Do Safety Net Transfers Improve Household Diets and Reduce Undernutrition? Evidence from Rural Ethiopia
48 Pages Posted: 9 May 2018 Last revised: 12 Jul 2018
Date Written: April 9, 2018
In this paper we examine the impact of the Ethiopia’s Productive Safety Net Program on household dietary diversity and child nutrition using both waves of the Ethiopian Socio-economic Survey. For identification, we use various methodologies. To estimate the effect of the program on household dietary diversity, we rely on the exogeneity of the change in the amount of money that kebeles (lowest administrative unit) have available to allocate among program beneficiaries, which depends on donor support. We present evidence that there is a discrete jump in the kebeles’ allocated budget between 2012 and 2014. We use the change in the amount of PSNP transfers in each kebele as an instrument for the change in the amount of the transfer received by each household. For robustness, we confirm our results using generalized propensity score matching with a continuous treatment. We find no effect of an increase in the amount of money received by households in the form of PSNP transfers on household dietary diversity. To examine the effect of PSNP participation on long-term child nutrition we use a difference-in-difference approach. We use children aged 6 to 24 months in 2012 as a baseline. The treatment group is children in beneficiary households between the ages of 6 and 24 months in 2014 because they were not born during the 2012 round of the survey, and the control group were children in the same age range in non-beneficiary households. We find no effect on height-for-age regardless of age cohort, model specification, or methodology. Results indicate consistently that PSNP has not had the desired effect on household dietary diversity or child nutrition, suggesting that perhaps the transfers need to be paired with additional interventions such as information about nutrition.
Keywords: Nutrition security, Dietary diversity, Impact, Continuous treatment, Dose-response function, Propensity Score Matching
Suggested Citation: Suggested Citation