Can Social Policies Reduce Income Disparities?

Seminar on "Facing Up to Inequality," Inter-American Development Bank and Organization for Economic Cooperation and Development, Paris, France, March 13, 1999

The question on the table seems simple: Can social policies reduce income disparities in Latin America? My answer is straightforward: Probably not, at least not for a very long time indeed. But the question is the wrong one. In these remarks I first reframe the question: What in fact is and should be the objective of social policies in Latin America? I suggest one objective (which indirectly subsumes various other ones) should be to ensure opportunities for all, making societies ever more meritocratic – whether or not income disparities decline. Second, I outline the constraint that income disparities create in meeting that social policy objective. The fact of huge income disparities creates economic, political and institutional barriers to broadening opportunities for the poor in the region. Thus social policies are in some sense an outcome of income differences as much or more than an input to reducing those differences; in the immediate future, it is non-"social" policies that are likely to be more effective instruments for reducing income disparities. Third, I conclude on an upbeat note: social policies are likely to work better, even in the face of continuing high income disparities, as economies become more open and polities more democratic in the region.

A central objective of social policy: Enhancing meritocracy. Social policies have two kinds of objectives. The fundamental objective in democratic societies is to ensure all citizens receive some minimal access to education, health, the right to associate with others in collective bargaining and so on, as legitimate ends (merit goods) in themselves. In modern western capitalist economies, social policies are also meant to smooth out the harsh realities of the unfettered market via what is often called a social compact. Modern Western capitalist systems are built, in fact, on the bedrock of competitive market incentives tempered by government-managed help to those unable (and sometimes unwilling) to cope.

I am going to take this fundamental objective as a given, and focus on what might be called instrumental objectives, borrowing from a term used by Sen in a somewhat similar context. The instrumental objectives of social policies include: to increase economic growth by exploiting potentially high returns and positive externalities to investments in people (human capital investments); to reduce poverty by directing those investments to those now poor so that they and their children can become more productive, both in the household and in the labor market, and thus eventually escape poverty; and to provide insurance against economic and other shocks for all members of society – both to protect existing social investments (e.g. so that children need not leave school and pregnant women need not suffer malnutrition – in both cases risking permanent losses to them and to society) and to minimize social and political tensions (which in turn could threaten the growth objective). These three instrumental objectives are easily summarized: growth, poverty reduction, and insurance.

The fact is that sound social policies can be highly successful in supporting growth, in improving the access and contribution of the initially poor to growth thus reducing poverty, and in providing insurance against shocks, without necessarily reducing the region’s huge income disparities. The potential and actual contribution of increased human capital to overall growth has been amply demonstrated, and I need not belabor it here. The contribution of education to growth in the income of the poor has also been recently documented; indeed cross-country analysis suggests that the initial level of education has twice as great a positive effect on the income growth of the poorest decile as on average income growth (table). The repeated finding that initial life expectancy and increases in life expectancy are closely associated with country growth is almost certainly related to the fact that once countries have exploited the new medical and public health technologies that contributed so greatly to higher life expectancy in the first three decades after World War II, additional changes in life expectancy are particularly sensitive to improvements in the income levels and overall health of the poor. And by definition, sound social policy implies that the institutions and programs exist for social insurance.

So social policies clearly are associated with increased income for the relatively poor (in normal times because investments in health and education increase productivity and opportunities and in bad times because a safety net maintains the minimal household income needed to sustain such investments). But higher income for the poor is not likely to reduce total income disparities, simply because in Latin America, though the income of the poor has been growing (in most countries), the income of the rich has been growing too, and as fast or faster. While the economic and political reforms which now dominate in most countries appear to generate healthy income gains for the poor, they are at least as supportive of income gains for the rich, and of course proportionate increases in the income of different income groups generate much bigger absolute gains for the rich than the poor. For example, household income grew by about 50 percent in Chile between 1987 and 1996, with the gains reasonably well distributed across deciles of the distribution in the two years. However, in absolute terms, the income of the richest increased by 180,000 (monthly pesos of November 1987), while the income of the poorest increased by 7,000 (figure). Those in the third decile of the income distribution were under the poverty line in 1987 and above it in 1996, an important achievement. However, their income in absolute terms was closer to the median household’s income in 1987 (about 20,000 pesos less) than in 1996 (about 30,000 pesos less). Meanwhile the income in absolute terms of the median household went from 290,000 less than that of the richest decile in 1987 to 440,000 less in 1996.

