Incentives and Compensation Schemes: An Experimental Study∗

When the performances of agents are correlated (because of a common random component) contracts that use information on both the relative and absolute performance of agents theoretically outperform rank-order tournaments and piece-rate schemes. Although the theoretical advantage of such contracts has long been noticed in the literature, the empirical papers that study this question have produced mixed results. Using a controlled laboratory experiment, we shed light on the performance of these contracts. We focus on the simplest version of such a contract: the Relative Piece-Rate (RPR) and compare its performance with that of tournaments and piece-rate schemes. We find that when the RPRs are imposed on the agents, they exert higher effort levels and give the principal (weakly) higher profits than tournament and piece-rate contracts. Second, we find that agents have little aversion to self-select into the RPR scheme when other alternatives are available. In our paper we fix the environment agents face and vary only their wage scheme. Our results support the theoretical predictions and suggest that principals can benefit from using RPR schemes in places where piece-rate and tournament contracts would otherwise be used.