基于Bayesian-Copula方法的商业银行操作风险度量

发布时间：2018-01-22 00:09

本文关键词： 操作风险贝叶斯理论损失分布 Copula　出处：《中国管理科学》2011年04期 　论文类型：期刊论文

【摘要】：本文在对损失分布法分析的基础上,将损失事件划分为内部欺诈、外部欺诈以及违规执行三种类型;引用两阶段分布拟合操作风险的损失强度分布,同时采用贝叶斯理论中的吉布斯抽样来获取参数估计值以减小低频率高损失数据不足带来的误差;考虑到操作风险各损失事件间可能存在的相关性,本文采用Copula函数对操作风险进行整合以获得联合损失分布函数,并计算出不同置信水平下我国商业银行操作风险损失的VaR值与CVaR值。实证研究的结果表明:基于贝叶斯理论的参数估计综合考虑了总体与样本等先验信息,估计出的参数值误差较小;Copula函数的引入与VaR值、CVaR值的测算,能在考虑了损失事件发生概率的同时,估测出操作风险潜在的损失大小,从而可以更准确度量操作风险。
[Abstract]:Based on the analysis of the loss distribution method, this paper divides the loss event into three types: internal fraud, external fraud and illegal execution. The two-stage distribution is used to fit the loss intensity distribution of operational risk, and the Gibbs sampling in Bayesian theory is used to obtain the parameter estimation value to reduce the error caused by the lack of low frequency and high loss data. Considering the possible correlation between operational risk and loss events, the Copula function is used to integrate the operational risk to obtain the joint loss distribution function. The VaR value and CVaR value of operational risk loss of Chinese commercial banks under different confidence levels are calculated. The empirical results show that:. The parameter estimation based on Bayesian theory takes into account the prior information such as population and sample. The error of the estimated parameter value is small; The introduction of Copula function and the calculation of VaR value and Cvar value can estimate the potential loss of operational risk while considering the probability of loss event. Thus, the operational risk can be measured more accurately.
【作者单位】：中南大学商学院;
【基金】：国家自然科学基金委创新群体科学研究基金项目(70921001) 国家自然科学基金面上项目(70973145,70771114) 中央高校基本科研业务费专项资金资助
【分类号】：F224;F832.33
【正文快照】： 1引言长期以来,人们对信用风险和市场风险的研究给予了极大关注,而对操作风险的重要性认识不足。20世纪90年代以来,以巴林银行和大和银行为代表的银行巨额损失事件一再发生,特别是2008年年初的法国“兴业银行事件”,以曾轰动全中国的广州商业银行的“许霆恶意提款事件”都说

【参考文献】