运用贝叶斯网络量化和控制商业银行操作风险——一个基于实际业务流程的例子  

Using Bayesian Network to Quantify and Control Operational Risk of Commercial Banks:A Practical Business Process Case

作　　者：刘睿[1] 巴曙松[2] 刘家鹏[3] 

机构地区：[1]云南大学工商管理与旅游管理学院 [2]国务院发展研究中心金融研究所 [3]清华大学公共管理学院博士后流动站

出　　处：《投资研究》2011年第7期106-117,共12页Review of Investment Studies

基　　金：云南省教育厅科学研究基金(编号:08Y0015)的资助

摘　　要：操作风险量化是为降低和控制风险服务的。运用贝叶斯网络分析量化操作风险并改进控制,对提升商业银行的操作风险管理水平是一条有效途径。论文以我国银行的一个典型业务—远期结售汇为例,研究了实践中建立和运用贝叶斯网络来估计操作风险发生频率的具体方法。以流程分析和映射为基础,阐述了风险映射与节点识别、网络结构建立、网络节点描述、节点赋值的实施步骤,并说明了利用贝叶斯网进行因果推理和诊断推理的方法。The quantification of operational risk serves the elimination and control of risk. Quantifying operational risk based on the Bayesian Network is an efficient way for commercial banks to improve operational risk management. This paper takes the case of a typical business process as an example, Sale and Purchase of Forward Foreign Currencies against RMB, to demonstrate how a Bayesian Network （BN） can be constructed and how operational risk frequency of a business process can be quantified. Based on process analysis and mapping, this paper demonstrates the procedure of risk mapping and node recognition, the establishment of network, the description of nodes and the valuation of nodes, providing the methodology of using Bayesian Network for causality and diagnosis analysis.

关 键 词：操作风险 贝叶斯网 业务流程 

分 类 号：F832.3[经济管理—金融学]