Measuring Multidimensional Poverty: Dashboards, Union Identification, and the Multidimensional Poverty Index (MPI)

We analyse three approaches to measuring multidimensional poverty, using a consistent set of data for 10 indicators in 101 developing countries. First we implement a simple dashboard of deprivations in ten indicators. While most dashboards stop there, we next describe the simultaneous deprivations experienced by people which conveys information on their joint distribution, yet fails to identify multidimensional poverty. We then implement a ‘union’ approach to measurement, and identify people as multidimensionally poor if they experience any one or more of the ten deprivations. The resulting Union headcount ratio of poverty is very high and may reflect errors of inclusion. We then implement an intermediary identification approach following Alkire and Foster (2011): the global Multidimensional Poverty Index (MPI). Exploring the censoring process of the intermediary identification, we observe that a Union MPI (or intersection) identification approach does not avoid normative choices as often claimed; rather these are made at the stage of indicator selection, and the identification process can be highly sensitive to these choices. The latter approaches often imply equal weights –which is itself a value judgement made out of the public eye. The global MPI clearly states value judgements, and performs robustness tests for them. The paper thus discusses strengths and challenges of different measurement approaches to multidimensional poverty. 

Citation: Alkire, S. and Robles, G. (2016). 'Measuring multidimensional poverty: Dashboards, union identification, and the Multidimensional Poverty Index (MPI)', OPHI Research in Progress 46a, Oxford Poverty and Human Development Initiative (OPHI), University of Oxford.