We consider the design of an optimal voting system when voting is costly. For a private values model with two alternatives we show the optimality of a voting system that combines three elements: (i) there is an arbitrarily chosen default decision and non-participation is interpreted as a vote in favor of the default; (ii) voting is sequential; (iii) not all voters are invited to participate in the vote. We show the optimality of such a voting system by first arguing that it is first best, that is, it maximizes welfare when incentive compatibility constraints are ignored, and then showing that individual incentives and social welfare are sufficiently aligned to make the first best system incentive compatible. The analysis in this paper involves some methods that are new to the theory of mechanism design, and it is also a purpose of this paper to explore these new methods.
