A Profit-Based Measure of Lending Discrimination

2025-12-23

AI Summary

This paper addresses implicit discrimination in loan pricing algorithms—specifically, systematic preferential treatment of male and Black borrowers—by proposing a novel fairness metric grounded in actual lending profitability differentials across demographic groups. Method: Leveraging approximately 80,000 personal loan records from a U.S. fintech platform, the study integrates financial econometrics, credit risk modeling, and fairness auditing frameworks. It employs empirical regression analysis and counterfactual calibration to assess model behavior under varying inclusion/exclusion of protected attributes (e.g., gender, race). Contribution/Results: Contrary to conventional “fairness through unawareness,” excluding protected attributes exacerbates miscalibration: the model significantly overestimates default risk for women and underestimates it for Black borrowers. In contrast, controlled incorporation of sensitive attributes enables effective risk recalibration, eliminating intergroup profit disparities. This work pioneers the use of real-world commercial profit—not just predictive accuracy or statistical parity—as the benchmark for fairness evaluation, demonstrating that sensitive attributes are both necessary and corrective for achieving risk-consistent fairness.

Technology Category

• Machine Learning: Ethics, Bias, and Fairness
• Natural Language Processing: Ethics — Bias, Fairness, Transparency & Privacy
• Computer Vision: Bias, Fairness & Privacy

Application Category

• Economics, Online Markets and Human Computation: Fairness, privacy, and diversity in economic environments
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Social Networks and Social Media: Fairness and bias in social network and social media analysis

Abstract

Algorithmic lending has transformed the consumer credit landscape, with complex machine learning models now commonly used to make or assist underwriting decisions. To comply with fair lending laws, these algorithms typically exclude legally protected characteristics, such as race and gender. Yet algorithmic underwriting can still inadvertently favor certain groups, prompting new questions about how to audit lending algorithms for potentially discriminatory behavior. Building on prior theoretical work, we introduce a profit-based measure of lending discrimination in loan pricing. Applying our approach to approximately 80,000 personal loans from a major U.S. fintech platform, we find that loans made to men and Black borrowers yielded lower profits than loans to other groups, indicating that men and Black applicants benefited from relatively favorable lending decisions. We trace these disparities to miscalibration in the platform's underwriting model, which underestimates credit risk for Black borrowers and overestimates risk for women. We show that one could correct this miscalibration -- and the corresponding lending disparities -- by explicitly including race and gender in underwriting models, illustrating a tension between competing notions of fairness.

Problem

• Develops a profit-based measure to detect lending discrimination in algorithmic underwriting
• Analyzes disparities in loan profitability across demographic groups using fintech data
• Examines tension between fairness notions by including protected attributes in models

Innovation

• Profit-based measure for lending discrimination
• Audit algorithm using miscalibration in risk estimation
• Include protected attributes to correct disparities