Impose coefficient restrictions in uncontrained VAR

[jfgeli]

Hi there,

I want to impose a linear restriction (lets say c(1) = 0) in one of the coefficients of an unrestricted VAR which I am estimating in a model. Is there a way to append (is that the command) the restriction inside the model and then estimate the coefficient? If so, could you give me a short example or let me know where in the documentation I can find an example?

Regards,

Fede

[EViews Gareth]

The only way to do it is to convert your VAR into a system (proc->make system) and then impose the restrictions in the system. Unfortunately you'll lose all of the post-estimation views and procedures available to a VAR (such as impulse responses).

[jfgeli]

I'll have to change the whole estimation strategy then...

Many thanks!

[oth76]

Hi everybody,
Is it possible to constraint the system using a command line in a relative automated manner ?
After an estimation of unconstrained VAR, let's say var1, many coefficients appear to be non-significant. I want to re-estimate the system by imposing them to be null.
After, I guess this first command:

var1.makesystem(n=sys1)

what command should I use ?
Any simple example please ?
Thank you for your help.

[EViews Gareth]

No, there is no easy way to do that.