Hi community,
I am conducting an Event Study in Stata by following the instructions of Princenton University (URL: https://dss.princeton.edu/online_hel...ventstudy.html) . I tested the abnormal returns in stata with the code provided by Princeton University. So that was no problem. As I have more events to test (243 Mergers and Acquisitions), I also had to calculate the average abnormal returns (AAR) & the cumulative average abnormal returns (CAAR). As the most important part of event studies is to test the significance of these returns, I really need your help.
Can the formula (from Princenton) also be applied to AAR`s or CAAR´s? Or do i have to use the so called "cross sectional t-test" provided by https://www.eventstudytools.com/significance-tests --> Second test described - [2] Cross-Sectional Test (Abbr.: CSect T) ? If have to use it how to calculate this one in Stata? I already tried to calculate it but I failed.
The code i used to calculate this cross sectional test is:
sort dif Date
(1) by dif: gen caar_test = (car_all - av_cumsum)^2
(2) gen sum_caar_t= sum(sd_caar_test)
(3) by dif: gen free= (1/(486*(486-2)))
(4) by dif: gen std_caar = sqrt(free*sd_caar_test)
(5) by dif: gen t_caar=av_cumsum/std_caar
To understand this code a short variable definition is provided for the above code:
dif = the difference of days to the event (e.g - 5 is 5 days before the event) and after the event (e.g. 5 is five days after the event).
Date = Date variable
car_all = cumulative abnormal return for the whole sample: calculated by summing up the abnormal returns over the event window
av_cumsum = The CAAR - cumulative average abnormal returns
I calculated the formula stepwise to get an understanding of it!
The calculation works but the t-values are changing for the different companies over the var. dif but this should not be as the CAAR are the same for the whole companies over the respective dif in the sample.
So I need a t-value for dif =- 5 for all companies the same and for dif=-4 the same t-value and so on!
Where is my mistake?
Does anyone have ever made such t-statistics in Stata and could please help me
?
Thanks in advance
Christoph
I am conducting an Event Study in Stata by following the instructions of Princenton University (URL: https://dss.princeton.edu/online_hel...ventstudy.html) . I tested the abnormal returns in stata with the code provided by Princeton University. So that was no problem. As I have more events to test (243 Mergers and Acquisitions), I also had to calculate the average abnormal returns (AAR) & the cumulative average abnormal returns (CAAR). As the most important part of event studies is to test the significance of these returns, I really need your help.
Can the formula (from Princenton) also be applied to AAR`s or CAAR´s? Or do i have to use the so called "cross sectional t-test" provided by https://www.eventstudytools.com/significance-tests --> Second test described - [2] Cross-Sectional Test (Abbr.: CSect T) ? If have to use it how to calculate this one in Stata? I already tried to calculate it but I failed.
The code i used to calculate this cross sectional test is:
sort dif Date
(1) by dif: gen caar_test = (car_all - av_cumsum)^2
(2) gen sum_caar_t= sum(sd_caar_test)
(3) by dif: gen free= (1/(486*(486-2)))
(4) by dif: gen std_caar = sqrt(free*sd_caar_test)
(5) by dif: gen t_caar=av_cumsum/std_caar
To understand this code a short variable definition is provided for the above code:
dif = the difference of days to the event (e.g - 5 is 5 days before the event) and after the event (e.g. 5 is five days after the event).
Date = Date variable
car_all = cumulative abnormal return for the whole sample: calculated by summing up the abnormal returns over the event window
av_cumsum = The CAAR - cumulative average abnormal returns
I calculated the formula stepwise to get an understanding of it!
The calculation works but the t-values are changing for the different companies over the var. dif but this should not be as the CAAR are the same for the whole companies over the respective dif in the sample.
So I need a t-value for dif =- 5 for all companies the same and for dif=-4 the same t-value and so on!
Where is my mistake?
Does anyone have ever made such t-statistics in Stata and could please help me
?Thanks in advance
Christoph
