Cash Transfers, Household Composition, and Human Capital Accumulation in sub- Saharan Africa: Experimental Evidence from Lesotho, Malawi, and Zambia Jacobus de Hoop 1 Sudhanshu Handa 2 Ramaele Moshoeshoe 3 VERSION 2: SEPTEMBER 2017 NOT TO BE SHARED OR QUOTED Abstract: This paper presents experimental evidence on the effects of large-scale, government-run, unconditional cash transfer programs in Lesotho, Malawi, and Zambia on children’s schooling. We further examine the heterogeneous effects of these programs by the child’s household vulnerability status, and whether they have any protective effects against parental loss. The paper documents pronounced positive impacts on school participation. The beneficial impacts on schooling outcomes are mainly concentrated among children who did not live with one or both parents at baseline - a sub-group targeted explicitly in two of the three study countries - confirming that cash transfers are a valuable instrument to support the most vulnerable children. ________________________________ 1 Corresponding author, UNICEF Office of Research - Innocenti, jdehoop@uncief.org 2 University of North Carolina (UNC), shanda@email.unc.edu 3 National University of Lesotho, rmoshoeshoe@gmail.com Funding and support for the implementation of the studies in the three countries were provided by a variety of partners under the Transfer Project and the Protection to Production project, including DfID, FAO, UNICEF, Save the Children UK, Sida, and UNC. We thank all colleagues (there are too many individuals to name here) who made the implementation of these studies possible. We thank Lisa Hjelm in particular for excellent research support. Funding for this particular paper was provided by the United States Department of Labour through Grant Number IL-26694-14-75-K-36 to UNICEF Office of Research- Innocenti. This document does not necessarily reflect the views or policies of the United States Department of Labour, nor does mention of trade names, commercial products, or organizations imply endorsement by the United States Government. 1
1. Introduction Cash transfer programs - social support programs that provide regular income transfers to poor households - have become an important component of social protection strategies in the developing world. An extensive literature examines the effects of cash transfer programs on the daily activities and human capital accumulation of children in developing countries. Key conclusions of this literature are that cash transfer programs: (i) tend to increase children's school participation (e.g. Baird et al., 2014; Fiszbein and Schady, 2009; Saavedra and Garcia, 2012), (ii) tend to have stronger effects on school participation if provided conditional on children’s regular school attendance (e.g. Akresh et al., 2013; Baird et al., 2011, Baird et al., 2014; and Benhassine et al., 2015), and (iii) appear to have, at best, modest effects on learning outcomes (Baird et al., 2014). Much of this literature focuses on Latin America, where most countries implement conditional cash transfer programs - programs that provide benefits subject to beneficiary households’ compliance with specified behavioral requirements, such as children’s regular school attendance. Unlike Latin American countries, over the past decade many Sub-Saharan African countries 1 have started to implement unconditional cash transfer programs as a social protection policy. The households that benefit from these programs are typically poor and tend to have a high ratio of dependents (children or elderly) to working age adults and a high share of children who have lost (or permanently live in the absence of) one or both parents. While many of these programs have improved education outcomes to be among their primary aims, it is important to establish the effects of these programs on children empirically. Because these programs are not conditioned on children’s school participation, households are free to consume the additional income directly or to invest it in productive activities 1 Including but not limited to, Botswana, Ghana, Ethiopia, Kenya, Mozambique, Namibia, Nigeria, Rwanda, South Africa, and Zimbabwe. 2
instead of child education (indeed, as mentioned above, programs that are not conditioned on school attendance tend to have weaker of effects on schooling outcomes). In fact, as the programs open up opportunities for credit constrained beneficiary households to invest in - and hence expand - their household entrepreneurial activities, a concern may be that children are withdrawn from school to engage in productive activities. This concern is particularly pressing for many programs implemented in Sub-Saharan Africa, because beneficiary households typically have few working age adults and may therefore rely on children to exploit new opportunities to invest in household entrepreneurial activities (see also de Hoop and Rosati, 2014, for a discussion of the channels through which a cash transfer program might affect children’s activities). Moreover, although there is strong evidence that the loss of a parent has negative implications for education outcomes (e.g. Gimenez et al., 2013; Beegle et al., 2010; Case and Ardington, 2006; Case et al., 2004), it is not clear a-priori that an income transfer is an appropriate tool to support the schooling outcomes of children who lost a parent. As summarized in Fitzsimons and Mesnard (2014) and explained in more detail in Case et al. (2004), Evans and Miguel (2007), and Gertler et al. (2004), the loss or absence of a parent can affect children’s activities and human capital accumulation for a number of reasons other than income loss, including detrimental “emotional and psychological consequences” (Fitzsimons and Mesnard, 2014, pp: 1) and changes in intra-household decision making dynamics. This paper examines the effects of large-scale, government-administered cash transfer programs on children’s schooling in three countries in Sub-Saharan Africa: Lesotho, Malawi, and Zambia. The paper further examines whether these programs have protective effects on children’s schooling against parental loss. Each of these programs provides regular income 3
support to poor and disadvantaged beneficiary households without requiring compliance with any behavioral conditions (related to schooling or otherwise). The programs either explicitly (in the case of Malawi) or implicitly target labor constrained households: the average child in our sample lives in a household with about 2.3 dependents (children or elderly) per working age adult. And the programs either explicitly (in case of Lesotho and Zambia) or implicitly target households with orphans and vulnerable children. Concomitantly, a large share of the children in our sample lived without their mother (about two in five) or their father (about three in four) at baseline. Each of the three programs was accompanied by an experimental impact evaluation comprising extensive baseline (pre-intervention) and follow-up data collection from beneficiary and comparison households. The evaluations were implemented as part of the Transfer Project, an initiative of Save the Children UK, UNICEF, and the University of North Carolina, evaluating the impact of cash transfer programs in (currently) 7 countries in Sub- Saharan Africa (See also: https://transfer.cpc.unc.edu). These three particular impact evaluations were incorporated in the study because (i) they follow a similar cluster randomized design, (ii) comprise comparable data on children’s activities allowing us to carry out pooled data analysis, and (iii) no peer-reviewed articles on child schooling outcomes had been published using the data from these experiments. 2 We document pronounced program impacts on self-reported school participation of older children (i.e. children in the age range at which school dropout starts). For these children, average school enrollment increases by about 9 percentage points and regular school attendance increases by nearly 6 percentage points. Impacts on their grade 2 There was no pre-registered plan for the analysis presented in this paper. However, analysis of program impacts on schooling and child work outcomes was a key objective of the impact evaluations of all three studies (as documented in the baseline reports available on-line: https://transfer.cpc.unc.edu). Moreover, as we shall show, findings based on pooled data from all three studies are reasonably consistent with findings based on data for the individual countries. The latter makes us confident that the presented findings represent substantive and valid patterns of program impact. 4
Recommend
More recommend
