Food Security and Agriculture in Developing Countries: Measurement and hypotheses for impact evaluations

This paper reviews the challenges related to establishing credible causal links between particular interventions and aggregate food security. A first set of challenges result from the lack of a common measurement of food security, with a multitude of indicators and definitions being used in different studies, making comparisons and broader inferences particularly hard. We discuss various measures and the existing evidence on their validity. We also line out a possible approach to validating some of the multi-dimensional measures in a more comprehensive way. A second set of challenges comes from the need to have credible exogenous variation in order to establish a causal relationship between an intervention and resulting food security outcomes. We review the literature and conclude that the literature to date leaves many open questions regarding the type of interventions that might be most effective to increase food security. This is due in part to the multitude of approaches to measurement of food security, and in part due to methodological concerns that limit causal inference in many of the existing studies. Likely, the optimal policy will also be strongly context-specific, and understanding the sensitivity of impacts to contextual changes hence is equally important.