Algorithmic Underwriting in High Risk Mortgage Markets

Abstract

We study the effects of a policy that increased reliance on algorithmic underwriting for low-credit-score, high-leverage mortgage borrowers. Using a bunching-based approach, we document a large policy-induced credit expansion among affected borrowers, with little changes in default risks given observables. The credit expansion is larger among White, Hispanic, and higher-income borrowers. Post-policy, low-credit-score individuals are more likely to move to better-rated school districts. A structural approach helps quantify the welfare implications of the policy. Our results suggest a limited role of human discretion for most borrowers in this market and highlight challenges in increasing financial inclusion for certain disadvantaged populations.

Keywords: Algorithmic Underwriting, FinTech, Household Leverage, Racial Inequality in Mortgage Markets, Mobility, Financial Inclusion.

JEL Classification: G18, G21, G51, O33