The Impact Of Investor Sentiment On Stock Returns

Investor sentiment is one of important research topics in behavioral finance.In recent years,many scholars have done a lot of research on the impact of investor sentiment on A-share price and made very fruitful research results.On the basis of previous studies,this paper studies the impact of investor sentiment on A-share return and distinguishes the impact on macroeconomic announcement days from the impact on non-macroeconomic announcement days,which contributes to the more accurate separation of the impact of investor sentiment on A-share price and more effective combination of traditional financial theory and behavioral finance.In the model aspect,this paper improves the two-factor model which includes market factors and investor sentiment factors.In sum,while previous studies focus on the factor loading on investor sentiment,we instead study the factor premium of investor sentiment.Besides,on the basis of two-factor model,this paper adds the size,value and momentum factors to construct five-factor model.This paper uses the investor sentiment index to measure the change of investor sentiment and the sample data is Shanghai A-share price from December 2003 to December 2016.To circumvent the problem that the BW sentiment-change index is only available at monthly frequency,we use the mimicking portfolio approach and construct a zero-investment portfolio.In the empirical test,this paper has done a lot of work.Firstly,this paper tests whether market beta and investor sentiment sensitivity in different types of transactions will change.The results show that the negative influence of investor sentiment on stock pricing is significant on non-macroeconomic announcements days.Important information about the state of the economy is revealed on macroeconomic announcement days.Thus,on non-macroeconomic announcements days,investors may be more driven by sentiment.Next,we use the standard Fama and MacBeth two-pass regression to examine above empirical test results and estimate risk premium separately for announcement and non-announcement days.Finally,we use kinds of methods to make robustness checks including conditional factor exposure,expanded sets of test assets and robust Fama and MacBeth cross-sectional regressions.The test results show that our results are robust to potential errors-in-variables and misspecification biases.Besides,we discuss if investor sentiment have marginal explanatory power relative to the Fama-French-Cahart factors(i.e.size,value,and momentum factors).The test results show that investor sentiment may be one factor underlying the Fama-French-Cahart factors.In this paper,we make lots of work on the impact of investor sentiment on China's A-share market in different trading days.Our findings can be summarized:(1)Investor sentiment has a significant impact on the price trend of China's A share market,which plays an important role in explaining the price volatility of A share market.(2)Compared to the traditional two factor model,the extended model of the five factor can be more accurate to capture the function of the investor sentiment in China's A share market pricing.(3)In the macroeconomic information announcement days,there is no significant correlation between investor sentiment and China's A share market average excess return,but in non-macroeconomic information announcement days,there is a negative correlation between investor sentiment and China's A share market average excess return.Our findings have important theoretical as well as practical implications.In terms of theoretical implications,our findings suggest that future theoretical models should try to take into account both fundamentals and investor sentiment,particularly how they affect the cross-section of stock returns on different trading days(i.e.,macroeconomic announcement days versus non announcement days).In terms of practical implications,our findings imply that in investment and risk analysis decisions,an empirical asset-pricing model that explicitly takes into account the role of investor sentiment may be more informative.