Quantification of model risk in stress testing and scenario analysis

Jimmy Skoglund

Abstract

Being able to understand and quantify the model risk inherent in loss-projection models used in macroeconomic stress testing and impairment estimation is a significant concern for both banks and regulators. The application of relative entropy techniques allows model misspecification robustness to be numerically quantified using exponential tilting toward an alternative probability law. Employing a particular loss-forecasting model, we quantify the worst-case-loss term structures of that model, yielding insights into the behavior of the worst-case scenario. In general, the worst case obtained represents an upward scaling of the term structure consistent with the exponential tilting adjustment. The relative entropy approach to model risk we use has its foundation in economics with robust forecasting analysis, and it has recently started to be applied in risk management. This technique can complement traditional model risk quantification techniques, where a specific direction or range of model misspecification reasons are usually considered, such as model sensitivity analysis, model parameter uncertainty analysis, competing models and conservative model assumptions.