| Traditional asset pricing models are usually based on the following assumptions:the market can quickly identify all new information and fully reflect them into asset prices to provide the best estimate of asset value.However,to complete this work requires a prerequisite:investors need to pay adequate attention to all information and accurately incorporate this information into investment decisions.In fact,attention is such a scarce resource.In addition to investing in financial markets,investors also need to allocate attention to study,life and work.Therefore,investors inevitably have to choose information when making investment decisions.It makes it easier for some prominent information to enter the eyes of investors,thereby affecting investment decisions.The limited attention is a common psychological phenomenon for investors.More and more literatures have studied the impact of limited attention on asset price changes.These studies have reached a unified conclusion:investors will pay more attention to some special stocks,thereby increasing the buying pressure to these stocks,pushing up the stocks’current prices,and causing thees stocks to perform poorly in the future.However,these studies mainly focus on the impact of limited attention on asset prices,and there are few literature studies on the impact of attention on asset pricing anomalies.Therefore,starting from the limited attention of investors,we study the impact of limited attention on asset pricing factor portfolios by constructing attention factors that express investors’attention,so as to give a certain explanation for the asset pricing anomaly in the Chinese stock market.At the same time,given that investors are more likely to pay attention to some market or industry clustering information,we will also verify the relationship between limited attention and the predictability of the market.First,how to measure attention is the focus of this research.Since investors can obtain information from multiple channels,it is difficult to combine information from all channels to measure attention.In order to make the results more accurate and robust,we use multiple indicators to measure attention.First,following Barber and Odean(2008),this article uses extreme daily returns,abnormal trading volume,and news attention as proxy indicators of attention.In the measurement of news attention,the information coverage is more comprehensive.we cover news coverage at three different levels:newspaper news coverage,analyst coverage,and stock bar discussion.At the same time,in order to eliminate the effective information contained in the proxy index,we also use the daily limit and the exogenous impact design of Longhubang to do a robustness test.Then,based on the attention measurement index,we construct attention factor and analyzes its performance.By buying low-profile portfolio and selling high-profile portfolio,we get the attention factor long-short portfolio.Analyzing the performance of the attention factor,whether it is the equal-weighted or value-weighted,the attention factor portfolio strategy can obtain stable excess returns.In order to prevent the return from being explained by FF three factors,we also calculate the risk-adjusted return after the FF three-factor adjustment,and the result is basically the same as that of the unadjusted return.We also find that the performance of the attention factor is persistent,that is,the arbitrageur does not make the return of the attention factor combination disappear immediately,but persists for a period of time.Finally,this article also compares the attention factor with common factor in the capital market.From the results,the focus factor portfolio has a higher return and a lower standard deviation,that is,a higher Sharpe ration.From the perspective of correlation,the correlation between the three attention measurement indicators is strong,but not completely related,indicating that each attention measurement indicator can play a unique role.In addition,the attention factor shows a positive correlation with the market portfolio and reversal factor,and shows a negative correlation with the momentum factor.Second,investors are concerned about how to influence the pricing of asset pricing anomalies.Following Peng and Xiong(2006),we first construct an investment decision model based on representative investors with limited attention to analyze how investors react to asset prices under limited attention,and then discusse the source of asset pricing anomalies.This model finds that investors with limited attention will pay more attention to clustering information about factor features,which amplifies the return of these anomalies to some extent.Then we carry out the empirical analysis.By re-copying the previous asset pricing anomaly factors,it was initially discovered that there are five types of significant anomalies in the Chinese stock market:scale,value,volatility,reversal,and liquidity.Regardless of whether it is equal-weighted or value-weighted,these anomalies can produce significant long-short returns.In order to analyze the relationship between attention and asset pricing anomalies,we use three methods(double sorting,factor model method,and Fama-Macbeth regression)to conduct empirical testing.The results of the double sorting method show that,in addition to the scale anomaly,the return effects of the other types of asset pricing anomalies are more significant in the high attention group,and the return effect in the low attention group becomes weaker or even no longer significant.This shows that the anomalies of value,reversal,volatility,and liquidity can be explained by investors’attention to a certain extent.The return of asset pricing factor anomalies may be driven by investors’limited attention.The factor model provides a way to verify the mutual interpretation of the factors.In the factor model that includes attention,except for the scale anomaly and the illiquid anomaly,the alpha of other anomaly factors is no longer significant,and even begins to change from positive to negative.For the scale anomaly,alpha has also been significantly reduced;from the perspective of the R~2,after the attention factor is included,the R~2 for all types of anomaly factors are significantly improved.When the attention is regressed on the factor model that incorporates various anomalies,the alpha of