The Complementarity Between Signal Informativeness and Monitoring

• Published date: 01 March 2023
• Author: PIERRE CHAIGNEAU,NICOLAS SAHUGUET
• Date: 01 March 2023
• DOI: http://doi.org/10.1111/1475-679X.12459

DOI: 10.1111/1475-679X.12459

Journal of Accounting Research

Vol. 61 No. 1 March 2023

Printed in U.S.A.

The Complementarity Between

Signal Informativeness and

Monitoring

PIERRE CHAIGNEAU∗AND NICOLAS SAHUGUET†

Received 29 December 2020; accepted 14 July 2022

ABSTRACT

A f‌irm that must decide whether to retain or terminate a manager can rely

on several sources of information to assess managerial ability. When it re-

lies on a performance signal and monitoring, we show that a more infor-

mative signal can surprisingly increase the value of monitoring. Then, sig-

nal precision and monitoring are complements. This happens if a more

precise information system makes some signals more negative indicators of

managerial ability that still do not trigger termination. When the turnover

cost is high enough and the manager is more entrenched after a positive

performance, an increase in signal precision increases expected monitor-

ing. In f‌irms with a high turnover cost, a less informative signal is com-

pounded by worse monitoring after a disappointing performance. This “bad

corporate governance trap” makes it hard for these f‌irms to eventually im-

prove performance.

∗Queen’s University, Smith School of Business; †Applied Economics Department, HEC

Montréal

Accepted by Haresh Sapra. We thank an anonymous reviewer,Jeremy Bertomeu, Matthieu

Bouvard, Jean-Étienne de Bettignies, Robert Göx (discussant), Adolfo de Motta, Naomi

Rothenberg (discussant), Luke Taylor, Mike Waldman, seminar participants at Erasmus Uni-

versity Rotterdam, McGill, Université Laval, the Accounting and Economics Society webinar,

the 12th Accounting Research Workshop, the 14th Workshopon Accounting and Economics,

and the HEC Montréal/McGill Workshop, for useful comments and suggestions.

141

© 2022 The Chookaszian Accounting Research Center at the University of Chicago Booth School of

Business.

142 p. chaigneau and n. sahuguet

JEL codes: D83, G34, M12, M41, M51

Keywords: corporate governance system; governance complementarity;

hard and soft information; informativeness; monitoring; skew-normal dis-

tribution

1. Introduction

Assessing the performance of a manager, and using this information to con-

tinue or terminate the manager, is a key aspect of corporate governance

(Laux [2014], Hermalin and Weisbach [2017]). To this end, investors can

rely on accounting measures (e.g., earnings), or market measures (e.g., the

stock price). In addition, investors or their agent (the board) can invest

resources to acquire additional information, including soft information,

about f‌irm performance and management quality. Tirole [2006] refers to

the f‌irst mechanism as “passive monitoring,” and to the second as “active

monitoring.” Investors then intervene on the basis of this information, most

notably by replacing the f‌irm’s manager.

An important question in the design of corporate governance systems

is whether these two types of monitoring are complements or substitutes.

According to Tirole [2006], this question “is central to the design of

the f‌inancial system (...) and yet it has not been investigated in detail

in the literature.” Leuz and Wysocki [2016] point out that changes in f‌i-

nancial reporting standards will likely affect other dimensions of the cor-

porate governance environment, and conclude that “we know relatively

little about the nature and importance of such institutional complemen-

tarities.” Armstrong, Guay, and Weber [2010] likewise argue that “we know

relatively little, however, about how f‌irms select among the various mecha-

nisms available to them, or how the various mechanisms interact and serve

as complements to and/or substitutes for each other.”

This question also has implications for the adequacy of f‌inancial report-

ing standards and corporate governance. When assessing the overall ef-

fect of an improvement in the quality of a performance signal, it is im-

portant to take into consideration the indirect effect of this improvement

on the intensity of active monitoring (henceforth “monitoring”), which

also provides information about the f‌irm’s management. If these sources

of information are substitutes, then a lower quality performance signal

is not highly detrimental for the termination decision because it will re-

sult in enhanced monitoring, that is, the endogenous increase in mon-

itoring counteracts the decrease in signal quality. If these sources of in-

formation are complements, however, then a lower quality performance

signal will lead to less monitoring, that is, the endogenous decrease in

monitoring compounds the information loss due to the decrease in signal

quality.

governance complementarity143

To study this question, we consider a stylized model of corporate gover-

nance. Our model can be viewed as an extension of Hermalin [2005] with

a signal of f‌irm performance, which can be interpreted as earnings or the

stock price. The f‌irm’s manager,whose ability affects the signal distribution,

can be terminated or reappointed. Even though the ability of potential

managers is unknown, the f‌irm can learn about the ability of the incum-

bent manager via two channels. First, the f‌irm observes the signal, which is

(imperfectly) informative about managerial ability. Second, after observing

this signal, the board must decide on the intensity of its monitoring, which

may provide additional information about the ability of its manager.

Based on all this information, the board has the option to terminate

the manager. Termination and transition to a new manager can involve

a turnover cost. As in Taylor [2010], we let the “effective turnover cost” in-

corporate both the cost to the f‌irm and the personal cost to the board—the

latter represents managerial entrenchment. We allow managerial entrench-

ment to be higher following a positive signal, which is good news about the

manager’s ability, rather than a negative signal.

We start with the simple case of a symmetric signal distribution, which

measures managerial ability, albeit imperfectly,and no turnover cost. Then,

an increase in the precision of the signal always reduces the value and there-

fore the intensity of monitoring. We then consider more general cases. This

allows us to identify sources of complementarity between the quality of the

signal and the intensity of monitoring. We say that the precision of the sig-

nal and monitoring are “complements” when an increase in the former

increases the latter, and “substitutes” otherwise.

First, we assume a positive effective turnover cost. Absent monitoring, the

manager is terminated whenever the performance signal is suff‌iciently bad

news about his ability. However, some signals that are bad news do not trig-

ger termination absent monitoring, as turnover is costly. When such a signal

is generated by a more precise information system, it is even worse news

about managerial ability. This increases the value of monitoring, which

provides additional information on managerial ability. In short, a positive

turnover cost creates complementarity between signal precision and mon-

itoring if the signal leads to negative belief updating on managerial ability

yet does not trigger termination.

Second, we use the skew-normal distribution to allow for potential asym-

metries in the signal distribution. This distribution generalizes the normal

distribution to allow for positive or negative skewness. With positive skew-

ness, higher precision raises the probability that some signals that are bad

news about managerial ability are generated by a high ability manager at

a higher rate than for a low ability manager. Then, these signals are now

paradoxically less informative, which raises the value of monitoring. A sym-

metrical effect is at play with negative skewness.

We then study the complementarity between signal precision and moni-

toring on average. That is, instead of asking whether monitoring increases

after a signal realization that was generated by a more precise information