The RATS Software Forum

Hi
I am trying to estimate a Structural VAR model with short and long run restrictions and I want one set of structural error to be correlated. Is it possible to do?

Yes, but it's usually the job of the short-run restrictions to get rid of the contemporaneous correlation. (That's how you end up identifying shocks when the long-run restrictions aren't enough).

Hi Tom
Thanks for your reply. I need structural shocks to be correlated. Cover et.al.(2005) did this in a three variable set up. Can you tell me a bit about how to do it in Rats? Many Thanks

Hi
They had a simple set up that was easy to solve. How to go when we have more number of variables?

Could you please give an actual reference?

Thanks Tom. Please find the attached references.
Identifying aggregate demand and supply shocks in a small open economy: Walter Enders and Stan Hurn (2005)

Using the Aggregate Demand-Aggregate Supply Model to Identify Structural Demand-Side and Supply-Side Shocks: Results Using a Bivariate VAR: Cover, Enders and Hueng (2003). Sorry I mentioned 2005 in my earlier post.

http://oep.oxfordjournals.org/content/59/3/411.full

See https://estima.com/forum/viewtopic.php?f=4&t=952

Hi Tom

What is the correct format of AFRML? As I am getting error "Identifier AFRML is Not Recognizable". Can I write one element as b(1)*-0.5 where b(1) is the element, I am going to estimate.

Hi Tom,

One more thing, how can we get VAR coefficients after model. I am not talking about print. I mean how we can use them in writing constraints.

abhishek wrote:Hi Tom,

One more thing, how can we get VAR coefficients after model. I am not talking about print. I mean how we can use them in writing constraints.

%VARLAGSUMS has all the information needed for doing long-run restrictions.

abhishek wrote:Hi Tom

What is the correct format of AFRML? As I am getting error "Identifier AFRML is Not Recognizable". Can I write one element as b(1)*-0.5 where b(1) is the element, I am going to estimate.

I assume because you named it something other than AFRML. The A option takes the form A=name of your FRML[RECT]. See the example in the other thread. Yes, you can use that as one element in the FRML.

Thanks Tom.

I am getting this error. Can't I do this.

I am trying a FIML on the lines of Keating (1990). I am not sure where Rats implements this with A restrictions.

abhishek wrote:Thanks Tom.

I am getting this error. Can't I do this.

Did you actually execute all those instructions? It looks like you tried to do the CVMODEL out-of-sequence. If you execute all those instructions, you can't get that error message.

Aside from that, that model can't possibly work---you have eight free parameters with five in the parameter set and three variances, which is two more than you can have for a three variable covariance matrix.

abhishek wrote:I am trying a FIML on the lines of Keating (1990). I am not sure where Rats implements this with A restrictions.

Again, I would ask you to give full references.