Lanne, Markku (contributor); Saikkonen, Pentti (contributor) - 2008
knowledge, all economic applications so far restrict themselves to causal autoregres-
sive models where the current value of the … horizons are considered. Thus,
for optimal prediction knowledge of the distribution of the error process"t is required
and …, even if this knowledge is available, prediction is not easy because, in general, the
prediction problem is nonlinear …