Whichever way we look at it, we always return to the same conclusion: namely that the social pact establishes equality among the citizens in that they all pledge themselves under the same conditions and all enjoy the same rights.

Jean-Jacques Rousseau, The Social Contract

It is precisely because the force of circumstances tends continually to destroy equality that the force of legislation should always tend to its maintenance.

Jean-Jacques Rousseau, The Social Contract

Introduction

The extent of inequality is of critical importance to development and development studies—above all, because high inequality is not compatible with a fair or just society, as Rousseau powerfully recognized in the first epigraph to this paper. From this perspective, preventing a high degree of inequality is an important objective in its own right. In addition, it is often viewed instrumentally as influencing other desirable goals, including poverty reduction, economic growth, health and education outcomes, and social and political stability. Here, predominant views today contrast with those of 50 years ago. Fifty years ago, inequality was viewed by development economists only from an instrumental perspective—almost exclusively in terms of the relation between inequality and the growth rate, then accepted as the predominant development objective. Moreover, rising inequality was argued to be a (possibly unavoidable) outcome of development (Kuznets 1955). This left policy makers as bystanders in an inexorable process—but one that would be reversed, according to Kuznets, as development proceeded. Since then, consideration of the instrumental aspects of inequality has extended well beyond economic growth; at the same time, high inequality is increasingly regarded as undesirable in itself, and consequently reducing inequality has become an important objective of development, as shown in Goal 10 of the United Nations’ Sustainable Development Goals: to “reduce inequality within and among countries.”

In parallel with its changing position in the array of development objectives, the study of inequality has also broadened with respect to two critical questions: inequality of what and inequality among whom. Nonetheless, the old questions remain as relevant as before, perhaps, even more relevant, in this broader context: First, what are the instrumental effects of inequality on other objectives (now including “new” objectives such as “happiness” and sustainability)? Second, how is inequality likely to evolve with development? Was Kuznets right that inequality would rise with more development? And has globalization affected its evolution?

Moreover, while reducing inequality has gained more prominence as an objective, in many countries, both rich and poor, inequality has risen significantly, bringing the question of policies to counter this to the fore. As Rousseau points out in this paper’s second epigraph, it is “because the force of circumstances tends continually to destroy equality” that legislation is needed to maintain it. Central to discussions of equality, then, is the need to identify appropriate equalizing policies and to understand policy formation, including both global influences on policy making in this area and domestic political forces. The “force of circumstances” that tends to destroy equality arises not only from the economic system but also in many cases from the global and domestic political system, which can prevent the changes needed to bring about a just distribution.

The aim of this essay is to review the changes that have occurred in consideration of equality over the past 50 years, with a particular focus on the areas to which Rousseau draws attention: why more equality is needed; whether the “force of circumstances” has generated rising inequality; and the implications for policy and politics.

I start with the questions of inequality of what and among whom. Clear answers to these questions are essential preliminaries to any discussion of inequality. I then consider why limiting the extent of inequality has increasingly come to be regarded as an intrinsic objective, a matter of justice. This is followed by a review of findings on some of the instrumental effects of inequality not only on growth but also on the many “new” development objectives. I then consider some of the connections between development, globalization, and inequality, ending with a discussion of policies and politics.

Conceptual Issues

Inequality Among Whom

In the nineteenth century, the consideration of inequality focused on factor shares, not personal distribution. According to Ricardo (1821, p. 5), “To determine the laws which regulate this distribution [between wages, profits, and rent] is the principal problem in political economy.” But this has not been the approach adopted by development studies. For much of the last 50 years, the prime interest has been inequality among individuals or households within a country, typically depicted in a Lorenz curve and summarized by the Gini coefficient. This is the type of inequality that concerned Kuznets (1955) and Galenson and Leibenstein (1955). And it remains the dominant concern among economists and policy makers. For example, only individual/household distribution is reported on in the widely used World Development Indicators produced by the World Bank. I shall term this vertical inequality (VI) to differentiate it from inequality among groups, which I define as horizontal inequality (HI). Interest in horizontal inequality, or distribution among groups rather than individuals, is more recent, although concern with one type of HI—gender inequality—dates back to the 1970s, reflecting the focus on gender that was stimulated by the feminist movements of the 1960s. Concern with horizontal inequalities more generally, with groups categorized along ethnic, racial, religious, or regional lines, initially arose out of work on violent civil conflict—which often follows group lines—but has since become of more general interest (Gurr 1993; Stewart 2000, 2001). Appropriate group categories for measuring horizontal inequalities differ across societies as the relevant categories depend on the salience of the categories as well as the question being asked. There are a variety of ways of measuring HIs: a population-weighted coefficient of variance is one (Mancini et al. 2008).

