Formal and relational incentives in a multitask model

Abstract

This paper studies the optimal contracts in a multitask model when a principal–agent relationship is long-term. If some outcomes are unverifiable, then the contracts have to satisfy the self-enforcing condition. I characterize the optimal contract in terms of the discount rate, the cost substitutes, and the weight of the unverifiable outcomes relative to the principal's payoff. Then, as the discount rate increases, the incentive to verifiable outcome (formal incentive) changes discontinuously and non-monotonically while the incentive to the unverifiable outcome (relational incentive) changes discontinuously but monotonically.