On the issue of application of the Gini coefficient and other inequality indices

Abstract

This article discusses a problem posed by G.L. Gromyko and I.L. Matyukhina in their publication «On the use of the Gini index in economic and statistical studies» (Voprosystatistiki, 2015, no. 9), and states a viewpoint regarding calculation of the Gini coefficient with the use of average, relative, and some other values. As the author points out, the Gini coefficient is merely one of many inequality measures; thus the statements regarding the Gini coefficient equally well relate to other indices having the same essence (e.g. the Theil index, Atkinson index, Herfindahl-Hirschman index, Hoover index, etc.). The author states that the criticism of the application of the Gini coefficient bases on fallacious premises, first of all, on erroneous interpretation of the notion of distribution in statistics. The distribution is interpreted as a way of dividing a good (with positive or negative utility) between some units (individuals, firms, regions, etc.). However, the statistics uses a different notion of distribution, empirical probability distribution. Besides, in author’s sight, the publication under consideration absolutizes the interpretation of the Gini coefficient as ‘concentration ratio’ and its connection with the Lorentz curve. The article provides specific examples that illustrate the adequacy of the use of the Gini coefficient and similar indices calculated from average and relative values for measuring inequality (differentiation, unevenness, dispersion, etc.).