Consolidating and improving the assets indicator in the global Multidimensional Poverty Index

This article traces a methodological path for constructing a statistically and normatively validated assets indicator, that in turn can be used within an internationally comparable measure of multidimensional poverty. The article validates a revision to the assets indicator of the global Multidimensional Poverty Index (global MPI) that makes the best possible use of existing data. Our normative focus is shaped by Amartya Sen’s capability approach, framing assets in relation to human activities. But surprisingly few asset items were available for 75 countries and 3.5 billion people, the standard set for indicator revisions, so the paper notes the lack of comparable data, as well as the lack of data on the quality, quantity or gendered ownership of assets. Drawing on empirical antecedents and complemented by normative reasoning, this article uses tetrachoric exploratory factor analysis, multiple correspondence analysis, classical test theory, item response theory and a non-parametric Mokken scale procedure to identify a set of items that proxy asset deprivations. Measures were trialled across 26 purposefully-selected countries. The final counting-based assets index includes nine statistically validated items (adding computer and animal cart to the original global MPI assets indicator). It has higher reliability than other options, and is arguably the most rigorous feasible indicator to compare asset deprivations that can be constructed from existing global MPI data sources. The methodological approach outlined here could be used to design and validate assets indicators within national, regional, or bespoke multidimensional poverty measures elsewhere.