VaR与CVaR的对比研究及实证分析  

A comparison between VaR and CVaR and empirical study

作　　者：刘小茂[1] 田立[1] 

机构地区：[1]华中科技大学数学系,湖北武汉430074

出　　处：《华中科技大学学报（自然科学版）》2005年第10期112-114,共3页Journal of Huazhong University of Science and Technology(Natural Science Edition)

基　　金：国家自然科学基金资助项目(70071012);广西科学研究与技术开发资助项目(0385008)

摘　　要：对两种风险度量方法———风险价值(VaR)和条件风险价值(CVaR)的性质和最优化算法进行了比较.首先对它们具有的性质如平移不变性等进行了比较,其中CVaR独有的次可加性最能显示它相对于VaR的优越性.然后,通过引入顺序统计量,把现实中基于CVaR的最优组合问题转化成线性规划问题,并说明只有当收益服从高斯分布时,基于VaR和CVaR的两种最优组合问题的解才是一致的,基于VaR的最优问题才有意义.最后对深市股票进行了实证分析,验证了上述结论,并分析了可能存在的影响结果的因素.The comparison between two widely used risk measures-VaR and CVaR on their properties ana optimization arithmetic was made in this paper. At first, their properties such as translation invariance were compared The subadditivity which only CVaR have can mainly manifest CVaR＇s advantage to VaR. Then, the optimization problem of portfolio under CVaR was transformed into the linear programming by the introduction of the order statistic. Only as the profit submit to Gaussian distribution, the solution of the optimizaion of portfolio under CVaR and that under VaR are coherent and thus the optimization based on VaR makes sense. At last, theempirical sutdy result on Shenzhen stock market approved the results, and the factors effecting the result were analyzed.

关 键 词：风险价值(VaR) 条件风险价值(CVaR) 投资组合 优化算法 

分 类 号：F830[经济管理—金融学]