Anderson, Richard G. (contributor); … - 2006
Westerlund and Edgerton (2005).
More precisely, define the linear projection of
t
y on its own history as
(12) ∑Γ …==
=
−−−
k
1i
iti2t1ttt
y),y,yy(Eyˆ " ,
where the projection coefficient matrices
i
Γ ’s are pp× . Then, we have:
(13 …)
ttt
eyˆy += ,
where
t
e is the projection error that is uncorrelated with
t
yˆ (or the lagged values of
t
y …