Multidimensional poverty indices

This paper explores the axiomatic foundation of multidimensional poverty indices. Departing from the income approach which measures poverty by aggregating shortfalls of incomes from a pre-determined poverty-line income, a multidimensional index is a numerical representation of shortfalls of basic needs from some pre-specified minimum levels. The class of subgroup consistent poverty indices introduced by Foster and Shorrocks (1991) is generalized to the multidimensional context. New concepts necessary for the design of distribution-sensitive multidimensional poverty measures are introduced. Specific classes of subgroup consistent multidimensional poverty measures are derived based on sets of reasonable axioms. This paper also highlights the fact that domain restrictions may have a critical role in the design of multidimensional indices. Received: 1 March 1999/Accepted: 1 August 2000