Delegating management to experts.

Owners of property and assets frequently delegate decisions about operating and maintaining their property to managers who are better informed about local market conditions. We analyze how owners optimally contract with managers who vary in their expertise at prescribing service. We show that the most expert managers offer the greatest variation in operating recommendations. Owners benefit from dealing with experts provided they contract sequentially, whereby terms are negotiated gradually as the manager acquires information.

1. Introduction

Property owners frequently delegate the operation of some segments of their business to local managers. This presumably arises because managers are better informed about local demand for the owner's services and about the costs of maintaining and improving the owner's assets. For instance, absentee landlords will frequently solicit the services of a property manager to arrange for the rental and upkeep of their properties. Similarly, the owner of a restaurant chain may rely on local franchisees to manage his chain in particular areas of the country. Business owners will frequently delegate decisions on the purchase of technical goods and services, including communications, computing, legal, and financial services, to experts who are better informed about the services that are available and their costs.

In these settings, we often observe that the authority of different managers to make decisions varies widely. In some instances a store manager may be given full authority to make whatever expenditures or purchases he thinks are warranted. In contrast, some managers are afforded very little autonomy to procure services or supplies they feel are required to run the business effectively. It seems likely that the degree of autonomy that is granted to the manager depends on his ability to make good decisions that benefit the owner.

While delegation of operating decisions is useful, it does pose some problems for the owner. For instance, suppose a landlord hires a property manager to make repairs and advertise his property to prospective renters. The owner may be unable to observe the manager's expertise at predicting the cost of repairs or advertising. Further, the owner may be unable to assess the accuracy of the manager's forecasts ex post, if he does not observe the actual cost the manager incurs. Added to this is the possibility that a privately informed manager may recommend services or repairs to maximize his profits rather than the owner's surplus. One often proscribed remedy for this is to separate the diagnosis from the supply of service. This rarely occurs in practice, though, because it dampens incentives for the expert to make accurate prescriptions of service. Another reason is that prescribing and supplying service are often complementary and therefore more costly to perform independently.

This article is motivated by the following questions: How do owners contract with managers who privately know their abilities to forecast and to prescribe service? What negotiation processes induce managers to provide accurate estimates of costs and to prescribe services most beneficial for the owner? When managers have similar but uncertain costs, what is the owner's preference for contracting with experts who vary in their ability to forecast future costs of service?

To address these questions, we extend the existing incentive contracting model to settings where the supplier's private information at contracting time is his ability to predict the eventual costs of supply. Starting with the seminal analysis of Baron and Myerson (1982), most studies focus on contracting with sellers who are perfectly and privately informed at the time they contract. (1) In contrast, we model a setting where managers are privately but imperfectly informed of market conditions at the time of contracting. We assume that managers differ in their expertise at forecasting, and privately know their forecasting ability at the time of contracting.

We model a sequential contracting process, where managers choose terms gradually over time as they receive more accurate forecasts of cost. Based on these forecasts they prescribe a quality of service for the owner to adopt. Our model of sequential contracting builds on the important insight, first recognized by Courty and Li (2000), that sequential contracts updated for the arrival of new information are not only superior in theory, but also seem to be used in practice for several important applications. (2)

Our analysis begins by defining what constitutes expertise in forecasting. All managers are presumed to draw their costs from the same distribution, and they are distinguished only by the forecast of costs that they receive. Intuitively, it seems that forecasts that track or systematically vary with actual costs are more accurate. When forecasts are relatively constant and invariant to actual costs, sellers are unable to adjust their supply decisions to take advantage of better cost information. Our results confirm this intuition in the following sense. We find that sellers prefer receiving more varied forecasts that enable them to earn greater surplus for any given supply contract. This ranking provides an ordering of forecast structures satisfying Rothschild's and Stiglitz's (1970) mean-preserving spread. That is, for a given distribution of actual costs, suppliers prefer one frequency of forecasts to another if the former is a mean-preserving spread of the latter.

To the extent that owners delegate decisions to managers, one might naturally presume that owners benefit from expert advice. After all, experts can prescribe more efficient levels of services to the owner. But they command greater information rents from owners in return. What we find is that the owner's ability to benefit from expert advice depends on the contracting sequence employed. When all contract terms are negotiated after the expert has observed his forecast, the owner may benefit little or not at all by this advice. An informed manager may extract too much rent from the owner to benefit from his superior advice. However, when contract terms are determined gradually as the expert acquires information about supply conditions, we find that the owner always benefits from dealing with expert sellers.

The owner's ability to benefit from expert advice depends on the contract he employs to delegate purchases. We demonstrate below that the owner optimally offers different contracts to managers with varying expertise. Interestingly, we discover that the contracts differ by the amount of decision authority the manager is delegated. Under the expert management arrangement, the manager is free to select the efficient level of service based on his forecast of cost. In contrast, under the nonexpert agreement, the manager's ability to adjust supply to his forecast of costs is constrained by the buyer. Offering different contracts with varying degrees of delegation authority allows the owner to screen managers by their ability to forecast. Naturally, the expert prefers the contract with the greatest delegation authority. This screening of managers is precluded, of course, when managers already know their forecast before contracting begins.

Before proceeding, we shall review how our findings relate to previous work. Our formal analysis of sequential contracting is preceded by the important articles of Courty and Li (2000) and Riordan and Sappington (1987), who were the first to demonstrate how agents may be assigned different contracts depending on their privately known distribution of preferences. Our model is closest to Courty and Li, who consider the optimal sorting of agents that differ according to the variability in their demand. Our analysis is similar to Courty and Li in that we demonstrate how managers are sorted according to the variability in the information signals they receive. In contrast to Courty and Li, however, our analysis is primarily concerned with how the expertise of the agent affects optimal contracts. Specifically, we demonstrate how contracts are structured to accommodate agents with different levels of expertise and determine to what extent owners benefit from contracting with expert managers.

At a general level, this study relates to the large and growing literature on delegation in agency relationships. Early articles in this area by Holmstrom (1984) and Armstrong (1994) analyze how to optimally delegate decisions to an exogenously endowed expert whose preferences are imperfectly aligned with the principal. These studies show that in the absence of payments, it is desirable to confine experts' decisions to certain areas. Our analysis complements these findings by showing that even when payments are permitted, it is optimal to constrain managers' decisions depending on their level of expertise. Aghion and Tirole (1997) and Szalay (2005) consider settings where an agent's expertise is acquired. They describe how the degree of decision authority afforded the agent determines an expert's acquisition of information. Our analysis shows that in a management setting, it is the targeting of compensation to certain tasks that determines incentives for sellers to acquire expertise.