Sources of uncertainty in modeling operational risk losses

Giulio Mignola and Roberto Ugoccioni

Abstract

ABSTRACT

Operational risk quantification techniques have been rapidly evolving since the first attempts in early 2000, when it appeared clear that this kind of risk would attract a specific capital requirement in the new prudential regulation. The basic component of most models developed by the industry is historical (or scenario) loss data. The modeling techniques used to obtain the risk measures are generally well developed and understood. In this work, assuming a simple but rigorous modeling framework, containing the basic features of the models which are generally observed in the industry, focus will be placed on the uncertainty of the model outputs. The sources of this uncertainty will be analyzed in a systematic way.