Hi all,
I am analyzing unbalanced panel investment data (8 year period 120
panels approximately 750 observations) with the xtabond2 command. I
notice that when I include dummy variables for each of the years the
Hansen J and AR(2) test values fall out of acceptable range, and
frankly the results are less significant (with no real change in
standard errors). Also, the year dummies are insignificant.
Is it absolutely necessary to control for the year? I cannot really
see any reason to include them. I've seen similar papers mention
controlling for years about 50% of the time (usually in a footnote). I
am assuming they are included when it helps the results, not included
when they do not. Any thoughts?
Thanks!
*
* For searches and help try:
* http://www.stata.com/help.cgi?search
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
