The Study Of Asset Pricing Based On Investor Attention

This dissertation expands the study of asset pricing based on investor attention.We theoretically and empirically explores the effects of investor attention on investortrading behavior and investment return. We also study the factors that affect investorinformation demand, and empirically test the theoretical models of Holthause andVerrecchia (1990) and Kim and Verrecchia (1997). Using information of the stockforum, we study the influence of investor attention and investor sentiment on theanalyst rating drift. The main contents and conclusions are as follows:Firstly, by introducing the heterogeneous attention, we expand the models ofKyle (1985) and Barber and Odean (2008). We theoretically find that when positiveinformation arrives, investor attention will enlarge the return of risky asset. However,when negative information comes, investor attention will amplify the loss. Usingsearch volume of Baidu to measure investor attention and combining with analysts’rating from2007to2010in Shanghai and Shenzhen A-share stock market, weconfirm the conclusions of the theoretical model.Secondly, we theoretically get that investor attention negatively affectsindividual investors’ investment return. Using abnormal search volume of Baidu as aproxy for individual investors’ attention, We find evidences support for Barber andOdean’s (2008) price pressure hypothesis and the conclusions of our theoreticalmodel.As search activities are investors in the process of getting information through asearch engine, we use search volume of Baidu to measure individual investorinformation demand. We find that acquisitions, dividends, earnings announcementsand performance forecasts significantly increase investors’ information demand. Wealso find that individual investors search for more information around earningsannouncements and especially for firms having more asymmetric information. In theearnings announcements date, search volume reaches the peak and is25.8%morethan the usual case. Consistent with the theoretical model of Holthause and Verrecchia(1990) and Kim and Verrecchia (1997), we find that the more individual investorssearch around earnings announcements, the more the stocks trading volume. Information demand before earnings announcements has no effect on the relationshipof contemporary abnormal return and earning surprise, but the postive associatonbetween abnormal return after earnings announcements and earning surprise is morepronounced when search before earnings announcements is more intense.Finally, we find that abnormal posting on East-Money stock forum issignificantly and positively correlated with other proxies of investor attention, andposting on the stock forum leads search activities and news reporting. Using abnormalposting on the stock forum as a proxy for investors’ attention, we find that investorattention can be an explanation for the drift of analysts’ ratings. By using SentimentAnalysis technology to analyze each posting, we construct investor sentiment indexfor each stock. By studying the effect of investor attention and investor sentiment onthe drift of analysts’ ratings, we also find that the higher attention and sentiment ofinvestors, the more stocks immediately react to analysts’ ratings. Subsequently, thedrift of stocks with high investor attention and sentiment is smaller than the lowinvestor attention and sentiment stocks.