VAR-Model - Pure Sign-Restriction Approach

[kristin]

In the course of a term paper I have to apply the pure sign-restriction approach.
As I got just a very small introduction to EViews including standard VAR-estimation with Cholesky decompostion I have no idea how to implement the pure sign-restriction approach in EViews.
I am very grateful for any ideas.

[trubador]

I am afraid pure sign restrictions are not straightforward to apply. There are not any built-in procedures in EViews for that purpose. You can write a program in EViews, but I am not really not sure of it, since Uhlig himself preferred to use Bayesian approach for the estimation and inference. I think it would better to dig into details of original work and carefully plan the steps of the estimation of pure sign restriction approach, before trying to implement it in EViews.

[macroWB]

In the course of trying to learn more about sign restrictions I ended up coding up a fairly general use sign restriction code in Eviews (https://sites.google.com/site/davidstephan/codes) I included a couple of examples including the Uhlig Monetary Policy VAR. I doubt the code is completely error free or understandable but I commented heavily so should be not too difficult to understand what is going on.

[trubador]

Thanks for the effort and sharing. It is indeed a very useful approach and therefore it is very nice to be able to carry out the analysis in EViews as well.

[Ablostu]

Hello, I tried to run the run_uhlig.prg proposed by macroWB just above (thank you for sharing!), but got an error message warring about a near singular matrix. But several colleagues of mine tried it as well, and some of them faced the same problem, but others did not have any problem. We use the same EViews 9 version. Where could that come from?