The Incentive Costs of Welfare Judgments

We ask which welfare judgments can be institutionalized without incentive costs when agents’ material and distributive preferences are private information. A policy rule is definitely implementable if it is implementable under arbitrary variation in the distribution of types: the normative end remains fixed, while the institutional means - the transfer scheme - may adjust to the distribution. We interpret an ex post budget imbalance, if required for definite implementation, as an incentive cost: it entails external subsidies or resource destruction. We find that a policy rule is costless only if it locally admits a welfare representation as aggregate material surplus plus a relational component. Every such relational component must obey a common normative grammar. A substantive subclass satisfying this grammar is globally attainable. Costlessness thus disciplines welfare evaluation without eliminating normative choice. We illustrate this normative freedom through three relational welfare judgments in public-good provision: political restraint, subsidiarity, and minority protection.