The Empirical Research On China’s Stock Market Bubbles

The stock market bubble issue is always the hot point in financial research and application, as bubbles’ emerge and burst will make significant impact on financial market and economy environment, the research on bubbles is very important with theoretical significance and realistic significance. This paper is based on the research in this field and focus on three questions:first, how to test the bubbles. Second, how big bubbles are. Third, how to tell bubbles burst time.There are three main parts in this paper. The first part is test the existence of bubbles in China’s stock market from1997to2011. Based on relevant theories derivation, we get Run Duration Dependence Model and Momentum Threshold Auto-Regression Model (MTAR). Then we use the two models testing the Shanghai composite index and Shenzhen composite index respectively. The second part is measure the size of bubbles. Price Earning Ratio is used to measure the size of bubbles from1997to2011at first and Residual Income Model (F-O) is amended to be applied to calculate bubbles’size indicator. The third part is use the latest research achievements in Econophysics to tell bubbles’burst time. Model derivation is made and we get the Logarithm Period Power Law Model (LPPL) which is used to forecast bubbles’ bust. Simplifying the LPPL model and using genetic algorithm to acquire unknown parameters, and then we can tell bubbles’bust time.Based on the research in this thesis, we come to the conclusion. Firstly, there are three bubbles in the period from1997to2011, which are from1997to2001, from2005to2008and from2009to2011. Secondly, Price Earning Ratio and MTAR came to the same conclusion. There is a big bubble in2000and2007respectively and the bubble in2007is the biggest in history. And we also conclude that the year2005is the best time to invest into China’s stock market. Thirdly, we can tell the bubbles’ burst time some time but not always. If the stock price’s trend is in accord with LPPL character, then we can get the best estimated value of parameter and bubbles’ burst times are acquired.