基于VaR—GARCH类模型的我国ETF市场风险管理研究.doc

内容摘要：自2004年末国内首只ETF—上证50ETF推出以来，11年间，ETF在我国不断发展，市场集中度日益提高。上证50ETF作为投资指数的一种工具,必然面临大盘涨跌带来的市场风险,所以很有必要深入研究其市场风险。本文在充分考虑金融时间序列“尖峰厚尾”特性、非对称效应的基础上，选取上证50ETF最新样本数据进行实证研究，构建基于三种分布下的GARCH族模型共9种，并分别测算风险价值VaR，运用Kupiec失败率检验法回测检验VaR，评估VaR参数法的准确性。经比较初步得出，当选择95%作为置信水平时，VaR-GARCH模型可以较为准确地衡量上证50ETF收益率波动的最大损失，其中尤以TGARCH(1,1)—GED模型表现最优。从而据此提出一些有关ETF市场风险控制的建议和措施，以期丰富我国ETF市场风险管理研究，推动我国ETF的健康发展。 关键词：上证50ETF 市场风险 VaR GARCH族模型 Abstract: Since the first ETF—the 50ETF in Shanghai Stock Exchange launched in China in late 2004, ETF has been enjoying rapid development and high market concentration in the past 10 years. As a tool which focuses on the index investment, SSE 50ETF is bound to undertake the market risk caused by the stock market’s fluctuations, so it is of great necessity to implement deep researches into it. Considering financial time series data’leptokurtic and asymmetric features, the paper chooses the latest data of SSE 50ETF as the sample data to calculate VaR, by using parametric method and building up 9 types of GARCH-models under three different distributions in empirical research. Then the paper uses Kupiec failure rate testing to evaluate the accuracy of VaR-GARCHs. According to the empirical research results, we can preliminarily draw the conclusion that VaR can measure the largest losses of the return volatility of SSE 50ETF effectively, under 95% confidence level, in particuar the TGARCH(1,1)—GED model performs best. Basing on the conclusion, the paper expects to offer some feasible suggestions on market risk control and management.Thus improving the system of market risk management of SSE 50ETF and promoting healthy development of ETFs. Key words: SSE 50ETF market risk VaR GARCH-models 目 录 一、引言 3 （一）研究背景 3 （二）研究意义 4 二、理论基础 4 （一）VaR理论 5 （二）残差的三种分布和GARCH族模型 6 （三）VaR-GARCH族方法和准确性检验 8 三、实证研究 9 （一）样本数据及其属性检验 9 （二）GARCH族模型的建立与对比分析 15 （三）求解