摘要
针对现有风险度量模型不能准确的模拟高维金融资产收益率风险,以上证指数、沪深300指数和股指期货指数为例,首先利用SVt和EVT对各序列的边缘分布进行建模,然后采用Vine Copula方法分析多序列之间的秩相关关系和极大似然值估计法估计参数,得到RVine,CVine和DVine三种不同树结构的分解模型,通过Monte Carlo模拟法计算出在同一边缘分布不同Vine Copula方法下和在不同边缘分布同一Vine Copula方法下单资产和投资组合的金融风险VaR.经实证检验并分析对比,VaR和返回式检验均表明SVt和EVT相结合对边缘分布有较好的拟合效果,再运用RVine描述资产间的相依结构在度量投资组合金融风险方面更准确合理.
Facing with the existing financial risk measurement model not accurate enough to measurement the risk of financial assets return in higher dimensional,taking the Shanghai index,CSI 300 index,and stock index futures as examples,using SVt and Extreme Risk Theory to model the edge distribution of each sequence,then using Vine Copula method to analyze the rank correlation between each sequences and maximum likelihood method to estimate parameters,the decomposition models of three kinds of different tree structures are obtained which are RVine,CVine and DVine,the Monte Carlo simulation method is used to calculate the single asset and investment portfolio financial risk VaR under the different Vine Copula method of the same edge distribution and the same Vine Copula method of the different edge distribution,respectively.Both VaR and return tests show that the combination of SVt and EVT has a better fitting effect on the edge distribution,and then use RVine to describe the dependence structure of the assets to be more accurate and reasonable in measuring the financial risk of the investment portfolio.
出处
《数学的实践与认识》
北大核心
2016年第9期104-112,共9页
Mathematics in Practice and Theory
基金
国家自然科学基金(11361044)
宁夏软科学计划项目(2015)
