Laboratories of Democracy: Policy Experimentation under Decentralization

Abstract

We develop a model of policy experimentation in which the policy decisions of local and national politicians are driven by career concerns. When politicians’ payoffs are convex in their reputation and policy-making is opaque, local politicians experiment by introducing more diverse policies than national politicians do. The welfare implications of decentralization depend on homogeneity between districts. In particular, learning about policies is greater under decentralization if and only if the districts are sufficiently homogeneous. The model delivers novel testable hypotheses. Decentralization should be positively correlated with more diverse (uniform) local policies among countries with opaque (transparent) policymaking. (JEL D72, D78, D83)