Comparing alternative mixing models for external operational risk data

This article presents a study of alternative mixing models for external data for a particular risk class using real operational risk data. Scaling through a proxy for size for this risk class does not seem to be a sensible technique; kernel-modified estimators and Bayesian estimators represent an improvement. The authors describe one additional technique capable of improving the treatment further

The regulatory definition of operational risk is the loss expected over the next 12 months resulting from inadequate or failed internal processes, people and systems or from external events (excluding reputational and strategic risk).

While market or credit risk can be thought of as a function of the bank's portfolio, operational risk should be treated as a function of the bank's processes and