The Research Of Measurement Of Commercial Banks' Operational Risk Based On Bayesian Network

Commercial bank is not only an important part of the modern financial industry, but also a high risk industry. Risk measurement and management is the effective core mission of the commercial banks. As one of the main risk of commercial banks, operational risk bursts more and more frequently and seriously, with the rapid development of world economy and financial industry. Because the losses of operational risk events all over the world, financial and regulatory authorities began to realize the need and importance for operational risk management. Operational risk management has become the core aspect of the risk control of commercial banks and other financial institutions.Currently, the banking sector credit risk and market risk measurement and management have been relatively mature, while the operational risk measurement research and practice is still in the exploratory stage. The main research methods include the loss distribution approach based on the value at risk, extreme value theory, income and expenditure models, and so on. However, the disadvantage of these common methods is to ignore the relationship between risk incentives. It is difficult to use these methods independently to describe and measure operational risks during the actual process of commercial bank operation.First, this paper introduces the basic concepts of operational risk, operational risk characteristics and classification. On this basis, we give an overview of all current operational risk measurement methods, describe the main methods in detail, and discuss the applicability of them. After the introduction of the traditional methods, we discuss the bayesian network approach, and the main principle, constructing methods, data requirements, data processing methods, application methods of the Bayesian network method. After that, we provide the averaged losses method to vanish different-time correlations.Then, we use the external operational risk loss data of commercial banks for the empirical analysis. LDA and Bayesian network are combined in this paper. The Bayesian network model is constructed using the data processed by averaged loss method. Using the model we estimate the VaR of the operational risk in commercial banks in China. According to Bayesian network inference, we simulate the process of the risk management of commercial banks and give the main application of the model in the measure of operational risk.Finally, the paper summarizes the research results, and suggests possible future research directions.