Human Capital Development and Operations Policy HCO Working Papers Protecting the Poor During Adjustment and Transitions K. Subbarao Jeanine Braithwaite Jyotsna Jalan July 1995 HCOWP58 Papers in this series are not formal publications of the World Bank. They present preliminary and unpolished results of analysis that are circulated to encourage discussion and comment; citation and the use of such a paper should take account of its provisional character. The findings, interpretations, and conclusions expressed in this paper are entirely those of the author(s) and should not be attributed in any manmer to the World Bank, to its affiliated organizations, or to members of its Board of Executive Directors or the countries they represent. Protecting the Poor During Adjustment and Transitions by K. Subbarao Jeanine Braithwaite Jyotsna Jalan A background paper for the World Bank contribution to the Social Summit, Advancing Social Development ABSTRACT This paper reviews the programs implemented in selected countries to protect the poor during normal times and during transition or adjustment. In particular, two questions are addressed: how have countries implemented the programs in practice, and what are the lessons. The paper reviews five sets of programs: (i) food subsidies and related interventions, (ii) food for works and other public works programs, (iii) credit-based self-employment programs, (iv) social funds and related interventions, and (v) child allowances. More representative than comprehensive, the selection of programs and countries is based on the availability of rigorous evaluations, the importance of the program, and the issue under consideration. In choosing between the instruments and their duration and coverage, governments face key issues, including how much to spend and how to finance programs; how to target assistance to the poor, within politically feasible limits; how to select the most appropriate delivery mechanism; and how to avoid adverse effects on incentives for labor supply and private savings and transfers. The reviewed program and country experience offers a few key lessons. First, it is essential to institute safety nets as an integral part of the overall package of economic reforms, instead of as "add ons" after the program begins. Second, the instruments should build on, and complement, existing arrangements, both formal and informal. Third, the outreach of programs and benefits must be held in check, and the targeting must be effective in order to avoid (i) adverse effects on incentives to work, and (ii) crowding out private transfers. In general, social assistance transfers are best financed from the general revenues. However, excessive funding for transfers would either (i) crowd out other developmental expenditures, or (ii) require heavy taxation that could kill incentives and distort results. Therefore, spending levels on poverty-targeted programs have to be low and consistent with the government's overall target for fiscal balance. 3 Table of Contents Introduction ........... 1 Safety Nets: An Overview Of Conceptual Issues .1 Safety Nets: The Range Of Instruments .1 Choosing Between Cash And In-Kind Transfers .I Coverage And Targeting ................................2 Financing Of Safety Nets .3 Current Levels Of Public Spending .3 Characteristics Of Vulnerable Groups .4 Private Safety Nets And Public Transfers .5 Safety Nets In Practice .6 Food Subsidies And Related Interventions .7 Public Works .10 Credit-Based Livelihood Programs .12 Social Funds And Related Interventions .13 Child And Family Allowances .15 Lessons From Country Experience On The Design Of Safety Nets .16 Introduction Protecting the poor and the vulnerable dunng normal times and during periods of economic reform is critical for alleviating poverty and sustaining the reform process. Inflation, contraction in output, decline in real wages, and other developments in some low-income adjusting countries, and countries in transition, underscore the need for public social safety nets to protect vulnerable groups from short-term poverty. Such safety nets are especially needed in countries where (i) the coverage of formal systems of social security is inadequate, and a large proportion of the population is poor and excluded from formal sector employment and consequently from social security benefits; (ii) formal systems are in disarray, fiscally unsustainable, and unable to protect vulnerable groups; and (iii) family-based social security arrangements are fading. In most developing countries, conventional programs of income maintenance and unemployment compensation are either infeasible fiscally and/or unable to handle the dual nature of economies with large informal sectors and widespread open unemployment. At the same time, the need for social assistance is acute because adjustment or transition may have lowered living standards of vulnerable groups. Moreover, such transfers may be necessary to build political consensus for sustaining economic reform. This paper briefly discusses the conceptual issues underlying the choice between various safety net instruments. It also reviews the country experience with selected safety net programs. A summary of some lessons from theory and practice is presented in the last section. Safety Nets: An Overview of Conceptual Issues Safety Nets: The Range oflnstruments Governments have the choice of two main instruments for reducing short-term poverty. The first instrument is an income-generation program; it carries some obligation from the recipient, such as the requirement to work This instrument consists either of credit-based self-employment programs or of labor-intensive public works programs. Both involve subsidies to a lesser or greater extent. Both can also be adopted either as temporary measures to ease hardship caused by economic shocks or, more broadly, as "direct" measures to alleviate poverty. The second instrument is a transfer. Unlike an income generation program, it carries no obligation from the recipient. Governments can transfer resources to the poor in cash or in-kind. A range of programs is available including (i) nontargeted in-kind transfer programs, such as general consumer subsidies; (ii) and weakly targeted cash transfers, such as universal family and child allowances (prevalent in Former Soviet Union [FSU] countries and Eastern Europe); (iui) better targeted programs, such as food stamps or cash compensations to poor households following the abolition of subsidies on food staples and severance payments, and (iv) cash transfers such as child allowances. Choosing Between Cash and In-kind Transfers Each type of transfer has particular merits and drawbacks. In general, cash transfers are preferred because they leave consumption decisions to the recipient This plan avoids distortions in the product markets that may occur with in-kind benefits (Newbery and Stern, 1987; Besley, 1993). Moreover, cash 2 transfers are generally more transparent in the budget, and the cost to the government can be estimated and known in advance. In some situations, however, in-kind benefits are the preferred choice. In-kind benefits can be direct subsidies on goods or services that can only be consumed by the recipient (the classic example is free school lunch); they can also be the provision of goods or services (or vouchers) easily converted into cash by the recipient. One immediate practical problems is: Which commodity to subsidize? The best choice is a good whose consumption will increase when its price is lowered (subsidized) proportionately more for the poor than the nonpoor. In theory, in-kind transfers are preferred if the good in question is in the nature of a public good, or if competitive markets for that good are absent (e.g., in energy distribution, scale economies may dictate that only one provider can operate efficiently) (Jimenez, 1993). However, there are other important considerations. In-kind benefits that have a negotiable cash value do not guarantee that recipients will use the benefit only for consuming the particular good. Countries seeking to influence consumption may find that providing goods or vouchers results in very little additional control and a lot more administrative work than cash benefits. On the other hand, the real value of in-kind benefits do not erode during inflation as nonindexed cash benefits do. In countries with weak institutional or administrative capacity, certain types of in-kind benefits may be easier to implement than cash transfers. Even many rich countries provide in- kind assistance (e.g., food stamps in the USA). Political considerations may force some countries to provide education in the public sector or other such services for achieving nationalistic goals. Finally, it is sometimes easier to target in-kind transfers than cash transfers because of the scope for self-selection (e.g., subsidizing coarse grains consumed only by the poor can ensure self-targeting). However, self-selection only helps to eliminate some people who do not truly need the benefit; self- selection does not guarantee 100 percent take-up rates by the targeted group. Self-selection can reduce costs and leakage of benefits to the better-off, but it does not ensure that benefits always go to the poorest Moreover, both cash and in-kind benefits can be structured for greater self-selection. Coverage and Targeting An important aspect common to both cash and in-kind benefits is targeting. Targeting can be based on an income-tested definition of the poor, or it can be focused on selected groups defined by category (e.g., children, the disabled). All targeted programs involve a trade-off between the desire to cover all the poor and the risk of wasting resources if the program benefits leak to the nonpoor. Financial constraints dictate the level of coverage provided, but the kind of targeting also affects costs. For example, it costs more per claim to identify smaler groups in geographically dispersed locations. Also critical are these two factors: the avoidance of disincentives to work for persons just above the income- level that qualifies for benefits, and the provision of incentives to get recipients to move off program rolls. While targeted approaches are preferable, targeting assistance-cash as well as in-kind--to the poor has never been easy. First, identifying the recipients on the basis of a means testing or by household characteristics is often difficult. For example, in a country like Albania, the scant knowledge available about the living standards, and income and expenditure patterns prevent any systemic efforts to identify 3 the poor and target assistance to them. Second, even when identification is possible, weak administrafive capacity proves a limiting factor. Third, governments are forced to initiate social safety net reform under many constraints, including a lack of public understanding about the necessity to target and the benefits of targeting. Even