Uncertainty modeling framework in operational risk

Tatiana Sakalo, Matthew Delasey

Abstract

ABSTRACT

The operational risk modeling approaches investigated in the literature are based on the assumption that it is possible to determine distinct parametric frequency and severity distributions based on the available data. In this paper this assumption is relaxed by developing a framework that allows for the incorporation of uncertainty by expressing inputs as p-boxes. Evidence theory is employed to combine the information obtained from different sources, such as internal loss data and quantitative risk assessment. The uncertainty is modeled at its source and propagated through the model to enable the calculation of bounds on the capital estimate, where the width of these bounds represents the level of uncertainty in the input data.