The Research Based On The Long Memory Of Chinese Security Market

Efficient Market Hypothesis is a cornerstone of modern finance theories. As the promotion of EMH, a series of theories that based on EMH are developed rapidly. However, EMH is faced with a lot of challenges because of too many hypotheses and difficulty of empirical tests. In order to reduce the hypotheses, FMH is raised in 1990s.As an important character of fractal market, long memory shows the correlative of long lags. We chose Shanghai Composite Index as the representative of Shanghai Stock Exchange and Shenzhen Component Index as the representative of Shenzhen Stock Exchange. By building SETAR models and ARFIMA models, we tried to find the long memory in Chinese stock market.In the test of long memory, we use the method of curve fitting. So the goodness of fit can be used to judge the long memory of markets. The result shows that the long memory of Shanghai and Shenzhen market is not apparent. But the long memory of the absolute value of yield sequences and the square value of yield sequences is typically strong.The long memory of market reflects the market is a nonlinear system, so a nonlinear model-SETAR is used here to analyze the stock market in China. We build SETAR model based on the absolute value of yield sequences and build ARFIMA model to estimate long memory parameters. The results of models shows that, in every threshold range, the strength of long memory is different. Both Shanghai market and Shenzhen market show that long memory becomes stronger as the yields are lower. At the same time, we build ARFIMA model based on yield sequences of Shanghai and Shenzhen market. Although the long memory parameters of yield sequences are obviously small, Shanghai and Shenzhen markets also have long memory.The empirical tests reveal Shanghai and Shenzhen markets show the feature of long memory. And also, the nonlinear model-SETAR describes Chinese market better.