Incentive contracts when agents distort probabilities

I show that stochastic contracts generate powerful incentives when agents suffer from probability distortion. When implementing these contracts, the principal can target probability distortions in order to inflate the agent's perceived benefits of exerting high levels of effort. This novel source of motivation is absent in contracts traditionally regarded as optimal. A theoretical framework and an experiment demonstrate that stochastic contracts implemented with small probabilities, which expose the agent to a high degree of risk, generate higher performance than cost-equivalent contracts with lower or no risk exposure. I find that probability distortions that result from likelihood insensitivity—cognitive limitations that prevent the accurate evaluation of probabilities—account for this finding. The results highlight the limits of contracts traditionally regarded as optimal.