Impacto de una pension minima en la poblacion de tercera edad y sus costos presupuestales. Evidencia para Latinoamerica.

• Author: Dethier, Jean-Jacques
• Position: Report
• Pages: 135(29)

The impact of a minimum pension on old age poverty and its budgetary cost. evidence from Latin America

1 Introduction

Alleviating old-age poverty requires different approach from other age groups. Since policies that go through labor and output markets and educational and training programs are ineffective, the only available instrument to alleviate old age poverty is a transfer of real income (possibly through price subsidies). In most developed countries, pension systems-which generally consist of a balanced combination of pay-as-you-go and funded schemes-include minimum pension schemes and are strongly redistributive, yielding a sizeable difference between poverty rates before and after transfer. (1) By contrast, in developing countries with pension systems, one observes that they have a limited potential to solve old-age poverty because of the low coverage of those systems. Coverage rates are below 30% in half Latin American countries. They range from around 10% of the labor force in Peru and Bolivia to about 60% in Chile. These figures are for 2006 and are based on the number of contributors (Mesa-Lago, 2004a; Rofman et al., 2008). As to the coverage of the elderly, the rates are extremely low in most Latin American countries. They range from 5% in Honduras to 85% in Uruguay. They are about 60% higher than 60% in the traditional four, Argentina, Brazil, Chile and Uruguay (ABCU, hereafter) plus Costa Rica and Bolivia.

More efficient solution consists of lump-sum transfers financed by tax receipts. These are pensions aimed at providing a replacement income to old persons under the poverty line and are of two types (Willmore, 2001; Holzmann et al., 2009). (2) The first type of minimum pension covers unconditionally all the elderly. Benefits are the same for everyone regardless of income, assets or work history. This distinguishes it from means-tested pensions which do not provide benefits (or provides reduced benefits) to those who have other income or assets, and from the minimum pension guarantee for which beneficiaries must have a history of contributions. In the OECD, only one country (New Zealand) provides a universal pension to its aged population with the objective to lift old persons above the poverty line. In low and middle income countries, only four countries have such universal minimum pension arrangements: Mauritius, Namibia, Botswana and Bolivia (On Mauritius, see Willmore, 2003). They are easy to administer and do not require information on the income of assets of the beneficiaries. They offer a pension which is relatively low and, with the exception of Mauritius, not high enough to lift its beneficiaries above the poverty line.

The second type of minimum pension is also universal but subject to means-testing. This welfare pension can be completed by housing subsidy or the possibility of being admitted in a public nursing home. (3) A number of developing countries have universal means-tested schemes although the means test applies to the household and not to the individual. The most famous examples are rural Brazil and South Africa. The South African minimum pension is quite generous in terms of level (about one-third of per capita income) and the number of beneficiaries is high reaching 88% of the covered population. The pension is paid to men aged 65 and women aged 60 and over. It is funded through general taxation. The Brazilian minimum pension, for which the eligibility age is 60 for men and 55 for women, corresponds to the minimum wage (Beltráo et al., 2004). It is also worth mentioning Mexico City (Federal District) and its program of transfers for food expenses to the elderly living in poor areas. A few studies examine the incidence of minimum pension schemes. Barrientos et al. (2003) studies the effect of social pensions on the poverty rate of elderly people in rural Brazil and South Africa and computes poverty rates and poverty gaps with and without means-tested minimum pension. He shows that, in both countries, the non-contributory pension reduces both the rate of poverty and the poverty gap. Rivera-Marques et al. (2004) study the incidence of Mexico City's safety net for the elderly and show that the program reduces poverty and inequality but that its performance in terms of poverty reduction is weaker as soon as the eligibility rules are relaxed (no means test and extension to non-poor areas). Other recent analysis of universal means-tested pension schemes (which are discussed below) include Carvalho and Evangelista (2001), Bertrand et al. (2003) and Duflo (2003).