with a well-informed public, social safety net reform (as other changes in economic regime) creates winners and losers, and governments can be faced with genuine popular opposition from groups who will lose existing benefits. For example, removal of open-ended general consumer subsidies tends to be politically unpopular. Typically, when a subsidy is abolished, most governments find it difficult not to meet the demands of vocal and organized groups for compensation, even though they are not poor. Governments must necessarily balance the interests of protecting the poor with the political imperative of compensating the vocal and organized groups. One way to strike a balance is to resort to a self-selection approach where feasible, rather than exclude some groups from the scope of programs altogether. Financing of Safety Nets In both adjusting and transitional economies, any transfer program that imposes a heavy fiscal drain simply has no chance of being sustained. The fiscal necessity of reducing entitlements needs to be balanced with the need to reach the poor, and governments must deal with the difficult choice of deciding who shall be protected. Such entitlements as pensions (typically the main fiscal burden) may be at the expense of assistance to other needy groups, such as poor children. Transfers may crowd out resources for investment; so there is a trade-off between providing temporary income transfer to vulnerable households and long-term economic growth that benefits all households. Such trade-off may be more stark in the FSU countries than elsewhere, given the large declines in output since 1990. High levels of transfers also require high levels of taxation and with such levies come inevitable distortionary and disincentive effects. Disincentives to work are most likely to emerge as incomes are raised and the transfer is withdrawn in countries with high implicit marginal taxation. Current Levels of Public Spending The amount of public spending on social protection, and who benefits from them, usually varies across countries depending on (i) the strength of private safety nets, (ii) the preexisting structure of poverty and the composition of vulnerable groups, (iii) the nature of programs and the extent of targeting, (iv) administrative capacity, (v) available financial resources, (vi) the speed and sequencing of adjustment or transition, and (vii) the political context of the reforms. Table 1 shows the current public spending on social protection for selected developing and developed countries. The data show only central government spending; they do not cover expenditure by state, provincial, or local governments. Therefore, the data gives an incomplete picture of how deeply involved a government is in the provision of social protection. Subject to this limitation, some patterns are worth noting. In general, public spending on social protection is high in OECD countries, Latin America and Caribbean, Eastem and Central Europe and Middle East, and North Africa countries. Public spending on social protection is lowest in relatively poor countries 4 of Sub-Saharan Africa and South Asia, as well as in relatively better-off countries of East Asia. For example, Malaysia, with a per capita GNP similar to Brazil, spends only a third of the latter. To understand these variations, it helps to know the differences in the characteristics of vulnerable groups, the strength of private safety nets, and the overall economic performance. Table 1: Central Government Spending on Social Protection for Selected Countries, 1992 Total Public Public Spending on Social Security, Region Country GNP per capita Expenditure Welfare & Housing (% of GNP) (% of GNP) OECD USA 23,240 25.3 7.3 Canada 20,710 23.9 8.7 UK 17,790 38.2 12.1 East Asia Singapore 15,730 22.1 1.8 Malaysia 2,790 29.4 3.4 Thailand 1,840 15.5 1.0 Philippines 770 19.4 1.0 Latin American and Caribbean Argentina 6,050 13.1 5.2 Mexico 3,470 17.9 2.3 Brazil 2,770 25.6 9.0 Eastern and Central Europe Russia 2,510 26.8 6.6 Poland 1,910 29.3 6.0 Middle East and North Africa Iran 2,200 19.7 3.9 Tunisia 1,720 32.8 6.1 South Asia SriLanka 540 28.2 4.5 Pakistan 420 21.7 0.7 India 310 16.8 1.0 Nepal 170 18.7 1.3 Sub-Saharan Africa Lesotho 590 33.2 1.8 Kenya 310 30.7 1.0 Malawi 210 26.6 0.9 Source: World Development Report, World Bank (1994). Note: The data should be interpreted with care. First, because of differences in coverage, the individual components of central government spending may not be strictly comparable across countries. Second, the data do not cover expenditure by state, provincial, and local govermments; this gap may seriously understate or distort the statistical portrayal of how resources are allocated for the various purposes, especially in countries where subsidiary levels of government are responsible for many social services. For further discussion of the construction and limitations of the data, see the technical notes in World Bank 1994. 5 Characteristics of Vulnerable Groups In all countries, three types of vulnerable groups exist the chronically poor, whose income levels are below an acceptable minimum (e.g., poverty line) even during periods of economic growth; the temporarily poor, whose income levels fluctuate above and below an acceptable minimum level dunng periods of natural or economic shocks; and those groups, poor or not, directly affected by adjustment (e.g., workers of a closed, loss-making public sector unit or those employed in the private sector during recessions). When creating policy, it helps to distinguish between those poor people with potential to move beyond poverty (e.g., those of working age, in good health) if appropriate short-term help were provided, and those poor in need of long-term support (i.e., the disabled, children, the infirm and the elderly) because of limited capacity to generate income. The affected vulnerable groups can be identified by their varied household characteristics.'For example, in most of Sub-Saharan Africa, the majority of groups adversely affected, in the short run, by adjustment are found in rural and urban informal sectors and among marginal farmers. In South Asia, the landless rural poor and the homeless urban poor (such as construction workers) are the hardest hit by short-term impacts of adjustment In the FSU countries, children, certain-kinds of pensioners,2 and the unemployed have taken the hardest blows from inflation and job losses. In Latin America, urban workers in both the formal and informal sectors, and landless rural people have been most severely affected by adjustment In some countries in transition (e.g., Romania and Mongolia), women, particularly female-headed households, have emerged as an important vulnerable group. Before the efficiency and adequacy of publicly provided safety net programs are reviewed, it is useful to know the strength of private safety nets and how they vary between countries. Private Safety Nets and Public Transfers Many developing countries have effective family support systems and private income transfers. Thus, the informal transfers on private accounts are considerable in the Philippines (Cox and Jimenez, 1992b). In China, not only are informal family-based support systems strong, but the structural features of the rural economy itself-access to land, either individually or collectively-guarantees economic security. To corrcty delineate household characteristics of vulnerable groups adversely affected by adjustment or transition in each country, reliable household data sets are required, based on household-level income and expenditure surveys. Unfortunately, such surveys are available for only a few countries to date. 2 Pensioners form a very heterogenous group in the FSU countries, because the term is used to denote cash transfer recipients in general, not solely old-age pensioners. Old-age pensioners themselves are not a homogenous group. Approximately one-quarter of them continue to work full-time and to receive a full pension, and less than one-fifth live alone (without direct access to wage income of other family members). Although savers saw the value of their nominal assets erode due to the veiy high inflation of 1992-93, some were able to shift wealth into real assets that maiined their value or even increased in nominal terms (e.g., real estate and private plots (agricultural land), as well as stores of soft goods and food). 6 In Sub-Saharan Africa, households and individuals cope with periodic crises by making claims for assistance on kin and fellow members of the same community. Such assistance may take the form of transfers of labor or, very often, of food. In practice, such networks remain the principal safety net for most of the poor, especially the rural poor. Thus in The Gambia, a relatively land-rich country, an institutionalized system of community-level food security and welfare can be found in the Zakat system. Under this system a tax revenue (10 percent of delivery obligation of staple food to the mosque) is redistributed to the poor and needy in the village community. Such intravillage communal systems reduce the pressure on household-level targeting in providing social assistance. However, for various reasons (e.g., adjustment-induced stress, sharp output fluctuations, and increasing poverty) such community and family support networks are eroding and are inadequate to cope with the situation. Countries with strong informal safety nets fall into two categories. One group consists of countries in East Asia and the Pacific Region, where impressive economic growth obviates the need for formal safety nets (the exception being the Philippines). For example, in Malaysia, there is virtually no publicly funded safety net: rapid economic growth has been the mainstay. This group of countries have avoided expensive public safety net programs by a combination of circumstances: buoyancy of private safety nets, rapid economic growth, and poverty reduction. Poverty incidence in East Asia fell from about one-third of the population in 1970 to one-tenth in 1990. The second group comprise countries where slow economic growth, poverty, and adjustment-induced pressures are harsh on vulnerable groups, notwithstanding the prevalence of informal systems of support In the latter group of countries, cost-effective, publicly funded safety nets are needed. Yet, designing a publicly funded and targeted safety net that does not displace private transfers is a challenge in these countries. Moreover, the overarching necessity to accelerate growth makes the policy for poverty-targeted safety net programs particularly difficult-an excessive reliance on safety nets may reduce investable resources for growth. Without growth, no amount of targeted programs can provide relief to the poor. In order to keep spending levels low, the size of benefits and the coverage of programs have to be carefully chosen. When making the choice, govermments have to bear in mind not only