A Practical Approach to Quantitative Model Risk Assessment

Model-based decisions are highly sensitive to model risk that arises from the inad-equacy of the adopted model. This paper reviews the existing literature on model riskassessment and shows how to use the theoretical results to develop a correspondingbest practice. Specifically, we develop tools to assess the contribution to model risk ofeach of the assumptions that underpin the adopted model. Furthermore, we introducenew model risk measures and propose an intuitive formula for computing model riskcapital. Some numerical examples and a case study illustrate our results.