Real-Time Credit Risk Monitoring: AI-Driven Early Warning Systems for Loan Portfolio Deterioration

References

[1] Accenture. (2023). AI in banking: Explaining the explainable. Accenture. https://www.accenture.com

[2] Australian Journal of Machine Learning Research & Applications. (2023). Dunka, V. K. Implementing AI-driven predictive analytics for credit risk management in banking. Sydney Academics.

[3] Bank for International Settlements. (2023). Operational resilience and machine learning governance. BIS. https://www.bis.org

[4] Basel Committee on Banking Supervision. (2016). Minimum capital requirements for market risk. Bank for International Settlements. https://www.bis.org

[5] Board of Governors of the Federal Reserve System. (2011). SR 11-7: Supervisory guidance on model risk management. Federal Reserve. https://www.federalreserve.gov

[6] 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. https://doi.org/10.1145/2939672.2939785

[7] De Finetti, B. (2017). Theory of probability. Wiley.

[8] Deloitte. (2024). Enhancing early warning systems with Gen AI. Deloitte UK. https://www.deloitte.com

[9] European Banking Authority. (2020). Guidelines on loan origination and monitoring (EBA/GL/2020/06). EBA. https://www.eba.europa.eu

[10] European Commission. (2022). Proposal for a regulation laying down harmonised rules on artificial intelligence (EU AI Act). https://eur-lex.europa.eu

[11] Federal Deposit Insurance Corporation. (2017). Adoption of supervisory guidance on model risk management. FDIC. https://www.fdic.gov

[12] Financial Times. (2024). Bank of England deputy governor on AI in stress tests. Financial Times. https://www.ft.com

[13] Hadji Misheva, B., Osterrieder, J., Hirsa, A., Kulkarni, O., & Fung Lin, S. (2021). Explainable AI in credit risk management. arXiv preprint. https://arxiv.org/abs/2104.08719

[14] Kahneman, D. (2011). Thinking, fast and slow. Farrar, Straus and Giroux.

[15] Ke, G., Meng, Q., Finley, T., Wang, T., Chen, W., Ma, W., ... & Liu, T.-Y. (2017). LightGBM: A highly efficient gradient boosting decision tree. Advances in Neural Information Processing Systems, 30, 3146–3154.

[16] Lundberg, S. M., & Lee, S.-I. (2017). A unified approach to interpreting model predictions. Advances in Neural Information Processing Systems, 30, 4765–4774.

[17] McKinsey & Company. (2024). Embracing generative AI in credit risk. McKinsey. https://www.mckinsey.com

[18] Neural Computing and Applications. (2022). Meta-analysis of AI in credit risk modeling. Springer. https://link.springer.com

[19] Newgen Software Technologies. (2024). Early warning system. Newgen. https://www.newgensoft.com

[20] OCC. (2023). Risk management of AI/ML in banking. Office of the Comptroller of the Currency. https://www.occ.gov

[21] OpenAI. (2023). Policy on fairness in models. OpenAI. https://openai.com

[22] Ribeiro, M. T., Singh, S., & Guestrin, C. (2016). “Why should I trust you?”: Explaining the predictions of any classifier. Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, 1135–1144. https://doi.org/10.1145/2939672.2939778

[23] SIGMOD. (2023). Streaming frameworks for real-time decisioning. Proceedings of the 2023 ACM SIGMOD International Conference on Management of Data.

[24] World Bank. (2023). Credit infrastructure in emerging markets. World Bank. https://www.worldbank.org

[25] [Author Unknown]. (2023). Human–AI interaction in loan approval. Journal of Behavioral Decision Making. Wiley. https://doi.org/10.1002/bdm.2345

@article{1710202,
      author = {Godwin David Akhamere},
      title = {Real-Time Credit Risk Monitoring: AI-Driven Early Warning Systems for Loan Portfolio Deterioration},
      journal = {Iconic Research And Engineering Journals},
      year = {2024},
      volume = {7},
      number = {9},
      pages = {553-563},
      issn = {2456-8880},
      url = {https://www.irejournals.com/formatedpaper/1710202.pdf},
      abstract = {Real time monitoring of credit risk is leading the financial innovation curve with the possibility of detecting the loan portfolio deterioration in advance through AI enabled early warning systems (EWS). This analysis formulates and tests models which constantly combine the indicators of the borrower behavior and macroeconomic signals to predict the risk of default even before occurring. Our proposed real-time solution integrates sequential deep learning, survival analysis, and hybrid micro-macro modeling and is operated through a streaming architecture facilitated by MLOps and explainability best practices. Evaluating predictive performance, lead time of early warnings, fairness to risky borrowers, and economic efficiency in the face of stress via the use of synthetic based and real-world data (with stringent privacy oversight), we compare the performance of the models against the current state of the art in the field. Our results show that real-time sequence models tend to predict a much longer lead in time than the traditional scorecards in normal conditions, and are more useful in predicting defaults during recessions as well since the macro-conditional hazard models are useful in such situations. The system has a high degree of calibration and fairness, and the explainability tools make model decisions transparent and ready to be presented to the regulator. We describe operation deployment considerations, regulator alignment (e.g. IFRS 9/CECL), cost benefit trade-offs and constraints. The findings emphasize the importance and effectiveness of AI-enabled, around-the-clock surveillance of EWS and how it can provide both theoretic and practical insights to adaptive risk management in the field with respect to financial risks.},
      month = {March},
  }