An analysis of the impact of finite horizons on macroeconomic control

Macroeconomic control exercises witch use large-scale non-linear consistent expectations models necessarily employ a finite time horizon since databases must be constructed to emulate such models. Expectations past the terminal date must be proxied by terminal conditions, which can distort the solution. In contrast, much of the analytical literature has focused on the linear-quadratic infinite time horizon problem, which for the closed form solution is easy to derive and offer convenient interpretation. This paper examines the extent to which the necessary imposition of a finite time horizon distorts the optimal policy not only through the terminal condition but through the impact the finite horizon can have on expectations. This is carried out by comparing the results obtained from a large non-linear consistent expectations model (the National Institute UK model) with a linearised representation of the same model.