Why Inequality Matters:

Some Economic Issues

Nancy Birdsall*

Introduction

High and growing inequality is the case not only in some industrialized countries, but

in many of the world’s developing countries and transitional, post-Communist

economies. This essay is about why economic inequality matters in the developing and

transitional economies—which by the middle of this new century will make up almost

95 percent of the world’s population.

Whether and why inequality matters is an old issue—going back to Condorcet

and Marx.1 Yet for most of the postwar period mainstream economists as a group have

not concerned themselves much with inequality. The assumption of textbook economics

is that inequality is likely to enhance growth by concentrating income among the rich, who

save and invest more, and by creating a necessary incentive for individuals to work hard.

Thus Finis Welch entitled his 1999 (Richard T. Ely) address to the American Economics

Association “In Defense of Inequality.” For mainstream economists, inequality has been

at worst a necessary evil and at best a reasonable price to pay for growth.

For development economists concerned primarily with the world’s poorer coun-

tries, the central issues have been growth and poverty reduction. Until the last few years

inequality was of sporadic interest at best—an unfortunate but immutable (and thus not

policy-relevant) outcome of growth.2 In the postwar period, mainstream development

economists were no doubt influenced by a Cold War environment in which talk of inequal-

ity invoked concerns about communism, statism, or at least fiscally imprudent populism.

Only in the 1990s did there emerge new thinking on inequality as a possible cause of low

growth—and thus a phenomenon that mattered, at least for understanding growth.

Income inequality matters if people (and nations, in the case of inequality

across countries) care about their relative income status. Inequality of absolute

* I am grateful to Christian Barry, Carol Graham, Steven Holmes, and participants in a Mt. Holyoke

conference, as well as anonymous referees, for comments on an earlier version.1 See Richard H. Tawney, Equality (New York: Barnes & Noble, 1952).2 Simon Kuznets, “Economic Growth and Income Inequality,” American Economic Review 45, No. 1

(1955), pp. 1–28; and Hollis Chenery and Moises Syrquin, Patterns of Development, 1950–1970 (New

York: Oxford University Press, 1975).

Reprinted from Ethics & International Affairs 15, no. 2.© 2001 Carnegie Council on Ethics and International Affairs

income may in itself bother people—both those at the low end, who may resent the

better-off, and those at the high end, who may enjoy their own affluence less if others

are visibly worse off. Reducing inequality may therefore be an end in itself for some

people and in some societies. There are also instrumental reasons why inequality mat-

ters—reasons that should concern us even if we neither care about relative income sta-

tus nor feel that relative status is a fundamental concern of development.

In this essay I discuss three instrumental reasons why inequality matters in

developing countries.

• Inequality can inhibit growth and slow poverty reduction.

• Inequality often undermines the political process: that may lead to an inade-

quate social contract and may trigger bad economic policies—with ill effects

on growth, human development, and poverty reduction.

• Inequality may undermine civic and social as well as political life, and inhib-

it certain kinds of collective decision-making; at the societal level it may also

generate its own self-justifying tolerance, perpetuating a high inequality equi-

librium despite the potential economic and political costs.

My concern is thus not with income inequality in itself but with its consequences,

in particular its consequences for how economies and societies operate and thus affect the

well-being of people. Does income inequality at the societal level reduce an economy’s

growth and make it more likely that poverty will increase or persist? Does income inequal-

ity lead to inequality in other domains (individual opportunities, political life)?3

I concentrate throughout on inequality within developing countries and tran-

sitional economies, but most of the points I raise apply in more advanced economies

as well. After defining inequality, poverty, and equity and illustrating some trends, I

present three reasons why inequality matters. I close with brief reflections on the

implications of inequality for the future and thus for policies on the part of develop-

ing countries and the international donor community.

D E F I N I T I O N S

Inequality

Inequality is of course a relative concept. This essay is about why “money” inequality mat-

ters, that is, inequality in income or in consumption (expenditures) per person. Money

inequality is not in itself necessarily a bad (or a good) thing. Other forms of inequality—

for example inequality of education or health or capacity to participate in political life—

Nancy Birdsall4

3 See Amartya Sen, Inequality Reexamined (Cambridge: Harvard University Press, 1992).

are more often viewed as bad in themselves. Most societies aim for equal outcomes in these

areas because they are important indicators of well-being in themselves. Income inequal-

ity is different from inequality of wealth, which is everywhere more concentrated; to the

extent that inequality affects policy through politics, as I suggest below, inequality of

wealth is probably as or more important than inequality of income or consumption.

Poverty

In contrast to money inequality, poverty as I use it here is not a relative concept, but refers

to the condition of people who are poor in absolute terms. The simplest and most wide-

ly used standard below which a person is defined as poor is a minimum money income

or command over consumption; the World Bank uses the standard of $1 a day (in 1985

dollars). More complete measures go beyond money-defined poverty to include below-

minimum access to health care and education, decent housing, and the like, as well as the

lack of capacity to use these resources.4 An example is the United Nations Development

Programme’s Human Development Index.5 But there is no widely accepted absolute min-

imum of these basic needs analogous to the $1 a day standard. Recently attention has

gone to even broader measures that would, for example, include as poor those who lack

resources to participate in civic or political affairs and to protect their own legal or other

rights6—though again, a minimum standard is hard to define across societies.

Across countries, there is at best a weak association between poverty and

inequality, as Figure 1 (p. 6) illustrates. Some poor countries such as India have rela-

tively low inequality; some rich countries such as the United States have relatively high

inequality. The association alone provides no insight into causal links.

Equity

I use the term equity, or social equity, as in an “equitable” society, to refer to the idea of jus-

tice or fairness in the processes that lead to outcomes such as income, as opposed to “equal”

outcomes or income per se. A society with relatively high income inequality might be an

equitable society if the observed inequality were the outcome of an entirely fair process—in

which all individuals enjoyed the same opportunities, but some worked harder or took more

economic risks with resultant greater economic gains than others.7 An equitable society can

have many poor people, if resources are limited but fairly shared. Many pre-industrialized

W H Y I N E Q UA L I T Y M AT T E R S 5

4 Amartya Sen, Development as Freedom (New York: Knopf, 1999).5 United Nations Development Programme, Human Development Report 2001, Making New

Technology Work for Human Development (New York: Oxford University Press, 2000).6 United Nations Development Programme, Human Development Report 2000, Human Development

and Human Rights (New York: Oxford University Press, 1999); and World Bank, World Development

Report 2000/01: Attacking Poverty (New York: Oxford University Press, 2000).7 See Amartya Sen, “Possibility of Social Choice,” American Economic Review 89, No. 3 (June 1999),

pp. 349–78.

societies were in this sense equitable.8 Equity may be a more satisfactory concept than equal-

ity from a normative point of view, in the sense that more of it is unambiguously a good

thing. However, equity is harder to measure. I argue below that money inequality can com-

plicate a developing country’s efforts to achieve desired outcomes, such as participation in

political and civic life, which some associate with a society’s degree of equity.

I N E Q UA L I T Y : L E V E L S A N D T R E N D S

Measures of income inequality are only as good as the data on income (or consumption)

on which they are based. Most of the inequality data referred to here come from house-

hold surveys that are reasonably comparable.9 However other evidence suggests that

income may be understated at the low end of the distribution, where much of it comes

from informal work and subsistence agriculture; and it may be even more understated at

the high end of the distribution, where respondents have incentives to underreport income

from work and may underreport or fail altogether to consider income from wealth.

Nancy Birdsall6

8 See Peyton H. Young, Equity: In Theory and Practice (Princeton: Princeton University Press, 1994). 9 Klaus Deininger and Lyn Squire, “A New Data Set Measuring Income Inequality,” World Bank

Economic Review 10, No. 3 (Sept. 1996), pp. 565–91.

