User’s Guide : Multiple Equation Analysis : Vector Autoregression (VAR) Models : Data Members of a SVAR
  
Data Members of a SVAR
 
Version Compatibility Notes
Structural factorizations of a VAR estimate several coefficients that differ from those found in a standard VAR.
To access the estimated A and B matrices from a structural factorization, you may use the @svaramat and @svarbmat data members:
matrix amat = var01.@svaramat
matrix bmat = var01.@svarbmat
The F and S matrices for long-run and short-run impulse responses may be obtained using the @svarfmat and @svarsmat data members:
matrix fmat = var01.@svarfmat
matrix smat = var01.@svarsmat
Note that there are many other data members which are SVAR specific (see “Var Data Members” a complete list of data members that are available for a VAR object).
Version Compatibility Notes
The increased flexibility of SVAR restrictions beginning in EViews 10 is accompanied by streamlined options in the svar estimation procedure and a new text language for specifying restrictions.
The options and text syntax used by previous versions of EViews will continue to work in EViews 10 with the same limitations, e.g., disallowing simultaneous short-run and long-run restrictions.
The new restriction capabilities are only available via the new interface. Furthermore, SVAR estimation results (View/Structural Factorization) produced in EViews 10 and saved to a workfile will be accessible in previous version of EViews only in the following cases:
The SVAR model was specified using only legacy options and text restrictions (if present). You may therefore use, for example, EViews 9 syntax to maintain workfile compatibility with EViews 9.
The SVAR A-B-only or F-only model was estimated by pattern matrix restrictions specified in matrix objects.