[Gretl-users] problem with system instrumental variables

[nadaud at centre-cired.fr]

Hello again, and thank you Sven for your advice
> I believe that variable 0 refers to the constant. And for TSLS there 
> was recently some change in whether the constant is automatically 
> added or not. Don't know whether this applies to 3SLS as well.
Yes that is right, I add the constant because in my AIDS system, 
demographics translate the constant.


>
> So until someone knowledgeable answers this, I can think of two 
> workarounds that may or may not work:
>
> 1) try leaving out the const and check afterwards whether gretl has 
> added it automatically
> 2) update to a newer version; if you already use 1.7.1 maybe use the 
> latest windows snapshot (or cvs, respectively).
>
> No guarantees, but good luck!

And after doing both, the problem persist !

I wonder if it is possible to estimate systems with just one different 
variables ? Looking at the example scripts it seems so. Also with the 
tsls syntax i can normally specify different instruments for both 
equations nope ?

Well let me try something else...

sincerely

Franck