The Value of Delegation in a Dynamic Agency

In this paper we analyze the value of delegation in a two-period agency. A central management hires an agent to perform a personal effort in each period. Due to time constraints or lack of ability this effort can not be performed by central management. Besides personal effort firm value is influenced by the decision to launch a project which has to be made at the beginning of period two. The project decision can either be delegated to the agent (decentralization) or it can be made by central management (centralization). Under decentralization the agent observes the project’s contribution before its decision. While this captures the benefit of delegation its cost is that the project decision is unobservable and must be motivated together with personal effort via the same incentive contract. In the centralized regime, in contrast, no incentives for the project decision are necessary, however, the project’s actual contribution will not be observed such that the project decision has to be made based on expectations. We analyze optimal performance measurement for both regimes in a linear contracting setting and analyze the variables that affect the value of delegation. We do this for two different contracting regimes: long-term commitment and long-term renegotiation-proof contracts. Trade-offs under both contracting environments differ substantially. In particular, under renegotiation-proof contracts, decentralization might become optimal even if its direct benefit in terms of acquiring specific knowledge about the project vanishes. The reason is that with delegation of the project decision central management implicitly commits to a higher second period incentive rate as personal effort and the project decision must be controlled via the same incentive contract. This is beneficial if renegotiation-proofness requires central management to set too low second-period incentives compared to long-term commitment. A necessary condition for that is, that intertemporal correlation is negative. Contrary to the classical view this result implies that the incentive problem under centralization may become more severe than under decentralization.