Scenario Analysis Modelling

Rafael Cavestany

Having presented methods for loss data modelling in Chapter 4, we now examine scenario analysis (SA) modelling for its integration into the capital model, either the hybrid model or scenario-based approach. In fact, SA permits us to incorporate, into a capital model, institution-specific, high-impact risk information that can be obtained only from expert judgement due to its low frequency. This requires the translation of scenario results in distributions models that can be incorporated into the joint loss distribution for the derivation of capital estimates

On this topic, BCSG-AMA says, “A bank should thus ensure that the loss distribution(s) chosen to model scenario analysis estimates adequately represent(s) its risk profile.” The process of choosing SA loss distribution that adequately represents the scenario risk profile can be performed through the following processes corresponding to this chapter:

• □

  “Translating scenario analysis results into distribution characteristics”;

• □

  “Fitting a full distribution to scenario analysis”;

• □

  “Distribution shape control during the scenario distribution fit”: this section describes how to arrive at a reasonable tail