Research On Risk Spillover Effect Of Security Market Based On Quantile Regression Method

Financial risks have the characters of rapid tarvelling,wide spreading and deep influencing.With the rapid development of financial integration in recent years,economic and trade cooperation between countries and regions is getting closer and closer.Once other country or region's security market is at risk,China's security market will be greatly affected.Therefore,quantitative characterization of risk is particularly important.The quantile regression method is favored by many scholars because of its robustness and its insensitivity to the distribution of residual terms.This paper applies the quantile regression method to the security market,mainly carried out from two aspects:the measurement of risk value and the depiction of risk spillover.First of all,considering that stocks are typical types of securities,they have an important position in the securities market.we select the S&P 500,Nikkei 225,Hang Seng and CSI 300 index to represent the four markets of the US,Japan,Hong Kong and Chinese mainland.The data range is from January 2005 to December 2017.We study the risk values and spillovers of these markets,which are based on the following three parts.Firstly,based on the quantile GARCH family models,the risk values of each stock index yield series are calculated,and the estimation results are compared from three aspects:confidence levels,residual distribution hypothesis and model forms.The results show that the estimation results of the ARCH model are slightly worse,and the estimations of other GARCH models are more accurate.Secondly,considering that use the quantile GARCH family models,the order determination is more complicated in the modeling process,while the CAViaR method directly models the conditional quantiles of the sequence,which reduces the systematic estimation error of the model.We add the liquidity and voladility index into classic CAViaR models,propose the CAViaR-LI and CAViaR-VI models.The empirical results show that,the optimal models for each market are not same,and the extened models estimate better than the classical models.Thirdly,based on the optimal VaR of each index,the risk spillover effects between various markets are studied by combining Granger causality test and quantile Co VaR method.The results show that the risk spillover between the markets with close geographical relationship will be greater,and when crises occur,the risk will spread to other markets faster and deeper.After analyzing the risks of various markets,we further study the risk values of various industries and the risk spillovers of industries on security markets based on the extended CAViaR models and CoVaR models.The CSI 300 industry indices are selected to represent various industries in Chinese mainland,and to make the results comparable,the data ranges are also from January 2005 to December 2017.The empirical results show that when an industry is in a state of risk,the risk of security market will increase to a certain extent.Therefore,when analyzing the risk of security market,the extreme situation of each industry in the risk state should be considered.And in general,the financial,pharmaceutical,information and consumer industries are relatively more important to the Chinese mainland security market.