Figure 1: Poverty and Inequality

0

10

20

30

40

50

60

70

80

90

100

0 5 10 15 20 25 30Inequality Ratio (ratio of household income per capita, richest 20% to poorest 20%)

Pove

rty

Hea

dco

unt

Rat

io,(

%o

fpo

pula

tio

nbe

low

$1pe

rda

yin

com

e)

Zimbabwe

BrazilChina

Uganda

IndiaKenya

Chile

Nigeria

Mexico

Russia

Compiled from:World Bank, World Development Indicators (CD-ROM, 1999); Klaus Deininger and Lyn Squire, in World Bank Economic Review (Sept. 1996)

◆

◆

◆

◆◆

◆

◆

◆

◆◆

◆

◆

◆

◆◆

◆◆◆

◆◆

◆ ◆

◆

◆

◆

◆

◆

◆ ◆

◆

◆

◆

◆

◆

◆

◆

◆

ZambiaGuinea-Bissau

Figure 2 summarizes information over time on inequality across regions,

including the advanced industrial economies, using the Gini coefficient.10 Two points

are noteworthy. First, there are marked and relatively stable differences across regions,

with Latin America and Sub-Saharan Africa the most unequal. Income inequality

tends to be higher in the developing world, and it is obviously more painful as average

income levels are lower and the poor are therefore much poorer. This makes any effect

of inequality on poverty particularly relevant for poor countries. In Europe, the

United States, and Japan the richest 20 percent of households are between five and ten

times as rich as the poorest 20 percent (not shown). In most countries of Latin

America, as well as in the Philippines, Thailand, Malaysia, and nine of the eighteen

countries of Africa for which data are available, the richest 20 percent of households

are more than ten times richer than the poorest 20 percent. In Brazil and South Africa,

the two countries with the highest measured inequality in the world, the richest house-

holds are more than twenty-five times richer than the poorest. In South Asia and

China, inequality is lower but the poor are much poorer. Second, within regions and

W H Y I N E Q UA L I T Y M AT T E R S 7

10 Among the many measures of income distribution, the Gini coefficient is the most sensitive to income

differences in the middle of the distribution. Other measures such as the ratio of the richest to poorest decile

or quintile of individuals or households are more sensitive to the bottom and the top of distributions.

Figure 2: Inequality by Region and Decade

20

25

30

35

40

45

50

55

1960s 1970s 1980s 1990sYears

GIN

Ico

effic

ient

deca

dala

vera

ge

Eastern Europe South AsiaOECD & High Income East Asia & PacificSub-Saharan Africa Latin America & Carribean

◆ ◆ ◆

◆

◆

Compiled from Deininger and Squire (1996)

more so within individual countries, inequality does change over time (the decadal

and regional averages in the figure minimize fluctuations); large increases in inequali-

ty in many countries of Eastern Europe between the 1980s and 1990s are one exam-

ple. Some inequality may be inevitable, but its level is not immutable.

Inequality Matters 1: It Inhibits Growth and Poverty Reduction

Sustainable economic growth is a goal for developing countries, both because it can

improve everyone’s lives and because it is necessary (if not sufficient) for reducing

poverty.11 Even if we do not care about income inequality (or its avoidance) as an end

in itself, we are likely to care about its effects on growth and on the income level of the

poor. Low income makes education and health and other indicators of well-being

more difficult to achieve for the poor. In addition, low income may undermine more

directly people’s sense of well-being; poor people when asked tend to name a good job

and steady income as their highest priority.12

TEXTBOOK ECONOMICS: THE MARKET WORKS AND INEQUALITY ENHANCES GROWTH

The assumption of textbook economics is that a tradeoff exists between augmenting

growth and reducing inequality. Kuznets and others explored causality in one direc-

tion—that growth in the early stages of development might cause inequality. The

assumption about the other direction of causality—that inequality would enhance

growth—is grounded in two simple observations. The first is that a high level of sav-

ings and resulting investment is a prerequisite of rapid growth, so that income must be

concentrated in the hands of the rich, whose marginal propensity to save is relatively

high compared to that of the poor.13 A related point is that large investments in infra-

structure and industry are critical to development; in the absence of a deep and well-

functioning capital market, wealth and income need to be concentrated to generate the

necessary minimum resources. (With this in mind, many developing countries

embraced the need for the state to assume the commanding heights of the economy

and to use tax resources for state-financed industrial investments—this approach in

some settings almost certainly, and ironically, led to concentration of income.)

Nancy Birdsall8

11 World Bank, World Development Report, 2000/01; see also David Dollar and Aart Kraay, “Growth

Is Good for the Poor,” World Bank Paper No. 2587, April 12, 2001. 12 Deepa Narayan, Voices of the Poor, Vol. 1. Can Anyone Hear Us? Voices from 47 Countries

(Washington, D.C.: World Bank, 1999).13 Nicholas Kaldor, “Capital Accumulation and Economic Growth,” in Nicholas Kaldor, ed., Further

Essays on Economic Theory (New York: Holmes and Meier, 1978), pp. 1–53.

The second observation is that inequality provides an incentive for individual

effort—for hard work, innovation, and productive risk taking—which ultimately

ensures higher output and increasing productivity, and thus higher average income and

rates of growth. For economists, these incentive effects of inequality are the backbone

of the moral hazard argument against tax-financed transfers to the poor—that such

transfers undermine individual responsibility and the work ethic.14

Textbook Inequality Is Mostly Constructive

That there are potential positive incentive effects of inequality suggests the usefulness of dis-

tinguishing between what might be called “constructive” and “destructive” inequality. Let us

define constructive inequality strictly as only that inequality which reflects differences in indi-

viduals’ responses to equal incentives (or opportunities). It would then be inequality that is

consistent with notions of equity as defined above, and also with efficient allocation of

resources in an economy. Destructive inequality would in contrast be inconsistent with equi-

ty and, reflecting privileges for the rich and blocked potential for productive contributions of

the poor, would be economically inefficient, reducing rather than enhancing growth.

The idea that some inequality is “constructive” is not confined to economists.

John Rawls argues that unequal systems of incentives and rewards may be justified if

they optimize the position of the least advantaged—for example by generating incen-

tives for wealth creation and raising the income of the least advantaged as they share

in overall growth.15 His fundamental point is that an increase in inequality can only be

justified if one outcome is an improvement in the welfare of the worst-off.

Many problems arise, however, in distinguishing conceptually between con-

structive and destructive inequality. Does a child neglected by her parents who begins

school already behind in social or other critical learning skills and ultimately acquires

fewer productive skills “deserve” a lower-paying job? If that child’s future income is

lower than it might have been, reflecting her preschool environment, is resulting

inequality “constructive”? Is the economy at its optimal allocation of resources?

Consider another example: that of inherited wealth. It seems a commonplace obser-

vation that inequality of wealth (and resulting inequality in income from wealth) that

results from bequests is not “constructive.” This is one (implicit) rationale for a high

estate tax. Yet even bequests may represent the outcome of incentives in the bequeath-

ing generation to work hard. And in societies where social insurance of any kind is

lacking, a bequest may encourage efficient risk taking and hard work in the receiving

generation—for example in obtaining education or starting a business.

W H Y I N E Q UA L I T Y M AT T E R S 9

14 Arthur M. Okun, Equality and Efficiency: The Big Tradeoff (Washington, D.C.: Brookings Institution,

1975).15 John Rawls, A Theory of Justice (Cambridge: Harvard University Press, 1971).

Equality of Opportunity and (Constructive) Income Inequality

One rationale for the existence of some income inequality is that it is equality of oppor-

tunity, not of income, that societies fundamentally care about. An equal opportunity

society is then seen as characterized by “equity,” or fairness, as defined above. Inequality

of income in an equitable society would presumably be wholly constructive in the eco-

nomic sense, that is, perfectly conducive to efficient use of resources and thus to growth.

The notion that the United States is a society of equal opportunity is deeply embed-

ded in this country’s self-image, and may contribute to tolerance of relatively high income

inequality here compared to Europe. But the challenge remains the same: to define concep-

tually “equal opportunity” and to assess empirically the relevance of unequal opportunity

to income inequality. Empirical work suggests that a good portion of measured inequality

of income in the United States is due not to differences in merit or effort but in fact to

unequal opportunity. In a series of papers John Roemer estimates that for blacks to earn at

wage levels equal to whites in the United States would require six to ten times more educa-

tional spending on blacks than on whites, given the differences in the wage levels by educa-

tion between blacks and whites matched by their centile rank in their own distributions.16

This result suggests something more is going on than purely constructive inequality.

