I am conducting an event study for different event studies. My event windows are [-45,45] ; ; [-5,5] ; [_2,2].
I conduct an event study about the abnormal returns of different deals around the announcement date. I have calculated the Abnormal Returns (AR'S) and the Cumulative Abnormal Returns (CAR's) at the end of every event window.
I have 9 deals. For every day I can calculate the AARs or Average Abnormal Returns. Subsequently I can calculate the Standard error for every day. Now the big question is about the CAAR's, or Cumulative Averrage Abnormal returns. I would like to show the significance of the CAARs by using a standard t-test.
As I understand, from other studies, this can be done by dividing the CAAR with (the Standard error of the AARs * the square root of T). Where T is the amount of days in the event window.
These test are all one day test statistics. I would like to show one complete number in a table for every event window. How can this be done? Only conducting the T-test on the last day of the event window (after the entire accumilation?) Summing all t-test of every eventday?
Let's assume you have average return score for each day, x1, x2, ..., xT. Is it possible that you average these scores to find a grand average which represents the return across days? The t-statistic for this grand average is the ratio of the grand average to the standard error of the grand average. The standard error of the grand average = standard deviation of x1, x2, ..., xT divided by square root of T. Where T is the number of days.
- Badges
- Science topic