At the same time, governments of developing countries have consistently been concerned with the gap between rich countries and the Third World, or intercountry distribution. Catching up is a major objective of many developing country policy makers. With globalization, and as data have improved, people are also beginning to measure global vertical (or personal) distribution (i.e., distribution among all persons or households in the world) (Milanovic 2005; Lakner et al. 2013).

Thus from a concern solely with distribution among individuals within a particular country, the study of inequality has shifted to encompass concerns with group distributions and global personal distribution as well. But Ricardo’s question—the division of national income between profits, rent, and wages—which must underlie and in part determine distribution among both individuals and groups—has been oddly neglected. While Piketty’s work is devoted to the issue of shares of capital and wages, to date it has focused on developed countries.

Inequality of What, or in What Space

This issue has much debated and is critical for questions of equity and policy. In the 1960s, the concern was with income inequality, since income was then accepted as the best shorthand for welfare. However, adopting an income or resources metric involved a narrow view of well-being that was widely criticized (Seers 1972; Morris 1977; Hicks and Streeten 1979; Sen 1977). Consequently, in a famous paper entitled “Equality of What?,” Sen (1980) advocated a much broader measure encompassing people’s capabilities or freedoms, namely all the things that people can be or do. In focusing on a range of capabilities, Sen abandoned the money metric of income and moved to a plural assessment of well-being and inequality, and by focusing on capabilities (i.e., what people could do or be, not what they actually do), he advocated opportunities rather than outcomes as the space for assessing inequality. In practice, however, measurement problems are such that those who adopt Sen’s plural approach typically measure inequalities in a range of outcomes (or functionings) rather than capabilities. Taking a different approach, Roemer (1998) argued that the objective of equality should be defined not in terms of outcomes but in terms of opportunities, defined as equalizing circumstances that are beyond the control of the individual. Following Roemer, Ferreira and Gignoux (2008, abstract) interpreted inequality of opportunities as occurring where there are “outcome differences that can be accounted for by morally irrelevant predetermined circumstances, such as race, gender, place of birth, and family background.” The approaches of both Sen and Roemer involve a multidimensional approach to identifying inequality—indeed, the issue becomes one of inequalities, not a unique measure of income inequality. Again, the salient dimension is likely to vary across societies. In many countries, health, education, access to basic services, and assets (including land) are clearly relevant and important inequalities. Inequalities in dimensions that lead to inequalities in other dimensions are particularly critical (Stewart 2009). For example, inequality in access to education can be the source of inequalities in health and in income, and there are also reverse links.

Important dimensions beyond the socioeconomic are also of special relevance to HIs. In particular, inequalities in cultural recognition and respect are important for the well-being and respect of cultural and religious minorities and help determine the ability of people from such minorities to function effectively. Further, political inequalities can be the source of many other inequalities. Lack of citizenship, for example, often debars people from many other entitlements. Lack of adequate representation in political bodies can lead to policy biases and discrimination as well as to lack of self-respect. These are important aspects in any society but were recognized as such as a result of work on conflict, since high inequalities in these dimensions are often a cause of conflict (Langer and Brown 2008; Stewart 2011).

Thus, over the decades, a major change occurred in defining the relevant space for assessing inequality, from the use of a unidimensional measure—income—to a multidimensional approach. Nonetheless, in practice, income inequality remains the dominant measure.

Inequality and Justice

Utilitarianism, which provides the foundations of welfare economics, can be interpreted to be strongly egalitarian (Pigou 1932), on the basis of the principle of diminishing marginal utility as incomes increase, but this view was countered by Robbins (1938, 1945), who argued that one could not compare utility across individuals: “In our hearts we do not regard different men’s satisfactions from similar means as equally valuable” (1945, pp. 156–57). Robbins’s interpretation was reinforced by those economists, such as Hayek, with libertarian views about the undesirability of restraints on individual actions.

Following Robbins, only efficiency considerations counted for economists, and the desirable extent of inequality became that which maximized output or was associated with a Pareto optimum: in other words, only instrumental reasons related to income were relevant. However, while this was generally accepted by economists, it was questioned by philosophers—notably by Rawls (1971), who, like Rousseau, adopted a social contract approach, arguing that a just distribution was one that maximized the position of the poorest within a nation. This pointed to a much more egalitarian objective than the Robbins view, with inequality justified only if it could be shown to improve the position of the poorest. In a development context, issues of justice were also behind the Declaration of Human Rights and a rights-based approach to development, as well as the human development approach. A rights-based approach to development requires universal rights in a number of dimensions that are particularly important for human well-being, including education, health, and shelter. Its implications for inequality depend on the (undefined) level of achievement needed to secure these rights. The higher the level, the more equality is required. But even relatively minimal rights imply more equality than is currently observed in many countries. The human development approach, initiated by the United Nations Development Programme in 1990, argued that “the basic objective of development is to create an enabling environment for people to enjoy long, healthy and creative lives” (UNDP 1990, p. 9). While this does not define a particular level of inequality, it rules out high inequality in poor countries, which is not compatible with the objective. The reduction of inequality as a major objective and means to human development, then, has been a recurrent theme of the UNDP’s Human Development Reports. The “equality of opportunities” approach also pushes in an egalitarian direction by claiming that only inequality due to factors within a person’s control is justified. In practice, a very large proportion of observed inequality is due to inherited privilege, wealth, and human and social capital, all outside the individual’s control.

