Variance decomposition under SVAR

For technical questions regarding estimation of single equations, systems, VARs, Factor analysis and State Space Models in EViews. General econometric questions and advice should go in the Econometric Discussions forum.

Moderators: EViews Gareth, EViews Moderator

Posts: 13
Joined: Tue Aug 15, 2017 7:56 am

Variance decomposition under SVAR

Postby bgauchan » Sun Oct 01, 2017 6:16 am


I am using real GDP of US and India to represent Global and Regional outputs respectively along with GDP of other six SAARC countries. The long run restrictions on SVAR is imposed following Blanchard & Quah (1989) procedure. Accordingly, it is assumed that in the long run the shock in the output of each of the SAARC countries will not affect both regional and global outputs and similarly, the shock in regional output (as represented by India) will not affect global output but will affect the output of other member countries. Now adopting Chow & Kim (2003) procedure, we shall decompose variability of domestic output into domestic, regional and global output shocks in the context of estimating optimum currency area (OCA) in South Asia. The idea is that if the countries in the region face higher degree of regional shock compared to country specific shock, they are considered to be better choice for common currency. Kindly advise how to perform variance decomposition under SVAR using the above case as an example given that the variance decomposition command in the eviews doesn't differentiate results between SVAR and simple VAR navigation...

I am using Eviews 9.0. I have enclosed my eviews file and a paper using aforementioned model for your ready reference.
(1.29 MiB) Downloaded 52 times
(33.21 KiB) Downloaded 53 times

EViews Matt
EViews Developer
Posts: 327
Joined: Thu Apr 25, 2013 7:48 pm

Re: Variance decomposition under SVAR

Postby EViews Matt » Mon Oct 02, 2017 1:15 pm


The variance decomposition can be performed using the structural factorization (assuming an SVAR has been successfully estimated) if that's what you want. Take a look at the documentation for the var::decomp procedure and option "imp=struct".

Return to “Estimation”

Who is online

Users browsing this forum: No registered users and 12 guests