Deliberation as Coordination through Cheap Talk∗

In contrast with other formal theory treatments of democratic deliberation in terms of (1) strategic information pooling or (2) relaxing the assumption of logical omniscience, this paper considers the role of deliberation in coordinating citizens on a higherwelfare equilibrium. The equilibrium in question belongs to a post-deliberation game of fundamental action vis-à-vis other individuals: giving, taking, harming, or producing private or collective goods. This action game has multiple equilibria, some of which achieve preferred outcomes for most or all individuals through coordinated actions. The present paper portrays this problem as a multi-player version of a battle-of-thesexes (BoS) game, and deliberation is represented as a sequence of pre-play proposals for various coordinated action patterns. Failure to reach sufficient agreement on these leads to play of a less favorable action profile, stylized as a mixed-strategy equilibrium. Thus the model extends Farrell’s (1987) model to the setting of more than two players, examining issues of how actions respond to proposals and how this connection bears on the aggregated effect of proposals–issues that do not arise in the two-player case but are fundamental to deliberation (as well as to endogenous-institution voting models). The BoS game for multiple players is first specified, and relevant equilibria derived. As a first cut at pre-play deliberation, we then add one pre-play stage of simultaneous proposals for action. Depending on whether the participants are swayed by agreement among a plurality of players, or instead require that a consensus is expressed, deliberation yields different results in terms of strategic proposals, likelihood of coordination, and expected payoff. The same questions are then addressed in a deliberation process consisting of multiple periods of such simultaneous proposals. The results match Farrell’s two-player conclusions in terms of expected payoff, but differ from his conclusions concerning deliberative strategies. Moreover, the length of deliberation necessary to approach the achievable payoffs depends strongly on the deliberation criterion (onetime consensus, or one-time plurality). ∗Prepared for presentation at Midwest Political Science Association annual meeting, April 20-23 2006, Chicago. A preliminary version was presented at the “Conference on Deliberation and Collective Choice,” Kellogg School, Northwestern University, May 20-21, 2005; thanks to David Austen-Smith and Timothy Feddersen for providing this opportunity. Thanks as well to seminar participants at the Hoover Institution, Stanford University.