Most attempts to define a just distribution are particularly egalitarian with regard to horizontal or group distribution, and less so with regard to vertical distribution. This is because group membership—as in the case of gender, race, ethnicity, class—is typically outside the control of the individual, so that equality of opportunities would rule out most horizontal inequality, whereas some vertical inequality would be justified as due to variations in individual choices and effort. Equally, incentives that may be needed to achieve a Rawlsian maximin are more relevant at an individual level and possibly not at all at a group level.Footnote 1

Rawls explicitly confined his principle to within nations or countries, since the implicit social contract on which the principle is based applies only to a society with a single government. In contrast, the human rights and human development approaches are global and require reduced intercountry as well as and intracountry inequality; the right to development, in particular, requires reduced intercountry inequality.

This view of what a just distribution might look like, together with rising inequalities observed in recent decades, led to criticisms of the UN’s Millennium Development Goals (MDGs) for including only absolute targets without reference to inequality. In practice, this meant, for example, that the target of halving infant mortality rates (or poverty rates) could be achieved at a country level while the rates for the lowest income category changed little or could even increase. Hence, the aim of reducing inequalities between people, groups, and countries has been adopted explicitly in the Sustainable Development Goals (SDGs) that succeeded the MDGs.

The Instrumental Effects of Inequality

The instrumental effects of inequality on economic growth have been a consistent concern of economists, while effects on other dimensions were initially of little concern. And as the objectives of development have widened, so has attention to the instrumental consequences for a broader set of objectives.

Economic Growth

The early focus on income as the objective of development was paralleled by an exploration of how income inequality affected GDP growth. There are reasons both for expecting a positive impact of inequality on growth and for expecting a negative impact. On the one hand, incentives and savings may be expected to be higher with more inequality; on the other, human capital may be less. In the 1950s, it was assumed that more unequal income distribution would lead to higher growth, via higher savings and incentive effects (e.g., Galenson and Leibenstein 1955; Okun 1975). This was challenged by Adelman and Morris (1973), who argued, with country evidence, that a more equal initial income distribution was associated with higher growth. Much of the later empirical literature has supported Adelman and Morris (e.g., Alesina and Rodrik 1994; Alesina and Perotti 1994; Persson and Tabellini 1994; Bénabou 1996).Footnote 2 But other evidence is more ambiguous (Fishlow 1995; Barro 2008; Li and Zou 1998; Forbes 2000; Banerjee and Duflo 2003). The relationship between inequality and growth may vary according to the level of inequality. For 1960–1998, Cornia (2004) found a concave relationship, with growth rising as inequality increases from very low levels, then declining with a further increase in inequality.

Besides the effect on the average growth rate, there is growing evidence that inequality is associated with greater instability in growth. Two economists at the International Monetary Fund found that “when growth is looked at over the long term, the trade-off between efficiency and equality may not exist. In fact equality appears to be important in promoting and sustaining growth” (Berg and Ostry 2011, p. 13; my italics). An example of how inequality may contribute to instability is the financial crisis of 2007–2008. Rising inequality in developed countries has been argued to be responsible for the crisis. Given the low wage share, the only way growth could be sustained prior to the crisis was through consumer credit (especially for houses in the USA), which eventually became nonviable, leading to the crash (Stiglitz 2012, 2013; Lansley 2012). This is one explanation for the nonsustainability of growth in very unequal economies.

Poverty

From the 1990s, the reduction—and, with the SDGs, elimination—of extreme poverty became a central development objective. Undoubtedly, inequality has an important effect on the achievement of this objective. For any given level of average per capita income, the higher the inequality, the higher the poverty rate is almost certain to be. In other words, the elasticity of poverty with respect to growth will be higher with lower inequality (Bourgignon 2003; Adams 2004; Fosu 2009). When growth is accompanied by rising inequality, the reduction in poverty may be very small or even negative, while conversely slow growth can be associated with large reductions in poverty if accompanied by substantial reductions in inequality (Cornia 2004). The importance of the level (and change) in income distribution for mediating the impact of growth on poverty has been illustrated by Dhatt and Ravallion (1998) for Indian states, by Fosu (2009) for African economies, and by Kalwij and Verschoor (2007) for 58 developing countries in the 1980s.

