Event study/variance equality cross section

[Genc007]

Hi everyone!

I have been on this problem for 3 days straight now and I have no idea how to solve this. So hopefully you can help me!

I am doing an event study on the stock market effects of a ceo succession announcement.
For this i have the following data:

1) 124 CEO announcements of stock listed firms.

2) Daily abnormal returns (AR).
I have basically 2 event windows: [-200,-26] which is the estimation window and [-15,+15] which is the event window.

So the structure of my excel file is:
_____ announcement 1 ________ announcement 2 ______ announcement 3______ annnouncement 4 _______ etc.
-200 ____ AR _________________________AR ___________________AR ____________________AR
-199 ____ AR _________________________AR ___________________AR ____________________AR
..
..
+14 ____ AR _________________________AR ___________________AR ____________________AR
+15 ____ AR _________________________AR ___________________AR ____________________AR

So basically I have 2 event windows that belong together. The estimation window and the event window.
Now I need to verify that the variance of the abnormal returns during the estimation window is not statistically different than the variance of the returns in the event window. And I need a Levene's test of variance equality for this.

My question is, how do I get Eviews to compare the variances of these two event windows?

[EViews Glenn]

Create a 0-1 dummy variable for your event window, say EVENT. Open the series you want to analyze, select View/Descriptive Statistics & Tests/Equality Tests By Classification...

Enter the name of the dummy in the edit field. Choose Variance test. The rest of the defaults should be fine. Click on OK.

[Genc007]

I tried that already, but I'll clarify my post.

Right now I have the excel file with the event dates on the row headers and the announcements themselves on the column headers. (i could also transpose this data).

So what I need to do is compare the average variance of the estimation window with the average variance of the event window.

So i have 124 events. This means I need to compare the average estimation window variance for all N announcements, through days [-200,-26] with the variance of the event window of [-15,15] for example.

Is this possible? Because right now, when I load the excel file into Eviews, I get either the variance per event, or per day.

EDIT: SO if i transpose my data, and average the estimation window returns, and the compare the variance of this series (through N events), with the variance of the averaged event window returns, will this work?
And will these be the correct variances?

[EViews Glenn]

I must admit that I'm totally confused about your data structure.

Your say that your row headers are event dates. You also said something about 124 events. Are there 124 rows corresponding to the event dates? Or are events the same as announcements where each of the 124 columns. IN the latter case, I'm assuming that the "event dates" represent something like days before and after the announcement?

[Genc007]

İ'll clarify.

İm doing an event study, and for this event study i calculated abnormal returns for the event window of [-200,+15],
With day 0 being the event days.

Now what i need to do is verify that on the event day, which is day 0(or for example The period [-1,,+1]), the variance was significantly different than the variance during the estimation window of [-200,-26].

So the rows represent the 124 event that İ identified and the columns represent the event dates (ranging from day -200 before the event to day 15 after the event)

So i need a Levene's test of equality of variances between for example the variance of the abnormal returns in the estimation window of -200 to -26 And the event period of [-1,+1].

How can i organize my data the best to do this? Should i maybe make a panel structure? Or just average the abnormal returns during the two periods and then compare the variances of these averages? İs this a valid test of equal period variances?
Or maybe just add all relevant days (for example -200 to -26 and. -1 to 1) to a group and perform a test of equality - variance?