XIX. Poverty and Social Security

Ke-young Chu, and Richard Hemming
Published Date:
September 1991
  • ShareShare
Show Summary Details

How is poverty measured? Which groups are most vulnerable and why?

How instructive is industrial country experience for designing social security programs in developing countries?

To what extent do informal arrangements—based upon the extended family and local communities—provide an adequate safety net?

How can social security be targeted to the poor?

Concern about poverty in developing countries has two related aspects. Over time, a sizable reduction in poverty is important, both in its own right and as a first step in providing the means for the self-improvement of the most disadvantaged groups in society. The policies most likely to achieve these objectives are those that stimulate growth and employment, and in so doing provide incomes for the poor and use some of the resources generated to make adequate social provisions for people who cannot benefit directly from rising incomes. However, the macroeconomic stabilization and structural adjustment policies that are necessary for growth may adversely affect some poor groups in the short run. There is therefore a need to put in place policies to protect the poor from the immediate consequences of adjustment. Social security addresses both issues: the provision of safety nets is required by the short-term concern, while more comprehensive social protection meets longer-term objectives.

Assessing Poverty


Poverty embraces material aspects that can be captured in terms of monetary indicators such as income or expenditure levels. It also includes nonmaterial aspects relating to quality of life, such as nutritional and health status and educational attainment. These are less easy to measure. Thus it is usual to supplement income-based measures of poverty with non-income indicators such as child or infant mortality, life expectancy, or educational status, to generate a more complete picture of poverty in a country. This may also have to be supported by sectoral (e.g., employment status) or regional (e.g., rural or urban) profiles of the poor.

The most common practice is to measure absolute poverty in relation to a subsistence level poverty line. Obviously, appropriately differentiated income or expenditure survey data have to be available, and it would normally be necessary to rely on data collected by national authorities or, in most cases, institutions such as the World Bank. Moreover, other UN agencies (UNICEF, WFP, and IFAD, for example) will also be a primary source of poverty estimates. However, such estimates have to be treated with caution. If the poverty line is defined with respect to minimum nutrition, clothing and shelter levels, serious problems arise in the definition of such a basket of goods given price changes, regional and cross-country variation, and household taste and composition differences. Additionally, when many people have an income close to subsistence level, minor variations in where the poverty line is drawn can make major differences to estimates of the incidence of poverty.

An alternative approach is to define relative poverty with respect to living standards that prevail in a particular society. An advantage of a relative poverty line is that it reflects changing perceptions of acceptable minimum living standards. Thus, while indoor sanitation or electricity may not be considered essential in very poor societies, these may constitute basic necessities in richer societies. A wider interpretation of relative poverty would also reflect nonparticipation in activities that are generally regarded as customary. In OECD countries, current practice is to define the poverty line as a proportion—usually 50 percent—of average income. However, in some instances the bottom 20-40 percent of households in the income distribution are defined as relatively poor. While this is generally difficult to defend, in developing countries with widespread poverty, the bottom 10-20 percent of households—all of which are poor in an absolute sense—may be an appropriate group to focus upon, since the ultra-poor rather than the poor are often the crucial target group.


Having decided upon a poverty line, the most widely used measure of the extent of poverty is the number of poor, known as the head count. This has the shortcoming that it is insensitive to the actual income levels of the poor. Thus a transfer from the poorest to the least poor, which raises the income of latter above the poverty line, would reduce the head count. However, if account is taken of the poverty gap, that is the average deviation of the incomes of the poor from the poverty line, it is less obvious that poverty has fallen. But this still gives equal weight to all poverty gaps. If deprivation increases more than proportionately with individual poverty gaps, then inequality among the poor matters, and more weight would need to be attached to the poverty gaps of poorer individuals. More sophisticated poverty indices reflect each of these three aspects of poverty. For example, the Sen poverty index combines the head count, the poverty gap, and the Gini index of income inequality among the poor. Measures can also be based upon alternative inequality indices, some of which do not require explicit specification of the poverty line (see the note on The Distributional Impact of Public Expenditure for a discussion of inequality indices). Only in a very few cases will it be feasible to obtain estimates of these more sophisticated indices.