Education, Health, and Nutrition

The human development approach, based on Sen’s capabilities approach, made improvements in the basic capabilities an intrinsic objective and consequently made the impact of inequality on these capabilities an important instrumental consideration. Extensive evidence shows that inequality in incomes tends to be reflected in inequality in education, including achievements at basic levels (such as literacy or completion of primary and secondary school) and unequal access to tertiary education (Filmer and Pritchett 1999; Bachmann and Hannum 2001; Birdsall and Sabot 1994). This arises because inequality affects both household decisions and public spending (Gutiérrez and Tanaka 2009). Figueroa (2006) has explored inequalities across groups in Peru, showing systematic differences not only in access to education but also in returns, partly due to discrimination in the quality of education and in employment between members of rich and poor groups, notably between poor indigenous groups and the more privileged nonindigenous.

An inverse relationship between average health outcomes and income inequality has been claimed but is not firmly established (Subramanian and Kawachi 2004). Wilkinson and Pickett (2006), in a review of 168 analyses, reported that of these, 52 % found evidence of a link between higher inequality and worse average health outcomes, 26 % were partially supportive, and 22 % provided no support. But Deaton (2003) questioned the data and methodology of some of this work and did not find a relationship between vertical inequality and health.Footnote 3 The fact that some studies find a negative impact of inequality on average health outcomes, and none a positive impact, suggests that decreasing inequality may improve, and will not worsen, a society’s average health. Moreover, given that most studies find a relationship between individual income levels and individual health outcomes, reducing income inequality is almost certain to contribute to improving the health of the poor. In addition, inequality has been shown to have a deleterious effect on nutrition of low-income groups, due to a combination of lack of income and poor maternal education (Murcott 2002; Moradi 2006). This in turn adversely affects health.

Human Security

Human security has come to be recognized as an important aspect of well-being and human development, for example in the 1994 and 2014 Human Development Reports (UNDP 1994, 2014), in the World Bank’s Voices of the Poor (Narayan 2000), and in the 2003 report of the Commission on Human Security (Ogata and Sen 2003). It is a clearly expressed priority for poor people themselves (Narayan 2000). Though human security goes well beyond physical security, criminality and civil war are critically important aspects of insecurity.

Criminality

Most research shows that greater vertical inequality is associated with higher levels of criminality as measured by intentional homicides and robbery. As Fajnzylber et al. (2002a, p. 1) stated after investigating the determinants of crime both across and within countries: “Crime rates and inequality are positively correlated within countries and, particularly, between countries, and this correlation reflects causation from inequality to crime rates, even after controlling for other crime determinants” (see also Krahn et al. 1986; Fajnzylber et al. 2002b; and many references cited in these articles). However, identification of the cause of this link between inequality and crime varies: economists cite the “economics of crime,” with more poor people having incentives to rob the rich in unequal societies, but sociologists point to the “sociology of crime,” with social control over conflict being weaker in unequal societies. Which of these mechanisms is dominant will determine appropriate policy responses. However, in either case, lower inequality should be associated with less crime.

Civil War and Other Group Violence

There is strong and accumulating evidence that higher horizontal inequalities are associated with greater likelihood of the outbreak of civil war (Gurr 1970; Stewart 2000; Mancini 2008; Cederman et al. 2013). However, perhaps surprisingly, on balance, the evidence does not support a connection between VI and civil war (Collier and Hoeffler 2004; Fearon and Laitin 2003), although Auvinen and Nafziger (1999) found a weak connection. The relationship between HIs and conflict arises because people mobilize behind a common identity (often ethnicity or religion), and such mobilization can be stimulated by sharp HIs. There is less potential for mobilization with high VI because of the absence of a unifying shared identity, although in some cases peasants’ or caste movements unify and mobilize the deprived. Research indicates that HIs are particularly likely to provoke conflict where there are simultaneously both economic and political inequalities in the same direction (Stewart 2008; Cederman et al. 2011; Østby 2008), since in those circumstances leaders have an incentive to mobilize because of political exclusion, while social and economic exclusion motivates their supporters. Inequalities in cultural recognition can also provoke conflict motivating both leaders and followers (Langer and Brown 2008). Further, there is evidence that horizontal inequalities are associated with forms of group violence other than civil war, including milder types of violence such as riots (Blau and Blau 1982) and horrendous forms like genocide (Fein 1993; Harff 2003; Stewart 2011). Some work too links gender inequality to domestic violence (Bailey and Peterson 1995; Yodanis 2004). The relationship between HIs and conflict provides a strong motive for action to correct such inequalities, even among people who are not concerned with equity. Yet action has to be introduced cautiously because the correction itself can be provocative.

