基于极值理论和多元Copula函数的商业银行操作风险计量研究  

Measurement of Commercial Bank's Operational Risk Based on Extreme Value Theory and Multivariate Copula Functions

作　　者：陆静[1] 张佳[1] 

机构地区：[1]重庆大学经济与工商管理学院,重庆400030

出　　处：《中国管理科学》2013年第3期11-19,共9页Chinese Journal of Management Science

基　　金：国家自然科学基金资助项目(71232004;71272085);国家社会科学基金资助项目(09BJL024);教育部人文社会科学基金资助项目(12YJA630135);重庆市自然科学基金资助(2009BB2042)

摘　　要：基于操作风险呈厚尾分布的特征,本文按照巴塞尔协议的要求,采用POT极值模型分别估计了多个操作风险单元的边缘分布,然后用多元Copula函数来刻画这些操作风险单元之间的关联性并计算在险价值。通过对中国商业银行1990-2010年操作风险数据的实证分析表明,Clayton Copula能更好地反映各操作风险单元之间的相关性结构,且采用Copula考虑操作风险相关性下的VaR值要比简单加总下的VaR值减少约32.3%。因此,应用Copula函数计量操作风险相关性,不仅可以提高估计的准确性,还能够达到资产组合的风险分散化效应,减少操作风险资本要求,为商业银行提升盈利能力创造条件。Owing to the fat tail of operation risk and based on the requirement of Basel Accord, marginal distributions of several operational risk cells are measured with the Peaks-Over-Threshold model of ex- treme value theory, dependency of operational risks is analyzed with multivariate copula functions and the accumulated VaR is calculated. By using the data of Chinese commercial banking from 1990 to 2010, it is shown that Clayton Copula can better reflect the dependent frame of operation risk and the VaR calculated with Clayton Copula is about 32.3% less than the one calculated by directly summing VaRs of all operation risk cells. Therefore, consideration on dependency of operation risk cells with Copula functions can meet the risk dispersion effect of asset portfolio, greatly reduce required operation risk capital and provide com- mercial banks with a better way to get more profits.

关 键 词：操作风险 极值理论 多元Copula函数 商业银行 

分 类 号：F832.33[经济管理—金融学]