2 Social Safety Nets in Economic Reform
- Ke-young Chu, and Sanjeev Gupta
- Published Date:
- April 1998
Social safety nets can be broadly defined as those instruments aimed at mitigating possible adverse effects of reform measures on the poor. These instruments include temporary arrangements, as well as existing social protection measures reformed and adapted for this purpose, such as limited food subsidies, social security arrangements for dealing with various life cycle and other contingencies (e.g., old age, disability, unemployment, sickness, and drought), and targeted public works.1
The chapter discusses major issues concerning social safety nets and reviews the IMF’s recent experience in this regard.2 It provides a rationale for social safety nets as an element of economic reform programs and assesses the main options in designing the social safety nets. It then summarizes the IMF’s efforts to integrate social safety nets in programs it has supported and the role of technical assistance in this regard.
Economic Reform and Social Safety Nets
Role of Social Safety Nets in Economic Reform
Economic reform measures are essential in establishing conditions for long-term sustained growth and poverty reduction and may benefit some of the poor directly. They are also essential in protecting the poor, who often suffer the most from open or repressed inflation during disorderly adjustment. Some of these measures, however, may precipitate an excessive decline in the living standards of certain poor or vulnerable groups of the population in the short run, although they may benefit some other poor groups. For example, although increasing prices toward remunerable levels benefits producers (e.g., small farmers), thereby immediately reducing poverty, net purchasers (e.g., of food) may suffer real income losses. If official consumer prices are not adjusted, however, raising producer prices will increase subsidy expenditures. Similarly, the reforms of inefficient state enterprises may generate unemployment, thus increasing poverty.
Reform programs should, where possible, be designed to minimize unnecessary adverse effects on poor or vulnerable groups. To this end, it is important that the mix and sequencing of reform measures be designed appropriately. Simple, broad-based, and equitable taxes with low rates, efficient public expenditure programs, and policies aimed at promoting appropriate relative prices constitute the major elements. In achieving a given fiscal deficit reduction target, it is important to explore alternative tax and expenditure policy mixes and to assess their implications for poverty as well as efficiency. The appropriate mix and sequencing of reform measures, however, cannot ensure that the adverse effects will be removed entirely, since the reform itself requires changes in relative income positions of various groups. Therefore, social safety nets are needed to mitigate possible short-run adverse effects on the poor.
Options for Cost-Effective Social Safety Nets During Reform
In designing social safety nets for a country, one has to consider a number of factor, including the composition of poor group, potential effects of reform policy measures, financial and administrative constraints, and existing social protection arrangements. Different circumstances will require different mixes of instruments. Many reforming countries have various private and public social security arrangements. Many of them have extended families and community-based arrangements. Some have ad hoc or permanent arrangements (such as targeted small-scale self-employment enterprises and public works programs) set up in cooperation with donor organizations. Most reforming countries have costly budgetary and implicit subsidies, which are inefficient and badly targeted. In addition, some countries, particularly middle-income developing and transition countries, have formal social security arrangements, including social insurance (such as public pension plans and unemployment benefits), family allowances, and assistance instruments. Whereas coverage of these arrangements in middle-income developing countries is limited to some formal sector workers, most economies in transition have close to universal coverage.
As they are, these arrangements may not be appropriate as social safety nets. Despite limitations, however, they may be reformed, or adapted, to provide a basis for cost-effective social safety nets. Newly designed social safety nets should complement, rather than replace, existing cost-effective institutions.3
The challenge is to design these instruments in a targeted and cost-effective fashion. Targeting, however, is difficult for two reasons: (1) defining and identifying the poor is difficult, and (2) the “middle-class capture” of the existing benefits that need to be reformed makes targeting a politically sensitive issue.4 Moreover, the requirement that the target groups include all the poor, not merely exclude the nonpoor, creates further administrative difficulties.
In principle, the poor could be defined as those whose consumption falls below a certain poverty line, based on a minimum consumption basket, and benefits could be provided as means-tested transfers, based on an income and asset threshold for the family or household unit. With a low poverty line, the financial resources required for the provision of benefits would be limited. In practice, however, defining the poor on this basis for the purpose of social safety nets would have demanding requirements in terms of criteria that remain controversial and detailed data on household incomes and expenditures.5 In many reforming countries, a marked concentration of per capita income is often noted at relatively low-income thresholds. If a slight change in the minimum threshold led to a substantial change in the share of the officially designated poor, it would be controversial and difficult to determine a minimum threshold, and, in many cases, the administrative costs of verification for means testing would be substantial.6 Moreover, marginal tax rates would be high at the point of withdrawal of a means-tested benefit based on an income and asset threshold for the family or household unit, and would generate major disincentives at low income levels (i.e., the “poverty trap”).
It is possible, however, to take a pragmatic approach and restrict benefits to those who are vulnerable to policy shocks, through regional targeting (e.g., to urban areas) or categorical targeting (e.g., to infants, schoolchildren, nursing mothers, and the elderly), or a mix of the two.7 The extent of protection provided on a categorical basis should be determined with reference to the actual consumption of the poorest deciles. The target groups should be determined on the basis of country-specific circumstances, including demographic characteristics, administrative capabilities, and the effects of reform policy measures.
There is a direct link between adequate and administratively feasible provisions for the vulnerable and the establishment of a simple tax system. If a system of social security or other measures are provided effectively for the vulnerable, then there would be less need to introduce complicated and differentiated tax structures for distributional objectives. And, as with the tax system, there would be a need to establish administratively simple expenditure mechanisms—without excessive differentiation or complexity in design. The net effects on equity would thus involve a joint consideration of tax and social protection measures.8
Limited Transitory Consumer Subsidies
The issues in designing limited subsidies during the period of reform include the choice of commodities and delivery instruments. Reformed subsidies could lead to substantial budgetary savings over general subsidies, even if the beneficiaries included the entire population, while providing the government with an opportunity to liberalize prices. If quantities subsidized were restricted to quantities consumed by the poorest income groups, which are often a fraction of the average consumption, it would protect the poor from the effects of price adjustments. Since household groups would receive the same benefits, this would be more progressive in incidence than a generalized subsidy, which generally provides greater benefits to the rich.
Choice of commodities. Ideally, commodities to be subsidized should be consumed only by the poor. In general, however, it would be difficult to identify items consumed only by the poor.9 Coarse grains and unrefined sugar are attractive in this respect, but many inferior commodities are not widely marketed.10 The alternative would be to distinguish marketable commodities that are important in the consumption baskets of all the poor, even if these were also consumed by the nonpoor; examples include the major staples such as wheat, maize, and rice.
