On the reliability of integrated risk measurement in practice

Peter Grundke

Abstract

ABSTRACT

For a correct aggregation of losses resulting from different risk types and, hence, a correct computation of total economic capital requirements, existing stochastic dependencies between risk-specific losses have to be considered by integrated risk measurement (IRM) approaches. Most of the literature that deals with IRM techniques is of a scientific nature. That part of the literature that describes frameworks that are intended to be used in practice by banks and insurers lacks any kind of (at least reported) quality control. The reliability of the total economic capital reductions, which practitioners claim to have due to diversification effects between the risk types, is therefore unclear. The main contribution of this paper is to conduct this kind of quality check for the IRM practice model proposed by Brockmann and Kalkbrener, who present the IRM model of the Deutsche Bank AG. For analyzing the accuracy of this specific IRM practice approach, a simple bottom-up approach is employed as a data-generating process. Within a simulation study and based on various robustness checks, it is shown that the practice approach, in spite of its simple nature, performs surprisingly well and is capable of producing total economic capital numbers that are similar to those of the bottom-up approach.