VECM in state space form
Posted: Thu May 08, 2014 8:01 am
Hi all, i have a question regarding getting time-varying adjustment terms for a VECM by writing out the model in state space form. So I am trying to estimate both the long-run and the short-term relationship in the same time. This is what i put in my ss object
Variables: A & B
@signal A(-1) = c(1) + c(2)*B(-1) + [var = exp(c(3))]
@signal d(A) = c(4) + beta1*(A(-1) - c(1) - c(2)*B(-1))+ c(5)*d(B(-1)) + c(6)*d(A(-1)) + [var = exp(c(7))]
@signal d(B) = c(8) + beta2*(A(-1) - c(1) - c(2)*B(-1))+ c(9) *d(B(-1)) + c(10)*d(A(-1)) + [var = exp(c(11))]
@state beta1 = c(12)+c(13)*beta1(-1) + [var = 1]
@state beta2 = c(14)+c(15)*beta2(-1) + [var = 1]
When I try to estimate it, it gives me an error:
Missing value found in signal innovation matrix!
I am not sure if I am specifying the cointegration equation correctly (with the lags).
Does anybody have an insight how I can do my estimation properly?
Variables: A & B
@signal A(-1) = c(1) + c(2)*B(-1) + [var = exp(c(3))]
@signal d(A) = c(4) + beta1*(A(-1) - c(1) - c(2)*B(-1))+ c(5)*d(B(-1)) + c(6)*d(A(-1)) + [var = exp(c(7))]
@signal d(B) = c(8) + beta2*(A(-1) - c(1) - c(2)*B(-1))+ c(9) *d(B(-1)) + c(10)*d(A(-1)) + [var = exp(c(11))]
@state beta1 = c(12)+c(13)*beta1(-1) + [var = 1]
@state beta2 = c(14)+c(15)*beta2(-1) + [var = 1]
When I try to estimate it, it gives me an error:
Missing value found in signal innovation matrix!
I am not sure if I am specifying the cointegration equation correctly (with the lags).
Does anybody have an insight how I can do my estimation properly?