of Political Order: From Prehuman Times to the French Revolution. Reprint edition. New York: Farrar, Straus and Giroux.

Hausmann, Ricardo, Lant Pritchett and Dani Rodrik. 2004. “Growth Accelerations.” National Bureau of Economic Research Working Paper 10566. Cambridge, MA: National Bureau of Economic Research.

Kitschelt, Herbert, and Steven I. Wilkinson (eds.). 2007. Patrons, Clients, and Policies. Cambridge: Cambridge University Press.

Lakner, Christoph, and Branko Milanovic. 2013. “Global Income Distribution: From the Fall of the Berlin Wall to the Great Recession.” World Bank Policy Research Working Paper 6719. Washington, D.C.: The World Bank.

Loayza, Norman, Jamele Rigolini, and Gonzalo Llorente. 2012. “Do Middle Classes Bring Institutional Reforms?” World Bank Policy Research Paper 6015. Washington, D.C.: The World Bank.

Lopez-Calva, Luis Felipe, and Eduardo Ortiz-Juares. 2011. “A Vulnerability Approach to the Definition of the Middle Class.” World Bank Policy Research Working Paper 5902. Washington, D.C.: The World Bank.

Lopez-Calva, Luis Felipe, Jamele Rigolini and Florencia Torche. 2012. “Is there such a thing as middle class values? Class differences, values and political orientations in Latin America.” Discussion Paper series, Forschungsinstitut zur Zukunft der Arbeit, No. 6292,

Lustig, Nora, and Darryl McLeod. 2009. “Are Latin America’s New Left Regimes Reducing Inequality Faster? Addendum to Poverty, Inequality and the New Left.” Washington, D.C.: Wilson Center.

Meyer, Christian, and Nancy Birdsall. 2012. “New Estimates of India’s Middle Class: Techncial Note.” Washington, D.C.: Center for Global Development. /doc/2013_MiddleClassIndia_TechnicalNote_CGDNote.pdf.

Milanovic, Branko. 2012. “Global Income Inequality by the Numbers: In History and Now.” World Bank Policy Research Paper 6259. Washington, D.C.: The World Bank.

Mullainathan, Sendhil, and Eldar Shafir. 2013. Scarcity: Why Having Too Little Means So Much New York: Times Books.

Pritchett, Lant, and Lawrence H. Summers. 2014. “Asiaphoria Meets Regression to the Mean.” NBER Working Paper 20573. Cambridge, MA: National Bureau of Economic Research.

Reddy, Sanjay G., and Thomas Pogge. 2010. “How Not to Count the Poor.” In Sudhir Anand, Paul Segal, and Joseph E. Stiglitz (eds.): Debates on the Measurement of Global Poverty. Oxford: Oxford University Press

Sachs, Jeffrey D. 1989. “Social Conflict and Populist Policies in Latin America.” NBER Working Paper 2897. Cambridge, MA: National Bureau of Economic Research.

Weitz-Shapiro, Rebecca. 2009. “Choosing Clientelism: Political Competition, Poverty, and Social Welfare Policy in Argentina.” APSA 2009 Toronto Meeting Paper.

Weitz-Shapiro, Rebecca. 2012. What Wins Votes: Why Some Politicians Opt Out of Clientelism. American Journal of Political Science 56 (July): 568-583.

Wiemann, Jürgen. 2015. “Sharing Global Responsibility. The role of the middle classes on the way to a just and sustainable global economy.” The European Journal of Development Research 27(2): 205–216.

World Bank. 2014. “New Country Classifications | Data.” Accessed December 2.


The growth projections behind Table 2. and Table 3 .

Birsdall, Lustig, and Meyer (2014) use global income and consumption distribution data for 2005 from the World Bank’s World Income Distribution database[22]. The shape of the underlying income distributions is assumed to be constant over time. The growth projections up to 2050 come from the Centre d’Etudes Prospectives et d’Informations Internationales (CEPII)[23]. Their growth forecasts are matched to the population growth forecasts of the United Nations.

Table A1 (below) shows the assumed average per capita growth rates for the countries in Table 2.

Table A1

[1] See Acemoglu and Robinson, 2013 and Fukuyama, 2012.

