摘要
以风险价值(value at risk,VaR)为金融风险度量,结合Copula函数及其相关函数建立金融风险模型.考虑到金融时间序列的时变性和厚尾特性,根据GARCH(generalized autoregressive conditional heteroscedasticity)模型和极值理论的POT(peak over threshold)模型,运用Copula方法来估计VaR的值.给出实例验证,将上述方法用于刻画美国纳斯达克指数和标准普尔指数的相关性,并计算了等权重下资产组合的VaR估计值.结果表明:VaR估计值的大小与所取的置信水平以及持有期有关;t-Copula和Clayton Copula方法较其他方法能更好地捕捉资产组合的相关关系,从而可以得到更好的VaR估计值.
Using VaR （value at risk） as the measure of financial risk, this thesis applies Copula function and its relevant functions to establish a financial risk module. Considering the time-varying and fat-tail features of the financial time series, it uses Copula method to estimate the value of VaR, according to GARCH （generalized autoregressive conditional heteroscedasticity） model and POT （peak over threshold） module in extreme value theory. The method is illustrated with an example that the correlativity of American NASDAQ index and Standard ＆ Poor＇s indices are indicated and the estimated VaR of the equal weight portfolio is calculated. The result shows that the estimated value of VaR is related to confidence level and holding period, t Copula and Clayton Copula methods are superior to others in capturing the correlativity of portfolio, hence the estimated VaR is better.
出处
《扬州大学学报（自然科学版）》
CAS
北大核心
2014年第4期25-29,共5页
Journal of Yangzhou University：Natural Science Edition
基金
江苏省自然科学基金资助项目(BK20141326)
教育部高等学校博士学科点专项科研基金(博导类)资助课题(20120092110021)
江苏省高等教育教学改革研究课题重点项目(2011JSJG085)
关键词
风险价值
金融时间序列
广义自回归条件异方差
极值理论
COPULA函数
value at rish
financial time series
generalized autoregressive conditional heteroscedas-ticity
extrme value theory
Copula function
