The Effects of Corporate Governance on Firms’ Credit Ratings

Hollis Ashbaugh-Skaife
University of Wisconsin – Madison
hashbaugh@bus.wisc.edu

Daniel W. Collins *
University of Iowa
daniel-collins@uiowa.edu

Ryan LaFond
Sloan School of Management, MIT

March 2004
Revised, September 2004, May 2005

_________________

* Corresponding author

We would like to thank Sanjeev Bhojraj, Bob Bowen, Tom Dyckman, Paul Hribar, April Klein,
S. P. Kothari, Charles Lee, Mark Nelson, Kathy Petroni, Shiva Rajgopal, D. Shores, Joe Weber,
Peter Wysocki, and seminar participants at Cornell, University of Illinois Audit Symposium,
Iowa State University, Lancaster University, London Business School, MIT, Washington
University, and the University of Washington for helpful comments and suggestions. We
especially thank Johannes Ledolter for useful discussions on implementation and interpretation
of ordered logit models and editor, Tom Lys, for detailed comments on our paper.
 The Effects of Corporate Governance on Firms’ Credit Ratings

Abstract

Using a framework for evaluating corporate governance recently developed by Standard and Poor’s, this
study investigates whether firms that exhibit strong governance benefit from higher credit ratings relative
to firms with weaker governance. We document, after controlling for risk characteristics, that firm credit
ratings are: (1) negatively associated with the number of blockholders that own at least a 5% ownership in
the firm; (2) positively related to weaker shareholder rights in terms of takeover defenses; (3) positively
related to accrual quality and earnings timeliness; and (4) positively related to over-all board
independence, board stock ownership and board expertise, and negatively related to CEO power on the
board. We also provide evidence that CEOs of firms with speculative grade credit ratings are
overcompensated to a greater degree than their counterparts at firms with investment grade ratings, and
that the overcompensation exceeds the CEO’s share of additional debt costs related to lower credit
ratings. Our study provides insights into the characteristics of governance that are likely to affect the cost
of debt financing and provides one explanation for why some firms continue to operate with weaker
governance when doing so may mean lower credit ratings.

JEL classification: G30; G32; M41

Keywords: Corporate governance, Credit rating, Executive compensation

 The Effects of Corporate Governance on Firms’ Credit Ratings

I. Introduction

This paper investigates whether firms that possess strong corporate governance benefit from higher

credit ratings relative to firms with weak governance. A firm’s credit rating reflects a rating agency’s

opinion of an entity's overall creditworthiness and its capacity to satisfy its financial obligations (Standard

and Poor’s 2004). Credit agencies are concerned with governance because weak governance can impair a

firm’s financial position and leave debt stakeholders (hereafter referred to as bondholders) vulnerable to

losses (FitchRatings 2004). To structure our analysis, we adopt a framework developed by Standard &

Poor’s for assessing firms’ corporate governance structures and practices (Standard & Poor’s 2002).

Standard & Poor’s (2002) framework focuses on four major components of governance: Ownership

Structure and Influence, Financial Stakeholder Rights and Relations, Financial Transparency, and Board

Structure and Processes. The governance attributes we examine within each of these components are

designed to increase the monitoring of management’s actions to promote effective decision making, limit

their opportunistic behavior and reduce the information asymmetry between the firm and its external

stakeholders. We investigate what effect, if any, these governance features have on firms’ overall credit

ratings.

Our analysis yields several key findings. First, we find variables that capture each of the four major

components of corporate governance enumerated above help explain overall credit ratings after

controlling for firm characteristics that prior research has shown to be related to debt ratings.

Specifically, we find that firms’ overall credit ratings are: (1) negatively associated with the number of

blockholders that own at least a 5% ownership in the firm; (2) positively related to weaker shareholder

rights in terms of takeover defenses; (3) positively related to the quality of working capital accruals and

the timeliness of earnings; and (4) positively related to over-all board independence, board stock

ownership, board expertise, and negatively related to CEO power on the board. To provide an indication

of the economic significance of our results, we find that moving from the lower quartile to the upper

1
quartile of the governance variables nearly doubles a firm’s likelihood of receiving an investment grade

credit rating--from .46 to .93.1 During the time frame of our analysis, the average yield for firms with

investment grade debt with a ten year maturity was approximately 6.00%. In contrast, the average yield

for firms with speculative grade debt with a ten year maturity was approximately 14.0%. This 800-basis

point spread translates into an annual interest cost differential of $38.4 million for the median firm in our

sample with a speculative grade credit rating.2

Our results suggest that weak governance can result in firms incurring higher debt financing costs. So

why are some firms willing to bear additional debt costs by not practicing good governance? We

approach this question by considering how CEOs can extract rents from weak governance. One way

CEOs can appropriate rents is through excess compensation. To investigate this conjecture, we estimate

CEO excess compensation following the work of Core, Holthausen and Larcker (1999). We document

that CEOs of firms with weaker governance (greater CEO power or management entrenchment) receive

more excess compensation relative to the CEOs of firms with stronger governance (less management

entrenchment). Furthermore, we show that firms with speculative grade debt have a greater propensity to

overcompensate their CEOs than do firms with investment grade debt. For firms with speculative grade

credit ratings, we then compare CEO excess compensation to their share of additional debt costs that

1
For purposes of this analysis, we hold the firm characteristic variables (ROA, LEV, SIZE, etc.) constant at the
mean values for the sample. For those governance attributes found to be positively (negatively) related to credit
ratings, our benchmark probability is determined by assigning governance values equal to the first (third) quartile
and then moving to the third (first) quartile value. For governance attributes measured as 0-1 dummy variables, the
benchmark probability is determined with the zero (one) value when the governance attribute is positively
(negatively) related to credit ratings.
2
To calculate this interest cost differential, we first determine the mean debt-to-asset ratio for our sample of
investment grade firms (0.28). We then multiply this ratio times the total assets of each speculative grade firm to
estimate the “as-if” capital structure if the speculative grade firm was an investment grade firm. Finally, we
multiply the estimated debt level times the 800-basis point spread to determine a speculative firm’s hypothetical
incremental interest cost due to weak governance. This additional interest can be viewed as a cost that shareholders
bear if the firm chooses not to implement stronger governance to preclude management from stealing firm assets.
However, this spread overstates the cost of weak governance from the shareholders viewpoint when weaker
governance allows wealth transfers from bondholders to shareholders to take place. To the extent that part of the
interest rate spread between investment grade and speculative grade debt is due to bondholders price-protecting
against potential wealth transfers between themselves and stockholders, this spread represents an upper bound on the
cost of weak governance from the shareholder’s perspective.

2
these firms bear due to weak governance. We find that the median excess compensation far outweighs

the CEO’s share of the additional after-tax interest cost from having speculative grade debt versus

investment grade debt, thus providing one explanation for why all firms do not practice good governance.

This paper makes several contributions to the extant literature on bond (credit) ratings and corporate

governance. Prior literature investigating firms’ credit ratings and debt costs models the cost of debt as a

function of issue characteristics and issuer risk attributes (see e.g., Horrigan, 1966 or Kaplan and Urwitz,

1979) while ignoring the governance mechanisms that are put into place to safeguard the assets of

the firm and ensure that bondholder interests are well-served. We extend the traditional bond

rating literature by incorporating governance features designed to control agency conflicts

between bondholders and management and between bondholders and shareholders.

Sengupta (1998) and Bhojraj and Sengupta (2003) explore the effects of corporate governance on

debt ratings and cost of debt financing, but restrict their analysis to a limited set of governance variables.

Sengupta (1998) finds a negative relationship between firms’ disclosure quality ratings and the cost of

debt financing as reflected in realized yields on new debt issues. Bhojraj and Sengupta (2003) find that

firms with a higher percentage of outside directors on the board and with greater institutional ownership

enjoy lower bond yields and higher ratings on their new debt issues. We extend these two studies by

evaluating a broader set of governance variables thereby providing a more comprehensive analysis of the

relevance of corporate governance from the perspective of bondholders.

Much of the prior literature that investigates the effect of corporate governance focuses on the value

of corporate governance from the shareholder’s perspective (McConnell and Servaes, 1990; Yermack,

1996; Karpoff, Malatesta and Walkling, 1996; Gompers, Ishii and Metrick, 2003). Our study focuses on

the value of governance from the bondholders’ perspective because debt is a major source of capital for

3
publicly traded firms.3 Thus, to the extent that governance is an important determinant of credit ratings, it

can have a significant effect on the cost of firms’ external financing costs.

In this paper, we investigate how various governance mechanisms that are intended to control agency

conflicts between management and all stakeholders impact credit ratings. In addition, our study provides

insights into the governance attributes that heighten or attenuate the potential conflict between

bondholders and shareholders. Although generally aligned, the interests of bondholders and shareholders

can diverge when there are differing stakes in firm performance and differing views on management’s

investment policies (FitchRatings, 2004). Gompers, et al. (2003) find that firms with stronger shareholder

rights have higher share values and enjoy a lower cost of equity capital. In this study, we find that firms

with stronger shareholder rights have lower credit ratings implying a higher cost of debt financing. Our

study is one of the first to demonstrate that governance mechanisms that benefit shareholders may do so

at the expense of bondholders.4 Thus, governance mechanisms designed to give more power to

shareholders can have wealth redistribution effects that leave bondholders worse off.

The remainder of the paper is organized as follows. Section II briefly describes the role of governance

in mitigating agency conflicts between bondholders and management and between bondholders and

stockholders. Section III sets forth the framework recently adopted by Standard and Poor’s for evaluating

the strength of firms’ corporate governance mechanisms and develops empirical proxies to capture

various elements within this framework. Section IV describes our sample, data sources, and variable

measurements and provides descriptive statistics. Section V presents the empirical models used to

investigate the relation between various corporate governance mechanisms and firms’ credit ratings along

with the main empirical results. In Section VI we present evidence on CEO excess compensation related

3
For example, in calendar year 2000, U.S. firms obtained $944.8 billion dollars in debt financing whereas equity
financing totaled $311.9 billion dollars (U.S. Census Bureau 2005).
4
Our results are consistent with a concurrent study by Klock, Mansi and Maxwell (2004) who find that firms with
stronger anti-takeover provisions (weaker shareholder rights) enjoy a lower cost of debt financing relative to firms
with weaker anti-takeover provisions.

4
to weak governance, address endogeneity issues, and conduct sensitivity analyses. Section VII concludes

and offers suggestion for future research.

II. Why Governance Affects Firms’ Credit Ratings

Firm credit ratings are determined by rating agencies’ assessment of the probability distribution of

future cash flows to bondholders, which in turn, depends on the future cash flows to the firm. A firm’s

creditworthiness is determined by assessing the likelihood that its future cash flows will be sufficient to

cover debt service costs and principal payments. As the mean of the firm’s future cash flow distribution

shifts downward or the variance of its future cash flows increases, the likelihood of default increases and

the firm’s credit rating will decline.

Within the Jensen and Meckling (1976) agency theory framework, bondholders, and more generally

debt stakeholders, face two types of agency conflicts that can increase the probability of default and,

hence, reduce the value of their claims. The first is the conflict between management and all external

stakeholders—both bondholders and shareholders. Separation of ownership and control in corporate

organizations leads to information asymmetry problems between external stakeholders and managers.

Information asymmetry creates a moral hazard problem when managers have incentives to pursue their

own interests at the expense of external stakeholders. Self-interested managerial behavior can take several

forms including shirking, consumption of perquisites, over compensation, and empire building, all of

which increase the agency risk faced by external stakeholders and decrease the expected value of the cash

flows to the firm and its external stakeholders. As the firm’s expect cash flows decline, the default risk of

bondholders increases leading to lower credit ratings.

The second agency conflict faced by bondholders is the conflict with shareholders. Shareholders in

levered firms have incentives to undertake actions that can transfer wealth from bondholders to

themselves. This wealth transfer can take several forms that affect the mean and the variance of the firm’s

future cash flows. For example, if shareholders demand direct payouts of firm assets (dividends or share

repurchases) as opposed to supporting manager’s investments in positive net present value projects, then

the mean of a firm’s future cash flow distribution will be lower. The reduction in a firm’s expected future

5
cash flows increases bondholders’ default risk. Likewise, if shareholders influence managers to invest in

riskier projects that increase the variance of a firm’s future cash flows, bondholders face greater default

risk. In both examples, bondholders bear greater risk that their fixed contractual claims on the firm’s cash

flows will not be paid while shareholders potentially are better off.

We hypothesize that governance features impact credit ratings by controlling agency costs that result

from conflicts between managers and all external stakeholders as well as between bondholders and

shareholders. Many of the governance features we examine are designed to reduce the agency conflict

between managers and all stakeholders. Governance mechanisms that provide independent monitoring of

management promote effective managerial decision making that increases firm value and guard against

opportunistic management behavior that decreases firm value. Governance mechanisms that promote

better managerial decision making and limit opportunistic management behavior benefit all stakeholders.

Conversely, we posit that if governance is weak, the firm’s distribution of future cash flows will shift

downward relative to what it would be with more effective governance. This increases the likelihood of

default resulting in a lower credit rating. For convenience, we refer to the role that governance plays in

mitigating the agency conflicts between management and all stakeholders as the “management

disciplining” hypothesis.

Shareholder and bondholder interests are generally aligned when better monitoring of management

occurs. However, certain elements of corporate governance have a more ambiguous impact on

bondholders (FitchRatings, 2004). For example, some features of governance can place greater power in

the hands of shareholders (or selected subsets of shareholders) who can assert their influence to obtain

preferential treatment at the expense of other stakeholders (e.g., greenmail or targeted share repurchases

[Dann and DeAngelo, 1983]). Alternatively, shareholders can use their voting power to encourage

management to undertake risky investments or engage in ownership changes that can harm bondholder

interests. Taking on riskier projects increases the likelihood of default, resulting in lower credit ratings.

Some of the governance features we consider below (e.g., shareholder rights) have the potential for

effecting wealth transfers between bondholders and shareholders. Hence, while beneficial from the

6
shareholders perspective, certain governance features potentially can be harmful to bondholders.5 Or,

alternatively, governance features that weaken shareholder rights may actually be viewed positively from

the bondholder’s perspective. We refer to this as the “wealth redistribution” hypothesis regarding how

governance impacts credit ratings.

In sum, the governance variables introduced in the next section proxy not only for the agency

conflicts between external stakeholders and management, but also potential conflicts between

bondholders and shareholders that can result in wealth transfer effects between these two stakeholder

groups.

III. Corporate Governance Attributes

Prior studies on corporate governance tend to focus on one attribute of governance, e.g., board

independence, as opposed to studying a set of governance attributes intended to protect stakeholders’

claims to firms’ resources. A limitation of this research is that some governance attributes may

complement each other in protecting stakeholders’ claims whereas other governance attributes may serve

as substitutes. As a result, inferences drawn from studying one attribute of governance may be limited.

