Poverty accounting

This paper proposes a new framework for poverty accounting, that is, the decomposition of poverty into its proximate components. Using aggregated household surveys from 124 countries, we estimate the potential impacts of income growth and redistribution on poverty rates, as well as their actual contributions to poverty reduction over the period from 1981 to 2010. Our fractional response approach shows that the potential impacts are highly non-linear and differ across regions and time. This non-linearity and variation need to be taken into account if empirical estimates are to inform development policy. Although historically growth has played the main role in poverty reduction, we find that initial inequality is a strong moderator of the impact of growth. In fact, there has been a shift towards pro-poor growth around the turn of the millennium, both at the $2 and at the $1.25 a day poverty line. Nevertheless, our projections of poverty rates until 2030 show that the end of extreme poverty within a generation, as put forth in the Sustainable Development Goals, is unlikely to materialize.