基于时变Copula与MCMC方法的债券市场风险实证研究

发布时间：2018-01-30 08:49

本文关键词： 时变Copula函数 MCMC方法债券市场 VaR Monte Carlo模拟　出处：《广西师范大学》2013年硕士论文　论文类型：学位论文

【摘要】：在金融危机背景下,虽然当前全球经济缓慢复苏,欧洲债务危机整体形势也日趋稳定,但全球经济和金融市场前景依旧不明朗。相对于我国持续低迷的股票、基金市场,国内债券市场却在迅速回涨。近年来,我国债券的发行期数和规模均呈现了较快的增长态势,然而在日益加剧的金融风险背景下,债券价格波动始终是一把双刃剑。债券市场的风险管理已经成为金融机构和投资者所面临的最重要问题之一。针对目前我国不断快速发展的债券市场以及债券指数走势相应的逐步转暖,债券组合风险的研究就显得非常具有现实意义。另一方面,Copula理论的出现并被引入金融研究领域,为弥补传统风险价值度量方面存在的一些缺点和不足提供了一条有效的途径。在已有的Copula理论和MCMC方法应用研究基础上,鉴于大都单独采用其中一种方法,而少有把两种方法结合起来进行应用研究,本文计划将时变Copula与MCMC方法结合起来研究我国债券市场风险。论文结合我国债券市场的实际特点,选取国债和企债指数,对债券投资组合建立了合适的时变Copula-GARCH-t模型,并根据MCMC方法确定了债券组合相应的配置比例,然后计算债券投资组合的VaR。最后与传统的静态常相关模式下Copula函数度量VaR进行分析比较,得出的结论是时变Copula-MCMC方法,不仅在组合配置的稳定性和精确度上要优于传统方法,而且能更准确的量化投资组合的市场风险,对投资组合风险研究效果有良好的改善,具有科学性和实效性。本文的特色主要体现在如下几个方面： 1.采用GARCH-t模型估计Copula函数的边际分布,即单变量债券收益率序列的分布,并且选取时变对称的Joe-Clayton Copula函数来描述债券间的相关结构,比较符合现实债券市场变量间相关性随时波动的特征。 2.根据资产收益率胜出的先验概率,结合贝叶斯推断中的MCMC方法确定债券组合的配置比例,在组合配置的精确度和稳定性上要优于传统方法。
[Abstract]:In the context of the financial crisis, although the global economy is recovering slowly and the overall situation of the European debt crisis is becoming more stable, the prospects of the global economy and financial markets are still uncertain. In the fund market, the domestic bond market is rising rapidly. In recent years, the number and scale of bond issuance in China have shown a rapid growth trend, but in the context of increasing financial risks. Bond price fluctuation is always a double-edged sword. Risk management in bond market has become one of the most important problems faced by financial institutions and investors. The number of trends corresponding to the gradual warming. The study of bond portfolio risk is of great practical significance. On the other hand, the emergence of Copula theory was introduced into the field of financial research. In order to make up for some shortcomings and shortcomings of traditional risk value measurement, this paper provides an effective way. On the basis of the existing Copula theory and MCMC method application research. In view of the fact that one of the methods is used alone and few of the two methods are combined for application, this paper plans to combine the time-varying Copula method with the MCMC method to study the risk of bond market in China. According to the actual characteristics of the bond market in China, the paper establishes a suitable time-varying Copula-GARCH-t model for the bond portfolio by selecting the bond bond and enterprise bond index. According to the MCMC method, the corresponding allocation ratio of bond portfolio is determined. Then the bond portfolio VaR is calculated. Finally, it is compared with the traditional Copula function metric VaR under the static constant correlation model. The conclusion is that the time-varying Copula-MCMC method is not only superior to the traditional method in the stability and accuracy of portfolio allocation, but also more accurate to quantify the market risk of portfolio. Research on portfolio risk has a good effect, scientific and effective. The main features of this paper are as follows: 1. Using GARCH-t model to estimate the marginal distribution of Copula function, that is, the distribution of univariate bond yield series. Moreover, the time-varying symmetric Joe-Clayton Copula function is used to describe the correlation structure between bonds, which accords with the characteristics of volatility between variables in the real bond market. 2. According to the priori probability of asset yield winning and the MCMC method in Bayesian inference, the allocation ratio of bond portfolio is better than the traditional method in terms of accuracy and stability of portfolio allocation.
【学位授予单位】：广西师范大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：F224;F832.51

【参考文献】