Evaluating Design Choices in Economic Capital Modelling: A Loss Function Approach

Nicholas M Kiefer and C Erik Larson

Economic capital models are complex, and by their design usually take as input the output of several other modelling exercises, including but not limited to the estimation of asset-level default probabilities (PD), loss-given default (LGD) rates, and cross-asset correlations of these same parameters.

Due to their inherent complexity, it is imperative that financial institutions develop an assessment of the model risk associated with the use of economic capital models, as well as their associated driver models. Model risk is defined for this purpose as incorrect predictions or incorrect decisions resulting from the misuse of models. Model risk is assessed in the context of the intended use of models and best-known practices used to build models. Credit-risk decision models are evaluated with respect to sample design, modelling techniques, validation procedures and revalidation procedures. Since all models are imperfect, model risk is best thought of as losses that might be sustained due to the overly broad interpretation or use of a model beyond the scope of application for which it was developed.

This chapter considers issues relating to the segmentation or grouping of credit