Gift Exchange in the Workplace: Addressing the Conflicting Evidence with a Careful Test

Tests of gift exchange, wherein agents receive excess wages which are noncontingent on performance in one-shot settings, have yielded contradictory evidence: they sometimes find effort boosts, consistent with gift exchange, whereas they sometimes find no effort increases, consistent with a standard model. We identify eight confounds that could have led to the mixed evidence—agent disutility from being viewed as selfish, small samples, insufficient wage raises, an effort ceiling, fatigue, selection of abler workers, reemployment concerns, and peer effects—and run a comprehensive test addressing them. Our test consisted of a field experiment hiring workers for a data entry job, followed by laboratory games assessing their prosocial behavior. After addressing these confounds, we find that behavior during the field test was consistent with a standard model: workers did not repay fixed wage raises with an effort boost, but they did raise effort in response to a piece rate. The piece-rate scheme was also more efficient: the effort boost came at lower expense than paying fixed wage raises. Further, workers who behaved prosocially in laboratory games did not behave prosocially in the field. Data and the online appendix are available at https://doi.org/10.1287/mnsc.2017.2801. . This paper was accepted by John List, behavioral economics.