Delivery instruments. To avoid a general subsidy, various instruments could be used. In the case of adjustments where there is little impact on consumer prices of items consumed by the poor, there would be scant justification for the introduction of the instruments discussed below.11 All of the instruments would require the maintenance of effective prices that deviate by varying degrees from market-clearing prices, and require special administrative arrangements (e.g., coupons and monitoring procedures); therefore, the potential for abuse (e.g., counterfeiting of coupons) and corruption would increase. These costs should be contained as much as possible but should be assessed against the potential benefits of the instruments. In many cases, a realistic comparison of budgetary costs of subsidies for the transition might not be between limited transitory subsidies and no subsidies, but between the former and existing generalized subsidies.
The various instruments discussed below resemble, but are not necessarily identical to, in-kind or cash provision. The choice between the two contrasting forms of transfer—limited cash provision compensating the removal of subsidies and in-kind provision—would depend on the nature of the risk facing consumers and the administrative capabilities of the country in question. Provided that adverse effects on production were avoided by assuring producers remunerative prices for their output, the welfare effect of cash and in-kind provision would be equivalent if prices were known with certainty and resale were permissible and costless. Cash compensation generally provides a greater choice to the consumer, hence the possibility of increasing welfare gains, and also might mean lower administrative and transaction costs than the alternatives. If the price adjustments were uncertain, then cash provision would lead to higher risk for the consumer than in-kind provision. The ability of the government to import subsidized goods would place an upper limit on budgetary outlays, and could also influence the variation in market prices faced by consumers. Thus, the risks facing the government would likely be much lower than those confronting an individual in an imperfect and unstable local market. Moreover, in periods of rapid and large price changes, the administrative costs of monitoring the price changes and ensuring the adequate payments to consumers could be substantial.
Rations with public distribution systems, where dual prices exist at the retail level, could pose administrative difficulties.12 This could encourage arbitrage and queues, and might not provide an adequate safety net for the vulnerable, as was the case in the Baltic States, Russia, and other former Soviet Union countries (BRO). Moreover, setting up a public distribution system from scratch might be expensive, time consuming, and inappropriate as a transitory mechanism that should be established rapidly. This system, however, could be effective in providing milk and preventive health care through schools or health clinics. An alternative, a system of quantity food stamps, using private distribution channels, could be categorically or regionally targeted and might be attractive on both administrative and distributional grounds. Beneficiaries could purchase these from local outlets (e.g., banks and post offices) at subsidized prices. The stamps could then be exchanged for given quantities of the specified items. The trader could be reimbursed in-kind (plus a handling charge), or in cash by the state. This arrangement would permit a single price to operate in the market, without the need for an expensive administrative machinery to oversee compliance. Resale of the quantity food stamps should be legal as there would be less danger of windfalls accruing to traders, or of corruption, compared with standard ration systems.13 This system would combine the administrative advantages with more adequate consumer protection than feasible with a pure cash transfer.
It would be possible to provide limited cash compensation directly or through coupons, the latter for purchasing a specified set of commodities. Such provision could be means-tested, categorically targeted, or universal, where extensive private retail outlets would be present to satisfy demand. The provision of cash-based food stamps has involved an element of means testing in some countries. For example, initially introduced in the United States because there was political support for in-kind transfers rather than cash transfers, such a food stamp scheme has evolved into a major income transfer program. Despite the administrative capabilities of the United States, the stigma attached to income testing has restricted take-up rates to around 40 percent of the eligible population. Sri Lanka and Jamaica are the only two developing countries that have tried to establish such food stamps on a nationwide scale. In Sri Lanka, lack of political support led to the erosion of benefits, and means testing was not very effective, although the community-based screening programs improved targeting. In Jamaica, half of the aid-financed food stamps were allocated on a categorical basis to mothers and children, the rest to the needy on the basis of a self-declared certification. While in Sri Lanka the nonpoor did not hesitate to declare incomes below the cut-off point, in Jamaica, the procedure requiring certification led to stigma, largely excluding the nonpoor, but excluding also many of the poor.
Social Security Arrangements
Existing permanent social security arrangements, as in the BRO countries, can be a useful basis for designing social safety nets.14 A major strand among social security systems is social insurance. This entails the financing of benefits by compulsory contributions or pay-roll taxes. It differs from private insurance in that contributions need not fully cover benefits for either certain groups15 or the total population covered. Including the poor and bad-risk individuals implies an element of redistribution inherent in social insurance within each generation as well as over succeeding generations.
In addition, there are targeted expenditures, which may be either means-tested, as with many local social assistance programs, or related to particular indicators of need (such as disability, female gender, and old or young age), which also meet the criterion of protecting the vulnerable.16
The social security systems of most countries combine the abovedescribed measures in various ways. Interactions between measures help determine whether the basic objective of “protecting” the vulnerable has been achieved. Focusing on one program, such as the public provision of retirement pensions, is therefore likely to be misleading, The fiscal evaluation of groups of related schemes and their alternatives is difficult but important, given the precarious budgetary positions of many countries and the escalating nature and extent of benefits and costs of provision.17
Public policy to ensure social security should take into account what individuals and households do to “protect” themselves. This involves an examination of (1) savings behavior and possibilities; (2) restrictions on the ability to smooth consumption over time, with limited access to credit markets; and (3) transfers and support mechanisms at the extended family or community levels. The scope for households or local communities to smooth consumption through accumulation and asset liquidation is limited, however, particularly in the case of major and repeated shocks.
Two issues that should be addressed in designing each instrument, depending on the administrative and demographic characteristics of each country, concern eligibility and structure of benefits.
Design options. An increase in the coverage of “formal” instruments in developing countries would take time. In transition economies, universal coverage might be reduced as private enterprises, particularly those in the informal sector, expand their activities.
Eligibility and the level and duration of benefits would determine the costs and adequacy of provision. Eligibility would need to be tightly defined. Whether benefits are flat or earnings-related would affect both administration and equity. In many countries, the dual objectives of the financial sustainability of social security funds and protection of minimum benefits imply that earnings-related benefits might have to be determined within a fairly narrow band, and thus resemble a flat benefit. Flat pension benefits would reduce the administrative complexity and costs of provisions, but might not be actuarially fair. Severely limited unemployment benefits might create a political bottleneck for enterprise reform or public sector retrenchment. A long duration of payments would necessitate high payroll contributions or large budgetary transfers, and would provide a disincentive to job search.
