Systemic Risk Measurement Of Shanghai Stock Market

With the development of market economy,the securities market has gradually improved and diversified.Various of market risks has become increasingly prominent.Therefore,the management and control of the risk in securities market has become very important.As an important part of the securities market,the stock market risk also affects everywhere in the whole market.Therefore,the study of the stock market risk can not only gain insight into the characteristics and laws of stock market risk,but also contribute to financial institution and investors controlling market risks reasonably.In the study of market risk,the traditional VaR method is a technology which is in wide use currently.But there are several assumptions in the traditional VaR methods do not match the reality,and obvious limitations in the calculation of multiple period risks.Early VaR model assumes a normal distribution on financial time series,however,most of the actual sequence of financial data exist "fat tail" nature and does not follow a normal distribution.At the same time,for a holding interval of the multi-period VaR calculation,the traditional econometric model mainly assumes that there is a simple linear relationship between multi-period VaR and single-period VaR.But reality is not in that case for the complexity of the financial time series.In order to measure the risks of the stock market better,we need to adopt broader the restrictions of the model.This paper used a semi-parametric quantile regression method on value at risk(Value at risk,VaR)measurement model as the main research method to study Shanghai stock market risk,which is based on the traditional VaR method and considering the "fat tail",biased and volatility clustering of financial return series.The specific study process is as follows:This paper selected the Shanghai Composite Index daily closing price index,from January 2,1991 to March 18,2016,a total number of 6166 samples,and obtained different holding period return sequence using quantile regression VaR model to do the empirical study.During the empirical study,this paper established the GARCH models under the assumption of Normal distribution,skewed Normal distribution,student's t-distribution and skewed student-t distribution separately and calculated VaR values by quantile regression,and tested the estimated effect of the model with Kupiec failure test method.Then,this paper compared estimated results under different assumptions to draw several general conclusions.Finally,this paper predicted the recent stock market risk of Shanghai with the fitted model and discuss the method and effect of the predictions briefly.Model in this paper had relaxed the assumptions of normal distribution in the traditional VaR model on financial time series.This paper also validated that it is not a simple linear relationship between multi-period VaR and single-period VaR.Empirical analysis proved that quantile regression VaR model are not sensitive to the distribution forms of residuals and the longer the holding period,the worse the estimated effect of the model.This model has a good universality and good quality in predicting risks.There are several merit in this paper.First,most of the existing literature assume a normal distribution in the VaR model,which is inconsistent with the actual situation.This paper has introduced other distributions based on the distribution character of Shanghai Composite Index return series to do a comparative study.It revised the limitations of the existing VaR model which is normally distributed,took into account the "fat tail" and bias of financial time series and improved the VaR model prediction accuracy.Second,most of the existing literature only study single period risk,but this paper added the length of the holding period as an explanatory variable in the model to calculate multiple of VaR and analyzed the effect of the length of the holding period to the model's accuracy.Third,the existing literature rarely mention the method and results on risk prediction via VaR model.This paper verified the predictive ability of the model in the empirical research,proposed a prediction method of the VaR model,and predicted the recent Shanghai Stock risk.In a word,quantile regression in this paper has a good quality on risk measurement.