VaR方法在沪深300中的实际应用与研究

发布时间：2018-06-11 11:56

本文选题：VaR + 风险管理　；参考：《首都经济贸易大学》2013年硕士论文

【摘要】：本文从VaR的三种基本的计算方法—参数法，历史模拟法，，蒙特卡洛模拟出发引出了第四种VaR计算方法---半参数法，然后结合了我国目前唯一的股指期货品种—沪深300，在显著性水平5%，持有期一天的情况下，使用了10种VaR的计算方法构建了19个不同的VaR模型。最后，用Kupiec检验法和VaR描述统计量研究了这19个VaR模型的效果及适用条件。首先，本文的检验结果表明，没有一个模型能在各个指标上都表现良好，每个模型都存在某种程度的缺点。这意味着，实际应用中要针对不同的情况选择不同的模型来进行风险度量测算。其次，本文的理论和实证分析表明了得下属一般性的结论：（1）就精度而言，蒙特卡洛方法和EVT方法最高，GARCH方法次之，历史模拟法最差。（2）就金融机构的风控管理成本而言，蒙特卡洛方法较高，历史模拟法、GARCH方法以及EGARCH法较低；加权历史模拟法权重越低模型的风险管理成本越高。EVT模型的阀值越高模型的风险管理成本越高。（3）对于大型金融机构，蒙特卡洛方法和EVT方法更适合；对于中型的金融机构，基于GED分布的GARCH模型或EGARCH模型更适合；对于小型金融机构，历史模拟法较适合。最后，本文仅仅对基本的GARCH模型，蒙特卡洛模拟，历史模拟法等共11种方法进行理论分析和实证研究，并未对基本方法混合改进后的衍生方法如GARCH-蒙特卡洛模拟等方法进行实证研究，如果能在计算方法的丰富性做进一步的改进，则还可能获得更加有意义的结果。
[Abstract]:In this paper, the fourth calculating method of VaR, the semi-parametric method, is derived from the three basic calculation methods of VaR, namely, parameter method, historical simulation method and Monte Carlo simulation method. Then combined with the only stock index futures in China-Shanghai and Shenzhen 300, in the case of significant level of 5, holding a day, using 10 VaR calculation methods to construct 19 different VaR models. Finally, Kupiec test method and VaR description statistics are used to study the effects and applicable conditions of these 19 VaR models. Firstly, the test results of this paper show that none of the models can perform well on every index. Each model has a certain degree of disadvantage. This means that different models should be chosen to measure the risk in practical application. Secondly, the theoretical and empirical analysis of this paper shows that the general conclusion: 1) in terms of accuracy, Monte-Carlo method and EVT method are the highest GARCH method, and historical simulation method is the worst. In terms of the cost of wind control and management of financial institutions, Monte-Carlo method is higher, historical simulation method is lower than GARCH method and EGARCH method. The lower the weighted historical simulation, the higher the risk management cost of the model. The higher the threshold value of the EVT model, the higher the risk management cost of the model.) for large financial institutions, the Monte Carlo method and the EVT method are more suitable; for medium-sized financial institutions, the Monte Carlo method and the EVT method are more suitable. The GARCH model based on GED distribution or EGARCH model is more suitable; for small financial institutions, historical simulation is more suitable. Finally, this paper only for the basic GARCH model, Monte Carlo simulation, There are 11 methods for theoretical analysis and empirical research, such as historical simulation and so on, and no empirical research has been carried out on the basic methods, such as GARCH-Monte Carlo simulation and so on, after mixed and improved derivation methods of the basic methods, such as GARCH-Monte Carlo simulation, etc. If further improvements can be made in the richness of the calculation methods, more meaningful results may be obtained.
【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】：2013
【分类号】：F224;F832.51

【参考文献】