Characteristics of the poor

In industrial countries, poverty is closely linked to employment status. Thus the majority of the poor are elderly, disabled, sick or unemployed heads of household and their dependents. Women and children account for a significant share of the poor. Low pay tends to result in poverty only when it is associated with some other contributing household characteristic such as a need to support a disabled spouse, elderly dependents or large number of children. By contrast, in developing countries poverty is widespread among the working population. The emphasis on formal employment is far less, and landless rural workers and smallholders constitute the majority of the poor. Employment status is more important in urban areas; however, poverty incidence is often high in the low-wage informal sector. While the nonworking population in developing countries relies heavily on the family or community for financial support, in South Asia and some other rapidly growing parts of the world, widows, orphans, the disabled, and the aged are among the poorest segments of society.

Social Security in Industrial Countries

Social insurance, income maintenance and child benefits

If there were perfect insurance markets, people would be able to protect themselves against many of the contingencies that result in poverty. However, insurance markets are subject to market failure, to which social insurance is a response (see the notes on Public Expenditure and Resource Allocation and Pensions for further discussion). In the majority of industrial countries, employees contribute to a social insurance program that pays retirement and disability pensions, sickness benefit, and unemployment benefit. Social insurance often provides a significant degree of income replacement, in that benefits are related to previous earnings.

For those who are not eligible for social insurance, or for whom it provides inadequate benefits, income maintenance is usually available. This takes the form of means-tested income supplements, which guarantee a minimum income close to the poverty line. In recognition of the vulnerable position of children, child benefits are often paid outside the social insurance and income maintenance programs. Moreover, these tend to take the form of universal benefits, paid to every family with children. However, in some countries child tax allowances (or tax credits) are available instead. These have the disadvantages that they offer no benefit to families where the head is not employed and that they confer greater benefits to the rich, who face higher marginal tax rates. Minimum wage legislation can also be regarded as part of the social security system.

The social security systems in industrial countries have evolved in line with varying needs, and consequently differ in major respects across countries, and over time in a given country. Thus improved provision for the aged in the United States has been instrumental in reducing poverty among this group since the 1960s. However, unlike in France and Britain, there is no comparable system of (universal) child benefits in the United States, which has a much greater incidence of child poverty as a consequence. Changing needs are also reflected in the widespread introduction of unemployment insurance after the Great Depression and widows’ pensions during World War II.

Problems and suggested solutions

The fact that poverty is not widespread in industrial countries can be viewed as a measure of the success of social security provisions. At the same time they are subject to criticism. These include the potentially adverse effects of unemployment compensation on voluntary unemployment, of sickness benefits on attendance at work, and of pensions on savings and the participation of the elderly in the labor force, although the evidence is far from unanimous. However, perhaps the most worrisome feature of social security programs in industrial countries is the persistence of significant pockets of poverty despite the availability of income support. The fact that this affects the elderly and children especially—groups who are least able to provide for themselves—is of particular concern. The principal shortcoming of social security programs in this regard concerns the operation and impact of means testing. First, the existence of the means test and the investigative manner in which it is often administered leads to stigma, with the result that not all the poor take up the benefits to which they are entitled. Second, the means test often implies that benefits are withdrawn at a high rate as other income rises. This benefit withdrawal rate is equivalent to a marginal tax rate that can be as high as 100 percent; moreover, even if it is lower, the combined impact of the withdrawal of a number of different benefits with income tax rates that are sometimes levied on very low incomes can impose effective income tax rates well in excess of 100 percent on the poor. Such punitive tax rates provide little incentive for the poor to work their way out of poverty; they therefore risk falling into a poverty trap.

