Hidden Markov Regimes in Operational Loss Data: Application to the Recent Financial Crisis

We determine whether there is an endogenous Hidden Markov Regime (HMR) in the operational loss data of banks from 2001 to 2010. A high level regime is marked by very high loss values during the recent financial crisis. There is therefore temporal heterogeneity in the data. If this heterogeneity is not considered in risk management models, capital estimations will be biased. Levels of reserve capital will be overestimated in periods of normal losses, corresponding to the low level of the regime, and underestimated in periods of a high regime. Variation in capital can go up to 30% during this period of analysis when regimes are not considered.