Path dependence in public-good games

This paper presents experimental evidence that contributions to a public good can bepath-dependent for a limited time span. We study a repeated linear public-good gamewith punishment opportunities. Our data shows that subjects who had experienced ahigher marginal return on public-good contributions in rounds 1-10 contributed more tothe public good in rounds 11 and 12, even though they faced the same marginal return asthe control group in these later rounds. In contrast, differences in contributions were notsignificant when comparing subjects bearing the same current costs of punishment points,but having had different costs in the past.