ABSTRACT
The ability of Bayesian Networks (BN) to generate decision support for complex risk management problems has been thoroughly demonstrated [Ale et al. 2007; Langseth 2002; Røed et al. 2009]. Recent research also illustrates the applicability of BNs within the domain of operational risk in financial institutions [Adusei-Poku (2005), Cowell et al. (2007), Mittnik & Starobinskaya (2007), Neil et al. (2009)]. Applications of BN models within the finance industry focus on establishing quantitative models for operational risk in an Advanced Measurement Approach (AMA) regime under the Basel II legislation (BCBS 2006). The unique ability of a BN model to visualise the causes of an event as well as simulating the impact of mitigating actions, is of particular interest as it facilitates sound risk management. BN methodology also provides an attractive solution to the problem of mixing input from different sources such as internal and external data and expert knowledge.