In many BRO countries, numerous and overlapping allowances and benefits, also largely financed by payroll contributions, have the effect of obscuring the net benefits conferred while adding to the complexity of administration and cost.
Policy tools. Overlapping policy tools would include a range of policies that could be grouped under (1) social insurance to provide cover for old age, disability, and health care; (2) minimum or basic benefits, categorically targeted, say, to the needy elderly through social pensions or to mothers with small children through family allowances; (3) unemployment benefits through severance pay and unemployment insurance; and (4) means-tested family support, which—in view of information costs and incentive effects—might well be feasible only at the local level with some community participation.
There would be a trade-off between average benefit levels and the overall costs, which would mean either higher payroll contribution rates or additional calls on budgetary resources. In a period of declining employment, real wages, and payroll contributions, it might not be feasible to index benefits fully to prices. Nonetheless, pensioners at subsistence levels would need to be assured access to minimum consumption requirements. In Eastern Europe, China, and BRO countries, the retirement age is a key factor in determining the cost of providing pensions for the elderly. There has already been a decline in the replacement rate (i.e., the value of average old-age pensions as a proportion of average wages) and the standard of living of pensioners in many countries.18 At the same time, these pension expenditures require relatively high contribution rates, in part, as a result of relatively low statutory retirement ages and generous eligibility criteria. Major reforms are needed to ensure the viability of pension funds simultaneously with effective provision for the elderly.
In both developing countries and transition economies, public sector restructuring may not be feasible unless provision is made for the unemployed. En the transition period, when the restructuring is likely to lead to widespread unemployment (10 percent unemployment or more is not uncommon, as seen in Eastern Europe), a normally defined unemployment insurance scheme (together with severance pay) might not be appropriate. If payroll contributions were constrained, fairly tight eligibility criteria and limits on both the level and duration of benefits would be needed. Despite these measures, there might be substantial start-up costs, which would require support from the budget.
This would leave the problem of assisting fresh entrants to the workforce and the longer-term unemployed whose insurance benefits have expired. Such persons could be provided with training or minimum benefits in a manner that would not create a disincentive to job search. Appropriately designed public works at low wages (see below) might provide a cost-effective screening or assistance mechanism.
Family allowances, in general, are well targeted to families with children.19 Such households tend to have low incomes relative to need and young children at an early stage of life cycle earnings; the arrival of additional children restricts opportunities for second-earner income. Provisions made directly to mothers would address the issue of intra-household distribution of incomes. Of course, universal benefits would also accrue to the well-to-do, and benefit levels should be related to available resources. However, it is unlikely that the level of family allowances could be expanded in line with inflation in the face of falling real revenue.20 A low level of family allowances, however, might increase the need for social assistance for certain groups, such as the unemployed with children.
Targeted Public Works
Targeted public works would represent a screening mechanism that provides assistance to the truly needy. The dual aspects of low wages and the tying-up of the workday are crucial in effective targeting. An assistance scheme would be subject to open-ended abuse if recipients were permitted to draw cash support while working in the informal sector. Also, the wage rate should be low to encourage the longer-term unemployed to participate actively in the labor market, or opt for retraining, and to make the public works more self-targeted, thus reducing administrative costs.21 High-wage public works would create opportunities for rent-seeking, which might be difficult to remove.
It is important to distinguish between public works as a safety net and public works designed for particular infrastructure needs. For the latter, skilled work at the appropriate market wage would be required; for the former, the creation of positive output should be considered an added bonus, although it should be sought where possible. For example, targeted public works programs, by improving feeder roads and other infrastructure, could reinforce the positive effects of pricing reform on supply. Safeguards would be needed to avoid transfers to phantom workers, discrimination in hiring, and other problems. For the most part, “safety net” public works could concentrate on low-skill employment. It would be advisable to prepare a list of “productive” projects that could be used for the safety net type of public works before a sharp increase in unemployment occurs.
Financing Social Safety Nets
The prospects for major increases in tax revenues in developing and transition economies are fairly limited in the short run. These prospects underscore the importance of improving the efficiency of public expenditure programs and restructuring them as a means of financing social safety net expenditures. In some countries, reforming generalized subsidies would generate enough resources to finance both targeted subsidies and some other productive government expenditures, or an improvement in the fiscal position.22 However, in some low-income countries, given the magnitude of the need, the gap between average consumption and the consumption of the poorest groups might be quite small, limiting the potential savings under a quantity food stamp scheme and requiring foreign assistance to cushion major policy or other shocks.
The cost of providing even minimum benefits associated with the formal social security instruments (pensions, allowances, and unemployment insurance) could be fairly substantial in a range of countries. As mentioned above, payroll taxes are a major source of financing for such expenditures in many countries. However, excessively high payroll taxes could discourage enterprise activities and introduce labor market distortions (e.g., the use of informal sector or contract workers for whom no contribution is paid).23 The use of general revenue to finance all or part of such expenditures is an issue faced in transition economies, particularly with regard to expenditures on account of unemployment. In all such cases, budgetary trade-offs among expenditure choices are quite stark, leading to a need to minimize social security outlays.
Social Safety Nets in the IMF’s Policy Advice
The IMF has advised the authorities, at varying levels of detail, on the integration of social safety nets in designing reform programs. Programs have paid increasing attention to social policies, in general, and to measures to mitigate the short-run adverse effects of economic reform measures on the poor, in particular. Many programs contain specific social safety net measures, which are sometimes based on detailed assessments that resulted from prior IMF technical assistance.24
The reform of consumer subsidies has figured prominently in both Stand-By Arrangements and Structural Adjustment Facility (SAF) and Enhanced Structural Adjustment (ESAF) programs in a number of developing countries (e.g., Algeria, Ethiopia, Ghana, Guyana, Jordan, Tanzania, and Uganda), as well as for Eastern European countries. In other countries, the restructuring of the public sector has been important in discussions with country authorities. Provisions for civil service reform have been featured in the programs for countries such as Barbados, Benin, The Gambia, and Sri Lanka. The restructuring of public sector enterprises has been important in many former centrally planned economies. Some IMF-supported programs have explicitly included measures for those who might become unemployed, and similar measures would also be needed in the formerly centrally planned economies in Africa, such as Angola, Ethiopia, and Mozambique. Such measures may also be required for other countries, such as India.
