An Empirical Study On The Heterogeneous Response Of Stock Returns To The Public Information

Whether stock returns reflects market information and how it reflects is a hotdiscussed issue in the theoretical and practical field. Fama (1973,1992) proposedEfficient Market Theory and Three-Factor Model to research these problems above.Besides, practitioners and investors also employed fundamental and marketinformation in stock returns in order to obtain earnings. With the rise of behavioralfinance in recent years, De Bondt (1985) studied whether investors had over-reactionand under-reaction to the market information from investors’ psychologicalperspective. He showed that market information had a heterogeneous influence on thestock returns, which was opposite to the Efficient Market Theory.Based on the theories of Fama and De Bondt on responding features of stockreturns, we employ Bayesian model to study whether stock returns can reflect marketinformation and how it reflects in Chinese stock market. In first place, we dividemarket public information into three categories, namely, corporate financialinformation, stock market information and macroeconomic information. Among these,we apply some model selection criteria, such as AIC and Bayes factor, etc to selectsignificant variables. The empirical results show that stock market information andmacroeconomic information have a greater impact on the stock returns, comparedwith the corporate financial information. Moreover, stock turnover rate, market risk,deposit reserve rate and money supply greatly influence stock returns. It indicates thatour investors pay more attention on market information and macroeconomicinformation when they make investment decisions. We think that these results mainlycome from short-term speculation of Chinese investors, imperfect stock dividendssystem and non-efficient financial information disclosure mechanism in Chinese stockmarket.In the empirical test on investor over-reaction and under-reaction, some scholarsmainly built "winner portfolio" and "loser portfolio" to compare investment returnsand then analyzed investor reaction problems. In this paper, we apply Bayesianquantile regression to study this problem in Chinese stock market. In the calculatingprocess, we employ Markov Chain Monte Carlo (MCMC) method and Gibbssampling to estimate parameters in regression model. The results show that investorstend to exaggerate good news, especially firm size, turnover and monetary supply, etc,when they face a soaring stock. This leads that the rising degree of stock returns is higher than the average rising degree when there exist a unit of positive shock.However, investors tend to underestimate good information above when they face aslumping stock. It causes that rising degree of stock returns is lower than the averagerising degree when there exist a unit of positive shock. Therefore, we acknowledgethat the investors in Chinese stock market also obtain over-reaction and theunder-reaction mentioned in the behavior finance. We think that these results mainlycome from investor psychological deviation, such as self-attribution bias, conservativebias, etc.In order to make Chinese stock market function more effectively on finance andinvestment, the regulatory agencies should improve stock dividends system andinformation disclosure mechanism. Meanwhile, investors also need to strengthen theirown ability to judge information and avoid investment losses caused by psychologicalbias.