Page 1 of 1

Issue with SVAR identifying restrictions

Posted: Mon Nov 09, 2015 12:19 am
by AlunaBic
Hi all,

I'm currently using EViews 8 and I'm attempting to estimate a SVAR model. However, when I input my identifying restrictions, I received the error message "Hessian matrix is near singular at final iteration parameter values". The restrictions I'm attempting to impose are:
R1 1 0 0 0 0 0 0
R2 NA 1 0 0 0 0 0
R3 NA 0 1 0 0 0 0
R4 NA 0 NA 1 0 0 0
R5 NA 0 0 NA 1 0 NA
R6 NA NA NA NA NA 1 0
R7 NA NA NA NA NA NA 1

I am only able to get the model to run if I eliminate the final 2 restrictions of R7, but unfortunately these are fairly important to the model. I've tried to change the starting values as well as manually input the restrictions in text form, but unfortunately neither of those have solved the issue. If anyone has any ideas how I can solve this problem it would be greatly appreciated.

Thanks,

Aluna

Re: Issue with SVAR identifying restrictions

Posted: Mon Nov 09, 2015 2:12 am
by trubador

Re: Issue with SVAR identifying restrictions

Posted: Mon Nov 09, 2015 2:28 am
by AlunaBic
Thanks for the response.

I've read all I can find on the forums to do with SVAR, including that post, but was hoping that maybe someone had found a solution since then.

So basically I can't estimate a non-recursive SVAR using EViews?