Persistent Information And Momentum Effect In Chinese Stock Market

Because of the limited cognitive resources, investors can not absorb and respond to all information. As frog-in-the-pan (FIP) hypothesis, investors may ignore the continuous occurrence of small magnitude information and produce under-reaction. This paper examines the frog-in-the-pan (FIP) hypothesis that continuous occurrence of small magnitude information may leading to under-reaction of investors.In order to make the frog-in-the-pan (FIP) hypothesis more specific, this paper provides a two-stage model to explain this hypothesis. This model divided investor into two categories:rational investors and FIP investors. Rational investors can absorb all the information processing, and FIP investors can not processe all the information. Due to the existence of FIP investors, these information cannot response in the price immediately. But these ignored information will eventually be reacted in the price, which will lead to better performance later. According to the frog-in-the-pan (FIP) hypothesis, the continuous occurrence of small magnitude information which may be ignored by investors will has strong predictive abilities of future return.In this paper, I use ID as a proxy of the degree of Information Discreteness. Smaller ID value is representative of continuous information and larger ID value is representative of discrete information. The empirical results of Chinese stock market in 2008-2015 show that the return of discrete information portfolio is -2.54%, and the return of continuous information portfolio improves to 1.17%. In addition, ID also has predictive abilities to long-term gains. In the smaller ID value group of continuous information, long-term momentum will not reversed; in the larger ID value group of discrete information, long-term momentum will reverse. This is also consistent with the frog-in-the-pan (FIP) hypothesis.Additionally, I use the ID value to build a momentum trading strategies. The backtesting result of A-share stock during 2008-2015 shows that smaller ID value group gains 23% higher than the larger ID value group.