【作者】 钟山；

【导师】 傅强；

【作者基本信息】 重庆大学 ， 技术经济及管理， 2014， 博士

【摘要】 伴随着世界经济全球化程度的日益提高和我国改革开放的进一步加深，金融市场作为经济发展的核心，在我国经济的发展和改革中起到了重要的枢纽作用和关键的推动作用。然而，经济快速发展所积累下来的金融风险得不到有效的控制，很容易引起连锁反映，从而引发全局性、系统性的金融危机，并殃及整个经济生活，甚至导致经济秩序混乱与政治危机。因此，在经济全球化的趋势和背景下，正确识别金融风险及其传导机制，及时、准确监测，采取适当措施防范和化解金融风险，进而维护国家经济安全，是一个具有重要理论意义和实践意义，又是急需解决的重要课题。本文在次贷危机的影响尚未完全消散，欧债危机刚刚有所缓和，全球金融市场仍处于恢复、调整期的严峻形势下，把目光汇聚于极端情况下金融市场风险的度量、管理及其传导机制的分析上，通过理论探讨和实证分析，研究行之有效的风险管理方法，为识别和管理风险创造良好条件，无疑具有极强的理论价值和现实意义。本文综合运用多个学科的理论，从金融风险的度量和溢出效应入手，先采用定性的方法了分析我国金融业存在的风险及主要影响因素，然后采用VaR、ES、CARE、CoVaR、GARCH、SV、Copula、MCMC等多种定量方法，来分析我国所面临的金融风险及其传导机制，立足我国经济转型的特征，探讨如何通过有效措施防范和化解金融风险、维护国家金融稳定，具体工作如下：(1)极端损失出现时，以分位数为基础的QVaR估计值会出现完全低估风险的情况。而从业人员和监管部门通常更关心就是在异常波动和极端损失（如经济危机、市场崩盘等）出现的情况下，金融资产所面临的最大潜在风险。为了避免QVaR模型的这些缺点，采用以Expectile模型为基础VaR度量方法(以下简称EVaR)来构建向量自回归期望分位数模型(CARE)，并以此来计算金融收益序列的VaR和ES，用以度量金融市场风险。通过对上证指数和深圳成指的实证分析发现：CARE模型在对金融收益序列的VaR估计与预测方面，明显优于金融风险管理实务界主流的RiskMetrics模型，也优于CAViaR模型，而且在ES度量方面也有着非常明显的优势。(2)从计量经济学的角度来看,考察两个市场间风险传导的实质就是检验其风险-Granger因果关系。采用Granger因果检验法，结合时变Copula-GARCH模型来分析中美股市之间的极端风险溢出效应。实证结果表明：长期来看，中国股市对美国股市的极端风险溢出并不明显，而美国股市对中国股市则具有显著的极端风险溢出效应。特别是经济危机时期，中国股市受美国股市的影响更加剧烈。(3)采用Granger因果检验法分析了汇率市场与黄金市场的均值溢出效应；并通过构建时变二元正态Copula-LSV-t模型，采用马尔科夫链蒙特卡洛（MCMC）方法对模型的参数进行贝叶斯估计，分阶段分析了汇率市场与黄金市场间的波动溢出效应。实证结果表明：均值溢出方面，次贷危机阶段，汇率市场对黄金市场存在着显著的均值溢出效应，而黄金市场对汇率市场则不存在均值溢出效应；欧债危机阶段，汇率市场与黄金市场之间存在着剧烈的均值溢出效应；波动溢出方面，经济危机阶段，汇率市场和黄金市场之间存在着明显的波动溢出效应，而在经济平稳时期，波动溢出效应则不明显。(4)时变相依结构的函数Copula方法可以灵活地选择资产边缘分布的具体形式，将资产的边缘分布与他们之间的时变相关结构分开考虑，能够捕捉变量之间非线性、非对称的动态相关关系；极值理论EVT具有很好的收益尾部拟合优度，不需要对整个分布进行建模，能够较好地克服其他度量方法在解决厚尾分布上的不足，是度量市场风险极端情形的一种有效方法；CoVaR方法不仅可以衡量单个金融市场的风险溢出，还能够捕捉系统风险的变化，是一种全面和有效的风险管理技术。采用极值理论，结合CoVaR和Copula模型，构建出EVT-Copula-CoVaR模型，分析了中国大陆、香港、美国和欧洲四个股票市场之间的风险溢出效应。实证结果表明：美国股票市场对中国股票市场存在着单向的正风险溢出效应，其次是香港股票市场，欧洲股票市场对中国股票市场的溢出效应很有限；中国大陆和香港股票市场对欧美股票市场的影响非常微弱；美国股票市场与欧洲股票市场之间存在着显著的双向溢出效应。这就意味着，我们通过可以分析信息更加透明的美国股市的历史信息，来预测中国股市的极端风险。最后，结合本文的理论分析和研究结论，提出了可行的、针对性的政策建议，并对未来的研究工作进行展望。更多还原

