Inequality Indices as Tests of Fairness

Abstract

Inequality indices are traditionally interpreted as measures of deviations from equality. This article interprets them instead as statistical tests for a null of fairness within well-defined income generating processes. We find that the likelihood ratio (LR) test for fairness versus unfairness within two such processes are proportional to Theil’s first and second inequality indices respectively. The LR values may be used either as a test statistic or to approximate a Bayes factor that measures the posterior probabilities of the fair version of the processes over that of the unfair. We also apply this perspective to measurement of inequality of opportunity.