the potential for crowding out private safety nets but also the sharp trade-offs between investments in growth and spending on poverty-targeted programs. Thus, in choosing between the instruments and their duration and coverage, governments have to confront the following key issues: how much to spend and how to finance programs; how to target assistance to the poor, within politically feasible limits; how to select a delivery mechanism that is most appropriate; and how to avoid adverse effects on incentives for labor supply and private savings and transfers. Also this important policy issue arises in the countries undergoing adjustment or transition: how to integrate a safety net program within the reform package, bearing in mind the above issues. Safety Nets in Practice What specific programs have been instituted to mitigate short-term poverty among the vulnerable groups? How successful have these programs been? Answers to these questions are found in five broad sets of programs reviewed in this section: (i) food subsidies and related interventions; (ii) food for works and other public works programs; (iii) credit-based self-employment programs; (iv) social funds, social action programs and related interventions to mitigate short-term adverse 7 consequences of adjustment; and (v) child allowances-the principal cash transfer in the FSU countries. Food Subsidies and Related Interventions Food subsidies have existed for more than four decades in some Asian countries. For example, India's Public Distribution System (PDS) goes back to 1942, when it was introduced to deal with wartime scarcities. Its objectives were to provide quantity rationing of scarce commodities to the population, stabilize consumer prices, and mitigate poverty and increase food consumption of the poor. The principal shortcomings of the PDS were inadequate coverage of the low-income households, inadequate supply of commodities consumed by the poor, supply of commodities consumed by the nonpoor (such as edible oils), and untargeted distribution of benefits (transfers) to higher income groups. The exception in the country is state of Kerala; here rations are spread fairly evenly between rural and urban areas; the poorest 60 percent receive 87 percent of benefits (World Bank, 1990; Subbarao, 1992). But for the country as a whole, subsidy costs grew, and the program became fiscally unsustainable. In 1992, a general political consensus emerged to limit the PDS to the most needy and deserving households in all states (along the lines in the state of Kerala), and the government announced a plan to revamp the PDS. Very poor blocks and households previously unprotected were now covered by PDS. Various targeting approaches based on region (the 1700 priority blocks), income, registered employment categozy, land holdings, and self-reported income are being pursued by different states. Bangladesh has a public food distribution along similar lines as India. Though some nonmonetary benefits did reach the poor, the absence of targeting led to substantial leakage of benefits to the nonpoor. Moreover, it is one of the most expensive programs in the region. Recent estimates suggest that for transferring one taka of income to an eligible household through a rural rational system, the program costs six times the income transfer. In Nepal, the National Food Corporation (NFC) provides food at below market prices. The benefits of the program largely go to the civil servants in Kathmandu, despite the lack of poverty in the area and the presence of a well-established market system. NFC food is flown at great cost to remote areas, and the transport costs are heavily subsidized. The food distribution has not had a significant impact on the poor, despite its high cost A generalized food subsidy has been in operation in the Philippines for several years. Recent analysis suggests that the poorest income groups and regions are missing out on their share of income transfer from the food subsidy program. The regional distribution of the food subsidy is extremely unequal. During the past three years, the National Capital Region and the Cagayan Valley, which account for only 2.8 percent of the nation's food poverty gap, received as much as 34.5 percent of the subsidized rice. The food subsidy program is also costly. It costs the Government three pesos to transfer one peso of income to the households that are not necessarily the poorest in the country. Untargeted and urban biased, the program fails to reach the households with the greatest need. Prior to 1979, Sri Lanka operated a general food price scheme that subsidized food in unlimited quantity to any buyer. The scheme had a large impact as a safety net for the poor because of its extensive coverage of different types of poor people and high value of benefits. The scheme's excessive cost prompted the government to target the program. Following the fiscal crisis of 1979-80, 8 Sri Lanka replaced its four-decade-old general food subsidy with a food stamp program. Ihe share of food subsidy dropped from 5.0 per cent in mid-1970s to 1.3 percent of GDP in 1984. But the real value of food coupons, which were fixed in nominal tenrs, was eroded; consequently the value of benefits fell by half. The program is income-tested, with the amount of food stamp receipts depending on the household's income, size and composition. Analysis based on the 1990-91 household-level survey data concludes that the programn is well-targeted, and food stamp receipts are progressive. Simulations also suggest that poverty with food stamps is lower than it would be under uniform transfers of the same gross budget (Datt and Gunewardena, 1994). Food security and providing adequate nutrition for the poor are perhaps the two primary concerns of the governments in Africa. Unfortunately, public programs to ensure food entitlements for a poor household have not been very successful in much of Africa. In general, the programs were poorly targeted. Both emergency and nonemergency initiatives were heavily dependent on foreign aid. Considering that foreign aid will continue to play a critical role in the future, the most urgent task is to improve the effectiveness of aid in improving household-level food security. In the past, the method adopted was direct provision of food to the households. This approach proved costly and also encountered logistical difficulties. Household surveys suggest that the poor spend as much as 85 percent on food. Food aid is currently sold in the market, often at less than market prices to promote the consumption of the poor, but this practice also led to disincentives to domestic production. People not consuming food adequately need not necessarily be given food, or rather, they can be aided by transferring cash, food stamps, or food rations. All of these methods may prove to be more cost- effective and may reduce distortions better than selling imported food at low prices (mostly in urban centers). If food aid continues to be used in a way that generates distortions and reduces producer incentives, it may significantly reduce the benefits from all other safety net programs aimed at short- term income transfers to the poor. In Africa, Botswana is an exception in the role of protecting vulnerable groups. In the 1980s, Botswana implemented a broad, integrated program for drought relief and recovery. The effort combined food supply management, employment generation, and agricultural assistance. Large-scale imports of food were combined with support for rural incomes through public works and grants. Food rations were given to households with children under 10, pregnant and lactating women, and destitutes. Intensive feeding was provided for malnourished children at health facilities, and a feeding program for primary school children was maintained. Botswana's integrated strategy of drought management contained the adverse effects of severe drought and protected the poor's entitlements to food and nutrition. The main objective of ration schemes is to provide fixed quantities of food at fixed prices. Such schemes aim to switch the impact of price changes from the buyers of rations to the government, thus reducing both chronic and transitory food insecurity among the poor. In Pakistan and Egypt, a food ration and subsidy system prevailed. In Pakdstan, three household surveys during 1977 and 1986 showed greater use of the ration system by the low-income groups, a result that indicated self-selection. In 1986, while 32 percent of urban households with monthly incomes of less than Rs.500 used the ration system, only 15 percent of households whose incomes were higher than Rs.2,000 did so (Alderman, 1991). Such self-selection took place when the government decided that it was no longer mandatory to purchase sugar (a commodity largely consumed by the nonpoor) at the ration depots. 9 The scheme was inefficient, nevertheless, since it was only partially targeted. The govermment switched from a partially targeted but inefficient subsidy to a general subsidy on wheat in 1987. This derationing significantly reduced unauthorized profits that had accrued to various intermediaries in the old system. In Egypt, a complex system of government control exists for the marketing and distribution of rationed quantities at low prices through cooperatives, flour stores, and bakeries. The regulations of the system vary by provinces; thus the system's impact on the poor is difficult to assess. Household data suggests that the poor in villages have less adequate access to the system than the poor in cities. However, the system as a whole favors the poorer groups of population more than the upper-income groups. Evidence also shows that even poor consumers appear to be less willing than the rest of the population to wait in line or be subject to other search costs (Alderman and von Braun, 1984). Overall, in countries lacking the infrastructure necessary to target programs to the families on the basis of means tests, self-selection by commodity choice appears to be the most appropriate vehicle for channeling food subsidies. Where such self-selection is hard to achieve and identification of the poor strains the administrative machinery, health and nutrition programs for children and women in regions of high mortality would probably serve the goal of household welfare more efficiently than either rationed food or general food subsidies. Where the program is based on carefully designed targeting principles, food security and nutrition programs have reached the vulnerable groups. In Malawi, for example, a major feeding program implemented through the hospitals and clinics successfully targeted malnourished children and pregnant and lactating women. To reach families far from hospitals and clinics, a pilot project is proposed to target farmiing families with young malnourished children. Benin and Burkina Faso adopted regional targeting. Certain areas of the countries were identified as