In SAF and ESAF countries, financial and administrative constraints tend to be more severe than in middle-income developing or transition countries. The SAF and ESAF programs have tended to adopt a broad approach to poverty and the social impact of adjustment, emphasizing—in policy framework papers (PFPs)—the need to increase expenditure on items such as education and health, and to include other specific social sector policy measures (e.g., policies addressing population, labor market, and women’s issues) as well as those to improve food security. Many PFPs have attempted to assess the social impact of adjustment and specify social safety net measures—often designed by donor agencies with the IMF’s cooperation.25 The donor-supported social investment funds that provide quick-disbursing soft loans, inter alia, to small-scale businesses may not actually reach either the poorest or those primarily affected by adjustment. Without an emphasis on cost-effectiveness, recent social funds (e.g., those in Egypt and a number of countries in Latin America) could weaken budgetary discipline.
Social safety nets are mentioned prominently in IMF staff reports on the former centrally planned economies, in connection with the need to control costs and to improve the effectiveness of formal social security measures. For example, in Hungary—which has a high level of information technology and administrative sophistication—the IMF has stressed the need to make the social security fund actuarially sound and to target assistance. Despite reforms to the pension and unemployment insurance systems, however, some problems remain. The principal problem concerns the implementation of social assistance mechanisms and means testing. The targeting of the provision of benefits for the long-term unemployed remains an issue in a number of Eastern European countries, and indexation procedures threaten financial stability in several cases. The IMF is increasingly cognizant of the need to incorporate reforms of institutions in programs, as indicated by the examples of Poland and other Eastern European economies.
The choice of social safety net instruments has varied with the circumstances of the country and the effects of the program. The Jordanian experience points to the importance of well-designed, limited, consumer subsidies as a key element of a social safety net in the face of large changes in relative prices. The Bangladeshi experience stresses the importance of foreign-assisted public works in circumstances of widespread poverty and no large changes in relative prices. The Polish experience suggests the importance of the cost-effective design of unemployment compensation in addressing the substantial unemployment in reforming transition economies.
Major issues addressed by the IMF’s technical assistance have ranged from reforming consumer subsidies to adapting permanent social security institutions to function as a social safety net. Most of the technical assistance work has highlighted the interactions between measures to mitigate the adverse effects of adjustment and formal social security instruments.
Reforming Consumer Subsidies
The IMF’s advice with respect to subsidies has been more or less unanimous: eliminate generalized subsidies as soon as possible to allow a reduction in production distortions without a major cost to the budget. Recommendations have depended on administrative capability as well as the extent and type of price adjustment envisaged. Some technical assistance reports have recommended the consideration of limited quantity food stamps, which may be targeted by region (e.g., urban households) or category (e.g., the unemployed, households with children), with additional purchases made at market prices. This would permit a rapid freeing of prices without adverse impact on the poor. A frequently proposed alternative has been to provide the cash equivalent transfer of the quantity food stamps following the removal of subsidies, provided that adjustments could be made to payments in time, and that there are no difficulties in effecting the payments. Some reports have supported the authorities’ intentions to introduce means-tested cash compensation while stressing caveats with respect to administrative capability and knowledge about incomes and cutoff points.
Many reports have recommended categorical transfers in-kind (e.g., milk) to young children and nursing mothers. Such provisions would tend to be better targeted than cash-based family allowances and would be more appropriate in countries with a relatively high birth rate because of lower budgetary outlays than with family allowances.26
Improving Social Security Arrangements as a Social Safety Net
Recommendations with respect to reform of the formal social security instruments have primarily concerned the adequacy of overall coverage and the financial implications of alternative measures. For pensions in BRO countries, recommendations have been that pension funds concentrate only on the financing and administration of pensions, and not allowances, to improve their overall financial viability. Most reports have commented on the need for pension funds to raise retirement ages and to limit early retirement provisions. Technical assistance reports, however, have been less unanimous with respect to indexation of benefits: some have argued for full indexing of minimum pensions with respect to price change and less than full indexation of the rest, to assure the stability of funds with a declining real revenue base. The difficulty arises when most pensioners are at minimum subsistence levels. In this case, with less than full indexation of wages (i.e., the revenue base for pension funds), full indexation of minimum pensions would generate a worsening of the financial position of the funds in the short term, which would be exacerbated by population aging in the longer term. However, a dilemma, particularly with high inflation and indexation procedures applied quarterly, is that Less than full indexation will erode the real value of pensions rather quickly, leading to a further deterioration of the living standards of pensioners. Some reports have argued for ad hoc adjustments in benefits to ensure that the growth in expenditures does not exceed the resources of the pension funds. The safety net for pensioners in this case may have to be provided by other schemes, such as quantity food stamps or their cash equivalent.
Recommendations concerning family allowances have varied. Some reports have argued for means testing, treating family allowances as social assistance. Others have taken the view that universal allowances have a role to play in reducing “need” and that means-tested assistance should be seen as a “local supplement” to the family allowances and other cash benefits. Under these circumstances, reports have recommended that the allowances be simplified—for example, a single allowance at a flat rate per child provided at a relatively low level commensurate with resources. With falling real revenues, however, full indexation of allowances would not be feasible. Some reports have cautioned against the use of such allowances in countries with high birth rates. Local social assistance requires institution building, and the scope for World Bank assistance in this context has been recognized.
Some countries are facing difficulty with the administration of earnings-related unemployment benefits, and some reports have supported the adoption of flat-rate benefits, set around the minimum wage; others have argued for a two-tier system, with low benefits (below the minimum wage) provided for fresh entrants to the labor force and the long-term unemployed. One problem, however, is that to ensure the political acceptability of unemployment, the flat-rate benefit for those laid off might have to exceed the minimum wage. To control costs and encourage job search, the duration of such highertier benefits would need to be limited to, say, six months. For the longer-term unemployed and fresh entrants, some reports argued for the adoption of targeted public works and the option to retrain if training objectives were made clear. The IMF’s technical assistance has recognized the importance of institutional developments, and has supported the efforts of the World Bank, the European Economic Council (EEC), and other agencies to help establish employment agencies.
The IMF’s technical assistance has taken into account the countries’ varying conditions, administrative capabilities, macroeconomic constraints, and aggregate resource potential. These considerations have been helpful in evaluating policy options for the short run, incorporating both temporary social safety nets and some longer-term goals. Increasingly, such information is being used in the design of IMF-supported programs.
