The estimation of the growth and redistribution components of changes in poverty: a reassessment

What are the respective contributions of growth and inequality changes to observed poverty variations? Many studies have attempted to provide some empirical evidence to answer this question using case studies with decompositions of observed poverty spells. Most of them rely on two decomposition frameworks suggested by Datt & Ravallion (1992) on the one hand, and Shorrocks (1999) and Kakwani (2000) on the other hand. However, despite their properties, these techniques are not appropriate for such an accounting exercise. Here, following Muller (2006), we propose an alternative decomposition procedure that is consistent with definitions of growth and inequality effects stemming from time-integral calculus. Contrary to the aforementioned methods, the proposed technique simultaneously fits the observed pattern of income distributions changes and does not produce large residual components.