No Refund, No Problem: Refund Institutions in a Combined Linear-Threshold Public Good Game

"We use a lab experiment to study a combined linear-threshold public good game, with linear returns to contributions and an additional higher return if a threshold is met. We experimentally vary handling of contributions that do not meet the threshold. Contributions are high, with almost all groups reaching the threshold. Refund institutions matter, but only a little: contributions are only weakly higher on average if insufficient contributions are automatically refunded than if they cannot be refunded. However, optional refunds chosen ex post perform relatively poorly. Risk averse people contribute less, but when a refund is possible, more risk tolerant people reduce their contributions relative to risk averse people. An online survey-experiment shows that people perceive refunds in this institution as desirable from a donor’s perspective, though a donor’s taking advantage of a threshold is perceived as non-normative, especially in a naturalistic charity context."