Comparative Analysis on the Risk Measure of China Stock Market
|School||Shandong University of Finance and Economics|
|Keywords||Risk Measure MIDAS（Mixed Data Sampling）Model Copula Model EEC（EGARCH-EVT-Copula）Model|
With the deepening development of economic globalization and financial liberalization,tremendous changes have taken place in the financial system, the prevention and measureof financial risk is becoming more complex. The influences for the financial marketvolatility include the development of information technology, the application of financialengineering, financial innovation, and raising the market at effectiveness.The financial market of our country has embarked on a compatible with the economicdevelopment and scale is constantly expanding. The financial risk has been paid moreattention. So it is necessary to establish suitable risk measurement models to assess the riskof financial market. The risk of financial market has wide range. This paper is based on thestock market as the main object of study. The stock market is an important component ofthe financial market, the volatility of the financial market significantly. This paper measureand analyse the risk of the stock market using the Shanghai Index(000001) and ShenzhenComponent Index(399001).At first, descriptive statistics is used to analyze the Shanghai index return series.Theresults show that the return series exist in leptokurtic phenomenon, volatility clusterphenomenon, long memory, conditional heteroskedasticity and leverage effect. Secondly,this paper introduces the theory of MIDAS model and Copula model.We build the MIDASmodel and EEC model using the Shanghai Composite Index and Shenzhen ComponentIndex return series to measure the risk of the stock market.The research results of this paper are mainly in the following three aspects. Firstly, thispaper measures of individual stock’s volatility combined with realized volatility of highfrequency data.The result confirmed that the MIDAS model can preferably describe stockmarket volatility. Secondly, bivariate normal t-Copula model can better fit the return seriescompared to bivariate normal Copula model. Thirdly, using high frequency data of MIDASmodel of risk measure result is superior to EEC model.