BVAR

[olsisthebest]

It is possible to impose restrictions in the parameters, as in SVAR, in a BVAR estimation?
I mean, how can I estimate a SBVAR in EViews 9?
Thank you.

[EViews Gareth]

No

[olsisthebest]

What are the differences between Normal-Wishart and Sims-Zha Normal Wishart priors?
Thank you.

[EViews Gareth]

http://www.eviews.com/help/helpintro.ht ... 01.html%23

[olsisthebest]

Can I impose diferent hiperparameter values across equations?

For instance, in equation 1 the dependent variable behaves like a random walk then I set mu=1, in equation 2 the dependent variable behaves like a white noise then set mu=0?
And the same for the other hyperparameters.

How can I do that?

[EViews Gareth]

No.

[olsisthebest]

I am using Sims-Zha NW prior to estimate a BVAR model.
When I use growth rates I don't have any problem. However, when I use the same variables but in levels I get the error "Near singular matrix".
Do you have any idea what is going on?

[EViews Gareth]

Not without seeing it.

[olsisthebest]

Please find attached the workfile.
I get the error message "near singular matrix" if i consider the variables in levels.
With growht rates (prefix "dl") I don't have any problem.
Any idea?
Thank you.

[olsisthebest]

Dear all,
Does anybody have an idea about the aforedmentioned problem?
Thank you.

[startz]

Is your version of EViews uptodate?

[olsisthebest]

Yes it is.
Did you try to estimate the BVAR with Sims-Zha Normal Wishart priors with variables in levels?

[startz]

Oh. I had just done a regular VAR. That works fine. I can confirm the Bayesian VAR does not work...and that does seem weird.

[olsisthebest]

Yes, it is really weird. It is suppose to work fine since it is not possible to have perfect multicollinearity.
Let's wait for forum moderation's insights.

[EViews Gareth]

Your hyperparameters are crazy, due to a bug in saving BVARs. We'll fix the bug, but in the meantime you'll need to re-specify the hyperparameters after opening the workfile.