An Application of Credal Networks on the Bank Operational Risks Management
Through a detailed classification of the factors causing operational risks, the author set up a topology structure based on bank business by introducing Credal Networks. The structure above is more powerful in expressing probabilities than Bayesian Networks, because it can precisely calculate the probabilities of the bank operational risks caused by different factors; then, on the basis of Delphim method, the author uses Assembly Algorithm and METLAB to calculate the probabilities of operational risks of different nodes. The paper does bottom-up and top-down reasoning, from the experiment, we find that the analysis of the operational risks by using Credal networks, in the absence of the data of previous losses, can effectively predict the potential risks, even if the experts are not capable of giving the exact probability of the risk factors and their mutual effect, so as to help the executives to make important decisions.
credal networks bank operational risks bayesian networks topology structure
XIAO Kuixi LI Haipeng LIAO Wenxiu
Centre of International Trade and Economics study, Guangdong University of Foreign Studies, P.R.China, 510006
国际会议
威海
英文
584-589
2010-07-24（万方平台首次上网日期，不代表论文的发表时间）