Working Paper 2018-457
A policy intervention entails a distribution of social resources to a target population in order to solve a particular problem they face. In that context, targeting is designed to assign treatment according to some eligibility criterion. While effective targeting is not an end in itself, it is central to both fairness and cost-effectiveness in policymaking. Decision makers and other stakeholders would therefore like to know the extent to which targeting respects eligibility. Answering this key question is essentially an exercise in associational inference. This paper proposes an evaluative framework for assessing targeting outcomes. The framework focuses on measuring, judging and explaining targeting performance. Among indicators of performance considered, the conditional probability of assignment emerges as a local measure which is more informative than global indicators. The paper advocates for the use of chance-corrected measures of interobserver agreement to judge the extent of agreement between assignment mechanisms, and demonstrates the use of bivariate probit regression analysis to identify proximate determinants of targeting outcomes. The proposed framework is applied to the evaluation of the targeting of cash transfers in the context of the Social Safety Nets Pilot Project (SSNPP) in Northern Cameroon.
Authors: B. Essama-Nssah.