Congratulations to Alkire and Foster for focusing much needed attention on the issue of multidimensional poverty and, moreover, for advancing the case for a single, multidimensional index to measure deprivation in the developing world. As seemingly most development economists recognize, poverty is more than a lack of income or inadequate consumption, but is composed a host of factors that simultaneously act to constrain capabilities and increase deprivation. Part of the debate around unidimensional or multidimensional metrics plays out something like this: from a policy perspective, if poverty is equated with lack of income, then policies that promote economic growth would appear to be all that are needed to reduce poverty. If, instead, poverty is a multidimensional phenomenon, then, as Kanbur and Grusky have put it, “The task (of remediating multidimensional poverty) …. requires targeting those aspects of inequality and poverty (e.g. residential segregation) that are causal with respect to many outcomes and hence likely to bring about cascades of change (my emphasis).” Our task as researchers and policymakers is to determine which aspects of poverty are causal with respect to many outcomes, and to make those aspects the targets of policy interventions. I do have a few questions, though, for the HDRO, particularly if, as noted in the press release back in June, the MPI will replace the Amartya Sen and Sudhir Anand-developed Human Poverty Index. In some of the literature that came out in June around the launch of the MPI, it was noted that “the MPI fixes weights between countries to enable cross-national comparisons; alongside this we strongly encourage countries to develop national measures having richer dimensions, and indicators and weights that reflect their context as Mexico did and Colombia is doing.” Does this mean that the HDRO will calculate its MPI for country X, while country X may calculate its own in any given year? If they differ, will the HDRO’s calculation be the MPI of record, or will the country’s be? What if country X takes advantage of the MPI’s flexibility in the choice of dimensions and indicators and, due to changes in political leadership, for example, chooses to calculate that country’s MPI with a different array of indicators the following year? What does this do to the ability of researchers to calculate change over time? In this case, would researchers simply resort to the UNDP’s calculation of country X’s MPI? And what about the data sources used to calculate the MPI? Ravallion notes that: “Rather (the indicators) were chosen because the methodology used by the MPI requires that the analyst has all the indicators for exactly the same sampled household. So they must all come from one survey. There is much better data available on virtually all of the components of the MPI, but these better data can’t be used in the MPI since they are only available from different surveys. This aspect of their methodology greatly constrains the exercise.” An advantage that the HPI has over the MPI is that it can be used in the absence of disaggregated data. As I understood it, HDRO statisticians collect new or projected data for each of the HPI’s 4 indicators from one year to the next—from the UN Department of Economic and Social Affairs Population Division’s analysis of national vital registration systems, from UNESCO’s Institute for Statistics, from WHO, etc. From a quick review of OPHI’s country profiles, it appears as though the data for individual country MPIs are drawn either from Demographic and Health Surveys (conducted every 5 years), or from the World Health Surveys (conducted irregularly), or from Multiple Indicator Cluster Surveys (conducted every 5 years or so), or… How will the MPI for any given country be calculated next year, given the infrequency of the surveys on which it depends? For those interested in longitudinal studies of changes in deprivation, how should the MPI be used? It would seem that an argument could be made that the loss in precision is made up for in the HPI’s ability to measure annual changes in deprivation, albeit at an aggregated level. On the question of decomposability, Alkire and Foster tout this feature of the MPI as one of its more significant advantages over other multidimensional indices that rely on aggregated data. However, Sen seems less convinced, noting that when decomposability is insisted on for all possible subgroupings, a basic conceptual problem emerges: " The mathematical form of decomposability has had the odd result of ruling out any comparative perspective (and the corresponding sociological insights), which is, in fact, fatal for both inequality and poverty measurement… It is easy to see why decomposability has such a strong appeal. It is ‘nice’ to be able to ‘break down’ the overall poverty of a total population into poverty in different subgroups of people that make up the total population. It gives, I suppose, some forensic satisfaction in solving a ‘whodunit’ (and by how much respectively)… (However), mathematically the demand that the breakdown works for every logically possible classification has the effect that the only measures of inequality or poverty that survive treat every individual as an island ....” (Sen, in Grusky and Kanbur, eds, Poverty and Inequality, 2006) Again, I applaud Alkire and Foster for bringing attention to the measurement of multidimensional poverty. However, readers should be reminded that theirs is only the latest in a long line of similar efforts, each with strengths and weaknesses. The UNDP may want to consider including the MPI to supplement its array of metrics, rather than to supplant the HPI. They measure different, but still important, things.