Blog Post

Social protection, household size, and its determinants: Evidence from Ethiopia

A recent working paper from IFPRI’s Ethiopia Strategy Support Program (ESSP) discusses how public policies, specifically those related to social protection interventions, may induce changes in household size or structure and how, in turn, these possibly unintended changes may impact the welfare consequences of the policies themselves.

Approximately one billion people benefit from social protection interventions worldwide, both in developed and in developing countries. These programs can affect household size in a variety of ways: fertility; child fosterage; employment patterns, including migration; marriage; and divorce and household dissolution. However, most of the literature to date only focuses on one single such effect, such as fertility; in addition, most of the studies done thus far focus only on programs in countries where fertility rates are already low or where child fosterage and the early marriage of girls is largely absent.

The paper tries to fill in some of these gaps by looking at Ethiopia’s Productive Safety Net Program (PSNP), a social protection intervention that includes a public works program enacted in Ethiopia, a developing country with high fertility rates.

The PSNP was created in 2002 in response to a major drought that caused more than 13 million people in Ethiopia to become reliant on emergency food aid. This crisis caused the government to look at ways to reduce general vulnerability to food shortages due to rainfall shocks. The PSNP targets particular woredas that are historically drought-prone and that have been previous recipients of food aid.

Participating households are selected based on specific criteria, such as being community members, facing continuous food shortages over a number of years, being acutely food-insecure due to a shock resulting in severe loss of assets, and lacking family support or social protection.  Each household’s assets (income, land, and livestock) are also assessed, as are their demographic characteristics, such as female-led, elderly, or disabled households. Although the program uses progressive targeting, the PSNP program budget is not large enough to include all eligible families; in addition, household size and the presence of young children are not considered when households are selected for the program.

The PSNP payments take the form of food and cash, with most beneficiaries receiving payments in return for undertaking public works. These public works, such as soil and water conservation projects or the building and maintenance of feeder roads, usually occur from January through July, outside of the growing season, when labor opportunities are low. There is also a Direct Support component of the program that is directed toward the elderly and disabled; under this component, payments are given without participation in public works. According to the ESSP study, historic payment levels have been consistently lower than what was set out in the program’s Implementation Manual, especially the Direct Support portion.

Data for the ESSP study comes from bi-annual PSNP surveys collecting longitudinal data from beneficiaries and non-beneficiaries from the same location. The ESSP study specifically uses four surveys from 2006 to 2012. The surveys are collected during the same time each year and include identical questions in each round that pertain household demographics, as well as a set of control variables.

The report presents a conceptual framework covering economic models that have been used to examine household demographics in previous studies.  According to the authors, several features of these models are relevant when looking at how PSNP may influence household size. First, some studies emphasize that as incomes rise, parents may be more likely to invest more in child quality (i.e. children’s health and education) and less in child quantity (having more children). Other studies suggest that if program benefits are tied to household size, PSNP payments may be seen as an incentive for households to have more children in order to receive higher benefits; however, the ESSEP study points out that, in practice, PSNP payments typically do not increase as household size rises. In addition, children under the age of two are not taken into consideration when calculation households’ PSNP entitlements.

The conceptual framework also highlights that social protection programs may influence when people choose to have children. The PSNP allows pregnant and lactating women to temporarily shift from the public works arm of the program to the direct support arm, thus reducing the likelihood that having children will result in lost income for a household. Because of this, and because households tend to move in and out of the PSNP over time as their economic situation changes, households may choose to reduce the spacing between their children and have children while they are PSNP beneficiaries in order to minimize income loss due to pregnancy.

Finally, some models may suggest that participation in the PSNP may help reduce the rural-urban wage gap, thus reducing out-migration from rural households and even encouraging urban workers to return to rural areas. While this pertains mostly to adults, it is possible that the PSNP may have similar effects for children; the additional income generated from the PSNP may make households less likely to send their children to other households to be fostered out of economic necessity or may even make households more likely to take in foster children from other areas.

These are all some theoretical ways in which the PSNP may influence household size and movement. How does these factor play out in the ESSP study? Using a difference-in-difference estimator, the study’s findings suggest that households participating in the public works component of PSNP undergo an increase in household size by 5.3 percent. Disaggregating this finding by sex, it appears that program participation results in an increase of 0.176 women (significant at the 5 percent level); the rise in men (0.095) is not statistically significant. Further disaggregating by age, the study finds that the increase in women in PSNP households stems entirely from an increase in adolescent girls (ages 12-18 years). Looking at the reasons behind this increase (either increased in-migration or reduced out-migration), the study finds that in households participating in the public works arm of the PSNP, the likelihood of girls aged 12-18 years leaving the household is reduced by 4.6 to 5.7 percentage points; there do not appear to be any effects on in-migration from this age group.

Thus, according to this study, the increased household size found in PSNP-participating households is due to solely to an increase in the number of adolescent girls (aged 12-18 years) staying in the home. Looking at PSNP survey records, which identify reasons why household members leave or enter the household, the survey authors find that for females, the most important factor driving out-migration is marriage. The authors posit that participation in the PSNP may cause households to delay the marriage of adolescent girls for two reasons. First, as noted previously, increased income may lead parents to invest more in their children’s education; thus, the PSNP may provide the opportunity for parents to afford to continue sending their female children to school into adolescence. Second, adolescent girls may be needed more at home to assist with household tasks while the older adults are participating in the public works portion of the program.

The ESSP study highlights the complexity of social protection programs like the PSNP and points to the need to consider multiple possible changes driven by these programs. Examining only the program’s effects on household size, for example, will not tell the whole story; to truly understand the impact of the PSNP on household demographics, it is necessary to dig deeper into the determinants of those effects.

By: Jenn Campus