Robust methods for analyzing income distribution, inequality and poverty

Abstract

The analysis of income distribution includes a long list of economic research topics. It is important to study how income is distributed in a population; for example, to determine tax redistribution policies to reduce inequality, or to carry out social policies that lead to poverty reduction. The information available comes mainly from surveys (and not from censuses, as is often believed) and this is the usual cause of long debates about its reliability, because the sources of errors are numerous. Moreover, the form in which the data are available is not always as one would expect, i.e. complete and continuous (microdata). In addition, one may have only data in grouped form (in income classes) and/or truncated data where a portion of the original data has been omitted from the sample, or simply not recorded.

Because of the characteristics of the data, it is important to complement the classical statistical procedures with robust foundations. This paper presents such methods, especially the selection of models, their fitting with various kinds of data, inequality and poverty analysis, as well as ordering tools. One approach is based on the influence function (IF) developed by Hampel (1974), further developed in Hampel, Ronchetti, Rousseeuw & Stahel (1986). It is also shown, through the analysis of real data from Great Britain and Tunisia, that vigorous techniques can give another picture of income distribution, inequality or poverty when compared to classical analyses.