EXPLAINABLE MACHINE LEARNING IN CREDIT RISK MANAGEMENT:IMPLICATIONS FOR FINANCIAL DECISION-MAKING AND REGULATORY COMPLIANCE
Keywords:
Credit Risk Modeling, Explainable Artificial Intelligence, Algorithmic Fairness, Machine Learning inFinanceAbstract
Financial decision-making in contemporary lending institutions are based on correct credit risk modelling where the selfdefault probability of the borrower and decision rules for lending decision are been determined. Traditional credit scoringmodels allow for transparency but have difficulties in extrapolating complex non-linear relationships from the borrowerfinancial data. Even though machine learning models may be able to bring a better predictive performance, their inabilityto explain its results and their potential for being unfair are a challenge in terms of regulatory enforcement, andresponsible lending. This study develops an integrated framework in a credit risk modeling approach, machine learningprediction, explainable artificial intelligence, fairness diagnosing, calibration analysis and policy-based thresholdevaluation. Using the Lending Club loan dataset three models were implemented for the prediction of default: Logisticregression, decision tree and XGBoost. Empirical results show thatthe XGBoost model showed the best predictive powerwith a ROC-AUC of 0.7139, the score achieved by logistic regression and decision trees was 0.7011 and 0.6883respectively. Explainability analysis offers the following important drivers of credit risk-interest rate, loan term, debt toincome, and loan amount. Fairness diagnostics illustrate differences in the rate of approvals of borrowers in differentincome groups, suggesting the need of the responsible model governance. On the whole the suggested framework can beconsidered a clear and policy implementable method of credit risk modelling.
Downloads
References
Bhatore, S., Mohan, L., & Reddy, Y. R. (2020).Machine learning techniques for credit risk evaluation: a systematic
literaturereview.Journal of Banking and Financial Technology,4(1), 111-138.
Bono, T., Croxson, K., & Giles, A. (2021).Algorithmic fairness in credit scoring.Oxford Review of Economic
Policy,37(3), 585-617.
Bussmann, N., Giudici, P., Marinelli, D., & Papenbrock, J. (2020).Explainable AI in fintech risk
management.Frontiers in Artificial Intelligence,3, 26.
Bussmann, N., Giudici, P., Marinelli, D., & Papenbrock, J. (2021).Explainable machine learning in credit risk
management. Computational Economics, 57(1), 203-216.
Chen, Y., Calabrese, R., & Martin-Barragan, B. (2024).Interpretable machine learning for imbalanced credit scoring
datasets.European Journal of Operational Research,312(1), 357-372.
Dumitrescu, E., Hué, S., Hurlin, C., & Tokpavi, S. (2022). Machine learning for credit scoring: Improving logistic
regression with non-linear decision-tree effects.European Journal of Operational Research,297(3), 1178-1192.
George, N. (2019).All Lending Club loan data[Data set]. Kaggle.
https://www.kaggle.com/datasets/wordsforthewise/lending-club
Guidotti, R., Monreale, A., Ruggieri, S., Turini, F., Giannotti, F., & Pedreschi, D. (2018).A survey of methods for
explaining black box models.ACM computing surveys (CSUR),51(5), 1-42.
Kannan, K., Lohia, P., Martino, J., Mehta, S., Mojsilovic, A., Nagar, S., ...& Zhang, Y. AI Fairness 360: An extensible
toolkit for detecting and mitigating algorithmic bias.
Karimi, A. H., Barthe, G., Balle, B., & Valera, I. (2020, June).Model-agnostic counterfactual explanations for
consequential decisions. InInternational conference on artificial intelligence and statistics(pp. 895-905). PMLR.
Kozodoi, N., Jacob, J., & Lessmann, S. (2022). Fairness in credit scoring: Assessment, implementation and profit
implications.European Journal of Operational Research,297(3), 1083-1094.
Liu, W., Fan, H., & Xia, M. (2022).Credit scoring based on tree-enhanced gradient boosting decision trees.Expert
Systems with Applications,189, 116034.
Lundberg, S. M., & Lee, S. I. (2017).A unified approach to interpreting model predictions.Advances in neural
information processing systems,30.
Moscato, V., Picariello, A., & Sperlí, G. (2021). A benchmark of machine learning approaches for credit score
prediction.Expert Systems with Applications,165, 113986.
Ribeiro, M. T., Singh, S., & Guestrin, C. (2016, August)." Why should i trust you?"Explaining the predictions of any
classifier. InProceedings of the 22nd ACM SIGKDD international conference on knowledge discovery and datamining(pp. 1135-1144).
Rudin, C. (2019). Stop explaining black box machine learning models for high stakes decisions and use interpretable
models instead.Nature machine intelligence,1(5), 206-215.
Shi, S., Tse, R., Luo, W., D’Addona, S., & Pau, G. (2022).Machine learning-driven credit risk: a systemic
review.Neural Computing and Applications,34(17), 14327-14339.
Szepannek, G., & Lübke, K. (2021). Facing the challenges of developing fair risk scoring models.Frontiers in
artificial intelligence,4, 681915.
Ustun, B., Spangher, A., & Liu, Y. (2019, January).Actionable recourse in linear classification. InProceedings of the
conference on fairness, accountability, and transparency(pp. 10-19).
Yang, K., Yuan, H., & Lau, R. Y. (2022). PsyCredit: An interpretable deep learning-based credit assessment approach
facilitated by psychometric natural language processing.Expert Systems with Applications,198, 116847.
