近年來，全球金融市場面臨諸多挑戰，大型銀行倒閉事件的重演引起重視。為了提前識別潛在的倒閉風險，本研究採用可解釋機器學習方法進行銀行倒閉預測並深入分析。本文選擇了懲罰型邏輯斯迴歸、隨機森林和極限梯度提升(eXtreme Gradient Boosting, 以下簡稱XGBoost)三種模型，對不同時間跨度(前一年T-1Y與前兩季T-2Q)的數據集進行分析，探討各財務變數對銀行倒閉預測的影響力。透過變數重要性和SHAP分析，本研究辨別了哪些變數對於銀行倒閉有顯著影響，並評估了這些變數與倒閉風險之間的正、負關聯性。研究結果表明，隨機森林模型在所有測試中表現最為出色，特別是在使用T-1Y資料集時，能夠展現出高度的預測準確性，而資本適足性及經營品質相關變數對於判別倒閉最為重要。此外，本研究的發現對於銀行風險管理與監理機構亦具有實務意義，有助於輔助其更早地識別潛在的倒閉風險並優化風險管理策略。

In recent years, the global financial market has faced numerous challenges, with the recurrence of major bank failures drawing significant attention. To proactively identify potential risks of failure, this study employs explainable machine learning methods to predict and thoroughly analyze bank failures. This paper selects three models: Penalized Logistic Regression, Random Forest, and XGBoost, to analyze datasets over different time spans (T-1Year and T-2Quarter), investigating the impact of various financial variables on bank failure prediction. Through feature importance and SHAP analysis, this research identifies variables that significantly influence the risk of bank failure and evaluates the positive or negative correlations between these variables and the risk of failure. The results indicate that the Random Forest model outperforms all others in every test, especially when using the T-1Y dataset, demonstrating high predictive accuracy. Variables related to Capital adequacy and Management quality are found to be the most critical in distinguishing failures. Moreover, the findings of this study are of practical significance to bank risk management and regulatory bodies, assisting them in identifying potential risks of failure earlier and refining risk management strategies.

Abedifar, P., Molyneux, P., & Tarazi, A. (2018). Non-interest income and bank lending. Journal of Banking & Finance, 87, 411-426.

Altman, E. I., Haldeman, R. G., & Narayanan, P. (1977). ZETATM analysis A new model to identify bankruptcy risk of corporations. Journal of Banking & Finance, 1(1), 29-54.

Avkiran, N. K., & Cai, L. C. (2012). Predicting bank financial distress prior to crises. New Zealand Finance Colloquium.

Beaver, W. H. (1966). Financial ratios as predictors of failure. Journal of Accounting Research, 71-111.

Bellovary, J. L., Giacomino, D. E., & Akers, M. D. (2007). A review of bankruptcy prediction studies: 1930 to present. Journal of Financial Education, 1-42.

Berger, A. N., & Bouwman, C. H. (2013). How does capital affect bank performance during financial crises? Journal of Financial Economics, 109(1), 146-176.

Betz, F., Oprică, S., Peltonen, T. A., & Sarlin, P. (2014). Predicting distress in European banks. Journal of Banking & Finance, 45, 225-241.

Board of Governors of the Federal Reserve System. (1996, December 27). Uniform Financial Institutions Rating System. Retrieved March 12, 2024 from https://www.federalreserve.gov/boarddocs/srletters/1996/sr9638.htm

Boyacioglu, M. A., Kara, Y., & Baykan, Ö. K. (2009). Predicting bank financial failures using neural networks, support vector machines and multivariate statistical methods: A comparative analysis in the sample of savings deposit insurance fund (SDIF) transferred banks in Turkey. Expert Systems with Applications, 36(2), 3355-3366.

Breiman, L. (2001). Random forests. Machine learning, 45, 5-32.

Breiman, L., Friedman, J., Olshen, R., & Stone, C. (1984). Classification and Regression Trees(1st ed.). Chapman and Hall/CRC. https://doi.org/https://doi.org/10.1201/9781315139470

Calmès, C., & Théoret, R. (2015). Product-mix and bank performance: new US and Canadian evidence. Managerial Finance, 41(8), 773-805.

Canbas, S., Cabuk, A., & Kilic, S. B. (2005). Prediction of commercial bank failure via multivariate statistical analysis of financial structures: The Turkish case. European Journal of Operational Research, 166(2), 528-546.

Carmona, P., Climent, F., & Momparler, A. (2019). Predicting failure in the US banking sector: An extreme gradient boosting approach. International Review of Economics & Finance, 61, 304-323.

Chen, T., & Guestrin, C. (2016). Xgboost: A scalable tree boosting system. Proceedings of the 22nd acm sigkdd international conference on knowledge discovery and data mining, 785-794.

Cihak, M. M., & Poghosyan, M. T. (2009). Distress in european banks: an analysis basedon a new dataset. International Monetary Fund.

Cole, R. A., & Gunther, J. W. (1998). Predicting bank failures: A comparison of on-and off-site monitoring systems. Journal of Financial Services Research, 13(2), 103-117.

