Admirably, our conference moderator, Jeff Sachs, wants to keep things simple by not adding to the list of what must be done in development. I do not want to add to that list either, but I must. Why? Frustration about Latin America-- where growth in the 1960s and 1970s did reduce poverty, but not by enough. Where early post-war growth was not sustained. Where growth in the 1980s collapsed, and the number of poor more than doubled from 70 to 150 million. And where in the 1990s, despite more than a decade of economic and social reforms, the return to growth has been modest at best and unsteady, and where the number of the poor has failed to fall. In Latin America, in short, though development has been a success when measured in terms of improving literacy and life expectancy, growth rates are still low and poverty persists.

So I want to add an ingredient to Jeff's list: the distribution issue. Not income distribution per se, but something more fundamental: the distribution of assets and opportunities, especially as it affects the poor. I want you to consider for a few minutes a possible lesson for Latin America from East Asia (where income inequality has for decades been much lower than in Latin America): that distribution of assets is relevant for ensuring that growth occurs from below, and therefore brings about poverty reduction.

Why has growth been consistently higher and poverty reduction so much greater in East Asia compared to Latin America? Why are the Koreans worrying about a drop in growth to 5 percent, while, outside Chile, 4 percent growth is considered a success in Latin America? The so-called Miracle study of East Asia's success, done at the World Bank, focused on broad-based, export-driven, shared growth. I believe behind shared growth was, in fact, what could be called growth from below; that is, growth fueled by increasing productivity of the poor in societies where the distribution of opportunities was relatively equal. To talk about distribution requires that I clarify its relation to the reduction of absolute poverty. Tonight I am fundamentally concerned with the reduction of absolute poverty. Societies tend to care about distribution, in addition to absolute poverty, if unequal distribution reflects destructive inequality, i.e. a lack of opportunities or a lack of mobility--which is likely to cause absolute poverty. There can also be constructive inequality, which provides incentives for mobility and rewards high productivity; this inequality usually reflects broad-based opportunities and is not associated with high absolute poverty and social immobility. Thus inequality in itself may or may not matter. But if it reflects or generates policies, programs and historical patterns in which the rich enjoy privileges and rents that ultimately undermine efficiency, growth, and poverty reduction, it certainly does matter.

World Bank Approach to Poverty Reduction:

At the IDB, my colleague Juan Luis Londoo and I recently prepared an assessment of World Bank poverty reduction policies for a session of the AEA conference held this past January on World Bank policies. One of our conclusions is straightforward: World Bank and other development economists have not focused enough on the fundamental issue of the distribution of assets and opportunities.

The World Bank's historic and continuing emphasis on growth as key to poverty reduction is absolutely correct. In 1968, the then-President of the World Bank, Robert McNamara, introduced the explicit goal of poverty reduction. But for all practical purposes in the analytic work and lending of the Bank, poverty reduction is seen as occurring through and because of growth. There has been some emphasis on distribution of income, but as an outcome rather than a determinant of growth or of poverty reduction. This lack of emphasis on income distribution and underlying asset inequality is not that surprising. Mainstream economic theory (in contrast to the Marxist tradition, which was certainly without influence in the World Bank) saw distribution as: (a) an outcome [Chenery et al., Redistribution with Growth]; or (b) as problematic in that redistribution through populist transfers has historically been a cause of destabilization and has inhibited sustainable growth. Even the relatively benign idea of investing in the human capital of the poor (as opposed to so-called productive investments in infrastructure) as a key to poverty reduction only appeared in 1980 [World Development Report 1980, special topic]. And in fact it was almost 10 years before Bank lending actually reflected this view--it took a decade to bring education and health lending up to a mere 10 percent of total lending.

In the 1980s, the strategy of growth and human capital accumulation as the means to reduce poverty was put on hold while analytical and operational work focused on adjustment issues. Then in 1990 came a second World Development Report with a special section on poverty. The report presented a three-pillar strategy for poverty reduction reflecting the history I have briefly outlined. The three pillars were growth, human capital accumulation via social programs, and safety net programs to protect the vulnerable and to alleviate poverty during periods of adjustment.

The Three Pillars in Latin America:

In Latin America, the three-pillar approach seemed an appropriate recipe, and it has in fact been implemented in the 1990s. Major economy-wide reforms, enacted starting in the mid to late 1980s, have brought a return to growth of 3-4 percent annually in the 1990s. Countries have also implemented major increases in their spending on human capital; social spending per capita (excluding pensions) has increased 22 percent in the 1990s, equivalent to an additional percentage point of GDP spent on health and education. Finally, most countries in Latin America have created some form of safety net, generally emergency funds for social protection.

