基于GARCH-EVT-COPULA模型的外汇投资风险测度研究

【作者】 谭芳；

【导师】 王宗润；

【作者基本信息】 中南大学 ， 管理科学与工程， 2009， 硕士

【摘要】 人民币汇率制度改革以来,我国外汇市场机制不断发展和完善,外汇投资由于其自身的一些特点已经成为继股票投资后的又一重要投资领域。而同时,外汇波动频繁而且波动幅度加大,导致外汇风险加大,如何度量和管理外汇风险成为业界人士关注的话题。本文采用近年来被国际市场广泛应用的VaR和CVaR方法度量外汇风险。GARCH模型和极值理论是两种测度外汇风险的好方法,GARCH模型能很好地描述收益率序列的动态波动集群现象,而极值理论能很好地描述收益分布的非正态厚尾特征。本文构建能增强模型预测能力的GARCH-EVT模型研究单一外汇动态风险,并用美元、欧元、日元、港币四种外汇进行实证分析。后验测试的结果表明,与股市风险的研究结果一样,GARCH-EVT模型能较好地预测单一外汇风险。考虑到外汇之间的相关性,同时投资者往往可以选择“一篮子”货币投资或规避风险,本文在GARCH-EVT模型的基础上引入了多元正态Copula、多元t Copula、多元Clayton Copula三种Copula函数对多元外汇投资组合风险进行研究。并以四元外汇为例,计算考虑相关结构时的单一外汇风险；计算四元外汇投资比重相等时的投资组合风险值；计算以风险最小为目的的四元外汇投资组合比例。结果表明t Copula和Clayton Copula能更好地描述外汇间的相关性,而且不管是哪种类型的Copula或者是哪种置信水平,最小风险的投资组合系数差别并不是很大,投资主要集中在美元。基于GARCH-EVT-COPULA模型的外汇风险价值更深层次地量化了外汇投资组合风险,为投资者和风险管理者提供了更全面的信息,便于风险决策,具有重大的实际意义。更多还原

【Abstract】 After the RMB exchange rate system reformation, the market mechanism of our foreign exchange market has gained continuous development and refinement. Foreign exchange investment has become a next important one after the stock investment. Meanwhile, the foreign exchange rates fluctuate frequently and the range of volatility also increases a lot, which lead the foreign exchange risk increase. Consequently, how to measure and manage foreign exchange risks has been concerned issue in financial field. In this paper we adopt VaR and CVaR to measure the foreign exchange rate risk and the VaR method has been widely used in the international market in recent years.GARCH model and EVT are two methods to measure foreign exchange risk. GARCH can describe the fact about returns time series-dynamic volatility resulting in the well-recognized phenomenon of volatility clustering, and EVT can better portray non-normality and fat tails of the return distribution. We compose the two methodologies of GARCH and EVT to come up with a robust model with much enhanced predictive abilities and apply it to study the risk of four foreign exchange including USD,EUR,JPY and HKD. The back test result shows that the GARCH-EVT model can also better predict the foreign exchange risk like describing share risk.With a view to the relevant changes between exchange rates, and at the same time investors can often choose "basket" currency to invest or avoid risk, this article introduces three kinds of multiple Copula functions including Normal Copula, T Copula, Clayton Copula to study multiple foreign exchange risk, and we choose the four kinds of foreign exchange as an example to do research. firstly calculate the single exchange risk through Copula model considering the relevant structure. Secondly compute the portfolio risk with the equal investment proportion. Thirdly find the optimal invest proportion aimed the least portfolio risk. The sesult shows that t Copula and Clayton Copula can better depict the correlation structure, the portfolio proportion of the minimal risk are approximate under which Copula or which confidence level. The investment focuses on USD asset.The foreign exchange risk value base on GARCH-EVT-COPULA model can better measure the portfolio risk, it can provide the investor and risk managers for more comprehensive information and help them risk decision-making, so it has great practical significance.更多还原