How Firms Use Director Networks in Setting CEO Pay

• Published date: 01 August 2019
• Date: 01 August 2019
• Author: Vladimir A. Gatchev,Ian Cherry
• DOI: http://doi.org/10.1111/fire.12190

The Financial Review 54 (2019) 501–540

How Firms Use Director Networks

in Setting CEO Pay

Ian Cherry∗and Vladimir A. Gatchev

University of Central Florida

Abstract

We examine how firms use the network of overlapping directorships to determine chief

executiveofficer (CEO) compensation. We contribute to related work by empirically exploring

two competing hypotheses. In the first hypothesis, networks propagate relevant information

used to establish good pay practices. In the second hypothesis, director networks are used

opportunistically to benefit the CEO. The empirical findings are generally consistent with

the first hypothesis. Yet, the importance of director networks is reduced when the CEO is

entrenched and when management hires a compensation consultant. The latter finding is

especially pronounced when director networks predict a reduction in CEO pay.

Keywords: executive compensation, director networks, corporate governance

JEL Classifications: G34, G38, J33, M12

∗Corresponding author: College of Business Administration, University of Central Florida, Orlando, FL

32816-1400; Phone: (407) 823-2977; Fax: (407) 823-6676; E-mail: ian.cherry@ucf.edu.

The authors are grateful to the Editor Richard S. Warr and two anonymous referees, whose insightful

comments have greatly improved the paper.For comments and suggestions on an earlier draft, we would

like to also thank Vladimir Atanasov, Honghui Chen, Melissa Frye, Daniel Green, Charles Schnitzlein,

Stephan Shipe, Ajai Singh, Geoffrey Turnbull, Qinghai Wang, and seminar participants at the University

of Central Florida, Boise State University, Loyola Marymount University, University of South Florida,

the 2014 Florida Finance Conference, the 2014 Financial Management Association annual meetings, the

2015 Midwest Finance Association annual meetings, and the 2015 Southwestern Finance Association

annual meetings. We thank Lalitha Naveenfor making available the CEO delta and vega data. We remain

responsible for any errors.

C2019 The Eastern Finance Association 501

502 I. Cherry and V.A. Gatchev/ The Financial Review 54 (2019) 501–540

1. Introduction

The interaction of overlapping directors forms a social network of influence

through which various corporate policies propagate. Research has established the

importance of director networks for firm policies such as options backdating (Bizjak,

Lemmon and Whitby, 2009), earnings management (Chiu, Teoh and Tian, 2013),

various measures of corporate governance practices (Bouwman, 2011), tax avoidance

strategies (Brown and Drake, 2014), board compensation (Boivie,Bednar and Barker,

2015), and the composition of executive compensation and incentives(Wong, Gygax

and Wang, 2015; Gallani, 2016).

To provide new insights in the role played by director networks, this paper per-

forms a detailed empirical analysis of how the chief executive officer (CEO) pay in

one firm is related to the CEO pay in other firms sharing its directors, which we label

networked CEO pay. Our interest in CEO pay is, in part, motivated by the ongoing

debate about what drives executive compensation.1We explore two competing hy-

potheses. The first hypothesis is derived under the optimal contracting view, which

argues that the board determines CEO pay in the best interest of shareholders. Accord-

ing to this view, director networks propagate relevant information used to establish

good pay practices. The second hypothesis is derived under the view that the board is

largely under the control of the CEO, and CEOs use their influence to set their own

pay and extract rents (e.g., Bebchuk, Fried and Walker, 2002; Bebchuk and Fried,

2003, 2004). According to this view, director networks are used opportunistically for

the benefit of the CEO.

