Optimal Contracts with Performance Manipulation

• Date: 01 September 2014
• Author: ANNE BEYER,IVÁN MARINOVIC,ILAN GUTTMAN
• DOI: http://doi.org/10.1111/1475-679X.12058
• Published date: 01 September 2014

DOI: 10.1111/1475-679X.12058

Journal of Accounting Research

Vol. 52 No. 4 September 2014

Printed in U.S.A.

Optimal Contracts with

Performance Manipulation

ANNE BEYER,

∗ILAN GUTTMAN,

†AND IV´

AN MARINOVIC

∗

Received 10 February 2012; accepted 29 May 2014

ABSTRACT

We study optimal compensation contracts that (1) are designed to address a

joint moral hazard and adverse selection problem and that (2) are based on

performance measures, which may be manipulated by the agent at a cost. In

the model, a manager is privately informed about his productivity prior to be-

ing hired by a firm. In order to incentivize the manager to exert productive

effort, the firm designs a compensation contract that is based on reported

earnings, which can be manipulated by the manager. Our model predicts

that (1) the optimal compensation contract is convex in reported earnings;

(2) the optimal contract is less sensitive to reported earnings than it would be

absent the manager’s ability to manipulate earnings; and (3) higher costs of

manipulating reported earnings (e.g., due to higher governance quality) are

associated with higher firm value, lower expected level of earnings manage-

ment, and higher output.

1. Introduction

In practice, many firms offer performance-based compensation schemes

to their executives. These compensation schemes are designed to align

managers’ incentives with those of shareholders. Specifically, performance-

based compensation contracts allow firms to attract productive personnel

∗Stanford University; †New York University.

Accepted by Haresh Sapra. We thank David Aboody, Baricz Arpad, Suren Basov, Jack

Hughes, Madhav Rajan, Stefan Reichelstein, Richard Saouma, Ilya Segal, Robert Wilson, and

two anonymous referees for detailed feedback, and workshop participants at UCLA and Stan-

ford University for many helpful comments.

817

Copyright C, University of Chicago on behalf of the Accounting Research Center,2014

818 A.BEYER,I.GUTTMAN,AND I.MARINOVIC

(adverse selection) and also to incentivize their employees to exert more

unobservable effort (moral hazard). However, performance-based compen-

sation is not only an instrument for addressing the agency problem be-

tween managers (agents) and shareholders (principal) but is also part of

the agency problem itself. The reason is that performance-based compen-

sation may provide incentives for the manager to manipulate financial re-

ports. Empirical evidence suggests that managers in fact do manipulate re-

ported earnings.

Our model captures these fundamental economic tradeoffs related

to performance-based compensation contracts of corporate executives.

Specifically, we model the following joint adverse selection problem and

multitask moral hazard problem. In our model, participation, compensa-

tion, production, and manipulation are endogenous. Managers are pri-

vately informed about their productivity, that is, managers know their per-

sonal cost of exerting productive effort prior to signing the contract with

the firm. This assumption reflects that managers are likely to possess su-

perior information about their personal skills and ability (e.g., Laffont

and Tirole [1986], Crocker and Morgan [1988]). The firm hires the man-

ager to produce output by exerting effort. The manager’s effort or, equiva-

lently, the firm’s true output, is unobservable to shareholders.1Therefore,

the contract must be written in terms of the only observable performance

measure—earnings reported by the manager. The manager does not have

to report earnings truthfully but may manipulate earnings at a personal

cost. When designing the manager’s contract, shareholders aim to max-

imize the expected output generated by the manager’s productive effort

net of his compensation. Anticipating that the manager may increase re-

ported earnings in two ways, shareholders take into account the effect of

the contract not only on the manager’s productive effort but also on his

decision to manipulate reported earnings. In this sense, the model is re-

lated to multitask moral hazard problems (e.g., H¨

olmstrom and Milgrom

[1991], Feltham and Xie [1994]). In addition to moral hazard problems,

when designing the contract, shareholders must consider the adverse se-

lection problem related to the manager’s productivity, which is privately

known to the manager.

In the model, we assume that the manager’s cost of productive effort is

quadratic in the manager’s effort, where the cost coefficient reflects (the

inverse of) the manager’s productivity/ability. Similarly, we assume that the

manager’s cost of manipulating earnings is quadratic in the manipulation

of reported earnings, where the cost coefficient reflects the effectiveness

of the firm’s corporate governance system in deterring the manager from

earnings manipulation. In contrast to the manager’s ability, we assume that

the manager does not possess any private information about the effective-

ness of the firm’s corporate governance system, that is, the manager’s cost

1Alternatively, output may become observable after the manager’s horizon, such that out-

put itself cannot be used for contracting.