Macroeconomic Policy Design with Incomplete Information: A New Argument for Coordinating Economic Policies

The harshest criticism of econometrically based policy evaluation is undoubtedly the sensitivity of its conclusions to assumptions concerning the information available to policy-makers or to misspecification of the econometric models used in the evaluation. Economists appear to have accepted this criticism without examining how far it is justified and to what extent it can be overcome. These sensitivities can be broadly classified into those due to information errors (non-controllable shocks external to the economy, as modelled) and those due to modelling errors (errors caused by differences between the true structure of the economy and the specification of the econometric model). This paper considers the former case and studies the sensitivity of macroeconomic policy design to information errors in a multicountry setting. It argues that a policy regime of cooperation produces more robust policy designs than does non-cooperative policy-making, particularly when policy-makers exploit their freedom to revise their decisions in the light of past errors. The investigation involves some theoretical analysis and numerical sensitivity analysis using two different empirical models (extending work done earlier with Andries Brandsma) and complements a study by Hughes Hallett and Holtham on robustness and model errors (presented at an International Economic Association conference, organized by CEPR on Global Macroeconomics: Policy Conflict and Cooperation, February 1987).