Re: [Gretl-users] Estimating Toda Yamamoto Causality Model
On Tue, 23 Oct 2018, Yusuf Abduwahab Hassan wrote:
Good morning day all,
Please how do i estimate the Augmented VAR (p+dmax) model proposed by Toda
Yamamoto 1995?
If i estimate a VAR model in levels using the Vector Autoregression menu,
can i add extra lags and interpret my results as using the Toda Yamamoto
approach to causality testing?
In fact, it's only a little bit more complicated than that: you have to
add one lag and then carry out the test on the "true" parameter order. For
example:
<hansl>
#
# example: classic 1980s-style monetarist cr*p
# (money Granger-causes income)
#
open denmark --quiet
# select optimal lag via HQC
var 8 LRM LRY --lagselect
scalar p = iminc($test[,3])
# "proper" regressors
list MLAGS = LRM(-1 to -p)
list YLAGS = LRY(-1 to -p)
# "extra" Toda-Yamamoto (aka Dolado-Lütkepohl) stuff
list EXTRA = LRM(-(p+1)) LRY(-(p+1))
ols LRY const MLAGS YLAGS EXTRA --no-df-corr
# GC test
omit MLAGS --chi-square --quiet
</hansl>
-------------------------------------------------------
Riccardo (Jack) Lucchetti
Dipartimento di Scienze Economiche e Sociali (DiSES)
Università Politecnica delle Marche
(formerly known as Università di Ancona)
r.lucchetti(a)univpm.it
http://www2.econ.univpm.it/servizi/hpp/lucchetti
-------------------------------------------------------