| The volatility is based on the conditional variance of the return on assets and is one of the important factors affecting the smooth operation of the financial market.With the ever changing global economic and financial environment,China’s financial market has been more and more impacted.Structural changes of volatility has been common in China’s Stock Markets China’s Stock Market due to sudden impacts.Althoughthe traditional GARCH model is widely used in financial time series,it can not do well in the stock market with switching regimes.This paper focus on the study of switching regimes on China’s stock markets’ volatility:First,it point out the shortcomings of single-regime GARCH models and Markov Switching GARCH models is more suitable for volatility of the market with switching regimes through combing the research results of domestic and foreign scholars in this field.In the method of parameter estimation,it is pointed out that the traditional maximum likelihood(ML)method is limited by path dependence and can not fit GARCH models well.The Markov chain Monte Carlo(MCMC)method based on Bayesian inference can effectively overcome the defects of ML method by distributed sampling and iteration to parameter convergence.Then,this paper expounds the basic ideas and structures of two types of Markov Switching GARCH models,one is Markov Switching GARCH(MS-GARCH)model with normal distribution innovations,the other is Markov Switching asymmetric threshold GARCH(MSTGJR)model with Student-t innovations,and gives the MCMC algorithm steps of these two models respectively.Third,in order to better illustrate the superiority of the MCMC algorithm,this paper programmed Bayesian MCMC algorithm to simulate the parameters of the two models,and compared with the parameter estimation results obtained by the ML method.The results confirm that the MCMC method is indeed better than the traditional ML method in the two types of Markov conversion GARCH models.Fourth,this paperbased on Shanghai composite index and Shenzhen composite index,thevolatility characteristics of China’s stock market are depicted and analyzed through the modeling of these two Markov Switching GARCH models.The results show that the return series of Shanghai Composite Index and Shenzhen Composite Index have obvious objective characteristics,such as volatility clustering,peak and heavy tail,sequence correlation,heteroscedasticity and system transformation,so it is more reasonable to introduce GARCH model of switching regimes to model and process.In addition,whether the Shanghai Composite Index or Shenzhen Composite Index,MSTGJR model can more accurately describe the objective characteristics of volatility in China’s stock market than MS-GARCH model.Finally,this paper put forward some suggestions for China’s stock marketand makeprospects for future research directions. |
