Page 1 of 1
Non-recursive overidentified SVAR
Posted: Thu Feb 06, 2014 4:27 pm
by researcher84
I'm trying to estimate Non-recursive over identified SVAR model. Can we estimate such model using eviews? if not then which program I should use? any resources on estimating this type of models? Thanks a lot for any feedback. I'm really stuck right now and in need for help.
Re: Non-recursive overidentified SVAR
Posted: Fri Feb 07, 2014 1:20 am
by trubador
Yes, you can. See the chapter on "Structural (Identified) VARs" in the manual.
Re: Non-recursive overidentified SVAR
Posted: Wed Feb 12, 2014 4:08 pm
by researcher84
The thing is once I enter the restrictions in eviews which are short-term restrictions. I get the msg "Hessian matrix is near-singular at final iteration parameter values". Please see the structure for the model I'm trying to estimate here
http://postimg.org/image/qyo59tfxv/ Thanks for your help.
Re: Non-recursive overidentified SVAR
Posted: Wed Feb 12, 2014 8:24 pm
by terrya
For what it's worth, I've had the same trouble with this specification in EViews but I can get output that seems plausible from RATS (using the CVModel).
Of course, I'm applying the model to different data (NZ-based).
Re: Non-recursive overidentified SVAR
Posted: Thu Feb 13, 2014 12:08 am
by researcher84
For what it's worth, I've had the same trouble with this specification in EViews but I can get output that seems plausible from RATS (using the CVModel).
Of course, I'm applying the model to different data (NZ-based).
Can you please tell me what is (CVModel), I just bought RATS. Thanks for your reply.
Re: Non-recursive overidentified SVAR
Posted: Thu Feb 13, 2014 1:38 am
by terrya
My best advice is to read the manuals and look at the examples supplied with Rats.