In 2002, Standard & Poor’s developed a framework for evaluating corporate governance that is based on

four governance components; ownership structure and influence, financial stakeholders rights and

relations, financial transparency and disclosure, and board structure and processes (Standard & Poor’s,

2002). Many of the individual governance mechanisms studied in the prior academic literature can be

classified into one of these four governance components. We use Standard & Poor’s (2002) framework,

along with prior literature, to identify governance attributes that potentially affect firms’ credit ratings.

The four dimensions of governance and the empirical proxies to capture the major attributes of

governance within each dimension are discussed below.

5
For example, shareholders will only approve mergers or acquisitions that serve their interests. But bondholders do
not always benefit under all takeover scenarios (see Asquith and Wizman 1990, and Warga and Welch 1993). So
giving shareholders greater power to determine ownership changes may well be viewed as an additional risk factor
by bondholders and rating agencies.

7
III.1 Ownership Structure and Influence

Typically, corporate governance is viewed from the perspective that publicly traded firms have

dispersed shareholders who demand governance to protect their residual claims. Governance mechanisms

that monitor management actions and limit their opportunistic behavior protect the interests of residual

claimants (shareholders) and the interests of bondholders as well. At times, however, the interests of

shareholders and bondholders can diverge. For example, shareholders with significant ownership

positions can exercise their influence to force management to take on more risky investments where

shareholders as a group receive the benefits of successful outcomes, but bondholders bear a

disproportionate share of the failures. Ownership structure and the influence that certain shareholders

exert on management play a key role in determining the potential for wealth transfers between

bondholders and stockholders, and are important elements of corporate governance.

Jensen (1993) and Shleifer and Vishny (1997) argue that blockholders or institutional investors that

hold large equity positions in a company are important to a well-functioning governance system because

they have the financial interest and independence to view firm management and policies in an unbiased

way, and they have the voting power to put pressure on management if they observe self-serving

behavior. Consistent with this view, Gordon and Pound (1993) find that the structure of share ownership

significantly influences voting outcomes on shareholder-sponsored proposals to change corporate

governance structures. Outside blockholders and institutions (when institutional holdings are relatively

concentrated) tend to align with the proposal sponsor, while insiders and outside directors who hold

significant stock positions tend to align strategically with management, who often oppose the shareholder-

sponsored proposals. Nesbitt (1994) finds that firms targeted by the California Public Employees’

Retirement System (CalPERS) experience positive long-run stock returns, and Opler and Sokobin (1997)

find that firms experience above-market performance the year after being targeted by the Council of

Institutional Investors. These results suggest that blockholders and active institutional shareholders lead to

more efficient monitoring of management and less managerial opportunistic behavior, which benefits all

stakeholders. We characterize this as the “management disciplining” role of governance and we predict a

8
positive relation between our institutional ownership and blockholder measures (described below) and

credit ratings.

A competing view in the literature suggests that concentrated ownership allows these shareholders to

exercise undue influence over management to secure benefits that are detrimental to minority

shareholders and bondholders (Shleifer and Vishny, 1997 and Bhojraj and Sengupta 2003). Examples

include greenmail and targeted share repurchases (Dann and DeAngelo 1983). Under the “wealth

redistribution” hypothesis, as the percentage of shares held by institutions increases or as the number of

blockholders increases, the likelihood of these shareholders using their influence to affect wealth transfers

from bondholders increases, suggesting a negative relation between credit ratings and our institutional

ownership and blockholder variables.

We capture the ownership effects of governance with three variables. BLOCK is the number of

outside blockholders that own 5% or more of a firm’s outstanding voting stock.6 %INST measures the

percentage of shares held by institutional investors. The relation between these two ownership structure

variables and firm credit ratings depends on whether these ownership concentrations, on average, are

beneficial to bondholders (management disciplining hypothesis) or further the interests of shareholders at

the expense of bondholders (wealth redistribution hypothesis). Because we have no way of predicting, a

priori, which effect is likely to dominate, we leave the prediction on these two variables unsigned. The

third variable, %INSIDE, is the percentage of shares held by officers or directors.7 We predict that

%INSIDE will be negatively related to RATING under the assumption that insiders will use their voting

power to expropriate firm resources for their personal benefit or resist shareholder-sponsored proposals to

6
An alternative construct to capture the power of significant ownership is to use the percentage of shares held by the
largest shareholder. Board Analyst has a variable labeled dominant shareholder, which reflects whether the firm has
a shareholder holding a significant proportion of shares. There are 146 of our 894 sample firms that have a
dominant shareholder owning more than 10% of the outstanding shares. When we estimate our model that includes
a dummy variable that captures firms that have a dominant shareholder, we find the coefficient on the dominant
shareholder variable to be insignificant.
7
Although this measure includes holdings by both officers and directors, the vast majority of %INSIDE is made up
of officer shareholdings. Thus, we expect this measure to largely proxy for managements’ self-interests rather than
board member incentives to monitor the actions of management.

9
increase the monitoring of their actions (Gordon and Pound, 1993), both of which are likely to lead to

greater agency risks for bondholders. In addition, we predict a negative relation between %INSIDE and

RATING because increasing insider ownership results in stronger incentives for officers and managers, as

residual claimants, to invest in projects that have very high returns when successful but low probabilities

of success (Jensen and Meckling, 1976)—that is, projects that increase bondholders’ risk due to the

differential payoff structure between bondholders and shareholders (wealth redistribution hypothesis).

III.2 Financial Stakeholder Rights and Relations

Financial stakeholder rights reflect the balance of power between stakeholders (bondholders and

shareholders) and management. A key element of this dimension of corporate governance is whether the

firm maintains a level playing field for corporate control and whether it is open to changes in

management and ownership that provide increased shareholder value. Gompers, Ishii and Metrick (2003)

construct an index based on 24 governance provisions, referred to as the G_SCORE, to measure the

power-sharing relationship between investors and management. The 24 provisions are classified into five

categories of management power: (1) tactics for delaying hostile bids; (2) voting rights; (3)

director/officer protection; (4) other takeover defenses; and (5) state takeover laws. Each firm’s

G_SCORE is the sum of points, where one point is awarded for the presence of each governance

provision. Thus, a higher G_SCORE indicates lower shareholder rights and greater management power.8

We use the Gompers, et al. (2003) G_SCORE metric to proxy for the stakeholder rights component of

governance.

Takeover defenses and other restrictions of shareholder rights like staggered terms of directors,

supermajority voting requirements for approval of mergers and ownership changes, and limits on

shareholders’ ability to meet and act places more power in the hands of management vis-à-vis

shareholders and can make it difficult to remove entrenched management that is acting opportunistically.

8
Using a sample of 1500 firms during the 1990s, Gompers, et al. (2003) find that taking a long position in firms
with the strongest shareholder rights and a short position in firms with the weakest shareholder rights yields an
average abnormal return of 8.5 percent per year.

10
Consistent with the management disciplining hypothesis, governance mechanisms tilted in favor of

opportunistic management can lower overall firm value, resulting in losses to both shareholders and

bondholders. This line of reasoning leads to a predicted negative relation between G_SCORE and credit

ratings. That is, firms with higher G_SCOREs (less shareholder power) are expected to exhibit lower

credit ratings because it is more difficult or costly to remove management that is acting opportunistically

when managers have greater power vis-à-vis shareholders. Alternatively, we may find a negative relation

between G_SCORE and credit ratings because firms with stronger shareholder rights (lower G_SCORES)

are likely to provide better monitoring and control over management leading to more effective and

efficient managerial decision-making, which in turn leads to better overall firm credit worthiness and

higher credit ratings.9

Giving greater power to shareholders to determine changes in management as well as ownership

control does not necessarily make bondholders better off (FitchRatings, 2004). For example, Asquith and

Wizman (1990) and Warga and Welch (1993) find that pre-buyout bondholders suffer significant wealth

losses in leveraged buyouts. These results suggest that bondholders do not always benefit under all

takeover scenarios. Therefore, governance mechanisms that place greater power in the hands of

shareholders may actually be viewed negatively by bondholders and credit rating agencies because it

increases the likelihood of ownership changes that can transfer wealth from bondholders to stockholders.

This line of reasoning predicts a positive relation between g-scores and credit ratings. That is, firms with

higher g-scores (weaker shareholder rights) are expected to have higher credit ratings. Consistent with

this conjecture, Klock, Mansi and Maxwell (2004) find that firms with stronger anti-takeover provisions

(weaker shareholder rights) have a lower cost of debt financing relative to firms with weaker anti-

takeover provisions.

9
Consistent with this conjecture, Gompers et al. (2003) find that firms with lower g-scores have higher firm values,
higher profits, higher sales growth, lower capital expenditures and lower corporate acquisitions. These factors
should lower firms’ credit risk leading to higher credit ratings. That is, firms with stronger shareholder rights are
expected to have higher credit ratings because of lower risk characteristics. Controlling for these additional risk
characteristics does not change the sign or significance of our g-score variable reported below.

11
 Given the mixed evidence on whether greater shareholder power translates into benefits for

bondholders (management disciplining hypothesis) or greater risk due to potential wealth transfers

between bondholders and shareholders, we make no directional prediction for the g-score variable.

III.3 Financial Transparency

Transparent financial reporting is critical to reducing the information asymmetry between the firm

and its capital suppliers. We posit that greater financial transparency facilitates the monitoring of

management actions and makes it less likely that management will act opportunistically. Sengupta (1998)

conjectures that firms with more timely and informative disclosures are perceived to have a lower

likelihood of withholding value-relevant unfavorable information. As a result, they are expected to be

charged a lower risk premium by creditors. Consistent with this prediction, he finds that firms with

higher AIMR disclosure ratings enjoy a lower effective interest cost of issuing new debt. As AIMR

disclosure ratings are no longer available, we use the quality of firms’ working capital accruals, WCAQ,

and the timeliness of firms’ earnings, TIMELINESS, to capture the transparency of firms’ financial

reporting. We describe the measurement of WCAQ and TIMELINESS in detail in Section IV. Briefly,

WCAQ is based on the standard deviation of the firm-specific residual from regressing working capital

accruals on past, contemporaneous, and future operating cash flows, where smaller residuals reflect a

better mapping of working capital accruals to cash flows (Dechow and Dichev, 2002). TIMELINESS is

the squared residual from regressing returns on earnings allowing for separate intercepts and slopes for

profit and loss firms (Gu, 2002). Earnings that better articulate with market returns are deemed to be more

transparent and timely in that they better reflect the economic events that are priced by the market. A

high squared residual indicates that earnings are less transparent/timely. To facilitate the interpretation of

these variables, we multiply the variables by negative one and predict a positive relation with firms’ credit

ratings.

The reliability of financial information is due, in part, to the quality and integrity of the audit process.

To proxy for the quality and integrity of the audit process, we use three measures: (1) the total fees (audit

plus non-audit) charged to the client firm divided by the total revenues of the audit firm (TOTFEES); (2)

12
%AUD_IND is the percentage of the audit committee made up of outside independent directors; and (3) a

dummy variable, FIN_EXPERT coded one if the firm’s audit committee has at least one individual

deemed to be a “financial expert,” and zero otherwise. Using the attributes of a financial expert set forth

by the Securities and Exchange Commission (SEC, 2003) this variable is coded one if the audit

committee has an outside independent director that is a CPA or who has experience as a chief financial

officer of another company.10

DeAngelo (1981) posits that auditor independence is threatened as the economic bond between the

auditor and client firm increases. Concern over economic bonding between the client firm and its auditor

was the major impetus behind the restrictions that Sarbanes-Oxley placed on the kinds of nonaudit

services that auditing firms can perform for their clients (U.S. Congress 2002). However, the evidence on

whether economic bonding between the audit firm and its client impairs auditor independence as proxied

by abnormal accruals is mixed. While Frankel, Johnson and Nelson (2002) find evidence consistent with

economic bonding impairing auditor independence, Ashbaugh, LaFond and Mayhew (2003), Chung and

Kallapur (2003) and DeFond, Raghunandan and Subramanyam (2002) do not find evidence of

independence impairment. We use our TOTFEES variable to measure the economic bonding between the

audit firm and its client.11 If credit rating agencies perceive that auditor independence, and thus the quality

of financial statements is impaired due to economic bonding, then we expect a negative relation between

this variable and firms’ credit ratings. However, if credit rating agencies perceive that the economic bond

between auditors and their audit clients do not threaten the quality of firms’ financial reporting, as some

10
The SEC recently adopted this provision of the Sarbanes-Oxley Act (SEC 2003) and defined an “audit committee
financial expert” to mean a person who has the following attributes:
(1) An understanding of financial statements and generally accepted accounting principles;
(2) An ability to assess the general application of such principles in connection with the accounting for estimates,
accruals and reserves;
(3) Experience preparing, auditing, analyzing or evaluating financial statements that present a breadth and level of
complexity of accounting issues that are generally comparable to the breadth and complexity of issues that can
reasonably be expected to be raised by the registrant’s financial statements, or experience actively supervising one
or more persons engaged in such activities;
(4) An understanding of internal controls and procedures for financial reporting; and
(5) An understanding of audit committee functions.
11
We consider alternative ways of measuring this construct in the “sensitivity analysis” section below.

13
of the studies noted above indicate, then we expect to find no relation between TOTFEES and credit

ratings.

The conventional wisdom is that audit committees more effectively carry out their oversight of the

financial reporting process if they include a strong base of independent outside directors. Klein (2002)

provides evidence to support this assertion. She finds a negative relation between audit committee

independence and abnormal accruals. She also reports that as audit committee independence declines,

abnormal accruals increase. To the extent that better monitoring of the financial reporting process leads to

less managerial opportunism and better financial transparency, this should lead to lower default risk for

bondholders. Accordingly, under the management disciplining hypothesis we predict a positive relation

between %AUD_IND and credit rating.

Several recent studies investigate whether including independent directors with financial expertise on

the audit committee is likely to improve the quality and integrity of a firm’s financial statements. Klein

(2003) finds that placing independent directors with professional certifications in accounting or law or

prior banking experience on the audit committee reduces earnings management. Agrawal and Chadha

(2003) provide corroborating evidence and find that companies whose boards or audit committees have an

independent director with a background in accounting or finance face a lower probability of earnings

restatements. DeFond, Hann and Hu (2004) and Davidson, Xie and Xu (2004) find a positive market

reaction to firms’ announcements that an accounting financial expert has been appointed to the audit

committee. To the extent that having a financial expert on the audit committee improves board

effectiveness and enhances the integrity of the financial reporting process, we predict this will lead to less

managerial opportunism. Accordingly, under the management disciplining hypothesis we predict a

positive relation between FIN_EXPERT and credit ratings.

