Osservazioni sull'intervallo di confidenza dell'indice di concentrazione di Gini, proposto da Gastwirth et al.

Abstract

This article points out that the interval of confidence for the Gini concentration ratio —proposed by Gastwirth et al.—is an asymptotically ultra-conservative confidence interval. It has in fact been shown the probability of the confidence interval tends asymptotically to one. This result has been empirically checked by using the methodology based on the simulation of the sample space. The simulations carried out on three models for income distribution that is Pareto, Log-normal and Dagum, allowed us to remark that the number of samples required to verify the asymptotic convergence increases with the increase of the number of income classes into which the sample is divided. The number of samples in which the confidence level of the interval examined was equal to one is considerably lower than the numbers usually considered in real analysis of income distribution.