Calculate value weighted return for event window
Posted: 16 August 2014 11:19 AM   [ Ignore ]
Newbie
Rank
Total Posts:  21
Joined  2014-07-18

I am trying to calculate the value weighted return for an event window.  The paper I am attempting to replicate has in the results VWRET and a t-stat for each event.  There are multiple events.  But, I am just looking at one right now.  The event window is 1/15/2002 - 1/18/2002 and the event is on 1/17/2002.  I have a sample of 6500 US companies returns and the file is set up with the returns for each day 1/15-1/18 and a variable called event.  This event is number 1.  I calculated the mean based on the event 1 for the companies in my sample.  Is this not the way to do it?  Because my result is -.002 and the result in the paper for the same window is -.0103. I am using proc means to get the mean over the four day window.  Also, what test do I run to get the t-statistic.  I am teaching myself how to do this.  Am I completely wrong in my approach?

I cannot use eventus because my control is foreign returns.  I also have to calculate the foreign returns for the event window.  Therefore, I need to know how to do this correctly.  I have researched on the internet, but have not found information that can help me.  Please help!

Thanks in advance!

Profile
 
 
Posted: 16 August 2014 02:33 PM   [ Ignore ]   [ # 1 ]
Administrator
Avatar
RankRankRankRank
Total Posts:  901
Joined  2011-09-19

hi Samme,

So, if I understand correctly, you have 6500 firms, with 4 days of stock return (1/15 - 1/18) for each firm. These returns need to be ‘aggregated’ into a single, value weighted return.

The first thing I would do is to compute an ‘aggregate’ return for each firm, where you add the 4 daily returns into a single return. (Either compounded, or the sum). Then you will have 6500 returns for 6500 firms.

It seems to me that the next step would be to weight each of these 6500 returns by firm value. So at this step you need market cap for each of the 6500 returns to use as a weight.

See here how to use proc means using weight: http://support.sas.com/documentation/cdl/en/proc/61895/HTML/default/viewer.htm#a002473731.htm

Scroll down to this example

proc means data=size maxdec=3 n mean var stddev;
   
weight precision;
   var 
objectsize;
   
output out=wtstats var=Est_SigmaSq std=Est_Sigma;
   
title1 'Weighted Analysis Using Default VARDEF=DF';
run

For an example how to do a t-test in SAS , see here: http://www.ats.ucla.edu/stat/sas/output/ttest.htm

Something like this, where dataset mydata holds a variable eventret (6500 obs):

proc ttest data=mydata H0=0/* test if eventret is 0 */
var eventret;
run

Hope this helps,

Joost

 Signature 

To reply/post new questions: Please use the group WRDS/SAS on Google Groups! http://groups.google.com/d/forum/wrdssas

Profile
 
 
Posted: 16 August 2014 03:55 PM   [ Ignore ]   [ # 2 ]
Newbie
Rank
Total Posts:  21
Joined  2014-07-18

Thank you for your response!  Do I need to calculate the value weighted raw return for the companies in CRSP?  Or can I use the value weighted return provided?

Thanks again!

Profile
 
 
Posted: 16 August 2014 06:47 PM   [ Ignore ]   [ # 3 ]
Administrator
Avatar
RankRankRankRank
Total Posts:  901
Joined  2011-09-19

hi Samme,

How would it be provided? Are these the full ‘universe’ of US stocks?

best regards,

Joost

 Signature 

To reply/post new questions: Please use the group WRDS/SAS on Google Groups! http://groups.google.com/d/forum/wrdssas

Profile
 
 
Posted: 16 August 2014 06:57 PM   [ Ignore ]   [ # 4 ]
Newbie
Rank
Total Posts:  21
Joined  2014-07-18

Hi Joost,

I think I now understand that I need to calculate the value weighted raw return.  The vwret in CRSP is for the all stocks, correct?  I didn’t realize that, but I am learning!  Thanks for your assistance and you patience!!

samme

Profile
 
 
Posted: 16 August 2014 07:52 PM   [ Ignore ]   [ # 5 ]
Administrator
Avatar
RankRankRankRank
Total Posts:  901
Joined  2011-09-19

hi Samme,

Value weighting or equal weighting is available for indices. If you have your own portfolio of firms, then you would have to compute it yourself. If your 6500 firms are some sort of index (like NYSE/Nasdaq), then you could use the value weighted returns that are provided in CRSP.

best regard,

Joost

 Signature 

To reply/post new questions: Please use the group WRDS/SAS on Google Groups! http://groups.google.com/d/forum/wrdssas

Profile
 
 
Posted: 16 August 2014 08:40 PM   [ Ignore ]   [ # 6 ]
Newbie
Rank
Total Posts:  21
Joined  2014-07-18

I do have my own portfolio of firms.  I am working on the calculation now.  Thank you so much for your assistance!

samme

Profile
 
 
Posted: 16 August 2014 09:06 PM   [ Ignore ]   [ # 7 ]
Administrator
Avatar
RankRankRankRank
Total Posts:  901
Joined  2011-09-19

You’re welcome - sounds like fun! smile

Joost

 Signature 

To reply/post new questions: Please use the group WRDS/SAS on Google Groups! http://groups.google.com/d/forum/wrdssas

Profile
 
 
Posted: 23 August 2014 05:25 PM   [ Ignore ]   [ # 8 ]
Newbie
Rank
Total Posts:  21
Joined  2014-07-18

Hi!  I have a few more questions..

I have calculated the value weighted U.S return for each event and my results are very close to paper I am attempting to replicate results.  There are 17 events.  I have 17 value weighted raw returns.

The author then appears to add the value weighted returns for four events she finds significant and calls this cumulative value weighted return.  There is a t-statistic associated with this number.  How would this be calculated?

Thank you!!

I

Profile
 
 
Posted: 23 August 2014 07:31 PM   [ Ignore ]   [ # 9 ]
Administrator
Avatar
RankRankRankRank
Total Posts:  901
Joined  2011-09-19

hi,

Would it be the case that the author just ‘added’ all the returns of the 4 events together as if it were a single event? That would be the easiest way at least. I don’t think that a t-test would be appropriate for testing if the sum of 4 mean returns are jointly different from zero. (If it were appropriate, it is probably a weak test/little statistical power). (I am not too sure about this though.)

Hope this helps,

Joost

 Signature 

To reply/post new questions: Please use the group WRDS/SAS on Google Groups! http://groups.google.com/d/forum/wrdssas

Profile
 
 
Posted: 24 August 2014 12:49 AM   [ Ignore ]   [ # 10 ]
Newbie
Rank
Total Posts:  21
Joined  2014-07-18

Thank you, Joost!

Profile
 
 
Posted: 24 August 2014 08:09 AM   [ Ignore ]   [ # 11 ]
Administrator
Avatar
RankRankRankRank
Total Posts:  901
Joined  2011-09-19

You’re welcome! smile

Joost

 Signature 

To reply/post new questions: Please use the group WRDS/SAS on Google Groups! http://groups.google.com/d/forum/wrdssas

Profile