III.4 Board Structure and Processes

This component of corporate governance deals with such things as: (1) board size and composition in

terms of proportion of inside, outside and affiliated directors; (2) board leadership and committee

structure; (3) how competent and engaged board members are; (4) whether there are a sufficient number

14
of outside independent directors on the board that represent the interests of all stakeholders, and how

those members are distributed across the various committees; and (5) whether board members are

remunerated and motivated in ways that ensure the long-term success of the company.

The first three elements address the board’s role and ability to provide independent oversight of

management performance and hold management accountable to stakeholders for its actions. Boards often

delegate oversight of key functions or decision making to standing committees—e.g., audit,

compensation, nominating or governance, finance and investment. These committees, made up of subsets

of board members, meet separately from the full board and generally have specific, narrowly defined

functions.

Prior research generally posits a positive relation between board and committee independence and

firm performance. Better firm performance should benefit all stakeholders leading to higher credit

ratings. However, research findings on the relation between board and committee composition and overall

firm performance are mixed. Baysinger and Butler (1985) and Hermalin and Weisbach (1991) find no

significant association between the percentage of outsiders on the board and same-year measures of

corporate performance. Bhagat and Black (2000) find no relation between overall board independence

and four measures of firm performance (Tobins’ Q, return on assets, market adjusted stock returns and

ratio of sales to assets) measured over a three year window. Agrawal and Knoeber (1996) investigate the

relation between firm performance (Tobin’s Q) and seven control mechanisms including percentage of

non-officer board members. Using a simultaneous equations framework to control for the

interdependence among the various control mechanisms, Agrawal and Knoeber find a significant negative

relation between outside membership on the board and firm performance, leading them to conclude that

boards seem to have too many outsiders.

Klein (1998) extends the previous research on board composition and firm performance by examining

the relation between the composition of the overall board and of various committees and firm

performance. Consistent with prior evidence, Klein finds no association between firm performance and

overall board composition. Moreover, she finds no association between the level of independence on

15
audit, compensation and nominating committees and firm performance. Interestingly, she does find a

significant positive association between the percentage of inside directors on finance and investment

committees and accounting and stock market performance measures. One explanation for this result is

that inside board members bring specialized institutional and industry-specific knowledge to the table that

helps these committees select long-term investment and financing strategies that enhance firm value.

Thus, inside board members appear to serve a useful role in overall corporate governance if strategically

placed on committees that have more of an operating focus than a monitoring focus.

Finally, and more germane to bondholder interests, Bhojraj and Sengupta (2003) posit that firms with

a greater proportion of outside directors on the board provide better monitoring of management actions.

Thus, bondholders face less agency risks vis-à-vis management. They posit that this, in turn, will lead to

higher bond ratings and lower debt yields. Consistent with this conjecture, they find that firms with a

higher proportion of nonofficer directors enjoy lower bond yields and higher ratings on new bond issues.

Based on the literature reviewed above, we use %BRD_IND to measure the percentage of board

made up of independent outside (nonaffiliated) directors. In articulating its core governance principles for

protecting bondholders, FitchRatings (2004) notes:

“Assessing a company’s governance practices begins with its board of directors.

An independent, active, and committed board of directors is an essential element of a
robust governance framework. A board that is not committed to fulfilling its
fiduciary responsibilities can open the door for ineffective, incompetent, and, in some
cases, unscrupulous management behavior.” (p. 5).

Consistent with this view and the literature reviewed above, we expect a positive relation between

%BRD_IND and credit ratings.

Imhoff (2003) argues that board governance is severely compromised when the current or former

CEO of the company also serves as chairman of the board. This is because the board chairman frequently

sets the board’s agenda and, therefore, controls issues brought before the board. Moreover, CEOs that

serve as board chair frequently have significant influence on the slate of candidates for board seats,

thereby increasing the risk that new board appointees will not be independent of management even

16
though they are “outsiders”. CEOs can also exert significant influence over the board through the

committees they serve on. We use CEOPOWER as a composite measure of the influence that the CEO

exercises over the board. A firm receives one point if the CEO is chairman of the board and one point for

each committee that the CEO serves on. We posit that CEO power on the board is likely to reduce the

board’s disciplining of opportunistic management. Accordingly, under the management disciplining

hypothesis, we predict CEOPOWER will be negatively related to credit ratings.

Ceteris paribus, we expect that boards comprised of members who are more competent or

knowledgeable will do a better job of monitoring the activities of management and make better decisions

leading to less default risk. Similar to Klein (1998), we measure board competency or expertise by the

percentage of outside board members that sit on boards of other companies (%BRD_EXPERT). We

predict a positive relation between this variable and credit ratings.

Board compensation is another element of the ‘Board Structure and Process’ component of

governance. Key issues are whether board members are remunerated and motivated in ways that ensure

the long-term success of the company. Jensen (1993) argues that boards with greater ownership in the

firm are more likely to do a better job of monitoring management and fulfilling their fiduciary

responsibilities. Consistent with this conjecture, Yermack (2003) finds that director stock and option

awards are positively related to firms’ investment opportunities and subsequent firm performance.

Yermack shows that tying directors’ pay more closely to stock performance through the use of options

and other equity awards generally leads to increased performance. We use %BRD_STOCK to measure

the percentage of directors that hold stock in the company and predict a positive relation between this

variable and credit ratings.

Recently the SEC endorsed the proposals of the NYSE and NASDAQ that firms adopt a formal

governance policy that outlines the roles and responsibilities of directors and establishes an explicit code

of business conduct and ethics for directors (SEC, 2003). We expect that having such a formal

governance policy places increased responsibility on board members and increases their legal liability

leading to greater attentiveness on the part of board members. Greater board attentiveness should lead to

17
better monitoring of management which, in turn, should lead to less management opportunism. We code

GOVERNANCE_POLICY with a one if a firm has a formal governance policy, and zero otherwise, and

predict a positive relation between this variable and credit ratings under the management disciplining

hypothesis.

Finally, we use %FINCOM_INSIDE to measure the percentage of insiders on finance committees.12

Based on Klein’s (1998) results that having insiders on finance and/or investment committees improves

firm performance, we expect this committee structure variable to be positively related to credit ratings

since improved firm performance is expected to improve a firm’s creditworthiness.

IV. Sample, Variables and Descriptive Statistics

IV.1. Sample and Data Sources

Data for this study are compiled from four sources:

• Governance measures, audit/non-audit fees and share ownership data – Board Analyst data base

and firm proxy statements

• G_SCORES – Gompers, et al. (2003)

• Credit ratings and accounting variables – Standard and Poor’s Compustat

• Stock return data- CRSP

We obtain the majority of the corporate governance measures from the Board Analyst data base

compiled by The Corporate Library, an independent research firm that provides data and analysis of

corporate governance issues.13 This data base contains detailed governance, audit/non-audit fee data and

stock ownership data (including institutional and inside ownership) for over 2000 U.S. companies and

profiles on over 22,000 individual directors. The data used in our primary analysis are from the 2003

proxy season covering the board and committee structures of firms for the 2002 fiscal year.

12
For those firms without finance committees, we used the percentage of insiders on the overall board for this
variable because, in the absence of a finance committee, the overall board would be charged with voting on financial
policy matters (see Klein (1998) for similar treatment).
13
Board Analyst does not provide information on finance and investment committees. This information was hand-
gathered from 2003 proxy statements.

18
 G_SCORES that measure the power-sharing relationship between investors and management were

obtained from Gompers, et al. (2003). These G_SCORES are available for approximately 1500 firms and

are based on the incidence of 24 governance provisions related to shareholder rights and take-over

defenses found in 2002 proxy statements.

For firm credit ratings (RATING) we use the long-term issuer credit ratings compiled by Standard &

Poor’s and reported on Compustat (data item 280). The ratings range from AAA (highest rating) to D

(lowest rating—debt in payment default). These ratings reflect S&P’s assessment of the creditworthiness

of the obligor with respect to its senior debt obligations. For purposes of our analysis, the multiple ratings

are collapsed into seven categories according to the schedule provided in Table 1. To facilitate the

discussion of the economic significance of our results, we also estimate our logistic regression model

using a two category classification scheme—investment grade and speculative grade. The assignment of

the credit rating groups into these two classifications are also shown in Table 1.

[Insert Table 1 here]

Table 2, Panel A summarizes the sample selection procedure and number of firms lost because of

minimum data requirements from each data source. Essentially, our final sample for the credit rating

analysis is determined by the intersection of firms for which required data are available on the four data

sources noted above.14

[Insert Table 2 here]

Panel B of Table 2 provides details on board and committee composition for our sample firms. Out of

894 sample firms, all have audit committees, 99.6% (890) have compensation committees, 90.9% (813)

have nominating committees, but only 27.1% (242) have finance committees. The average board

(committee) size is 10 (4) directors. The incidence of firms having at least one insider on a committee

ranges from 0.7% (6 / 894) for audit committees to 30.2% (73 / 242) for finance committees. The much

14
In general, our sample firms are larger than the average firm on Compustat with sample means of assets, sales,
market value of equity, and long-term debt (in millions) of $20,765, $7,502, $8,982, and $4,021, respectively. In
addition, 84%, 15% and 1% of the sample firms’ shares trade on the New York Stock Exchange, NASDAQ, and the
American Stock Exchange, respectively.

19
higher incidence of insiders on finance committees presumably reflects the fact that insiders bring

valuable institutional and industry-specific knowledge and expertise to this committee (Klein, 1998).

Roughly 73% of our sample firms have CEOs that serve as Chairman of the Board, and the more common

committees that CEOs serve on are the nominating and finance committees.

IV.2. Independent Variables

Corporate Governance Measures

The variables identified in Section III that we use to capture key governance attributes within the

S&P framework are summarized in Panel A of Table 3 along with their predicted relation with RATING.

Except for our measures of financial transparency, the variable measurements were described in detail in

Section III when introduced, so we do not take time to repeat them here.

[Insert Table 3 here]

Our first measure of financial transparency, working capital accrual quality (WCAQ), is an

accounting based measure of financial reporting quality based on the work of Dechow and Dichev (2002).

Dechow and Dichev (2002) derive a measure of working capital accruals quality that is a function of how

well working capital accruals map into cash flows. The calculation of WCAQ is a two-step process

beginning with the cross-sectional estimation of the following OLS model:

WCAt = β 0 + β 1CFOt −1 + β 2 CFOt + β 3 CFOt +1 + ε (1)

where

WCA=working capital accruals (-(Compustat # 302+ Compustat # 303+ Compustat # 304+ Compustat #
305+ Compustat # 307)) scaled by average total assets (Compustat # 6),
CFO= cash flow from operations (Compustat # 308) scaled by average total assets.

We estimate regressions by three, two, or one-digit SIC codes conditional on having at least ten firms in

each SIC group. WCAQ is the standard deviation of the firm-specific residual from the prior three to five

years. Firms with smaller WCAQ values report working capital accruals that better map into cash flows,

i.e., are considered to have higher quality working capital accruals. To facilitate discussion of our results,

20
we multiply WCAQ by negative one. Thus, larger WCAQ values imply higher quality working capital

accruals.

Our second measure of financial transparency, TIMELINESS, is derived from the following

regression equation:

RETit = β 0 + β1 NIBEit + β 2 LOSS it + β 3 NIBEit * LOSS it + β 4 ∆NIBEit + ε it (2)

where

RETit = the market adjusted return for firm i over fiscal year t (from CRSP),
NIBEit = net income before extraordinary items (Compustat # 18) scaled by beginning of period market
value of equity for firm i in period t (Compustat # 25 * Compustat # 199),
LOSSit = one if NIBE is negative, zero otherwise,
∆NIBEit = the change in net income before extraordinary items (Compustat # 18) scaled by beginning of
period market value of equity (Compustat # 25* Compustat # 199),
NIBEit* LOSSit = interaction term that allows for a differential market reaction for loss versus profit firms.

We estimate the above regression cross-sectionally within one, two and three digit SIC codes requiring a

minimum of ten firms in each industry grouping.

Gu (2002) argues that the squared residuals from equation (2) can be conveniently interpreted as the

degree of price movement (returns) that is not explained by contemporaneous accounting earnings.

Higher squared residuals imply less timely earnings. To facilitate interpretation of our results, we multiply

this measure by negative one. Thus, larger (less negative) values imply more timely earnings. One can

think about financial statement transparency as encompassing the relevance and reliability dimensions of

accounting information. The more information about the firm’s current economic activities that is

embedded in current earnings and the more precise that information (i.e., the more relevant and reliable it

is), then the more transparent the economic activities of a company is to its stakeholders. Higher quality,

more transparent earnings information means less information asymmetry between the firm and its

bondholders, leading to less uncertainty about default risk which, in turn, should lead to higher credit

ratings. Barth and Landsman (2003) provide empirical support for this claim in that they find that firms

with more value relevant earnings enjoy a lower cost of debt. We use the Gu measure as one proxy for

21
financial transparency as it captures both the timeliness of firms’ financial information and relevance of

the financial information for assessing firms’ current economic conditions. 15

Control Variables—Firm Characteristics

Additional firm-specific explanatory variables are included in the RATING models based on a survey

of prior research on the determinants of corporate bond ratings (e.g., Horrigan 1966, Kaplan and Urwitz

1979, Boardman and McEnally 1981, Lamy and Thompson 1988, and Ziebart and Rieter 1992). The

accounting-based ratios of debt-to-assets (LEV), return-on-assets (ROA), and interest coverage

(INT_COV) are used to proxy for firms’ default risk, where higher LEV and lower ROA and INT_COV

values reflect greater default risk. We use LOSS, a categorical variable set equal to one if the firm reports

negative earnings in the current and prior fiscal year, as another control for default risk given that the

likelihood of default is greater for firms that are unprofitable.

Firm size, SIZE, is included as a control variable because larger firms face lower risk, and thus are

expected to have higher credit ratings. We also control for differences in firms’ debt structure by

including SUBORD, which is coded one if the firm has subordinated debt. The debt structure of a firm

with subordinated debt is considered to more risky due to the differential claims to assets by debt

providers. Firms’ capital intensity (CAP_INTEN) is included in the model to control for differences in

firms’ asset structure, where firms with greater capital intensity present lower risk to debt providers, and

thus are expected to have higher credit ratings. Finally, FIN_UTILITY is set equal to one if a firm is a

financial institution or utility firm, (zero for firms not in these two industries) to control for lower default

risk for firms operating in regulated industries.

The specific measurements of the variables capturing differences in firm characteristic along with

their predicted relation with RATING are summarized in Table 3.

Descriptive Statistics

15
To validate this construct, we regress the AIMR disclosure ratings on TIMELINESS for the same periods. The
coefficient is 13.26 with a t-statistic of 3.89. Thus, this measure of financial transparency and AIMR disclosure
rankings appear to be capturing similar constructs.

