The consequences of a refund in threshold public good games

We revisit the question of whether a refund increases efficiency in threshold public good games. New experimental evidence is presented on the effect of endowment size. We demonstrate that a refund increases efficiency if and only if the endowment is small relative to the threshold. We also propose a novel way to analyze the effect of a refund. Specifically, we argue that a refund increases efficiency only if significantly many groups converge towards zero contributions in the absence of a refund.