基于时变混合Copula模型的市场间极端风险溢出度量

发布时间：2018-01-14 23:01

本文关键词：基于时变混合Copula模型的市场间极端风险溢出度量　出处：《浙江工商大学》2012年硕士论文　论文类型：学位论文

更多相关文章： Copula CoVaR 极端风险溢出 返回测试

【摘要】：2007年美国次贷危机爆发并迅速蔓延至其它金融市场,最终导致了一场席卷全球的金融危机。这一事实充分表明全球金融市场并不是相互孤立的,随着经济全球化和金融一体化的迅猛发展,全球金融市场之间的相互依存性日益增加,金融风险的表现形式也日趋复杂化和多样化,在这样的形势中缺乏对极端条件下金融市场之间风险溢出的考量,可能会在很大程度上低估金融市场的风险水平,造成灾难性的后果。如何有效地度量金融市场之间的极端风险溢出水平成为了时下风险管理机构和金融监管当局亟须解决的问题。随着金融创新的飞速发展,尤其是近年来,金融风险原有的一些分析方法如基于线性相关的分析方法等已经不再能够适应这一要求,而一种新的,可以用于研究非线性、非对称相依关系的Copula方法,在国际上被迅速应用到金融市场研究的各个领域,成为资产定价、金融风险监管、管理与防范以及保险定价的有效工具。本文尝试性地提出了一种时变混合Copula模型,这种模型能够在分布的不同区域分别选择不同种类的时变Copula函数对金融变量进行描述。模型使用时变Gumbel Copula函数对联合分布上尾部的相依关系进行描述,使用时变Rotated Gumbel Copula函数对联合分布下尾部的相依关系进行刻画,在分布的其他区域则使用时变混合Copula函数来捕捉相依关系的变化。这种模型不仅克服了单—Copula函数只适于描述分布特定区域相依性的缺陷,相较于混合Copula函数,这种模型在不同时期的相依参数还具有时变性,特别适用于研究非常时期的金融变量建模问题。时变混合Copula模型不仅能够捕捉变量之间相依关系的变化,还可以捕捉到相依模式的变化,特别是在非常时期,更加适用于对金融变量联合分布建模。本文基于时变混合Copula模型,提出了一种新的、针对极端风险溢出指标CoVaR进行估计的方法,并使用这种方法对美国股票市场、中国大陆股票市场、英国股票市场以及香港股票市场之间的极端风险溢出效应做了实证分析研究。最后通过返回测试检验表明：基于Normal Copula、Time-Varying Normal Copula、Gumbel Copula、Time-Varying Gumbel Copula以及混合(Mixed) Copula模型计算得到的市场间极端风险溢出指标CoVaR均不能通过检验,均在不同程度上低估了市场间的风险溢出效应,不能作为CoVaR计算的有效工具。而基于时变混合Copula的模型明显优于其它模型,使用该模型计算得到的金融市场间极端风险溢出指标CoVaR全部通过了检验,因此证实了本文提出的时变混合Copula模型在度量金融市场间极端风险溢出方面具有显著的优越性,可以将其作为度量金融市场间极端风险溢出水平的有效工具。
[Abstract]:In 2007, the subprime mortgage crisis in the United States broke out and spread to other financial markets, which eventually led to a financial crisis sweeping the whole world. This fact fully shows that the global financial markets are not isolated from each other. With the rapid development of economic globalization and financial integration, the interdependence of global financial markets is increasing, and the manifestations of financial risks are becoming more and more complicated and diversified. In such a situation, the lack of consideration of risk spillover between financial markets under extreme conditions may greatly underestimate the risk level of financial markets. How to effectively measure the level of extreme risk spillover between financial markets has become an urgent problem to be solved by risk management institutions and financial regulatory authorities. With the rapid development of financial innovation, especially in recent years, some of the original financial risk analysis methods, such as linear correlation analysis methods can no longer adapt to this requirement, and a new. The Copula method, which can be used to study the nonlinear and asymmetric dependency, has been rapidly applied to various fields of financial market research, such as asset pricing and financial risk supervision. This paper presents a time-varying mixed Copula model. The model can select different kinds of time-varying Copula functions to describe the financial variables in different regions of the distribution. The model uses time-varying Gumbel. The Copula function describes the tail dependency on the joint distribution. The time-varying Rotated Gumbel Copula function is used to characterize the tail dependency under joint distribution. In other regions of the distribution, the time-varying mixed Copula function is used to capture the variation of dependency. This model not only overcomes the defect that the single-Copula function is only suitable for describing the dependence of a particular region of the distribution. . Compared with the mixed Copula function, the dependent parameters of this model are time-varying in different periods. The time-varying mixed Copula model can not only capture the changes of the dependent relationships between variables, but also capture the changes of dependent patterns. Especially in the unusual period, it is more suitable for modeling the joint distribution of financial variables. Based on the time-varying mixed Copula model, a new model is proposed in this paper. The method of estimating the extreme risk spillover index (CoVaR), and using this method in the US stock market, the Chinese mainland stock market. The extreme risk spillover effect between the UK stock market and Hong Kong stock market is analyzed empirically. Finally, the result of return test shows that: based on Normal Copula. Time-Varying Normal Copula,Gumbel Copula. Time-Varying Gumbel Copula and mixed. The Copula model calculated the inter-market extreme risk spillover index CoVaR can not pass the test. The risk spillover effect between markets is underestimated to some extent, and can not be used as an effective tool for CoVaR calculation. However, the model based on time-varying mixed Copula is obviously superior to other models. Using this model to calculate the financial market risk spillover index CoVaR all passed the test. Therefore, it is proved that the time-varying mixed Copula model proposed in this paper has significant advantages in measuring the extreme risk spillover between financial markets. It can be used as an effective tool to measure the level of extreme risk spillover between financial markets.
【学位授予单位】：浙江工商大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：F830.91;F224

【参考文献】