Stressing Dynamic Loss Models

We propose a reverse stress testing framework for dynamic loss models (in particular, compound Poisson processes over a finite time horizon). We define the stressed model as the probability measure under which the process satisfies the constraints and which minimizes the Kullback-Leibler divergence to the reference compound Poisson model. We solve for the stressed model and illustrate the dynamic stress testing by considering stresses on risk measure constraints such as Value at Risk.