Does Board Independence Reduce the Cost of Debt?

• DOI: http://doi.org/10.1111/fima.12068
• Author: Dong Chen,Michael Bradley
• Published date: 01 March 2015
• Date: 01 March 2015

Does Board Independence Reduce

the Cost of Debt?

Michael Bradley and Dong Chen∗

Using the passage of the Sarbanes-Oxley Act and the associated changes in listing standards as

a natural experiment, we find that while board independence decreases the cost of debt when

credit conditions are strongor leverage is low, it increasesthe cost of debt when credit conditions

are poor or leverage is high. We also document that independent directors set corporate policies

that increase firm risk. These results suggest that independent directors act in the interests of

shareholders and are increasingly costly to bondholders with the intensification of the agency

conflict between these two stakeholder groups.

Prior literature suggests that independent directors generally perform a better monitoring role

than affiliated or inside directors in increasing shareholder welfare (Fields and Keys, 2003;

Hermalin and Weisbach,2003). This is consistent with the reputational concer ns of these directors

as professional referees (Fama, 1980). Some monitoring functions, such as the reduction of

managerial shirking or perquisite consumption and the improvement of financial disclosure

and transparency, may also benefit creditors (Goh, Ng, and Yong, 2012; Armstrong, Core, and

Guay, 2013). Other potential actions that an independent board may take, however, may harm

bondholders when the interests of shareholders and bondholders diverge, leading to agency costs

of debt (Jensen and Meckling, 1976; Myers, 1977). Most studies find that board independence

is negatively related to the cost of debt (Bhojraj and Sengupta, 2003; Anderson, Mansi, and

Reeb, 2004; Ertugrul and Hegde, 2008) suggesting that, on average, the benefits of independence

outweigh its costs.

The focus of extant studies examining the overall effect of board independence on the cost

of debt does not preclude the possibility of a differential impact of independence under differ-

ent circumstances. If the difference between the benefits and costs of board independence on

bondholders is not the same in all states of the world, such a differential impact is expected. It is

plausible that the costs of board independence will increase with the severity of the shareholder-

bondholder (S/B) conflict (commonly characterized by the closeness to default or the level of

financial leverage) (Bodie and Taggart, 1978; John and John, 1993). The benefits of board in-

dependence, such as improved financial disclosure, may also be more important under financial

distress as there is generally greater uncertainty about the future profitability of a distressed

firm than a f inancially sound firm (Sengupta, 1998). Because the rates of changes in the costs

and benefits of board independence may differ as the S/B conflict intensifies, the net effect of

independence on the cost of debt may differ with respect to the degree of the S/B conflict. For

example, suppose the costs of board independence are less than the benefits at mild degrees of

This is a substantially revised version of the paper circulated earlier under the title “Corporate Governance, Credit

Condition, and the Cost of Debt.” Wethank an anonymous referee, Raghavendra Rau (Editor), and YanyanLiu for very

helpful comments. Wealso appreciate comments by Kiyoung Chang, Ivana Vitanova, GregoryNagel, and the participants

of the Midwest Finance Association and the Eastern FinanceAssociation 2014 Annual Meetings.

∗Michael Bradley is the F.M. Kirby Emeritus Professor of Investment Banking and Professorof Law at Duke University

in Durham, NC. Dong Chen is an Assistant Professor of Financeat the University of Baltimore in Baltimore, MD.

Financial Management •Spring 2015 •pages 15 - 47

16 Financial Management rSpring 2015

the S/B conflict, but increase at a faster rate than the benefits, eventually exceeding those benefits

after a certain level of the conflict. Then the impact of independence on the cost of debt should

be negative if the S/B conflict is mild, but positive if it is severe. These potentially differential

and particularly divergent impacts of board independence on the cost of debt have not received

much attention in the existing literature. This study attempts to fill this gap.

Using a sample of S&P 1,500 firms from 2001 to 2005 with senior unsecured bonds outstanding,

we examine the differential impact of board independence on the cost of debt, conditional on

the expected severity of the S/B conflict. We utilize the passage of the Sarbanes-Oxley Act

of 2002 and the associated changes in the listing standards instituted by the major exchanges

(abbreviated collectively as SOX subsequently) as a natural experiment. Employing propensity

score-weighted difference-in-differences (DID) regressions (Hirano, Imbens, and Ridder, 2003;

Imbens and Wooldridge,2009), we examine the effect of a plausibly exogenous increase in board

independence on the change in the cost of debt. This methodology distinguishes our study from

the rest of the literature, which relies on cross-sectional identification strategies and, as such, is

susceptible to the concerns of endogeneity.

To identify the noncompliant (treatment) firms, we focus on the SOX requirements that the

board of listed companies be comprised of a majority of independent directors and that all

of the members of a firm’s audit committee be independent. In contrast to most prior studies

(Bhojraj and Sengupta, 2003; Anderson et al., 2004; Ertugrul and Hegde, 2008), we do not find a

significant overall effect of board independence on the cost of debt, suggesting that on averagethe

benefits approximately equal the costs of independence. However, consistent with a differential

impact of board independence on the cost of debt, we find that greater independence results in a

significantly lower cost of debt when the expected S/B conflict is mild, but a significantly higher

cost of debt when the expected conflict is severe.

