The impact of Ethiopia’s Productive Safety Net Program on schooling and child labor JOHN HODDINOTT , DANIEL O . gILLIgAN AND
Introduction The period since the early 1990s has seen an expansion of social protection programs in developing countries aimed at protecting households from economic shocks, reducing poverty and improving the well-being of the poor. As research on the effects of these programs has increased, interest has shifted to the benefits of these programs for child welfare. Much of the literature has examined whether program transfers to households lead to increased human capital investment in children through greater access to education and improved health (Duflo 2000). The proliferation of conditional cash transfer programs (CCTs), which make payments conditional on such investments by tying transfers to school attendance and child growth monitoring, has further increased interest in impacts of social protection programs on human capital investment (Schultz 2004; Levy and Ohls 2006). Few studies have examined whether social protection programs have an effect on child labor (Ravallion and Wodon 2000; Maluccio and Flores 2004). Child labor, especially in its worst forms, directly reduces the current welfare of children. It may also reduce time available for schooling, which has long-term effects on future welfare. Social protection programs may reduce child labor time by increasing income or by directly tying transfers to school attendance. Public works (PW) programs, a form of social protection that remains common in Sub-Saharan Africa (SSA) and India, provide households with income transfers in exchange for participation in labor-intensive work projects.2 These programs have multiple objectives, including providing income to poor workers, smoothing household consumption and constructing or rehabilitating public infrastructure. They act as a form of employment insurance for the poor and may include on-the-job training to reintegrate low-skilled workers into the labor force (Grosh et al. 2007). Despite the popularity of PW programs, little is known about how they affect child welfare. Economic reasoning tells us there will be two types of effects: an income effect from the transfers received and a substitution effect from the additional labor demand. The income effect of PW on schooling and child labor is similar to that of other social programs. The literature on human capital investment posits that if child schooling is a normal good, the increased income will
result in increased levels of school participation (Behrman and Knowles 1999). In the child labor literature, Basu and Van’s (1998) “luxury axiom” states that child labor will decrease as income increases above a subsistence threshold. An important difference between PW and other social protection programs is that the labor requirement in PW leads to substitution effects that can be detrimental to child welfare. Public works programs directly increase demand for household labor and may alter the intra-household division of labor between adults and children in ways that could adversely affect both schooling and child labor. This relation follows from what Basu and Van call the “substitution axiom”, that adult and child labor are substitutes. The income and substitution effects have opposing influences on child labor; a priori, it is not possible to tell which effect will dominate. The magnitude of these effects will depend on the amount of labor required under the program, the wage paid, the opportunity cost of adult household member time and the child’s productivity in household activities. We can learn something about the likely effects from the debate on the likely effects of rising economic status on child labor cited in Edmonds (2005, 2007). If the source of income growth is driven by increasing returns to labor time, the impact on child labor is ambiguous. Similarly, it is not possible to tell whether the increased demand for adult labor from PWs will increase or decrease school participation, which depends on the opportunity cost of schooling and future returns to schooling. Moreover, because parents can alter the amount of leisure that their children consume, it is possible for both child labor and school attendance to increase, with leisure time reduced (Edmonds 2007). While theory provides a useful guide in framing the possible impacts of PWs on the well-being of children, empirical evidence is needed to determine which effect is greater. However, on this topic, no such evidence exists. Recent reviews of PWs (Grosh et al. 2007; Subbarao 1997, 2003) provide no evidence, nor does Edmonds’ (2007) exhaustive review of the child labor literature. The objective of this research is to begin to fill this knowledge gap. We examined the impact of Ethiopia’s Productive Safety Net Program (PSNP) and complementary transfers on child schooling and child labor. The PSNP, the result of a re-engineering of food aid in Ethiopia in 2005, provides food or cash transfers to households through large PW projects or through a smaller program of direct transfers to households with elderly or disabled members. The PSNP is the cornerstone of the Ethiopian government’s food security program, which now operates as a standing safety net, rather than a system based on emergency appeals. Since 2005, it has provided assistance to more than seven million people, with annual transfers averaging $300 million. To our knowledge this makes it the largest social program in Africa after South Africa’s pension program. Using quasi-experimental evaluation techniques based on matching, we estimated the size of the effects of participation in PWs under the PSNP on child school attendance and child labor hours devoted to farm work or domestic tasks. We differentiated the impacts by gender and age of the child, since the division of household labor, returns to labor activities and returns to school participation
all differ by gender and may vary by the age of the child. We also investigated how these impacts vary by the intensity of participation in the PW projects and by the use of production credit from a complementary food security program that boosts demand for farm labor. We found that participation in public works led to a moderate reduction in agricultural labor hours on average for boys aged 6 to 16 years and a reduction in domestic labor hours for younger boys aged 6 to 10 years. Boys in households receiving more regular transfers (at least 90 birr per member) showed large increases in school attendance rates and, at the younger age, a significant reduction in total hours worked. When public works was coupled with agricultural packages designed to boost farm productivity, there was no effect on boys’ schooling, and labor hours decreased only for younger boys in domestic chores. For girls, the measured effects were weaker, but differences emerged between younger (aged 6 to 10) and older (aged 11 to 16) girls. Younger girls experienced worse outcomes, with lower school attendance on average and increased child labor in households participating in PWs and the Other Food Security Program (OFSP). Older girls benefited, with a reduction in labor hours on average and an increase in school attendance in households receiving larger transfers.