Balancing Act: Correcting for Data Imbalance in Bank Failure Prediction

52 Pages Posted: 13 May 2023

See all articles by Rebel A. Cole

Rebel A. Cole

Florida Atlantic University

Jon Taylor

Florida Gulf Coast University - Department of Accounting, Finance & Business Law

Date Written: January 13, 2023

Abstract

Researching the determinants of bank failure is an important task, yet the extant literature on bank failure early warning models fail to identify which model technique, sampling methodology, or set of coefficients provides the most accurate model when predicting failure on out-of-sample data. In this paper, we examine previously published studies on bank failure prediction to determine with statistical significance which among the chosen set is most accurate. We also examine the effects of bias-adjusting models from the Machine Learning literature to determine if bias-correcting sampling algorithms improve accuracy. Using oversampling techniques to correct for the inherent bias present in bank failure prediction datasets, we examine the predictive accuracy of gradient boosting decision tree models and show with statistical significance that machine learning techniques can improve out-of-sample predictive accuracy for early warning bank failure models. We find that bias-adjusting techniques do improve accuracy, primarily among healthy banks.

Keywords: banking crisis, bank failure, bank supervision, CAMELS, failure prediction, financial crisis, logit model, offsite monitoring

JEL Classification: G17, G21, G28

Suggested Citation

Cole, Rebel A. and Taylor, Jon, Balancing Act: Correcting for Data Imbalance in Bank Failure Prediction (January 13, 2023). Available at SSRN: https://ssrn.com/abstract=4446836 or http://dx.doi.org/10.2139/ssrn.4446836

Rebel A. Cole

Florida Atlantic University ( email )

College of Business
777 Glades Road
Boca Raton, FL 33431
United States
1-561-297-4969 (Phone)

HOME PAGE: http://rebelcole.com

Jon Taylor (Contact Author)

Florida Gulf Coast University - Department of Accounting, Finance & Business Law ( email )

Ft. Myers, FL 33965-6565
United States
239-590-7316 (Phone)

HOME PAGE: http://www.fgcu.edu

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

Paper statistics

Downloads
36
Abstract Views
163
PlumX Metrics