High Frequency Data Based Research On Extreme Risk Spillovers And The Magnet Effect

With the continuous enrichment of the financial market types and the increasing complexity of financial product innovation,extreme price movements occur frequently in global financial markets.In severe cases,these extreme movements may lead to a financial crisis and even economic crisis.However,the risk measurement models and risk identification methods applied in traditional risk regulation framework have exposed many shortcomings and deficiencies.Under this realistic background,it is necessary to accurately measure and predict extreme price movements of financial assets,and the extreme risk has gradually become an important topic in academic research field.With the continuous development of information technology,the application of high frequency data in financial risk management is becoming more and more extensive.High frequency time series analysis could accurately capture the dramatic changes in financial risk during a short time period,and provide a more precise method for the study of extreme risks in financial markets.Under the research framework of extreme risk analysis using high frequency data,this paper takes the high frequency data modeling methods as the starting point and analyzes three financial realistic issues from the perspective of extreme risk,i.e.,‘Intraday risk spillovers between the Chinese stock market and index futures market',‘Risk spillover effects between different sectors of the Chinese stock market',and ‘The magnet effect of circuit breakers'.We use different econometric models to analyze the risk interactions and extreme risk characteristics of Chinese financial markets,in order to provide useful advice and reference for market investors' asset allocation and market regulators' policy formulation.The main findings of this paper as follows:First,this paper extends the theoretical definition of CoVaR measure and proposes a predictive CoVaR measure which could capture nonsynchronous spillover effects,and constructs a high frequency MV-CAViaR model framework to estimate multivariate CoVaR directly,this model could eliminate the bias caused by intraday transaction pattern.Moreover,this model is applied to empirically analyze the intraday extreme risk spillovers between the Chinese stock market and index futures market.Regression results show that the risk spillovers between the stock market and index futures market are asymmetric under different risk tails,market conditions or trading rules.Specifically,there exists significant downside spillover effects and insignificant upside spillover effects between the spot market and futures market.The stock market plays a dominant role in risk transmission during bullish periods and the futures market plays a dominant role in risk transmission during bearish periods.Finally,high margin requirements would weaken the spillover effect of the futures market,but it would also strengthen the spillover effect of the stock market.Second,this paper incorporates high frequency data into the CAViaR model which merely model low frequency data,and constructs a MIDAS MV-CAViaR model that jointly model high frequency and low frequency data,the proposed model is applied to empirically analyze the extreme risk spillovers between the GEM and main board in the Chinese stock market.Regression results show that considering high frequency data information in the tradition CAViaR model could significantly improve the explanatory power of price shocks and the overall model performance.Moreover,the GEM plays a dominant role in the extreme risk transmission mechanism and it has a unidirectional spillover effect on the main board market,the price shocks occurred in the GEM market would significantly reduce the extreme risk of the main board market.Furthermore,we consider the spillover effect of intraday rage shocks in the MV-CAViaR model framework and find that the intraday range shocks occurred in the GEM market would also reduce the extreme risk of the main board market significantly.Finally,this paper uses high frequency transaction data of Chinese stock index futures contracts and empirically examines the magnet effect hypothesis of circuit breakers from three perspectives,i.e.,price trend,market volatility and extreme risk.Regression results show that the magnet effect of circuit breakers does not exist in the form of the price trend or market volatility,because when CSI 300 index decreases and close to the triggering limit,the probability of a price decreases and the level of market volatility remain unchanged.On the contrary,empirical results show that the magnet effect exists in the form of extreme risk because as the CSI 300 index approaches the breaker limit,the probability of detecting a price jump(especially negative jump)in a stock index futures contract increases significantly.This indicates that extreme market events would occur more frequently and the probability of triggering circuit breakers would increase in the future.In addition,our study finds a significant increase in liquidity demand and insignificant change in liquidity supply ahead of a market-wide trading halt,suggesting that the deterioration of market liquidity may play an important role in explaining the magnet effect.