With economic reforms, the region achieved some positive growth in the early 1990s, so that per capita income recovered to its 1980 levels. But average growth rates have been anemic, and some portion of the growth achieved reflects catch up effects after a period of no growth. Moreover, the overall results of the economy-wide reforms and increased social spending have been less than satisfactory for poverty reduction. With the possible exceptions of Chile and Colombia, countries in the region have managed little or no reductions in poverty in the 1990s.

Two other trends are worrisome. First is evidence of worsening income distribution over time, and its link to the minimal progress against poverty. If the economies of Latin America had maintained the same income distribution throughout the 1980s and 1990s as in 1970, the increase in poverty would have been smaller by almost half in the years 1983 to 1995. In other words, at least half of the rise in poverty since the 1970s is associated with a deterioration in the distribution of income. Second is evidence that the distribution of education itself is worsening. Using data on the education of adults (years of school completed) over the last three decades, we estimate that average education, though it has increased from about 3 years in 1970 to more than 5 years today, is becoming more and more unequally distributed (in that the standard deviation of average adult education has increased). Thus the pattern of inequality is being repeated over time. In sum, the three-pillar recipe in Latin America is not delivering the desired results, at least not yet. Moreover, the worsening trends in income distribution and distribution of education suggest that the recipe of growth, human capital accumulation and safety nets may not alone address Latin America's underlying problems of poverty and high inequality.

Growth and Inequality Across Countries:

In our work at the IDB, we have examined the relationship between economic growth, the income of the poor and inequality across a group of 43 countries over the past three decades. Using the high-quality income distribution data compiled by Klaus Deininger and Lyn Squire, we selected countries with Lorenz curves available for two periods of time separated by at least five years. For the resulting sample of 43 countries, we also use income estimates per capita in international purchasing power prices, and information on physical capital investment, the education of the labor force, land distribution, and trade indicators. Our regressions confirm the now standard results of growth analysis: economic growth reduces poverty, and income inequality reduces economic growth. So less income inequality would reduce poverty by increasing growth.

But there is more to the story. To the standard regressions we added variables measuring the initial distributions of land and human capital. We find that a more equal distribution of assets matters. It reduces poverty not only indirectly by accelerating economic growth, but directly by enhancing income growth of the poorest groups. In fact, the positive effect of lower asset inequality on income growth is almost twice as great for the poor as for the population as a whole.

Thus we have two virtuous circles: a more equal distribution of assets reduces poverty (1)indirectly by enhancing aggregate growth which in turn reduces poverty, and (2) directly. But the mirror image of these is a vicious circle where high initial asset inequality inhibits asset accumulation which traps the poor in poverty and, by limiting aggregate growth, reduces society's capacity to help the poor.

These findings are obviously relevant for Latin America. Growth has been anemic in the region and poverty reduction has been minimal. Income inequality is high and procyclical. These outcomes are related to weak asset accumulation (particularly of human capital) and high inequality of assets (land and human capital).

Compared to the East Asian economies, Latin America has had much lower physical capital accumulation. The region also had a lower level of initial human capital and higher initial asset inequality in human capital and land. Consider this. With East Asia's distribution of assets--land and education--in 1960, Latin America would have half the number of people living in poverty today. The number of poor would likely be even lower for two reasons. First, the number of poor would be lower if we were to take into account the effects of greater asset equality on growth, and of growth on poverty reduction. Second, the number would be lower still if, in fact, physical and human capital accumulation were a function of initial inequality (as could be tested in a structural model).

Lessons for the Multilateral Development Banks (MDBs):

Several lessons emerge for the multilateral development banks. First, the emphasis on growth and human capital accumulation as key to poverty reduction makes sense. Second, and less positive, more attention should be focused on a second key determinant of poverty reduction and aggregate growth: the distribution of assets, both physical and human capital. The long-standing inattention of all the MDBs to inequality in the distribution of assets, especially education, has been costly. More concern earlier with the causes and the consequences of income inequality would have called greater attention to a fundamental constraint on poverty reduction: the poor's lack of access to the assets necessary for increased productivity and income.

In the context of Latin America, the multilateral development banks have long decried populist transfers. There is an alternative: to focus on programs that put productive assets in the hands of the poor. This means focusing not only on expanding education, but on its distribution as well. It means seeking other mechanisms beyond education to increase the access of the poor to productive assets: land reform, reform of legal systems, credit, and fair competition. All of these can create opportunities in previously unequal societies, eliminating the hidden privileges in asset markets historically enjoyed by the rich. The growing support of the MDBs for microenterprise programs acknowledges the relevance of access to assets and opportunities for income growth among the poor. Similarly, new emphasis in the banks on political and economic decentralization and on participation of the poor in the design and implementation of social and economic programs, with real voice and the power of choice, can be effective in poverty reduction. In democratic societies only political access and economic freedom can help ensure greater access to the assets that will raise incomes.

Thank you.