基于Copula函数的我国创业板与主板市场风险关系研究

发布时间：2018-05-24 22:05

本文选题：主板 + 创业板　；参考：《北京工商大学》2014年硕士论文

【摘要】：2009年10月30日，创业板市场首批28家上市公司在深圳交易所正式挂牌交易，标志着我国创业板时代正式来临。与主板市场相比，创业板市场在上市企业规模、行业分布、投资者结构、退市制度等方面存在较大的区别，但两个市场均属于我国多层次证券市场的一部分，都要面临一些共同的影响因素，那么两个市场的走势是否相关，如果存在相关，那么在不同情形下相关程度如何，比如在暴跌或暴涨的情形下相关程度是否提高，这种“尾部相关性”，对于投资者和风险管理者是一个至关重要的问题。因此，本文对主板和创业板之间的相关结构进行深入研究。本文采用Copula函数技术对两个市场的相关结构进行分析，并且结合风险价值VaR进行风险预测能力分析。首先，本文分析了我国主板和创业板市场收益率各自的分布特征。在Q-Q图等检验基础上，通过对风险价值VaR的预测能力进行比较，发现GARCH-Normal模型不能很好的捕捉创业板市场的尾部特征，而较好的模型为GARCH-T模型。但在主板市场上，相比GARCH-T模型，简单的GARCH-Normal能够更准确的捕捉尾部特征。即与主板市场相比，创业板市场存在更厚的尾部。其次，在对主板和创业板市场收益率分布特征有效估计的基础上，对不同Copula函数进行估计。通过检验发现，混合Copula模型更适合用来描述我国主板和创业板市场之间的相关结构，两个市场之间整体具有正相关关系，但存在非对称的尾部相关，即两个市场在暴涨或暴跌的情况下，相关程度明显提高，并且在暴跌时的相关程度高于暴涨时的相关程度。并在Copula基础下，利用蒙特卡洛模拟方法对投资组合的VaR进行分析，发现无论在样本内还是样本外混合Copula的风险预测效果最优。
[Abstract]:On October 30, 2009, the first batch of 28 listed companies in the gem market were officially listed on the Shenzhen Stock Exchange, marking the formal arrival of the gem era in China. Compared with the main market, the gem market has great differences in the scale of listed enterprises, industry distribution, investor structure, delisting system and so on. However, the two markets are part of the multi-level securities market in China. All have to face some common influencing factors. Well, is the trend of the two markets relevant? if there is a correlation, what is the degree of correlation in different situations, for example, whether the correlation degree is increased in the case of a slump or a surge? This tail correlation is a crucial issue for investors and risk managers. Therefore, this article carries on the thorough research to the main board and the growth enterprise board correlation structure. In this paper, the Copula function technique is used to analyze the related structure of the two markets, and the risk forecasting ability is analyzed by combining the risk value VaR. First of all, this paper analyzes the distribution characteristics of the return rate of China's main board and gem market. On the basis of Q-Q chart and other tests, it is found that the GARCH-Normal model can not capture the tail characteristics of gem market well, and the better model is GARCH-T model by comparing the prediction ability of VaR with risk value. But in the main board market, compared with the GARCH-T model, the simple GARCH-Normal can capture the tail features more accurately. That is, compared with the main market, gem market has a thicker tail. Secondly, on the basis of the efficient estimation of the return distribution characteristics of the main board and the growth enterprise market, different Copula functions are estimated. It is found that the hybrid Copula model is more suitable to describe the correlation structure between the main board and the gem market in China. There is a positive correlation between the two markets, but there is an asymmetric tail correlation between the two markets. That is, the correlation between the two markets in the case of a sharp rise or fall, significantly increased, and the correlation in the collapse is higher than the correlation in the skyrocketing. On the basis of Copula, Monte Carlo simulation method is used to analyze the VaR of the portfolio, and it is found that the risk prediction effect of mixed Copula in and out of the sample is the best.
【学位授予单位】：北京工商大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：F832.51;F224

【参考文献】