Self-Nudging vs. Social Nudging in Social Dilemmas: An Experiment

The exogenous manipulation of choice architectures to achieve social ends ('social nudges') can raise problems of effectiveness and ethicality because it favors group outcomes over individual outcomes. One answer is to give individuals control over their nudge (`self-nudge'), but the trade-offs involved are poorly understood. We examine how subjects self-nudge in a paradigmatic social dilemma setting and whether outcomes differ between the self-nudge and two exogenous nudges in line with perfect free-riding or full cooperation. Subjects recruited from the general population play a ten-round VCM online in fixed groups of four with one daily contribution decision. The nudge takes the shape of a non-participation default contribution, comparing zero, full, and self-determined levels. We find that the average self-nudge is 44% of the endowment and only 7% of subjects choose one of the two exogenous defaults. Yet, there is a hard trade-off between ethicality and effectiveness: Self-nudging groups do not better than groups under the perfect free-riding nudge. The reason is that non-defaulting subjects contribute less. Groups under the full cooperation default exhibit no reactance against the nudge and outperform both alternative choice architectures.