DESIGNING AND EVALUATING SOCIAL SAFETY NETS: THEORY, EVIDENCE AND POLICY CONCLUSIONS

This paper reviews the literature on the performance of commonly found socialsafety net programs in developing countries. The evidence suggests that universal foodsubsidies have very limited potential for redistributing income. While targeted foodsubsidies have greater potential, this can only be realized when adequate attention isgiven to the design and implementation, as well as to the social and political factorsinfluencing the adoption, of these programs. Although well-designed public worksprograms have impressive targeting performance, they have large non-wage costs; thus,to be cost-effective, they need to produce outputs that are especially beneficial to poorhouseholds. Social funds, which emphasize both community involvement and assetcreation, have been cost-effective, but they are difficult to target to extremely poorhouseholds. Traditional public works programs are particularly attractive for addressingvulnerability, but they require flexibility regarding choice of output. Targeted human capitalsubsidies appear to have great potential for addressing extreme poverty; but again,their design needs to reflect the human capital profile of countries and the administrativecapability of the government.