22
 Table 4 presents descriptive statistics for the various governance and firm-characteristic variables.

Within the “Ownership Structure and Influence” component of governance, we find the average (median)

number of blockholders that own 5% or more or the firm’s stock is 4.5 (4.0). The average (median)

percentage of shares held by institutional investors is 63% (68%) while the average (median) percentage

of shares held by insiders (officers and directors) is 8% (4%). For the “Financial Stakeholder Rights”

dimension of corporate governance, the average G_SCORE of our sample firms is 9.59, which is similar

to the mean G_SCORE reported by Gompers, et al., of 9.15. Sixty-one of our sample firms (6.8%) fall

into Gompers et al.’s dictatorship portfolio (G_SCORES > 13 indicating greater management power)

while forty-two firms (4.7%) fall into their democracy portfolio (G_SCORES < 6 indicating greater

shareholder rights).

[Insert Table 4 here]

Turning to the “Financial Transparency” dimension, the mean (median) WCAQ and TIMELINESS

value is -0.04 (-0.03) and -0.10 (-0.03), respectively. Recall the larger (i.e., less negative) values of these

variables reflect higher quality financial reporting. The measure of economic bonding between the firm

and its auditor is our TOTFEES variable.16 Because of its small magnitude, we multiply this variable by

100. Before this scaling adjustment, the median firm’s total fees paid to its auditor amount to only .04%

of the audit firm’s total revenues. Ninety-two percent of the average firm’s audit committee is comprised

of outside independent board members, with over three-quarters of the sample firms having 100%

independent audit committees. The little variation in %AUD_IND is explained by the fact that the listing

rules of the New York Stock Exchange and NASDAQ require listed firms to have audit committees

comprised of at least three independent directors. There are, however, exceptions to the rules that result

in not all firms having 100 percent independent audit committees (see Klein (2003) for an overview).

Finally, 26% of our sample firms have an outside financial expert (CPA or CFO) serving on their audit

committee.

16
Recall we measure this as total fees (audit and non-audit) paid by the client divided by the audit firm’s total
revenue.

23
 Within the “Board Structure and Process’ dimension, the descriptive statistics indicate that the

average (median) percentage of outsiders on the board is 70% (73%) and the lower quartile value is 58%.

Consistent with the evidence in Table 2, the majority of our sample firms have CEOs that also serve as

Chairman of the Board or on other board committees. On average, 36% of outside directors serve on other

boards and 87% of the directors hold stock in the company. Forty-two percent of the sample firms have a

formal governance policy. The average percentage of outsiders on the compensation (nominating)

committee is 90% (79%), while the average percentage of insiders on the finance committee is 15%.17

For brevity, we do not take time to describe the summary statistics for the firm characteristic

variables. Turning to the dependent variables, we note that the median credit rating is 4.0 implying a debt

rating in the BBB+ to BBB- range, and that sixty-three percent of our sample firms have an investment

grade credit rating.

Table 5 presents correlations among the firm characteristic variables (Panel A) and governance

variables (Panel B) and between these variables and credit ratings. The upper right hand portion of each

panel presents Pearson product-moment correlations while the lower left hand portion presents the

Spearman rank-order correlations. In Panel A, the simple correlations between each of the firm

characteristics and our RATING variable are in the predicted directions and are statistically significant at

the .01 level or below except for the capital intensity variable (CAP_INTEN), which is negative and

insignificant. Specifically, we find that ROA, INT_COV, SIZE and FIN_UTILITY are significantly

positively correlated with credit ratings whereas LEV, LOSS, and SUBORD are significantly negatively

correlated with ratings. Not surprisingly, several of the firm characteristic variables exhibit high

intercorrelations.

[Insert Table 5 here]

17
Recall that in coding this variable, if a firm does not have a standing finance committee, we used the percentage of
insiders on the overall board for %FINCOM_INSIDE because the board de facto votes on all major financing
decisions in the absence of a finance committee. This explains why the percentage of insiders on this committee
appears to be smaller than the numbers imply in Table 2.

24
 Panel B of Table 5 presents the correlations between the various governance variables and between

these variables and RATING. Fourteen of the sixteen governance variables exhibit Pearson correlations

with the RATING variable that are significant at .01 or below. The correlations among the various

governance variables generally fall below .30 except for the board and committee independence measures

(shown in shaded cells) which are generally in the .38 to .56 range. The high intercorrelations between the

committee and board independence measures are to be expected because the committees are drawn from

the board membership. Because of these high correlations, we include only the board and audit committee

independence measures in our logit model.

V. Empirical Tests and Results

V.1 Ordered Logit Results

Our empirical tests are derived from a general model that represents credit ratings as a function of

corporate governance components and firm characteristics.

Credit rating = f (corporate governance components, firm characteristics).

To test the predicted relations between corporate governance components and credit ratings, we

estimate a series of ordered logit models. We use ordered logit models because the seven categories of

credit ratings convey ordinal risk assessments; we can rank order firms’ preferences across the rating

categories but cannot assume uniform differences in benefits (costs) between the categories.

[Insert Table 6 here]

We begin by estimating the model using only the firm characteristic variables to provide a benchmark

from which to assess the incremental effect of various corporate governance mechanisms on credit

ratings. The benchmark results are reported in the Model 1 column of Table 6. All of the estimated

coefficients on the firm characteristics have the expected sign and are significant at the 0.01 level or

better. The results document that credit ratings are positively related to ROA, INT_COV, SIZE,

CAP_INTEN, and negatively related to LEV, LOSS and SUBORD. We also document that utilities and

financial institutions are likely to have better credit ratings. The benchmark model yields a Likelihood

ratio χ2 of 660.55, which is significant at the .01 level, and has a generalized R-square of 52 percent.

25
 The remaining columns of Table 6 report the results of testing whether the various components of

corporate governance within the S&P framework are associated with firms’ credit ratings. In Column 2

of Table 6, we report the results of estimating the model incorporating the “Ownership Structure and

Influence” component of corporate governance. We find a significant negative coefficient on BLOCK, a

significant positive coefficient on %INST, and a marginally significant negative coefficient (p < .10) on

%INSIDE. The significant positive coefficient on %INST is consistent with our “management

disciplining” hypothesis the conjectures that institutional investors contribute to more efficient monitoring

of management and that the benefits of better monitoring are shared by all stakeholders. The negative

coefficient on BLOCK indicates that firms with a larger number of blockholders have lower credit

ratings. This finding is consistent with our “wealth redistribution” hypothesis that blockholders, as

influential shareholders, can exercise influence on management to secure benefits that are detrimental to

bondholders. This result also corroborates the findings of Bhojraj and Sengupta (2003) who document

that blockholders have an adverse impact on bond ratings. The marginally negative coefficient on

%INSIDE implies that inside ownership adversely affects bondholders providing additional evidence

supporting our “wealth redistribution” hypothesis. The Wald χ2 of 39.64 (significant at .01) indicates that

the addition of the ownership structure and influence variables, as a group, add significant incremental

explanatory power to the benchmark credit rating model.

The results of estimating the model using G_SCORE as our proxy for “Financial Stakeholder Rights

and Relations” are reported in the Model 3 column of Table 6. We find a positive and highly significant

coefficient on G_SCORE and the Wald χ2 of 13.47 is statistically significant at the .01 level. Recall that

the smaller the G_SCORE, the stronger the shareholder rights. Our results suggest that stronger

shareholder rights are associated with lower firm credit ratings. Gompers, et al. (2003) find that firms

with stronger shareholder rights have higher firm value and higher profits. Thus, this finding provides

strong support for the “wealth redistribution” hypothesis that posits that bondholders may suffer potential

wealth transfer effects associated with stronger shareholder rights and that from the bondholders’

perspective the risks of wealth transfer outweigh the positive firm value effects documented in Gompers,

26
et al. (2003). Our finding of a positive association between G_SCORE and credit ratings is consistent

with the work of Asquith and Wizman (1990) and Warga and Welch (1993) who find that certain kinds of

ownership changes can result in significant wealth transfers from bondholders to shareholders. Our results

are also consistent with Klock, et al. (2004) who find that firms with stronger anti-takeover provisions

(weaker shareholder rights) have a lower cost of debt financing relative to firms with weaker anti-

takeover provisions.

The Model 4 column of Table 6 displays the results from estimating the credit rating model using the

“Financial Transparency” variables after controlling for firm characteristics. As predicted, we find a

positive relation between RATING and the transparency of firms’ financial reporting as measured by

WCAQ and TIMELINESS. We also find evidence that the quality of the audit process affects a firm’s

credit rating in that firms having more independent directors serving on their audit committees and having

an independent financial expert on the audit committee have better credit ratings. We fail to find a

significant association between TOTFEES and RATING. The Wald χ2 of 77.16 indicates that the

variables comprising the financial transparency and information disclosure component significantly

improve the explanatory power of the RATING model.

The results of investigating whether the “Board Structure and Processes” component of corporate

governance affects credit ratings are reported in the Model 5 column of Table 6. In general, the estimated

coefficients on the “Board Structure and Processes” variables support our “management disciplining”

hypothesis that better board structure and processes contribute to more efficient monitoring of

management that protects all stakeholders’ interests. As predicted, we find a positive coefficient on

%BRD_IND, which indicates that the greater the board’s ability to provide independent oversight of

management the better the credit rating. This result is consistent with Bhojraj and Sengupta (2003) who

find that firms with a greater proportion of independent outside directors on the board have higher bond

ratings. The positive coefficient on %BRD_EXPERT indicates that when a greater proportion of the

board is comprised of knowledgeable individuals, as proxied by their service to other boards, the higher

the firm credit rating. We also document a positive relation between %BRD_STOCK and RATING. This

27
result indicates that as more members of the board have an equity stake in the firm, they have greater

incentives to restrict managerial opportunism or to monitor management decision making leading to

lower default risk. Finally, the documented positive coefficient on GOVERNANCE_POLICY suggests

that firms receive benefits in the form of better credit ratings by having formal governance policies.

Overall, the Wald χ2 of 42.66 indicates that the board structure and processes component is a significant

determinant of firms’ credit ratings.

The last column of Table 6 reports the full model, where we jointly test whether the four components

of the S&P corporate governance framework are associated with firms’ credit ratings. The model is

highly significant with a Wald χ2 of 139.93. While the coefficients on the firm characteristic variables

remain significant and in the predicted relation to credit ratings with the exception of FIN_UTILITY, the

results indicate that within each component of S&P governance framework, there appears to be at least

one dominant governance mechanism that affects firms’ credit ratings. Specifically, after incorporating

all four components of governance into the RATING model, we find BLOCK, G_SCORE, WCAQ, and

TIMELINESS are governance attributes that are significant determinants of credit ratings. In addition,

we find that four of the six governance provisions related to board structure and processes are significant.

Specifically, we find the coefficients on %BRD_IND, %BRD_EXP, and %BRD_STOCK to be positive

and significant at conventional levels. We also find a marginally significant negative coefficient on

CEOPOWER. This latter result suggests that it is costly for firms, in terms of default risk, to cede the

chief executive officer with too much board control.

V.2 Investment vs. Speculative Grade Analysis

As stated above, credit ratings convey ordinal risk assessments. Because of the difficulty in

quantifying the marginal effects of changes in each governance variable on credit ratings with multiple

categories, we use an alternative classification scheme that partitions credit ratings into two categories--

investment grade or speculative grade. Many bond portfolio managers are restricted from owning

speculative grade bonds (Grinblatt and Titman, 2002), and as such, firms incur significant costs if they

28
receive a speculative bond rating. Furthermore, using a dichotomous credit rating classification allows us

to more readily assess the economic impact of corporate governance on firms’ expected cost of debt.

Table 7 displays the results of estimating six logitistic regressions using INVESTMENT_GRADE as

the dependent variable, where INVESTMENT_GRADE is coded one if the firm’s credit rating is BBB- or

better, and zero otherwise. The results are similar to the results of the RATING analyses reported in

Table 6 with a few exceptions. First, the coefficient on INT_COV is insignificant in all of the

INVESTMENT_GRADE analyses whereas it was highly significant in all of the RATING analyses.

Second, when considering the financial transparency component of corporate governance in isolation, the

coefficients on %AUD_IND and FIN_EXPERT are not significant. Third, the coefficient on

%BRD_EXPERT is not significant in either Model 5 or Model 6. Finally, unlike the RATING analysis,

the coefficients on %BRD_IND and CEOPOWER are insignificant in the full model. When we estimate

the INVESTMENT_GRADE model that incorporates all four corporate governance components, we find

once again that BLOCK is negatively related to credit ratings and G_SCORE, WCAQ, TIMELINESS,

and %BRD_STOCK are positively related to credit ratings.

[Insert Table 7 here]

In order to provide some insight into the economic significance of our results, we calculate the change

in probability of receiving an investment grade credit rating as a result of changing the levels of various

corporate governance variables. The change in probability is calculated using the following steps. First,

we calculate the probability of achieving an investment grade credit rating from our logitistic regression

model using the following expression:

π ( X ) = e β ' X (1 + e β ' X ) (3)

where β is the vector of coefficients from Model 6 in Table 7 and X is the vector of independent variables

set equal to their mean values. Next, we calculate the marginal changes in the probability of a firm

receiving an investment grade credit rating as a result of a one unit change in each of our governance

variables. This marginal effect is measured by ∂π ( X ) / ∂xi = β iπ ( X )[ 1 − π ( X )] , which is again calculated at

29
the mean value of the regressors. These marginal effects are reported in column 3 of Table 8 for the

governance variables after standardizing each non-binary variable by its mean and dividing by its

standard deviation.18 The marginal effects measure the change in the probability of receiving an

investment grade rating for a one standardized unit change in each governance variable while holding the

firm characteristics at their mean values.

An alternative way of assessing the effect of various governance variables on the likelihood of

receiving an investment grade credit rating that is easier to interpret is to calculate the values of the logit

function, π ( X ) , at selected xi values such as their lower and upper quartiles (Agresti 2002, p. 167). This

entails substituting the quartile values for each xi explanatory variable into eqn. (2) while holding the

other variables constant at their means. The linear approximation to changes in π ( X ) is obtained by

multiplying the interquartile range of xi values (see Table 4 for the interquartile ranges) by the marginal

effects based on the unstandardized value of the variables (Agresti 2002, Chapter 5). These values are

reported in the last column of Table 8.

[Insert Table 8 here]

Moving from the first quartile to the third quartile of BLOCK decreases the probability of receiving

an investment grade credit rating by approximately .19. The change in probabilities for G_SCORE,

WCAQ, and TIMELINESS are approximately .04, .05, and .08, respectively, while the change in

probabilities for %BRD_IND and %BRD_STOCK are .048 and .052 respectively. Although the

probability changes due to any one governance variable may not appear to be all that dramatic, the

aggregate effect across all dimensions of corporate governance can be substantial.

