Abstract
We present evidence on the operation of incentive pay from a field experiment in Pakistan, looking at piece rates and pay based on rank achieved in a tournament. Importantly, some workers are in contracts ‘tying’ them to the employer for several picking seasons; others are 'untied' in the sense of being employed for only the current season. We find that incentive pay (of either type) improves productivity by 30% on average, but that there are important differences across the types of workers: in particular, tournament incentives are less effective amongst the tied workers. We suggest that our main results have implications for tournament theory and the design of incentive pay schemes, particularly to the extent that they may discourage some workers and, thus, reduce incentives.