priority areas based on certain parameters related to food insecurity. Once the areas were defined, households with certain characteristics (i.e., dependency ratio, access to credit, literacy) were identified as the most vulnerable and targeted for food subsidy. As noted, some countries have switched to food stamp programs as a substitute to open general subsidies and/or rationing schemes. Sri Lanka switched from the ration system to food stamps in 1979; as a result, total government spending fell from 14 percent to 7 percent of the budget. The program was means-tested. Nonetheless, the administrative cost was only 2 percent of the program cost (Cornia and Stewart 1993). In 1984, Jamaica launched a food stamp program. The beneficiaries of the program were pregnant and nursing mothers and young children, regardless of household income. The program was effective in restricting access to the poor, perhaps because the stamps could be collected by the categories of households only at the primary health clinic in exchange for immunization of children and pregnant women, and nutrition counseling of mothers. Since the nonpoor did not frequent these clinics, the program effectively reached the poor, and the administrative cost was 4 percent of program cost (Grosh, 1994). An early assessment of a similar program in Bangladesh (known as Food For Education Program) where stamps were linked to households' obligation to send children to school, suggests that the targeting was successful (IFPRI, 1994). Moreover, the program effectively promoted enrollments and reduced drop-out rates among the children of poor households. These cases illustrate that food-based interventions can be targeted and implemented cost effectively. 10 If not implemented and targeted well, programs like food stamps can easily lead to fraud. Zambia is a case in point Large-scale counterfeiting of food coupons in Zambia led to virtual elimination of the program. The program also suffered from the absence of an innovative targeting mechanism. Overall, one conclusion stands out: food stamps yield a more progressive pattem of transfers than food rations or a general food subsidy. However, only a few countries succeeded in controlling administrative costs and achieving a reasonable measure of success in targeting. It is possible, evidence suggests, to combine a food stamp program with innovative requirements that can ensure better targeting and lower administrative costs. Public Works South Asia pioneered public employment programs, which now constitute an important safety net in India and Bangladesh. With the labor force rising by six to seven million new entrants each year, open unemployment standing at 17 million, and employment growing less rapidly than the labor force, the Indian government's most important safety net program has been the creation of employment, especially in rural areas. India has an impressive record of generating 875 million person-days of work in 1991, with 55 million persons gaining some employment under the program. A recent evaluation shows that the welfare benefits of the program are reasonably well-targeted to the poor; more than 70 percent of the person-days are used by persons below the poverty line, and more than half are used by the scheduled castes and tribes and the landless who account for the highest incidence of poverty. Public employment programs in India have effectively reduced the adverse consequences of droughts (Rao, Subbarao, and Ray, 1988). Yet, the program abounds in shortcomings: women do not fare well (receiving only 25 percent of benefits); the assets created are of poor quality; maintenance is inadequate; and wages offered are the "minimum wages," usually above the market rates, thus reducing the scope for self-targeting (Subbarao, 1992; Kakwani and Subbarao, 1993). The Maharashtra Employment Guarantee Scheme (MEGS) is now internationally known as one of the most effective safety nets for the poor in the state of Maharashtra, India. One reason for the initial success of the program was the setting of the wage rate lower than both the minimum and market wage rate. This peg ensured self-targeting and limited outreach. The gap between the scheme's wage rate and the minimum wage has now narrowed, if not closed, resulting in some reduction of the self-targeting feature of the program. Recent evidence, based on household data, suggests that the poor's participation in MEGS was highest, and self-targeting worked much better in this scheme than in a means-tested credit program like the Integrated Rural Development Program mentioned later (Datt, 1994). MEGS, over the last twenty years, considerably expanded the irrigation infrastructure, thus expanding the area under double-cropping. Despite its success and popularity, the scheme is criticized for creating low-productivity assets, which have been poorly maintained. Also employment generated under the scheme has been steadily declining. This decline is interpreted by some analysts to be indicative of rationing following wage increases (Ravillion, Datt and Chaudhri, 1993). Other analysts attribute the decline to a drop in the demand for MEGS employment due to an increase in the market demand for agricultural laborers. Notwithstanding this controversy and the criticisms about the scheme, the program uniquely demonstrates how to combine short-term 11 employment generation with infrastructure development, and at the same time how to be an effective safety net for the poor. Bangladesh has also experimented with public employment programs since 1962. Like they did in India, the benefits did reach the poor, especially in recent years. Unlike in India, the program was largely financed by external food aid. With the redirection of food aid to other countries, it is unlikely the program will be maintained at levels recorded in recent years. Another important public employment program in the region is Sri Lanka's Janasaviya Program designed to cushion the adverse- effects of the countrys structural adjustment program The program is generally labor-intensive and self-targeted, since the wage rates are maintained at levels below the market wage rates. The program contains two interesting features: (i) it combines the existing food stamp program by rendering the entitlement to food coupons conditional upon a household's supplying 24 days of labor every month at a rural public works, such as road construction; and (ii) it allows the participating household to receive a monthly cash payment (of Sri Lankan Rupees [SLR]. 2,500) for a period of two years, 58 percent of which is to be used for consumption and 42 percent to be saved for investment In other words, at the, end of two years, family members would have enough accumulated resources for investment in the income-generating small enterprises in which they had been receiving technical training during the two-year period. The program has enlisted seven million food stamp holders. Unlike other countries in the region, Sri Lanka has integrated this program with macroeconomic stabilization policies. The Philippines counts more than two decades of experience in labor-based public works. Until now, no careful evaluation of the program has been attempted. Evidence strongly suggests, however, that the program was never managed to ensure employment among the poorest groups. A major problem with the program has been the level of the wage rate. The total wage rate (food plus money wage) for the Food-for-Works Program remained higher than the market wage, destroying the scope for self-targeting. In Sub-Saharan Africa, public works were implemented both within the purview of Social Funds/Social Action Programs and outside the purview of these funds. The countries involved include Botswana, Cape Verde,.Ghana, Kenya, Lesotho, Malawi, Mozambique, Tanzania, and Zimbabwe. The goal of the programs is to achieve the dual objective of infrastructure building (rural roads in particular) and to generate employment. Priority in site selection was assigned to areas with high agricultural potential. Wage rates were linked to minimum wages in specific rural regions. In general, the poorest adults-both male and female-participated in the projects. In Botswana, for example, female participation is higher among the poor, especially in areas close to towns and big villages (Teklu, 1994). One reason behind this behavior is that female-headed households (especially young females with large number of dependents) are willing to accept lower wages than males. However, net employment gain is smaller than net income gain due to the shift from activities that pay even lower than public works. The programs suffered from weak synchronization between agricultural work and project work. In many countries, projects in the poorest neighborhoods are implicitly targeted so that the project's benefits (assets) can accrue disproportionately to the poor. 12 In Chile, Bolivia and Peru, one important safety net program generates massive employment for unskilled workers. Each country's experience with the program is varied, however. In Chile, wages were kept low, and the activities were decentralized and given low profile to keep them unattractive to the nonpoor; in Peru, poor supervision, overcentralization, and poor combination of inputs contributed to gross inefficiencies; and in Bolivia, the jobs were created at excessive cost Experience gained with respect to public works in different country environments suggests that the program, if properly designed, can simultaneously address four problems: food insecurity, lack of employment opportunities, environmental degradation, and poor infrastructure (van Braun et al, 1992). All these problems are more prevalent in Sub-Saharan Africa than in other parts of the world. When wages are set low and the created assets are maintained, public works can target the poor (and women among them) and generate both transfer benefits and stabilization benefits; the result often helps alleviate poverty and prevent famine. The emergence of such institutions as The Program of Action to Mitigate the Social Costs of Adjustments (PAMSCAD) in Ghana; the Directorate for Foods, Roads, and Rural Infrastructure in Nigeria; and Agence d'Execution des Travaux d'Inte'ret Public (AGETIP) in Senegal suggests that public response to cost-effective public works is now forthcoming. The learning costs for these newly emerging institutions can be kept low by carefully sifting and analyzing the experience in public works gained in Africa and Asia. Credit-based Livelihood Programs India is one of the earliest countries to experiment with a credit-based income-generating program known as the Integrated Rural Development Program (IRDP). Entirely financed by the government, the program is now the cornerstone of the govemrnent's antipoverty policy. Under the program, the poor are assisted with credit after a means test is conducted. The program has channeled an unprecedented