Dear Stata-listers,
In my dissertation data, I have an unbalanced panel of 439 firms
tracked across 25 years (a total of 3946 firm-year observations).
Since the main explanatory variables of interest are time invariant, I
am using random effects estimation (xtreg, re). My question pertains
to how Stata treats time effects.
I read in Kennedy, 2003 (p.311) that "In the random effects estimation
procedure the time period specific error component is added instead of
a more familiar year dummy variable used in fixed effects". My
question is - does Stata xtreg, re, automatically "know" -i.e.
include- the period error in the estimation or do I need to create
year dummies after all and include them in the analysis?
although my results are for the most part consistent whichever option
I choose - there is slight difference in significance of some
variables, and I would like to make sure I am specifying the model
correctly. I tried searching for an answer on statalistserv, but had
no luck finding it. Thank you in advance for your time and effort, any
help would be sincerely appreciated.
Have a wonderful day,
Smiles
Hana
*
* 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/
