基于贝叶斯极值估计的商业银行内部欺诈风险度量研究  

A Study of Internal Fraud Risk Measurement of Commercial Banks Based on the Bayesian Extreme Value Estimations

作　　者：欧阳资生[1] 黄颖 

机构地区：[1]湖南商学院财政金融学院,湖南长沙410205

出　　处：《运筹与管理》2017年第12期126-134,共9页Operations Research and Management Science

基　　金：国家社科基金重点项目(17ATJ005);National Planning Office of Philosophy and Social Science(11BTJ011);湖南省高等学校科技创新团队~~

摘　　要：内部欺诈风险是我国商业银行面临的一个重大风险来源。本文针对内部欺诈具有的低频率高损失的特点,采用不同分布分段刻画其损失统计分布规律,对于低于和高于门限值的样本点,采用Box-Cox变换和全Paretian分布模型进行分析,然后采用贝叶斯估计对全Paretian分布模型的参数进行估计,接着在此基础上对建立了一个内部欺诈风险度量模型,然后使用所构建的风险度量模型对操作风险在险风险值、经济资本和最大可能损失进行了测算,最后提出了防范操作风险的政策建议。Internal fraud risk is a significant risk source for the banking industry in China. Using the loss data of internal fraud of Chinese commercial banks, we present a segmented distribution model through the choice of exact threshold. We use the Box-Cox change for the data less than the threshold, and a full Paretian distribution for the data greater than the threshold. Then we use the Bayesian methods to estimate the parameters of the full Paretian distribution. The object of this work is to present a full Paretian model for measuring the operational risk of a bank. The model is used to estimate the value-at-risk, economic capital and probable maximum loss. At last, proposals for Chinese commercial banks are given to avoid operational risks.

关 键 词：内部欺诈 操作风险在险风险值 全Paretian分布模型 

分 类 号：F222.3[经济管理—国民经济] F83