Sustainability

The broadening development agenda now gives environmental sustainability a central role, essential if any current development is to be shared by future generations. More inequality is likely to make it more difficult to achieve sustainability goals. This is for two related reasons: the issue of “catch-up” and that of fairness. Despite technological advances in carbon saving, higher economic growth remains tied to greater carbon emissions, and a slowdown in global economic growth is likely to be an essential component of achieving sustainability. Yet both within and across societies, a major motive for growth is to catch up to those people/countries with higher incomes. So long as inter- and intracountry inequality remains constant or increases, catch-up cannot be achieved whatever the growth rate, and people and countries at lower levels in the income hierarchy will constantly press for growth to try to approach the (ever higher) living standards of those above them. Greater equality within and between countries not only will itself bring about some catch-up but also will make a lower rate of growth more acceptable. Moreover, it will do so without reducing “happiness,” according to empirical research showing that at high income levels increased incomes do not raise happiness (Easterlin 1974; Graham and Felton 2006). The issue of fairness is related to this. Given an absolute limit on safe carbon emissions at a global level, the scarce emissions need to be distributed fairly, from the perspectives of both justice and acceptability. Prima facie, a fair distribution of a scarce global resource would be one of equal per capita distribution. This might need to be qualified to encourage efficiency of use, but we can safely conclude that a fair distribution would be far more equal than the present one. A fairer distribution would also be more likely to make limitation of carbon use acceptable to poorer people and countries, and hence more likely to be implemented. In short, reduced inequality in incomes and carbon use is an important part of achieving sustainability (Neumayer 2011; Stewart 2015).

In sum, while considerations of justice call for much more equality between and within countries, even without any such ethical stance, there are strong instrumental reasons for having more equality, since it would be likely to contribute to the achievement of agreed-upon objectives, such as more sustained economic growth, more educated and healthier populations, less criminality, and, with greater horizontal inequality, less civil war. In addition, greater equality would make it easier to achieve environmental sustainability. Thus most arguments favor more equality, both vertical and horizontal. We may not know the ideal distribution, but we do know the desirable direction of change. Yet Kuznets held that development would be associated with more inequality, while many point to the way globalization seems to be increasing inequality. Is the world economy unavoidably moving in the “wrong” direction? This is discussed in the next section.

Is Rising Inequality Inevitable?

In a seminal paper derived from cross-country data and evidence on the historical evolution of a few developed countries, Kuznets (1955) suggested that as incomes rose inequality would first rise and then fall along the famous inverted U-shaped curve. Apart from being an empirical generalization, Kuznets’s underlying explanation was that development consisted in a switch of employment from a low-productivity traditional sector to a high-productivity modern sector, which would generate first rising, then falling inequality as the relative size of the two sectors changed. However, increasingly sophisticated tests of the hypothesis have shown that its validity depends on the model specification, the sample, and the time period adopted. Tests using trends within countries have given it little support (Anand and Kabur 1993; Fields 2001; Bruno et al. 1998). The failure to find consistent evidence of a Kuznets curve relationship has been attributed to policy differences across countries (Kanbur 2011). Moreover, recent trends in developed counties show an upturn in inequality—indicating a sideways S-curve rather than an inverted-U—while Piketty has argued that the sharply reduced inequality in the first half of the twentieth century was due to special factors, including the aftermath of two world wars, and could not therefore be expected to be replicated.

In the 1980s and 1990s, the majority of countries (developed and developing) showed a rise in inequality. Indeed, Cornia and Sampsa (2001) found that the Gini coefficient rose between 1980 and 2000 in two thirds of the 73 developing countries for which they had data. It has been suggested that globalization, which accelerated during these decades, was and would continue to be responsible for rising inequality. According to Mazur (2000), “Globalization has dramatically increased inequality between and within nations,” and Gallas et al. (2014, p. 146) found that “neoliberalism and financialization are accompanied, in most countries, by self-reinforcing tendencies towards higher levels of economy inequality.” Of the many aspects of globalization that may affect distribution, two important ones are greater trade openness and freeing of capital movements. Following Heckscher-Ohlin, it was expected that freer trade would improve the distribution in labor-rich countries (i.e., poor countries) and worsen it in capital-rich ones (i.e., high-income countries). Yet this was not observed. Barro (2000), among others, “finds that the relationship between openness and inequality is positive for low-income countries and negative for high-income countries” (quoted in Kremer and Maskin n.d.). Szekely and Mendoza (2016) find that “more trade openness leads to a worsening in the income distribution” among Latin American countries and cite other studies coming to the same conclusion for Latin America. For the late twentieth century, the balance of the evidence suggests that more trade openness is associated with worsening inequality. Kremer and Maskin (n.d.) develop a model based on labor complementarities that explains why inequality may rise in both rich and poor countries with greater openness. Harrison et al. (2010) survey other theories pointing in the same direction.

