st: lagged indep vars and panel data

[Kit Baum <baum@bc.edu>]

Cordula said

> I have a question about a panel data model. I want to estimate a 
> regression
> containing lagged x's, but no lagged y's, i.e.
>
>  frag fdi l.fdi l2.fdi & some dummies
>
> I'm not sure how to do that with Stata. As far as I can see, the 
> xtabond
> command is only for dynamic panel data models with lagged dependent
> variables. I've tried the usual xtreg..., but then Stata tells me that 
> I
> can't use time series operators. Does anyone know the command I should 
> use?
Problem is that the xtreg command is very, very old code that was 
written before L. D. F.
It will work just fine if you create new variables (with generate) 
equal to the appropriate lags: e.g.

gen lfdi = L.fdi

and so on, and then include those variables in the xtreg regressor 
list. Other 'xt' commands are more recent inventions, and know how to 
use the time series operators.

If xtserial detects nonindependence in the errors, it is signalling 
that the errors within a single panel are not serially independent. 
Presumably xtgls would be an appropriate estimator in this context. See 
the discussion in Wooldridge cited in whelp xtserial for more details.

Kit



*
*   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/