Do you believe this: The poor suffered severly during the 1980s, and much of their suffering can be attributed to adjustment. Or this: Some of the poor---but certainly not all or even most---had their living standards fall, and without adjustment the suffering would have been greater.
The two poles anchor one of the biggest recent debates in development. To begin resolving the debate requires knowing what happened to the poor during the 1980s, a factual answer that emerges from careful analysis of the available information. To resolve it fully requires knowing what effect adjustment policies had on poverty. Not simple, for that requires knowing what would have happened without adjustment. The articles in the May 1991 issue of the World Bank Economic Review address both questions (see pages 8-10).
Analysis of how the welfare of the poor changed in the 1980s is complicated by four major measurement issues.
Looking at income. Well-being is the product of a range of factors, including adequate consumption of goods and services, health, status, achievement, and security. Because most of those factors can be bought in the market, income generally determines access to them---and thus is the most commonly used measure of well-being.
Sustained increases in the incomes of the poor require broadly based economic growth. Since the distribution of income changes only slowly, most reductions in poverty come mainly from increases in average income. The great majority of the poor are in South and East Asia (520 million and 280 million people in 1985), regions that sustained annual growth in per capita gross domestic product (GDP) of 2.8 and 7.0 percent from 1980 to 1988. Sub-Saharan Africa and Latin America, with 180 million and 70 million poor, experienced declines of 2.4 and 0.7 percent during the same period.
The majority of the world's poor---those in Asia---may have benefited from broadly based economic growth during the 1980s. In Sub-Saharan Africa and Latin America, however, the 1980s can be aptly labeled "a lost decade." Analysis of those (few) countries with reliable data drawn from household surveys confirms this general picture.
In Eastern Europe too, poverty increased. With the distribution of income remaining more or less unchanged, poverty moved in line with average income---increasing in Poland and Yugoslavia and remaining roughly constant in Hungary. Two patterns stand out. First, poverty became an increasingly urban phenomenon. And second, the increase in poverty was associated with a decline in real wages rather than a rise in unemployment.
Policymakers in these countries chose explicitly to distribute the social costs of adverse economic developments more evenly by allowing real wages to fall rather than unemployment to rise. Heavy public employment made such a policy feasible. But in most countries this is not an option, and the ultimate impact on poverty, especially in urban areas, depends on the flexibility and responsiveness of the labor market.
This point emerges clearly from analysis of the developed countries. Despite social insurance, poverty increased during recession---in Australia and the United Kingdom mainly because of increased unemployment and in the United States mainly because of lower wages among unskilled and low-skilled workers.
Unlike the other industrial countries, the impact of recession on the poor in Sweden has been much less severe, apparently reflecting the combination of short-lived unemployment insurance and the guarantee of a public sector job. This gives protection but minimizes adverse effects on incentives because one cannot receive benefits indefinitely. Most developing countries are unable to administer or finance such systems, but the basic principle underlies some public programs, such as the Maharashtra State Employment Guarantee Scheme in India.
The general relation between growth (or decline) and poverty notwithstanding, it is still possible for the poor to suffer a decline in well-being even when average incomes are rising. This seems to have happened in China, which undertook big policy reforms in 1979-83, especially in rural areas. Between 1978 and 1988 real average net income for rural residents approximately doubled, and real average expenditure for urban residents may also have doubled.
The dramatic improvement in average personal income produced a big decline in poverty in rural areas through 1983, a trend partly reversed thereafter. From 1981 to 1984 poverty declined significantly in urban areas. Although urban poverty was also lower in 1988 than in 1985, it rose after 1986. These setbacks originate in the dismantling of some aspects of the system of social security and services---and in stagnant agricultural production and accelerated inflation. The poor were more susceptible to adverse economic events.
What, then, of Africa? For five countries there, changes in relative prices led to no significant welfare gains or losses for poor rural smallholders despite the overall decline in average incomes. Most of Africa's poor are smallholders, but what happened to the urban poor? They and their fellow urban dwellers apparently suffered most from adjustment there.
