DATA for: Punishment piggybacks on reputation to sustain efficient cooperation

Data and analysis files for "Punishment piggybacks on reputation to sustain efficient cooperation", by Antonio Arechar, Antonio Espin, Jeroen Niebor, Chris Starmer, and Simon Gächter.

ABSTRACT: Cooperation for the common good is required in communities of familiar individuals, such as workplaces, and in fleeting, anonymous encounters, such as those online. Yet how informal social sanctions can restrain socially harmful selfishness (‘first-order free-riding’) across this spectrum of interactions remains unclear, because costly punishment itself poses a ‘second-order free-riding’ problem. Here we report an experimental study (n = 1,305) that systematically varies anonymity, action observability, interaction length and punishment severity in repeated public goods games with and without costly punishment. We find that stable and efficient cooperation emerges only when individuals are identifiable, actions are observable, interactions are repeated, and punishment imposes greater costs on the punished than on the punisher. Under these conditions, punishment is used rarely (<4% of opportunities) yet remains highly effective. Our results show that punishment operates primarily as a credible threat that piggybacks on reputational incentives, sustaining cooperation while largely avoiding its own costs. These findings identify a behavioural mechanism through which reputation and punishment jointly support cooperation and help explain why anonymous environments entail substantial social costs in terms of foregone cooperation and excessive punishment.