1 Chapter 14 Managing Operational Risks with Bayesian Networks

This chapter introduces Bayesian belief and decision networks as quantitative management tools for operational risks. Bayesian networks are already well-established for use in the risk management of large corporations and the aim of this chapter is to describe how these powerful statistical tools may be applied to operational risk management in banks and other financial institutions. In order to manage operational risks effectively, the factors that are thought to influence the risk must be identified. These can be the “key risk drivers ” of the firm’s operations (see sections 12.4.4. and 13.7) or they can be classified into a separate category of their own (see section 12.2). Ideally, by exerting some control over these factors, operational risks will be reduced. Before attempting this, some important questions should be addressed: 1. Effectiveness: What effect will the controls have on the risk? It is one thing to identify the factors that contribute to the risk, but quite another to actually quantify the effect that changes in a factor will have upon the risk. For this one needs a quantitative model that relates the risk to the factors. This is precisely what a Bayesian network is. 2. Dependency: Is it possible that by reducing one risk, another risk will be increased