Cole, R. A., & White, L. J. (2012). Déjà vu all over again: The causes of US commercial bank failures this time around. Journal of Financial Services Research, 42, 5-29.

Cox, D. R. (1958). The regression analysis of binary sequences. Journal of the Royal Statistical Society Series B: Statistical Methodology, 20(2), 215-232.

Davis, J., & Goadrich, M. (2006). The relationship between Precision-Recall and ROC curves. Proceedings of the 23rd international conference on Machine learning, 233-240.

Federal Deposit Insurance Corporation. (2024, March 12). 12 CFR Part 327. Retrieved March 18, 2024 from https://www.ecfr.gov/current/title-12/part-327

Friedman, J. H. (2001). Greedy function approximation: a gradient boosting machine. Annals of statistics, 1189-1232.

Gogas, P., Papadimitriou, T., & Agrapetidou, A. (2018). Forecasting bank failures and stress testing: A machine learning approach. International Journal of Forecasting, 34(3), 440-455.

Huang, J., & Ling, C. X. (2005). Using AUC and accuracy in evaluating learning algorithms. IEEE Transactions on knowledge and Data Engineering, 17(3), 299-310.

Kuhn, M., & Johnson, K. (2013). Applied predictive modeling (Vol. 26). Springer.

Lopez, L., & Delouya, S. (2024, April 26). FDIC says Republic First Bank is closed by Pennsylvania regulators. Cable News Network. Retrieved June 3, 2024 from https://edition.cnn.com/2024/04/26/business/regulators-seize-republic-first-bancorp/index.html

Lundberg, S. M., & Lee, S.-I. (2017). A unified approach to interpreting model predictions. Advances in neural information processing systems, 30.

Manthoulis, G., Doumpos, M., Zopounidis, C., & Galariotis, E. (2020). An ordinal classification framework for bank failure prediction: Methodology and empirical evidence for US banks. European Journal of Operational Research, 282(2), 786-801.

Martin, D. (1977). Early warning of bank failure: A logit regression approach. Journal of Banking & Finance, 1(3), 249-276.

Mayes, D. G., & Stremmel, H. (2012). The effectiveness of capital adequacy measures in predicting bank distress. 2013 Financial Markets & Corporate Governance Conference.

Mehmood, A., & De Luca, F. (2023). How does non-interest income affect bank credit risk? Evidence before and during the COVID-19 pandemic. Finance Research Letters, 53, 103657.

Meyer, P. A., & Pifer, H. W. (1970). Prediction of bank failures. The Journal of Finance, 25(4), 853-868.

Min, J. H., & Lee, Y.-C. (2005). Bankruptcy prediction using support vector machine with optimal choice of kernel function parameters. Expert Systems with Applications, 28(4), 603-614.

Nguyen, H. H., Viviani, J.-L., & Ben Jabeur, S. (2023). Bankruptcy prediction using machine learning and Shapley additive explanations. Review of Quantitative Finance and Accounting, 1-42.

Ohlson, J. A. (1980). Financial ratios and the probabilistic prediction of bankruptcy. Journal of accounting research, 109-131.

Petropoulos, A., Siakoulis, V., Stavroulakis, E., & Vlachogiannakis, N. E. (2020). Predicting bank insolvencies using machine learning techniques. International Journal of Forecasting, 36(3), 1092-1113.

Shapley, L. (1997). 7. A Value for n-Person Games. Contributions to the Theory of Games II (1953) 307-317. In K. Harold William (Ed.), Classics in Game Theory (pp. 69-79). Princeton University Press. https://doi.org/doi:10.1515/9781400829156-012

Sinkey Jr, J. F. (1975). A multivariate statistical analysis of the characteristics of problem banks. The Journal of Finance, 30(1), 21-36.

Tanaka, K., Kinkyo, T., & Hamori, S. (2016). Random forests-based early warning system for bank failures. Economics Letters, 148, 118-121.

The Federal Reserve System. (2023, May 18). Evolution of Silicon Valley Bank. The Federal Reserve System. Retrieved March 06, 2024 from https://www.federalreserve.gov/publications/2023-April-SVB-Evolution-of-Silicon-Valley-Bank.htm

Tran, K. L., Le, H. A., Nguyen, T. H., & Nguyen, D. T. (2022). Explainable machine learning for financial distress prediction: evidence from Vietnam. Data, 7(11), 160.

Yoon, I. H. (2006). Financial statement analysis for differentiating between failed and surviving merchant banks. Institute of East and West Studies, Yonsei University (Institute of East and West Studies), 18(2), 131-160.

Zhao, H., Sinha, A. P., & Ge, W. (2009). Effects of feature construction on classification performance: An empirical study in bank failure prediction. Expert Systems with Applications, 36(2), 2633-2644.

Zou, H., & Hastie, T. (2005). Regularization and variable selection via the elastic net. Journal of the Royal Statistical Society Series B: Statistical Methodology, 67(2), 301-320.