Recent experience suggests that it is possible to integrate a minimum set of social safety nets in economic reform programs. While detailed household income and expenditure data would be helpful, it is often possible even without such data to identify appropriate target groups and instruments, assess potential costs, and design social safety nets.
The mix of social safety net instruments should be determined by a number of factors: reform measures and their effects, the composition of vulnerable groups, and administrative and financial constraints. For example, with a large increase in food prices in real terms, it would be critical to introduce measures to provide the vulnerable with the means of obtaining basic foodstuffs; with a major public sector retrenchment program, it would be important to introduce unemployment benefits, but with due attention paid to promoting fast employment of the unemployed. Social safety nets should be aimed at mitigating the adverse effects of reform measures on the poor. Thus, a large increase in the consumer and producer prices of foodstuffs implies the need to target social safety nets to the vulnerable in urban areas. A high child dependency ratio implies the need to assist families with children; a high old-age dependency ratio implies the need to target the elderly. Limited subsidies are helpful, partly because governments often already have the necessary administrative arrangements in place. They need to be streamlined for the administration of transitory subsidies. As the reform proceeds longer-term approaches to social protection should be explored.
Recent experience has provided a number of lessons that are summarized here:
Whereas targeting is important for cost-effectiveness, experience suggests that targeting by categories defined by household income and assets is administratively complex and might not be feasible in many cases. Alternatives are available, such as categorical targeting of limited subsidies on key commodities (e.g., through quantity food stamps) or cash transfers to certain groups (e.g., urban households, the elderly, women with children, and the unemployed); these alternatives could be effective in protecting the poorest, although some of the nonpoor might also benefit.
Over the longer term, governments should be encouraged to enhance the administrative capacity for targeting and for providing effective social assistance. In many cases, targeted public works programs could fulfill the assistance role with relatively low overhead but might need to be augmented by categorical transfers to those unable to participate (e.g., orphans or the disabled). The interactions among instruments is important and should be considered on a case-by-case basis.
Social Safety Net Instruments
There are a variety of informal and formal social security arrangements, including community-based arrangements, formal social security institutions, and those arrangements set up by bilateral and multilateral governmental and nongovernmental agencies. Where possible, social safety nets should be built on existing arrangements, with inefficient programs reformed or adapted to address the concerns for the poor.
Limited Transitory Consumer Subsidies
Whether transfers are made in-kind or through cash should be determined pragmatically, with administrative costs and other factors, such as efficiency and equity (including intrahousehold equity), taken into account. Cash transfers provide consumers with a greater choice but could force the beneficiaries to bear the risk of unexpected price changes and might not be equitable for some family members, such as female children. Generalized consumer subsidies would be expensive and could generate considerable production disincentives.
Major budgetary savings could be achieved by a shift to more targeted provisions, either through limited cash compensation following a removal of subsidies, in-kind transfers (e.g., milk provision targeted to schoolchildren or lactating mothers, through primary health clinics), or quantity food stamps (which would limit the extent of the transfer an individual could receive and could be targeted regionally). In this manner, a rapid transition to free market pricing would be feasible. At the same time, cushions against price shocks would reduce the need for indexed adjustments in pensions, allowances, and unemployment compensation.
Formal Social Security Institutions
Adapting or reforming formal social security institutions as an element of social safety nets is important. However, due attention should be paid to the coverage of the formal social security institutions. In terms of coverage, formal social security institutions might not benefit a large number of the vulnerable in the formal and informal sectors. However, an expansion of coverage of formal social security arrangements should take into account the existing informal and traditional social security arrangements.
In reforming formal social security benefits, short-term and long-term aims might diverge. While nominal benefits would have to be adjusted for higher prices in the long run, the indexation of benefits would exacerbate financing difficulties in a transition. For pensions, in the longer term, the role of the public and private sectors should be identified. For example, the public sector might be able to provide only a limited flat-rate universal pension, with additional earnings-related benefits as an option. In the short run, there would be little scope for private pensions, and earnings-related benefits might be administratively complex. A flat-rate pension on its own, however, would reduce the fairness of the system.
An important element in reducing pension fund imbalances in many transition economies would be to increase gradually the statutory retirement age. However, this solution might aggravate unemployment problems. This stresses the importance of efficiently functioning labor markets.
Unemployment benefits might be provided in combinations of lump-sum severance payments and benefits over time. There would be less unfairness in flat-rate benefits for unemployment than for pensions, but eligibility should be defined tightly, as should the duration of the benefits. Fresh entrants to the workforce or the long-term unemployed could be assisted through other targeted measures, such as public works programs, and financed from general revenue rather than from the unemployment insurance fund.
Considerably more work is needed to identify effective administrative arrangements for social protection. It is important not only to integrate social safety nets in reform programs but also to review implementation experience.
Role of Member Governments
Experience shows that governments should play an active role in designing and integrating social safety nets in reform programs. Given the increasing body of international experience, it is important for member countries to seek assistance, for example, from international agencies, including the IMF, and bilateral donors.
Potential Role of Existing Instruments as Social Safety Nets
Many countries use consumer subsidies as a means of social protection. The aim is often to ensure that vulnerable groups have access to low-priced basic necessities, such as foodstuffs, energy products, and transportation services. To achieve this goal, many countries rely on administrative controls on consumer prices set well below marketclearing levels. While such general subsidies can ensure wide coverage, the associated provision costs can place unsustainable burdens on the budget or the balance of payments. A general subsidy may also undermine the establishment of appropriate prices for farmers, given arbitrage possibilities. In some cases, compensatory input subsidies are provided to farmers, causing further production distortions and budgetary cost and creating a barrier to the effective operation of market signals. The administration of consumer subsidies through generalized price controls creates conditions for the wasteful consumption of subsidized commodities.
When unrealistically low consumer prices coexist with realistic producer prices, explicit consumer subsidies imply a substantial budgetary burden. In many cases, subsidies are implicit and not transparent, as governments finance them by either keeping domestic producer prices at low levels or applying an overvalued exchange rate to the imports of the subsidized products—thus creating allocative distortions, domestic shortages of subsidized products, and parallel markets where subsidized commodities are traded at much higher prices. Quite often, unrealistically low domestic prices result in the wasteful feeding of wheat to livestock and the smuggling of food and energy products.27
Generalized consumer subsidies on essential commodities (e.g., basic foodstuffs), although distortionary, do help the poor. These subsidies tend to provide wealthier groups with larger absolute benefits, but the benefits for poor groups account for a large share of their expenditure. It is important to phase out subsidies for items consumed primarily by the rich and to replace consumer subsidies with more targeted measures.
