Dear all,
I've got several questions about panel data models.
1. Is it a common view for practicing econometricians that Cross
Section models are appropriate for making inferences in the long run,
while Time Series and Panel models are best for the short run
inferences (Baltagi, Griffin (1984), Green, Kim, Yoon (2001))?
2. Intuitively this separation seems somewhat strange to me, for
example, if researcher has a panel model in which some regressors are
changing over time rather slowly (e.g., fertility, urbanization rate,
or literacy rate in developed countries), how can the corresponding
estimated coefficients be called short run estimates? Besides I doubt
that there is a clear-cut criterion of short/long run period, is
there?
3. If long run and short run estimates concept from point 1 is
correct, then using a panel dataset one can (1) estimate a within
(fixed-effects) model and acquire short run estimates, (2) estimate a
between-effects model to acquire long run estimates, and (3) estimate
a random-effect model. Since the latter is a weighted average of FE &
BE models, does it mean that one gets a mean estimate of long run and
short run estimates?
4. Does existence of individual effects mean that once one believes in
their presence and estimates a panel model, it automatically prohibits
him/her to estimate a cross section model, since the letter will
result in biased estimates because of omitted variable (individual
effect) bias? Or a researcher can suppose that in the long run
individual effects are insignificant and estimate a cross section
model, and at the same time suppose that in the short run the effects
are significant and estimate a panel model?
Thank you in advance,
Vladimir Dashkeyev.
*
* 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/