amount of funding through loans and subsidies, enabling poor to obtain nonland assets. Though India pioneered studies on measurement of poverty, many costly concurrent evaluations of the schemes were done without baseline control groups. Consequently an objective assessment of the program has been lacking. Though most assessments show some benefits of the program, a major criticism has been that the scheme benefitted poor closer to the poverty line than the ultrapoor. The ultrapoor families, who generally lacked collateral, continued to have difficulties in obtaining working capital from the credit agencies even after the households promptly repaid their IRDP loans. This problem was successfully resolved by the Grameen Bank in Bangladesh through group lending. Outside the publicly supported safety nets, Bangladesh's Grameen Bank has been a unique experiment in alleviating poverty through an income-generating route (as opposed to "hand outs"). The Bank lends to groups of five to eight self-selected persons who have agreed to form a group, and who have guaranteed to monitor each other. Thus, credit is channeled to persons who had been considered "risky" by other commercial banks owing to absence of collateral. Grameen Bank's significant impact as a safety net for vulnerable households (women in particular) has been clearly demonstrated (Khandker, Khalily and Khan, 1994). Compared to nonparticipant control groups, Grameen Bank participant's incomes rose by more than 50 percent, food and nutrient intake per capita by 9 percent, and expenditures on health and education by 18 percent. Women's employment rose by 36 percent compared with 18 percent in villages without a branch of the Grameen Bank. 13 In the Philippines, several line ministries of the government have implemented credit-based livelihood programs tied to nongovernmental agencies (NGOs) that perform banking functions. A portion of the cost of operations of the NGOs is subsidized by the government. Evidence suggests that the programs benefitted the near-poor and the nonpoor more than the ultrapoor. The programs were implemented in regions already prosperous. The benefit-cost ratios of the major livelihood programs were unfavorable, and the incremental employment and income effects were negligible. Even the modest gains disappeared in less than two years after the projects were started; hence the record on sustainability was poor. The repayment rates were also unsatisfactory. A number of countries have programs for microenterprise development as one of the main strategies under Social Funds. Social Funds and Related Interventions In response to adjustment-induced hardships on the poor, the most notable forms of intervention have been social action programs (SAPs), social funds (SFs), and social investment funds (SIFs). For example, fifteenr out of twenty countries in the Latin American region have one of these interventions. The activities under these funds include infrastructure building and public works, child and maternal feeding and nutrition programs, support for setting up of microenterprises, support to small farmers, miners and fishermen, and immunization and other health care programs. The principal source of finance for the funds are the contributions from external donors (about 70 percent), with government support amounting to about 15 percent. SAPs are designed as regular investment projects; each component is fully appraised and all implementing agencies are identified before the project starts up. SFs are designed to fund local organizations. Both respond to funding requests from local agencies, but neither identify nor implement projects. Many programs devised under the SAPs/SFs are targeted to urban areas partly because of the perception that adjustment and economic degradation is felt more severely by the urban population. Targeting in these projects has primarily been in the form of geographical targeting, self- targeting, targeting through subprojects or components and through the use of implementing agencies, such as NGOs. Both programs have an important role to play in the short and medium run to provide quick support to the poor, as well as to create a close nexus between the central government and local organizations (public and private). Since most of these programs are new, however, their long-term impact on poverty cannot as yet be evaluated. The funds vary in degree of autonomy, ability to reach the poorest groups and regions, the extent of decentralization, and financing (especially the role of donor financing). Depending on these factors, the overall impact of the funds on the poor also varies across countries. Bolivia's Emergency Social Fund (ESF) has been much recognized for its demand-based approach, transparency and efficiency of operations largely outside the government bureaucracy. The ESF benefitted more than one million poor by its projects in infrastructure, including health posts and schools, school lunch programs, and job creation activities. By far the most lasting contribution of the ESF has been its efforts, strongly backed by the NGOs, to provide the poor with a means to help themselves. Yet, even in Bolivia, the poorest groups (particularly women) and poorest regions benefitted the least from ESF. 14 Drawing on the experience of Bolivia, El Salvador introduced The Fund for Social Investment (FIS) in 1989. The program targeted the vulnerable groups (e.g., pregnant women and children under age 5), and also vulnerable regions. FIS has a good record of working effectivey with the NGOs. Despite these efforts, FIS did experience difficulties in reaching the poorest groups, and poor municipalities received disproportionately low allocations. Nonetheless, FIS's achievements are impressive, judged against the massive need for reconstruction in the country. Under Senegal's SF, a nongovernmental organization, AGETIP, was set up to select and implement public works and services programs using labor-intensive techniques. The program aims to provide employment and subsistence to unskilled workers located in urban and periurban areas. The project has had low management costs and has involved many small-scale construction agencies. The success of AGETIP has prompted Chad, Mali, Tfhe Gambia, and other countries to learn from the experiences of AGETIP while formulating public works programs. In Ghana PAMSCAD was set up as part of governmental efforts to alleviate the hardship of vulnerable households. It benefitted from donor finance. PAMSCAD consisted of 23 antipoverty measures; the cost of implementing the measures was estimated at $84 million over two years. The program suffered from administrative bottlenecks and inability to coordinate numerous projects. The program failed to reach the poorest rural groups. For example, two-thirds of its funds were spent in urban areas, whereas the rural North, where the majority of the poor lived, received only three percent of the funds. In some countries, several programs were introduced under one umbrella to protect the poor from the short-term adverse effects of adjustment; health and nutrition interventions were central to the package. The Honduras Family Assistance Program currently benefits about 100,000 mothers and young children who represent about a quarter of the population at risk of malnutrition. In parallel, the food coupons program benefits 200,000 children. The impact of these programs has not been evaluated. Mexico's PRONASOL supported a variety of antipoverty programs in poor areas, including health and nutrition projects, employment projects, and projects that improved the quality of education in rural schools in poor areas. Beneficiary participation, a central feature of this program, definitely improved the quality of the program by providing productive employment opportunities for the poor. PRONASOL substantially expanded Mexico's safety net, which has been confined merely to food subsidies. Bangladesh's Vulnerable Group Development (VGD) programn provides supplementary feeding to 500,000 destitute women and children in distressed areas. The program operates on a cost- effective basis. To ensure that the receipt of food has an impact beyond the period food is received, VGD provides a complementary package of development services. These services include functional skills in literacy and numeracy, health and nutrition training, market-oriented income earning skills, group formation, access to credit and savings. Thus, VGD effectively offers "food" for "training." Unlike rationing schemes in Bangladesh which transfer 1 Taka of income to the poor at a cost of 6.6 Taka, the VGD progran transfers 1 Taka of income at a cost of only 1.4 taka. Reduced leakage, 15 combining training and beneficiary supervision, and monitoring contribute to the cost-effectiveness of VGD. The program is managed and administered in a way that makes the beneficiaries feel "empowered"-an aspect missing in most food subsidy programs (IFPRL 1994). However, the program is heavily driven by external finance; to that extent, its sustainability is in doubt, unless the government reduces its commitment to the expensive general food subsidy and diverts resources to the VGD program. To sum up, the demand-based nature of social funds is clearly a departure from past attempts to address short-term poverty. This tack, however, does not guarantee a fiscally viable safety net First, a social fund, however well designed, cannot bring much relief to the poor if the programs are manipulated or inefficiently implemented. Second, even if effectively implemented, the social fund cannot protect the poor fully unless social expenditures are redirected toward the poor. Third, poor targeting, especially the inability to reach women, continues to be a weakness in social funds and in many other safety net initiatives. Fourth, to the extent most of the Funds are externally funded, with very little government support, their sustainability is threatened, unless governments or NGOs fund the projects after extemal support ends. Child and Family Allowances Direct cash transfer in the form of child allowance is the most significant form of social assistance in countries of the FSU countries and in Eastern Europe. Before 1991, famnily allowances in the FSU were limited to a one-time payment upon the birth of a child (instituted during WWII). The pronatalist birth allowances had eroded in real terms because they were not indexed and because the poverty benefit was inadequate even in 1974. Legislation was enacted in 1991 with the intent of transferring back to the population the entire nominal amount of aggregated price increases; the legislation also aims to establish an automatic index system of wages, transfers, and savings deposits. A universal system of family allowances was adopted at the end of 1991 as the vehicle to compensate the FSU population for anticipated price increases.3 Rapid decline in output and fiscal contraction