Nor is Chile an unusual example. Overall income distribution is more equal in Costa Rica. There the income of the poorest 40 percent of households increased by about 130 percent between 1981 and 1995 while the income of the richest 10 percent grew about 100 percent (figure). Yet the absolute difference in income between the income of the third decile and the median household increased from about 1600 to about 2200 (1990 annual), and the difference between the median household and the richest decile increased from less than 13,000 to more than 26,000.

The fact that in many countries of Latin America, income is much more highly concentrated in the richest decile than for example in the United States or Great Britain, a point assessed so well in the 1998 IPES, contributes substantially to the absolute income disparities I have outlined. I return in a few minutes to the question of whether these differences matter.

Of course, income growth across the board, including for the rich, is not a bad thing in itself. Increasing income across the board is likely to reflect the benefits of a more dynamic, competitive economy, and to a large extent that is the story of Latin America in the 1990s, despite the recent problems associated with financial crises outside the region and in Brazil. As long as income gains are shared, and not confined largely to the already richer households, it is difficult to decry the changes, and silly to blame insufficient or ineffective social policies for failure to reduce overall income disparity.

Though reducing income disparities may not be a sensible objective of social policy, especially in the short run, reduced inequality can be a happy outcome of good social policy in the long run. Heavy publicly financed investments today and throughout the next decade in education and health, well-targeted to embrace the poor, will eventually (several decades hence) bring an improvement in the distribution of income and a gradual reduction in absolute income disparities in Latin America. The evidence from East Asia, where education expanded much faster and from a higher initial level, is suggestive. In East Asia, the distribution of education improved much more rapidly (figure) and the wage premia to skilled labor have increased much less. The result: income inequality (increasingly in Latin America and elsewhere, largely a function of labor income inequality), at least through the 1980s, did not decline and may even have improved. The rapid and broad increases in education, combined with macroeconomic stability and a labor market that functioned virtually without distortions explain much of the difference in both growth and the level of income inequality between Latin America and East Asia (figure).

In the short run, however, the reality in Latin America is that income distribution and income disparities are not likely to change much due to social policy nor even to general economic policy. On the contrary, it should not surprise us that market reforms provide signals that reward those who already have assets that generate income, be those assets land, education, or physical capital. It is well established that inequality of land is high in Latin America. It is also the case that though the distribution of education is almost certainly improving, since most countries are now on the good side of the inverted U-shaped Kuznets-like curve, it is still improving too slowly to prevent rapidly increasing returns to the educated and skilled, as the demand for skills in the labor market continues to be robust. As a result, real wage growth has been growing much faster for skilled than for unskilled labor in most of those countries for which we have data (figure).

What does all this imply? Probably that the formulation of an instrumental objective of social policy should stay as close as possible to the goal of improving opportunities across the board, including for the poor, thus making societies fundamentally more meritocratic – where any inequality reflects more differences in individual motivation, talent and willingness to work rather than differences in access to assets associated with the initial family income of the household into which an individual was born (or associated with ethnic or racial group or gender).

If the objective of social policy is to guarantee equal opportunity in a more meritocratic society, then the mechanisms of social policy would no doubt rely more on generating human capital through public investment than on guaranteeing entitlements per se, or guaranteeing transfers that are not need-based, i.e. targeted to the poor. Even programs of social insurance, including pension systems, unemployment insurance and other programs under the rubric of a social safety net, could be guided by the logic of ensuring maintenance of adequate opportunities (to work, to save, to invest in schooling and health at the time of economic or other shocks). This would be an approach to social policy, broadly conceived, which is more rooted in the Anglo-Saxon tradition than in the continental European tradition. It thus may represent more of a departure from traditional conceptions of social policy in at least some countries of Latin America, than its simple description in words conveys.

Constraints to effective social policy: Income disparities are a problem. It is now widely acknowledged that the design and implementation of effective social policy in developing countries is no easy task. In the 1980s and early 1990s, the widespread assumption was that macroeconomic and structural reforms could be easily complemented by increased social expenditures as a way to put a human face on the adjustment process. That was what social policy meant. Now there is increasing focus on the difficulties of implementing the "second stage" of reforms, where institutional capacity and broader consensus is needed to overcome formidable technical and political barriers to modernization and rationalization of government social programs that are technically complicated and often governed as much by the needs of special interest groups, including the public service workers who administer and staff programs, and by the political demands associated with clientalism.