Mobility and Constructive Inequality

Another rationale for inequality (consistent with most inequality being constructive) is

that inequality of current income may not reflect underlying opportunities that would be

better revealed by information on lifetime or intergenerational mobility. Information on

lifetime mobility would reveal the extent to which a person during his lifetime changes

places in the overall income distribution of his reference group. Similarly, information on

intergenerational mobility would reveal the extent to which children’s place in the distri-

bution of lifetime income is independent of their parents’ place. Inequality of income

would matter less, and be more acceptable, if economic or political change were gener-

ating more opportunities and thus more mobility, including mobility that is downward.

This might be the explanation for voters’ continued endorsement of market reforms in

Latin America and Eastern Europe (even as inequality remains high in the former and has

risen substantially in the latter)—that the reforms are creating new opportunities in more

meritocratic systems, and that market signals are perceived to reward hard work, inno-

vation, and talent more fairly than more centralized and statist economic systems did.17

Nancy Birdsall1 0

16 Julian R. Betts and John E. Roemer, “Equalizing Opportunity Through Educational Finance

Reform” (paper presented at the First Workshop of the LACE/IDB/World Bank Inequality and Poverty

Network, Buenos Aires, Argentina, August 25, 1998). 17 Nancy Birdsall and Carol Graham, “Mobility and Markets: Conceptual Issues and Policy Questions,” in

Nancy Birdsall and Carol Graham, eds., New Markets, New Ideas: Economic and Social Mobility in a Changing

World (Washington, D.C.: Brookings Institution and Carnegie Endowment for International Peace, 2000), pp. 3–21.

But in fact social mobility does not seem to have been a major factor offset-

ting measured inequality. The common view of the United States as a highly mobile

society compared to Western Europe is the result of its higher average income growth,

which has lifted all boats, not a greater amount of switching of the positions of indi-

viduals or their children in the overall income distribution.18 Panel data on incomes of

workers in the United States indicate that in the 1980s, when wage inequality was

increasing, those initially at the bottom were more likely to be moving downward over

five-year periods than those in the middle or at the top.19 In Latin America, educa-

tional opportunities are increasing, but it is still the case that the amount of education

children receive in virtually every country is greater the higher the income and educa-

tion of their parents.20 If we assume that children’s educational achievement is a good

predictor of their future position in the ranking of lifetime income, then we have to

conclude that intergenerational social mobility is relatively limited.

Moreover, the magnitude and strength of the relation between parents’

characteristics and children’s education is not the same across countries. In Latin

America it depends, among other things, on macroeconomic conditions and on

countries’ education policies.21 This suggests that differences among countries in

market conditions and social policy affect the extent to which inequality is destruc-

tive versus constructive. Roemer’s numbers and Sawhill’s analysis show that in the

United States, some portion of wage inequality is in fact due to destructive inequal-

ity—the absence of equal opportunities. 22 This is likely due to some combination

of current discrimination in the labor market or in schooling, and past discrimina-

tion affecting children through its effects on their parents’ and their communities’

ability or willingness to provide the home inputs that complement schooling itself.

In South Africa, Brazil, and elsewhere, income inequality is also associated with

ethnic and racial heterogeneity.23

W H Y I N E Q UA L I T Y M AT T E R S 1 1

18 Isabel V. Sawhill, “Opportunity in the United States: Myth or Reality,” in Birdsall and Graham, eds.,

New Markets, New Ideas. 19 Gary Fields, “Income Mobility: Concepts and Measures,” in Birdsall and Graham, eds., New

Markets, New Ideas, pp. 101–34.20 Jere Behrman, Nancy Birdsall, and Miguel Szekely, “Intergenerational Mobility in Latin America:

Deeper Markets and Better Schools Make a Difference,” in Nancy Birdsall and Carol Graham, eds., New

Markets, New Opportunities? Economic and Social Mobility in a Changing World (Washington, D.C.:

Brookings Institution and Carnegie Endowment for International Peace, 1999), pp. 135–67.21 Ibid.22 Sawhill, “Opportunity in the United States: Myth or Reality,” pp. 22–35.23 Nancy Birdsall and Lesley O’Connell, “Putting Education to Work in Egypt,” Carnegie Endowment

Working Paper #5 (Washington, D.C.: Carnegie Endowment for International Peace, October, 1999); and

David Lam, “Generating Extreme Inequality: Schooling, Earnings and Intergenerational Transmission of

Human Capital in South Africa and Brazil,” PSC Report, No. 99–439 (Ann Arbor: Population Studies

Center, University of Michigan, 1999).

The Measurement Problem

In the end, “equity,” equality of opportunity, and the “right” amount of mobility in a soci-

ety are difficult to measure. Though analyses of social mobility and of inequality in

opportunities for education provide insights into the weight of destructive inequality with-

in and across societies, it is difficult to decompose measured income inequality into these

two components. Thus it is difficult to distinguish effectively between the constructive

inequality of the textbook story and destructive inequality. Under certain conditions,

some portion of income inequality may be constructive, reflecting and reinforcing what

economists would call efficiency in a competitive market system; but another portion is

likely to be destructive, reflecting inefficiencies in the market that inhibit growth.

Examining the effects of inequality on growth within and across countries can

provide insight about the degree to which inequality in different settings is growth

inhibiting and, at least in that sense, destructive, and why. This brings us to the crux of a

fundamental reason why inequality matters: its effects on growth and poverty reduction.

L E S S S A N G U I N E E C O N O M I C S : T H E M A R K E T I S I M P E R F E C T A N D

I N E Q UA L I T Y M AY I N H I B I T G ROW T H

Rapid growth combined with low levels of inequality in the so-called miracle economies

of East Asia24 raises the question whether the textbook tradeoff between efficiency or

growth on the one hand, and equality on the other, makes sense, especially in developing

countries. The contrast between fast-growing East Asia, with its low inequality, and slow-

growing Latin America, with its high inequality, needs explanation (see Figure 3).

Cross-country analyses of growth during the postwar period done in the early

and mid-1990s tended to confirm an association of high income inequality with lower

growth. Initial theorizing put the explanation in the political sphere. The median voter

in a more unequal society is relatively poorer and favors a higher overall tax burden to

finance more redistribution. A higher overall tax burden is likely to reduce the econo-

my’s efficiency and growth.25 Inequality thus reduces growth because it encourages

inefficient redistribution. There are two problems with this theory, however, as it

applies in developing countries. First, it is not clear that policies are primarily shaped

by the representative median-income voter, as opposed to special interests with the

resources and connections to influence policy. This is probably more true in developing

country democracies compared to the mature democracies of the most advanced

economies. I return below to the question of how inequality affects policy-making.

Nancy Birdsall1 2

24 World Bank, The East Asian Miracle: Economic Growth and Public Policy (New York: Oxford

University Press, 1993).25 Alberto Alesina and Dani Rodrik, “Distributive Politics and Economic Growth,” Quarterly Journal

of Economics 109, No. 2 (May 1994), pp. 465–90.

Second, other empirical evidence suggests that redistributive policies, measured in

terms of the marginal tax rate, are not in fact associated with lower growth.26 Indeed,

in the case of East Asia and other developing economies, it seems that redistribution in

the form of land reform and mass education supported growth.27

More recent economic models focus on the likelihood that inequality will exac-

erbate the effect of market failures on growth. Among capital market failures, a weak or

incomplete capital market is the most obvious shortcoming behind the growth-reducing

effect of inequality.28 If creditworthy borrowers cannot borrow because they are too

poor, without collateral to comfort lenders (given imperfect information, another “fail-

ure” in the competitive model), then their low income necessarily limits their ability to

invest—in their own farms, small businesses, and in the health and education of their

children. The inability of their parents to borrow to send children to school (to com-

W H Y I N E Q UA L I T Y M AT T E R S 1 3

26 William Easterly and Sergio Rebelo, “Fiscal Policy and Economic Growth: An Empirical

Investigation,” Journal of Monetary Economics 32, No. 3 (1993), pp. 417–58.27 Nancy Birdsall, David Ross, and Richard Sabot, “Inequality and Growth Reconsidered: Lessons

from East Asia,” World Bank Economic Review 9, No. 3 (1995), pp. 477–508.28 See Roland Benabou, “Unequal Societies,” NBER Working Paper #5583 (Cambridge: National

Bureau of Economic Research, May 1996); and Samuel Bowles and Herbert Gintis, “Escaping the

Efficiency-Equity Trade-off: Productivity-Enhancing Asset Redistribution,” in Gerald Epstein and Herbert

Gintis, eds., Macroeconomic Policy After the Conservative Era: Studies in Investment, Saving and Finance

(New York: Cambridge University Press, 1995), pp. 408–40.

