Page 1 of 2
Posted: Fri Jan 31, 2014 2:03 am
This thread is about svarpatterns
add-in for just-identified SVAR models.
The add-in allows you to impose both short-run and long-run restrictions to obtain non-recursive orthogonalization of the error terms (as opposed to recursive Cholesky decomposition) for impulse response analysis that would make more sense from a macroeconomic/structural point of view. In order to use the add-in, you should first estimate a regular VAR model. After that, you can either supply the name of your model or the covariance matrix. The output will be a factor matrix, which can further be used in generating impulse responses, but NOT in conducting variance decomposition
(see the picture below). In short, this add-in aims to extend the current functionality of EViews' Structural VAR estimation toolbox.
Posted: Sat Feb 01, 2014 3:48 pm
Can this add-in be called inside a programme? I've searched the forum and googled in vain to find an answer to the more general question of whether add-ins can be called inside programmes rather like sub-routines.
Posted: Sat Feb 01, 2014 4:01 pm
Of course they can. Just use whatever command the add-in has assigned to it.
Posted: Sat Feb 01, 2014 4:05 pm
Thanks. I suppose I just missed this point somehow.
Posted: Mon Feb 17, 2014 6:12 pm
Does anyone know how to do variance decomposition after using SVARpatterns?
EViews only allow variance decomposition via Cholesky or structural decomposition (short OR long-run), but not when SVARpatterns is used.
Posted: Sun Apr 27, 2014 9:02 am
How do i specify the long run and short run matices in the svarpatterns? e.g var1.svarpatterns(options).
Also, how do I do the impulse response functions after this?
Posted: Sun Apr 27, 2014 2:39 pm
The options are clearly shown in the documentation. What is lacking now is how to do impulse response functions and variance decomposition after the svarpatterns. Any one has ideas?
Posted: Sun Apr 27, 2014 2:44 pm
variance decomposition in SVARpatterns
Posted: Wed Jan 14, 2015 9:08 pm
Is it possible to have the variance decomposition with the user specified factor matrix (the one obtained from LR and SR restrictions)??
Posted: Sat Aug 15, 2015 2:31 pm
The SVAR add-in that help to estimate both short and long run restrictions, is it possible to have the variance decomposition?
And the output graphs, the dashed lines indicate a 95% confidence interval, or the one standard error bounds?
Posted: Mon Nov 23, 2015 7:23 pm
When I use a matrix object (a variance-covariance matrix) to call the SVARpattersn addin, I got the following error message:
"Please supply the matrix for VAR coefficents."
Do you know how to fix it?
Posted: Tue Nov 24, 2015 2:11 am
wilshire wrote:Hi all,When I use a matrix object (a variance-covariance matrix) to call the SVARpattersn addin, I got the following error message:"Please supply the matrix for VAR coefficents."Do you know how to fix it?ThanksWilshire
Please do not post the same question twice.
You need to explicitly provide the estimated VAR coefficients as a matrix as well.
Posted: Tue Nov 24, 2015 6:25 am
I have another question: Almost every time when I run the addin, I got the error "Estimation not converge". But when I supplied a starting values matrix, I got another error which say "THETAVEC is not defined."
Can you please help?
Posted: Tue Nov 24, 2015 2:36 pm
Estimation from covariance matrix is a little bit complicated. I cannot locate the source of the problem without seeing the actual workfile. In the meantime, you can use the add-in through the VAR object, if possible.
Posted: Wed Nov 25, 2015 6:45 am
Actually I used the VAR object to perform SVARPatterns. Please see attached the workfile. The VAR object is VAR_COMM, and pattern matrices are pattern_sr for short-run, and pattern_lr for long-run restriction. I have tried many ways to use the SVARPetterns but each time gives the error message "estimation not converges". I donot know why.