Explaining Denials: Adverse Action Codes and Machine Learning in Credit Decisioning

46 Pages Posted: 17 Jun 2022

See all articles by George Krivorotov

George Krivorotov

Government of the United States of America - Office of the Comptroller of the Currency (OCC)

Jeremiah Richey

Office of the Comptroller of the Currency

Date Written: June 10, 2022

Abstract

As ML/AI usage expands to a variety of high-stakes decisioning, there is an increasing need to provide transparent explanations. In credit decisioning, regulation has long stipulated the provision of "Adverse Action Codes" (AACs) explaining denial of credit. However, there is potentially a wide range of acceptable AAC methodologies that could be used. This paper compares AACs derived from four common methods - an axiomatically-backed Shapley-based approach, a Most Points Lost-based approach, a difference from Mean approach, and a Univariate binning approach - based on two XGBoost risk models predicting credit card and mortgage risk respectively. We find that, overall, the Univariate approach deviates the most from the Shapley-based approach, all derived differences are "significant" within our novel placebo testing framework, differences are more pronounced for lower-risk customers on the border of the reject boundary, and the Most Points Lost and Mean approach are less robust to data perturbations than Shapley.

Keywords: Machine Learning Explainability, Credit Risk Modeling, Shapley, Adverse Action Codes

JEL Classification: D81, C4, G5

Suggested Citation

Krivorotov, George and Richey, Jeremiah, Explaining Denials: Adverse Action Codes and Machine Learning in Credit Decisioning (June 10, 2022). Available at SSRN: https://ssrn.com/abstract=4133915 or http://dx.doi.org/10.2139/ssrn.4133915

George Krivorotov (Contact Author)

Government of the United States of America - Office of the Comptroller of the Currency (OCC) ( email )

400 7th Street SW
Washington, DC 20219
United States

Jeremiah Richey

Office of the Comptroller of the Currency ( email )

250 E Street, SW
Washington, DC 20219
United States

Do you have negative results from your research you’d like to share?

Paper statistics

Downloads
140
Abstract Views
731
Rank
372,809
PlumX Metrics