To provide some evidence that some actions that independent directors take may benefit share-

holders, but not necessarily bondholders, we examinethe effects of increased board independence

on the firm’s subsequent risk profile and payout policies; two actions that maybenef it one stake-

holder group, but hurt the other. Consistent with this view, we find that an exogenous increase

in board independence leads to an increase in risk-taking behavior, which presumably benefits

shareholders at the expense of bondholders. However, we do not find a significant effect of

independence on payout policies (dividends and repurchases).

We contribute to the literature across several dimensions. The divergent effect of board inde-

pendence on the cost of debt, depending upon contingencies, adds to the literature that argues that

“one-size-fits-all” regulations may have unintended consequences (Chhaochharia and Grinstein,

2007; Wintoki, 2007; Linck, Netter, and Yang, 2009; DeFond et al., 2011; Chen, 2014). While

both this paper and Chen (2014) analyze the differential effect of board independence on credit

risk, Chen (2014) focuses on credit ratings and confirms a non-monotonic impact. In contrast,

we examine bond spreads and study the cross-sectional variation of the impact conditional on the

expected S/B conflict.

Our finding of an opposite effect of board independence on the cost of debt depending on the

credit condition of the firm also provides new insight into the literature. While a few studies

have also examined the differential effect of corporate governance on the cost of debt with the

deterioration of credit quality (Bhojraj and Sengupta, 2003; Klock, Mansi, and Maxwell, 2005),

they focus on the stronger impact of governance for more distressed firms, with the implication

that the direction of the impact should be the same regardless of the circumstances. In contrast,

our results demonstrate that not only the magnitude, but also the sign of the effect of corporate

governance on the cost of debt may change with a firm’s credit condition.

Bradley & Chen rDoes Board Independence Reduce the Cost of Debt? 17

The risk-increasing effect of board independence also adds to the related literature. In contrast

to our findings, most studies find either a negative (Bargeron, Lehn, and Zutter, 2010; Ni

and Purda, 2012) or an insignificant correlation between independence and risk-taking (Cheng,

2008). However, with the exception of Bargeron et al. (2010), most studies relyon cross-sectional

variations that are susceptible to the concerns of endogeneity.1Although a few studies employ

an instrumental variable (IV) approach, there is still concern about the validity of their IVs.

For example, Ni and Purda (2012) use chief executive officer (CEO) tenure as an IV for board

independence. However, Chen and Zheng (2014) argue and find evidence that CEO tenure matters

for risk-taking behavior.It is notable that in analyzing the 2007-2008 f inancial crisis, Beltratti and

Stulz (2012) find that banks with more shareholder-friendly boards (including more independent

boards) are riskier. Consistent with our view, Beltratti and Stulz (2012) argue that these firms took

excessive risks ex ante in the interests of their shareholders, but, ex post their stocks performed

worse due to the unexpected financial crisis. The positive effect of board independence on

managerial risk-taking also provides a different perspectivewhen evaluating the effect of SOX as

compared with the existing research. Most studies find that SOX discourages risk-taking (Cohen,

Dey,and Lys, 2009; Bargeronet al., 2010). Consistent with these studies, we f ind that, overall,risk-

taking decreased post-SOX. However, firms that were not in compliance with the independence

requirement before SOX experienced a smaller decrease. Therefore, our results suggest that while

many provisionsin SOX, such as increased executive and director personal liabilities and the rules

related to internal controls, may discourage managerial risk-taking (Bargeron et al., 2010), the

requirement of higher board independence nevertheless promotes risk-taking.

The remainder of the paper is organized as follows. Section I describes the data, variables,

and summary statistics. This section also describes the timelines and the provisions of SOX that

are relevant for our study. Section II presents the empirical specifications and results. Section III

conducts robustness checks based on propensity score matching, while Section IV concludes.

I. Data, SOX Provisions, Variables, and Summary Statistics

A. Data and Sample

The appendix provides detailed descriptions of the variables used in our empirical analysis,

as well as their data sources. Consequently, we provide only a brief description of each variable

here.

Following the literature, we use bond yield spreads as a measure of the cost of debt. Our

bond data are taken from a proprietary database from S&P that contains the prices and issue

characteristics of all publicly traded senior unsecured corporate bonds as of the end of March

from 2002 to 2006. The year range and the inclusion of only senior unsecured bonds are a

result of data limitations, but we note that these bonds are the most prevalent of corporate

1In unreported analysis, we find some potential explanations for the difference in results between our study and Bargeron

et al. (2010), although both employ SOX as a natural experiment. First, while 2002 is the treatment year in our study,

Bargeronet al. (2010) def ine 2001 as the treatment year,b utstill set the post-SOX period to start from 2003. This mismatch

may be problematic since some noncompliant (compliant) firms at 2001 would become compliant (noncompliant) by

2002 and, as such, should no longer be treated as noncompliant (compliant). In addition, Bargeron et al. (2010) include

fewer control variables. Moreover, Bargeron et al. (2010) do not employ the propensity score method to address the

issue of the dissimilarity between noncompliant and compliant firms. Also, in Bargeron et al. (2010), standard errors

are not adjusted for heteroskedasticity and autocorrelation. Finally, whilewe use the requirements for both the majority

independence of entire boards and the full independence of audit committees to identify noncompliant firms, Bargeron

et al. (2010) focus on the majority independence standard.