A fundamental problem that has been the focus of much work in agency theory has been the design of contracts to provide insurance to risk-averse agents and to elicit appropriate levels of effort.' In the context of insurance, the problem is viewed as one of moral hazard whereby the insured agent will reduce the level of his precautions to prevent an accident if he is insured against the adverse outcome. In the labor market context, the problem is one of providing effective work incentives while at the same time promoting the risk-sharing element of contracts. The overall structure of the analyses is quite similar whether or not the focus is on the insurance market, the labor market, or principal-agent problems in general. For concreteness, this paper addresses the labor market incentives problem. The labor market problem is complicated by the firm's inability to observe the worker's ability and effort and also by stochastic elements that impede a firm's attempts to make more indirect inferences using output to assess the worker's productivity-related efforts. This combination of uncertainty and the need to create incentives takes on added importance in the case of workers who are risk-averse. The presence of risk aversion often mitigates the emphasis the firm can place on incentive creation, as there is a desire on the part of workers to have stable income streams. Complete equalization of one's income level across states to promote insurance will, however, remove the differential rewards needed to provide an incentive for individuals to expend effort. This inevitable tradeoff between the work incentive and insurance function of contracts has been studied in detail for single-period contracts. The focus of my analysis here will be on how these influences affect the multi-peiiod wage structure. Although there has been research on the multi-period incentives problem2 and on the role of moral hazard in multiperiod contexts,3 there is no literature on the optimal design of a merit rating system over