基于Copula变点原理金融危机传染性的实证分析

发布时间：2018-06-05 02:07

本文选题：变结构copula + 金融危机传染　；参考：《南京大学》2014年硕士论文

【摘要】：对国内外金融领域的研究者来说,金融危机传染是一个重要的课题,要深入研究金融危机传染效应,必然要熟悉金融市场内部变量的相关性,Copula函数是研究金融变量之间关系的一个重要工具。本文主要研究了Copula理论在金融危机传染问题中的应用,在系统地介绍了Copula理论之后,对九组不同的股指序列构建了基于核密度估计的Copula模型,并具体讨论了各股指序列的分阶段建模问题。本文的核心工作可概括如下：首先采用核密度估计方法对10个国家(地区)的典型股票指数的对数变换序列进行边缘分布估计,并采用K-S检验及Box-Ljung独立性检验确定这些核密度估计值,是否满足Copula函数边缘分布的要求；然后根据平方欧式距离法和条件分布函数K-S检验法确定最佳Copula模型,对模型中的参数和相依系数值进行了估计；最后使用极大似然法对各股指收益率相依性的Copula函数进行变点检测,依据变点的位置将时间序列分段,并得到每个时间段中Copula函数的参数及尾部相关系数值。本文以2006年1月3日-2014年1月10日期间的美国标准普尔500指数(SP500)、俄罗斯RTS指数、韩国KOPSI指数、香港恒生指数、沪深300指数、孟买SENSEX300指数、日经225指数、希腊ASE指数以及英国金融时报指数(FTSE100)每日的收盘价数据为实证分析对象,对SP500指数与其他股指直接的相依性关系和金融危机的传染性进行了深入的测算。本文的主要研究结论：(1)SP500指数与其他各股指(恒生指数除外)的变化趋势都是一致的,除中国内陆和日本之外,其他国家(地区)和美国股市变动都显示了很强的相关性。就Copula函数尾部相关系数结果来看,德国、俄罗斯、韩国、英国和希腊各国与美国股市变动的协同性较强,一方股市的金融危机很容易传染到另一个股市中；(2)九个国家(地区)的指数收益率与SP500指数收益率之间相依结构都存在变结构点,且这些变点发生的时刻和全球危机产生影响的时间基本上是一致的,这些国家(地区)或多或少地受到美国次债危机的影响和冲击,而欧债危机对各国(地区)的影响有限。
[Abstract]:The contagion of financial crisis is an important subject for the financial crisis researchers at home and abroad. To study the effect of the financial crisis, we must be familiar with the correlation of the internal variables in the financial market. The Copula function is an important tool for the study of the relationship between financial variables. This paper mainly studies the Copula theory in the contagion of the financial crisis. In the application of the problem, after systematically introducing the Copula theory, the Copula model based on the kernel density estimation for nine different stock index sequences is constructed, and the stage modeling of each stock index sequence is discussed. The core work of this paper can be summarized as follows: first, the kernel density estimation method is used for the typical shares of the 10 countries (regions). The logarithmic transformation sequence of the ticket index is estimated, and the K-S test and Box-Ljung independence test are used to determine the kernel density estimates and whether the requirements of the edge distribution of the Copula function are satisfied. Then the optimal Copula model is determined by the square Euclidean distance method and the conditional distribution function K-S test. According to the value of the system, the Copula function is detected by the maximum likelihood method, the time series is segmented according to the location of the variable points, and the parameters of the Copula function and the tail correlation coefficient in each time period are obtained. This paper is based on the American Standard of January 3, 2006 -2014 January 10th. The SP500 (SP500), the Russian RTS index, the Korean KOPSI index, the Hongkong Hang Seng Index, the Shanghai and Shenzhen 300 index, the Mumbai SENSEX300 index, the Nikkei 225 index, the Greek ASE index and the UK Financial Times Index (FTSE100) daily closing price data are the empirical analysis of the direct relationship between the SP500 index and other stock indexes and the financial crisis. The main conclusions of this paper are as follows: (1) the SP500 index is consistent with the other indexes (except the Hang Seng Index). Except for inland and Japan, the changes of other countries (regions) and the US stock market are strongly correlated. The results of the tail correlation coefficient of the Copula function Look, the countries of Germany, Russia, Korea, Britain and Greece have strong synergies with the changes in the US stock market, and the financial crisis of one stock market is easily transmitted to another stock market. (2) there is a variable structure point between the index return and the SP500 index yield of the nine countries (regions), and the time and the global change occur. The time for the impact of the crisis is basically consistent, and these countries (regions) are more or less affected by the American subprime crisis, and the European debt crisis has a limited impact on countries (regions).
【学位授予单位】：南京大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：F224;F831.5

【参考文献】