Counterfactual Decomposition of Pro-Poorness Using Influence Functions

Abstract

Poverty reduction has emerged as a fundamental social objective of development, and has become a metric commonly used to assess the performance of public policy. This paper adapts the methodology of Firpo, Fortin and Lemieux (2009) [2009. “Unconditional Quantile Regressions.” Econometrica 77 (3): 953–973] to the measurement of the pro-poorness of income growth. The method allows the analyst to identify covariates that affect poverty reduction. The methodology is policy-relevant because policy-makers can better target these covariates than the average level of income, or the level of inequality. We demonstrate this by application to Bangladesh 2000–2010.