On the incentive effects of uncertainty in monitoring agents: a theoretical and experimental analysis

When two or more agents compete for a bonus and the agents' productivity in each of several possible occurrences depends stochastically on (constant) effort, the number of times that are checked to assign the bonus affects the level of un-certainty in the selection process. Uncertainty, in turn, is expected to increase the efforts made by competing agents (Cowen and Glazer (1996), Dubey and Hai-manko (2003), Dubey and Wu ( 2001)). Theoretical predictions were derived and experimental evidence collected for the case of two competing agents, with the bonus awarded to that agent who outperforms the other. Levels of uncertainty (sampling occasions of productions, 1 or 3), cost of production (high or low), cost symmetry (asymmetric or symmetric), and piece-rate reward were manipulated factorially to test the robustness of the effects of uncertainty. For control, a sin-gle-agent case was also theoretically analyzed and empirically tested. The re-sults indicate that, for tournaments, greater uncertainty does indeed lead to greater than expected effort and lower unit variable costs.