the attention factor does not significantly decrease,which shows that the attention factor can explain the return of various anomalies,while the various anomalies cannot explain the attention factor return.Fama-Macbeth regression connects attention pricing and asset pricing anomaly pricing,and therefor can study their mutual influence.The regression results show that the attention significantly reduces the pricing of asset pricing anomalies.The sub-sample results show that the pricing effect of asset pricing anomalies in the high attention group is stronger,while in low attention groups,the pricing effect has become weaker or even no longer significant,which once again shows that the benefits of asset pricing anomaly can be explained by the attention.Since limited attention strengthens the mispricing of the anomalous factor of asset pricing,when the mispricing caused by the limited attention is corrected,the mispricing of the anomaly factor will also be eliminated.Finally,in order to measure the explaination power,we use the coefficient split method to test.The result of the coefficient split shows that the attention factor has shown a good prospect in explaining the asset pricing anomaly.Expect for the liquidity anomaly,the attention factor can explain the asset pricing anomaly premium by more than 25%,attention factor plays a strong role in the pricing of asset pricing anomalies,and investor attention is the main reason for these asset pricing anomalies.At the same time,in order to eliminate the information contained in the attention index,this article then uses the exogenous impact of attention to deal with.China’s unique price limit system design and the announcement of the Dragon and Tiger list.The effect of portfolio stocks on asset pricing is basically consistent with the results of the previous three attention indicators.Third,how investors attention affects the predictability of the market’s aggregate returns.Given that investors pay more attention to some market-level clustering information,there is a connection between investor attention and the predictability of aggregate market returns.First,this article constructs a comprehensive index of market attention.According to the characteristics of the attention index,we construct a three-dimensional market attention index,which is return-based,trade-based and news-based,and uses them to predict the total excess return of the market.The prediction results in the sample reveal that the attention index has a significant negative prediction effect on the total excess return of the market in the next month.High market attention is often accompanied by low market excess returns in the future.The forecast cycle is expanded.The degree index also has a long-term predictive effect on the market’s excess returns.From 1 month,to 3 months,6 months or even 1 year,the negative forecast effect of attention on market returns has always been significant,which shows that attention is driven by mispricing.It cannot be eliminated in a short period of time;it takes a long time to digest.Next,we divide the sample period into an upward cycle and a downward cycle to further analyze the intra-sample forecast results.The results show that the prediction effect of the attention index on the total excess return of the market is more significant in the downward market.In the upward market,due to the existence of investor disposal effects,high attention will trigger the sale of stocks,thus leading to lower returns in the future;while in the downward market,high attention will often trigger stock purchases.Exacerbated the stock market crash,which led to further declines in future earnings.In order to make the forecasting effect more reliable,this paper conducts out-of-sample forecasts and conducts out-of-sample analysis with a 60-month window period.The results show that the attention index shows better out-of-sample forecasting capabilities for the entire stock market.Furthermore,this article discusses the difference between attention and sentiment indicators.This article finds that investor sentiment has no predictive power on short-term market returns,but has a significant negative predictive effect on long-term returns;while attention has a negative forecast in the short-term,Will disappear for a long time,and there will be a significant difference between the two,which shows that the prediction of emotion is relatively early,and attention is not a substitute measure of emotion,it can directly capture the attention of investors’behavior.At the same time,this article explores the source of the predictability of the attention index.With the help of the discount model,the company’s intrinsic value depends on the discount of future cash flows.This article believes that this predictability may come from high earnings estimates during periods of high attention.The contributions of this article are:(1)Different from the previous methods,when measuring investor attention,we do not consider a single proxy indicator of attention,but comprehensively considers each indicator,using three levels of attention,namely return,trading,and news,respectively.Accurately reflect the role of attention;in addition,we also adopt the exogenous design of"daily limit"and"dragon tiger list"as a robustness test.Then we construct a long-short combination of attention factors,which can earn higher excess returns with lower volatility,this can provide a reference for quantitative practice.(2)The previous research on the degree of attention focused on changes in asset prices caused by limited investors,focusing on changes in individual asset prices.Instead of raising it to the level of asset portfolio or even market portfolio,the research in this article attempts to explain the typical asset pricing anomaly from the perspective of investor attention,expands the research on asset pricing anomalies,and provide a unified framework for explaining asset pricing anomalies.(3)The research in this article extends the research on the predictability of market returns.In the past,the research on market predictability focused on the market’s own indicators and macroeconomic indicators,while ignoring the relevant indicators of investor decision-making behavior.Starting with limited attention,it provides new evidence for the predictability of aggregate returns in the Chinese stock market. |
