On the measurement of multidimensional poverty in multiple domain contexts

Abstract

13è Meeting of the Society for Social Choice and Welfare. Lund, 28 Juny-1 juliol

We develop the measurement of multidimensional poverty for the case in which the different dimensions taken into account are partitioned in several domains-an issue with crucial implications for the identification and aggregation of the poor which has been neglected in the literature. First, we introduce a general method to identify the poor that makes room for the non-trivial interactions that might exist between dimensions depending on the domains they belong to. Consistent with the former, we then present a new aggregation method that allows for the possibility of having domain-specific elasticities of substitution among pairs of dimensions. Our empirical findings using 48 Demographic and Health Surveys across the developing world suggest that when considering the alternative identification and aggregation methodologies proposed here, the set of households that are identified as poor and the corresponding multidimensional poverty levels can differ to a considerable extent with respect to currently existing approaches.