To demonstrate this point, we first calculate the probability of receiving an investment grade credit

rating for a hypothetical firm that takes on the lower (upper) quartile values of governance variables that

are positively (negatively) related to credit ratings while holding all the firm-specific variables at their

18
We use standardized values because the various governance variables are measured in different units. Without
standardization the marginal probabilities are difficult to compare and interpret (Agresti 2002, Chapter 5).

30
mean values.19 This yields a probability of receiving an investment grade credit rating of .46. We next

repeat this process but now use upper (lower) quartile values of governance variables that are positively

(negatively) related to credit ratings. This yields a probability of receiving an investment grade credit

rating of .93. Thus, a firm approximately doubles its probability of receiving an investment grade credit

rating by implementing better governance along multiple dimensions.20 During the time frame of our

analysis, the average yield for firms with investment (speculative) grade debt with a ten year maturity was

approximately 6.00% (14.0%). This 800 basis-point spread translates into $38.4 million in before-tax

interest costs for the median firm in our investment grade sample.21 This cost estimate, however,

overstates the cost of weak governance from shareholders’ perspective when weak governance allows

wealth transfers from bondholders to shareholders can take place. To the extent that part of the interest

rate spread between investment grade and speculative grade debt is due to bondholders price-protecting

against potential wealth transfers between themselves and stockholders, the amount computed above

represents an upper bound on the cost of weak governance from the stockholder’s perspective.

Nevertheless, governance mechanisms that increase firms’ likelihood of receiving an investment grade

debt rating have significant implications for assessing debt financing costs.

VI. Additional Analyses

VI.1 Why Don’t All Firms Practice Good Governance?

The preceding analysis suggests that firms with weak governance have a lower probability of

receiving an investment grade credit rating and, as a consequence, incur significantly higher debt costs.

19
For governance attributes measured as a 0-1 dummy variable, the benchmark probability is determined with the
zero (one) value when the governance attribute is negatively (positively) related to credit ratings.
20
We hasten to note that this illustration does not reflect the typical firm in our sample because any given firm will
likely not start from a position of having weak governance (low quartile) along all of the multiple dimensions we
consider. Nor is it likely that any given firm would be able to move to a position of having strong governance along
all dimensions (upper quartile).
21
To calculate the incremental interest costs, we first determine the mean debt-to-asset ratio for our sample of
investment grade firms (0.28). We then multiply this ratio times the total assets of each speculative grade firm to
estimate the “as-if” capital structure if the speculative grade firm was an investment grade firm. Finally, we
multiply the estimated debt level times the 800-basis point spread to determine a speculative firm’s hypothetical
incremental interest cost from having weak governance.

31
This raises the question of why some firms are willing to bear additional debt financing costs by not

practicing good governance. One way to think about answering this question is to consider how

managers can appropriate some or all of the rents from outside stakeholders by resisting better

governance.

Evidence by Core, et al. (1999) suggests one way that managers can extract rents from weak

governance. They find that CEOs with greater power over the board or that are more entrenched earn

greater compensation after controlling for standard economic determinants of pay. Moreover, they find

that the estimated component of overcompensation is significantly negatively related with subsequent

firm operating and stock performance. Their results suggest that firms with weaker governance structures

exhibit greater overcompensation of CEOs and face greater agency problems. Generalizing the Core et al.

(1999) results to our research setting, it may be rational for managers to resist efforts to improve

governance as long as they receive more overcompensation relative to their share of increased debt costs

due to weaker governance. Moreover, various frictions in the market for corporate control such as anti-

takeover amendments, super-majority voting rules for ownership changes, poison pills and golden

parachutes contribute to management entrenchment and allow overcompensation and non-optimal

governance structures to persist.

To investigate whether overcompensation helps explain why some firms don’t seek stronger

governance, we model CEO pay in 2002 as:

CEO_PAY = f (economic determinants, board and ownership structure attributes).

Following Core et al. (1999), we measure CEO_PAY in three different ways: Salary, Salary+Bonus

and Total Compensation. The definitions of the alternative measures of CEO_PAY, the specific board

and ownership structure variables and the economic determinants are detailed in the Appendix. As noted

in Core et al. (1999), the portion of CEO_PAY explained by the board and ownership variables represents

overcompensation. Under the optimal contracting view there should be no association between these

governance provisions and CEO compensation, i.e., CEO compensation is only a function of economic

determinants.

32
 The results of estimating the cross-sectional models of CEO_PAY for sample firms (Equation A1) are

presented in Panel A of Table 9. As shown, the economic and board and ownership variables explain

from 39% (for Total Compensation) to 52% (for Salary + Bonus) of the variation in CEO pay. The F-tests

on the incremental explanatory power of the set of board and ownership variables relative to the economic

determinants of pay are significant at .001 level and below.

[Insert Table 9 here]

To estimate the proportion of the CEO’s compensation that represents overcompensation

(OVERCOMP), we first calculate for each sample firm the predicted excess compensation by multiplying

the estimated board and ownership coefficients by the sample firm’s board and ownership variables’

values. We then scale predicted overcompensation (OC) by the relevant CEO_PAY value (which we

label OC_Salary, OC_Salary+Bonus and OC_TotalComp).

Our major findings related to overcompensation are summarized in Panel B of Table 9. The mean

(median) overcompensation percentage for firms in our Investment Grade sample ranges from 34.9%

(32.7%) for OC_Salary to 56.1% (50.9%) for OC_TotalComp. For the firms in our Speculative Grade

sample, the corresponding mean (median) overcompensation percentages range from 39.1% (35.5%) for

OC_Salary to 59.2% (60.8%) for OC_TotalComp. For two of the three compensation measures, the

Speculative Grade sample exhibits significantly greater overcompensation of CEOs relative to the

Investment Grade sample after controlling for standard economic determinants of pay.

To give some idea of the incentives that CEOs of Speculative Grade firms have for trading off higher

firm debt cost due to weaker governance for higher overcompensation paid directly to them, we scale the

CEO’s overcompensation by his share of additional financing costs that the firm incurs by having

Speculative Grade debt. To calculate the CEO’s share of the higher financing costs, we multiply the

CEO’s percentage ownership in the firm by the after-tax cost of the additional financing costs to the

33
firm.22 The results of this comparison are presented in Panel C of Table 9. The median value of

OC_Salary is $240,000 (not tabled) which is roughly 123 times the CEO’s share of the additional after-

tax interest cost from having speculative grade debt versus investment grade debt. For OC_Salary+Bonus,

the median overcompensation is $646,000 with a multiple of 334, and for OC_TotalComp the median

overcompensation is $1,808,000, which translates into a multiple 905 times the CEO’s share of the after-

tax interest cost differential.23

To statistically test the CEO’s benefits-to-cost ratio of weak governance, we partition firms in our

Speculative Grade sample into two groups. The first group represents the sub-sample of firms for which

the CEO’s overcompensation is less than or equal to the CEO’s share of the interest expense. The other

group is the sub-sample of firms for which the CEO’s overcompensation is greater than his portion of

interest costs due to weak governance. The latter group reflects the firms for which the CEO benefits

outweigh the costs of weak governance. Consequently, these are firms where the CEO is expected to

impede governance improvements as they would result in a net loss to the CEO, given his degree of

overcompensation and ownership stake in the firm. The χ2 test results suggest that for the vast majority

of firms in the Speculative Grade debt sample, the degree of CEO overcompensation outweighs his share

of the additional debt costs that may result from weaker governance.

A competing explanation for why CEOs may be willing to have the firm incur higher interest costs

due to weaker governance is that the higher interest costs are associated with the firm taking on riskier

projects with a higher expected return, which benefits shareholders as the residual claimants. The higher

interest costs could be offset by higher shareholder value, which can also benefit the CEO indirectly via

continued employment or directly via his personal ownership stake in the firm. Moreover, under this

22
The additional after-tax financing costs are determined by multiplying the firm’s long-term debt at the beginning
of 2002 times 8% (the spread between investment grade and speculative grade debt with 10 year maturity in 2002)
times 65% (1 – marginal corporate tax rate).
23
A major reason for these relatively large multiples is because CEOs of most firms in the speculative grade sample
hold such a small percentage of the firm’s shares. For example, 169 of the 245 speculative grade firms have CEOs
that own less than one-tenth of 1% of the firm’s stock. Thus, the typical CEO’s share of the higher after-tax interest
costs is quite small.

34
competing explanation the “excess compensation” that we have estimated may simply be compensating

managers for bearing higher risk.

If this risk explanation for higher compensation is valid, we should observe firms with lower credit

ratings exhibiting higher future performance. That is, we would expect riskier firms to earn higher

accounting rates of return. Table 10 provides evidence on this competing explanation for our sample of

firms. Panel A shows that next year’s ROA (ROE) is positively associated with firms having investment

grade credit ratings after controlling for the current year ROA (ROE). Thus, it is not the case that firms

with lower credit ratings have higher shareholder value due to stronger future performance. Consistent

with results in Core et al. (1999), Panel B shows a significant negative relation between the amount of

overcompensation and future period ROA. Moreover, our excess compensation results are not due to

risk premia—that is, compensating managers for bearing higher risk. The risk effects on

compensation are controlled for through two of the economic determinants of compensation

included in the model--the standard deviation of ROA (STD_OPROA) and the standard deviation

of stock returns (STD_RET). These results are consistent with the conjecture that firms with weaker

governance have greater agency problems and the resulting CEO overcompensation is a dead weight cost

that is borne by all stakeholders.

[Insert Table 10 here]

If it is easier for CEOs of firms with weaker governance to garner excess compensation, as the results

presented here suggest, and if the CEO’s share of the additional debt costs are low, then there are clear

incentives for managers to resist efforts to strengthen governance. Thus, CEO expropriation of rents

through excess compensation provides one potential explanation for why some firms continue to operate

with weaker governance when doing so may mean lower credit ratings.

VI.2 Endogeneity

The preceding analysis treats governance attributes as being exogenously determined. Under the

assumption of optimal contracting, a firm’s governance structure is unique in equilibrium and

35
endogenously determined (Bushman, Chen, Engel and Smith, 2004; Hermalin and Weisbach, 2003). If

governance provisions are endogenously determined such that there is a factor or set of factors that affect

governance and also affect credit rating agencies’ assessment of firms’ creditworthiness, then our study

suffers from a potential correlated omitted variable problem. This misspecification causes the parameter

estimates to be inconsistent, which clouds the interpretation of results.

The econometric solution for endogeneity is to use two-stage procedures that rely on instrumental

variables to generate predicted values of the independent variables (in our case, the set of governance

variables) that are uncorrelated with the error term in the structural model. Unfortunately, instrumental

variables are very difficult to identify in most accounting research settings (Ittner and Larcker, 2001).

This is particularly true with respect to governance attributes in that there is no well developed theory or

model of the economic determinants of governance (Hermalin and Weisbach, 2003).

The lack of theory on the determinants of corporate governance draws into question the adequacy of

any instrumental variable approach to deal with potential endogeneity issues in our setting. However,

there is limited empirical evidence (Hermalin and Weisbach, 1991; Bhagat and Black, 2000), that poor

past performance (both accounting and stock market) leads to increases in board independence.

Therefore, past performance is potentially a correlated omitted variable, at least with respect to our board

independence measure.

In Table 11, we expand our full model (Model 6 in Table 6) to include stock returns (PP_RET) as a

past performance measure.24 This measure is industry-adjusted and we present results for one, three and

five-year prior performance horizons.25 With one exception (G_SCORE, 5-year prior performance) all of

the variables that were significantly related to credit ratings in our original model continue to be

24
Including past performance measures in our base model along with the set of governance variable is equivalent to
using two stage procedures where we first regress each of the governance variables on the past performance
variables and then include the predicted values from the first stage model into the structural model. We choose the
one-step approach because it is simpler to implement given the fourteen separate governance variables in our
structural model.
25
We also conducted analyses using raw performance and market-adjusted performance measures and the results are
qualitatively the same as those reported here.

36
significant in the augmented model. Thus, the inclusion of a past performance measure has little

influence on our conclusions regarding the importance of governance on credit ratings.

[Insert Table 11 here]

Another potential competing explanation for our results is that having debt in a firm’s capital

structure may act as a substitute for strong governance. Jensen (1986) argues that having debt in the

capital structure can reduce the agency costs that external stakeholders face when firms have free cash

flows available for discretionary spending by managers. The threat caused by failure to make debt service

payments serves as an effective motivating force for managers to invest firms’ resources in ways that

maximize the cash flows available to external stakeholders. Firms with weak governance may choose to

have managers pre-commit to achieving a minimal cash flow by issuing debt. Thus, Jensen’s free cash

flow theory of capital structure implies that firms with weaker governance will have more debt

outstanding, resulting in lower credit ratings and higher interest costs.

Merging governance and the traditional capital structure hypothesis suggests that there is a trade-off

between governance and capital structure. In other words, strong governance and leverage are negatively

related. Thus, the higher interest costs we attribute to having weaker governance may well be costs

associated with firms choosing to operate with higher leverage which serves the same role as governance

in disciplining managements’ opportunistic tendencies towards over investing. However, there are several

pieces of evidence in our results that are inconsistent with this competing capital structure hypothesis.

First, differences in sample firms’ capital structures are controlled for by including leverage as one of

the firm-specific control variables in our credit rating model. That is, the effects of the various governance

variables on credit ratings are after taking into account how leverage affects credit ratings. Second, in

unreported results, we find the correlations between the various governance variables and leverage to be

quite low, generally falling in the range of + or - 0.05. Only one of the governance variables

(%BRD_EXPERT) exhibits a significant negative correlation with leverage (ρ=-0.09). As a group, the

governance variables explain only 4% of the cross-sectional variation in leverage. Therefore, for our

sample, governance does not appear to determine leverage. Finally, firms that issued debt during our

37
sample period actually tended to have stronger governance along most dimensions than those firms that

did not issue debt.26 Again, these results are inconsistent with debt acting as a substitute for strong

governance.

