The Impact Of Media Reports On Stock Prices And The Reversal And Continuation Of The Impact

Since John Maynard Keynes created the word "animal spirit",economists have been trying to understand the determinants of irrational behaviors in stock market,which cannot be explained by fundamental information of the public company and there have been number of papers describing irrational phenomena in the field of investment.Then the subject behavioral economics has been created.According to theories in behavioral finance,human behavior is not perfectly rational.In 1979,combining behavioral finance with traditional finance,Kalman proposed Prospect Theory,in which people care about the difference between the real income and the prospected income,rather than the absolute value of the real income itself.In addition,emotions often undermine human self-control,leading people not to make rational decisions.People cannot fully distinguish between genuine information and fake information from the news that they obtained.In concern of the spread of information,the most popular circulation is mass media.Many investors,especially individual investors,obtain information about relevant companies from the mass media which is often considered to influence the stock market by affecting investor’s sentiment.In this context,this thesis examines how the mass media influences the stock market.This thesis explains the problem from two aspects,one is whether the news media will have an impact on the stock market,and the other is how the news media influence the stock market.Stocks from CSI 800 Index is taken as the sample stocks in this thesis.The time range of the data is from October 7,2015 to October 7,2018,and the time frequency is daily.We first uses Python to crawls for texts on Xinhua.com,China News Network,Securities Times,21 st Century Business Herald,Sina Finance and other websites and record the publication time of the texts.Then we use Python’s third-party tools to analyze the emotions indicated by the texts.After removing the auxiliary words and prepositions in the article,we use the default dictionary tool to score the texts.According to the scores,the articles are categorized as news with negative tone and news with positive tone.After obtaining specific data,we regresses stock return on the lags of proportion of negative news in panel data and find that,the increase in the proportion of negative news will lead to a significant decline of return the next day.The 1st,3rd,and 5th order lags of proportion of negative news are significant at the critical values of 0.1%,10%,and 5%,respectively,the coefficients of which are all negative.However,when the VAR model analysis was performed separately for the entire period of each company,the first-order lags of only six companies are significantly at the 10% confidence level,which contradicts the conclusion of the regression of panel data..In this case,we refers to the relevant literature,and proposes the assumption that the relationship between mass media reported on specific companies and company-level stock’s return is time-varying and episodic rather than continuous.Then we use the rolling-window VAR,which means to move the window one quarter forward every day,to verify the assumption.The result shows that the relationship between negative news reports and stock returns is episodic.This shows that there are few information in the news media,and there are many emotional parts.However,most of these stock returns will reverse,which is caused by the increase in negative news,resulting in a rise in stock prices.Media sentiment and information have no lasting impact on stock prices,which means there are little vital information in the news reported,but the news contains many emotional information.However,most stocks are often reversed immediately after being shocked by the news,which suggests that what media says is not always noise.It can sometimes contain new information,but the impact on stock prices is still not enduring.