Figure 3. GDP Growth Per Capita and GINI Coefficients Latin America and East Asia, 1960s-1990s

Venezuela

Bolivia

Philippines

Peru

MexicoBrazil

Colombia

Chile

Singapore

Indonesia

Hong Kong

Korea

Thailand

0

5

10

15

30 40 50 60Average Income GINI 1960-1990s

Ave

rage

GD

PPe

rC

apit

ac

Gro

wth

1960

-199

0s

Compiled from:World Bank, World Development Indicators, (CD-ROM 1999); Deininger and Squire (1996).

pensate for lost work income of children as well as to finance direct costs) may condemn

the children of the poor to limited education and low future income, generating a self-

perpetuating “poverty trap.”29 Consistent with the idea of a poverty trap is evidence that

where financial markets are deeper (and the poor better able to obtain credit), the asso-

ciation between parents’ income and education and that of their children is weaker.30

Similarly, poor insurance markets may trap the poor in inefficient informal systems of

risk sharing such as low-interest savings clubs.

Government or policy failures can also exacerbate the effect of inequality on

growth. In developing countries, government failures often reinforce the problems that

weak markets create. Lack of adequate public spending on basic education, and

repressed interest rates and directed credit programs that end up limiting access to

borrowing except to privileged insiders, will reinforce the negative effect of market

failures on growth in the presence of inequality.

Some inequality of income is explained by inequality of assets—such as land,

housing, and the asset of “human capital” (most easily measured at the individual level by

years of education). Inequality of assets is empirically more closely linked to low growth

in developing countries than inequality of income. Deininger and Olinto show that across

countries initial land ownership inequality is associated with low growth; they argue that

inequality of land is linked to rural poverty that in turn limits human capital accumula-

tion and thus growth.31 Carter shows that concentration of land ownership is associated

over long subsequent periods with concentration of income, even in countries where the

economic relevance of agriculture has diminished.32 This suggests that inequality of assets

like land affects the evolution of all kinds of political and social institutions that end up

limiting growth. Birdsall and Londoño show that across countries inequality in the distri-

bution of education—a good proxy for the asset of human capital—reduces growth, and

that once inequality of land and education are accounted for, inequality of income wash-

es out as a factor affecting growth.33 These arguments for the effect of inequality of

income and of assets on growth indirectly put substantial blame not only or primarily on

Nancy Birdsall1 4

29 Karla Hoff, “Market Failures and the Distribution of Wealth: A Perspective from the Economics of

Information,” Politics and Society 24, No. 4 (1996), pp. 411–32.30 Behrman, Birdsall, and Szekely, “Intergenerational Mobility in Latin America: Deeper Markets and

Better Schools Make a Difference,” pp. 135–67.31 Klaus Deininger and Pedro Olinto, “Asset Distribution, Inequality and Growth: An Illustration of

World Bank’s Inequality Data” (Washington, D.C.: World Bank mimeo, 1999).32 Michael Carter, “Land Ownership, Inequality and the Income Distribution Consequences of

Economic Growth” (paper presented for the conference “Rising Income Inequality and Poverty

Reduction,” WIDER, Helsinki, December 11–13, 1999).33 Nancy Birdsall and Juan Luis Londoño, “Assets Inequality Matters: An Assessment of the World

Bank’s Approach to Poverty Reduction,” Applied Economics in Action Papers and Proceedings (May 1997),

pp. 32–37; and Vinod Thomas and Yan Wang, “Revisiting the Lessons of Development” (Washington,

D.C.: World Bank mimeo, 1999).

inequality per se but on high rates of poverty and weak markets. They provide a com-

pelling logic for reducing poverty in the interests of growth, and for concentrating on

improved capital and insurance markets as a sensible route to reduced poverty. With bet-

ter markets, even the initial distribution of land and of education would not lock the poor

out of the growth process. In that sense these arguments bring us back to the fundamen-

tals of market-driven development: Get markets working better and poverty and inequal-

ity will not in themselves inhibit growth. Is that all there is to the story?

No, not necessarily. Even without recourse to politics or to the evolution of institu-

tions, and quite independently of poverty levels, there are other arguments for why inequal-

ity itself inhibits growth. Aghion, Caroli, and García-Peñalosa set out a model in which given

diminishing returns to capital (and given some degree of imperfection in capital markets,

which is reasonable, since capital markets are not perfect anywhere), inequality implies that

those with more wealth will have a lower marginal productivity of investment, while those

with less wealth cannot exploit higher return investments. This unequal distribution of

wealth reduces aggregate returns to investment in an economy, and thus reduces growth.34

This could apply in advanced industrialized economies as well as in poorer economies.

Aghion, Caroli, and García-Peñalosa assume an essentially static shelf of

investment opportunities. But their approach is consistent with one that explains growth

in settings such as East Asia in terms of the appearance of new, higher-return invest-

ments (because of a new agricultural technology, or because a school has been put into

a village—as in the model of Birdsall, Pinckney, and Sabot.35 The new investment pos-

sibilities available generate a sufficient return to induce a new round of savings and self-

financed investment (including by lower-income households who cannot easily borrow),

thus raising average growth. The model assumes that the poor household’s discount

rate, initially high because it is poor, falls as its initial investment increases income in

subsequent periods. The successively lower discount rates imply that in each period, new

investments with lower but positive returns become attractive. The underlying point

holds across all economies; the richly endowed (with capital, including human capital)

are likely to be investing in projects with lower returns, while the poor are forced by lim-

ited resources to pass up higher-return investments. Low (but still positive and secure)

returns to a limited circle of investors is one explanation for how so-called cronyism in

the form of close relations among business, banking, and government in Asia may have

contributed to the recent financial crisis. The implication is that redistribution would

increase the number of investors and raise average investment returns, and thus growth.

W H Y I N E Q UA L I T Y M AT T E R S 1 5

34 Philippe Aghion, Eva Caroli, and Cecelia García-Peñalosa, “Inequality and Economic Growth: The

Perspective of the New Growth Theories,” Journal of Economic Literature 37 (December 1999), pp. 1615–60.35 Nancy Birdsall, Thomas C. Pinckney, and Richard Sabot, “Natural Resources, Human Capital and

Growth” (paper prepared for the UNU/WIDER project on Environmental, Export and Human Capital

Accumulation Problems of Resource Based Growth Models, 1999).

Finally, high inequality may make it difficult for an economy to manage external

shocks. High inequality is likely to complicate management of the distributional conflicts

that typically intensify during economic downturns. Latin America, for example, the region

with the world’s highest levels of inequality, has a history of economic volatility, including

populist periods in which loss of fiscal control has ultimately undermined growth.36

The bottom line is not complicated. Cross-country studies provide increasing

evidence that high levels of inequality in developing countries inhibit growth.37 The

results are consistent with the high growth rates of low inequality countries of East

Asia over three decades, and the low growth rates of high inequality Latin America

(Figure 3). Theorists have provided various stories to explain the link. Some of the sto-

ries in fact reflect the negative effects of poverty—in the face of various capital mar-

ket failures, the poor have difficulty acquiring and using productive assets and thus

cannot easily get on the growth train, and their resulting low productivity inhibits

overall growth. Others do not require that some members are poor in an absolute

sense, only that not all members have the same access to the credit market. Given

inequality, the effects of market failures are compounded when policy fails as well and

if countries are vulnerable to internal and external shocks of various kinds.

L E S S S A N G U I N E E C O N O M I C S : I N E Q UA L I T Y I N H I B I T S P OV E RT Y R E D U C T I O N

Inequality is likely to inhibit poverty reduction in two senses.

