Algorithmic Underwriting in High Risk Mortgage Markets
80 Pages Posted: 10 Nov 2023 Last revised: 26 Dec 2023
Date Written: October 14, 2023
We study the effects of a policy that shifted from pure human underwriting to human-augmented algorithmic underwriting for low-credit-score, high-leverage mortgage borrowers. Estimating the bunching of loans around the policy's debt-to-income threshold, we find a large credit expansion to affected borrowers with little changes in default risks or interest rates among the affected group. Such effects are more pronounced among non-Hispanic White borrowers and higher-income borrowers. Consequently, low-credit-score households are more likely to move to better school districts. We use a structural approach to quantify the welfare implications of the policy change and isolate the credit supply channel. Overall, our results suggest that automated underwriting systems (AUS) can help increase financial inclusion while controlling risk. However, it can also generate disparate impact across racial groups and along the income distribution.
Keywords: Algorithmic Underwriting, FinTech, Household Leverage, Racial Inequality in Mortgage Markets, Mobility, Financial Inclusion.
JEL Classification: G18, G21, G51, O33
Suggested Citation: Suggested Citation