ROBUSTNESS PROPERTIES OF INEQUALITY MEASURES

Inequality measures are often used to summarize information about empi rical income distributions. However the resulting picture of the distr ibution and of changes in the distribution can be severely distorted i f the data are contaminated. The nature of this distortion will in gen eral depend upon the underlying properties of the inequality measure. We investigate this issue theoretically using a technique based on the influence function, and illustrate the magnitude of the effect using a simulation. We consider both direct nonparametric estimation from th e sample, and indirect estimation using a parametric model; in the lat ter case we demonstrate the application of a robust estimation procedu re. We apply our results to two micro-data examples.