Powered by machine learning and AI and built on ever-increasing computational capabilities, complex mathematical models are rapidly becoming one of the linchpins of risk management and, by extension, directly impacting banking operations.
Techniques from mathematical statistics, big data analytics, and artificial intelligence are employed to predict likely future outcomes based on historical datasets. Nowadays, a wide range of mathematical models is used, from more traditional methods such as linear or logistic regression, principal component analysis and hierarchical clustering to more advanced machine learning techniques, including random forests, gradient boosting, neural networks and Bayesian updating.
As such, credit risk models play a crucial role in various aspects of risk management, accompanying a client’s lifecycle from loan origination via both credit scoring as well as risk- based pricing, to ongoing monitoring of clients’ risk profile. These models are essential for estimating both expected credit losses (for Loan Loss Provisioning) and unexpected credit losses (for Capital Management under the IRB approach).
Key focus areas
1. Credit scoring models
2. Risk-based pricing
3. Loan loss provisioning
4. Internal ratings-based approach (IRB)
5. Model risk management