This section is a source of information for students who want to perform an event study in Finance. An event study is a popular tool for investigating the impact of events on stock returns. There are several ways of doing this, and the procedure described below is just an illustration of how to get started. This may be helpful while reading some of the methodological papers mentioned below.
In order to perform an event study, you need data on events and data on security returns. Information on security returns can be obtained from several sources, including Refinitiv’s Eikon or Datastream. However, there are several attractive alternatives. Consider for example Yahoo Finance, where you can find a lot of information on US stocks for free. And for those focussing on the US markets, the many univeristy libraries also offers Wharton’s WRDS database.
The following list of links might be useful:
- CPSC: Consumer Product and Safety Commission. Information on product safety issues, lawsuits, etc..
- EM-DAT: The international disaster database. Large dataset with a lot of information on accidents and disasters. Maintained by the Université catholique de Louvain.
- Global Terrorism Database
- Moody’s: Credit ratings of corporate and government bonds. Rating changes.
- Recalls.gov : Data on product recalls in the UK.
- Spactrack: Website tracking Special Purpose Acquisition Companies.
- Zephyr: Database with several corporate events. Available on the Library website of the university.
Event studies are designed to measure the impact of an event on a security’s return. There are several ways of doing this, and the procedure described below is just an illustration of how to get started. This may be helpful while reading some of the methodological papers mentioned below.
After identifying the event date or the announcement date, we need to establish the so-called event window. Next to the event date, the event window typically includes several days surrounding the event or announcement date. Then we calculate cumulative returns over the days in the event window as the object of analysis.
Since the event return may be affected by well know factors (in particular systematic or market risk). For this reason, we adjust the event returns for these factors by subtracting the normal returns. The resulting returns are also known as adjusted returns or abnormal returns. The normal return is the expected return as if the event did not take place. In calculating the normal returns, we often use information on the returns of the stocks in estimation window, which is the period preceding the event window.
The estimation window should be chosen in such a way to make sure that the returns in the estimation window are not affected by the event. To accomplish this, it is sensible to have the estimation window end several days before the event, and it should not include any dates that are also in the event window. To be perfectly safe, it makes sense to leave on day between the end of the estimation window and the start of the event window. The event window should also be long enough to calculate meaningful estimates of normal returns. In practice, researchers choose estimation windows ranging from 90 to 200 days.
Brown and Warner (1980) report three different ways of adjusting for normal returns:
1. The mean adjusted return model:
ARi=Ri,0 – E[Ri] , (1)
where t is the event date, Ri,t is the return on the even date and E[Ri] is the mean return in of stock i in the estimation window. The example in the excel sheet attached illustrates how to calculate the mean adjusted return and the test statistic for evaluating average abnormal returns at the event date. In this spreadsheet, there are 94 securities, centred around an event date at t=0. The average abnormal returns are based on a three day event window [-1,+1] and are presented in column EH. The average abnormal return over this three day event window equals 1.87% and is presented in cel EH98. The standard deviation of the abnormal return is presented in cel EH99, and the t-statistic equal 4.18 and is presented in cel EI101.
Significance of results
In order to test for the significance of the cumulative abnormal returns, we first need to check whether the cumulative abnormal returns have a normal distribution or not. Checking for normality can be done with the Jarcque-Bera test. When returns follow a normal distribution, we can use the regular t-test for testing the hypothesis that the cumulative abnormal returns are equal to zero.
In the example, the outcomes of this t-test was presented in cell D26 and cell H26 of the spreadsheet. Notice that the test statistic includes the standard deviation of the mean cumulative abnormal return, which is the standard deviation of the individual CAR’s divided by the square root of the number of observations.
When the returns do not follow a normal distribution, you have to use a non-parametric test, such as the Corrado (1989) test. The original Corrado test was developed for analyzing abnormal returns on the day of the event. However, in practice event windows often have a length of more than one day. Cowan (1992) and Kolari and Pynnonen (2010) discuss how to deal with event windows of more than one day.
The Corrado test ranks the return(s) in the event window relative to the period including both the estimation window and the event window. The Corrado test includes the following test statistic:
Finding literature for you study
In writing your thesis, it is important to acquaint yourself with the research that has been done in the field before. The leading journals in Finance are:
- Journal of Finance
- Journal of Financial Economics
- Review of Financial Studies
- Financial Management
The leading journals have a strong focus on the US. Other journals that publish research in finance (and in particular with a more European focus) are:
- European Journal of Financial Management
- Journal of Banking and Finance
- Journal of Business, Finance, and Accounting
- Barber, Brad M., and John D. Lyon, 1997, ‘Detecting long-run abnormal stock returns: The empirical power and specification of test statistics’, Journal of Financial Economics, Vol. 43, No. 3, pp. 341-372.
- Benninga, S., and T. Voetman, 2008, ‘Event studies’, Chapter 14 In: Financial Modelling by S. Benninga, The MIT Press, Cambridge MA.
- Brown, S.J., and J.B Warner, 1980, ‘Measuring security price performance’, Journal of Financial Economics, Vol. 8, pp. 205-258.
- Brown, S.J., and J.B. Warner, 1985, ‘Using daily stock returns: the case of event studies’, Journal of Financial Economics, Vol. 14, pp. 3-31.
- Corrado, Charles, 1989, ‘A nonparametric test for abnormal security-price performance in event studies’, Journal of Financial Economics, Vol. 23, No. 2, pp. 385-395.
- Cowan, Andrew. R., 1992, ‘Non-parametric event study tests’, Review of Quantitative Finance and Accounting, Vol. 2, No. 343-358.
- Kolari, James W., and Seppo Pynnonen, 2010, ‘Nonparametric Rank Tests for Event Studies’, SSRN Working Paper.
- Kothari, S.P., and Jerold B. Warner, 1997, ‘Measuring long-horizon security price performance’, Journal of Financial Economics, Vol. 43, pp. 301-339.
- Kothari, S.P. and Jerold B. Warner, 2006, ‘Econometrics of Event Studies’, In: Handbook of Corporate Finance: Empirical Corporate Finance, Ed. B. Espen Eckbo.
- MacKinlay, A. Craig, 1997, ‘Event studies in economics and finance’, Journal of Economic Literature, Vol. 35, No.1 , pp. 13-39.