Multidimensional Poverty and Its Discontents

Abstract

More data on non-income dimensions of poverty are available than at any previous time in history. Alongside this, multidimensional measurement methodologies have advanced considerably in the past fifteen years. These advances together have created new possibilities to measure multidimensional poverty at the local, national and international level. Yet the fact that overall measures can be constructed does not mean that they will necessarily add value. This paper focuses on the question of when, how and why certain multidimensional poverty measures add value, sketches the limits of the contribution, and introduces a set of standing questions. The key value-added of a rigorously implemented multidimensional poverty index is that it conveys additional information not captured in single-dimensional measures (or in a dashboard) on the joint distribution of disadvantage and the composition of poverty. It also provides a consistent account of the overall change in multidimensional poverty across time and space. The paper discusses the joint distribution approach to multidimensional poverty measurement and presents one class of poverty measures within this approach. It then introduces one recently implemented measure within this family: the 104-country Multidimensional Poverty Index 2010 and uses concrete examples to explain its construction further. For example, without weights one can only identify the multidimensionally poor by the union or the intersection approaches; by these approaches the 2010 MPI would have identified an average of 58% or 0% of people across the 104 countries as poor. It also shows how to ‘unfold’ the MPI by sub-group or dimension, and also by intensity – because similar ‘intensities’ of poverty can conceal different distributions of intensity among the poor. Pointing out the added value of multidimensional poverty indexes is not to suggest that single-dimensional measures be abandoned but rather supplemented. Investing further in multidimensional measures has the potential to generate significant advances in understanding and useful policy tools.