| The stock market bubble has always been empirical research on financial markets,Study the production, inflation and burst of the stock market bubble about the influence of China’s stock market, has important theoretical significance and practical significance to the healthy development of China’s securities market,Therefore, we need to conduct the thorough research to the stock market bubble.At first, this paper carries out theoretical analysis of the research background of the bubble, and analysis the definition, classification, formation mechanism of bubble in system. Research both home and abroad are summarized, the meaning of the prose is given.Secondly, in this paper, the stock market bubble detection methods are summarized, especially for the inspection of theoretical thinking, as well as the advantages and disadvantages of inspection methods are carried on the thorough analysis. It is concluded that there are indirect and direct detection methods, and found the direct method is more suitable for the study of China’s stock market.Therefore, using the method of the inner bubble, p/e ratio, tobin Q, measure the bubble dynamic behavior to directly test bubble, and the stock market in our country has carried on the existence of a speculative bubble with econometric analysis.IHMM is on the basis of markov model with hierarchical Dirichlet process, and establishing a state of infinite hidden markov area time-varying autoregressive model,And it combine with the AR process, paragraphs intercept and random disturbance variance breakpoint probability and the area structure to realize compatible non-stationary data in the process of stock market bubble measure model. And the setting of the model also considers bubble instability as a measure of the time series,uncertainty of random perturbation terms, as well as the process of the structural instability, then embedded in the IHMM model of explosive and nonlinear characteristics of the bubble. The model also through hierarchical structure fitting posterior distribution, and get a posteriori estimation results using Gibbs samplingestimation, estimate the lag coefficient, it is the measurement indicators for the measurement of bubble-temperature coefficient, and through the posterior distribution of the temperature coefficient of area system analysis identify the bubble area system dynamics, and offer us about the size of the bubble burst power as well as the overall dynamic information asset prices. We use markov monte carlo(MCMC) method to estimate IHMM, By mixing Gibbs algorithm of hierarchical structure completed all infinite state model of the posterior unbiased estimate of the transfer system variables.The innovation of this article is to use the IHMM to empirical studies on China’s stock market bubble. And estimate model using mixed Gibbs algorithm of hierarchical structure, which can not only avoid additional processing of the original data information loss, also can realize time varying estimates for correlation coefficient.Finally, we based on IHMM model to simulate the three bubble measurement methods and the results of the empirical testing is reviewed. Determine the Shanghai a-share in the sample period each time whether there is bubble, it was found in September 2006 in May- 2007, May 2007- December 2007, April 2009- July 2009 three time bubble. And we found that the bubble has certain periodicity, within the state, is a bubble and composition of negative bubbles appear alternately. Bubble formation, expansion is not instantly you can do, and are sustainable. Is not a bubble burst, all but part of burst, and for a period of time gradually disappear. Bubble indicators measured temperature coefficient of the posterior mean and high volatility in the posterior probability of point and point of slump in share prices, use IHMM to identify the Chinese stock market bubble can better describe the development of Chinese stock market speculation bubble and evolution, and has good prediction. |
PDF Full Text Request