Combining Observational and Experimental Data to Evaluate Adaptive Social Protection Interventions: The Case the Nutrition Improvement through Cash and Health Education (NICHE) in Kenya

Social protection systems have served as an instrument to mitigate the socioeconomic impact of shocks and avoid a sharp decline in living standards. Adaptive social protection aims to strengthen the ability of social protection systems to anticipate, prepare for and respond to shocks while contributing to resilience of poor and vulnerable households to covariate shocks. The United Nations Children’s Fund (UNICEF) works with governments and partners to scale up social protection responses in both one off and protracted crises. One of these responses is the Nutrition Improvement through Cash and Health Education (NICHE) programme, implemented by the Government of Kenya with the support of the UNICEF, integrates nutrition, social protection and child protection into an adaptive cash transfer programme to allow households better prepare to weather-related shocks, such as droughts and floods. In this paper, we systematically combine and analyze experimental and observational survey data to obtain credible estimates of the causal effect of adding weather-related adaptive features to a social protection intervention (NICHE) in Kenya and separately estimate the overall effects of the NICHE programme from observational data. Both observational and experimental samples contain data about a treatment, observable household and individual characteristics, and short-term outcomes focused on children’s well-being. To estimate the effect of the adaptive features of social protection on the child outcomes, we use an A/B test experiment in an experimental sample. Then we use the relationship between the treatment and the secondary outcomes from the experimental sample to estimate the overall effect of NICHE on child outcomes while addressing the potential confounding in the observational sample. The paper showcases a rigorous evaluation strategy that can be implemented in fragile settings and that is not as intrusive to programme implementation and contributes to the understanding of the effects of adaptive interventions on child outcomes and resilience.