EViews.com

hello,
when I estimate a VAR model whith time varying parameters
y(t)=(x(t),z(t))'=A(t) y(t-1)+u(t)
I proceed equation by equation as follow:
in a new Sspace object, I wrote:
@signal x=a11*x(-1)+a12*z(-1)+[var=exp(c(1))]
@state a11=a11(-1)+[var=exp(c(2))]
@state a12=a12(-1)+[var=exp(c(3))]

and then in an another Sspace object I wrote the second equation:
@signal z=a21*x(-1)+a22*z(-1)+[var=exp(c(4))]
@state a21=a21(-1)+[var=exp(c(5))]
@state a22=a22(-1)+[var=exp(c(6))]

Please who knows if is it the right method to treat this kind of models? or have you an other approach to do it?

Thanks a lot.