SVAR

[Scalper]

Hello,

I am with a doubt about SVAR matrices. Thus, in EViews the expression is Ae=Bu , for short-term.
However, i am folowing Basher et al (2011) since the expression is Au=e ( "u" are the reduced - form errors and "e" strucutural shocks).

As in eviews the matrice A is incorporated e and in Basher et al (2011) the matrix A is incoporated u, when i put my matrix A in eviews?!

In A field or B field?! Because, when i put in A matrix, the eviews shows up " WARNING:The matrix B is fixed and the structural innovations variaces are not estimate"

best regards,

[trubador]

There may be notational differences among different sources, especially regarding the residuals or errors. However, I do not think you have to change to assignment of matrices. The message you receive is probably due to your specification of B matrix, where you should assign missing values (NA) to the elements if you want them to be estimated. Otherwise, non-missing values are held fixed at the specified values.

[Scalper]

For instance,

Killian (2010) writes et=A0^-1 εt , and Basher et al (2011) writes Aut = εt,

This way, if i want to follow one of them, i have to put my matrix restrictions in field A and B as a identity matrix. Am i correct?

Thank you for your time and consideration trubator.

Best regards

[trubador]

I am not familiar with the papers you cited, but yes I think so...

[arthur]

Hi,

Please help me with a matter. I have estimated Structural VAR and have to perform dynamic forecast. When I did it and compared with simple VAR system with the same variables and lags, I discovered that the results of SVAR and VAR are the same. I became a little bit doubtful about the results of SVAR forecast.
So, I would like to ask you, if after estimating structural parameters, Eviews does the forecast via VAR?

B.R.
Arthur