Multidimensional Inequality Measurement: a Proposal

Two essential intuitions about the concept of multidimensional inequality have been highlighted in the emerging body of literature on this subject: first, multidimensional inequality should be a function of the uniform inequality of a multivariate distribution of goods or attributes across people (Kolm, 1977); and second, it should also be a function of the cross-correlation between distributions of goods or attributes in different dimensions (Atkinson and Bourguignon, 1982; Walzer, 1983). The present paper proposes a general method of designing a wider range of multidimensional inequality indices that also respect both intuitions, and illustrates this method by defining two classes of such indices: a generalization of the Gini coefficient, and a generalization of Atkinson ; s one-dimensional measure of inequality.