• First, in an accounting sense, it is obvious that at any given growth rate, coun-

tries with a less equal distribution of income will reduce poverty less.38

• Second, inequality may undermine the effect of aggregate growth on income

growth among the poor for substantive reasons. The market and government

failures discussed above, that trigger inequality’s growth-reducing effects, are

likely to affect the poor disproportionately and thus to reduce their income

growth disproportionately. Birdsall and Londoño report that land and education

Nancy Birdsall1 6

36 Ricardo Hausmann and Michael Gavin, “Securing Stability and Growth in a Shock-Prone Region:

The Policy Challenge for Latin America,” in Ricardo Hausmann and Helmut Riesen, eds., Securing

Stability and Growth in Latin America: Policy Issues and Prospects for Shock-Prone Economies (Paris:

OECD, 1996), pp. 23–64; Dani Rodrik, “Where Did All the Growth Go? External Shocks, Social Conflict,

and Growth Collapse” (Cambridge: Harvard University mimeo, 1999); and Aghion, Caroli, and García-

Peñalosa, “Inequality and Economic Growth: The Perspective of the New Growth Theories,” pp. 1615–60. 37 This is not necessarily true for the more industrialized countries—probably because both markets

and government institutions work better there and, as discussed in the text, the negative effect of inequal-

ity on growth operates through weak markets and weak government institutions. 38 Martin Ravallion, “Can High-Inequality Developing Countries Escape Absolute Poverty?”

Economic Letters 56, No. 1 (September 1997), pp. 51–57.

inequality reduce the income growth of the poorest quintile about twice as much

as they reduce average income growth for all quintiles. They argue that initial

inequality of education generates inequality of income, and that in a vicious cir-

cle, inequality of income induces a new round of unequal education. This is con-

sistent with the idea that initial inequality of assets sets the tone for policy and

for the evolution of institutions (rules, social norms, and the role of the state)

that can lock in inequality of income. In most developing countries, agricultur-

al growth is associated with a reduction in poverty because the poor tend to be

concentrated in agriculture and in rural areas. However, in Latin America with

its highly unequal distribution of land and of income, agricultural growth in the

1970s and 1980s failed to reduce poverty.39 Landowners’ increasing incomes did

not create consumer demand for the small-scale labor-intensive products that

would have generated jobs and raised incomes of the poor.40

Market failures are more likely and more prevalent in developing countries—where

institutions that compensate for such failures are weak, inequality is higher, and the poor are

poorer. So in the real world (beyond the textbook), where competitive markets do not exist,

inequality matters because above some level it reduces the potential growth rate of the world’s

poorer countries, where the bulk of the world’s poor that would benefit from growth live.

Inequality Matters 2: It Undermines Good Public Policy

Behind the evidence that inequality inhibits growth is the likelihood that concentration

of income and assets at the top not only interacts with market failures to reduce

growth, but also leads to government failure. Public-choice models attribute poor pub-

lic policy to government regimes in which bureaucrats and insiders face no real checks

on the pursuit of their own interests.41 If the rich favor public policy that preserves priv-

ileges even at the cost of growth, inequality not only inhibits growth given government

W H Y I N E Q UA L I T Y M AT T E R S 1 7

39 Alain de Janvry and Elisabeth Sadoulet, “Growth, Poverty, and Inequality in Latin America: A

Causal Analysis, 1970–1994,” in Nora Lustig, ed., Shield the Poor: Social Protection in the Developing

World (Washington, D.C.: Brookings Institution, 2000). 40 This high-end inequality or concentration of income at the top may have slowed average growth,

too, by preventing the dynamic multiplier effects that consumer demand would have generated for the

industrial sector. See John Mellor, “Pro-Poor Growth: The Relation Between Growth in Agriculture and

Poverty Reduction” (report prepared for USAID/GEGAD, November 1999); and Birdsall, Ross, and Sabot,

“Inequality and Growth Reconsidered: Lessons from East Asia,” pp. 477–508. This may be part of the rea-

son for the finding of Birdsall and Londoño, reported above, that inequality of education and land are

associated with less growth of income among the poor than for other income groups.41 James M. Buchanan, Robert D. Tollison, and Gordon Tullock, eds., Toward a Theory of the Rent-

Seeking Society (College Station: Texas A&M University Press, 1980).

failure, but contributes itself to government failure. The problem seems especially great

when concentration at the top is combined with substantial poverty at the bottom, and

there is not a large middle class to demand accountability from government.

T H E S O C I A L C O N T R AC T A N D P RO - P O O R P O L I C Y

Consider the effect of inequality on the social contract and on the extent of what has been

called pro-poor policy. Spending on education and health is often viewed as win-win, that

is, as spending that improves efficiency by raising productivity while at the same time

improving equity. Yet studies suggest that public spending in these areas does not reap the

efficiency and equity returns it might42 and tends to reflect the interests of the rich even

where more spending in the interests of the poor would be more efficient overall.43

What lies behind the possible unfortunate association of income inequality with

inadequate social policy? First consider the supply side, for example the supply by the pub-

lic sector in the case of basic education. When the distribution of income is highly

unequal, the provision of subsidized basic education (and health and other services) to all

households implies a relatively large tax burden on the rich. High-income households in

many developing countries have succeeded in resisting that burden, in part because they

see limited benefits to themselves. (For many reasons, overall government revenues and

public social expenditures as a percent of GDP are much lower in developing than in

advanced economies.) The most notable case may be Guatemala, where income is highly

concentrated at the top, and government revenues as a percent of GDP remain among the

lowest in the world at about 10 percent; the current government has been unable to com-

ply with its commitment in the context of the peace agreements of several years ago to

raise tax revenues. In addition, the rich may succeed in channeling public-sector spending

to those programs from which they are most likely to benefit. In Latin America and in

Francophone Africa, a high share of public spending on education is allocated to univer-

sity education—from which the rich benefit most because their children are most likely to

complete secondary school and enjoy better-quality schooling, most likely to pass restric-

tive university entrance exams. Venezuela is an extreme; in the early 1990s Venezuela allo-

cated 35 percent of its public education budget to higher education, compared with

Korea’s allocation of 8 percent. Public expenditure on education as a percentage of GNP

was actually higher in Venezuela (5.1 percent) than in Korea (4.5 percent). However, after

Nancy Birdsall1 8

42 Lant Pritchett and Deon Filmer, “What Educational Production Functions Really Show,” World

Bank Policy Research Working Paper, 1975 (Washington, D.C.: World Bank, 1997).43 Nancy Birdsall and Estelle James, “Efficiency and Equity in Social Spending: How and Why

Governments Misbehave,” in Michael Lipton and Jacques van der Gaag, eds., Including the Poor:

Proceedings of a Symposium Organized by the World Bank and the International Food Policy Research

Institute (Washington, D.C.: World Bank, 1993), pp. 335–58.

subtracting the share going to higher education, public expenditure available for basic

education as a percentage of GNP was considerably higher in Korea (3.6) than in

Venezuela (1.3). Compounding the problem on the supply side is the fact that it generally

costs more to reach the poor in the first place—because the poor are more likely to live in

rural areas and because poor parents are usually less productive in “producing” at home

the human capital that complements public expenditures. Thus, good social policy

requires that the public sector “swim against the tide” and spend more for the same out-

come on the poor.44 This is a politically difficult challenge, and more so if political influ-

ence and power are correlated with income and income is highly concentrated. Efforts to

target public expenditures for some programs almost exclusively to the poor—an

approach vigorously encouraged by the World Bank and the international donor commu-

nity—can reduce the overall fiscal burden, but may then lose the political support of the

working and middle class. Income inequality also complicates social policy because it

affects household demand for education (and health). For any average income in a devel-

oping country, higher inequality implies more households that are poor and thus less like-

ly to have the resources to keep their children in school. In 1989 Brazil and Malaysia had

similar levels of per capita income. But the poorest quintile in Brazil had only about one-

half the absolute income level of the poorest quintile in Malaysia. Given an income elas-

ticity of demand for secondary education of 0.50 (a conservative figure), if the distribution

of income had been as equal in Brazil as in Malaysia, secondary enrollment among poor

