【作者】 李刚；

【导师】 李卫兵；

【作者基本信息】 华中科技大学 ， 西方经济学， 2013， 硕士

【摘要】 近年来，伴随着信息技术的发展，全球化进程的脚步也越来越快，经济全球化使得各个国家和各个市场之间的联系更为紧密，单一市场波动所带来的影响能够轻而易举地跨越市场、跨越国界，对处在全球一体化下的其他国家和市场产生不同程度的影响。而股票市场作为支持一国经济发展并控制国家经济命脉的最重要的资本市场，其重要性不言而喻。因此，研究股票市场的风险并以此来防范其对国家经济建设可能带来的不利影响就显得尤为重要。理论部分首先介绍了金融风险的定义和分类，其中VaR因其容易理解、易于操作且能够量化风险等诸多特点现已成为风险监管机构主流的风险测度方法，即便VaR方法有着这些优点，但与此同时其也有不足的地方，于是在此基础上提出了Copula-VaR方法来对股票市场进行风险分析。其次对Copula函数的性质和不同类型Copula函数的特征进行了详细的介绍，因为Copula在测度资产组合的风险时是不需要资产收益率服从正态分布的假设前提，而且其在结构上能较好的拟合资产之间的联合分布，并且也能消除单纯的VaR因不能解决资产之间的非线性相关性所引起的不合理测度风险等特点，因此Copula-VaR方法能得到更加精确的投资组合VaR值。实证部分选择我国股票市场的上海综合指数和深圳成分指数进行研究分析。根据样本数据联合密度函数所具有的特征，本文选用的是具有尾部对称特征的二元t-Copula函数作为上证指数和深成指数的联合分布函数，利用Copula函数对其各自变量的求偏导后服从0-1均匀分布的特点，用蒙特卡洛模拟法求出了在不同置信水平下的VaR。最后，变动投资组合中上证指数和深成指数的比例就得到了一组VaR值并将得到的结果绘制成曲线，随着模拟次数的增加，得到了近似直线的投资比例和组合的VaR值，直线的斜率即是单位上证指数百分比变动所引起的组合VaR值变化，也就是全文的结论。文章中所用到的Copula方法是最近发展起来的一种更加精确的用来测度相关性的方法，不仅适用于股票市场，对于其他资本市场同样适用，另外，这种方法还可以解决多变量相关性的问题。更多还原

【Abstract】 In recent years, with the development of Information&Technology, the steps of theglobalization process is becoming more and more quickly. Economic globalization makesthe connection between countries and markets more closely, the effects caused byfluctuations of the single market can easily exports to the other countries which are underthe same economic environment. The stock market as the most important capital market toa country, whose importance is self-evident, because it supports a country’s economicdevelopment and controls the country’s economic lifeline. Therefore, it is of vitalimportance to research the changing of the stock market’s risk and prevent the adverseimpact to the economic construction of a country.In the theory section,I introduce the definition and classification of financial riskfirstly,and VaR is becoming a mainstream risk measurement all over the world because ofits easily being understood, operated and being able to quantify risk and so on. Even if theVaR method has these advantages, but it also with its own deficiencies at the same time.Based on that I introduce the Copula-VaR method to analyze the VaR of the stock market.Secondly, introduce the theory of the Copula and characteristics of different types ofCopula functions in detail.Because using copula measure the risk of the portfolio risk hasits own features,such as the return of the assets are not necessarily obey the normaldistribution hypothesis, and it can better fit the joint distribution between the assets and itcan also eliminate the unreasonable risk measurement caused by the nonlinear correlationbetween assets and so on. So the Copula-VaR method can obtain more accurate VaR of theportfolio.In the empirical section, I choose Shanghai composite index and The Shenzhencomponent index of our country for this paper. According to the sample data’s feature of thejoint density function, I choose the tail symmetry characteristics of binary t-Copulafunction as the distribution function of Shanghai composite index and Shenzhen component index. Taking the advantage of the partial derivatives of the copula function obey theU[0,1],then I make use of the Montecarlo simulation method work out the VaR underdifferent confidence level. Finally, by changing the weight of Shanghai composite indexand Shenzhen component index in the investment portfolio we can get a set of VaR value,then we can draw a curve according to the result. With we increase the number ofsimulation, we can obtained the approximate linear about proportion of the investment andportfolio VaR. The linear’s slope is the changing portfolio VaR caused by the unitpercentage change in the Shanghai composite index, it is also the conclusion of the paper.The Copula method,developed recently, used by this paper is a accurate method ofmeasuring correlation, it is not only applied to the stock market, it can also apply to othercapital market, in addition, this method can solve multivariate correlation.更多还原