Counterfactual decomposition of pro-poorness using influence functions

Poverty reduction has emerged as a fundamental objective of development and hence a metric for assessing the aggregate performance of public policy. Declaring a policy outcome pro-poor on the basis of changes in an aggregate indicator may hide more than it reveals about the heterogeneity of impacts underlying the aggregate outcome. This paper demonstrates the use of influence functions to link poverty-focused evaluation functions to individual or household characteristics and perform counterfactual decompositions in order to identify and analyze the endowment and structural effects and their determining factors that ultimately drive pro-poor outcomes. An empirical illustration presents an analysis of the pro-poorness of the growth pattern in Bangladesh in 2000-2010. We find that socioeconomic arrangements in Bangladesh have become more progressive over time.