Macroeconomic control exercises using large-scale non-linear consisten
t expectations models necessarily employ a finite time horizon. By con
trast, much of the analytic literature has focused on the linear-quadr
atic infinite horizon problem, for which closed form solutions are eas
y to derive and offer convenient interpretation. This paper examines t
he extent to which the imposed finite time horizon distorts the optima
l policy. We compare the effects on the reputational time inconsistent
solution with those on time consistent strategies using both a large
non-linear consistent expectations model and a compact linearised vers
ion.