Re: st: re: Re: Panel data and lagged dep variable

[Clive Nicholas <clivelists@googlemail.com>]

Kit Baum replied to Pan Wang:

> If there are significant time-specific common factors (e.g., business cycle factors) appearing in the response variable,
> then by omitting time dummies you are misspecifying the model, and estimates of any coefficients (including
> the coefficient of the LDV) are likely to be biased and inconsistent. Include the dummies, do an F-test for their
> joint significance, and if you don't need them, out they go. But determining whether a static or dynamic model
> is needed should be performed in the context of a well-specified model.

Absolutely right, of course. But frequently I find myself estimating
unbalanced pooled-panel models with singleton unit dummies, which
naturally drop out of the model. Consequently, I have to circumvent
this issue by demeaning the explanatory variables within units and
then re-estimating. But, as you'll know, that typically comes at the
cost of reduce the reducing the amount of variation in the response
variable my model would have had a chance of explaining. Your answer
to this might be: "Then collect more data". But, as you'll also know,
that's not always possible.

-- 
Clive Nicholas

[Please DO NOT mail me personally here, but at
<clivenicholas@hotmail.com>. Please respond to contributions I make in
a list thread here. Thanks!]

"My colleagues in the social sciences talk a great deal about
methodology. I prefer to call it style." -- Freeman J. Dyson.
*
*   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/