Assessing the sensitivity of poverty to inequality: A method and an application to Brazil, 1981–2022

This paper proposes an analytical approach to assess the sensitivity of poverty to inequality variations. It introduces a general transform of the Lorenz curve that delivers point Gini elasticities of poverty corresponding to observed distributional changes. This overcomes the limitations of existing methods, due to reliance on rigid and inadequate redistribution processes or dependence on the magnitude of inequality changes. The proposed framework enables pro-poorness benchmarking via extremely regressive and progressive counterfactual redistribution scenarios. A Changes-in-Changes construction is also considered for comparisons across time or space while holding the initial income distribution fixed. Finally, the paper investigates the relationship between inequality changes and the incidence, intensity, and inequality dimensions of poverty. An empirical illustration for Brazil (1981–2022) shows that an increasing share of poverty reduction was associated with inequality variations; inequality fell and poverty declined sharply, notably between 2003 and 2014. Moreover, irrespective of initial conditions, the sensitivity of poverty to inequality increased over the four decades. Yet, distributional change remained below pro-poor benchmark frontiers, reflecting insufficient relative gains for the worst-off.