To improve the effectiveness of social security in industrial countries, attention has focused on alternatives to means testing and the elimination of high marginal tax rates. The most radical proposals require the introduction of integrated tax-benefit schemes whereby benefits are paid to everybody—thus eliminating the means test—and then recouped through the tax system from those who do not need them. This avoids unintentionally high marginal tax rates on low incomes. However, the problem is that if the minimum income guaranteed under the scheme is to be sufficient to prevent poverty, general tax rates may have to be high. There is therefore a trade-off between equity (which is best served by universal provision of a high guaranteed minimum income) and efficiency (which requires low tax rates).

Social Security in Developing Countries

Constraints on social security policy

In addition to the different characteristics of the poor described above, social security policy in developing countries is constrained by a number of other factors. One important difference between industrial and developing countries relates to the severity of poverty in the latter and widespread failure to secure minimum food requirements. Other constraining factors include the limited range of activities in the formal sector, resource availability, and administrative capacity in developing countries. These differences have a number of implications for the design of social security in developing countries. Since employment status provides a poor guide to poverty risk, the primary focus will not be on providing social insurance. Instead, the combination of extensive poverty and limited resources suggests that the focus should be on establishing a safety net that guarantees access to minimum requirements, in particular food. In this connection, the targeting of benefits should be attached considerable importance. Administrative difficulties both in identifying beneficiaries and delivering benefits—together with cost considerations—suggest that existing community-based arrangements should be an integral part of the social security system.

Lessons from industrial countries

Insofar as the focus of social security policy in developing countries is on the provision of safety nets, the industrial country experience is of only limited relevance. In the early stages of development, there will be little overlap between social security systems in developing and industrial countries. However, one objective of developing countries should be to shift eventually from the direct provision of food and other goods and services to income maintenance, which is preferable from an allocative viewpoint. Also, as the formal sector grows, social insurance will become increasingly important. As the social security system develops in these directions, industrial country experience becomes more relevant.


Perfect targeting occurs when only the target group is covered by a program and each member of the target group receives the benefits to which they are entitled. These two aspects of targeting come into conflict. Attempts to target too precisely result in errors of exclusion, as deserving beneficiaries are denied coverage. However, efforts to improve the delivery of benefits invariably lead to errors of inclusion, as coverage widens beyond the target group. The first type of error reduces program effectiveness while the second type adds to the program’s financial cost.

Means testing ensures that only the poor benefit from social programs intended to help them. Moreover, if all the poor took the means test they would receive appropriate assistance. However, as the industrial country experience illustrates, means testing is stigmatizing, which leads to low take-up, and can give rise to high marginal tax rates, which have disincentive effects. Taking into account also its steep administrative costs, extensive means testing is probably neither feasible nor desirable in developing countries. Reliance will have to be placed instead on self-targeted forms of social security and targeting based on proximate indicators of need. There are discussed below.

Role of community-based provision

The extended family is still the basis of support for contingencies such as old age, disability, or widowhood in most developing countries. This may be augmented by community (i.e. local) arrangements, for those with inadequate family support, and such measures often replicate provisions of formal social security. Local communities are relatively efficient in identifying and providing for those in need, and therefore targeting is fairly close. There is often also a greater willingness to pay for what is readily seen to be mutual insurance. However, these arrangements may not be adequate when a particular adversity affects entire communities; for example, in drought-prone regions there is clearly a role for government in transferring consumption across regions. Moreover, as social insurance becomes more important, the role of local communities will necessarily recede. Indeed, it should be noted that while the origins of social security systems in industrial countries can be traced to charitable giving at the local level, private sector involvement is now modest and tends to be employer based. Because of insurance market failure, the government is always likely to play an increasing role as development proceeds.