I want simply to focus for a few minutes on the constraint that high-income inequality poses to implementation of effective social policy. The constraint arises on the demand side and on the supply side.

Consider first the demand for human capital. Latin America's large endowment of natural resources historically has limited society's demand for education. The socioeconomic arrangements that accompanied large-scale agricultural production and natural resource extraction involved relatively few owners of capital and many unskilled workers. There has thus been little demand for skilled workers, in part because natural resources tend to be complementary to capital, not skilled labor, in production. This is one message of the IPES 1998. Perhaps as a result, governments and families in Latin America have invested relatively little in education, given average income levels, seeing relatively higher returns to physical capital. A rich natural resource base in the region also minimized the need to develop competitive nontraditional exports in the early postwar period, thus perpetuating traditional production arrangements.

Second, high-income inequality in Latin America has implied that more households are liquidity-constrained, unable to borrow and without the resources necessary 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 elasticity of demand for secondary education of 0.50 (a conservative figure), if the distribution of income had been as equal in Brazil and Malaysia, secondary enrollments among poor Brazilian children would have been more than 40 percent higher. There is some evidence that, among the poor, the income elasticity of demand for basic schooling exceeds 1.0, in which case secondary enrollments among poor Brazilian children would have been more than 80 percent higher.

Third, household demand for education is not only a function of household income and household access to borrowing. It is also a function of expected returns to the family from schooling, in the form of higher future income for educated children. Two different public policies have systematically reduced the demand for basic education and other forms of human capital among the poor by reducing its expected returns.

First, postwar Latin American governments pursued import-substituting industrialization policies in an attempt to shift away from exporting primary commodities and to promote local manufacturing. These ISI policies resulted in large subsidies and protection for the owners of capital, but did not promote demand for labor (Schiff and Valdes 1992). As increased profits accrued to the owners of capital, real wages declined for the unskilled workforce. Relatively low wage growth among workers, combined with high returns to capital, did nothing to encourage demand for basic education among the poor. Additionally, some Latin American labor markets have discriminated against certain ethnic, linguistic or racial groups that also tend to be poor. This discrimination has reduced the expected returns to education among these groups and further reduced the demand for education among the poor.

The second problem has been educational policy itself. Low and declining quality of basic education in Latin America has reduced returns to basic schooling in the region, especially for poor households whose children are likely to attend the lowest-quality schools. High repetition and dropout rates in Latin America, especially among the poor, are sad testimony to parents' initial efforts to enroll children, and their growing discouragement as low quality and low achievement limit their children's learning and the expected economic returns.

At the same time, the supply of education in Latin America has itself been affected by the region's high-income inequality. When the distribution of income is highly unequal, providing subsidized basic education and health to a large proportion of poor households implies a relatively large tax burden on the rich. Though it is difficult to directly document, it seems likely that high income families have succeeded at least in some countries in resisting that tax burden, in part because they foresaw limited benefits to themselves. One result was, until recently, the underfunding of education, especially primary education, even in the prosperous years before 1982 —and a decline in quality. A second result has been the channeling of public subsidies to higher education, where children of the rich are more likely to benefit. In fact, a high share of the region's public spending on education is allocated to higher education—more than 20 percent on average, compared to 15 percent on average in East Asia. Venezuela and Korea are extreme examples. While in the early 1990s Venezuela allocated 35 percent of its public education budget to higher education, Korea allocated just 8 percent of its budget to postsecondary schooling. Public expenditure on education as a percentage of GNP was actually higher in Venezuela (5.1) than in Korea (4.5). However, after 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 costs more to reach the poor. Given any percentage of government expenditures raised via taxes and devoted to social programs, the country with a larger proportion of households below the poverty line faces higher costs in producing the same human capital outcome compared to a country with a smaller proportion below that line. Why? The poor are likely to reside in rural areas, where the costs of attaining the same degree of public access are greater. More important, the poor are usually less productive in "producing" human capital, of which the publicly financed input is only one and often a relatively small input. Health and education outcomes are in fact produced by households, using a combination of inputs, including housing, food, parental attention and so forth as well as public health and education services. Of course this also implies that at the margin the return to a public expenditure on social programs may be higher for the poor than for the nonpoor – though even this may only be true once above a minimal level of home-produced input.

