Rational ignorance is not bliss: When do lazy voters learn from decentralised policy experiments?

A popular argument about economic policy under uncertainty states that decentralisation offers the possibility to learn from local or regional policy experiments. We argue that such learning processes are not trivial and do not occur frictionlessly: Voters have an inherent tendency to retain a given stock of policy-related knowledge which was costly to accumulate, so that yardstick competition is improbable to function well particularly for complex issues if representatives’ actions are tightly controlled by the electorate. Decentralisation provides improved learning processes compared to unitary systems, but the results we can expect are far from the ideal mechanisms of producing and utilising knowledge often described in the literature.