in most FSU countries have raised doubts whether the program can be sustained in its present form. One option often advocated is to refocus family and child allowances more closely on the groups determined most vulnerable by a means test. Yet this option has inherent problems, since large families are more likely to be in poverty than small families. Single mothers with large families are even more likely to be in poverty. Categorical targeting, such as geographic targeting or limiting family allowances to three or more children (except for a child or two children in single parent families), offers a possible approach Another option, which may help reduce the dangers of excluding some deserving households, is to restrict income (means) testing to families with one or two children. Country experience suggests that means tests may not work in FSU countries for political and practical reasons. First, a means test can be unpopular, especially since family allowances existed for 3Payment was on a per-child basis, with differentiation according to the age of the child (under 6 and 6-17 years old). There was no income test for this universal benefit 16 nearly 75 years. Second, the test can lead to perverse results if it is based only on reported money income. There may be difficulties in measuring incomes from livestock and agricultural activities, particularly when a large amount of food is in the form of in-kind consumption from private agricultural land. Moreover, expansion in privatefmformal sector activities in the FSU countries has attenuated comparisons of money income indicators. The gap between reported income and actual money income in countries undergoing transition tends to widen markedly due to the dramatic expansion of (monetized) informal sector activities. Third, information gaps render a means test administratively more complicated and expensive in FSU countries than in countries with reliable income/expenditure data. For these reasons, it seems appropriate to adopt some kind of categorical targeting with requirements such as those adopted in some Latin American countries. Moreover, there is a possibility that a means test might cause some deserving families to be excluded. In Romania, recent evidence suggests that a household headed by a female with a low level of education is unlikely to receive child allowances (Subbarao and Mehra, 1995). Lessons from Country Experience on the Design of Safety Nets To mitigate the short-term adverse effects of major economic reform measures on the poorer members of society, it is necessary to introduce safety nets. Countries can rely on various policy tools-a range of cash and in-kind transfer instruments is available, including family and child allowances, consumer subsidies, and public works. The choice and scope of instruments depend on particular country circumstances, including the mix of adjustment policies, the speed and sequencing of measures, the composition of target groups, the extent and depth of poverty, administrative capacity, the availability of financing and political economy factors. Experience from a variety of country situations reviewed in this paper suggests a few key lessons to keep in mind when designing safety net programs. First, it is important to integrate safety net programs into economic reform programs instead of introducing them as "add ons" after the reform is initiated. Country experience suggests that it is possible to do so, bearing in mind the key issues discussed earlier. Thus, in Sri Lanka, the program of economic restructuring is simultaneously accompanied by a safety net program targeted to the poor (see Box 1). Second, the instruments should build on, and complement, existing arrangements, both formal and informal. For example, when extended families help their members during crises, and community- based mechanisms allow members to pool risks, safety net instruments should not replace these existing informal arrangements. Third, where targeting is imperative and central or local governments are weak, local communities and nongovernmental agencies have to be relied on as intermediaries for channeling assistance. Fourth, certain safety nets should remain transitory measures; they should not become permanent. Fifth, it is best to finance safety nets from tax revenues collected from normal domestic sources. Under extreme conditions of fiscal contraction, as in the FSU countries, donor assistance may be needed in the interim. 17 Box 1: Integrating Safety Net Program with Economic Reform Program: The Case of Sri Lanka .:Sri Lanka is one of the few countrics with a long history of substantial public investmentr in.' health and education. In .1989, Janasaviya.Program (JSP) was introduced alongside structural:refoTrms Under the progran, the poor were provided fixed monthly cash grants for a period of two years, in the expectation that they would acquire some skills to become microentrepreneurs.: Sixty percent of the :monthly grant was actually released to the households, but the balance of 40 percent-.the Uviigis': component-vas made availabTe at'the end of the'two-year period.: With this lump sum, the househoids could start an enterprise after the end of the iwo-year period; The interesting feature, however, is that, in rcturn for the cash payments during the two years, beneficiary households must provide 20 days of labor per.month, which can:be spent either on training of. the beneficiaries themselves, or working on community infrastructure projects. The program is being introduced throughout the country (especially in rural areas) in successive, geogr-aphically-defined "rotnds," NNith the least affluent areas being covered first. A total of cleven rounds are planned; five arc already completed. Prior to JSP, the government had put in place.a food stamp program (FSP) to replace the earli&r, open-ended fobd subsidies. However, the FSP was poorly targeted, and benefits. leaked substantially to the. nonpoor. In order to limit coverage of FSP, households sclected for inclusion in the JSP are taken off the FSP rolls. Unlike in FSP, community participation methods. are now being used to. identify eligible households for JSP. Available evidence indicates that JSP succeeded in adopting tighter targeting than FSP. Moreover, thc employment schemes under JSP have generally been labor-intensive, involving wages: at rates below thc mark-et level so that the program attracted only the truly poor. Sri Lanka's experience demonstrates that (i) it is possible to integrate poverty reduction goals:. with the design of macroeconomic stabilization programs; (ii) it is possible to target better and reduce costs by community-based identification procedures; and (iii) thc poor can be protected during cconomic reforms without necessarily inicreasingpublic transfers as a perccntage of GDP. Experience with different types of safety nets yields the following lessons. Targeted subsidies are preferred to general subsidies in some countries, but the measures are more cost-effective where a certain degree of self-targeting is achieved through commodity selection (e.g., low-grade bread and barley in Jordan [see Box 2]). Where the data are more reliable and administrative capacity exists alongside a political willingness to target, distributing vouchers/food coupons redeemed through private traders, without distorting market prices may be another option. In general, in-kind transfers of food items can be cost-effective when they use existing channels and local communities (such as women's groups). Immediately following price liberalization, safety nets in FSU countries need to address restricting consumer subsidies, unemployment compensation, and training programs; they also need to introduce public works where regional pockets of unemployment are high (see Box 3). Providing cash compensation when a major subsidy is removed is another safety net option worth exploring, especially in countries where a reasonable degree of price stability was achieved. As policy reform gives rise to unemployment, retraining and severance pay may be helpful. Though such training may not immediately create jobs, it can alleviate skill shortages in growth areas with some indirect effect on unemployment. 18 Box 2: Jordan: Successful Revamping of Transfers to Cut Costs In Jordan, open-cnded food subsidies prior to the reform was the single largest item (3.2 percent ofl;DP) contributing to thc fiscal deficit. Following the rcform, generalized subsidics on brcad and barley-the food staples of the poor-were retained, but Lheir prices were increased lo cover part of the cost of transfcr. Clcarly, it was politically incxpedient to get rid of generalized subsidies an bread and barley, However, generalized subsidies on iice, sugar and milkwuere replaced by a system of quantity- specific food cotipons, giving duc weight to the size of thc family. Subsidy cost fcll dramatically from 3.2 percent of CDP to 1.4 percent of GDP. Jordan also introduced a new cash transfer-some 25,000 families idcnliflcd below the poverty linc received direct income support. Thus, Jordan presents an interesting example of cushioniing vulnerable groups from short-run adverse impact of adjustment by appropriately designed safety nct involving both a change in the existing in-kind bencfit, and introducing a ncw cash benerit. When reform entails major short-term unemployment especially in some regions (for example, following the closure of a mine in a region), social safety nets include labor-intensive public works to create job opportunities for the poor (as in Chile, Bolivia, India, and Mozambique). However, whether introduced as a general strategy to alleviate poverty or as a transitory safety net, public works are effective in targeting benefits to the poor only if low wages are offered to ensure self-targeting (see Box 4). When public works run efficiently, they can strengthen existing infrastructure, improving the supply response from the economy and providing visible evidence of reform benefits. In general, social assistance programs are best financed from the general revenues. However, excessive levels of funding for transfers will either (i) crowd out other developmental expenditures, given current levels of taxation, or (ii) require heavy taxation that can have well-known disincentive and distortionary effects. Therefore, spending levels have to be consistent with the government's overall target for fiscal balance. Public expenditure reviews should help governments choose among competing uses for its scarce fiscal resources. At the same time, an analysis of revenue-raising efforts of governments is also critical to get a balanced picture of the fiscal crisis. To keep spending levels down, appropriate designing of safety nets is crucial. Thus, transfer programs should in general be temporary; programs are best designed when self-targeted, and have built-in incentives for households to graduate out of the transfer