A Simple, Transparent and Rational Weighting Approach to Combining Different Operational Risk Data Sources

Abstract

According to Basel financial regulations, the use of external data is indispensable to the implementation of an advanced measurement approach (AMA) for calculating operational risk capital. This paper aims to provide a simple solution to perhaps the most challenging step in building any capital model for operational risk: the integration of internal and external data. We propose a generic weighting function based on a non-parametric approach that can be used to weight the different distributions; this is in line with regulatory requirements under the AMA. After analyzing the different driving factors and considering the desired sensitivities of the weights, we build and calibrate a weighting function to match all necessary and relevant conditions. This approach is completely flexible (it can be used with different constraints or adapted so that its sensitivity depends on the risk profile of the bank) while at the same time highly tractable.

Keywords: operational risk, mixing, loss-distribution approach, weighting, Anderson–Darling statistics, integration of external data