Key points are not available for this paper at this time.
Holmstrom (1982) has shown that a non-budget-balancing contract induces a team of risk-neutral agents to choose the first-best effort levels. This is not generally true when agents are risk averse. Furthermore, a "massacre" contract, which punishes all but one agent when the outcome is low, can attain the first best over a wider range of parameters than any other budget-balancing contract.
Eric Rasmusen (1987) studied this question.