In summary, income disparities complicate the lives of well-intentioned designers and administrators of sound social policy. Income disparities are usually associated with higher costs of reaching those who have the least income and higher average costs for producing the same increment in human capital per unit of public spending. At the same time, income disparities may increase the resistance to government’s raising and directing revenues to expenditures on social programs, especially those most likely to benefit most the poor.

The high concentration of income in Latin America and the increases in absolute income disparities associated with the recent period of reform-supported growth are thus a cause for constant vigilance if sound social policy is to be preserved and in fact strengthened as it must be in Latin America. This is a bottom line even though the reduction of income disparities is not in itself an objective of social policy. This is all the more case if, as Atkinson has suggested, societies’ tolerance for income inequality is a function in part of changes in inequality, so that one-time increases in inequality may alter social norms and make the new higher level of inequality more acceptable. (Atkinson compares changes in inequality, social acceptance, and government policy in the United Kingdom to other countries in Europe.)

Unfortunately, external shocks, be they financial or in the form of natural disasters, compound the challenge that social policymakers face. In Latin America’s emerging markets, with their dependence on foreign capital inflows, it is difficult to maintain social spending per beneficiary when overall deficits must be cut to reassure foreign markets that debt levels are sustainable (the problem of procyclicality noted in the 1997 IPES). And the shocks themselves in fact usually raise the amounts that need to be spent per beneficiary, especially per poor beneficiary. Thus reducing the region’s vulnerability to volatility could contribute greatly to sound social policy.

So where is social policy headed, and will income disparities grow inexorably? There are reasons to be optimistic. First, the opening of trade and capital markets means the business sector has an increasing interest in supporting education, health, child development and other social programs. Structural reforms and rapid technological change are already making the lack of secondary school graduates a badly felt bottleneck in some countries. It is increasingly understood that to maintain competitiveness in a global economy requires a more skilled and flexible labor force, especially given the relatively high wage levels in Latin America compared to Asian competitors in export markets.

Second, declining fertility rates in the 1980s are finally diminishing what had been for three decades tremendous pressure on the school system to accommodate more and more children every year. As quantity levels off, there is now room to focus on quality. This is one aspect of the demographic window of opportunity that was discussed this morning.

Third, the region's democratization is spurring the growth of civil society groups (NGOs, community groups, and increasingly pluralistic labor movements) that are more effective constituencies for better education and other social investments, especially among the poor, where individual parents have traditionally had little voice. Democratization has also brought to power Latin American presidents emphasizing social policy, such as Brazil's Fernando Enrique Cardoso and Bolivia's Gonzalo Sanchez de Lozada. Ministers of education, health and social welfare are no longer the last cabinet members appointed, and are thus more likely to bring technical and policy leadership as well as political strength to their immense challenges.

Two examples illustrate the potential for a new generation of effective social policy in Latin America. The first is Chile, one of the earliest reformers and one of the most open economies in the region. In the 1990s a combination of targeted money transfers and targeting of social program expenditures increased the share of income of the poorest 20 percent of households from 4.3 percent to 6.3 percent, and reduced the share of the richest households from 57 to 54 percent. Thus the ratio of income of the richest to the poorest went from more than 13 to one in terms of own-earned income, to 8.6 to one in total real income. Relative to initial own-income, the resultant increases in income and well being for the bottom quintile were large. Those increases, too easily summarized in dry numbers, in fact reflect a massive and concerted effort in Chile during the 1990s, with the transition to democracy, both to improve the targeting and effectiveness of all kinds of social expenditures and socially-motivated subsidies and to raise the amounts of public revenues dedicated to social programs.

A second example comes from cross-country analysis of the effects of economic and schooling policies on intergenerational mobility in Latin America. Analysis indicates not surprisingly that parent’s income and schooling are critical determinants of children’s schooling, but that other factors matter as well. In countries of the region with deeper financial markets, and where spending is higher on primary education per school-age child, parents’ characteristics matter relatively less; variation across countries within the region indicates that the policy seeds of increased intergenerational mobility – which lie both in the extent of internal market development, and in social policy – are bearing fruit in those countries where they have been planted.