基于Copula函数的商业银行整合风险研究  

Integrated Risk Research in Commercial Banks Based on Copula function

作　　者：刘祥东[1] 刘澄[1] 王洋洋 陆嘉骏[3] 

机构地区：[1]北京科技大学东凌经济管理学院,北京100083 [2]Viterbi School of Engineering,University of Southern California,Los Angeles,USA 90089 [3]浙江大学经济学院,杭州310027

出　　处：《管理评论》2013年第10期20-27,共8页Management Review

基　　金：国家自然科学基金项目(71173012);中国博士后科学基金项目(2013M540855)

摘　　要：整合风险的量化管理已成为现代商业银行风险管理的发展趋势。本文选择信用风险和市场风险作为整合风险的影响因素,针对小样本且不满足正态分布的情况,采用核密度估计来对各自边缘分布进行拟合。根据平方欧式距离选取出最优Copula函数,使用半参数法将不同边缘分布连接成二元联合分布,并选用条件风险价值CVaR作为衡量整合风险的指标。通过对建立的Copula-CVaR模型进行Monte Carlo模拟,遍历搜索不同权重的模拟结果,找出银行最优风险资产组合,进而对我国12家上市商业银行整合风险水平进行评估。实证结果表明,整合风险低于单一风险之和,潜在的信用风险要高于市场风险,并且国有银行的整合风险普遍大于其他上市的股份制银行。Quantitative management of integrated risk has become a trend of risk management in modern commercial bank. In this paper, we choose credit risk and market risk as the component of integrated risk. As to the condition that they are little sample and don＇t satisfy normal distribution, we use the Kernel density estimation method to estimate marginal distributions respectively. According to the square Euclidean distance principle, we choose the best Copula function and use semi-parametric model to connect the two marginal distributions into a bivariate joint distribution and use conditional value at risk CVaR as the index to measure integrated risk. Through conducting Monte-Carlo Simulation on established Copula-CVaR model, we traverse the search of simulation results in different weights and find out the best risk portfolio. Furthermore, we evaluate the integrated risk level of 12 listed commercial banks of China. Empirical results show that integrated risk is lower than the sum of single risks, potential credit risk is higher than market risk and integrated risk of state-owned bank is higher than that of other listed joint stock banks.

关 键 词：COPULA函数 商业银行 整合风险 CVAR MONTE CARLO模拟 

分 类 号：F224[经济管理—国民经济] F830.33