Social Security Arrangements
Existing institutions include informal arrangements in traditional societies and formal sector social security, with or without budgetary support.
Informal Social Security Arrangements in Traditional Societies
The extended family remains the principal source of support for the elderly in societies as diverse as rural China,28 Indonesia, India, sub-Saharan Africa, and some BRO countries. This typically reinforces the desire and need for male children, limiting prospects for a reduction in the birth rate. Problems arise when there are no male offspring or when there is a breakdown in family relationships—often connected with an inability to provide support. In Africa, the term for being poor is often synonymous with the absence of family or friends.29
Support for the needy that have no sources of family support tends to be based on community-level food security arrangements. For example, in The Gambia, deliveries of staple food for the needy are based on the precept of zakat (a religious tax earmarked for the needy). The zakat is formalized in some Muslim countries and based entirely on personal responsibility for others. Also, informal arrangements of mutual labor transfers may provide up to half of the total labor input at the community level. For example, when able-bodied persons are sick, or if there is a local landslide, neighbors provide labor assistance (see von Braun, 1991). In rural China, a community-based system of wu bao (five guarantees) provides social assistance to the needy who have no sources of family support. This system, however, suffers from stigma and take-up is low. Nevertheless, in the Islamic Republic of Iran, there is almost full take-up of community-based local support for categories of widows, orphans, and the elderly.
Coping mechanisms that work relatively well in normal times may fail in times of severe stress. For example, communities with marginal capacity to deal with normal shocks can be hard pressed to protect themselves in the face of consecutive shocks. With the evolution of urban, nuclear households, some family and ethnic ties have dissolved, weakening community support systems in many countries.
While such community-based social security provisions are often based on fairly sophisticated redistributive and insurance aspects, these provisions do not exist in all countries. Even where they do exist, traditional social security mechanisms based on the family encourage population growth and may lead to pressures on community resources.
While the local provision system can be effective in identifying the needy, without wider public support, it tends to break down if local shocks are widespread. Thus, formal mechanisms are needed to protect the vulnerable on a consistent basis, and particularly to ensure that the financing needs are met.
Formal Social Security Arrangements in Developing Countries
Although a number of countries have some formal social security arrangements (pensions, unemployment insurance, family allowances, and social assistance), the coverage is often limited to a part of the formal sector or the public sector. As such, they cannot protect broad groups of the population from the effects of economic shocks that may arise as a result of stabilization or structural adjustment in a country.
There is scope for formal instruments in low-income countries, keeping in mind overall fiscal constraints. In India, for example, formal sector employment, although a small part of the total labor force, accounts for 27 million persons. As in many other developing countries, the restructuring of the public sector is handicapped by the absence of unemployment insurance mechanisms, which in the past were considered inappropriate for low-income countries. Fairly rapid urbanization and the spread of wage-based activities in rural areas in India have increased the prospects for contribution-based social security systems, which, given demographic patterns, should generate surpluses for pensions in the immediate future. Noncontributory assistance schemes are also growing in importance, and the rural widows’ pension programs in several states attest to the importance given to this vulnerable group.30
Many francophone countries in Africa have family allowance systems that provide cash transfers to children of covered individuals. As with pensions in Latin America, the crucial issues are limited coverage and cost. A system—financed through general revenue—under which only formal sector wage earners and their offspring are eligible would be badly targeted. Moreover, a pronatalist instrument may be an inappropriate transplant in countries where birth rates are already high.31
Many Latin American countries have programs with relatively extensive coverage and aging populations. Most began with full funding of social security through trust funds.32 However, their financial positions have been weakened through the forced holdings of government paper at below-market interest rates, rising payroll tax rates (which have encouraged tax evasion), inappropriate investments, and the proliferation of benefits. A typical practice has been for governments to use pension fund surpluses to cross-subsidize burgeoning health care expenditures. The schemes have now become mainly pay-as-you-go (PAYG) and face financial crises—notably Argentina, Brazil, Mexico, and Uruguay, and also in Bolivia, Costa Rica, and Peru. Indexed benefits, in an inflationary context, together with declining contributions33 have increased the pressure on general revenue to meet the deficits. There has been, however, an inability to raise resources to meet the current obligations and projected costs34 because of the high rates of indirect, income, and payroll taxes.
The high payroll tax rates—in excess of 30 percent of the wage bill in many Latin American countries—have led to evasion, reducing the base for contributions and increasing the incentive to hire uninsured workers from the informal sector or use independent contractors. Thus, the potential for providing for the elderly has not always been realized, given the limitations of coverage to the formal, mainly urban, sector. However, relatively extensive coverage has been a major element in improving equity in Chile and Costa Rica,35 where the labor force is relatively well organized in both the urban and rural sectors.36 A number of countries in Latin America are making efforts to reform their social security systems.
Social Security Arrangements in Transition Economies
Formal social security arrangements in transition economies provide only partial social protection. In many former centrally planned economies, social security objectives have been met essentially through a system of employment guarantees, the provision of basic consumer goods at artificially low relative prices, and access to fairly costly social security programs, including various allowances administered by state-owned enterprises. However, such systems have not succeeded either in preventing poverty or in ensuring the availability of consumer items. The provision by state-owned enterprises of social functions such as housing, medical attention, and care for the elderly has contributed, in part, to a “soft” budget constraint,37 impeding their speedy transformation. Reform issues thus include a reevaluation of the existing social security arrangements, adequate benefits to remove the responsibility of provision from enterprises beyond employers’ contributions, and coverage of newly recognized contingencies (e.g., unemployment). These requirements have to be juxtaposed against the cost of provision and the disincentives created by excessively high rates of payroll contributions.
During the transition, a fall in the real level of revenues has often been observed, whereas expenditure commitments have continued unabated.38 There are, therefore, severe budgetary constraints in meeting new expenditures on social protection. Moreover, it might not be possible to continue to increase payroll taxes, which are, in some cases, already as high as 50 percent or more of the wage bill.
To achieve cost-effective social protection for vulnerable groups in transition economies, it is evident that temporary measures (e.g., targeted food subsidies) might need to be considered, together with a reform of formal social security arrangements.
