VaR的计算及其在风险管理中的应用

发布时间：2019-02-22 10:24

【摘要】：VaR作为一种新兴的风险度量方法，较之传统的风险度量方法，，如情景模拟法、压力测试法、灵敏度方法等，因其方法直观、结果量化、易懂等特点，受到风险管理者们的青睐，也得到众多学者的研究。本文首先介绍了VaR的基本理论及一些传统的计算方法，随后，选择上证180指数、深圳成份指数和香港恒生指数为研究对象，对这些指数的收益率序列进行了基本的统计分析，表明这些市场的收益率序列具有尖峰厚尾性和具有ARCH效应，因此认为对这三个市场使用GARCH族模型是合适的。接着，使用GARCH模型和APARCH模型在正态分布、t分布、GED分布、偏态分布和偏态GED分布下来计算各个指数的VaR值，作了初步的模型和结果分析，同时，使用MCMC方法来估计GARCH模型的参数，并与传统的极大似然估计法来比较。在本文的最后，将多个回测方法结合在一起，从准确性、保守性和有效性等三个方面来评价各模型，得出的主要结论有：(1)由于通过三个证券市场的收益率数据所计算的APARCH模型的参数1均非零，且为正值，说明上海、深圳和香港证券市场都存在明显的“杠杆效应”。(2)利用MCMC方法来估计GARCH模型的参数，在不同置信水平下，对上证180指数和恒生指数进行VaR的计算，表明GARCH-N-MCMC模型的有效性比另外两个模型高，在模型准确性得到保证的前提下，使用该方法，可以让投资者有最小的准备金机会成本。(3) APARCH模型与GARCH模型的比较：当置信水平较高时，一方面，APARCH模型在准确性和有效性方面比GARCH模型有所提高，但同时APARCH模型也相对保守些；另一方面，基于偏态广义误差分布的APARCH模型更能捕捉到金融市场的各种特性，如波动聚集性，尖峰厚尾性等，且APARCH模型较GARCH模型更有效，采用APARCH模型结合偏态分布来分析计算这三个市场的VaR值效果更好。本文对VaR的计算方法作了一些尝试，期望能给风险管理者提供一些决策支撑。
[Abstract]:Compared with the traditional risk measurement methods, such as scenario simulation, stress test and sensitivity, VaR, as a new risk measurement method, is favored by risk managers because of its intuitive, quantitative and understandable characteristics. It has also been studied by many scholars. This paper first introduces the basic theory of VaR and some traditional calculation methods, then selects the Shanghai Stock Exchange 180 Index, Shenzhen component Index and Hong Kong Hang Seng Index as the research object, carries on the basic statistical analysis to these index return series. It is shown that the yield series of these markets have spikes and thick tails and have ARCH effect. Therefore, it is considered appropriate to use the GARCH family model for these three markets. Then, the GARCH model and APARCH model are used to calculate the VaR values of each index in normal distribution, t distribution, GED distribution, skew distribution and skewness GED distribution. The parameters of GARCH model are estimated by MCMC method and compared with the traditional maximum likelihood estimation method. At the end of this paper, several methods are combined to evaluate the models from three aspects: accuracy, conservatism and effectiveness. The main conclusions are as follows: (1) the parameters of the APARCH model calculated through the return data of the three securities markets are all non-zero and positive, indicating Shanghai. There are obvious "leverage effects" in both Shenzhen and Hong Kong stock markets. (2) the parameters of GARCH model are estimated by MCMC method, and the VaR calculations of Shanghai Stock Exchange 180 Index and Hang Seng Index are carried out under different confidence levels. It is shown that the GARCH-N-MCMC model is more effective than the other two models, and this method is used to ensure the accuracy of the model. (3) the comparison between APARCH model and GARCH model: when the confidence level is high, on the one hand, the accuracy and validity of APARCH model are higher than that of GARCH model. At the same time, the APARCH model is more conservative. On the other hand, the APARCH model based on skew generalized error distribution can capture the characteristics of financial market, such as volatility aggregation, peak and tail, and APARCH model is more effective than GARCH model. It is better to use APARCH model and skew distribution to analyze and calculate the VaR value of these three markets. In this paper, some attempts are made on the calculation method of VaR, which is expected to provide some decision support for risk managers.
【学位授予单位】：华南理工大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：F224;F830

【参考文献】