BVAR (Bayesian VAR)

[EViews Gareth]

You're right, the current version of MSBVAR does support exogenous variables. When I have time, I'll see if I can add support for them. In the mean time, if any one else want to tinker with the add-in a bit and re-submit it, I'll gladly accept it!

[donihue]

Oh well ....

In any case, many many thanks for this excellent initiative. This (and the VAR forecast) add-in has made EViews (and indirectly also MSBVAR!) vastly more useful for many of us.

Regards
Donihue

[sjamescppib]

I have just installed the Bayesian VAR add-in. Whenever I try to estimate I receive the message "syntax error." Does anyone have any thoughts on what might be causing this?

[EViews Gareth]

Perhaps you could provide the workfile and detailed instructions on what you are doing.

[dsinigaglia]

This initiative is great, but I guess there are still some important functionalities there are missing. Quick questions/comments:

a) Does the add-in support exogenous variables other than a constant (like, dummy variables, for example)? It seems R can handle but not the add-in. I think that would represent a important advantage over the competition.
b) Is there anyway I could use the add-in to make forecasts? The great limitation, it seems, is that there is no such thing as a BVAR object that can be used for model forecasting, as in the case of a standard VAR. Can that be corrected somehow?
c) There are no confidence intervals in the IRFs. It would be interesting if that could be incorporated, accounting for both parameter and shocks uncertainties. Could that also be corrected?
d) Finally: estimation output only gives point estimates, which does not make any sense. Could the output display some convergence criteria (such as prior and posteriors) over the estimated coefficients’ distributions? Can that be done?

I would appreciate any comments answers.

[EViews Gareth]

I hope, within a month or two (which I know is a long time!) that the BVAR Add-in will get some more attention. My first attempt was a sort of proof of concept - I'm not an R expert, nor a BVAR expert, and I was sort of hoping someone that is could take up the mantle! Given that hasn't happened, I/we'll look into it a bit more to see if we can improve it.

With regards to the exogenous variables, as mentioned above, at the time I wrote the Add-in the R package didn't support them. That has now changed.

[Mihaain]

Hello,
Recently I`ve installed the BVAR add-in, but when I try to perform an analysis I do not find any output in the IRF tables.
I`ve read your previous post regarding this problem, but I did not manage to find a solution. Any suggestions?

Best regards

[EViews Gareth]

The Bayesian VAR Addin is currently being completely overhauled. If you can wait a few more days...

[EViews Gareth]

A new version of the Add-in has been released. Since it is so different, we've started a new thread