'Social poker' - A laboratory test of predictions from club theory

The theory of clubs addresses the gap between purely private and purely public goods, being concerned with how groups ('clubs') form to provide themselves with goods that are available to their membership, but from which others (non-members) can be excluded. Despite 35 years of formal development, there have been virtually no laboratory studies of club formation. We develop the 'social poker' laboratory paradigm toward filling this gap, and test the predictions from club theory that populations will partition into a privately and socially optimal set of clubs. The experiment included three conditions: (1) 'Single shot' with one trial of club formation; (2) 'iterated' with a sequence of four trials; and (3) 'iterated dollar-guarantee', with four trials in which participants who were not included in clubs still earned a small amount of money. In all conditions, clubs were frequently larger than was privately or collectively optimal: in the second condition, clubs were increasingly likely to include unnecessary members across trials. After clubs formed, members had the opportunity to 'overclaim' - to take more than their agreed-upon share of the club good. Although the incidence of overclaiming was low, it was more common in larger clubs, further reducing the collective earnings of participants.