Priority-Based Multidimensional Poverty

In this paper, we propose a new methodology of multidimensional poverty measurement. Our approach is based on using self-stated responses to household spending priorities. This information is first used to know what the relevant deprivations are. In particular, the stated top priorities allow us to identify the poor. For each household, the deprivations are aggregated according to relative weights computed from the data on priorities. Such procedure can be justified by a set of axioms. Using the priorities data in such a way allows us to: (1) eliminate ‘Command variables’ in favour of ‘Intrinsic welfare variables’, which correspond better to the ‘basic needs approach’; and (2) solves problems arising with needs heterogeneity, (3) avoids the arbitrariness that typically arises in many stages of construction of multidimensional poverty indices. Finally, we propose an empirical application to Seychelles that: (1) elicits what the main deprivation dimensions are, (2) delivers a realistic estimated incidence of poverty, as opposed to what is found with the usual approaches, (3) shows the approach consequences in terms of targeting statistics of using multidimensional poverty rather than monetary poverty. We also use data on public budget to exhibit the ‘implicit multidimensional poverty index’ of the State in Seychelles.