eThesis - Kauppakorkeakoulun sähköiset gradut

The purpose of this thesis is to study news-related reference point formation by market investors. I hypothesize that investors form reference points following salient, surprising firm-specific events, which are in this study represented by earnings announcements. To assess the existence of the phenomenon, I run tests based on abnormal trading volume around crossings of hypothesized reference points, which jointly test for the existence of a reference point and the disposition effect, i.e. the tendency of investors to sell winning stocks prematurely and to hold on to losing stocks. I also link investor sentiment to these tests, and argue that the reference point formation effect should be found to occur less under adverse market conditions. The tests performed in this thesis were created by Kliger and Kudryavtsev (2008), whose study remains to date the only one to study this phenomenon. The scope of this thesis is further expanded by testing if the magnitude of the surprise element in the earnings announcements has an effect on reference point formation. Moreover, it is tested if firm characteristics, namely size and beta, have an effect on the phenomenon.

The data for this study comprises all the earnings announcement events by New York Stock Exchange (NYSE) listed companies during the time period between January 2006 and December 2008, totaling 11,622 events after necessary eliminations. The key test metric is trading volume, which is for each event tracked for 75 days post-event and 250 days pre-event. In addition, stock price data, analyst consensus EPS forecasts, and actual EPS figures are used for identifying hypothesized reference points and surprise element in company-specific news content. The data is collected from the Datastream, IBES History, and First Call databases of Thomson Financial.

The results indicate that individuals are indeed prone to the disposition effect, and that surprise, measured in this study with both stock market reaction and actual earnings relative to forecasted earnings, seems to be important in driving reference point formation. This finding is highly significant with relevant t statistic values ranging from 3.33 to 6.55. However, I find evidence against my initial hypothesis of less reference point formation occurring under adverse market conditions. On the other hand, the data provide support for the hypothesis that the magnitude of the surprise inherent in the earnings announcement increases reference point formation. In addition, the phenomenon is found to be driven by small firms.