The Time Is Now: Advancing Fairness in Lending Through Machine Learning

68 Pages Posted: 22 Nov 2022

See all articles by Vitaly Meursault

Vitaly Meursault

Federal Reserve Banks - Federal Reserve Bank of Philadelphia

Daniel Moulton

Federal Reserve Banks - Federal Reserve Bank of Philadelphia

Larry Santucci

Federal Reserve Bank of Philadelphia

Nathan Schor

Federal Reserve Banks - Federal Reserve Bank of Philadelphia

Date Written: November 1, 2022

Abstract

Machine learning (ML) models generate credit scores that predict default better overall than traditional models, but they raise concerns from regulators about their effect on protected groups. Recent literature shows that better credit scoring models only marginally improve the relative accuracy of default prediction for historically under-served groups – they do little to overcome these groups’ noisier credit files. We show that by pairing ML models with group-specific thresholds based on protected attributes, such as a consumer’s location, lenders can both improve fairness and increase profits relative to baseline models. Such a process runs contrary to the existing requirement to exclude demographics and geography from lending decisions. However, our results highlight the potential benefit of a public policy framework that relaxes this requirement in a way designed to improve the outcomes for historically under-served groups and simultaneously encourages the adoption of ML in credit underwriting.

Keywords: Credit Scores, Group Disparities, Machine Learning, Fairness

JEL Classification: G51, C38, C53

Suggested Citation

Meursault, Vitaly and Moulton, Daniel and Santucci, Larry and Schor, Nathan, The Time Is Now: Advancing Fairness in Lending Through Machine Learning (November 1, 2022). FRB of Philadelphia Working Paper No. 22-39, Available at SSRN: https://ssrn.com/abstract=4283403 or http://dx.doi.org/10.21799/frbp.wp.2022.39

Vitaly Meursault (Contact Author)

Federal Reserve Banks - Federal Reserve Bank of Philadelphia ( email )

Ten Independence Mall
Philadelphia, PA 19106-1574
United States

Daniel Moulton

Federal Reserve Banks - Federal Reserve Bank of Philadelphia ( email )

Ten Independence Mall
Philadelphia, PA 19106-1574
United States

Larry Santucci

Federal Reserve Bank of Philadelphia ( email )

Ten Independence Mall
Philadelphia, PA 19106-1574
United States
2155746014 (Phone)

HOME PAGE: http://https://www.philadelphiafed.org/our-people/larry-santucci

Nathan Schor

Federal Reserve Banks - Federal Reserve Bank of Philadelphia ( email )

Ten Independence Mall
Philadelphia, PA 19106-1574
United States

Do you have a job opening that you would like to promote on SSRN?

Paper statistics

Downloads
164
Abstract Views
636
Rank
337,393
PlumX Metrics