Research On The Impact Of Investor Sentiment On Stock Returns Based On Partial Least Squares

Due to the uncertainty in the future,it’s difficult for investors to make an accurate judgment on the future cash flow and income in the capital market,.Generally speaking,they usually only form a belief or preference,which is the comprehensive judgment of investors on future cash flow based on subjective and objective factors.Therefore,even for the same financial asset,different investors will have different beliefs and preferences,which is called"sentiment".The uncertainty of sentiment,which affects investors’behavior by affecting their judgment of expected returns,may eventually cause stock prices continuously deviate from their value.Investor sentiment is difficult to measure.Traders in the market often feel its presence but unable to quantify it.Baker（2006）proposed principal component analysis（PCA）,which is widely used in the literature since then.This paper selects the partial least square method（PLS）which is first applied by Huang（2015）.And using principal component method was used as the control.In terms of the selection of agent indicators,five indicators including turnover rate（TURN）,average ADL index（MAADR）,new investor account（NIA）,price/earnings ratio（PE）and consumer confidence index（CCI）are finally selected according to relevant literature and the basic situation of China’s capital market.Using these five indicators as proxy variables,S^PLS and S^PCA are constructed respectively.In terms of fitting effect and forecasting ability to the market,empirical study shows that S^PLS is better than S^PCA.Then we use the constructed sentiment index to empirically research the impact of investor sentiment on stock returns.It is divided into two parts:first,the overall effect of investor sentiment,which means the effect of sentiment on the overall stock market.This paper analyzes the persistence of the influence of investor sentiment on stock market earnings through the regression of market returns in different time spans.The empirical results show that there is a short-term earnings inertia and a long-term earnings inversion between investor sentiment and the overall stock market.Approximately for the next nine months,sentiment has maintained its predictive power over the stock market.Second,the cross-sectional effect of investor sentiment,which means the degree to different stocks affected by sentiment fluctuations.Market value,stock returns volatility,profitability,age listed are used as characteristic variable.The paper divides all the Shanghai A-share stocks into 10 groups in ascending order of each characteristic value.Finally,we get the conclusion:small firm,high volatility,less profitable stock is higher sensitivity to the emotions.But conditional influence of stock age using domestic securities market data is difficult to verify.The empirical results basically verify the conclusion that investor sentiment has a stronger impact on stocks that are difficult to value and arbitrage.In conclusion,S^PLS can better predict the overall stock market returns,and partial least square method is more suitable to construct the sentiment index than the principal component analysis method.Investor sentiment and the overall stock market are shows short-term earnings inertia and long-term earnings reversal.Investor sentiment has a greater impact on stocks that are difficult to value and arbitrage in general.Therefore,Chinese investors should fully consider the impact of investor sentiment on future returns and formulate appropriate investment strategies to obtain excess returns or avoid risks.