Social progress. Economic growth does not translate automatically into comparable improvements in life expectancy and primary school enrollment. And temporary economic decline is not necessarily associated with a reversal in social progress. Yet, much of the debate has focused on the poor's access to basic social services during the 1980s.
When changes in measures of well-being under adjustment (such as those in Jamaica in 1984 and 1985) are viewed relative to the long-term trends of the measures rather than as changes in levels, they are much less significant than previously suggested. Measures related to educational and health outcomes have not deviated much from the trend, but the nutrition among small children has deteriorated somewhat.
There still is no definitive answer to the question of how poverty changed in the 1980s, but as more careful empirical analysis becomes available, some of the more extreme views of the 1980s will have to be modified. What's the agenda for future work?
Start with the absence of reliable data. Add the need to specify what would have happened if the adjustment policies under consideration had not been pursued. This need to compare results with an appropriate counterfactual applies to all analysis of economic policy, but it has often been forgotten in the debate on adjustment.
Among the policies common to structural adjustment programs is a devaluation of the domestic currency and a reduction of the disincentives to international trade. This pair of reforms is intended to increase foreign exchange earnings (improving the current account balance) and the perceived creditworthiness of the country (increasing capital inflows). It is also intended to induce a more efficient allocation of resources domestically to reflect the comparative advantage of the country. What is not known is whether these policies reduce welfare for the poor in the short run more than a continuation of the existing policy regime or some other package of policies.
Reform of fiscal (and related monetary) policies raises the same issue. Clearly these programs can reduce welfare in the short run, but in most cases they were preceded by falling terms of trade, reduced external finance, increased debt service costs, and unsustainable growth in external and fiscal deficits and inflation. Thus any analysis of the effects of structural adjustment on poverty must compare the outcome not with the preceding period, but with the outcomes that could be expected from alternative policies that would have been economically and politically feasible under the difficult conditions of the 1980s.
By stressing what would have happened without adjustment, the issue is correctly posed as one of choosing among alternatives.
A permanent deterioration in the terms of trade that reduces income throughout the economy also raises the marginal value of income to the poor, thus increasing the social value of income transfers to them. But the marginal value of investment also increases because the structural adjustment that is now required realistically can only take place through new investment. The tradeoff between the use of limited resources for these two options is the fundamental dilemma confronting the policymaker.
Should resources be transferred to the poor immediately following the shock, through a system of taxes and transfers that will inadvertently send some of the benefits to the nonpoor? And, if so, should the transfer be in the form of current income or productive assets, such as investment in human capital?
Modeling counterfactuals. In the first few years of adjustment, according to the Bourguignon model of the efficiency costs of poverty-reducing redistribution policies in an economy undergoing adjustment, the net social benefits of transfers to the poor will be less than the social benefits of investment. Then, when recovery is under way, investment in education and training of the poor will provide greater social benefits than will transfers of current income to them. The results depend critically on the assumed values of a few key variables, especially the increased productivity of capital during restructuring and the efficiency cost of taxes and transfers.
The level of abstraction in the Bourguignon model is its strength---and its weakness. To provide more realism, Bourguignon, de Melo, and Suwa analyze models representing a low-income African economy and a middle-income Latin American one. Their approach allows a richer representation of the underlying economic structure.
Relative to the Latin American model, the African one is characterized by greater wage and price flexibility, a lower stage of institutional development, and less integration with international financial markets. Expose the two models to the same external shock, and then explore two cases. The first is a no-adjustment case, in which the authorities rely on import rationing to support a constant real exchange rate. The second is a set of adjustment packages, including real devaluation, tax and tariff reform, and compensatory income transfers. These models trace the consequences of alternative adjustment paths on income distribution. This simultaneously reveals the impact on the poor and the sources of likely political resistance to adjustment (such as those benefiting from prior policies).
Two key points. The analysis of options reveals two key points. First, the no-adjustment option increases poverty: during the seven-year simulation period, the head count index of poverty increases from 30 to 40 percent in Africa and from 12 to 36 percent in Latin America. This result flows directly from the assumed reliance on import controls, a policy that creates rents that mainly benefit the rich.
