Measuring unambiguously pro-poor growth

This paper reports on a failed attempt to replicate work by Son (2004) on the measurement of the pro-poorness of growth. Son promotes a new “poverty growth curve” methodology for measuring the pro-poorness of growth, in part because of its ability to conclusively categorize 82% of the growth spells she examines as either unambiguously pro-poor, not pro-poor, or immiserizing, regardless of the chosen poverty level. Further analysis of her original data set reveals numerous errors in the classifications of the growth spells, such that the proposed methodology can only conclusively categorize the pro-poorness of 62% of the growth spells examined. This brings into question Son’s assertions that her method is “powerful” and “conclusive.” The poverty growth curve method is nevertheless found to have a high degree of conclusive results when it is extended to measure separately the impacts of growth on poverty and on income inequality, something that Son does not test, and yet which can still be of interest to development economists.