Many workplace incentive systems involve opportunities for gaming. One of the most well-known Achilles heels of incentive systems is the �ratchet effect,� in which agents have a perverse incentive to reduce effort, in order to induce the principal to set easier incentive targets. This paper first reports results from two large-scale field experiments in a warehouse, with random assignment to incentive schemes. The results show that workers respond to the static incentive embodied in the level of the piece rate, but they exhibit only a modest sized ratchet effect, and then only after a considerable amount of time. To pinpoint mechanisms, the paper turns to evidence from online experiments conducted with the warehouse workers. These suggest that only a minority of workers recognize the ratchet effect in the context of the firm�s relatively complex incentive scheme. Workers with higher cognitive ability are more likely to exhibit the ratchet effect online, and making the incentive scheme simpler allows workers to recognize the ratchet incentives. Among workers who participated both online and in the warehouse experiments, we confirm that higher cognitive ability and recognition of ratchet incentives were relevant for exhibiting the ratchet effect in the field experiments. The results suggest that bounded rationality can make incentives work better that predicted by standard theory, due to making gaming opportunities �shrouded attributes.� The results imply that incentive systems may work in different ways depending on the cognitive sophistication of the workforce, and raise the possibility that there may be an optimal degree of complexity in incentive design for mitigating gaming behavior. |