EViews.com

I searched the forum and did not find a response to this specific question.

I have estimated a VAR and want to calculate the IRF from a 1 unit shock, not a 1 s.d. shock. I know I can do that using the "Residual - One Unit approach, but I would like to choose the Cholesky ordering. To solve this problem, I only need to know the s.d. of the innovation, and simply rescale the IRF.

Where can I find the s.d. from the innovation?

Thanks

Hello,

Are you referring to the diagonal elements of the Cholesky factorization of the VAR's residual covariance matrix? You can calculate that factorization matrix explicitly, e.g., @cholesky(v.@residcov), or retrieve it after running the impulse responses, e.g., v.@impfact.

I am not sure...


I just want to know the s.d. of the innovation, in order to be able to rescale the IRF to get the response to a 1 unit impulse, not 1 s.d. impulse. But I still want to choose "Cholesky - dof adjusted" option, not "Residual - One Unit".

I think may be I can get this innovation s.d. just looking at the s.e. of the residuals in the specific VAR equation for the variable that I want to apply the innovation shock.

Thanks

Okay, what I've indicated should be what you want. As you've pointed out, for the Cholesky decomposition this should be the same as the residual standard deviations, e.g., the square roots of the diagonal elements of the residual covariance matrix.

Try the sirf add-in

Thanks!!