Bayesian Adaptive Decision Theory Versus Dynamic Games as Models for Economic Planning and Policy-Making under Uncertainty

One of the main problems in theoretical and applied studies of quantitative economic policy and planning is concerned with the potential achievements of stabilization policies aimed at controlling a dynamic economic system and guided by an intertemporal objective function, which exhibits trade-offs between different target variables. Optimization methods, particularly those of optimal control theory and dynamic programming including adaptive control theory, have been applied to many theoretical and empirical models in order to obtain insights into this question. During the last years, however, this research has come under increasing attack from several authors. One of the main arguments against these optimization studies is the assertion that optimizing stabilization policies cannot achieve their aims because of the high degree of uncertainty inherent in socio-economic (as opposed to physical) systems. But if the basic decision-theoretic framework of the theory of economic policy is accepted, this claim is largely lacking a theoretical foundation. In particular, it can be shown by methods of adaptive (dual) optimal control theory that a combination of cautious active policymaking and learning about the system response in general can improve the performance to be achieved, even under substantial uncertainties of several kinds (see, e. g., Kendrick, 1981).