Sur la d\'ecomposabilit\'e empirique des indicateurs de pauvret\'e

TL;DR

This paper investigates the empirical decomposability of poverty indicators, demonstrating that some widely used measures like Sen and Shorrocks are nearly decomposable in practice, enabling effective sectoral poverty analysis.

Contribution

It provides empirical evidence that certain poverty indicators are practically decomposable, supporting their use in sectoral poverty reduction policies.

Findings

Sen and Shorrocks indicators are nearly decomposable in practice

Decomposability error is approximately one to two per thousand

Empirical analysis based on 1996 Senegalese survey data

Abstract

We study the empirical decomposition of poverty indicators. This property is very important and convenient in the context of the fight against poverty. Indeed, it makes it possible to put in place sectoral poverty reduction policies on the basis of a relevant stratification laid down at the outset. The simultaneous impacts of these policies, measured as reduction gains over the population as a whole, is then obtained by aggregating those obtained at each stratum by a relatively simple formula. It turns out that indicators as important as those of Sen and Shorrocks do not verify this property contrary to the elements of the class of Foster - Greer and Thorbecke. Given the data from the 1996 Senegalese Survey of Households (ESAM), we show that the lack of decomposability of these indicators on the income variable for several types of population stratification is practically zero , of the order of one to two per thousand. This makes it possible to use the decomposition of the Sen and Shorrocks indicators without any untoward consequences. An explanatory model of these results is presented for future research.