In addition to the supplemental results noted above, there are other features of our setting that suggest

correlated omitted variables are not driving our results. In Table 6 we show that there are seven distinct

governance variables that are significantly related to credit ratings. There is at least one variable from

each of the four S&P components of governance that exhibits significant explanatory power, and there is

relatively low correlation among these seven governance variables (see Table 5). Thus, there is no single

omitted variable that could simultaneously be correlated with all seven of these governance variables in

such a way to provide an alternative explanation for our results. Moreover, it is hard to imagine that there

would be a set of omitted economic variables that would be highly correlated with our governance

variables and be correlated with credit ratings in a fashion that’s consistent with our findings.27

Another feature of our setting that suggests that we have appropriately modeled credit ratings is the

evidence from credit rating agencies themselves that indicates that governance features are an important

input into the credit rating process. For example, in a recent special report on credit policy entitled,

“Evaluating Corporate Governance: The Bondholders’ Perspective”28, Fitch Ratings (2004) states the

following:

“The purpose of this global criteria report is to inform the marketplace of Fitch
Ratings’ approach to evaluating and incorporating the quality of a company’s corporate
governance within the overall credit ratings process. While Fitch always has taken
aspects of corporate governance into account, this report formalizes a more systematic
framework for reviewing governance practices that affect credit quality . . .” Fitch’s
framework is grounded in agency theory and defines corporate governance from a
creditor perspective. . . . .Ultimately, companies that are found to have exceptionally

26
These results are available from the authors upon request.
27
We also acknowledge that there could be other economic variables omitted from the model that are correlated
(some positively and some negatively) with credit ratings. We have included all the major economic determinants of
credit ratings in our model based on evidence provided in prior research. If there are major economic variables that
have been omitted from our model, then these have been systematically overlooked by a vast literature on
determinants of debt ratings, and we believe this is unlikely.
28
Fitch Ratings, Credit Policy, Special Report, “Evaluating Corporate Governance: The Bondholders’ Perspective,”
April 12, 2004.

38
 weak corporate governance (or disclosure practices) could face a downgrade or other
negative rating action, while those with very strong practices might warrant a special or
favorable mention in the credit analysis.” (p. 1).

Statements like these by major credit rating agencies clearly indicate that governance factors are direct

inputs to the credit rating process. Moreover, three major rating agencies (S&P, Moody’s and Fitch

Ratings) have developed infrastructures and have invested significant resources to evaluate firms’

governance structures. These actions clearly signal that governance is important to the credit rating

process.

VI.3 Sensitivity Tests

In our original model, we use total audit fees paid to a firm’s auditor to proxy for the economic bond

between the auditor and client, which potentially threatens auditor independence. Contemporaneous

literature investigating the quality of accounting information in the presence of threats to auditor

independence uses alternative measures of economic bonding (e.g., Frankel, et al., 2002; Ashbaugh, et al.,

2003). To test the robustness of our results related to the quality of the audit function in governance, we

substitute two alternative proxies of economic bonding for the TOTFEE variable in the full model (not

tabled). The first substitution uses the ratio of non-audit fees to total fees (FEERATIO) for TOTFEE.

The second substitution is a dummy variable coded one if TOTFEE is in the upper quartile of the

distribution of TOTFEE and zero otherwise (FEEDUMMY). We expect observations falling into the

upper quartile of TOTFEE to be firms where auditor independence is more likely to be threatened. The

coefficients on FEERATIO and FEEDUMMY are insignificant at conventional levels. Thus, we continue

to find no evidence that measures of economic bonding between the auditor and client firm adversely

affects credit ratings.

Our second set of sensitivity tests relate to our proxy for transparent and timely financial reporting.

Recall that TIMELINESS is defined as negative one times the squared residual from a cross-sectional

regression of returns on the levels and changes in earnings. We substitute two alternative measures of

TIMELINESS in our full model. The first substitution defines TIMELINESS as negative one times the

variance of the residuals from firm-specific time-series regressions, where we require firms to have a

39
minimum (maximum) of eight (ten) years of data to estimate equation (2). The coefficient on this

specification of TIMELINESS is positive and significant at the .001 level or better.

Gelb and Zarowin (2002) provide an alternative specification of the relation between returns and

earnings. Specifically, they posit and provide evidence that current price is reflective of the

informativeness of future earnings. Thus, our second substitution for TIMELINESS is the negative

squared residual from the regression of returns on contemporaneous and future earnings and earnings

changes after controlling for future returns. Once again, the coefficient on this specification of

TIMELINESS is positive and significant at the .001 level or better.

We use the %BRD_EXPERT, the percentage of outside directors that sit on other boards, as a

measure of board competency or expertise following Klein (1998). However, there is evidence in the

literature that when board members sit on too many boards monitoring of management is compromised

and, as a consequence, firm performance deteriorates (Bhagat and Black, 1999; Klein, 1998). As an

additional sensitivity test (not tabled), we include a variable for the percentage of board members that sit

on four or more boards. We find no evidence that board members being “too busy” adversely affects

credit ratings, and adding this variable does not detract from the significance of our %BRD_EXPERT

variable.

The results of these sensitivity tests indicate that our inferences are robust to alternative measures of

governance attributes.

VII. Summary and Conclusions

Weak corporate governance has been singled out as the leading cause for recent high-profile cases of

corporate fraud. Using a framework for evaluating corporate governance structures recently developed by

Standard & Poor’s, this study investigates whether firms that exhibit strong governance benefit from

higher overall firm credit ratings relative to firms with weak governance. We present compelling

evidence that a variety of governance mechanisms do help explain firm credit ratings after controlling for

firm characteristics that prior research has shown to be related to debt ratings. Specifically, we find that

firm credit ratings are: (1) negatively associated with the number of blockholders that own at least a 5%

40
ownership in the firm; (2) positively related to weaker shareholder rights in terms of takeover defenses;

(3) positively related to the degree of financial transparency; and (4) positively related to over-all board

independence, board stock ownership and board expertise and negatively related to CEO power on the

board. We show that a hypothetical firm that possesses desirable governance characteristics from the

bondholder’s viewpoint approximately doubles its likelihood of receiving an investment grade credit

rating. Given the spread between investment grade and speculative grade bond yields, better governance

can translate into significant debt costs savings for firms.

Our primary analysis documents that firms’ governance affects firms’ credit ratings. Our secondary

analysis provides insights into why all firms do not possess strong governance. We note that the cost of

weak governance is borne by all stakeholders whereas the benefits of weak governance can accrue to

managers when they can appropriate some or all of the rents from outside stakeholders by resisting better

governance. We report compelling evidence that suggests that CEOs of weak governance firms can

garner overcompensation in excess of their share of debt costs due to weak governance. Thus, we provide

one explanation for why all firms do not practice good governance.

A number of organizations and companies (Standard & Poor’s, Board Analyst, The Board Institute,

Moody’s Investor Services, and FitchRatings) have begun to compile company ratings of corporate

governance practices along several dimensions. Investigating whether these composite ratings are useful

determinants of credit ratings is one avenue of future research. Another avenue of future research is to

focus on the benefits of governance to equity stakeholders by investigating the relation between

governance and firms’ cost of capital.

41
 References

Aboody, D. M. Barth and R. Kasznik. 2004. “Do Firms Manage Stock-Based Compensation Expense
Disclosed Under SFAS 123?” working paper, University of California at Los Angeles and Stanford
University.

Agrawal, A. and S. Chadha. 2003. “Corporate Governance and Accounting Scandals,” Working
paper, University of Alabama.

Agrawal, A. and C. Knoeber. 1996. “Firm Performance and Mechanisms to Control Agency
Problems between Managers and Shareholders,” Journal of Financial and Quantitative Analysis 31:
377-397.

Agresti, A. 2002. Categorical Data Analysis, John Wiley and Sons, New York.

Ashbaugh, H., R. LaFond and B. Mayhew. 2003. “Do Nonaudit Services Compromise Auditor
Independence? Further Evidence,” Accounting Review 78: 611-640.

Asquith, P. and T. Wizman. 1990. “Event Risk, Covenants, and Bondholder Returns in Leveraged
Buyouts,” Journal of Financial Economics 27: 195-213.

Barth, M. and W. Landsman. 2003. “Cost of Capital and the Quality of Financial Statement
Information,” Working paper, Stanford University.

Baysinger, B. and H. Butler. 1985. “Corporate Governance and the Board of Directors: Performance
Effects of Changes in Board Composition,” Journal of Law, Economics and Organizations 1: 101-
124.

Bhagat, S. and B. Black. 1999. “The Uncertain Relationship Between Board Composition and Firm
Performance,” Business Lawyer 54: 921-963.

Bhagat, S. and B. Black. 2000. “Board Independence and Long-Term Performance,” Working paper,
Stanford Law School, Stanford, CA.

Bhojraj, S. and P. Sengupta. 2003. “Effect of Corporate Governance on Bond Ratings and Yields:
The Role of Institutional Investors and the Outside Directors,” The Journal of Business 76: 455-475.

Bushman, R., Q. Chen, E. Engel and A. Smith. 2004. “Financial Accounting Information,
Organizational Complexity and Corporate Governance Systems,” Journal of Accounting &
Economics, forthcoming.

Boardman, C. and R. McEnally. 1981. “Factors Affecting Seasoned Corporate Bond Prices,” Journal
of Financial and Quantitative Analysis 16: 207-216.

Chung, H., and S. Kallapur. 2003. “Client Importance, Nonaudit Services, and Abnormal Accruals,”
Accounting Review 78: 931-955.

Core, J., R. Holthausen and D. Larcker. 1999. “Corporate Governance, Chief Executive Officer
Compensation, and Firm Performance,” Journal of Financial Economics 51: 371-406.

42
Dann, L. and H. DeAngelo. 1983. “Standstill Agreements, Privately Negotiated Stock Repurchases,
and the Market for Corporate Control,” Journal of Financial Economics 11: 275-300.

Davidson, W., B. Xie and W. Xu. 2004. “Market Reaction to Voluntary Announcements of Audit
Committee Appointments: The Effect of Financial Expertise,” Journal of Accounting & Public
Policy 23: 279-293.

DeAngelo, L. 1981. “Auditor Size and Audit Quality,” Journal of Accounting & Economics 3: 183-
200.

Dechow, P. and I. Dichev. 2002. “The Quality of Accruals and Earnings: The Role of Accrual
Estimation Errors,” Accounting Review 77: 35-59.

DeFond, M., R. Hann and X Hu. 2004. “Does the Market Value Financial Expertise on the Audit
Committee of Boards of Directors?” forthcoming, Journal of Accounting Research (Supplement).

DeFond, M., K. Raghunandan and K.R. Subramanyam. 2002. “Do Non-Audit Service Fees Impair
Auditor Independence? Evidence from Going Concern Audit Opinions,” Journal of Accounting
Research 40: 1247-1274.

FitchRatings, 2004. Credit Policy Special Report, “Evaluating Corporate Governance: The
Bondholders’ Perspective”, New York.

Frankel, R., M. Johnson and K. Nelson. 2002. “The Relation Between Auditors’ Fees for Nonaudit
Services and Earnings Management,” Accounting Review 77: 71-105.

Gelb, D. and P. Zarowin. 2002. “Corporate Disclosure Policy and the Informativeness of Stock
Prices,” Review of Accounting Studies 7: 33-52.

Gompers, P. J Ishii and A. Metrick. 2003. “Corporate Governance and Equity Prices,” Quarterly
Journal of Economics 118: 107-155.

Gordon, L. and J. Pound. 1993. “Information, Ownership Structure, and Shareholder Voting:
Evidence from Shareholder-Sponsored Corporate Governance Proposal,” Journal of Finance 48:
697-718.

Grinblatt, M. and S. Titman. 2002. Financial Markets and Corporate Strategy, McGraw-Hill Irwin,
2nd Edition.

Gu, Z. 2002. “Cross-Sample Incomparability of R2s and Additional Evidence on Value Relevance
Changes Over Time,” Working paper, Carnegie Mellon University.

Hermalin, B. and M. Weisbach. 1991. “The Effect of Board Composition and Direct Incentives on
Firm Performance,” Financial Management 21(4): 101-112.

Hermalin, B. and M. Weisbach. 2003. “Boards of Directors as an Endogenously Determined

Institution: A Survey of the Economic Literature,” Economic Policy Review 9: 7-26

Horrigan, J. 1966. “The Determinants of Long-Term Credit standing with Financial Ratios,” Journal
of Accounting Research 4: 44-62.

43
Imhoff, E. A. 2003. “Accounting Quality, Auditing, and Corporate Governance,” Accounting
Horizons Supplement 17: 117-128.

Jensen, M. 1993. “The Modern Industrial Revolution, Exit, and the Failure of Internal Control
Systems,” Journal of Finance 48: 831-880.

Jensen, M. and W. Meckling. 1976. “Theory of the Firm: Managerial Behavior, Agency Costs and
Ownership Structure,” Journal of Financial Economics 3: 305-360.

Kaplan, R., and G. Urwitz. 1979. “Statistical Models of Bond Ratings: A Methodological Inquiry,”
Journal of Business 52: 231-261.

Karpoff, J., P. Malatesta, and R. Walkling. 1996. “Corporate Governance and Shareholder
Initiatives: Empirical Evidence,” Journal of Financial Economics 42: 365-395.

Klein, A. 1998. “Firm Performance and Board Committee Structure,” Journal of Law and Economics
41: 275-303.

Klein, A. 2002. “Audit Committee, Board of Director Characteristics, and Earnings Management,”
Journal of Accounting & Economics 33: 375-400.

Klein, A. 2003. “Do Audit Committees with Financially Literate Directors Experience Less Earnings
Management,” Working paper, Stern School of Business, New York University.

Klock, M., S. Mansi and W. Maxwell, 2004. “Corporte Governance and the Agency Cost of Debt,”
Working paper, George Washington University.

Lamy, R., and R. Thompson. 1988. “Risk Premia and the Pricing of Primary Issue Bonds,” Journal
of Banking and Finance 12: 585-601.

McConnell, J. and H. Serves. 1990. “Additional Evidence on Equity Ownership and Corporate
Value,” Journal of Financial Economics 27: 595-612.

Nesbitt, S. 1994. “Long-Term Rewards from Shareholder Activism: A Study of the ‘CalPERS’
Effect,” Journal of Applied Corporate Finance 6: 75-80.

Opler, T. and J. Sokobin. 1997. “Does Coordinated Institutional Activism Work? An Analysis of the
Activities of the Council of Institutional Investors,” Working paper, Ohio State University.

Securities and Exchange Commission. 2003. NASD and NYSE Rulemaking: Relating to Corporate
Governance. Release No. 34-48745. Washington, D. C.: Government Printing Office.

Sengupta, P. 1998. “Corporate Disclosure Quality and the Cost of Debt,” Accounting Review 73:
459-474.

Shleifer, A. and R. Vishny. 1997. “A Survey of Corporate Governance,” Journal of Finance 52: 737-
783.

Standard & Poor’s. 2002. “Standard & Poor’s Corporate Governance Scores: Criteria, Methodology
and Definitions”, New York. McGraw-Hill Companies, Inc.

44
U.S. Census Bureau. 2005 Statistical Abstract of the United States: 2004-2005. Section 25 Banking,
Finance and Insurance.