Brazilian children would have been more than 40 percent higher.45

The demand for education will also be a function of expected returns in the labor

market. Indeed, education is probably treated primarily as an investment by poorer house-

holds, while for the rich it may also have consumption value. Low public spending on pri-

mary and secondary education has implied low-quality public schooling in many

developing countries, reducing the return to basic education and further discouraging

demand among households most concerned with the economic benefits of schooling and

most dependent on public schools. This may help explain the high dropout rates in some

countries (again, especially in Latin America), even in the face of high returns on average

to those who manage to complete secondary school.46 Racial or ethnic discrimination in

labor markets also implies lower average returns to education for some groups; thus, to the

extent that income inequality reflects discrimination, there can result a vicious circle of low

W H Y I N E Q UA L I T Y M AT T E R S 1 9

44 Nancy Birdsall and Robert Hecht, “Swimming Against the Tide: Strategies for Improving Equity in

Health,” in Christopher Colclough, ed., Marketising Education and Health in Developing Countries

(Oxford: Oxford University Press, 1997), pp. 347–66.45 Nancy Birdsall, David Ross, and Richard Sabot, “Inequality and Growth Reconsidered: Lessons

from East Asia,” pp. 477–508.46 Jere Behrman and Nancy Birdsall, “Quality of Schooling: Quality Alone Is Misleading,” American

Economic Review 73, No. 5 (1983), pp. 928–46.

education and persistent inequality, even as the discrimination itself recedes. Some would

argue that this is happening in the United States—that demand for education among blacks

is lower because of past if not current job discrimination, and is perpetuating current lev-

els of income inequality by reducing average education of blacks compared to whites.

P E RV E R S E E C O N O M I C P O L I C Y A N D T H E RO L E O F T H E M I D D L E C L A S S

Some of the unfortunate effects of inequality on public policy cannot easily be distin-

guished from the effects of a society’s simply having a large number of poor (in

absolute terms) with relatively weak political voice. However, it is also possible for

income inequality to contribute to poor policy even where there is little or no absolute

poverty. If large differences in income and wealth lead some people to “feel inferiori-

ty and shame at the way they must live,” then these people are likely to feel and be less

politically active.47 An extreme example is the untouchables, or scheduled castes, in

India; even the nonpoor members of this group have not been able to use their poten-

tial political power to exploit fully what are aggressive quotas for government jobs and

other forms of affirmative action.48

In many other settings where there is little absolute poverty (including most if

not all advanced economies), relatively low income is combined with racial or ethnic

minority status (in part reflecting past if not current discrimination), and with weak

political voice. Though it is not clear whether minority status or relatively low income is

the real culprit, income inequality probably compounds if not causes political weakness.

In the extreme, the political weakness of the relatively poor may mean that the

advantaged exercise sufficient control over others to constitute an “abridgement of lib-

erty.”49 Sometimes that control occurs as a result of financial contributions to political

campaigns, access to the media, or less benign exercises of political influence such as

bribes and extortion. If these kinds of influence in turn affect job availability, workplace

safety, local environmental conditions, or simply “what kind of life one can live,” then

large inequalities can be said to matter because of their political repercussions.50

There may also be bad political outcomes when those who feel disadvantaged

do have political voice. Because people care about inequality, inequality raises the like-

lihood of perverse policy responses—economic policies that make inequality worse

Nancy Birdsall2 0

47 Thomas M. Scanlon, cited in Charles R. Beitz, “Does Global Inequality Matter?” Metaphilosophy

32 (January 2001), pp. 95–112.48 William Darity Jr., “Intergroup Inequalities Across Countries” (paper presented at the

Macrodynamics of Economic Inequality conference, Jerome Levy Economics Institute, Bard College,

Annandale-on-Hudson, New York, October 28–29, 1999).49 Beitz, “Does Global Inequality Matter?”50 See Thomas M. Scanlon, What We Owe to Each Other (Cambridge: Harvard University Press, 1998).

while also inhibiting growth. Put another way, high inequality may generate not only

inadequate social policy through perverse effects on political participation. It may also

lead to positively perverse economic policies.51 There are many examples. Populist pro-

grams designed to attract the political support of the working class hurt workers in the

long run—this was the case in Garcia’s Peru in the 1980s and Peron’s Argentina. When

financed by unsustainable fiscal largesse they bring the inflation or high interest rates

that exacerbate inequality. (The rich can protect themselves from inflation with indexed

financial assets and by placing capital abroad; and from high interest rates by pressing

for privileged access to credit.) Price controls imposed on food products consumed by

the urban middle class and the poor hurt rural producers, or they lead to the disap-

pearance of products from stores as they are hoarded and resold at higher prices out of

reach of the majority. A high minimum wage may make it harder for the unemployed

to find work. High interest rates penalize the small business sector. Regulatory privi-

leges, trade protection, and special access to cheap credit and foreign exchange—all

bad economic policies—inevitably increase the profits of a wealthy minority.

Three other examples of self-defeating policies are worth noting.

• Protectionism is not a surprising reaction to the growing wage gap between the

skilled and unskilled. Yet developing countries that have been most open to

trade have had the fastest growth; those least integrated into global markets,

such as many African economies, have remained among the world’s poorest.52

Increases in trade and economic integration in poor countries, though associ-

ated with high wage inequality, may actually reduce inequality of income and

consumption. There are two possible reasons. First, as obstacles to imports fall

and price competition intensifies, prices drop—a boon for the poor, who use

most of their income for consumption. Second, trade liberalization and open

markets in general weaken the unfair advantages enjoyed by the rich and con-

nected, undermining the economic privileges and monopolies (reflected in

wealth, not wage gaps) that otherwise perpetuate high inequality.

• Worker entitlements that are legislated by the state rather than negotiated

through bargaining between employers and workers are another example.

Throughout Africa and Latin America, the injustices associated with high

W H Y I N E Q UA L I T Y M AT T E R S 2 1

51 See Nancy Birdsall, “Life Is Unfair: Inequality in the World,” Foreign Policy (Summer 1998), pp.

76–93. The essays in Nancy Birdsall, Carol Graham, and Richard Sabot, eds., Beyond Tradeoffs: Market

Reforms and Equitable Growth in Latin America (Washington, D.C.: Inter-American Development Bank,

Brookings Institution Press, 1998) discuss a range of policies that can enhance both efficiency and equity,

yet are not always chosen.52 See Jeffrey Sachs and Andrew Warner, “Sources of Slow Growth in African Economies,” Journal of

African Economics 6, No. 3 (October1997), pp. 335–76.

inequality contribute to legislation that protects a few workers, especially

those in state-owned enterprises and other protected sectors. Resulting “stan-

dards” raise the cost of labor, inducing employers to invest in labor-saving

technologies and to avoid formal hiring in favor of short-term job contracts

without any benefits at all. The loss of new jobs hurts mostly the poor and

unskilled whose main asset, after all, is their own labor. The reliance on tem-

porary labor discourages employer investments in training.

• Artificially low prices for water, electricity, and other public services are anoth-

er perverse reaction to the problem of inequality and associated political pres-

sures. Prices that are too low undercut the commercial viability and financial

sustainability of utilities; that leads to substandard services or none at all in

poor neighborhoods. The result? Cheap electricity and water are available to

powerful industrial interests, while the poor in slums rely on jerrybuilt con-

nections and buy bottled water at high prices from private trucks.

It would be silly to blame all bad policy on income inequality, but it would also

be foolish to ignore the potential effect of inequality and the frustrations it produces.

Easterlin shows that at low income levels, people care most about their absolute

income. But as their absolute income increases, relative income status matters more.53

In developing countries, it is members of the middle stratum that are most likely to care

about their relative status, and who are likely to be aware of changes in their income

status relative to the rich with changes in economic policies. A large or growing gap

between the middle and the rich is fertile ground for the perverse choices listed above.

