Logit estimation of conditional cooperationin a repeated public goods experiment

A conditional cooperator in a public goods game wants to match hispartners' expected contribution. We investigate theoretically and empiricallywhether (and to what extent) conditional cooperation can explainhow individual contributions evolve in a repeated two-person public goodsexperiment using a perfect strangers design. To identify a random utilitymodel including non-pecuniary preferences we elicit participants' beliefs.Our econometric results show that the distribution of preferences in thepopulation can be captured by a latent-class mixed logit specificationwith three subpopulations, and that 55% of participants can be regardedas conditional cooperators. Thus, the decline in average contribution levelsmay be attributed to the presence of conditional cooperators who haveto revise their expectations about the others' behavior.