[2] For a short version outlining the distinction, see the blog post entitled “’Who You Callin’ Middle Class?’ A Plea to the Development Community.” Center for Global Development. April 18, 2014. /blog/who-you-callin%E2%80%99-middle-class-plea-development-community

[3] In Mexico, a comparison of consumption and income survey results from 2012 suggests that at $10 a day of household income per person, consumption is equivalent to about $7.50 a day. Author’s estimate based on the Encuesta Nacional de Ingresos y Gastos de los Hogares from Mexico. In the tables below, all country data is identified as based on either income or consumption surveys of households. Regarding India especially, there is considerable controversy concerning the purchasing power parity numbers; see Deaton and Heston (2008) and Chen and Ravallion (2008).

[4] Using World Bank classification of low-income countries (World Bank, 2014). The regional averages are weighted by country.

[5] Table 1 is based on PovcalNet data. Birdsall (2012) reported 32 percent for Brazil based on 2009 data; Meyer and Birdsall (2012) reported just under 6 percent for India, based on the Indian National Sample Survey in 2009/10.

[6] See “’Who You Callin’ Middle Class?’ A Plea to the Development Community.” Center for Global Development. April 18, 2014. /blog/who-you-callin%E2%80%99-middle-class-plea-development-community ., which includes further citations on this subject.

[7] See Loayza, 2012; they show using cross-country data an association between the size of the $10-$50 middle class in developing countries and public spending on social progams as well as lower corruption and indicators of governance.

[8] See Weitz-Shapiro 2009 and 2012; and Kitschelt and Wilkinson (eds.), 2007. These and others report that as incomes rise, clientelism tends to decline.

[9] See Birdsall et. al 2014,

[10] Demonstrations that are triggered by a single event affecting the insecure strugglers can put larger issues like corruption, inequality, and access to education into the spotlight; this may be what happened in Egypt at the time of the Arab Spring.

[11] See Figure 8: Redistributive impact of changes in social spending budget share by quintile in Birdsall, Lustig, and McLeod, 2011. See also Lustig and McLeod, 2009.

[12] Hausmann, Pritchett and Rodrik, (2004) show that growth can accelerate for many reasons, but is more likely to persist beyond a few years where institutions are better (using conventional measures of “institutions”).

[13] Note that even if these are underestimates for countries where data are for consumption, few of those with less than 20 percent middle class now would have more than 20 percent even if their numbers were increased by 25 percent –as the Mexico comparison (see footnote 3) might suggest.

[14] However a recent paper of Pritchett and Summers, (2014) emphasizes how rare it has been for high country growth rates as in China to persist rather than converge. Despite that projections invariably extrapolate, as these do, from recent rates.

[15] The countries where the middle class will comprise at least 20 percent of their total population by 2030 and by 2050 represented close to 50 percent of the developing world population in 2010.

[16] These projections were originally developed for Birdsall et. al (2014), which provides details on their construction. The relevant tables are included in the CGD Working Paper version: /sites/default/files/new-poor-latin-america_1.pdf

[17] See footnote 12.

[18] de la Torre and Messina, 2013.

[19] Brynjolfsson and McAfee, 2014.

[20] India’s survey-measured consumption appears to be particularly low compared to national accounts-based consumption estimates vis-à-vis other countries, which indicates an overestimation of poverty and underestimation of the middle class (Deaton 2003). Reddy and Pogge (2010) argue that consumption PPPs may understate the true consumption of the very poor, which implies that there could be an understatement of the total consumption of the poor and the struggler groups in India relative to countries with a larger share of their population living on at least $10 a day. Together these could help explain the low estimate of the size of the middle class in India in Table 2. Meyer and Birdsall (2012) estimate the size of the Indian middle class to be at 6 percent of the total population (see footnote 5).

[21] The projections from Birdsall, Lustig, and Meyer (2014) use 2005 as their baseline year, which is responsible for the discrepancies between the estimates for the share of the middle class in Table 1 and Table 2. The country-specific distributions of income are as of 2005. The country-specific growth rates are the actual growth rates for 2005-2010 and from 2010 onwards are those as listed in the Appendix.

[22] This harmonized global dataset of household consumption and income surveys, compiled by Branko Milanovic (2010), is freely available at

[23] The model is fitted with publicly available data and uses a transparent methodology, see for a full description

Rights & Permissions

You may use and disseminate CGD’s publications under these conditions.