This article investigates a repeated employment relationship between a principal
and a team of agents hired to solve a series of problems. With non-verifiable
output, rewarding agents based on team performance can relax the principal�s
credibility constraint by smoothing bonus payments over time. Team incentives
also induce free-riding, but the principal prefers them to individual incentives if
effort costs are relatively high and problems difficult to solve. We show that a
simple mixture of an individual and team bonus constitutes the optimal relational
contract under joint performance evaluation. The optimal contract may be inefficient
when team size is endogenous, as rewarding team performance forces
the principal to share surplus with agents, but may allow him to motivate a larger
group.