A major Midwestern bank retained Prescio to perform a review of their Basel II AMA Operational Risk Model. The Bank’s model followed the loss distribution approach to modeling operational risk incorporating multiple severity distributions with the Poisson frequency distribution.

The approach relied on a Monte Carlo simulation to generate the aggregate loss distribution along with the structure scenario simulations for modeling the low frequency, high severity loss events of greatest interest. The Bank was also exploring the effects of using truncated versions of the severity distributions along with mixture models to capture the multi-modal characteristics of some of their loss data.

Our review included benchmarking the Bank’s approach against our own internally developed operational risk model. We performed a series of back tests for stability of the coefficient estimations. We conducted a series of simulations to confirm the stability and robustness of the Monte Carlo simulation for generating both the structured scenario data and the aggregate loss distributions. Our review also included a series of stress tests and scenarios designed to test the overall robustness of the modeling approach.