There is also general agreement that capital account openness and the financialization that accompanies it are likely to increase inequality (IMF 2007; Gallas et al. 2014). A further inequality-inducing factor associated with globalization is the weakening of workers’ movements and rights, especially in developed countries. To increase competitiveness, labor market reforms have been introduced, including measures that worsen the contractual conditions of labor and depress unskilled wages, while subsidies to investment are frequently granted in order to secure (internationally mobile) capital. Global competition for the highly skilled has tended to bid up their rewards, thus increasing inequality within the wage sector. Most countries saw rising inequality in the 1980s and 1990s; some also experienced a rise in inequality in the 2000s, with that rise being particularly sharp in China, India, the USA, and the UK. Nonetheless, despite the universal pressures from globalization, in the 2000s most countries in Latin America and the majority in sub-Saharan Africa saw a fall in inequality (Cornia 2014; Lopez-Calva and Lustig 2012; Cornia and Martorano 2012; see also Roser 2015). For the world as a whole, moreover, global personal income distribution has not risen and may have fallen from 1988 to 2008, as declining intercountry inequality—especially due to the rise in Chinese incomes—offset the worsening intracountry inequality (Lakner and Milanovic 2013).Footnote 4

The different trajectories among countries suggest that rising inequality is not inevitable even in the context of increasing globalization and competitive pressures. The egalitarian changes in Latin America in the 2000s have been attributed largely to policy differences (Cornia and Martorano 2012; Cornia 2014), and Kanbur (2011) makes a similar point with respect to countries that avoid the rising inequality predicted by Kuznets. This suggests that while there are strong tendencies making for rising inequality in a globalizing capitalist world, governments can counter this by well-designed policies. Changing perspectives on policies—and on the politics underlying these policies—are therefore discussed in the next section of this paper.

Policy Perspectives

Reducing inequality did not form part of the policy agenda of the donor community—led by the World Bank—for most of the past 50 years. In the 1960s and 1970s, the major emphasis, shared by developing-country governments, was on economic growth; the 1980s were the Washington Consensus years, with stabilization and adjustment as the overriding objectives and liberalization of various sorts as the means. The devastating impact of these policies on poverty was gradually recognized (Cornia et al. 1986), and from 1990 onwards poverty reduction became a central goal, culminating in 2000 in the MDGs, in which, as noted above, inequality played no part. Reducing inequality has crept up the agenda at an international level since the 2000s. For example, the World Development Report 2006 was entitled Equity and Development (World Bank 2006a), and reducing inequalities within and among countries is one of the SDGs. Yet this has not been paralleled by any global efforts to define and advocate policies to reduce inequality. Indeed, many of the policies that continue to be advocated by global institutions—and are even insisted upon in conditional loans—are likely to be inequality increasing, such as reducing public expenditure, switching from direct to indirect taxation, and reducing high marginal rates of income tax. Labor market reforms and privatization also tend to be unequalizing. Similar policies are demanded by European institutions and governments. There are no signs of the policy set advocated by these global institutions being revised, despite apparent acceptance of the need to reduce inequality.

At a country level, there has been a much more varied approach. In the early decades after the second world war when countries became independent, land reforms and nationalization of assets pushed in an egalitarian direction, but these were reversed with privatization or transformed (to market-based land reforms) to become unegalitarian. Nonetheless, particular countries have adopted and maintained egalitarian policies over the years. For example, South Korea and Taiwan introduced radical land reforms in the late 1940s and early 1950s, and many other countries adopted less comprehensive land reforms (El Ghonemy 1990). Early on, Costa Rica introduced an array of pensions and transfers that have been sustained. Affirmative action programs have reduced HIs (and VI), most comprehensively in Malaysia but also partially in many other countries (e.g., India, Nepal, Brazil) (Brown et al. 2012). Cuba sustained a highly egalitarian distribution through nationalization of assets, limits on wage inequalities, and strong comprehensive social policies. In the 2000s, Latin American countries have introduced/raised minimum wages, adopted universal social protection programs, and extended education, all contributing to reducing inequality in disposable incomes (Cornia 2014). Many countries have reduced education inequalities as they move toward universal primary and secondary education, and some countries have targeted health policies to improve health equality, notably Thailand.

Policies that reduce inequality can be categorized as affecting predistribution or redistribution and as universal or targeted. Predistribution policies are those that affect the primary distribution of income, before taxes or transfers. For example, changes in the quantity or returns to education affect the distribution of wage income; the distribution of income from assets is affected by policies toward the distribution of asset ownership and those affecting the returns to assets—for example, through interest rate or competition policies. Taxes and transfers affect the redistribution of income and can be designed to reduce inequality substantially.Footnote 5 Universal policies are policies that apply to everyone in the country, such as universal pensions or child support, or universal health and education services. Targeted policies are directed toward particular segments of the population. For VI, these target people below a particular income, and for HI they target particular groups—for example, education quotas for disadvantaged individuals or groups.

