Endogenous Group Formation and Public Goods Provision: Exclusion, Exit, Mergers, and Redemption

Abstract

We test an assortative mechanism whereby groups are formed endogenously, through the use of voting. Once formed, groups play a public-goods game, where the social value of an incremental contribution to the group account increases with the size of the group. Societies of nine people are initially formed randomly into three groups of three people, with identification numbers, and play the game for three periods. Individuals then learn about the average contribution of each individual in one's current own group, as well as the average contribution in other groups, and can decide whether to exit the group. Remaining group members choose whether to exclude any current members from the group; the new groups then choose whether to merge with other existing groups. The exclusion/merge process takes place after every three periods; with 15 periods in one segment and another 15 periods in a second segment (with a re-start) with new identification numbers.

We find a great degree of success for this mechanism. The average contribution rate to the group account (excluding the final three periods of a 15-period segment) is 90.2%. We see large and stable groups forming, promoting efficiency: Overall, after the first three periods of a segment, there is at least a 5-person group formed 96% of the time, with at least a 7-person group 66% of the time, and the nine-person grand coalition forming 27% of the time. After the re-start, these figures are even higher: at least a 6-person group formed 100% of the time, at least a 7-person group formed 79% of the time, with the nine-person grand coalition forming 43% of the time. Results from our control sessions indicate that this success is due to the voting mechanism.

Keywords: Public goods, Endogenous group formation, Experiment, Exclusion, Ostracism, Merger, Efficiency, Voting

JEL Classification: A13, B49, C79, C91, C92, D71, H41