Optimal fiscal policy in times of uncertainty: a stochastic control approach

Abstract

This paper deals with the possibilities of designing optimal fiscal policy under uncertainty. First, different forms of uncertainty are discussed for economic policy analysis and design. For dynamic models under uncertainty, a stochastic optimum control framework is presented. Algorithms for nonlinear models are briefly reviewed: OPTCON1 for open-loop control, OPTCON2 for open-loop feedback (passive learning) control, and OPTCON3 for dual control with active learning. The OPTCON algorithms determine approximately optimal fiscal policies. The results from calculating these policies for a small macroeconometric model for Slovenia serve to illustrate the applicability of the OPTCON algorithms and compare their solutions. The results show that the most sophisticated and time intensive active-learning solution, which requires the use of an extremely small and simple model of the economy, is not necessarily superior to the simpler solutions. For actual policy design problems and policy advice, it will often be better to neglect the stochastic uncertainty and use deterministic optimization instead, especially since in practice, the most important forms of uncertainty are not stochastic but relate to the model specification, the behaviour of other policy makers or other agents, or fundamental uncertainty that cannot be dealt with at all.