Country evidence shows that these policies can be effective in reducing inequality. Given the strong evidence, summarized above, that more equality would be desirable from an instrumental perspective as well as from that of advancing justice, the question arises as to why such redistributionary policies have not been adopted more widely. To answer this, we have to turn to political economy, which is becoming a central issue for those currently investigating inequality.

Several strands of thought are relevant here: the median-voter issue; attitudes toward cross-group redistribution; the apparently increasing role of money in politics; and the role of social movements. Although these appear to be separate considerations, analysis suggests they are strongly interconnected.

Political Economy

Although political economy was probably at the forefront of Rousseau’s thought—as exemplified by the second epigraph to this paper—political economy considerations were not common 50 years ago (with the exception of Marxist thinkers), partly because redistribution was not a prime objective and partly because authoritarian regimes were dominant. With the widespread advance of democracy, however, a central paradox emerged: If the majority of people have below-average incomes, why didn’t democracy lead to redistributionary policies? As Kenneth Roberts points out in this issue (quoting Schmitter 2014), the puzzle is that democratic institutions “have done little to bring about ‘significant changes’ in power relations, property rights, policy entitlements, economic equality, and social status.” This contradicts the median-voter hypothesis, which hypothesizes that in a democracy the greater the inequality, the more approval for redistribution (Meltzer and Richard 1981). Empirical evidence for that hypothesis is weak and inconclusive, although Schmidt-Catran (2014) finds supporting evidence for long-term within-country (but not cross-country) preferences. A possible reason for the lack of support for the hypothesis is resistance to across-group redistribution in multiethnic societies, which is explained by psychologists as due to the way people limit their “domain of justice” to their own group (Schmidt-Catran 2014; Clayton and Opotow 2003; Quadagno 1994). In democracies where a majority group dominates, the government is unlikely to support redistribution to poorer minority groups (see Rodrik in this issue). Nonetheless, survey evidence for some African countries shows extensive support for redistribution toward poorer groups among the population, though this may not be translated into government action (Langer et al. 2015).

Robert Wade (2014) argues that the median-voter hypothesis involves a misconception of how people identify themselves: the median voter, he hypothesizes, may believe that if redistribution occurs it will not be from the rich to the median (or middle class), but more likely from the middle class to the poor. The median voter would therefore actually lose by such a redistribution and therefore resists it. Wade supports the validity of this view by reference to the apparent constancy of the share of the middle 40 % of the distribution observed by Palma (2011). Haggard et al. (2013), investigating the median-voter hypothesis, found that preferences for redistribution vary according to occupation and place of residence, and not simply according to a person’s location in the income hierarchy.

Another explanation, also shared by Wade, is the power of money in politics. As Crouch (2014, p. 4) puts it, increasingly “free market democracy” is being replaced by “corporate authoritarianism,” as companies dominate political finance and control the media and many think tanks. Moreover, the financial sector exerts power through fear by threatening to withdraw if radical redistribution occurs. As Gallas et al. (2014, pp. 144–45) state: “In situations of extreme wealth disparity, those at the top find it relatively easy to influence political decisions, while those at the bottom struggle to make themselves heard.”

Poor people individually are unable to resist these powerful forces. In principle, political parties may empower them by enabling them to unite behind a progressive agenda. However, political parities have been shown to be weak in many of the new democracies and to be patrimonial rather than class based (Roberts in this issue). In many African countries, political ties are forged through shared ethnic rather than class identities (Posner 2005).

Yet this account of the underlying political economy of policies toward redistribution overdetermines resistance to redistribution. In practice, a number of developing countries have seen decreases in inequality (shown for Latin America in Fig. 1). Investigations of cases of effective redistribution invariably find that social and political movements have played an important role in bringing about progressive change. For example, this explains the historic emergence of the welfare state in Europe (Haggard and Kaufman 2008). In Korea too, a powerful workers’ movement emerged that gave rise to welfare state provisions (Koo 2001). In Brazil and Bolivia, worker and peasant movements formed political parties that eventually gained power (Hunter 2006; Crabtree and Whitehead 2001). In some cases, single-issue movements had an influence: for example, promoting India’s National Employment Guarantee Scheme and the Right to Food (Dreze and Reetika 2011).

Fig. 1
figure 1

Changing income distribution in selected Latin American countries. Source: Roser (2015)

Full size image

When it comes to correcting HIs, the motive is rather different: it seems that fear of violence is often a powerful motivation. Identity groups by their nature have a natural basis for political mobilization, which may be peaceful but is sometimes violent. The riots of the late 1960s provided the backdrop to the Malaysian policies, while Northern Ireland’s policies were introduced in the hope of ending the conflict. In Nigeria, constitutional provisions limited political inequalities with the intention of keeping the country together after the Biafran war. In the USA, affirmative action was introduced after city riots by Blacks. In Brazil, however, workers’ movements played a role supported by a discourse of fairness and justice. In South Africa, the deprived group also formed the majority, providing a political basis for redistribution, although policies have been unequalizing within groups.