U.S. Congress. 2002. The Sarbanes-Oxley Act of 2002. 107th Congress of the United States of
America. H.R. 3763. Washington, D.C.: Government Printing Office.

Warga, A. and I. Welch. 1993. “Bondholder Losses in Leveraged Buyouts,” Review of Financial
Studies 6: 959-982.

Yermack, D. 1996. “Higher Market Valuation of Companies with a Small Board of Directors,”
Journal of Financial Economics 40: 185-211.

Yermack, D. 2003. “Remuneration, Retention, and Reputation Incentives for Outside Directors,”
forthcoming, Journal of Finance.

Ziebart, D., and S. Reiter. 1992. “Bond Ratings, Bond Yields and Financial Information,”
Contemporary Accounting Research 9: 252-282.

45
Appendix

We use the following alternative measures of CEO pay (see Core, et al. 1999; Aboody, Barth and

Kasznik, 2004) to derive measures of overcompensation:

Salary = the dollar value ($000) of the base salary (cash and non-cash) earned by the CEO during fiscal
2002,
Salary+Bonus = current compensation ($000) comprised of salary and bonus earned by the CEO during
fiscal 2002,
TotalComp = total compensation ($000) earned by the CEO during fiscal 2002, comprised of the
following: salary, bonus, other annual, total value of restricted stock granted, total value of stock options
granted (using Black-Scholes), long-term incentive payouts, and all other total.

Following Core, et al. (1999), we use the following variables to measure the economic

determinants of CEO pay:

Sales = the natural log of Compustat #12,

MB = the average market to book ratio as of the end of fiscal 2001 (Compustat #25 x #199 divided by
#60) where firms are required to have a minimum of three and a maximum of five years of prior data,
OPROA = operating income divided by total assets (Compustat #13 minus #14 divided #6) for the 2001
fiscal year,
RET = the buy and hold return over the 2001 fiscal year,
STD_OPROA = the standard deviation of OPROA as of the end of the 2001 fiscal year where firms are
required to have a minimum of three and a maximum of five years of OPROA,
STD_RET = the standard deviation of buy and hold returns for the fiscal year where firms are required to
have a minimum of three and a maximum of five years of RET.

We use the following board structure and ownership structure variables to capture CEO

power/management entrenchment [the predicted relation with CEO compensation is shown in

parentheses—see Core, et al. (1999) and previous discussion in this paper]:

CEOPOWER = Composite score representing the power of the CEO where a firm receives one point if
the CEO is the chairman of the board, and one point for each of the committees (compensation,
nominating, audit) that the CEO sits on. (In some instances the CEO does not have voting power yet still
is identified as being part of the committee) (source Board Analyst) (+),
%COMP_CEOAPP = the percent of outside independent directors on the compensation committee
appointed by the CEO (+),
BRDSIZE = the number of directors on the board (+),
%BRD_INSIDE = the percent of the board made up of insiders (?),
%OUT_BUSY = the percent of outside independent directors that sit on more than 4 boards (number of
outside independent directors sitting on more than 4 boards divided by total outside independent
directors) (+),
%BRD_ActiveCEOs = the percent of the board that are currently employed as CEO’s (+),

%INST = % of shares held by institutional investors multiplied by 100 (source Board Analyst) (-),

46
BLOCK = Number of block holders, where block is defined at the 5% ownership level (source Compact
Disclosure) (-),
IND_DUM = industry dummies based on firms’ two digit SIC codes and are included for each two-digit
SIC group having at least 10 observations, for a total of 23 industry dummies.

The following cross-sectional OLS model is used to estimate the determinants of CEO

compensation using fiscal year 2002 data for our sample firms:
23
CEO _ COMP = ∑ b IND _ DUM
i 0I ni
+ b Sales + b MB + b OPROA + b RET
1 i 2 i 3 i 4 i
n =1

+ b STD _ OPROA + b STD _ RET + b CEOPOWER

5 i 6 i 7 i
+ b % COMP _ CEOAPP
8 i

+ b BRDSIZE + b % BRD _ INSIDE + b % OUT _ BUSY + b % BRD _ ActiveCEOs

9 i 10 i 11 i 12 i

+ b CEO _ Ownership
13 i
+ b % INST + b BLOCK + ε
14 i 15 i i
( A1)

47
Table 1: Credit Rating Classifications

S&P Compustat Issuer Long Term

Debt Rating Data280 Credit Rating Grade
AAA 2 7 Investment
AA+ 4 6 Investment
AA 5 6 Investment
AA- 6 6 Investment
A+ 7 5 Investment
A 8 5 Investment
A- 9 5 Investment
BBB+ 10 4 Investment
BBB 11 4 Investment
BBB- 12 4 Investment
BB+ 13 3 Speculative
BB 14 3 Speculative
BB- 15 3 Speculative
B+ 16 2 Speculative
B 17 2 Speculative
B- 18 2 Speculative
CCC+ 19 1 Speculative
CCC or CC 20,23 1 Speculative
C 21,24 1 Speculative
D or SD 27,29,90 1 Speculative

48
 Table 2 Sample Details

Panel A: Sample Construction (fiscal year 2002)

Number Firms
of Firms Lost

Number of firms in the main Board Analyst Data Set 2050 0

Number of firms having complete governance data from Board Analyst 1867 183

Number of firms having necessary Compustat data for ratings model 1693 174

Number of firms having g-scores 1392 301

Number of firms having credit ratings on Compustat 894 498

Panel B: Sample Firms’ Board Composition

Breakdown of Inside, Outside And Affiliate Directors by Board and by Committees

for Sample Firms in Fiscal Year 2002

Audit Compensation Nominating Finance

Board Committee Committee Committee Committee
Number of firms having a board (committee) 894 894 890 813 242
Average number of directors on board
(committee) 10.05 3.93 3.72 3.97 4.31
Average number of insider directors on board
(committee) 1.67 0.01 0.02 0.06 0.39
Average number of outside directors on board
(committee) 7.02 3.63 3.37 3.46 3.42
Number of firms having at least one insider on
the board (committee) 885 6 17 42 73
Number of firms having the CEO as chairman of
the board (committee) 652 3 10 31 65

Notes: This table presents the sample selection and data requirements. Panel B presents the size of the board and
committees for the sample of firms for which the specific committee exists. The difference between the average
number of directors on the board (committee) and number of insiders and outsiders on the board (committee) is due
to affiliated directors.

49
Table 3 Variable Definitions
Predicted
Variables Sign Definitions and Data Source
Ownership Structure and
Influence:
BLOCK Number of block holders, where block is defined at the 5% ownership level
?
(source Compact Disclosure)
%INST ? % of shares held by institutional investors (source Board Analyst)
%INSIDE − % of shares held by insiders (officers and directors) (source Board Analyst)
Financial Stakeholder Rights &
Relations:
G_SCORE ? Shareholder rights governance score (source Gompers, Ishii and Metrick 2003)
Financial Transparency:
WCAQ Negative one times the standard deviation of the firm-specific residual from the
prior three to five years, where residuals are from the following cross sectional
estimations of Dechow and Dichev’s (2002) model:
WCAt = β 0 + β 1CFOt −1 + β 2 CFOt + β 3 CFOt +1 + ε
+ where regressions are estimated by three, two, or one-digit SIC codes conditional
on having at least 10 firms in each SIC group. WCA=working capital accruals (-
(Compustat # 302+ Compustat # 303+ Compustat # 304+ Compustat # 305+
Compustat # 307)); CFO= cash flow from operations (Compustat # 308); all
variables are scaled by average total assets (Compustat # 6)
TIMELINESS Negative one times the squared residual from the following regression
RET = β 0 + β 1 NIBE + β 2 LOSS + β 3 NIBE * LOSS + β 4 ∆NIBE + ε
where the regression is estimated by three, two, or one-digit SIC codes conditional
on having at least 10 firms in each SIC group. RET= the market adjusted return
+ over the fiscal year (from CRSP); NIBE= net income before extraordinary items
(Compustat # 18) scaled by beginning of period market value of equity
(Compustat # 25* Compustat # 199); LOSS= one if NIBE is negative, zero
otherwise; ∆NIBE= the change in net income before extraordinary items
(Compustat # 18) scaled by beginning of period market value of equity
(Compustat # 18* Compustat # 199)
TOTFEES Total fees paid by the firm to its auditor divided by total revenues of the audit
− firm multiplied by 100 (source firms’ proxy statements, Accounting Today Top
100 firms and D&B’s Million Dollar Database)
%AUD_IND + % of audit committee made up of independent directors (source Board Analyst).
FIN_EXPERT One if the firm has an independent financial expert on the audit committee, where
+ financial expert is defined as an audit committee member being a CFO or having a
CPA, zero otherwise (source Board Analyst)

50
Board Structure and Processes:
%BRD_IND + % of independent directors on the board (source Board Analyst)
CEOPOWER Composite score representing the power of the CEO where a firm receives one
− point if the CEO is the chairman of the board, and one point for each committee
(compensation, nominating, audit) that the CEO sits on (source Board Analyst)
%BRD_EXPERT % of independent directors that hold seats on other firms’ boards (source Board
+
Analyst)
%BRD_STOCK + % of directors that own stock in the firm (source Board Analyst)
GOVERNANCE_POLICY One if the firm has a formal governance policy, zero otherwise (source Board
+
Analyst)
%FINCOM_INSIDE + % of insiders on the finance committee (source proxy statement)
%NOM_IND + % of independent directors on the nominating committee (source Board Analyst)
%COMP_IND % of independent directors on the compensation committee (source Board
+
Analyst)
Firm Characteristics:
LEV Total debt (Compustat #9 plus Compustat #34) divided by total assets (Compustat
−
#6).
ROA + Net income before extraordinary items (Compustat #18) divided by total assets
LOSS One if the net income before extraordinary items is negative in the current and
−
prior fiscal year, zero otherwise
INT_COV Operating income before depreciation (Compustat #13) divided by interest
+
expense (Compustat #15) or (Compustat #339)
SIZE + Natural log of total assets
SUBORD − One if the firm has subordinated debt, zero otherwise
CAP_INTEN + Gross PPE (Compustat #7) divided by total assets
FIN_UTILITY One if firm is a financial institution (one-digit SIC code 6) or a utility (two-digit
+
SIC code 49), zero otherwise

51
Table 4 Summary Statistics on Credit Rating Variables

Standard
Variables Mean Deviation Median 25% 75%
Ownership Structure and Influence:
BLOCK 4.46 2.80 4.00 2.00 6.00
%INST 0.63 0.24 0.68 0.51 0.80
%INSIDE 0.08 0.11 0.04 0.02 0.08

Financial Stakeholder Rights& Relations:

G_SCORE 9.59 2.59 10.00 8.00 11.00

Financial Transparency:
WCAQ -0.04 0.03 -0.03 -0.05 -0.02
TIMELINESS -0.10 0.17 -0.03 -0.10 -0.01
TOTFEES 0.11 0.22 0.04 0.02 0.10
%AUD_IND 0.92 0.16 1.00 1.00 1.00
FIN_EXPERT 0.26 0.44 0.00 0.00 1.00

Board Structure and Processes:

%BRD_IND 0.70 0.16 0.73 0.58 0.83
CEOPOWER 0.78 0.50 1.00 0.00 1.00
%BRD_EXPERT 0.36 0.22 0.36 0.20 0.50
%BRD_STOCK 0.87 0.20 0.92 0.82 1.00
GOVERNANCE_POLICY 0.42 0.49 0.00 0.00 1.00
%FINCOM_INSIDE 0.15 0.12 0.13 0.08 0.22
%NOM_IND 0.79 0.32 1.00 0.67 1.00
%COMP_IND 0.90 0.20 1.00 0.83 1.00

Firm Characteristics:
LEV 0.31 0.17 0.30 0.19 0.41
ROA 0.02 0.08 0.03 0.01 0.06
LOSS 0.13 0.34 0.00 0.00 0.00
INT_COV 10.78 18.77 5.16 2.79 10.23
SIZE 8.46 1.49 8.20 7.38 9.44
SUBORD 0.18 0.39 0.00 0.00 0.00
CAP_INTEN 0.54 0.40 0.48 0.19 0.84
FIN_UTILITY 0.23 0.42 0.00 0.00 0.00
RATING 3.83 1.11 4.00 3.00 5.00
INVESTMENT_GRADE 0.63 0.48 1.00 0.00 1.00

Notes: RATING= S&P LT Domestic Issuer Credit Rating (Compustat #280), see Table 1 for numeric coding;
INVESTMENT_GRADE= 1 if a firm’s credit rating is investment grade as noted in Table 1, zero otherwise.
See Table 3 for other variable definitions.

52
Table 5 Correlations

Panel A: Firm Characteristics

RATING LEV ROA LOSS INT_COV SIZE SUBORD CAP_ FIN_

INTEN UTILITY
RATING -0.28 0.44 -0.42 0.27 0.50 -0.19 -0.03 0.24
LEV -0.26 -0.22 0.21 -0.42 -0.03 0.21 0.28 -0.03
ROA 0.39 -0.23 -0.60 0.34 0.06 -0.04 -0.01 0.00
LOSS -0.41 0.17 -0.55 -0.18 -0.16 0.01 0.06 -0.18
INT_COV 0.45 -0.52 0.72 -0.42 -0.02 -0.14 -0.07 -0.13
SIZE 0.51 -0.06 -0.09 -0.15 -0.04 -0.02 -0.20 0.41
SUBORD -0.21 0.19 -0.13 0.01 -0.19 -0.06 -0.09 -0.04
CAP_INTEN -0.04 0.33 0.01 0.08 0.03 -0.19 -0.10 -0.18
FIN_UTILITY 0.26 -0.03 -0.16 -0.18 -0.23 0.39 -0.04 -0.23

Notes: Bold text indicates significance at the 0.01 level or better.

