Both dynamic and static forecast?

[luching]

In a VAR system, is there any inbuilt command or a quick way to do a dynamic forecast for some equations and static forecast for the rest of the equations?

[TomDoan]

If you want to do a dynamic forecast for some variables, while using the actual values for the others, you just need to define a model which includes only the equations (from the VAR) defining the variables that you want to treat dynamically. In other words, you "exogenize" some subset of the variables.

[luching]

Here's what I have done:

group smodel %modeleqn(varmodel,1) %modeleqn(varmodel,2)
forecast(model=smodel)

I think the issue is that when I try to forecast the smaller model, it cannot read the 3rd variable which is the one I want to use as actuals.

It's something else. It can't even run this:

group smodel %modeleqn(varmodel,1) %modeleqn(varmodel,2) %modeleqn(varmodel,3)
forecast(model=smodel)

[TomDoan]

Here's what I have done:

group smodel %modeleqn(varmodel,1) %modeleqn(varmodel,2)
forecast(model=smodel)

I think the issue is that when I try to forecast the smaller model, it cannot read the 3rd variable which is the one I want to use as actuals.

It's something else. It can't even run this:

group smodel %modeleqn(varmodel,1) %modeleqn(varmodel,2) %modeleqn(varmodel,3)
forecast(model=smodel)

Use the following as a workaround:

Code: Select all

modify %modeleqn(varmodel,1) eqn1
modify %modeleqn(varmodel,2) eqn2
group smodel eqn1 eqn2
forecast(model=smodel)

[luching]

Thanks. This is a lot more elegant than the one I wrote using matrices.