An Alternative Formulation of the Esteban-Gradín-Ray Extended Measure of Polarization

Measuring the extent to which distribution is polarized has recently attracted some attention from economists. Polarization, though related to inequality, is a different concept and implies a tendency towards clustering of observations into groups. It is of interest because it may be related to the generation of tensions and social unrest. In independent studies, Esteban and Ray (1994) and Wolfson (1994) conceptualized the notion of polarization and developed the corresponding measures. Since their papers, many other indices have been proposed. Particularly, in a recent paper in this journal, Grad´ın (2002) uses an extension of the Esteban and Ray measure. In this paper we point out some problems with this extended measure of polarization and explore an alternative formulation for the extension so as to solve the problem identified. An illustration of the use of these tools taking data from 22 Latin American countries shows the possibilities of this new measure and how polarization and inequality can often differ in practice.