Comment on multi-dimensional indices

Abstract

Do composite indices (that combine various measures of different phenomena)have merit? The three Forum papers in the 2011 June issue of the Journal ofEconomic Inequality are rich in ideas and technically well-argued; they elucidatenicely fundamental conceptual as well as measurement issues.Ravallion [9] argues the merits of the conventional single-dimension incomeor expenditure-based measure of poverty at the household level, as long as it iscomplemented with analysis incorporating non-market-price based dimensions ofpoverty such as schooling and health of household members (the “dashboard” view).Alkire and Foster [1] explain the logic of combining various dimensions of povertyinto a multi-dimensional index, in which the initial identification of who is poor isbased not solely on income or expenditures but on additional dimensions of being“poor”, with transparent weights chosen for all dimensions. Klugman et al. [7]setout the logic of adjustments to the well-known Human Development Index (HDI),and argue the merits of the multi-dimensional HDI despite its inelegant combinationof stocks and flows, and of inputs to well-being (income) and outcomes (schooling).On the measurement of poverty, Ravallion’s dashboard view wins in my reading,if only because it makes tradeoffs based on market prices explicit, and forces theanalyst to make explicit tradeoffs on non-market goods in assessing who is poor. But Inote that Alkire and Foster’s emphasis on the joint distribution of various dimensionsof poverty is fundamental and illustrates the benefits of a single data source atthe household level. Incidentally, I would like to see an assessment of the relativesensitivity of identification of the poor to the World Bank’s conventional measure ofpoverty (often without any accompanying dashboard statistical analysis), comparedto one or another version of the Alkire and Foster approach, for a country overmultiple years, using panel data. Does a multi-dimensional identification do betterat understanding the fundamentals of long-term poverty and low inter-generationalmobility within families?