Research On The CVaR Of Stock Market Risk Based On The Semiparametric GARCH And Copula Connect Function

Establishing a suitable risk measurement model has always been the key and difficult points in financial research.This thesis takes the daily returns of the CSI 300 Index and HSI Index as the research sample.The semi-parametric GARCH model for one indices and the Copula-semi-parametric GARCH model for the two indices are established respectively.Then the VaR and CVaR of are calculated.The thesis is divided into six chapters,and the main work is carried out from three levels.The first level(chapter 2):Giving the theoretical basis and data analysis foundation.The theoretical basis includes GARCH model,semi-parametric GARCH model,Copula function,VaR and CVaR.The basic data analysis includes data selection,basic statistical analysis and statistical test.It IS found by analysis that the characteristics of the data are consistent with the semi-parametric GARCH model's characteristics.The second level(chapter 3 and chapter 4):The model is built to calculate VaR and CVaR based on the first level.Specifically,the semi-parameter GARCH model of CSI 300 and HSI are respectively established in chapter 3.Replacing the p and q parts in the form of GARCH(p,q),we obtain two different types of semi-parametric models,and the model parameters are estimated by two-stage iterative algorithm;Four prediction error values are used to test the model.Through R language software,the semi-parametric GARCH model with p as the non-parametric form is better;The ARCH coefficient of CSI 300 ? = 0.9214 × 10 is small,shows that the volatility of the return series on the impact-response is quick.The GARCH coefficient of HSI ? = 0.9212 is greater than the CSI 300 ? = 0.8501,indicates that the impact of the return series to the conditional variance is more durable.The chapter 4 is the establishment of the Copula-semi-parameter GARCH model.Using the peak method to segment the data,the binary normal Copula function was selected according to AIC and BIC criteria,and the two indexes have strong positive correlation.The third level(chapter 5):Calculate the CVaR of stock market risk based on the model constructed in the second level.This includes the calculation of CVaR of the single asset and the two portfolios.The former is calculated according to the semi-parameter GARCH model built in chapter 3,and the latter is calculated according to the Copula-semi-parameter GARCH model built in chapter 4;The VaR values of the samples and the extra-samples are estimated and predicted respectively.Through the test we conclude that the CSI 300 actual failure rate is greater than the given expectations of failure rate under the 99% confidence level,the portfolio and HSI actual failure rate is less than the given expectations of failure rate under the 95% confidence level.It shows the CSI 300 take is more appropriate to choose the confidence level of 95%,and the asset portfolio and HSI under the confidence level of 99% can more accurately measure the market risk.