Page 1 of 1

unrestricted VAR: possible to impose another constant value when forecasting?

Posted: Mon Nov 18, 2019 1:25 pm
by maragloria
Hi there,

I would like to know if it would be possible to change the constant in an unrestricted VAR to a different value prior to performing (dynamic) forecasts?
No other coefficient estimates need to change besides the constant.

Many thanks,

Mara

Re: unrestricted VAR: possible to impose another constant value when forecasting?

Posted: Mon Nov 18, 2019 3:10 pm
by EViews Matt
Hello,

The only way that occurs to me is to 1) create a model object from the estimated VAR, 2) break the link between the model and the VAR, 3) edit the individual equations within the model to replace the constant terms as you see fit, and finally 4) perform the "forecast" as a normal model solve.

Re: unrestricted VAR: possible to impose another constant value when forecasting?

Posted: Tue Nov 19, 2019 8:46 am
by maragloria
Hello Matt,

Thanks for your prompt reply. Is there a model command I could use to overwrite the constant value? I'll be performing rolling regressions, and I'll like to program the 4 steps you suggest.

Thanks again,

Mara

Re: unrestricted VAR: possible to impose another constant value when forecasting?

Posted: Tue Nov 19, 2019 10:05 am
by EViews Matt
Unfortunately, there is not. At that point, the model will contain text specifications for each equation, so you'll be forced to parse each equation as a string and replace the constant term. At least the constant term should always be the very last term, which will make the parsing easier.

Re: unrestricted VAR: possible to impose another constant value when forecasting?

Posted: Tue Nov 19, 2019 11:51 am
by maragloria
Thank you! That will do it.