GARCH类-Copula在金融市场相关性分析中的选择及动态Copula研究
发布时间：2019-03-06 19:39
【摘要】：金融市场相关性对于研究信息传导机制、风险度量甚至金融投资等方面都具有重要的意义。 GARCH类模型具有明确的经济涵义，并且能很好地描述金融时间序列的条件异方差性。Copula函数不限制边缘分布的选择，具有很好的适用性。因此，本文用GARCH类模型描述边缘分布，用Copula函数描述联合分布。 本文总结提出了四步判断准则以选择最佳的GARCH类模型，利用2拟合优度检验选择最佳的Copula函数。金融市场是不断变化的，因此本文还着重研究了动态Copula。动态Copula分为时变Copula和变结构Copula两类。在时变Copula研究中，本文提出了补充估计值的固定窗口法和半固定窗口法以估计时变参数序列，总结了时变参数的演化方程的形式，并研究了最佳窗口长度的选择问题。在变结构Copula研究中，，本文给出了基于B-G算法构建变结构Copula模型的方法。最后，本文通过度量VaR比较了常相关Copula模型和动态Copula模型在刻画金融风险方面的能力，分析了窗口长度对模型刻画金融风险能力的影响。
[Abstract]:The relevance of financial market plays an important role in the study of information transmission mechanism, risk measurement and even financial investment. The GARCH class model has a clear economic meaning and can well describe the conditional heteroscedasticity of financial time series. The copula function has good applicability because it does not limit the choice of edge distribution. Therefore, in this paper, the GARCH class model is used to describe the edge distribution, and the Copula function is used to describe the joint distribution. In this paper, a four-step judgment criterion is proposed to select the best GARCH model, and the optimal Copula function is selected by using the 2-fit goodness-of-fit test. The financial market is constantly changing, so this paper also focuses on the dynamic Copula.. Dynamic Copula can be divided into time-varying Copula and variable structure Copula. In the study of time-varying Copula, the fixed window method and semi-fixed window method are proposed to estimate the time-varying parameter series, the evolution equation of time-varying parameters is summarized, and the selection of the optimal window length is studied. In the study of variable structure Copula, this paper presents a method of constructing variable structure Copula model based on BJG algorithm. Finally, this paper compares the ability of constant correlation Copula model and dynamic Copula model in describing financial risk by measuring VaR, and analyzes the influence of window length on the ability of model to depict financial risk.
【学位授予单位】：暨南大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：F224;F830.91
本文编号：2435834
[Abstract]:The relevance of financial market plays an important role in the study of information transmission mechanism, risk measurement and even financial investment. The GARCH class model has a clear economic meaning and can well describe the conditional heteroscedasticity of financial time series. The copula function has good applicability because it does not limit the choice of edge distribution. Therefore, in this paper, the GARCH class model is used to describe the edge distribution, and the Copula function is used to describe the joint distribution. In this paper, a four-step judgment criterion is proposed to select the best GARCH model, and the optimal Copula function is selected by using the 2-fit goodness-of-fit test. The financial market is constantly changing, so this paper also focuses on the dynamic Copula.. Dynamic Copula can be divided into time-varying Copula and variable structure Copula. In the study of time-varying Copula, the fixed window method and semi-fixed window method are proposed to estimate the time-varying parameter series, the evolution equation of time-varying parameters is summarized, and the selection of the optimal window length is studied. In the study of variable structure Copula, this paper presents a method of constructing variable structure Copula model based on BJG algorithm. Finally, this paper compares the ability of constant correlation Copula model and dynamic Copula model in describing financial risk by measuring VaR, and analyzes the influence of window length on the ability of model to depict financial risk.
【学位授予单位】：暨南大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：F224;F830.91
【参考文献】
相关期刊论文 前10条
1 韦艳华,张世英;金融市场的相关性分析——Copula-GARCH模型及其应用[J];系统工程;2004年04期
2 张瑞锋;张世英;;基于VS-MSV模型的金融市场波动溢出分析及实证研究[J];系统工程;2007年08期
3 罗付岩;邓光明;;基于时变Copula的VaR估计[J];系统工程;2007年08期
4 汪昌云;李楠;;基于二维跳扩散模型的股市相关性研究[J];经济理论与经济管理;2010年07期
5 修晶;周颖;;人民币离岸市场与在岸市场汇率的动态相关性研究[J];世界经济研究;2013年03期
6 秦伟良;颜华实;达庆利;;基于多分辨分析的沪深股市相关性分析[J];数理统计与管理;2009年03期
7 梅世强;徐梅;;基于小波多分辨分析的中国股票市场因果关系分析[J];天津大学学报(社会科学版);2011年01期
8 吴礼斌;崔岩岩;;基于小波方差分解的沪深综指序列的特性分析[J];统计与决策;2010年23期
9 包卫军;徐成贤;;基于SV-Copula模型的相关性分析[J];统计研究;2008年10期
10 郭立伟;韩兆洲;;沪港股市动态相关性和大风险溢出的实证研究[J];产经评论;2010年04期
本文编号：2435834
本文链接：https://www.wllwen.com/jingjilunwen/touziyanjiulunwen/2435834.html
