Kalman Filter

[istatistik]

i wan't to estimate like a model ARX(1) with time-varying parameter but i can't make state space definition in eviews.

model y=a+b*y(-1)+c*x(-1)+u ; y:dependent variable, x:random variable, a,b and c time-varying parameter with
a=c(1)*a(-1)+u1
b=c(2)*b(-1)+u2
c=c(3)*c(-1)+u3

how is the signal and state equation specification in auto-specification menu?

Thank you.

[istatistik]

is it

@signal y=sv1+sv2*y(-1)+sv3*x+[var=1]

@state sv1=c(1)*sv1(-1)+[var=1]
@state sv2=c(2)*sv2(-1)+[var=1]
@state sv3=c(3)*sv3(-1)+[var=1]

i can try this but it is the same with OLS estimation.

[EViews Glenn]

I'm not quite certain what the last question means, but for AR(1) time-varying coefficients:

@signal y = sv1 + sv2*y(-1) + sv3*x(-1) + [var = exp(c(1))]

@state sv1 = c(3) + c(4)*sv1(-1) + [var = exp(c(2))]
@state sv2 = c(6) + c(7)*sv2(-1) + [var = exp(c(5))]
@state sv3 = c(9) + c(10)*sv3(-1) + [var = exp(c(8))]


Note that I got this using the auto-spec proc. This may not be the model you want, as it is difficult for me to determine from your comments the exact specification that you are trying to create.