The Value of Precontract Information About an Agent's Ability in the Presence of Moral Hazard and Adverse Selection

• Author: MASAKO DARROUGH,RAJIV D. BANKER,SHAOPENG LI,LUCAS THREINEN
• DOI: http://doi.org/10.1111/1475-679X.12290
• Date: 01 December 2019
• Published date: 01 December 2019

DOI: 10.1111/1475-679X.12290

Journal of Accounting Research

Vol. 57 No. 5 December 2019

Printed in U.S.A.

The Value of Precontract

Information About an Agent’s

Ability in the Presence of Moral

Hazard and Adverse Selection

RAJIV D. BANKER,

∗MASAKO DARROUGH,

†SHAOPENG LI,

∗

AND LUCAS THREINEN

‡

Received 6 April 2015; accepted 12 September 2019

ABSTRACT

We analyze the expected value of information about an agent’s type in

the presence of moral hazard and adverse selection. Information about the

agent’s type enables the principal to sort/screen agents of different types.

The value of the information decreases in the variability of output and the

agent’s risk aversion, two factors that are typically associated with the severity

of the moral hazard problem. However, the value of the information about

agent type first increases but ultimately decreases in the severity of adverse se-

lection. The decrease comes about because the means available to the princi-

pal to induce effort—namely, the pay–performance sensitivity—must also be

used to sort/screen agents, and these two goals conflict. This decline in value

occurs despite the monotonically increasing importance of the information

in determining the principal’s expected profits. Further, we show that the

peak value of information occurs at a predictable level of adverse selection.

These results imply that over some range, the importance of the information

∗Fox School of Business, Temple University; †Zicklin School of Business, Baruch College;

‡Macalester College.

Accepted by Haresh Sapra. We acknowledge helpful suggestions from the editor and an

anonymous referee that helped shape this paper.

1201

CUniversity of Chicago on behalf of the Accounting Research Center, 2019

1202 R.D.BANKER,M.DARROUGH,S.LI,AND L.THREINEN

will be increasing, and the value of the information will be simultaneously

decreasing, in the severity of adverse selection.

JEL codes: D82; J33; M52

Keywords: principal and agent; precontract information; moral hazard; ad-

verse selection

1. Introduction

In the process of hiring a new manager, a principal typically acquires, be-

fore contract negotiation, a variety of information about the candidate

manager’s ability.1Information may be hard or soft. Hard information,

such as the candidate’s educational background or past performance at

prior positions, is quantifiable and verifiable. Soft information, such as the

candidate’s reputation, may not be easily quantifiable or verifiable. Yet, in

either case, the compensation contracts offered by the firm may incorpo-

rate such acquired precontract information. If so, contracts can differ based

not only on postcontract, future outcomes, but also on the precontract

information. For example, contingent compensation (such as a bonus)

may depend on realized outcomes, but also on the information known

about the manager at the time of contract negotiation. Needless to say, the

fixed compensation (such as a salary) can only depend on the precontract

information.

In this paper, we develop a model to analyze the value of precontract

information in a principal-agent setting with both moral hazard and ad-

verse selection. A moral hazard problem may emerge if a principal cannot

observe her agent’s action, such as his effort level. The principal can ad-

dress this problem by making the manager’s compensation a function of

some observable outcome that depends on the manager’s effort (typically

referred to as pay–performance sensitivity or PPS). An adverse selection

problem may arise when some relevant traits of the agent cannot be ob-

served by the principal (e.g., the manager’s ability) so that the agent can

misrepresent himself. The principal can address this type of problem by

offering a menu of contracts, each of which would only be chosen by an

agent of a particular level of ability.

In designing a menu of contracts under asymmetric information, we

show that the principal faces a trade-off in structuring incentive contracts

to deal with both moral hazard and adverse selection simultaneously. When

adverse selection is overlaid on top of moral hazard, we find that the best

option available to the principal is to modify the compensation structure

that would have been offered if the principal is faced with only a moral haz-

ard problem. The result is a weaker link between pay and outcome in some

1We model “information” as simply anything that could improve the principal’sknowledge

about the agent’s managerial ability.

VALUE OF PRECONTRACT INFORMATION 1203

of the contracts, which decreases the incentive of the agent to exert effort.

But if the principal did not modify the contracts in this way, the agent might

agree to an even worse (from the principal’s perspective) contracting out-

come. Replacing an underperforming CEO with a new CEO, for example,

can be extremely costly (see Taylor [2010]). Thus, information about an

agent’s ability is fundamentally valuable because it allows the principal a

freer hand to focus the compensation structure on mitigating the moral

hazard problem.2

By analyzing a model with both moral hazard and adverse selection,

we first derive the ex ante value of precontracting information about the

agent’s ability.We show that, because of the dual purpose served by the PPS

in addressing both the adverse selection and moral hazard problems, the

value of the precontracting information on ability depends, interestingly,

on features related to the moral hazard problem: The value decreases in the

noisiness of the measure of effort and the risk aversion of the agent. This

is because the PPS in the menu of contracts, which would have been used

strictly to address the principal’s uncertainty regarding the agent’s postcon-

tracting action in a pure moral hazard setting, is now also the primary tool

available to the principal to sort/screen the agents by ability.3The princi-

pal achieves sorting by distorting the PPS in the menu of contracts from

the optimal value in a strictly moral hazard setting.4Thus, better precon-

tract sorting comes at the expense of providing suboptimal postcontract

incentives to the agent.

Because of this tension between precontract sorting and postcontract in-

centivization, the principal gains from better precontract information. If

the principal becomes more convinced that the agent is of a particular type,

the expected cost of modifying contracts on the menu decreases. In partic-

ular, the informed principal can gain by modifying contracts for other types

in such a way that they become more costly to the principal if implemented

but relatively more attractive to the types for which they are intended. This

is because it is less likely that any of these other contracts will ever take ef-

fect (because it is unlikely that the agent is of any of the types for whom

these other contracts are intended). This modification allows the principal

2By “ability,” we mean any trait of the agent such as intelligence, managerial talent, fit for

the position, or other personal characteristics that will, for a given level of managerial effort,

positively influence the expected outcome of interest to the principal. We also refer to the

realization of this trait for a particular agent as the agent’s “type.”

3Although “sorting” and “screening” are often used interchangeably in the literature, we

use sorting hereafter. We examine a setting in which the principal designs a mechanism so

that agents sort themselves. Eeckhout and Kircher [2010] refer to the mechanism as ex ante

sorting as opposed ex post screening, in which, for example, buyers reveal their reservation

values in an auction setting.

4Although in our model, the principal moves first by designing and offering a menu of

contracts to the agent, it is possible to have the agent move first by signaling his type at some

cost as in Spence [1973]. Such signals may be interpreted as a special case of the information

modeled in our paper so long as they are fixed by the time of contract negotiation.