GJR-Copula模型在投资组合的风险管理中的应用

发布时间：2018-02-03 18:28

本文关键词： GJR-Copula 投资组合非对称性风险度量　出处：《中央民族大学》2013年硕士论文　论文类型：学位论文

【摘要】：近些年来,国际金融形势发生了深刻的变化。金融市场的波动愈加频繁,危机经常发生,对风险的度量已成为行业关注的话题之一。传统的风险度量手段建立在马克维茨的投资组合理论的基础上,资产组合之间为线性相关关系,资产组合的收益率服从多元正态分布,使用VaR指标对单一资产和资产组合进行风险度量。但大量的事实证明,金融资产的收益率存在明显的非正态的尖峰厚尾的特征,资产之间的相关关系非线性,传统的正态模型会低估资产风险。另一方面,VaR方法存在一定的建模缺陷,不能满足管理者对风险控制的需要。因此,本文引入Copula模型和CVaR技术,对于单个金融资产的收益率,CVaR方法考虑到了其尾部损失的均值,更加适于测度风险；对于投资组合联合收益率,建立Copula模型基于金融资产的非线性相关性,不限制边缘分布从而构建了相应的的联合分布损益函数。本文首先讨论了Copula模型在金融市场尾部相关性度量上的应用；接着基于VaR与CVaR的风险测度方法,依次运用历史模拟法、方差协方差法、蒙特卡罗模拟法和极值理论测度道琼斯工业指数和香港恒生指数的VaR和CVaR,结果表明,四种方法的计算的VaR都低估了实际股指的风险。然而,CVaR的失败率大大降低,提高了对未来变动的预测准确度。考虑到资产收益率这一时间序列的非对称的时变特征,文章对单个股指的收益率序列建立GJR模型,并在此基础上引,Copula方法,建立GJR-Copula模型,度量投资组合的VaR和CVaR,并通过返回检验验证模型。结果证明,GJR-Copula模型的CVaR可以精准地测度收益率具有非对称性的资产的风险价值以及投资组合的风险价值,为风险管理着的投资决策提供信息支持。
[Abstract]:In recent years, the international financial situation has undergone profound changes. The traditional risk measurement method is based on Markowitz's portfolio theory and the relationship between portfolio is linear. The return rate of portfolio is from multiple normal distribution, using VaR index to measure the risk of single asset and portfolio. But a lot of facts prove that. The return rate of financial assets has the characteristic of non-normal peak and thick tail, the correlation between assets is nonlinear, the traditional normal model will underestimate the risk of assets. VaR method has some defects in modeling and can not meet the needs of risk control. Therefore, this paper introduces Copula model and CVaR technology to the return rate of a single financial asset. The CVaR method takes into account the mean value of its tail loss and is more suitable to measure risk. For the joint return rate of portfolio, the Copula model is based on the nonlinear correlation of financial assets, and the corresponding joint distribution profit and loss function is constructed without limiting the edge distribution. This paper first discusses the application of Copula model to the measurement of tail correlation in financial markets. Then the risk measurement method based on VaR and CVaR, using historical simulation method, variance covariance method in turn. Monte Carlo simulation and extreme value theory are used to measure the VaR and Cvar of the Dow Jones Industrial Index and the Hang Seng Index in Hong Kong. The results show that the VaR calculated by the four methods underestimate the risk of the actual stock index. The failure rate of CVaR is greatly reduced, which improves the accuracy of predicting future changes, considering the asymmetric time-varying characteristics of the time series of asset return. This paper establishes the GJR model for the return sequence of a single stock index, and on the basis of this, establishes the GJR-Copula model to measure the VaR and CVaR of the portfolio by using the Copula method. The results show that the CVaR of GJR-Copula model can accurately measure the risk value of assets with asymmetric return rate and the risk value of portfolio. Provide information support for risk-managed investment decisions.
【学位授予单位】：中央民族大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：F224;F830.59;O211.67

【参考文献】