Page 1 of 1

Diebold et al. (2006) - Estimation problem

Posted: Sun Nov 13, 2016 1:22 pm
by sican
Hi guys!

I am trying to do a similar stuff as Diebold, Rudebusch and Aruoba (2006) made, but I keep getting an error of:
WARNING: Singular covariance - coefficients are not unique .
I checked the forum and saw that sometimes it has to be with the initial coefficient values, but I don't know how to define the best ones to use, so if you can help me, I'd really appreciate it.

Please find my workfile attached, so you can see if I'm doing something wrong.

Thanks in advance!

Re: Diebold et al. (2006) - Estimation problem

Posted: Tue Nov 15, 2016 6:36 am
by sican
Just to bring the topic up. Does someone knows how to help me?
Thanks!

Re: Diebold et al. (2006) - Estimation problem

Posted: Tue Nov 15, 2016 7:40 am
by dakila
The problem is related to Eviews optimization algorithm, but the initial coefficients values.

Re: Diebold et al. (2006) - Estimation problem

Posted: Tue Nov 15, 2016 11:50 am
by sican
The problem is related to Eviews optimization algorithm, but the initial coefficients values.
Sorry, but I didn't understand. My english isn't that great. :) Did you mean that the problem is related with the initial coefficient values?

Re: Diebold et al. (2006) - Estimation problem

Posted: Wed Nov 16, 2016 7:33 am
by dakila
No

Re: Diebold et al. (2006) - Estimation problem

Posted: Wed Nov 16, 2016 1:32 pm
by sican
No
So what exactly is the problem regarding the eviews optimization algorithm? I'm using the Marquardt algorithm, with a maximum number of iterations of 500 and a convergence tolerance of 10 E-6. Btw, I'm using Eviews 8.

Re: Diebold et al. (2006) - Estimation problem

Posted: Sat Nov 19, 2016 9:31 am
by sican
Anyone? :(