Empirical Study Of Complexity Of Chinese Stock Market

This dissertation, based on the complexity theories, explores the Chinese stock market with the help of chaos and fractal. In the past 40 years, the linear paradigm has dominated financial economics. But many times in the capital market, when an action occurs, there follows an exponential and non linear reaction. The traditional capital theories made simplifying assumptions and used linear models to describe capital market. Needless to say, linear models are easy to handle. However, excessive simplifying assumptions are incorrect, since, more often than not, the results of theories can not explain the realities. The efficient market hypothesis is the cornerstone of traditional capital market theories. However, many phenomena, such as the peaked and fat-tailed distribution of stock returns, sudden crash of the stock market ,the long memory of stock prices series, can not explained by EMH .EMH treat these occurrences as anomalies .But the important is whether these occurrences are anomalies or the truths? Maybe a new mode of thinking will give us a better understanding.The development of complexity science provides a new perspective to the study of capital market. Complexity science breaks through the linear paradigm, and brings a radical transform of methodology .We find we can better understand the stock market under the framework of complexity science. This dissertation, regarding the stock market as a complex, non-linear and dynamic system, studies the Chinese stock market. The purposes of this dissertation are following:(1) Can Chinese stock market be explained by the traditional capital market theories?(2) Can we apply complexity science to Chinese stock market?(3) How can we measure the complexities of Chinese stock market?(4) How can we predict the stock prices and how to model the volatilities of stock returns under the frame of complexity science?Some conclusions are drawn from our study. There are:(1) The behavior of Chinese stock market does not conform to EMH, and traditional capital market theories under linear frame can not explain the Chinese stock market.(2) Chinese stock market is a complex, non-linear and dynamic system. The fractal structure and chaotic features can be detected. The movement of stock prices is fractal and persistent. There are fractal attractors in the evolution of Chinese stock market. And the averagenon-periodic cycle of shanghai stock market is about 150 days and 180 days for Shenzhen stock market.(3) The combination of phrase construction and artificial neutral network can provides a good non-linear prediction of short-run stock prices. However, it is impossible to make precise long-run prediction because of the existence of chaos.(4) The volatilities of Chinese stock market do not satisfy the (?)T law, which once againshows Chinese stock market is not efficient. Instead, the volatilities of Chinese stock market are persistent, time-varying, clustering and conditional heteroskedastic. We find the ARFIMA-GARCH model can better describe the volatilities.(5) As for a complex stock market, the standard deviation and the beta coefficient are not good risk measurements. Instead, the non-linear variable ,H exponent is a good replacement for risk measurement.This dissertation, instead of using traditional methodology, which is static, equilibrium and linear, adopts complexity science as theoretical foundation. Complexity reflects Nature and human society. We fully prove that the distribution of Chinese stock market returns in non -normal, but non-linear, self correlative and heteroskedastic based on strict econometrics methods. We use rescaled range analysis to demonstrate that there are self-similarity, long memory and sensitive to initial value in the time series of Chinese stock returns. The models we use to forecast short-run stock price and volatilities are advanced tools in the field of financial economics. We write some programs to analyze stock data, which are of practical value.The achievements of this dissertation provide a new perspective for the study of Chinese stoc...