GARCH-VaR模型在我国ETF风险测量中的应用研究

发布时间：2018-04-29 03:36

本文选题：EFT + VaR　；参考：《辽宁科技大学》2012年硕士论文

【摘要】：ETF作为一种金融投资工具,存在的风险是不言而喻的,如何有效的规避和预测ETF市场风险就成为了一个亟需解决的问题。传统的风险测量方法,如标准差、β系数等方法不仅适用范围有限,并且适应的金融工具也非常少,因而不能全面的反映市场风险。本文从定量分析的角度,对ETF的市场风险进行研究。选取上证50ETF从发行日到2011年9月6日的每日成交价格,缺省数据用邻近的上日成交价格代替,选用研究对象的对数收益率作为样本数据。用EViews 6.0对时间序列进行处理,分析发现,上证50ETF的收益率具有尖峰厚尾性、不对称性、波动集聚性。据此,本文选择能够较好的描述厚尾性的t分布和GED假定,选择能够消除自相关性的GRACH族模型来拟合数据。运用EViews 6.0软件的proc功能直接得到各模型的条件方差,同时,运用MATLAB的逆累积分布函数值的计算功能和数值积分功能分别计算出t分布和GED的分位数,分别代入到VaR的计算公式中,得到上证50ETF在不同分布下的风险值。最后,选用Kupiec(1995)提出的失败率检验法,分别对GRACH族模型在厚尾分布下计算的VaR值和采用历史模拟法、蒙特卡罗模拟法计算的VaR值的准确性进行检验,认为GED下的GRACH(1,1)模型计算的风险值能较好的反映市场风险。
[Abstract]:As a financial investment tool, the risk of ETF is self-evident, how to effectively avoid and predict the market risk of ETF has become a problem that needs to be solved. Traditional risk measurement methods, such as standard deviation, 尾 coefficient and so on, are not only limited in scope, but also have very few suitable financial instruments, so they can not reflect market risk comprehensively. This paper studies the market risk of ETF from the angle of quantitative analysis. The daily transaction price of Shanghai 50ETF from the date of issue to September 6, 2011 is selected. The default data is replaced by the neighboring transaction price of last day, and the logarithmic rate of return of the research object is selected as the sample data. Using EViews 6.0 to deal with the time series, it is found that the return rate of 50ETF in Shanghai has the characteristics of peak, thick tail, asymmetry and volatility agglomeration. Based on this, this paper selects the t distribution and GED assumption which can describe the thick tail well, and selects the GRACH family model which can eliminate the autocorrelation to fit the data. The conditional variance of each model is obtained directly by using the proc function of EViews 6.0. At the same time, the calculation function of MATLAB's inverse cumulative distribution function and the function of numerical integration are used to calculate the t distribution and the quantile of GED, respectively. The risk values of 50ETF in different distributions of Shanghai Stock Exchange are obtained by inserting them into the formula of VaR. Finally, the failure rate test method proposed by Kupiecke 1995 is used to test the accuracy of the VaR value calculated by GRACH family model under the thick tail distribution and the VaR value calculated by historical simulation method and Monte Carlo simulation method, respectively. It is considered that the risk value calculated by the GED model can better reflect the market risk.
【学位授予单位】：辽宁科技大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：F832.51;F224

【参考文献】