Mechanism design and voting for public-good provision

We propose a new approach to the normative analysis of public-good provision. In addition to individual incentive compatibility, we impose conditions of robust implementability and coalition proofness. Under these additional conditions, participants' contributions can only depend on the level of public-good provision. For a public good that comes as a single indivisible unit, provision can only depend on the population share of people in favour of provision. Robust implementability and coalition proofness thus provide a foundation for the use of voting mechanisms. The analysis is also extended to a specifi cation with more than two public-good provision levels.