The VaR Method Based On Fluctuate Model And The Empirical Research Of The Financial Market In China

Financial risk is due to the fluctuation of financial assets. Therefore the core of riskmeasure is to estimate and forecast the fluctuation of price. The estimation of fluctuation rate,which methods have gained great development, all along is one of the most active fields todemonstrating finance and computation economics in the past.In this thesis, the distributing characters of the rate of return serial of SHI and SZI arestudied, and the basic statistical characters of the fluctuation in stock's return rate, whichinclude non-normal, autocorrelation and stationary, are analysed by descriptive statistics, suchas mean, skewness and kurtosis. VaR of SHI and SZI under two error distributing and twobelieve levels is calculated by combination GARCH, which is compared with the VaR of SHIand SZI in the same case by GARCH model, EGARCH model, TGARCH model. Finally, theconclusion that the combination GARCH model is more effective to calculate the VaR thanothers.There are five parts in this paper. In the first part, we introduce developmentalbackground of the finance risk value and the current study at home and abroad. In the secondsection, we introduce the fluctuating estimate theory and the VaR calculated method. In thethird section, we study the distribution character of the returning rate's serial about SHI andSZI, and show that they are all skewness, kurtosis and non-normal. Through the test we getthat their rate of return aren't autocorrelation. But they are stationary serial. In the fourthsection, four models are established, by which based on t distributing and GED distributingVaR of SHI and SZI are individually calculated under believe level being ninety- nine percentand ninety-five percent. The risk measure accuracy of VaR are compared and the models'validity are tested. The last part is a conclusion. Fluctuating models are all effective. Thecombination GARCH model is more effective to calculate the VaR than others.