That gap is greater in general in developing compared to developed countries, and is

growing in the former socialist economies of Eastern Europe and Russia.54

Inequality Matters 3: It Erodes Social Capital and InhibitsEffective Collective Decision-Making

Amartya Sen proposes that the idea of individual well-being can be described in terms

of sets of capabilities to realize basic human “functionings.”55 Those basic function-

ings are often thought of in terms of an individual’s health, self-respect, and so on,

Nancy Birdsall2 2

53 Richard Easterlin, “Will Raising the Incomes of All Increase the Happiness of All?” Journal of

Economic Behavior and Organization 27, No. 1 (June 1995), pp. 35–47.54 See Nancy Birdsall, Carol Graham, and Stefano Pettinato, “Stuck in the Tunnel: Have New Markets

Muddled the Middle Class?” Working Paper Number 14 (Washington, D.C.: Brookings Institution and

Carnegie Endowment for International Peace, 2000), pp. 3–21.55 Amartya Sen, Inequality Reexamined.

but they also include the ability to participate in the life of the community. But par-

ticipation in the life of the community suggests there are assets that are not held indi-

vidually but only in relation with others. The ability to participate may be enhanced

or circumscribed by characteristics of the community as well as the individual. The

idea of an asset held at the community or society level, or of social capital, reflects the

reality that certain assets (capital) are by definition relational—they exist only to the

extent they are shared with others. Putnam refers to social capital as “features of

social organization, such as trusts, norms, and networks, that can improve the effi-

ciency of society, facilitating coordinating actions.”56

Social capital has economic value because it is likely to reduce the cost of

transactions and contract enforcement (where the parties rely on trust and custom, for

example, with reduced costs of monitoring) and, as Putnam emphasizes, is likely to

encourage the formation and sustainability of the strong regulatory and government

institutions that enable market economies to function. From India and elsewhere

comes evidence that social capital in agricultural communities encourages the more

rapid uptake of new technologies57; others argue that where common property

resources are important social capital enhances collective decision-making.58

Social capital may also be an end in itself. Galbraith laments that private

affluence in the United States has coincided with poverty in public life.59 Kaus pro-

poses that those concerned with redistribution in the United States focus on increas-

ing not income equality or even equality of opportunity but “social equality,” that

is, enhanced ability of Americans to participate in common spheres of community

life, independent of differences in their income and socioeconomic status.60 He

notes examples such as the military draft, which at one time created a domain in

which all draftees were equal regardless of background; public schools to the extent

they serve families of different economic status; and mass transit as long as it is

clean and safe enough to retain middle-class users. Some such communal domains

are less directly dependent on public decisions but are private—the local tavern, the

community softball league, commercially successful urban sites such as Inner

Harbor in Baltimore. The pleasures of these places and states of being reflect and

reinforce “social equality.”

W H Y I N E Q UA L I T Y M AT T E R S 2 3

56 Robert Putnam, Making Democracy Work: Civic Traditions in Modern Italy (Princeton: Princeton

University Press, 1993).57 D. Parthasarathy and V. K. Chopde, “Building Social Capital: Collective Action, Adoption of

Agricultural Innovations, and Poverty Reduction in the Indian Semi-Arid Tropics” (paper for the Global

Development Network meeting, Tokyo, Japan, December 2000).58 Elinor Ostrom, Governing the Commons: The Evolution of Institutions for Collective Action (New

York: Cambridge University Press, 1990).59 John K. Galbraith, The Affluent Society, 3d ed., rev. (Boston: Houghton Mifflin, 1976).60 Mickey Kaus, The End of Equality (New York: Harper Collins, 1992).

The question then arises whether high money inequality has an adverse effect on

social capital. It seems likely that in highly unequal societies, marked differences in way of

life—in housing, transportation, types of vacation, and entertainment—may imply less of

the “social equality” just described, and less social capital. Similarly, with high inequality,

asymmetries of information in social exchange are more likely to arise, hindering cooper-

ation and reducing efficiency as well as undermining “social equality” per se.61 There is

indeed some evidence that income inequality adversely affects some of the inputs or cor-

relates of social capital. For example, in Tanzania, the degree of participation in groups

that provide economic benefits or informal insurance to their members is higher in com-

munities where income inequality is lower.62 In one study of sugar cooperatives in India,

the more unequal the land ownership among members, the less productive the coopera-

tive overall.63 Similarly, in the United States the percentage of households that participate

in various membership organizations is higher in metropolitan areas with lower income

inequality—controlling for racial and ethnic heterogeneity, income, education, and other

household characteristics. The effect is substantial. An increase in the Gini coefficient of

inequality by one standard deviation leads to a reduction in the probability of participa-

tion of 24 percentage points—more than two times the effect on participation of an indi-

vidual going from the status of high school dropout to high school graduate or more.64

A large literature in the field of local public finance indirectly addresses the

same issue by assessing the link between income levels and the formation of com-

munities with different amounts of heterogeneity in terms of income, race, educa-

tion, and other individual or household characteristics. A typical finding is that the

quality of publicly provided education is inversely related to income inequality, con-

trolling for average income. The same relationship holds between the provision of

other public goods and racial or ethnic heterogeneity.65 The connection is plausible;

inequality is likely to affect cooperation by creating psychological distance among

community members, reducing trust, and raising the costs of interaction. In a field

Nancy Birdsall2 4

61 Alberto Alesina and Eliana La Ferrara, “Participation in Heterogeneous Communities,” NBER

Working Paper #7155 (Cambridge: National Bureau of Economic Research, June 1999); available at

www.nber.org/papers/w7155. See also Pranab Bardhan, Samuel Bowles, and Herbert Gintis, “Wealth

Inequality, Wealth Constraints and Economic Performance,” in Anthony Atkinson and François

Bourguignon, eds., Handbook on Income Distribution (Elsevier, North Holland, 1999), pp. 541–603.62 Eliana La Ferrara, “Inequality and Participation: Theory and Evidence from Rural Tanzania,”

unpublished paper, cited by Alesina and La Ferrara, “Participation in Heterogeneous Communities,” p.14.63 Abhijit V. Banerjee, Dilip Mookherjee, Kalvan Munshi, and Debraj Ray, “Inequality, Control Rights

and Rent-Seeking: A Theoretical and Empirical Analysis of Sugar Cooperatives in Maharashtra,” mimeo,

1998, cited in Abhijit V. Banerjee and Esther Duflo, “Inequality and Growth: What Can the Data Say?”

NBER Working Paper #7793, (Cambridge: National Bureau of Economic Research, July 2000).64 Alesina and La Ferrara, “Participation in Heterogeneous Communities,” p. 23.65 Raquel F. Fernandez and Richard R. Rogerson, “Income Distribution, Communities and the Quality

of Public Education,” Quarterly Journal of Economics 111, No. 1 (1996), pp. 135–64.

experiment putting together small groups of participants from rural Peru, in which

the groups differed in terms of the degree of inequality of income across members,

the groups with lower inequality were more likely to find ways to cooperate and

optimize their use of a common property resource. The author refers for an expla-

nation to lack of trust and poor communication among group members of different

income levels; in the example he uses, a low-income individual was unwilling to

trust the advice of a high-income group member, though the latter was proposing a

win-win solution.66

Still another example of the possible link between money inequality and loss

of social capital comes from studies of crime and violence. Crime and violence clear-

ly undermine the possibilities for communal life. And income inequality appears to be

a critical determinant of crime. Fajnzylber, Lederman, and Loayza assess the impact

of inequality on homicide rates for a cross section of forty-five developed and devel-

oping countries over the period 1965–95.67 Income inequality measured by the Gini

coefficient has a significant and positive effect on homicide rates, robust to a variety

of specifications. Alternative measures—relative poverty inequality of education, and

ethnic polarization—do not change the measured effect of income inequality. The

authors also report that ratios of income of contiguous quintiles starting with the sec-

ond quintile (that is, third to second, fourth to third, and fifth to fourth) exacerbate

crime, and at an increasing rate. In other words, it is not inequality among the poor

that explains crime, but the disparity between the middle strata and the rich. The ratio

of income of the fifth to the first quintile has an effect on violent crime that is less

than a third of the effect of the ratio of the fifth to the fourth quintiles. These find-

ings provide indirect evidence for the potential problem for policy-making set out

above, of a middle-income group with unrealized expectations and resulting possible

frustrations—even as their income grows in absolute terms.

RISING INEQUALITY MAY INCREASE A SOCIETY’S TOLERANCE FOR HIGH INEQUALITY

If inequality matters for any of the reasons above, then the possibility that it can persist

matters too. What if high or rising inequality justifies itself at the societal level through

a change in social norm, perpetuating or extending its other negative effects?