Targeted Public Works
Targeted public works could provide continued subsistence transfers where unemployment insurance is not available or is too costly. Such schemes could provide needed assistance rapidly at low cost, and could be effectively self-targeted if employment were provided at subsistence wages. Nonetheless, permanent employment guarantees, through wage subsidies and support to industries and sectors that are uncompetitive in an open trading environment, are unlikely to be sustainable in budgetary terms and have stifling effects on incentives and initiative. Furthermore, more efficient activities are penalized for the benefit of the inefficient.
Public works employment at subsistence wages and distribution of relief are also important for famine prevention,39 as seen with the Maharashtra Employment Guarantee Scheme in India in 1972–73 and again in the early 1980s. The Maharashtra drought was of greater severity than those in sub-Saharan Africa, which resulted in several severe famines. Use of public works in Maharashtra and other similar projects prevented famine and appeared to be fairly well targeted. Similarly, in Bangladesh, the rural works program proved to be an effective safety net in preventing a repeat famine in 1988, under conditions similar to those which led to famine in 1974. Public works employment for famine prevention was also strikingly successful under very difficult conditions in the African context in the 1980s, such as in Botswana and Cape Verde (see Drèze and Sen, 1989).
Chile, which has one of the oldest and most advanced social insurance systems in Latin America, used self-targeted employment programs during two major crisis periods, 1974–77 and 1982–84, to compensate for severe unemployment and a decline in the real wage. The Programas de Empleo (EEP) employed at low wages, largely in the form of basic food products, around 6 percent of the labor force by 1976. At that stage, open unemployment was 16 percent of the labor force. During the second shock, a third of the labor force became unemployed, but only half sought protection from the EEP; others sought assistance elsewhere. Because of the self-selecting nature of the EEP, the total budgetary cost was 1.4 percent of GNP at the height of the crisis in 1983. The EEP was gradually run down with a tightening of the labor market, and was withdrawn completely in 1988 as open employment returned to around 6 percent, or close to 1970 levels (see World Bank, 1990).
AhmadEhtisham1992“Poverty, Demographic Characteristics and Public Policy in CIS Countries,”paper presented at the 48th Congress of the International Institute of Public Finance Seoul.
AhmadEhtisham and AtharHussain1991“Social Security in China: A Historical Perspective,” in Social Security in Developing Countriesed. byAhmadEhtisham and others (Oxford, England: Clarendon Press).
AhmadEhtisham and YanWang1991“Inequality and Poverty in China: Institutional Change and Public Policy, 1978 to 1988,”World Bank Economic Review (International)Vol. 5 (May) pp. 231–57.
AhmadEhtisham and Ke-youngChu1993“Russian Federation: Economic Reform and Policy Options for Social Protection,” in Transition to Market: Studies in Fiscal Reformed. byVitoTanzi (Washington: International Monetary Fund). pp. 135–54.
AtkinsonA.B.1992“The Western Experience with Social Safety Nets,”Welfare State Programme Discussion Paper No. 80 (London: Suntory-Toyota International Centre for Economics and Related Disciplines, London School of Economics).
AtkinsonA.B.1993“On Targeting Social Security: Theory and Western Experiences with Family Benefits,”Welfare State Programme Discussion Paper No. 99 (London; Suntory-Toyota International Centre for Economic and Related Disciplines, London School of Economics).
BeveridgeBaron William Henry1942Social Insurance and Allied ServicesCmnd. 6404 (London; HMSO).
CoxD.R. and EmmanuelJiménez1992“Private Transfers and the Effectiveness of Public Income Redistribution in the Philippines” (unpublished; Washington: World Bank).
DrèzeJean1990“Widows in Rural India,”Development Economics Research Program Discussion Paper No. 26 (London: London School of Economics and Political Science).
DrèzeJean and AmartyaSen1989Hunger and Public Action (Oxford, England: Clarendon Press).
GroshMargaret E.1995“Toward Quantifying the Trade-Off: Administrative Costs and Incidence in Targeted Programs in Latin America,” in Public Spending and the Poor: Theory and Evidenceed. byDominiquevan de Walle and KimberlyNead (Baltimore: Johns Hopkins University Press for the World Bank).
GuhanS.1988“Social Security in India,” in Poverty in India: Research and Policyed. byB.HarrissS.Guhan and R.Cassen (New Delhi: Oxford University Press).
GuptaSanjeev and KarimNashashibi1990“Poverty Concerns in Fund-Supported Programs,”Finance & DevelopmentVol. 27 (September) pp. 12–14.
IliffeJohn1987The African Poor; A History (New York; Cambridge; Cambridge University Press).
LeGrandJulian and DavidWinter1987“The Middle Classes and the Defence of the British Welfare State,” in Not Only the Poor: The Middle Classes and the Welfare Stateed. byRobert E.Goodin and JulianLeGrand with JohnDryzek and others (London; Boston: Allen & Unwin).
Mesa-LagoC.1991“Social Security in Latin America and the Caribbean: A Comparative Assessment,” in Social Security in Developing Countriesed. byEhtishamAhmad and others (Oxford, England: Clarendon Press).
MilanovicBranko1992“Distributional Impact of Cash and In-Kind Transfers in Eastern Europe and Russia,”Research Paper Series No. 9 (Washington: World Bank).
NashashibiKarimSanjeevGuptaClaireLiuksilaHenriLorie and WalterMahler1992The Fiscal Dimensions of Adjustment in Low-Income Countries IMF Occasional Paper No. 95 (Washington: International Monetary Fund).
TanziVitoed. 1992Fiscal Policies in Economies in Transition (Washington: International Monetary Fund).
von BraunJoachim1991“Social Security in Sub-Saharan Africa: Reflections on Policy Challenges,” in Social Security in Developing Countriesed. byEhtishamAhmad and others (Oxford, England: Clarendon Press).
Note: An earlier version of this paper was discussed by the IMF (Executive Board on June 2, 1993. The paper represents a collective effort by the staff of the Expenditure Policy Division of the IMF’s Fiscal Affairs Department. Ehtisham Ahmad prepared the initial draft.
