Research On Co-movement Of Multiple Stock Markets

With the rapid development of economic globalization and the acceleration of financial integration,it is of great theoretical value and practical significance for investors' investment decision-making and supervision departments' risk management to explore the co-movement between the international stock market and the Chinese stock market.Based on vine Copula model and time-varying multiple Copula model,this paper studies the static and dynamic co-movement among the returns of five stock indices,namely:Hang Seng Index(HK),Dow Jones Index(DQS),Shanghai Stock Exchange Index(SH),Shenzhen Composite Index(SZ)and Nikkei 225 Index(JAP).First,descriptive statistical analysis of returns shows that there are strong autocorrelation effects and ARCH effects in each series.Therefore,we use EGARCH(1,1)-skew t model to conduct empirical research and test,and find that the model is well fitted.At the same time,we find that the tail of the marginal distribution in Shanghai stock market is fat and prone to extreme events.Second,relying on the marginal probability density function of returns estimated by the EGARCH model,we establish the vine Copula model.And the differences of R-vine Copula,C-vine Copula and D-vine Copula are compared from goodness of fit test and risk measurement.The risk measurement is analyzed by Monte Carlo simulation.The results show that the R-vine outperforms C-vine and D-vine both in terms of goodness of fit test and portfolio risk prediction.With the increasing number of variables,R-vine structure and C,D-vine structure will be more and more different.Then the advantages of R-vine structure will be more prominent.Finally,we analyze the co-movement among stock markets from the static and dynamic perspective.This paper studies tree structure and three correlation coefficients in static co-movement analysis.And we find that Shanghai stock market is the central node of the five stock markets.And with the increasing number of conditional markets,the conditional correlation between markets is decreasing.In five stock markets,the probability of falling at the same time is greater than the probability of simultaneous rising.In terms of dynamic co-movement,a time-varying Copula model which based on rolling window is established.We find that the dynamic co-movement among these five markets is relatively stable except during the period of financial crisis,the dynamic correlation time-varying characteristics of each node in each tree structure are not obvious,and the correlation coefficient changes within the range of 0.2.In whole co-movement process,the three correlation coefficients are basically in the state of "lower-tail correlation coefficient>rank correlation coefficient>upper-tail correlation coefficient",which shows that the impact of "bear market" is greater than "bull market".With the continuous entry of conditional markets,the dynamic correlation between stock markets will be significantly reduced,and even the state of asymptotic independence will appear.Facing the problem of "curse of dimensionality" in vine-Copula,this paper introduces a time-varying multivariate DCC Copula model.Based on the calculation of dynamic co-movement,we find that the time-varying multivariate DCC Copula is appropriate to describe the dynamic co-movement among these five stock markets.And according to AIC,BIC criteria and logarithmic likelihood values,the time-varying t-DCC Copula is superior to the time-varying Gaussian-DCC Copula,and slightly better than the R-vine Copula.Then the expression of the dynamic co-movement correlation coefficients among five markets is obtained,which provides the corresponding theoretical basis for preventing financial risks.