Using Shifted Distributions in Computing Operational Risk Capital

Abstract

A common problem in applying Loss Distribution Approach to estimating operational risk capital is that loss event data is only collected above certain threshold. This paper demonstrates how conditional distributions obtained from data shifted by subtracting the threshold can be used to estimate operational risk economic capital. The key is to utilize the aggregated data of small impact events below the collection threshold which is usually collected by banks but neglected in the analysis. The advantages of this method, which make it a method of choice in many situations, are relative ease of implementation, the utilization of all the available data, the reduced errors in estimated distribution parameters and more efficient capital calculation due to lower annual loss intensity.

Keywords: Operational risk, Economic capital, Loss distribution approach, Truncated data

JEL Classification: C13, G10, G21