【Abstract】 As the rapid development of the world economic globalization and China’s reformand opening up policy, the financial market plays an essential role in the developmentand opening up of our economy. However, the financial risk accumulated by the rapiddevelopment of economy can not be effectively controlled, thus it will lead to afinancial crisis, which will destroy the balance of our economy. Moreover, it will lead toeconomic disorder and political crisis. Therefore, it is important to guard againstfinancial venture and maintain the national economic security. Meanwhile, this is themajor issue we need to solve.The focus of this paper is on measurement, management and analysis of thefinancial markets risk in an environment that the global financial market would meet theU.S. national debt crisis again. At the same time, the influence of the subprime crisisand the European debt crisis is not disappear. The thesis creates favorable conditions foridentification and management of risk, which is of great theoretical significance andrealistic value.The paper starts from estimating the spillover effect of financial risk. Then, itclarifies the risk and effect in economic system. Next, it analyzes the economic risk byusing some quantitative methods such as VaR, ES, GARCH, SV, COPULA, MCMC.Finally, it discusses how can we use effective ways to guard against financial ventureand vindicate the stability of the economic system. The next paragraph will give usmore details.When Extreme loss occurs, the estimate based on QVaR will underestimate the risk.However, the financial regulators care more about the maximum potential risk that thefinancial assets will face in an huge economic loss. In order to avoid thesedisadvantages of QVaR model, this paper uses the CARE model which is based onExpectile and CAViaR model to calculate the return of VaR and ES. At the same time,the thesis uses the result to measure the financial risk. The finding of the empiricalanalysis is that the CARE model is better than the RiskMetrics and CAViaR model inestimating and predicting the return of VaR. Moreover, the CARE model has advantagein estimating the ES.In Econometrics, the essence of analyzing market risk is to examine itsrisk-Granger causality. This paper use the Granger causality and Copula-GARCH modelto analyze the Extreme risk spillover effect between the China’s and American stock market. The result shows that the extreme risk spillover effect of Chinese stock marketmade by American stock market is obvious than that China makes to America.Especially, the extreme risk spillover effect is more dramatic in economic depression.This paper points out the mean spillover effect between currency market and goldmarket by using Granger causality. Furthermore, it elaborates the spillover effectbetween currency market and gold market.The empirical results show that the foreignexchange market has strong mean spillover effect on the gold market, while the goldmarket has no spillover effect on the foreign exchange market in the subprime crisis.Furthermore, there is dramatic mean spillover effect between the currency and goldmarket in the European debt crisis. However, the volatility spillover effect between thecurrency and gold market is obvious in economic crisis, while in an steady economicperiod,the effect is not obvious.Time-varying Copula function can flexibly select the particular form of marginaldistribution assets and consider the time-varying correlation between the marginaldistribution assets and the structure with them separately. Moreover, it can capturenonlinear and asymmetric dynamic relationship in variables. EVT extreme value theorywhich can overcome disadvantages of other methods is an effective way to measuremarket risk in extreme circumstances. CoVaR not only can measure the risk of a singlefinancial market overflow, but also be able to capture the change of the systematic risk,is a comprehensive and effective risk management techniques. This paper studies therisk spillover effect between the the stock market in China’s mainland, Hong Kong, theUnited States and Europe by using EVT-Copulas-CoVaR model which made up ofextreme value theory, CoVaR model and Copulas.The empirical results show that the U.S. stock market has one-way risk spillovereffect on China’s stock market, followed by the Hong Kong stock market,whereas thespillover effects of European stock is very limited. The impact that China’s mainland andHong Kong stock market brings to European and the United States stock market is veryweak. However, the two-way spillover effect is significant between The U.S. andEuropean stock market. This means that we can predict the extreme risk of Chinesestock market by analyzing the history of the U.S. stock market.Overall,the thesis proposes workable and feasible policy recommendations andpoints out the future research work by summarizing the research content of the wholedissertation.更多还原