The Sensitivity Functional Forms of Earnings: Earnings Functions to Specification

Abstract

The purpose of this paper is to examine the nature of the latent component in the distribution of earnings. The distribution of earnings can be deconvoluted into a latent and an observable component. The observed component is generated by those measurable characteristics of the individual that are interpreted as indicators of productivity. The latent component is generated by misinformation, discrimination and ignorance on the part of the employer. In this paper we employ the distribution of the residuals of a series of Mincer type equations as the proxy for the distribution of the latent component in earnings. By using various specifications and by computing distributions across occupations we are able to remove the confounding influence of potential specification bias. The residuals are examined using the Bera-Jargue test for normality and Elderton’s Kappa criterion. The results are reported for 34 occupations using the 1990 CPS data.