Hello,
I should say at the outset, that I'm new to Stata and so I still learning my
way around the command structure etc. That said, I am attempting to estimate
a panel data set. I have 401 firms, with quarterly data for 10 years.
Preliminary diagnostic tests indicate that the disturbances exhibit serial
correlation and heterskedasticity. Moreover given that many of these firms
operate in the same industry, there is likely to be strong cross correlation
of the disturbances as well. The possibility (and tests confirms this as
well) that there is also likely to be fixed effects ( the hausman test
suggested that the fixed effect model was preferred to the random effects
model).
My first question is, how reliable are tests of poolability (ANOVA), or fe
versus re in the presence of serial correlation, non constant variance and
cross correlation in the disturbances? My suspicion is that they are not,
but I haven't been able to confirm this. Given this suspicion, I decided to
first use FGLS (xtgls) estimate the model as this procedure is robust to
violations in the error terms. I was thinking that I could then employ the
option for "fe" or "re" however, neither of these options is available for
xtgls.
I'm somewhat stumped at this point as to how I might proceed. Any
suggestions would be greatly appreciated.
Thanks
_________________________________________________________________
FREE pop-up blocking with the new MSN Toolbar � get it now!
http://toolbar.msn.com/go/onm00200415ave/direct/01/
*
* 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/