As indicated above, a primary objective of social security in developing countries is to provide a safety net, not only for the aged and those unable to participate in the labor market, but also for large groups of the working poor. Food relief—in the form of food subsidies, rations, or more targeted provisions like food stamps—is usually the principal form of assistance for vulnerable groups. General subsidies are self targeting if they apply to inferior marketable commodities, but examples are not easy to find (see the note on Price Subsidies for further details). Rations approximate universal benefits in kind, and may be preferable to general subsidies for most normal goods, both on budgetary grounds and from the nutritional point of view if the rationed commodities would otherwise be priced out of affordable range for the poor owing to excess demand by the better-off consumers. Food stamps allow more precise targeting, but inevitably involve means testing. Supplementary feeding programs are often used to address severe malnutrition, providing food directly to those groups which face the greatest health risk, such as pregnant women, children, and the aged. However, because malnutrition often reflects not only food shortages but also the presence of disease, such programs have to be closely coordinated with the delivery of primary health care.

Public employment programs are also a commonly used form of social security. Poverty is reduced by providing employment to those in need, usually on labor-intensive infrastructure projects such as water and soil conservation, irrigation, flood control, roadworks, and reforestation. To be a cost-effective means of poverty relief, such programs should pay only subsistence wages, and so effectively target the poor through self-selection. Moreover, if importance is attached to food provision, the programs can be partly a form of food-for-work. Public employment programs offer the additional advantage that they often create infrastructure and other capital that can have second-round effects on the incomes of the poor.

Notwithstanding the availability of private transfers, food relief, public employment opportunities, and whatever more formal social security arrangements are in place, a number of people will be inadequately provided for. It is important to try to isolate the characteristics of the remaining poor. These may be fairly general (related to age, gender, or landholding, for example) or country specific (related to location, religion, race, or class, for example). If the most vulnerable and unprotected groups can be identified by such characteristics, this can provide the basis for a targeted safety net. Indeed, where finer targeting is impractical, such proximate indicators of poverty risk may have to be mainstay of affordable social security provision.

Country Illustration

Employment programs in Chile and Tunisia

Despite the fact that Chile has one of the oldest and most advanced social insurance systems in Latin America, the use of self-targeted employment provision at low wage levels played an important role in supplementing unemployment insurance during two major crises in 1974-77 and 1982-84. The Programas de Empleo (EEP) were instituted in 1974, and by 1976 employed around 6 percent of the labor force. At that stage, open unemployment was 16 percent. By the end of 1982, when the second major adjustment was forced, open unemployment was 16 percent. By the end of 1982, when the second major adjustment was forced, open unemployment rose to 31 percent. EEP was doubled in 1983 to cover just under half of the unemployed. However, because of the self-selecting nature of EEP, the total budgetary cost was 1.4 percent of GDP at the height of the crisis in 1983. The majority of workers in two of the main EEP schemes were in the lowest quintile of the income distribution. The EEP was gradually run down with a subsequent tightening of the labor market, and was withdrawn completely in 1988 as unemployment returned to around 6 percent. Not all public works programs are as effective in providing a relatively cheap social safety net. For example, one such program in Tunisia during the 1950s and 1960s paid such high wages that employment was rationed on a rotational basis. While the program covered up to 39 percent of the unemployed, its cost reached 5 percent of GDP. The contrast between the Chilean and Tunisian experiences emphasizes the desirability of targeting through low wages, which not only limits overall cost but also makes it easier for the scheme to be wound down when not needed.


    Ahmad Ehtisham “Social Security and the Poor,” World Bank Research Observer (WashingtonWorld Bank1990).

    Atkinson A.B. “Income Maintenance and Social Insurance,” in Handbook of Public Economics Vol 2ed. by A.J. Auerbach and M. Feldstein (Amsterdam: North-Holland1987).

    • Search Google Scholar
    • Export Citation

    Besley T. and R. Kanbur“The Principles of Targeting,” in Poverty in Developing Countries ed. by M. Lipton and J. van der Gaag (Washington: World Bank and IFPRI1990).

    • Search Google Scholar
    • Export Citation

    Kanbur S.M. Ravi “Measurement and Alleviation of Poverty: With an Application to the Effects of Macroeconomic Adjustment,” IMF Staff Papers 34 (Washington: International Monetary Fund1987).

    • Search Google Scholar
    • Export Citation

    Other Resources Citing This Publication