Polarisation and Poverty Reduction in Africa: The Devil Is in the Choice of Equivalence Relation

Abstract

The focus of this paper is the low observed mean consumption elasticity of poverty in Africa, and the suggestion that polarisation of national distributions, specifically the non-parametric ‘relative distribution’ method, is essential to understand the low regional elasticity. The version of the methodology adopted results in a measure of absolute polarisation. We show that the results obtained for 24 countries in the region are entirely a product of this choice, and while preference for translation invariance is a normative matter, claims regarding changes in distributions are not. There is no evidence of distributional changes unaccounted for by standard measures of inequality and mean consumption. These variables also explain the evolution of poverty levels in the 24-country sample. Given that changes in mean consumption and inequality account for both the changes in the chosen measure of polarisation and the evolution of poverty, there is no distinct role for the chosen measure of polarisation in accounting for the evolution of poverty in the region.