A Robust Multidimensional Poverty Profile for Uganda

We compute a national multidimensional poverty index (MPI) for Uganda following the approach proposed by Alkire and Foster (2007). Using household survey data, we show how the incidence of multidimensional poverty has fallen in recent years, and we use the decomposability features of the index to explain the drivers of the reduction in multidimensional poverty. We extend the standard application of the MPI to distinguish between domains and dimensions, which is particularly important given the high degree of multiple deprivations within the standard-of-living domain. We also compare the results from Uganda with other countries for which the MPI has been computed, and we note some caveats in such a comparison. The robustness of our estimates is tested in a stochastic dominance framework as well as through statistical inference. Notably, we extend the one-dimensional analysis of stochastic dominance to take into account household size in a second dimension, which is particularly important as some of the MPI indicators are sensitive to the number of household members. By exploiting a unique sub-sample of the integrated household survey programme in Uganda, which has not previously been analysed, we are also able to match the data set used for the MPI with data used to compute the conventional estimates of monetary poverty. This enables a more robust assessment of the complementarities of the two types of poverty measures than has been previously possible.