Confidence Interval Estimation for Inequality Indices of the Gini Family

In this paper we present some nonparametric bootstrap methods to construct distribution-free confidence intervals for inequality indices belonging to the Gini family. These methods have a coverage accuracy better than that obtained with the asymptotic distribution of their natural estimators, typically the standard normal. The coverage performances of these methods are evaluated for the index R by Gini with a Monte Carlo experiment on samples simulated from the Dagum income model (Type I), which is usually used to describe the income distribution.