Social Preferences and Collusion: A Laboratory Experiment

We investigate whether other-regarding preferences toward passive consumers affect seller behavior in a repeated duopoly game. Our experimental design assigns a human subject to the role of consumer in one (human) treatment, but not in the other (robot) treatment. In the human treatment the consumer's payoff is zero when the sellers successfully collude at low quantities, and positive otherwise. We find that sellers produce collusive quantities less frequently and earn lower profits in the human treatment. Our results provide evidence that social preferences can affect market outcomes.