EViews.com

Hi everyone!
I am trying to relate the potential output with a simple Phillips curve and the okun law.
My model is
@signal dlgdp = sv1 + sv2
@signal dlcpi = c(1)*dlgdp (-1) + c(2)*sv5 +c(3)*lltcr+ sv3
@signal ump= c(9)*sv5 + [var = exp(c(10))]

@state sv1 = sv1(-1) + [var = exp(c(4))]
@state sv2 = c(6)*sv2(-1) + [var = exp(c(5))]
@state sv3 = c(8)*sv3(-1) + [var = exp(c(7))]
@state sv4 = sv2(-1)
@state sv5= sv4(-1)

Where dlgdp is the gdp growth, dlcpi is the inflation, ump is the natural unemployment.
My question is:
in the signal equation of dlcpi and ump I have to use the lag of the sv2, I was wondering if I correctly specified the signal equation with the sv5..
In others words if I want to use a lag of a state equation, It is this the right approach?
Furthermore, I use the lag of the dlgdp in the second signal equation. It is correctly specified?

Thank you very much!

Your model seems allright in terms of EViews syntax. However, I think the decomposition of gdp growth (first signal equation) is ill-defined and might lead to a technical problem in estimation. You can find useful discussions on similar topics if you search the forum.

Thank you Trubador!!!! It help a lot to read others disscusion!!
Thanks again!

I have another question.. The terms [var = exp(c(10))]
Imply the error term of the equation, so the value of c(10) must be not significant, this will indicated that the error term it has zero mean. If I have all the other terms significant but c(10) also is significant, something is missing in my equation right?
Thank you!

I have another question.. The terms [var = exp(c(10))]
Imply the error term of the equation, so the value of c(10) must be not significant, this will indicated that the error term it has zero mean. If I have all the other terms significant but c(10) also is significant, something is missing in my equation right?
Thank you!

No, c(10) tells you about the variance not the mean. And in this formulation the test of significance is about whether the variance equals 1.0, which is unlikely to be interesting.

Thank you!!!