Risk Measurement And Influencing Factors Analysis Of Chinese Stock Market

With the rapid development of stock market in China,a relatively complete framework of it has already been established so far.However,owing to various factors,the stock market prices fluctuate greatly,so the greater importance has been attached to the stock market risk measurement.Therefore,this article focuses on building a quantified model of stock market risk based on the price of the Shanghai and Shenzhen 300 Index,an important sector of the Chinese stock market.The article mainly includes the influencing factors of China’s stock market risk and the comparison of different value at risk(VaR)prediction methods.There is little research and application on the stock market VaR influencing factors analysis and the quantile estimation of linear models with GARCH-X error.Therefore,in the first section,this article introduce the possible risk factors affecting stock market to the model based on the literature research results,and then establish a linear regression model including GARCH-X error terms,which means that the realized volatility is added as a covariate in the GARCH volatility model.Parameter estimation of QR-GARCHX model is realized by two-step estimation and joint estimation,and then the estimated and predicted VaR of CSI 300 assets at a certain level can be calculated.Given the 5%quantile,it can be inferred from the parameter estimation results that high-frequency intraday trading information on the stock market will have a significant impact on stock market fluctuations.The change rate of the exchange rate of the US dollar against the RMB is negatively correlated with stock market risk,and the change rate of interest rates is positively correlated with stock market risk.With further investigation on the estimated values of various parameters at different quantiles,it was found that realized volatility can capture stock market volatility under conditions other than extreme risk appetite,the exchange rate has a stable negative correlation with stock market risk and there is no obvious correlation with investors ’risk appetite.The effect of interest rates on stock market risk has a strong correlation with investors’risk awareness:the more cautious the investor’s attitude towards risk,the more likely that the change of interest rate will cause China’s stock market to fluctuate greatly.The characterization of stock market risk under extreme risk aversion is difficult.Comparing the parameter estimates of stock indexes of different sizes,it is found that the realized volatility is more suitable for capturing the volatility of large-scale stock indexes.The impact of the exchange rate on medium and large market value stock indexes is opposite to that of small market value stock indexes,and the significance of its impact declines as the market value of stock indexes decreases.The interest rate has a positive impact on the risks of the three scale indexes,but the impact on small-scale stock indexes is slightly greater than that on large-scale stock indexes.Finally,the significant influencing factors of the risk of CSI 300 industries are given.It’s shown that the realized volatility has a significant impact on the volatility of all the industries,the significant influencing factors of different industries are quite different while different explanatory variables have the same impact on various industries.The second part of this article comprehensively analyzes the effect of QRGARCHX method in China’s stock market risk measurement.The rolling window prediction method is adopted.At the same time,five models including Risk Metrics,HS(Historical Simulation),FHS(Filtered Historical Simulation),QR-GARCH,and QR-GARCHX are used to realize the prediction of 5%VaR.Finally,the prediction results are compared horizontally and vertically and the comparison between the VaR and the yield series is used to show the prediction effect.At the same time,the necessity of introducing a conditional heteroscedastic structure is illustrated intuitively through the diagram.After comparing the empirical coverage ratio and backtesting results of each model vertically,it’s concluded that the four models have passed two VaR backtesting,and QR-GARCHX method has obtained the most accurate empirical coverage rate.The QR-GARCHX prediction effect has been greatly improved then QRGARCH,which indicates the effectiveness of introducing realized volatility as a covariate.