Clustering - Petersen 2009
Posted: 27 May 2012 07:32 PM   [ Ignore ]
Administrator
Avatar
RankRankRankRank
Total Posts:  901
Joined  2011-09-19

Petersen, 2009, Estimating Standard Errors in Finance Panel Data Sets: Comparing Approaches
In: Review of Financial Studies, January, 2009, Volume 22, pp 435-480.

Overview: http://www.kellogg.northwestern.edu/faculty/petersen/htm/papers/standarderror.html
and http://www.kellogg.northwestern.edu/faculty/petersen/htm/papers/se/se_programming.htm
Scroll down for SAS.

SAS code, John McInnis, https://webspace.utexas.edu/johnmac/www/cluster code.htm and
http://www.kellogg.northwestern.edu/faculty/gow/htm/GOT/

 Signature 

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

Profile
 
 
Posted: 11 January 2013 06:50 PM   [ Ignore ]   [ # 1 ]
Sr. Member
RankRankRankRank
Total Posts:  169
Joined  2011-09-20

Very important topic.
It seems stata is easier for this kind of task.


here are simulation data and results from professor Petersen’s website:
http://www.kellogg.northwestern.edu/faculty/petersen/htm/papers/se/test_data.htm

some other related sas code from professor Stoffman:
http://kelley.iu.edu/nstoffma/

 Signature 

Zenghui
A humble student of business

Profile
 
 
Posted: 11 January 2013 07:10 PM   [ Ignore ]   [ # 2 ]
Administrator
Avatar
RankRankRankRank
Total Posts:  901
Joined  2011-09-19

Indeed; the SAS code is unstable. It is throwing weird errors by slight modifications of the model..

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: 14 January 2013 04:59 PM   [ Ignore ]   [ # 3 ]
Sr. Member
RankRankRankRank
Total Posts:  169
Joined  2011-09-20

Joost,

Is it correct that clustering (one or two dimentions) will not change the degree of freedom of regression?

Thanks,
Zenghui

 Signature 

Zenghui
A humble student of business

Profile
 
 
Posted: 15 January 2013 10:19 AM   [ Ignore ]   [ # 4 ]
Administrator
Avatar
RankRankRankRank
Total Posts:  901
Joined  2011-09-19

hi Zenghui,

I am not sure if this is by definition, but what I have been seeing is that the coefficients remain unchanged, but the standard errors get larger. (would mean that degrees of freedom drops; I am not 100% sure though).

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