A full account of the determinants of redistributionary politics and policies is thus complex: put rather simplistically, it is a matter of the power of money balanced against the resistance generated by people’s movements, and how well each organizes, in the context of increasing global pressures. In rhetoric, these global pressures are more and more on the side of equality and redistribution, as articulated in the SDGs; but in practice, the influence of competitive pressures, global finance, and corporate power weighs the balance against any radical redistribution.

Conclusions

Fifty years ago, the question of equality was predominantly viewed as an intellectual project: What is the extent of inequality? What is the optimum inequality? And what policies might bring it about? We have made considerable progress in answering these questions in relation to vertical inequality of incomes, mainly measured by the Gini coefficient, but much less progress on how to achieve something nearer the optimum, given the political forces and the unfavorable global environment.

There is increasing acceptance that the Gini coefficient itself is not well adapted to current inequalities because it focuses too much on the middle, and other measures more sensitive to the top and bottom of the distribution (such as the Palma set of measures) are beginning to be used.Footnote 6 It is also widely accepted that we should broaden the dimensions in which we measure inequality and should focus on horizontal as well as vertical inequality, while there is growing attention to global personal distribution. But in practice, much less attention has been given to these broader aspects in either measurement or policy analysis.

Moreover, there has been little analysis of the Ricardo question—what the shares of output going to profits (domestic and foreign), rents, and wages are and what determines them. This is an outstanding question that urgently needs to be explored, as it underlies distribution among people and groups and across countries. A major determinant of personal income distribution is the distribution of output between wages, profits, and rents. In a twenty-first-century context, rents are not mainly a matter of returns to land but include returns to technology (protected as intellectual property) and to (often temporary but recurring) monopolies (Kaplinsky 2005). Whereas Ricardo analyzed the question in a national context, globalization means that it needs to be analyzed both globally and nationally. The global capitalist system allows free movement of capital, which can move from one location to another to maximize profits and minimize taxation. Yet the same system limits worker mobility and encourages governments to keep wages and taxes low and minimize worker protection so as to attract the mobile capital. Intercountry distribution is affected, as returns to capital largely accrue to people in the countries where the headquarters of the multinational corporations are located and not in the countries where the profits are earned. Labor migration has led to a similar situation in that earnings in one country affect distribution in another. Remittances from migrants’ wages alter the distribution in the countries from where the migrants come, often in an equalizing direction. In Nepal, for example, remittances account for as much as one quarter of national income and are estimated to have accounted for one third to one half of poverty reduction between 1996 and 2004, according to one study (World Bank 2006b). A full understanding of the distribution of income within and between countries, and globally, therefore requires the measurement and analysis of factor shares.

This paper has drawn attention to what might be termed the inequality paradox: there are strong reasons for supporting more equality for justice and instrumental effects. Yet within countries, the forces making for more inequality are rarely countered, partly because of weak political capacity and appetite for such a move, encouraged by the broadly unequalizing policies supported by the most powerful global institutions. The contradiction between what is, and is accepted to be desirable, and actual policy making is the central inequality paradox. It is clear from experience that inequality-reducing policies need to be supported and claimed by strong political action, based on associations of the deprived and their allies. They are not likely just to happen because global rhetoric favors more equality.

A critical question is whether concern with equality and inequality should focus on national domains or the global arena. The natural reaction to this question is to argue for a focus on global justice and distribution. Yet contractarian theories of justice relate to the national level—including those of Rousseau and Rawls. And many of the instrumental reasons for concern with inequality also relate to the national sphere. Indeed, this paper has been largely concerned with national distributional issues. A further reason for concern with the national is that policy making occurs mostly at the national level. Nonetheless, shared humanity suggests that the global level is the right one from a normative perspective, as recognized by the international human rights approach. In addition, the fact that environmental sustainability can be achieved only at a global level requires concern with global distribution. That between-country inequality is estimated to account for around three quarters of global personal inequality and intracountry inequality only one quarter reinforces the importance of a global perspective. Even if we completely eliminated intracountry inequality, we would be left with very high levels of global interpersonal inequality (Lakner and Milanovic 2013).

Clearly, we need to be concerned with both intra- (vertical and horizontal)—and intercountry inequality. The main effective arenas of politics and policy are national, and it is here that people live their daily lives. For most people, it is the national level that generates a sense of living in a just or unjust society. National cohesion therefore requires attention to national distribution. But global distribution increasingly pervades people’s consciousness. It affects political and economic stability and environmental sustainability. A just, peaceful, and flourishing world requires a substantial reduction in inequality at both levels. And as globalization advances, with further increases in global capital, labor, and technology flows, it becomes more difficult to tackle inequality at either level without also tackling the other.