An absolute multidimensional poverty measure in the functioning space (and relative measure in the resource space): An illustration using Indian data

In this paper we develop a multidimensional poverty measure that attempts to capture absolute poverty in the functioning space. As suggested by Sen, if the measure aims to be absolute in the functioning space, it needs to be relative in the resource space. To generate a relative measure, this measure adapts the poverty cut-off in resource-related indicators in a multidimensional poverty measure to prevailing standards in a region. As illustration, this poverty measure utilizes the Indian Demographic and Health Survey (DHS) and is based on UNDP's global Multidimensional Poverty Index (MPI). Similar to the global MPI, we apply the Alkire-Foster dual cut-off approach (Alkire and Foster, 2011a) and broadly follow the global MPI in the choice of indicators, weights, and overall cut-off. However, adaptable indicator thresholds are considered when appropriate. We argue that global MPI indicators in the health dimension are not open to a relative assessment, as they reflect specific health functionings (i.e. being free from premature mortality and being well nourished). In the education and standard of living dimensions, we set indicator thresholds at the median of the reference population, while experimenting with different reference populations. Empirically we find that the overall ranking of poverty in India does not change using our relative MPI, but the differentials in poverty are substantially smaller between states and rural and urban areas, also depending on the choice of the reference population.