Stata The Stata listserver
[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

st: -areg-, -xtreg- and -cluster- (again)


From   "Richard Upward" <[email protected]>
To   <[email protected]>
Subject   st: -areg-, -xtreg- and -cluster- (again)
Date   Thu, 08 Jul 2004 12:47:36 +0100

Apologies for revisiting a topic which has been visited on several occasions over the last couple of years, but could I ask if anyone can clarify the current thinking regarding the relationship between 

-areg-
-xtreg-
and the -,cluster()- option.

(1) Since -xtreg, fe- and -areg- are equivalent, why is it not possible to request robust standard errors with -xtreg, fe- but it is possible with -areg-?

[This question was also asked on Statalist by <[email protected]> in November 2003, but I could not track down any answer.]

(2) The manual warns us that the number of levels in the absorb() variable should not exceed the number of clusters. ([R] areg, p.85).  However, there is a FAQ at

http://www.stata.com/support/faqs/stat/aregclust.html

which concludes that "Further review has shown that cluster() can be combined with areg, even if the clusters match the fixed-effects."  

This FAQ is dated 18 May 2004.  Is the caution in the manual still good advice?  Does the FAQ imply that one can actually have more fixed-effects than clusters?   

It seems that this would be a common requirement in a model where one has more than one level of aggregation (e.g. workers and firms, pupils and schools), and one estimates a model which controls for the lower level using fixed-effects but allows for correlation in the errors at the higher level.  Thus the number of clusters would generally be (much) smaller than the number of fixed-effects.

(3) Finally, is the problem simply one of a degrees-of-freedom correction which can easily be implemented post-estimation?  For example, could one "within-transform" the data and estimate a simple -regress, cluster()- command and then do the standard D.o.F. correction which takes into account the number of fixed-effects?  Or could one do this correction directly after -areg, cluster()- without transforming the data?

Any clarification on these issues would be gratefully received.  

Richard


Richard Upward
School of Economics
University of Nottingham
University Park
Nottingham NG7 2RD
Tel: +44 (0) 115 95 14735

http://www.nottingham.ac.uk/economics/staff/details/richard_upward.html


This message has been scanned but we cannot guarantee that it and any
attachments are free from viruses or other damaging content: you are
advised to perform your own checks.  Email communications with the
University of Nottingham may be monitored as permitted by UK legislation.


*
*   For searches and help try:
*   http://www.stata.com/support/faqs/res/findit.html
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/



© Copyright 1996–2024 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index