Page 1 of 1
Structural Residuals
Posted: Mon Aug 12, 2013 8:25 pm
by rzulfadin
Dear Tom,
Let the SVAR model:
B*y_t= A_1 y_{t-1}+ A_2 y_{t-2} +A_3 y_{t-3}+ A_4 y_{t-4} + V_t
How can I extract the A and B matrices? I want to obtain the V's using B*U = V to check the correlation matrix of V's. Is it uncorrelated by assumption of orthogonalized shocks?
Thank you.
Rahadian
Structural Residuals
Posted: Mon Aug 12, 2013 10:03 pm
by TomDoan
rzulfadin wrote:Dear Tom,
Let the SVAR model:
B*y_t= A_1 y_{t-1}+ A_2 y_{t-2} +A_3 y_{t-3}+ A_4 y_{t-4} + V_t
How can I extract the A and B matrices? I want to obtain the V's using B*U = V to check the correlation matrix of V's. Is it uncorrelated by assumption of orthogonalized shocks?
Thank you.
Rahadian
In this GARCH model?
Structural Residuals
Posted: Tue Aug 13, 2013 3:45 am
by rzulfadin
TomDoan wrote:
In this GARCH model?
Actually for the original VAR model without the GARCH term. I think I've found what I am looking for. So below is how I can transform the U's to the V's in the original VAR. Is that right?
TomDoan wrote:
compute factor=%decomp(%sigma)
@structresids(factor=factor) u gstart gend vresids
done immediately after the original VAR is estimated. This transforms the OLS residuals (u) to the structural residuals (vresids). As written, this is specific to the Cholesky factor structural model used in the oil-GDP paper. With a different structural model, you would get a different factor matrix for the covariance matrix %SIGMA.
Can I also use %decomp and @structresids to transform the OLS/SUR residuals to the structural shocks for a non-recursive VAR or near VARs? In that, I just need to adjust the 'factor', is that right? Thanks.
Structural Residuals
Posted: Tue Aug 13, 2013 5:07 pm
by TomDoan
There are some examples of @StructResids at
http://www.estima.com/forum/viewtopic.php?f=4&t=684.
Yes. You can apply this to the residuals from any type of SVAR, as long as you have a factor for the covariance matrix. If you use CVMODEL, you need to use the FACTOR option to get that. And it can also be applied to the results from a SUR.