Abstract:
Manipulating choice architectures to achieve social ends ('social nudges') raises problems of ethicality. Giving individuals control over their default choice ('selfnudges') is a possible remedy, but the trade-offs with efficiency are poorly understood. We examine under four different information structures how subjects set own defaults in social dilemmas and whether outcomes differ between the self-nudge and two exogenous defaults, a social (full cooperation) and a selfish (perfect free-riding) nudge. Subjects recruited from the general population (n = 1,080) play a ten-round, ten-day voluntary contribution mechanism online, with defaults triggered by the absence of an active contribution on the day. We find that individuals' own choice of defaults structurally differs from full cooperation, empirically affirming the ethicality problem of social nudges. Allowing for self-nudges instead of social nudges reduces efficiency at the group level, however. When individual control over nudges is non-negotiable, self-nudges need to be made public to minimize the ethicality-efficiency trade-off.