In many countries, formal public social protection schemes, including consumer subsidies and social security arrangements, coexist with traditional or informal private or community-based schemes. Social security arrangements in this chapter refer to social insurance and social assistance programs. The former (e.g., pensions and unemployment insurance) may not necessarily be targeted to the poor, while the latter generally are targeted to the poor. Note that the term social security is defined broadly in this chapter, although in countries such as the United States, it is often synonymous with state pensions. Social security arrangements provide cover against the loss of income resulting from various life cycle, natural, and other contingencies, even in the absence of economic reform; social safety nets as discussed in this chapter refer primarily to measures that protect vulnerable groups from the impact of reform, while, subject to resource constraints, also providing for other contingencies, See Atkinson (1992) for an alternative definition of social safety nets.
It does not address the issue of the design of the macroeconomic and structural reform package, including the implications of its scope and speed for demands on social safety nets, but assesses appropriate social safety nets that would underpin the required adjustment.
The appendix to this chapter provides a brief discussion of major social protection institutions existing in various countries, focusing on consumer subsidies, social security arrangements, and certain targeted public works that proxy social assistance. Eventually, some of these measures (e.g., consumer subsidies) should be phased out, as more effective transfer mechanisms are established. See Cox and Jimenez (1992) for a discussion of the possibility that public transfers could crowd out private transfers.
The World Bank has been helping countries enhance their capacity to compile household income and expenditure surveys and to define minimum consumption baskets. Eventually, this work will be helpful to the IMF’s work.
See, for example, Atkinson (1992b) for a discussion of administrative difficulties of means testing in industrial countries.
The presumption is that rural residents generally benefit from the increases in procurement prices. This leaves the problem of identifying and providing benefits to net purchasers in rural areas. With a very substantial output response in agriculture, real food prices in China have fallen consistently in recent years, which together with the increase in the price of subsidized rations has led to a situation where the ration is being withdrawn without much of an impact on urban households. See Ahmad and Wang (1991).
New Zealand is an example of a country that introduced a flat-rate value-added (VAT) on all transactions but transferred all social security expenditures to the budget.
If a comprehensive income tax system could be established, it would be possible to “claw back” the provisions to middle-class and wealthy beneficiaries through taxation. In many developing and transition economies, however, it might be unrealistic to expect such a system to be established in the near future.
This criterion also indicates that commodities not generally consumed by the poor should not be subsidized. While consumption patterns vary among countries, subsidies on frozen meats and macaroni in Egypt, and on sardines, chicken, and salted fish in Jamaica, are inappropriate, as such commodities are not generally consumed by the poor.
Coarse grains and unrefined sugar are known as “gur” in India and Pakistan.
This was the case in many “adjusting” countries in South Asia in the late 1980s. For example, during this period in Bangladesh, a special ration provided benefits to civil servants and the military, and formed part of a real wage package but could not be conceived of as a social safety net.
This has been used in market economies at times of stress or disruptions of supplies, such as in the United Kingdom and its colonies during World War II.
Jordan has successfully introduced this system.
The earliest objectives of formal social security included maintaining—without stigma—minimum living standards, and replacing income in times of adversity. However, social security also entails a smoothing of consumption or a spreading of income over the life cycle, and reducing risk. Often there is a redistribution of income between groups with differing needs. Moreover, the different configurations of political and interest groups influence the directions that social provision has taken in different countries. This variety of objectives (in conjunction with political realities) has led to the different manifestations of social security.
For example, old people that are already privately uninsurable but are strong candidates for social insurance.
While means-tested social assistance, often provided at the local level, is an important element in the social security systems in many industrial countries, the development of social insurance and other categorical transfers, such as universal child benefits, are the result of “[t]he strength of popular objection to any kind of means test.” See Beveridge (1942).
This happens, for example, as increases continue in the proportion of old people, as in many Eastern European and Asian countries, or in the proportion of children under age five, as in Africa.
For example, in BRO countries, the replacement rate for pensions fell from 60 percent in the late 1950s to around 30 percent by the late 1980s.
They are, in general, “pro-poor.” See Milanovic (1992).
In some countries, such as in Central Asia or North Africa, high levels of family allowances might be quite inappropriate given demographic realities and fiscal constraints. See Ahmad (1992).
See Grosh (1995) for a comparison of administrative costs for various targeting mechanisms in Latin American countries.
Under generalized food subsidies, the share of the benefits received by families that belong to the poorest quintile could be as low as 8 percent; for the wealthiest quintile, it could be as high as 36 percent (see Grosh, 1995). In Jordan, the replacement of generalized subsidies with a system of quantity food stamps led to a reduction in consumer subsidies from 3.2 percent of GDP in 1990/91 to 1.4 percent of GDP in 1991/92. Calculations for Russia and Algeria suggest that similar schemes could generate a 50–60 percent reduction in outlays on subsidies without adversely affecting the food consumption of the poorer groups of the population. See Ahmad and Chu (1993) for calculations for Russia.
For example, payroll tax rates for pension funds in a 30–40 percent range are not uncommon in BRO countries.
It is also noteworthy that family allowances in many of these countries are pronatalist.
In some BRO countries, low producer prices have induced farmers to feed subsidized bread to their livestock or to sell wheat to neighboring countries rather than to the central procurement agency, forcing the governments to allocate scarce foreign exchange reserves to the importation of foreign wheat.
See Iliffe (1987).
Widows without grown sons tend to face difficulties and, thus, are among the most vulnerable, even if their extended families are relatively prosperous (see Drèze, 1990). The Indian state of Kerala, which has performed exceptionally well in terms of social indicators, has maintained an old-age pension scheme for the unorganized sector since 1961, but destitution has not been a criterion for eligibility (see Guhan, 1988). Most other Indian states have been less successful in terms of coverage, as stringent means tests have restricted eligibility and have been administratively cumbersome.
Children in large families are often quite vulnerable. However, the provision of family allowances with universal coverage may not be administratively feasible in some poor countries.
This is due to collection lags (Tanzi effect) and increasing noncompliance in many cases, as well as to the shrinking formal resource base.
An aging population can put additional pressure on the schemes. The relevant consideration, even with such demographic profiles, is to bring younger groups of workers into the system and to increase retirement ages. In countries with younger demographic profiles, there are likely to be surpluses in the initial stages. However, even in such countries, the viability of the social security system depends on the liabilities to be incurred relative to the contribution base of the covered workforce, and a young overall demographic structure does not guarantee surpluses.
Evaluations of the distributional consequences of social security provisions would entail the examination of the entire system of benefits, funding, and taxes, and of the implications for the population as a whole.
Because of the social services provided, additional subsidies or tax reductions were required to keep the enterprises functioning.
See Tanzi (1992).
See British India (1880).