Stll, I don't know if analogue is true for pvars ((trend)-stationary, case I in Hsiao): Let  we have pvar(1) with only x_it and y_it

If estimate dynamic panels for x_it on x_i(t-1) and y_i(t-1) and y_it on x_i(t-1) and y_i(t-1) separately is the same as system gmm,

almost only thing to do is organize bootstrap ci for irfs.

I think, two step approach for panel cointegration is a good thing, because without good shortcuts the general formula is intractable.

Resent survey by Canova: http://www.ecb.europa.eu/pub/pdf/scpwps/ecbwp1507.pdf (Suspect, you know him personally)

This semester I have improbable 26 hours of lectures and labs per week, and spring will be completely empty. At last, I am

going to fini! sh uk.po. I hope I know how to write heteroscedastic (cross-sectional) logit and  probit and mfx for cross-sectional

probit. If it is desirable, I think I can manage functions for them.

 

>> panel data VAR (as described in Hsiao, for example)
>
> don't know

We do have GMM estimators for dynamic linear models for panel data 
(commonly known as GMM-DIF and GMM-SYS). It could be nice to package some 
panel cointegration procedure: Pedroni's apporoach was based on two-step 
estimation, which is clearly easily feasible with the techniques we 
already have. I'm not aware of more recent methods in widespread use. 
Ideas, anyone?

-------------------------------------------------------
   Riccardo (Jack) Lucchetti
   Dipartimento di Scienze Economiche e Sociali (DiSES)

   Università Politecnica delle Marche
   (formerly known as Università di Ancona)

   r.lucchetti@univpm.it
   http://www2.econ.univpm.it/servizi/hpp/lucchetti
-------------------------------------------------------

_______________________________________________
Gretl-users mailing list
Gretl-users@lists.wfu.edu
http://lists.wfu.edu/mailman/listinfo/gretl-users