In a discussion of changes in wage inequality in the OECD countries,

Anthony Atkinson makes the point that widening wage dispersion can result not just

W H Y I N E Q UA L I T Y M AT T E R S 2 5

66 Juan-Camilo Cardenas, “Real Wealth and Experimental Cooperation: Evidence from Field

Experiments” (paper for Global Development Network Meeting, Tokyo, Japan, December 2000).67 Pablo Fajnzylber, Daniel Lederman, and Norman Loayza, “Inequality and Violent Crime”

(Washington, D.C.: World Bank mimeo, 1999).

from shifts in the demand for skill, arising from trade liberalization or from skill-

biased technical change, but also from changes in social norms. He quotes Sir John

Hicks: “There are important social (and expectational) elements in the ‘free market’

part of wage determination. Even there, wages are not simply determined by supply

and demand.”68 Trade union bargaining, statutory wage determination, and elements

of convention or fairness may all be important across the society as a whole in deter-

mining wages and wage differentials. If individual utility depends in part on reputa-

tion and on conformity with social norms, then loss of reputation is greater the more

people believe in the norm. But the norm is undermined the more people cease to

observe it,69 for example under pressure from an exogenous shift in demand and from

an increase in competition with the globalization of markets. Then the socially

acceptable range of wage differential may widen, and a new high-wage differential

equilibrium may emerge. Atkinson may have in mind his own country, the United

Kingdom; he shows that wage dispersion has increased more there than elsewhere in

Western Europe in the last two decades.

This point raises the specter of a vicious circle: Are the pressures of global-

ization (wage dispersion, a decline in the tax burden on so-called footloose capital)

creating shifts in norms and expectations, making it more acceptable for governments

as well as employers to restrict or even abandon any redistributive role? This would be

ironic, given some evidence that in fact redistribution through progressive taxes and

public investments in education are associated with higher growth.70 If a society’s con-

cept of equity, that is, of what is just and fair, changes because it becomes accustomed

to more inequality, then inequality—along with its costs set out above—becomes

harder to reduce through decisions made at the social and public-policy level.

Endogenous social norms suggest the disturbing possibility that inequality breeds

inequality, and that societies that are already unequal can get stuck in an “unequal

equilibrium trap.” An unequal equilibrium trap is especially worrying for developing

countries, to the extent that it is associated with destructive inequality—with inequal-

ity of opportunity that blocks the productive potential of the poor, undermining

aggregate growth and perpetuating poverty.

Nancy Birdsall2 6

68 Anthony Atkinson, “Equity Issues in a Globalizing World: The Experience of OECD Countries,”

in Vito Tanzi, Ke-young Chu, and Sanjeev Gupta, eds., Economic Policy and Equity (Washington, D.C.:

World Bank, International Monetary Fund, 1999), pp. 63–80; quotation at p. 68.69 George Akerlof, “A Theory of Social Custom, of Which Unemployment May Be One

Consequence,” Quarterly Journal of Economics 94, No. 4 (June 1980), pp. 749–75.70 Easterly and Rebelo, “Fiscal Policy and Economic Growth: An Empirical Investigation,” pp. 417–58;

Robert Barro, The Determinants of Growth: A Cross-Country Empirical Study, Lionel Robbins Lectures

(Cambridge: MIT Press, 1997); and see Birdsall and Londoño, “Asset Inequality Matters,” pp. 32–37, and

Thomas and Wang, “Revisiting the Lessons of Development.”

Implications for the Future and for Policy

I have demonstrated that high and rising inequality matters, especially in developing

countries, and especially if inequality is primarily “destructive”—in economic terms,

inefficient. Even “constructive” inequality, for example wage gaps due to skill differ-

ences, can have political and social costs if it undermines good policy and increases tol-

erance for itself. But I have said little about causes of inequality, except indirectly.

Without systematic analysis of causes, it is not possible to do more than speculate

about future trends in inequality and the resultant effects on other aspects of develop-

ment. A pessimistic case can be made that the recent trend of increasing inequality in

some countries mirrors that in the United States, the country at the frontier of the

information age, and thus will persist. The absence of explicit policies to deal with dis-

tribution, globalization, and the increasing salience of new technology in the global

economy could lock in the forces of the digital divide. These forces favor the skilled over

the unskilled, ensuring growing wage gaps in a world where more and more income is

based on work and less and less on land or the inheritance of other forms of wealth.

The market reform process rewards those who have the human capital asset, and pro-

vides them and their children with permanently high returns to that asset. Children’s

access to that asset is mostly a function of their parents’ stock of human capital. Some

market reforms, moreover, because they are not done in ideal textbook fashion, exacer-

bate the problem—in Russia the privatization process was corrupted by insider privi-

leges, and the process locked in those privileges; in Asia the incomplete opening of the

capital account was good for a small minority but not for the average household.

A more optimistic case takes into account the reality that most economic sys-

tems are at least in part self-correcting. Rising wage returns to education increase house-

hold demand for children’s schooling, and the competitive pressures of more open

economies increase demand from the business sector for public education systems that

are more efficient and more effective in producing a flexible and skilled labor force. Open

markets, a seemingly ineluctable trend, help expose and undo the corruption that thrives

on protectionism and insider rents. Some self-correction comes through the political

process, as more democratic systems ensure that one-person, one-vote will correct in part

for the reality that political power is partly a function of economic power. Meanwhile,

technology and the globalization process itself, including the dramatic decline in the cost

of information and communication with the Internet, increase the influence of informal

civil society and other democratizing social forces; the digital divide disappears.

The reality, of course, lies somewhere in between. A base case is that as in the

past, inequality will change slowly if at all in the absence of explicit policy efforts to com-

W H Y I N E Q UA L I T Y M AT T E R S 2 7

bat it. For currently high-inequality countries, that implies ongoing risks of the costs of

inequality, and of the policy traps I have outlined. For other developing countries where

inequality is lower, it implies new risks to the extent that some economic reforms, along

with growth and technology change, tend to increase inequality—-at least in the short

run, as those with skills and other assets make rapid gains in new markets. Much then

depends on the nature and implementation of political reforms, especially those that

enhance democratic voice, and on the political impetus for egalitarian policies.

What might those be? History suggests that certain kinds of apparently redis-

tributive policies can go awry; redistributive policies have long been feared on the assump-

tion that higher taxes and transfer programs would increase distortions and undermine

incentives, compounding the political and economic difficulties of achieving the equitable

growth that would reduce inequality. This is clearly so in the case of perfectly competitive

markets. It is less clearly so in the real world of imperfect markets. Market failures, more-

over, are often compounded by government failures—that is, missing policies or bad poli-

cies—and those very government failures may be particularly acute where income

inequality is high and elites have more influence than does the representative median voter.

This is not the place to set out in detail the potential for efficiency or growth-enhancing

distributive and redistributive policies. More, and more effective, public spending on edu-

cation and health programs is an obvious way to build human capital and gradually to

make more equal the overall distribution of productive assets. Land titling and land reform

are mechanisms to distribute assets. Public expenditures on social insurance and social

safety nets can help households retain productive assets during economic downturns.

There may be limits to redistribution on the expenditure side, but more progressive tax

systems are also a potential vehicle for some redistribution—in some circumstances the

resulting efficiency gains may outweigh distortionary effects. The distributive effects of

market reforms are worth closer scrutiny—not only at the country level but also by the

donor community that supports such reforms.71

During a long transition from agriculture to industry, changes in production

and in the structure of employment caused wrenching inequality. Much inequality

today may be the result of an analogous transition from an industrial to an informa-

tion age. The issue is whether the transition will be slowed or speeded by decisions

made in the political and policy spheres, and whether those decisions can minimize

the risks and reduce the costs that citizens of the earlier transition bore.

Nancy Birdsall2 8

71 Francois Bourguignon and Thierry Verdier, in “The Political Economy of Education and

Development in an Open Economy,” (unpublished mimeo, Delta, Paris, November 2000), model the effects

of open capital markets and other liberalization measures on redistribution, including through political

change. See also Nancy Birdsall, “Managing Inequality in the Developing World,” Current History

(November 1999), pp. 336–81.