Estimation and Evaluation of Loan Discrimination: An Informational Approach

Abstract

Many recent studies have analyzed whether lending discrimination exists. In all previous studies, theresearcher faces constraints with the available data or modeling problems. In this article, we use anew informational-based approach for evaluating loan discrimination. Given limited and noisy data, wedevelop a framework for estimating and evaluating discrimination in mortgage lending. This newinformational-based approach performs well even when the data are limited or ill conditioned, or whenthe covariates are highly correlated. Because most data sets collected by bank examiners or banks sufferfrom some or all of these data problems, the more traditional estimation methods may fail to providestable and efficient estimates.This new estimator can be viewed as a generalized maximum likelihood estimator.We provide inferenceand diagnostic properties of this estimator, presenting both sampling experiments and empiricalanalyses. For two of the three banks analyzed, we observe some evidence of potential racial discrimination.