Page 1 of 1
SVAR residuals
Posted: Fri Mar 23, 2012 1:06 pm
by bab
Hello,
I've estimated a SVAR and need to get the orthogonalized residuals (because they have the interpretation of structural shocks). Especifically, I need the series of structural shocks. Does anyone know a simple way to get that?
Thank you in advance!
Re: SVAR residuals
Posted: Wed Apr 04, 2012 8:47 am
by bab
Ok, maybe not that simple! Could somebody help me?
Specially with Cholesky orthogonalization
Re: SVAR residuals
Posted: Fri Apr 13, 2012 3:43 am
by trubador
HDecomp add-in lets you save the structural residuals with respect to your choice of orthogonalization.
Re: SVAR residuals
Posted: Thu May 10, 2012 12:37 am
by mth
Hi,
I am having a similar problem. I would like to obtain a historical decomposition for a SVAR with two endogenous variables and a long-run restriction. I successfully managed to run a structural decomposition with the HDecomp add-in. But does anyone know how to interpret the output? (for the two-variable VAR I get six columns).
Many thanks in advance!
Re: SVAR residuals
Posted: Fri May 11, 2012 5:33 am
by trubador
The document hdecomp.pdf in the add-in folder includes a case study that makes of this method, which might be helpful in this regard. Moreover, you can refer to W. Enders' "Applied Econometric Time Series" textbook for a detailed explanation of the approach along with several examples.
Re: SVAR residuals
Posted: Mon May 21, 2012 9:05 am
by mth
Many thanks! That helps a lot. I hadn't noticed the add-ins folder before.
Re: SVAR residuals
Posted: Sun Apr 21, 2013 1:31 pm
by mmm
Hello!
I also estimated a SVAR with two endogenous variables and long run restrictions and used the hdecomp add-in in order to obtain the structural residuals. I obtained that matrix with six columns and i am not very sure what are the series of my structural residuals (I should get two in my case...). I have also checked the hdecomp.pdf example but I am still a little confused.
If someone could help me figure it out, it would be much appreciated.
Thank you very much!