Bookmark and Share

Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down on April 23, and its replacement, statalist.org is already up and running.


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

st: Autocorrelation in Panel Data - XTGLS first differenced or corr(ar1)?


From   "Levchak, Philip J" <philip-levchak@uiowa.edu>
To   "statalist@hsphsun2.harvard.edu" <statalist@hsphsun2.harvard.edu>
Subject   st: Autocorrelation in Panel Data - XTGLS first differenced or corr(ar1)?
Date   Wed, 19 Dec 2012 05:02:44 +0000

Hello,

I am analyzing a dataset of 66 nations over 20 years (unbalanced) - using xtgls.  My biggest concern with the data is the presence of autocorrelation.  I test for autocorrelation using Woolridge's test for serial correlation.  The regression is in first differences and is weighted by population size.  It indicates no panel autocorrelation.  Do I need to keep the regression in differenced form for xtgls?  I have tried using xtgls, specifying corr(psar1) in non-differenced form.  However, the results from the differenced regression and the non-differenced (psar1) regression are quite different.  Is there a reason that I should use one method over the other?  

I also tried the user-written test for autocorrelation (xtactest) following xtgls (link below).  Regardless of what method I use (differenced or psar1), the test indicates serial correlation in the residuals.
  
http://www.stata.com/statalist/archive/2003-11/msg00722.html
   
Thanks.

Phil Levchak
Ph.D. Candidate
Department of Sociology
University of Iowa
Iowa City, IA 52245
*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/


© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   Site index