New approaches to operational risk modeling

Estimation of economic capital of a financial institution requires modeling of operational losses of the business units of the organization. Operational losses of the financial institution are usually represented as an aggregate sum of the losses incurred by the operational events of the business units. Simulation of these events requires the introduction of a co-dependence structure for a more realistic description of the loss process. In this paper, we discuss some of the approaches to this modeling problem. We propose a semi-analytical framework for evaluating the operational losses and for estimating the economic capital. This framework includes a backward simulation method for modeling the dynamics of the operational risk events, described as a multivariate Poisson process. We consider the corresponding calibration problem for modeling the correlation structure of the multivariate Poisson process. In addition, we discuss the methods for aggregating the loss distributions of the business units and for computing the distribution of the operational losses of a financial institution. A dynamic approach to the simulation of the operational losses might be beneficial for developing strategies in allocating economic capital.