开放式基金风险价值测度模型的优化与实证

发布时间：2017-12-30 21:10

本文关键词：开放式基金风险价值测度模型的优化与实证　出处：《首都经济贸易大学》2014年硕士论文　论文类型：学位论文

【摘要】：我国开放式基金的风险管理需要更高的风险测量技术。VaR是目前金融风险度量的主流指标，其计算有很多种方法。Copula模型可以刻画变量间非线性、非对称的相关结构，并且不限制边缘分布的类型，使得基于Copula理论的风险价值测度模型相比传统多元分布模型测度更能准确的测量投资组合的风险价值，本文针对现有的GARCH-Copula模型，通过引进高斯核函数估计和GPD模型，提高对多元变量边缘分布的尖峰厚尾特征拟合的准确程度，，并考虑了波动集聚性的杠杆效应，构建了TARCH-GPD-Copula优化模型。本文提出的优化模型保留了原模型能够刻画边缘分布的波动集聚性的优点，同时弥补了原模型不能充分刻画边缘分布尖峰厚尾特征的缺陷，提高了对多元变量联合分布的拟合准确性，从而能够更加准确的计量投资组合资产的VaR值。优化模型的理论意义还在于具有普遍适用性，适用于各类金融资产投资组合风险价值的计算。本文首先从我国开放式基金风险管理现状出发，引入Copula理论和极值理论，提出采用Copula理论研究风险价值测度的必要性。其次在金融风险测度的框架下，深入分析了GARCH-Copula模型优缺点，在此基础上提出测度VaR的TARCH-GPD-Copula优化模型，并给出其求解步骤及源程序。本文实证研究中，以华夏大盘精选混合开放式基金前十只重仓股投资组合为研究对象，运用TARCH-GPD-Copula优化模型，以蒙特卡洛模拟方法计算其VaR值。同时对比计算了GARCH-Copula模型、方差协方差法、历史模拟法下投资组合的VaR值。通过Kupiec检验比较这四种模型的计算的VaR值的有效性。结果表明TARCH-GPD-Copula优化模型样本外数据测试效果最好，估计的风险价值最接近真实风险价值，失败率最低。实证结果与理论一致，从实证角度验证了理论推测的正确性，希望能为我国的基金管理公司或基金监管部门对开放式基金投资组合进行风险测度提供参考依据。
[Abstract]:China's open-end fund risk management needs a higher risk measurement technology. VaR is the mainstream index of financial risk measurement. There are many methods. Copula model can describe the nonlinear, asymmetric correlation structure between variables, and does not limit the type of edge distribution. The risk value measurement model based on Copula theory can measure the risk value of portfolio more accurately than the traditional multivariate distribution model. Based on the existing GARCH-Copula model, Gao Si kernel function estimation and GPD model are introduced to improve the accuracy of feature fitting for the edge distribution of multivariate variables. The leverage effect of volatility agglomeration is considered. The optimization model proposed in this paper retains the advantage that the original model can describe the volatility agglomeration of the edge distribution. At the same time, it makes up the defect that the original model can not fully describe the sharp and thick tail features of the edge distribution, and improves the accuracy of fitting the joint distribution of multivariate variables. Therefore, the VaR value of portfolio assets can be measured more accurately. The theoretical significance of the optimization model lies in its universal applicability, which is applicable to the calculation of portfolio risk value of all kinds of financial assets. This paper firstly introduces Copula theory and extreme value theory from the current situation of risk management of open-end funds in China. The necessity of using Copula theory to study risk value measurement is put forward. Secondly, under the framework of financial risk measurement, the advantages and disadvantages of GARCH-Copula model are deeply analyzed. On this basis, the TARCH-GPD-Copula optimization model of measure VaR is proposed, and its solving steps and source program are given. In this paper, we use TARCH-GPD-Copula optimization model to study the portfolio of the top ten heavy stocks in China's large market selected open-end funds. Monte Carlo simulation method is used to calculate the VaR value, and the GARCH-Copula model, variance covariance method, is compared and calculated. The VaR value of the portfolio under the historical simulation method. The validity of the calculated VaR value of the four models is compared by Kupiec test. The results show that the TARCH-GPD-Copula optimization model is effective. The best test effect is out of sample data. The estimated value of risk is closest to the real value of risk, and the failure rate is the lowest. The empirical results are consistent with the theory, which verifies the correctness of the theoretical speculation from the empirical point of view. This paper hopes to provide a reference for fund management companies or fund regulators to measure the risk of open-end fund portfolio.
【学位授予单位】：首都经济贸易大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：F224;F832.51

【引证文献】