Quantifying Discrimination in Home Mortgage Lending: Estimation of Loan Price Elasticities Across Products and Races
Abstract
This paper uses individual loan data for a national lender to estimate loan price elasticities across loan products and racial groups. The data contain detail on closing dates and prices, enabling control for market volatility and construction of pseudo-prices for foregone products. The empirical mode is a mixed multinomial and conditional logit which is rooted in a structural utility framework. Asymptotic standard errors for the elasticity estimates are generated with the bootstrap and allow the inference that minorities are nearly perfectly insensitive to price, whereas whites are highly sensitive to price. The racial differences in price elasticities may indicate pre-application discrimination and provide incentive for rational (price) discrimination.They also suggest that empirical assessment of discrimination must confront non-random sorting of borrowers across products to avoid biased estimates of discrimination.