Dynamic risk-sharing with two-sided moral hazard

A group of risk-averse agents repeatedly produce a perishable consumption good; individual outputs are observable but efforts are not. The contracting problem admits a recursive formulation, and the optimal value function is the fixed point of a contraction mapping. When the agents can be punished to the full extent in a single period, every continuation contract of an optimal contract is itself optimal; the marginal utility ratio between one agent and another is a submartingale. The results imply that it is in general important to restrict an agent whose moral hazard constraint is binding from saving through another agent within the risk-sharing group. Limited commitment and long-run implications of optimal contracting are also examined.