Credit Risk Models

The reliable evaluation of an obligors’ creditworthiness and the estimation of their probability of default (PD), exposure at default (EAD) and expected recovery (RR, LGD) are crucial components within credit risk management.

Key Project Experience

• We supported a banking client in building rating models compliant with Basel regulatory requirements. Regression analysis was used to identify obligors’ characteristics with strong discriminatory power and combine them in a stable rating model. Various stakeholders were involved in the model building process to ensure that the statistically significant variables are also in line with economic intuition. Approved rating models were calibrated to representative data and documented.
• We helped to establish a model risk management framework for credit risk models and set up a regular credit risk model validation and monitoring processes. As part of the project we drew up model validation standards for PD, LGD and EAD models. These standards contained procedural steps for qualitative model validation, statistical tests for quantitative performance assessment and detailed guidelines for the consistent interpretation of the validation results.