Stock returns around earnings announcements: an empirical analysis of the Helsinki Stock Exchange
Pelttari, Teemu (2004)
Kuvaus
Kokotekstiversiota ei ole saatavissa.
Tiivistelmä
Since Ball & Brown (1968), the continuation of abnormal returns after earnings an-nouncement has been notified in various studies. The purpose of this study is to find out, whether the post announcement drift exists in the Helsinki Stock Exchange. The four-factor time-series model introduced by Kim & Kim (2003) is used as a risk estima-tion model. The factors included into the four-factor model are book-to-market value, size, earnings surprise and return of a market portfolio. The single-factor market model is used as a reference model.
The firms included in the study are arranged annually on the basis of earnings surprise. After that, two extreme portfolios of unexpected earnings (positive and negative) are formed for each event year. The post announcement drift is then studied within these two extreme portfolios. The calculation of unexpected earnings is done with annual earnings per share figures, and average earnings per share forecasts collected from the Arvopaperi-magazine. The unexpected earnings are studied with the annual earnings announcements released in the period from 2002 to 2004. As a whole, the data is col-lected from the period between 2000–2004.
In brief, the results show that there is a notable post announcement drift in the Helsinki Stock Exchange. However, it might be so small, that the traders can’t utilize it. One of the main findings of this study is also that the four factor model doesn’t provide signifi-cantly different results, when the positive and negative earnings announcements are tested separately. For both models, the cumulative abnormal return after the earnings announcement is significant for the portfolio of positive earnings surprises. The market reaction to the negative earnings announcements seems to be more accurate than market reaction to the positive earnings announcements. In the case of negative earnings an-nouncements, the most of the drift occurs before the event day. The results of testing the difference between cumulative abnormal returns for the portfolio of positive and nega-tive unexpected earnings show that the four-factor model provides better results, when results are cumulated before the event day. This leads to a conclusion, that the post an-nouncement drift might exist due to the misspecified model used in risk estimation.
The firms included in the study are arranged annually on the basis of earnings surprise. After that, two extreme portfolios of unexpected earnings (positive and negative) are formed for each event year. The post announcement drift is then studied within these two extreme portfolios. The calculation of unexpected earnings is done with annual earnings per share figures, and average earnings per share forecasts collected from the Arvopaperi-magazine. The unexpected earnings are studied with the annual earnings announcements released in the period from 2002 to 2004. As a whole, the data is col-lected from the period between 2000–2004.
In brief, the results show that there is a notable post announcement drift in the Helsinki Stock Exchange. However, it might be so small, that the traders can’t utilize it. One of the main findings of this study is also that the four factor model doesn’t provide signifi-cantly different results, when the positive and negative earnings announcements are tested separately. For both models, the cumulative abnormal return after the earnings announcement is significant for the portfolio of positive earnings surprises. The market reaction to the negative earnings announcements seems to be more accurate than market reaction to the positive earnings announcements. In the case of negative earnings an-nouncements, the most of the drift occurs before the event day. The results of testing the difference between cumulative abnormal returns for the portfolio of positive and nega-tive unexpected earnings show that the four-factor model provides better results, when results are cumulated before the event day. This leads to a conclusion, that the post an-nouncement drift might exist due to the misspecified model used in risk estimation.