Journal of Operational Risk

Risk.net

Hidden Markov regimes in operational loss data: application to the recent financial crisis

Georges Dionne and Samir Saissi Hassani

  • We propose a method to consider business cycles in operational risk data.
  • Statistical tests do not reject asymmetric distribution in U. S. banks losses.
  • Capital estimations is biased if this heterogeneity is not considered.
  • Additional capital reaches 30% when regimes are neglected.

We propose a method to consider business cycles in the computation of capital for operational risk. We examine whether the operational loss data of US banks contains a hidden Markov regime-switching feature from 2001 to 2010. We assume asymmetric distribution of monthly losses. Statistical tests do not reject this assumption. A high-level regime is marked by very high loss values during the recent financial crisis, confirming temporal heterogeneity in the data. If this heterogeneity is not considered in risk management models, capital estimations will be biased. Banks will hold too much capital during periods of low stress and not enough capital in periods of high stress. Additional capital reaches 30% during this period of analysis if regimes are not considered.

To continue reading...