Operational risk measurement based on multi-time scale dependence

Abstract

Most existing studies simply use the annual data recommended by the Basel Committee on Banking Supervision (BCBS) to measure operational risk under dependence, and few are aware of the impact of the dependence at different time scales on measurement results. This paper proposes to measure the operational risk under dependence from the perspective of multi-time scales. Firstly, frequency dependencies are modeled based on the annual, semi-annual, and quarterly risk data to analyze the features of multi-time dependencies. Secondly, the loss distribution approach (LDA) is employed to calculate the operational risk under dependencies on the different time scales and discuss the impact of the time scale selection on operational risk measurement results systematically. The empirical study is based on the Chinese Operational Loss Database (COLD), which is one of the largest Chinese external operational risk databases. The results reveal that the dependencies of operational risk on different time scales indeed have distinct differences and significantly influence the operational risk measurements.