Question on an event study with foreign exchange data

[Yoran Verbloem]

Hello!

I am trying to do an "event study" on foreign exchange data from the euro usd pair. I have a number of announcements, which are the events, and only 1 asset, the forex pair. I am currently trying to setup the code so that it calculates for this pair the abnormal returns by taking the average of 3 days after the event and substracting the average of the returns 3 days prior to the event. The event is the announcement day. After this I will run a regression on these abnormal returns but that is not the problem at this moment.

I have forex data, with the date and the returns, and i have a list of event dates, here are the first rows:

Fxdate	Return
03/01/2005	-0,007
04/01/2005	-0,0142
05/01/2005	-0,0008
06/01/2005	-0,0069

QEannDate

14/01/2005

22/08/2007

08/11/2007

23/11/2007

I have tried the package eventstudy, but this is not usable because it only allows to look at 1 event, and eventstudy2 does not seem to work for my case as well.

Could someone help with pointing at either a package that can do this or a guide that walks through the needed code for this? I am also not sure if I should try to code myself or use a package for this.

Any help is of course greatly appreciated and thank you for your time!

-Y

[Thomas Kaspereit]

Dear Yoran,

eventstudy2 can be used for this.

There, you have to use the COMEAM model and select [-3,-1] as the estimation window, using the eswlbband eswub option. Then define one of the CAR windows as [+1,+3] using the car*lb car*ub options.

Best Thomas Kaspereit