In Latin America, five countries-Argentina, Brazil, Chile, Costa Rica and Uruguay-have non-contributory pensions (Bertranou et al., 2004). These programs generally have a social assistance character. In that they are targeted at the poor and disabled who have no contributory capacity. In Brazil and Costa Rica, part of the social assistance pension benefits is financed by cross-subsidies from social insurance programs. In terms of coverage, Chile, Uruguay and Costa Rica offer the greatest coverage but, in absolute terms, Brazil has a social assistance program with more than 2 million beneficiaries and, if the rural pensions program is included, the number of beneficiaries exceeds 8 million. Table 1 gives the main features of Latin American pension policies. For more details, see, e.g., Holzmann et al. (2009) and Mesa-Lago (2004a). Even with high rate of coverage poverty will only be eradicated if benefits are high enough and the family structure is not too burdensome.

2 Evidence on poverty in old age

At the international level, surprisingly little evidence is available on poverty in old age. For example, in its statistical publications, the World Bank does not report poverty rates for all age groups (World Bank, 2005). Data on child poverty are published separately but not data on poverty in old age. Only recently have there been efforts to publish internationally comparable indicators of welfare from an age-specific perspective for rich and poor countries (see for example HelpAge International (2004) and Kakwani et al. (2004)).

In developed countries, the old age poverty rates are generally not much lower than those for the total population but this is a relatively recent trend. A few decades ago, the average income of the elderly was substantially lower than that of other age groups and their rate of poverty much higher (Förster et al., 2003). In developing countries, patterns are different and there is no obvious trend. As far as Latin America is concerned, the poverty headcount for the elderly is clearly lower than for the population average in the cases of ABCU and to a lesser extent in Nicaragua and Panama. It is higher in the other countries as shown in figures 1 and 2 below. (4) These four countries, Argentina, Brazil, Chile and Uruguay, which are among the richest in our sample, will often behave differently from the rest.

Poverty in old age can still be observed even in countries--for instance in the OECD--that have generous transfers for the elderly including targeted minimum pensions. This seems puzzling at first sight since the pension is universal and its level is above the poverty line (generally half the median income). There are at least three reasons for this apparent puzzle: family composition (if the other family members do not have any resource. the equivalent income of each member can be below the poverty line); take-up issue (when the pension is means-tested some individuals can be reluctant to claim it) and finally, given that it is subject to a means-test, some elderly people prefer to keep their assets even if these assets don't generate much income.

2.1 Old age poverty rates under current policies

Figure 1 presents the poverty ratio for the persons older than 60 and for the whole population in Latin America. The poverty ratio is based on a poverty line equal to half the median income of the household. Figure 2 present the poverty ratio using a different definition of the poverty line, namely a poverty line equal to $2 a day. The equivalence scale we use is the OECD scale that is equal to 0.5 + 0.5 x number of adults +0.3x number of children (up to age 16). (5)

With the poverty rate calculated with the OECD scale and a poverty line equal to half the median income, Brazil, Chile, Uruguay, Argentina (and to a lesser extent Nicaragua) have low poverty rates comparable to most OECD countries (below 11%). (6) These four countries are often associated as having the same "mature" treatment of old age. It is worth noting that they do not all belong to the richest Latin American countries as one can see on table 5 in the appendix. Mexico and Venezuela are richer than Brazil and Argentina. For the other 14 countries the poverty rates are quite higher and in most cases higher than for the rest of the population. With the US$2 a day poverty threshold the poverty rates in ABCU become negligible (

We draw three main conclusions from the comparison of old age poverty in these Latin American countries. First, poverty rates are consistently lower for the elderly than for the whole population in Argentina, Uruguay, Brazil and Chile. (7) Second, in the other countries, the situation is heterogeneous and depends on the poverty line chosen. Using half the median income, Bolivia, Colombia, Costa Rica, Honduras and Mexico have comparable overall levels of poverty in old age and the elderly are poorer than the rest of the population. Finally, the difference between old age and overall poverty rates is not very high for all countries with limited pension systems.

Old age poverty is computed here using household surveys. As pointed out by Angus (1997), when household per capita income (or expenditure) is used as the main welfare indicator, the assumption made about the way in which resources are shared in the family to which an elderly belongs affects the quality of the estimates. The assumptions made by...