as income levels improve. Public works programs at lower than market wage illustrate this design issue: workers who manage to get employment at the going market wage automatically exit out of the program. Social assistance channeled through community-based targeting can keep overall coverage and costs low. Governments have to make difficult political choices. Resource constraints have reinforced the prospects for a targeted strategy in many countries (including such countries as India, which rarely, if ever, adopted a targeted approach in the past) as the constituency for reform was built up. Unless the outreach of programs and benefits are held in check, and targeting is effective, adverse effects on incentives to work are inevitable. Interestingly, this lesson also comes from Sri Lanka's past experience with the food subsidies. Notwithstanding a move to food stamps in 1980, and even as late as 1991, the number of beneficiary households were twice as high as the households under the poverty line. It is estimated that the loose targeting of the food stamp program has reduced work effort (labor supply) of 19 Box 3: Options for Affordable Safety Nets in the FSU Countries . . a. The greatest challenge faced by the ESU countries is how to best reform and redesign the existing ormil system of social assistance to adequately protcct newly vulnerablc groups in an affordabl, manner, The existing formal system, consisting of generalized consumer subsidies on food, hiousing, utilities, and. urban mass transportation, universal family allowances, and cash transfers, is not alTordable, nor has it . kept the lower income groups from severe declines in their standard of living. Eliminating generalized consumer subsidies is politically difricult, but can be achieved.. when governmcnts arc able to adequalely inform and involve thc population in understanding the trade-offs. Generalized food subsidies can be replaced by quantity- specific food coupons for the poor, covering a limited range of staples, based on family: size. Another option is to offer a one-time cash bencrit targeted to those who fall below thc poverty linc (an incomes test) when subsidics are climinated. Tn the current environment, a mcans test might not be as cffcctivc and might be substituted or replaced by an assets test. Carc must be taken, though, to ensure that rcsourccs arc available to nmaintain the real value of such a cash benefit. Family allowances are weakly targeted forms of social assistance. The element of targeting is vcry important (especially considering the lack of targeting of other cash benefits) and should be preserved. . Measuring family allowances by means testing can be vcry complicated; if implemcntable, it might providc a vehicle for some cost savings, but given (he extent of possible complications, a categorical approach mighlt be more cffcctive. Options for categorical targeting include gcographic targeting and restricting allowances to three or more children, except in the case of single parent families or, optionally, those one- and two-child familics that arc below thc poverty linc (incomes test). * Reforming existing unemployment bencefits is something that many governments can not dclay until the mcdium termn. Unemployment bencfits are currcntly too low and coverage is -unnecessarily restricted, creating :enornious hardship for families -with unemployed members wvho have no access to other cash transfcrs. . * Public works programs are an important option for both rehabilitating the aging and inadequate infrastructure, and for targeting regions with severe unemploynieii(, due to: restrncturing. Carc must bc taken to see that such programs arc genuine job crcation activities, and not disguised subsidies to state-owned enterprises in financial disarray, Nor should public works-covcrage and wage fixation policies in parlicular-interfere with the normal functioning of labor markets. * Serious public expenditure reviews are necessary in all the FSU countries, as spending pattcrns continue to reflect the outmoded priorities of the old system inherited from the USSR. No system of social assistance can be adequately financed if governments do not recogni7t the trade-offs in public expenditures.. In particular, governments will have to consider booth pension reforn and Lhe reduction of nonessential spending in order, to provide social assistance. 20 Box 4: Public Works: When Are They Relevant? In widely differing countrv situations in Latin America, Africa, and South Asia, public vworks have been resorted to provide temporary incomes (or food) to vulnerable households. When are public works to be recommended? What are the ingredients of success? And what arc the impacts of public cmplovmcnt creation programs on Lhe labor market? In the early 1950s. public works were started in predominantly agrarian economies of South Asia to provide seasonal employment for about seven months a year wvhen there was no farming activity (due to the absence of a second crop). The intensity of public wvork-s operations sharplv increased during periods of droughlt when even the single crop, usually rice, failed due to monsoon failure. Public n-orks enabled the develcpment of irrigation-related infrastructure and rural rbads. Since monsoon failures and droughts werc typically a regional phenomenon, public works tended to be regionally concentrated. Thus, public works w,ere founid most feasible when unempluyment iuas seasonal and heavily concenirared regionanlv. In the wvake of adjustment-induced pressures, public works can be one of the many policies to protect vulnerable groups, provided the unemployment situation in the country is characterized by regional concentration (e.g., felloine the closure of a mine in the public sector) and/or seasonality. E\pcrience with public work-s suggests that the level at *%hich the wage is fixed could be critical in ensuring its success. If public works are to be used to supplement, not to substitute, the employment generated through normal economic activities, the program should be targeted only to those individuals wLvho are unablc to gain employmcnt in normal economic activities at the going market wage. Such self- selection can be ensured onlv if the wvaLe is maintained at a level lower than the market wage. Low w-ag,es attracted w,romen to the work sites, and also kept the overall size of the program limited and fiscally sustainable. In some countries (as in Zambia), pavment of woages in-kind rendered the program less attractive to men (who prefer cash wages) and enabled more womcn to join the program, improving household-level food securitv. In practice, however, maintaining low sages, and payment of wages in-kind, has not proven cas. hIore recently, governments are coming under increasing pressure to pay the minimum wage, uhich is typically higher than even the market wage. If this happens, not only would the self-targeting characteristic of the program disappear, but the program would also seriously interfere with the normal functioning of the labor marklet bv artificiallv raising the supply price of labor across the board. Paymcnt of wa!ges in-kind cncountered serious adnministrative problerns. W\hile public works at low wages on a limited scale in regional pockets of high unemployment can act as an important safety net for the vulnerable households, guaranteed employment in public works at high (minimum) wages can scriously hamper efficient functioning of the labor markets. participants by 3 to 5 days per month (Sahn and Alderman, 1992). By contrast, food stamps targeted to pregnant and lactating women and to children under five registered at the primary health centers in Jamaica were more successful in restricting the scope of the program to the most needy. The success of safety net programs depend critically on whether or not the programs are depoliticized, decentralized, and are strongly attached to the already strong local, community-based initiatives and organizations. While the demand-driven nature of social funds is clearly a departure from other programs, it does not guarantee a success. The social fund in Bolivia was very successful 21 because it enlisted NGO participation in reaching the poor. Beneficiary participation is the central feature behind the success of Mexico's PRONASOL-an umbrella program to deliver health, nutrition, education, housing and infrastructure to the poor. Though El Salvador's Fund for Social Investment was set up as recently as 1990, the Fund has set an impressive record of working with NGOs in targeting assistance to the poor. Avoiding economic analysis and political debate on options for reforming the social safety net, or introducing safety nets long after the reforms, virtually amnounts to countries relying on inflation to decide winners and losers. Such a policy is sure to hurt the poor in the short run and impair the reform in the medium run. In the final analysis, the poor benefit best when governments openly consider the costs and benefits of safety net reform, and introduce cost-effective safety nets especially where private transfers are the least. 22 References Alderman, Harold. 1991. "Food Subsidies and the Poor," in G. Psacharopoulas (ed.) Essays on Poverty, Equity and Growth. 172-202. . Pergamon Press. New York Alderman, Harold and J. von Braun. 1984. The Effects of the Egyptian Food Ration and Subsidy System on Income Distribution and Consumption. IFPRI. Research Report 45. Washington, D.C. Barr, N. 1993. The Economics of the Welfare State. Weidenfeld and Nicolson. Besley, Timothy. 1993. "The Principles of Targeting." in M. Lipton and J. van der Gaag (eds.) Including the Poor. The World Bank. Washington, D.C. Braithwaite, J. 1994. "The Old and New Poor in Russia: Trends in Poverty." Education and Social Policy Department. The World Bank. Washington, D.C. Carvalho, S. 1994. "Social Funds: Guidelines for Design and Implementation." HRO Working Paper 34. The World Bank. Washington, D.C. Chen, S., G. Datt and M. Ravallion. 1994. "Is Poverty Increasing in the Developing World?" The World Bank. Washington, D.C. Mimeo. Cornia, G. A. and S. Sipos (eds.). 1991. Children and the Transition to the Market Economy. UNICEF. England. Cox, D. and E. Jimenez. 1992a. "Social Security and Private Transfers in Developing Countries: The Case of Peru." The World Bank Economic Review. 6(1):155-69. Cox, D. and E. Jimenez. 1992b. Private Transfers and Public Income Redistribution in the Philippines. Paper presented at the Workshop on Public Expenditure and Poverty. The World Bank. Washington, D.C. Cox, D. and E. Jimenez. 1990. "Social Objectives through Private Transfers: A Review." The World Bank Research Observer. 5(2):205-18. Datt, G. 1994. "Maharashtra Employmnent Guarantee Scheme." Unpublished note for the Hunger Conference. The World Bank, Washington, D.C. Datt, G. and D. Gunewardena. 1994. "Some Aspects of Poverty in Sri Lanka: 1985-90." The World Bank. Washington, D.C. Mimeo (Draft). Gill, I., E. Jimenez and Z. Shalizi. 