Strategic Behavior and Learning in Repeated Voluntary-Contribution Experiments

We present an experiment investigating why contributions decline in repeated public goods games. To distinguish between alternative explanations for this decrease we use a strategy method to elicit strategies in a simple two-stage public goods game. By repeating the game with new opponents participants have the opportunity to learn across games. We find that the behavior elicited using the strategy method is consistent with that of a direct response version of the game. Contributions in stage two are around 45% lower than contributions in stage one. While this pattern of declining contributions is robust across repetitions of the two-stage game, the participants’ strategies are not. Changes in individual strategies across successive repetitions sometimes increase and sometimes decrease stage-one contributions, on average contributions decrease by 7% per game. Thus experience with the game leads to an erratic and less pronounced deterioration in contributions, compared with the systematic and more marked deterioration generated by submitted strategies.