Second, initial conditions have an important effect on the outcome (including the distributional consequences) of adjustment policies. Thus the option including a real devaluation helps most of the poor (those living in rural areas) in the African case through higher export earnings but hurts those in the informal sector. The same package, however, hurts the rural and urban poor in Latin America. Here the poor rely more on wage income. And because market rigidities prevent the economy from responding to the new structure of incentives, employment and real wages fall. Experience in countries as diverse as Costa Rica, Ghana, and Indonesia reinforces the conclusions from these models.
Back to growth. By far the most important point, and the main message of the Bourguignon model, is that the primary objective of adjustment should be to move the economy as quickly as possible to a growth path that allows effective implementation of the strategy outlined in the World Development Report 1990.
Three measures seem especially important:
In some cases (such as Indonesia) this combination of measures may reduce poverty even in the transition. In other cases (such as Costa Rica) poverty will intensify in the short run. Nevertheless this package holds the best prospect for a least-cost return to a growth path consistent with sustained increases in the well-being of the poor.
Temporary safety nets. Although it is important to keep this central point in mind, many countries have experimented with compensatory programs designed to act as a safety net during the transition. In designing such schemes, attention must be paid to their success in reaching the intended beneficiaries, their fiscal cost, and their impact on incentives. Public employment schemes are often proposed as cost-effective ways to transfer resources to the poor, if the offered wage is high enough to meet basic needs but not so high that it attracts the relatively rich.
Bolivia's Emergency Social Fund (ESF)---established as a temporary program during Bolivia's economic crisis to finance requests from local authorities for small, labor-intensive projects---is an example. Unlike most such schemes, the projects were carried out, and the workers were hired, by private subcontractors. The program has been very successful in attracting and disbursing substantial external funds. But has it affected the earnings of the workers hired?
At first glance it appears that the ESF has not done much in reaching the poor because most of the participants fall in the middle rather than the lowest deciles of the earnings distribution. But the definition of earnings in this calculation includes the earnings from the program. Estimates of the earnings of these workers without the ESF indicate, however, that about 30 percent of the participants would have fallen in the two lowest deciles of the earnings distribution and more than 70 percent in the bottom four deciles. This conceptually more accurate assessment of the fund's success in reaching the poor emerges through specification of a counterfactual.
The key element that determines the distributional effect of the program is the wage. Because private subcontractors are used, and because they pay market rates, the wage fails to act as a screening device. Thus, although the average participant in the program earns 51 percent more, about two-thirds of the transfer is required to replace forgone income, and not all of the net transfer accrues to the lower income groups.
The scheme, although reaching many relatively poor people, is not an especially cost-effective way to transfer income to the poor. Other programs, such as the Employment Guarantee Scheme in Maharashtra State in India, set wages low enough so that only the poorest choose to participate. In principle, this increases the share of the transfer reaching the poor and reduces the fiscal cost of any trans-fer to the intended beneficiaries.
No answers---yet. There cannot yet be a conclusive answer to how adjustment policies have affected the poor. But there is a correct way to approach the problem---through careful comparison of the results of alternative policies. For future work, one possibility is to assess the impact of alternative adjustment policies on key relative prices rather than on income. At the very least this approach would make best use of the available data while still emphasizing the importance of comparing alternatives.
But not all effects of adjustment work through prices. At least two other areas warrant more work. The first is the need to assess the distributional consequences of public expenditures, especially those for basic health and education. Although an analysis by income class may be feasible only rarely, examining the distribution of expenditures by region (most of the poor reside in rural areas) and by service (preventive versus curative care, for example) may be possible. It can also provide the basis for a more equitable (and efficient) allocation of spending under severe macroeconomic conditions.
The second area for added work is examining the effectiveness of programs designed to compensate those most vulnerable to the adverse effects of adjustment. Many countries are now pursuing schemes similar to the ESF to cushion the impact of adjustment on the poor. The fundamental issue in assessing these schemes is their cost-effectiveness in achieving their objective relative to other options. Subjecting them to counterfactual analysis could increase their effectiveness in transferring income to the poor.
From Lyn Squire's intoduction to the May 1991 issue of the World Bank Economic Review.