Overcontribution and decay in public goods experiments: a test of the heterogeneous agents hypothesis

standard attempts to explain the phenomenon of decaying contribution in repeated linear public goods games are based on a ‘representative agent’ approach, with either selfish or altruist agents and an ‘error’ component. In this paper we try to test by purely experimental means the alternative hypothesis that in experimental public goods games there are at least three types of player: free riders, cooperators, and reciprocators. We try to identify the various types by means of four classification methods, and then play the public goods game with homogeneous groups. We observe that (1) the average contribution level is enhanced in this setting; (2) the decay phenomenon is replicated in groups of ‘pure’ free riders, whereas in groups of cooperative and reciprocating players the contribution is high and fairly stable throughout the game.