We examine how CEO pay depends on networked CEO pay. Controlling for

year and industry effects and for other relevant firm, CEO, and board characteristics,

we find that CEO pay increases by 5.4% for a one-standard-deviation increase in

networked CEO pay.We then test whether boards take into account the characteristics

of firms in the director network when propagating CEO pay practices. To that end,

we estimate a predictive model and decompose networked CEO pay into a predicted

component and a residual component. The predicted component is CEO pay predicted

by firm, CEO, and board characteristics, and the residual component is the residual

from the model. We find CEO pay is more sensitive to the residual component than

to the predicted component of networked CEO pay.Further analysis suggests that the

significance of the predicted component of networked CEO pay is due to latent firm

characteristics rather than an actual influence on CEO pay.

Toensure our findings are robust to endogeneity concerns, all analyses control for

the endogenous selection of directors to boards using a Heckman approach (Heckman,

1979). We further verify the importance of director networks while focusing on

directors with ongoing directorships at the firm of interest and new directorships at

1For a review of this extensive literature, see, for example, Frydman and Jenter (2010). A press paper

by Steven Clifford provides further anecdotal evidence on how companies decide to pay their CEOs:

(https://www.theatlantic.com/business/archive/2017/06/how-companies-decide-ceo-pay/530127/).

I. Cherry and V.A. Gatchev/ The Financial Review 54 (2019) 501–540 503

other firms. In such cases, the appointment at new firms is not under the control

of the firm of interest and so endogeneity concerns should be mitigated. As another

approach to handle endogeneity, we confirmthat our findings are robust to geography

fixed effects, firm fixed effects, and CEO fixed effects.

The relevance of director networks is relatively stable over time, even though

the past 20 years have experienced significant changes in governance practices such

as the Sarbanes-Oxley Act (SOX) of 2002, the accompanying changes in NYSE and

NASDAQ-listing requirements for board independence, and the Dodd-Frank Wall

Street Reform and Consumer Protection Act of 2010 (Dodd-Frank). To examine

whether our findings capture a mechanical link due to broadly rising pay levels, we

estimate both the importance of direct links and indirect links between firms. A direct

link exists when two firms share a director, while an indirect link exists when two

firms do not share a director but have directors who jointly serve on the boards of

other firm(s). Direct links should provide astronger communications channel between

firms (Larcker, Richardson, Seary and Tuna,2005). Indeed, we find (1) indirect links

do not have a significant effect on CEO pay, and (2) the importance of direct links is

not diminished if we control for indirect links.

Further analyses produce several new findings, which contribute to the under-

standing of how firms use director networks to set CEO pay. First, we find CEO pay

responds significantly to residual networked CEO pay both when residual networked

CEO pay is negative (i.e., belowaverage) and when it is positive (i.e., above average).

In some specifications, the response to positive shocks is twice as large as that to

negativeshocks, but this difference is not statistically significant. Second, we find that

director networks have a more pronounced effect on CEO pay for firmswith stronger

corporate governance than for firms with weaker governance. Notably, the CEO pay

of firms with weaker governance is not sensitive to negative shocks in networked

CEO pay. Third, we explore how the use of compensation consultants changes the

role of director networks in determining CEO pay. We find that when management

hires consultants, CEO pay is sensitive to networked CEO pay only when networked

CEO pay is relatively high but not when it is relatively low, and that management is

more likely to hire a compensation consultant when networked CEO pay is relatively

low.

A potential concern is that our findings may be affected by the use of compen-

sation peers.2While the benefits of using compensation peers are still debated, it is

widely accepted that peer groups are significant in explaining CEO pay practices.

Faulkender and Yang (2010), Bizjak, Lemmon and Nguyen (2011), and Elson and

Ferrere (2013) find that firms tend to select highly paid peers, creating a natural

2This concern is somewhat mitigated by findingsin Cremers and Grinstein (2014) that CEO talent markets

do not span across industry lines, so industry effects should capture some of the peer effects. Moreover,

due to antitrust laws (e.g., Section 8 of the Clayton Act), most directors who serve on multiple boards also

serve on boards across different industries, so director network effects should be largely independent of

industry peer effects.