Constructing strange(?) VAR model
Posted: Tue Apr 29, 2014 8:33 am
Dear members,
I have hypothesis that A affect B, B affect C, and C affect D. While A, B, C can affect in reverse way with a lag, I would like to make the VAR so that D does not give any feedback to A, B, C. So, A -> B -> C -> D, but D // C-> B->A, with a lag (so Cholesky decomposition does not do it, as it only does contemporaneous restriction).
Does anybody has an idea how to implement it in Eviews, or perhaps in Matlab or STATA? (Eviews is preferred, of course) :D
Pleeeeaase help! Thanks a lot for your answer in advance!
I have hypothesis that A affect B, B affect C, and C affect D. While A, B, C can affect in reverse way with a lag, I would like to make the VAR so that D does not give any feedback to A, B, C. So, A -> B -> C -> D, but D // C-> B->A, with a lag (so Cholesky decomposition does not do it, as it only does contemporaneous restriction).
Does anybody has an idea how to implement it in Eviews, or perhaps in Matlab or STATA? (Eviews is preferred, of course) :D
Pleeeeaase help! Thanks a lot for your answer in advance!