1990. "Targeting Consumer Subsidies for Poverty Alleviation: A Survey and a Primer of Basic Theory." Policy Research Department. The World Bank. Washington, D.C. 21 because it enlisted NGO participation in reaching the poor. Beneficiary participation is the central feature behind the success of Mexico's PRONASOL-an umbrella program to deliver health, nutrition, education, housing and infrastructure to the poor. Though El Salvador's Fund for Social Investment was set up as recently as 1990, the Fund has set an impressive record of working with NGOs in targeting assistance to the poor. Avoiding economic analysis and political debate on options for reforming the social safety net, or introducing safety nets long after the reforms, virtually amounts to countries relying on inflation to decide winners and losers. Such a policy is sure to hurt the poor in the short run and impair the reform in the medium run. In the final analysis, the poor benefit best when governments openly consider the costs and benefits of safety net reform, and introduce cost-effective safety nets especially where private transfers are the least. 22 References Alderman, Harold. 1991. "Food Subsidies and the Poor," in G. Psacharopoulas (ed.) Essays on Poverty, Equity and Growth. 172-202.. Pergamon Press. New York Alderman, Harold and J. von Braun. 1984. The Effects of the Egypiqian Food Ration and Subsidy System on Income Distribution and Consumption. IFPRI. Research Report 45. Washington, D.C. Barr, N. 1993. The Economics of the Welfare State. Weidenfeld and Nicolson. Besley, Timothy. 1993. "The Principles of Targeting." in M. Lipton and J. van der Gaag (eds.) Including the Poor. The World Bank. Washington, D.C. Braithwaite, J. 1994. "The Old and New Poor in Russia: Trends in Poverty." Education and Social Policy Department. The World Bank. Washington, D.C. Carvalho, S. 1994. "Social Funds: Guidelines for Design and Implementation." HRO Working Paper 34. The World Bank. Washington, D.C. Chen, S., G. Datt and M. Ravallion. 1994. "Is Poverty Increasing in the Developing World?" The World Bank. Washington, D.C. Mimeo. Comia, G. A. and S. Sipos (eds.). 1991. Children and the Transition to the Market Economy. UNICEF. England. Cox, D. and E. Jimenez. 1992a. "Social Security and Private Transfers in Developing Countries: The Case of Peru." The World Bank Economic Review. 6(1):155-69. Cox, D. and E. Jimenez. 1992b. Private Transfers and Public Income Redistribution in the Philippines. Paper presented at the Workshop on Public Expenditure and Poverty. The World Bank. Washington, D.C. Cox, D. and E. Jimenez. 1990. "Social Objectives through Private Transfers: A Review." The World Bank Research Observer. 5(2):205-18. Datt, G. 1994. "Maharashtra Employment Guarantee Scheme." Unpublished note for the Hunger Conference. The World Bank, Washington, D.C. Datt, G. and D. Gunewardena. 1994. "Some Aspects of Poverty in Sri Lanka: 1985-90." The World Bank. Washington, D.C. Mimeo (Drafl). Gill, I., E. Jimenez and Z. Shalizi. 1990. "Targeting Consumer Subsidies for Poverty Alleviation: A Survey and a Primer of Basic Theory." Policy Research Department. The World Bank. Washington, D.C. 23 Government of Bangladesh. 1993. "Report of the Working Group on Targeted Food Interventions." Graham, C. 1994. Safety Nets, Politics and the Poor. Brookings Institution. Forthcomning. Grosh, M. 1994. Administering Targeted Social Programs in Latin America: From Platitudes to Practice. The World Bank. Washington, D.C. Intemational Food Policy Research Institute. 1994. Options for Targeting Food Interventions in Bangladesh. Washington D.C. Jimenez, E. 1993. "Cash versus In-kind Transfers." Unpublished lecture notes. The World Bank. Washington, D.C. Kakwani, N. and K. Subbarao. 1993. "Rural Poverty in India," in M. Lipton and J. van der Gaag (eds.). Including the Poor. The World Bank. Washington, D.C. Khandker, S. B. Khalily and Z. Khan. 1994. Sustainability of Grameen Bank: What Do We Know? Education and Social Policy Department, World Bank. Krumm, K., B. Milanovic and M. Walton. 1994. "Transfers and the Transition from Socialism: Is a Radical Altemative Necessary?" The World Bank. Washington, D.C. Mimeo. Kwon, S. 1994. "The Role of Public and Private Transfers to Vulnerable Groups with Special Reference to Experiences of the Eastem Asian Countries." The World Bank. Washington, D.C. Mimeo. Lipton, M. and J. van der Gaag (ed.). 1993. Including the Poor. The World Bank. Washington, D.C. Marc, A., C. Graham and M. Schacter. 1993. "Social Action Programs and Social Funds: A Review of Design and Implementation in Sub-Saharan Africa." AFTHR Technical Note 9. The World Bank. Washington, D.C. Milanovic, B. 1993. "Social Costs of Transition to Capitalism: Poland 1990-91." The World Bank. Washington, D.C. Newbery, David M. and N. Stern. 1987. The Theory of Taxationfor Developing Countries. Oxford University Press. New York. Rao, C.H.H., K. Subbarao and S.K. Ray. 1988. Unstable Agriculture and Droughts: Implications for Policy. Vikas: Delhi. 24 Ravallion, Martin, G. Datt and S. Chaudhuri. 1993. Higher Wages for Relief Work Can Make Many of the Poor Worse Off Recent Evidence from Maharashtra's "Employment Guarantee Scheme". Working Paper Series No. 568. World Bank. Sahn, David and H. Alderman. 1992. The Effect of Food Subsidies on Labor Supply in Sri Lanka. World Bank. Processed. Subbarao, Kalanidhi. 1992. "Interventions to Fill Nutrition Gaps in India," in B. Harriss, S. Guhan and R H. Cassey (eds.). Poverty in India: Research and Policy. Oxford University Press. Bombay. Subbarao, Kalanidhi and K. Mehra. 1995. SocialAssistance and the Poor in Romania World Bank. Washington, D.C. Subbarao, Kalanidhi and K. Ezemenari. 1995. Transition, Poverty and Social Assistance in Mongolia. ESP Discussion Paper 55. World Bank. Teklu, T. 1994. "Labor-Intensive Rural Roads in Kenya, Tanzania and Botswana: Some Evidence on Design and Practice." International Food Policy Research Institute. Washington D.C. Mimeo. UNCTAD. 1994. "The Effects on Poverty Alleviation of Structural Adjustment Programs." Geneva. Mlmeo. van de Walle, D., M. Ravallion and M. Gautam. 1994. "How Well Does The Social Safety Net Work?" The World Bank. Washington, D.C. von Braun, J., T. Teklu and P. Webb. 1992. "Labor-intensive Public Works for Food Security in Africa: Past Experience and Future Potential," International Labor Review, vol. 131 (1). WorldBank. 1994a. WorldDevelopmentReport. WorldBank:Washington,D.C. World Bank. 1994b. Averting the Old Age Crisis: Policy Options for the Graying World. Washington, D.C. World Bank. 1994c. Poverty in Poland. Washington, D.C. World Bank. 1990. World DevelopmentReport Washington, D.C. Human Capital Development and Operations Policy Working Paper Series Contact for Title Author Date paper HROWP24 Concepts of Educational Marlaine E. Lockheed March 1994 M. Verbeeck Efficiency and Effectiveness Eric Hanushek 34821 HROWP25 Scientific Research for Erik W. Thulstrup March 1994 L. Malca Development 37720 HROWP26 Issues in Education Finance Stephen P. Heyneman April 1994 B. Cassorla and Management in ECA and 37172 OECD Countries HROWP27 Vocational Education and Julian Schweitzer April 1994 A. Gonzalez Training: The Role of the 37799 Public Sector in a Market Economy HROWP28 Social Security Issues and Nguyen X. Nguyen May 1994 M. Espinosa Elements of Reform 37599 HROWP29 Health Problems and Policies Mary Eming Young May 1994 0. Shoffner for Older Women: An 37023 Emerging Issue in Developing Countries HROWP30 Language and Education in S.M. Cummings May 1994 M. Espinosa Latin America: An Overview Stella Tamayo 37599 HROWP31 Does Participation Cost the Jesko Hentschel June 1994 D. Jenkins World Bank More? Emerging 37890 Evidence HROWP32 Research as an Input into Harold Alderman June 1994 P. Cook Nutrition Policy Formation 33902 HROWP33 The Role of the Public and Deepak Lal June 1994 M. Espinosa Private Sectors in Health 37599 Financing HROWP34 Social Funds: Guidelines for Soniya Carvalho July 1994 K. Labrie Design and Implementation 31001 HROWP35 Pharmaceutical Policies: Graham Dukes July 1994 0. Shoffner Rationale and Design Denis Broun 37023 HROWP36 Poverty, Human Development Harsha Aturupane August 1994 P. Cook and Growth: An Emerging Paul Glewwe 30864 Consensus? Paul Isenman HROWP37 Getting the Most out of Helen Saxenian September 1994 0. Shoffner Pharmaceutical Expenditures 37023 HROWP38 Procurement of Denis Broun September 1994 0. Shoffner Pharmaceuticals in World 37023 Bank Projects HROWP39 Notes on Education and Harry Anthony Patrinos September 1994 I. Conachy Economic Growth: Theory and 33669 Evidence HROWP40 Integrated Early Child Mary Eming Young October 1994 0. Shoffner Development: Challenges and 37023 Opportunities Human Capital Development and Operations Policy Working Paper Series Contact for Title Author Date paper HROWP41 Labor Market Insurance and Deepak Lal October 1994 M. Espinosa Social Safety Nets 37599 HROWP42 Institutional Development in Alberto de Capitani October 1994 S. Howard Third World Countries: The Douglass C. North 30877 Role of the World Bank HROWP43 Public and Private Secondary Marlaine E. Lockheed November 1994 M. Verbeeck Schools in Developing Emmanuel Jimenez 34821 Countries HROWP44 Integrated Approaches to T. Paul Schultz November 1994 M. Espinosa Human Resource Development 37599 HROWP45 The Costs of Discrimination in Harry Anthony Patrinos November 1994 I. Conachy Latin America 33669 HROWP46 Physician Behavioral Nguyen X. Nguyen December 1994 M. Espinosa Response to Price Control 37599 HROWP47 Evaluation of Integrated T. Paul Schultz January 1995 M. Espinosa Human Resource Programs 37599 HROWP48 Cost-Effectiveness and Health Philip Musgrove January 1995 0. Shoffner Sector Reform 37023 HROWP49 Egypt: Recent Changes in Susan H. Cochrane February 1995 0. Shoffner Population Growth Ernest E. Massiah 37023 HROWP50 Literacy and Primary Kowsar P. Chowdhury February 1995 M. Espinosa Education 37599 HROWP51 Incentives and Provider Howard Barnum March 1995 0. Shoffner Payment Methods Joseph Kutzin 37023 Helen Saxenian HROWP52 Human Capital and Poverty Gary S. Becker March 1995 M. Espinosa Alleviation 37599 HROWP53 Technology, Development, and Carl Dahlman April 1995 M. Espinosa the Role of the World Bank 37599 HROWP54 International Migration: Sharon Stanton Russell May 1995 0. Shoffner Implications for the World 37023 Bank HROWP55 Swimming Against the Tide: Nancy Birdsall May 1995 A. Colbert Strategies for Improving Equity Robert Hecht 34479 in Health HROWP56 Child Labor: Issues, Causes Faraaz Siddiqi June 1995 I Conachy and Interventions Harry Anthony Patrinos 33669 HCOWP57 The World Bank's Relationship Roberto Gonzales July 1995 K. Schrader with South Africa: A Cofino 82736 Successful Approach to Partcipation HCOWP58 Protecting The Poor During K. Subbarao July 1995 K. Labrie Adjustment and Transitions Jeanine Braithwaite 31001 Jyotsna Jalan