53
Panel B: Governance Variables

A B C D E F G H I J K L M N O P Q R
RATING A -0.36 0.11 -0.16 0.19 0.30 0.38 0.26 0.03 0.06 0.18 0.16 0.11 0.07 0.31 0.25 0.26 -0.10
BLOCK B -0.39 0.22 0.14 -0.05 -0.18 -0.09 -0.20 -0.01 0.02 -0.05 -0.06 -0.01 -0.06 -0.06 -0.12 -0.10 0.04
%INST C 0.01 0.35 -0.18 0.13 -0.08 0.07 0.02 0.06 0.08 0.18 0.21 0.12 -0.02 0.19 0.04 0.14 -0.04
%INSIDE D -0.33 0.29 -0.08 -0.23 -0.10 -0.04 -0.15 -0.04 -0.12 -0.39 -0.31 -0.21 -0.06 -0.24 -0.13 -0.15 0.19
G_SCORE E 0.21 -0.06 0.09 -0.18 0.10 0.14 -0.07 0.09 0.09 0.27 0.25 0.18 0.05 0.20 0.23 0.12 -0.20
WCAQ F 0.32 -0.23 -0.18 -0.22 0.07 0.16 0.09 -0.01 0.07 0.08 0.02 0.07 0.02 0.07 0.17 0.07 -0.04
TIMELINESS G 0.28 -0.11 0.02 -0.06 0.06 0.16 0.02 -0.04 0.03 0.02 0.02 0.07 0.08 0.04 0.13 0.10 0.01
TOTFEES H 0.25 -0.15 0.05 -0.29 0.04 0.03 -0.01 -0.06 0.06 0.09 0.07 0.05 0.07 0.28 0.06 0.20 -0.05
FIN_EXPERT I 0.04 -0.01 0.05 -0.05 0.10 -0.01 -0.04 -0.02 0.09 0.10 0.14 0.09 0.01 0.05 0.01 0.04 -0.05
%AUD_IND J 0.03 0.04 0.08 -0.11 0.07 0.00 0.04 0.05 0.09 0.51 0.38 0.41 0.04 0.27 0.06 0.09 -0.12
%BRD_IND K 0.19 -0.04 0.15 -0.38 0.27 0.06 0.05 0.13 0.10 0.46 0.56 0.55 0.13 0.54 0.16 0.23 -0.49
%NOM_IND L 0.14 -0.04 0.18 -0.23 0.20 0.04 0.01 0.06 0.13 0.42 0.55 0.44 0.02 0.36 0.17 0.24 -0.26
%COMP_IND M 0.10 0.00 0.10 -0.18 0.16 0.03 0.06 0.06 0.10 0.43 0.51 0.50 0.00 0.31 0.09 0.14 -0.14
CEOPOWER N 0.07 -0.06 -0.01 -0.12 0.06 0.02 0.06 0.09 0.01 0.05 0.16 0.01 0.02 0.12 0.07 0.06 -0.09
%BRD_EXPERT 0 0.31 -0.05 0.17 -0.35 0.21 0.01 0.03 0.35 0.05 0.27 0.54 0.36 0.31 0.13 0.14 0.33 -0.33
%BRD_STOCK P 0.28 -0.16 -0.02 -0.19 0.28 0.19 0.05 0.13 -0.01 0.08 0.24 0.19 0.15 0.10 0.22 0.15 -0.09
GOVERNANCE
POLICY Q 0.25 -0.10 0.11 -0.21 0.12 0.05 0.07 0.21 0.04 0.08 0.22 0.21 0.14 0.08 0.33 0.18 -0.23
%FINCOM_INSID
E R -0.15 0.05 -0.01 0.24 -0.22 -0.06 -0.02 -0.16 -0.06 -0.11 -0.49 -0.18 -0.14 -0.07 -0.35 -0.18 -0.23

Notes: Bold text indicates significance at the 0.01 level. RATING= S&P LT Domestic Issuer Credit Rating (Compustat #280), see Table 1 for numeric coding. See Table 3 for other
variable definitions.

54
Table 6 Logistic Regression Results of the Effects of Corporate
Governance Attributes on Firms’ Credit Ratings (Dependent Variable = RATING)

Predicted Estimated Coefficient

Variables Sign
Model 1 Model 2 Model 3 Model 4 Model 5 Model 6

Firm Characteristics

LEV − -2.223*** -2.018*** -2.222*** -2.302*** -2.133*** -2.014***

ROA + 0.108*** 0.112*** 0.107*** 0.098*** 0.103*** 0.094***
LOSS − -1.073*** -1.018*** -1.019*** -1.000*** -1.159*** -0.986***
INT_COV + 0.015*** 0.013*** 0.016*** 0.016*** 0.016*** 0.017***
SIZE + 0.869*** 0.777*** 0.867*** 0.805*** 0.778*** 0.643***
SUBORD − -0.744*** -0.654*** -0.684*** -0.747*** -0.673*** -0.604***
CAP_INTEN + 0.863*** 0.811*** 0.825*** 0.712*** 0.738*** 0.538***
FIN_UTILITY + 0.436*** 0.312** 0.438*** 0.259* 0.501*** 0.214

Governance Attributes
Ownership Structure and Influence
BLOCK ? -0.162*** -0.165***
%INST ? 0.594** 0.390
%INSIDE − -0.778* 0.442

Financial Stakeholder Rights &

Relations
G_SCORE ? 0.094*** 0.037*

Financial Transparency
WCAQ + 6.911*** 7.372***
TIMELINESS + 3.571*** 3.463***
TOTFEES − 0.495 0.340
%AUD_IND + 0.801** -0.036
FIN_EXPERT + 0.203* 0.126

Board Structure and Processes

%BRD_IND + 1.041** 0.950*
CEOPOWER − -0.098 -0.177*
%BRD_EXPERT + 0.794** 1.003***
%BRD_STOCK + 1.385*** 1.109***
GOVERNANCE_POLICY + 0.235** 0.179
%FINCOM_INSIDE + 0.449 0.317

Generalized R-square 0.52 0.54 0.53 0.56 0.54 0.60

Likelihood ratio χ2 660.55*** 701.67*** 673.95*** 738.46*** 702.51*** 811.55***

Wald χ2 39.64*** 13.47*** 77.16*** 42.66*** 139.93***

Sample Size 894 894 894 894 894 894

Notes: Tabled values are coefficient estimates from the following ordered logit models:

Model 1 RATING = f(Firm Characteristics)

Model 2 RATING = f(Firm Characteristics, Ownership Structure and Influence)
Model 3 RATING = f(Firm Characteristics, Financial Stakeholder Rights & Relations)
Model 4 RATING = f(Firm Characteristics, Financial Transparency)
Model 5 RATING = f(Firm Characteristics, Board Structure and Processes)
Model 6 RATING = f(Firm Characteristics, Governance Attributes)

***,**,* indicates significance at the 0.01, 0.05, and 0.10 level or better, respectively. The Wald χ2 statistic tests whether
the governance attributes added in each model as a whole explain a significant portion of the variation in firms’ credit ratings.
RATING= S&P Long Term Domestic Issuer Credit Rating (Compustat #280), see Table 1 for numeric coding. See Table 3
for other variable definitions.

56
Table 7 Logistic Regression Results of the Effects of Corporate
Governance Mechanisms on Firms’ Credit Ratings (Dependent Variable = INVESTMENT_GRADE)

Predicted Estimated Coefficient

Variables Sign
Model 1 Model 2 Model 3 Model 4 Model 5 Model 6

Firm Characteristics:

INTERCEPT ? -8.321*** -6.635*** -9.540*** -7.418*** -10.202** -7.125***

LEV − -3.450*** -3.262*** -3.385*** -3.864*** -3.533*** -3.545***
ROA + 0.142*** 0.145*** 0.137*** 0.132*** 0.137*** 0.130***
LOSS - -1.400*** -1.263*** -1.346*** -1.240*** -1.507*** -1.111***
INT_COV + 0.006 0.005 0.010 0.004 0.007 0.007
SIZE + 1.104*** 0.990*** 1.080*** 1.048*** 1.066*** 0.913***
SUBORD − -1.473*** -1.445*** -1.385*** -1.501*** -1.412*** -1.416***
CAP_INTEN + 1.353*** 1.253*** 1.267*** 1.150*** 1.286*** 0.925***
FIN_UTILITY + 0.592** 0.393* 0.599** 0.441* 0.488** 0.194

Governance Attributes
Ownership Structure and Influence
BLOCK ? -0.219*** -0.226***
%INST ? 0.680* 0.583
%INSIDE − -1.324* -0.207

Financial Stakeholder Rights &

Relations
G_SCORE ? 0.145*** 0.067*

Financial Transparency
WCAQ + 7.066** 6.995**
TIMELINESS + 4.215*** 4.108***
TOTFEES − 0.661 0.110
%AUD_IND + 0.481 -0.434
FIN_EXPERT + 0.064 -0.004

Board Structure and Processes

%BRD_IND + 1.270** 0.885
CEOPOWER − -0.086 -0.168
%BRD_EXPERT + 0.248 0.698
%BRD_STOCK + 1.540*** 1.320***
GOVERNANCE_POLICY + 0.182 0.066
%FINCOM_INSIDE + 0.026 -0.224

Generalized R-square 0.43 0.45 0.43 0.45 0.44 0.48

Likelihood ratio χ2 495.10*** 530.51*** 509.57*** 536.46*** 516.25*** 587.35***

Wald χ2 33.04*** 14.02*** 33.62*** 20.76*** 72.84***

Sample Size 894 894 894 894 894 894

Notes: Table values are coefficient estimates from the following logit models:

Model 1 INVESTMENT_GRADE = f(Firm Characteristics)

Model 2 INVESTMENT_GRADE = f(Firm Characteristics, Ownership Structure and Influence)
Model 3 INVESTMENT_GRADE = f(Firm Characteristics, Financial Stakeholder Rights & Relations)

57
Model 4 INVESTMENT_GRADE = f(Firm Characteristics, Financial Transparency)
Model 5 INVESTMENT_GRADE = f(Firm Characteristics, Board Structure and Processes)
Model 6 INVESTMENT_GRADE = f(Firm Characteristics, Governance Attributes)

***,**,* indicates significance at the 0.01, 0.05, and 0.10 level or better, respectively. The Wald χ2 statistic tests whether
the governance attributes added in each model as a whole explain a significant portion of the variation in firms’ investment
grade ratings. INVESTMENT_GRADE= 1 if a firm’s credit rating is investment grade as noted in Table 1, zero otherwise.
See Table 3 for other variable definitions.

58
Table 8 Assessment of Changes in Probabilities of Receiving an Investment Grade Credit Rating for
Selected Changes in Governance Variable Values

Predicted
Marginal Effect Change in Probability
Variables Sign
Standardized Variables Q1 vs. Q3 Values
Governance Attributes

Ownership Structure and Influence:

BLOCK ? -0.136 -0.194
%INST ? 0.030 0.036
%INSIDE − -0.005 -0.003

Financial Stakeholder Rights & Relations:

G_SCORE ? 0.037 0.043

Financial Transparency:
WCAQ + 0.042 0.045
TIMELINESS + 0.152 0.081
TOTFEES − 0.005 0.002
%AUD_IND + -0.015 0.000
FIN_EXPERT + -0.001 -0.001

Board Structure and Processes:

%BRD_IND + 0.031 0.048
CEOPOWER − -0.018 -0.036
%BRD_EXPERT + 0.033 0.045
%BRD_STOCK + 0.056 0.052
GOVERNANCE_POLICY + 0.014 0.014
%FINCOM_INSIDE + -0.006 -0.007

Notes: The Marginal Effects column shows the effects of receiving an investment grade credit rating due to a one
unit change in the variable of interest after standardizing the independent variables. Marginal effects are computed
as: π ( X ) = e β ' X (1 + e β ' X ) where β’X is evaluated at the mean values of X. Tabled values in the Change in
Probability column show the change in the probability of receiving an investment grade credit rating as a result of
moving from the first to the third quartile value of the variable of interest, holding all other variables constant at
their mean values. See Table 3 for variable definitions.

59
Table 9 Money on the Table Analysis
Panel A: OLS Model of CEO Compensation
Dependent Variable
Salary Salary+Bonus Total Compensation
Economic Determinants
SALES 123.143*** 331.884*** 1873.306***
MB -1.589 13.675 104.749*
OPROA 223.394 813.621 -774.941
RET 29.145 199.105*** -287.823
STD_ OPROA -57.652 -251.676 3489.100
STD_RET -43.506 -54.487 1130.478**

Governance Determinants
CEOPOWER 51.177*** 103.168 583.542*
%COMP_CEOAPP 78.960*** 485.927*** 905.928**
BRDSIZE 13.835*** 45.726*** 146.444**
%BRD_INSIDE -98.787 -248.887 -853.844
%OUT_BUSY 116.613 590.452** 1552.127
%BRD_ActiveCEOs 43.982 64.081 684.285
CEO Ownership -244.240** 190.693*** 22.806
%INST 1.121** 3.832** 11.731
BLOCK -4.229 -47.030*** -202.682***

Adj R2 0.47 0.52 0.39

F-stat Governance Determinants 5.65*** 8.21*** 2.73***
Sample Size 737 737 737

60
Table 9 continued
Panel B: Estimates of Overcompensation by Grade
Overcompensation estimates INVESTMENT_GRADE SPECULATIVE_GRADE P-value
n=492 n=245 (two-tailed)
OC_ Salary % (mean) 0.349 0.391 0.00
OC_ Salary % (median) 0.327 0.355 0.01
OC Salary+Bonus %(mean) 0.535 0.600 0.00
OC Salary+Bonus %(median) 0.478 0.573 0.01
OC_ Total Comp %(mean) 0.561 0.592 0.24
OC_ Total Comp %(median) 0.509 0.608 0.26

Panel C: Overcompensation to Share of Interest Costs Ratio – Speculative Grade Firms (n=245)
Non Investment Grade Firms (245) Q1 Median Q3
OC_ Salary/CEOSHARE_INTEXP 51.242 122.816 435.501
OC Salary+Bonus /CEOSHARE_INTEXP 133.467 334.230 1212.920
OC_ Total Comp /CEOSHARE_INTEXP 365.773 905.243 3180.150

Panel D: Test of Significance

# firms where ratio is # firms where ratio is greater χ2
less than or equal to 1 than 1

OC_ Salary/CEOSHARE_INTEXP 5 240 225.41***

OC Salary+Bonus /CEOSHARE_INTEXP 4 241 229.26***
OC_ Total Comp /CEOSHARE_INTEXP 4 241 229.26***

Notes: Panel presents the coefficient estimates for OLS regression, see Appendix. ***,**,* indicates significance
at the 0.01, 0.05, and 0.10 level or better, respectively. Salary is equal to the dollar value in thousands of the base
salary (cash and non-cash) earned by the CEO during fiscal 2002. Salary+Bonus is equal to the current
compensation in thousands comprised of salary and bonus earned by the CEO during fiscal 2002. Total
Compensation is equal to total compensation in thousands earned by the CEO during fiscal 2002, which is
comprised of the following: salary, bonus, other annual pay, total value of restricted stock granted, total value of
stock options granted (using Black-Scholes), long-term incentive payouts, and all other total pay. OC_ Salary % is
defined as the dollar amount of Salary due to the governance determinants divided by salary; Salary+Bonus % is
defined as the dollar amount of Salary+Bonus due to the governance determinants divided by Salary+Bonus; OC_
Total Comp % is defined as the dollar amount of Total Compensation due to the governance determinants divided
by Total Compensation. INVESTMENT_GRADE and SPECULATIVE_GRADE are based on a firm’s credit
rating as noted in Table 1. Median differences are assessed using the Wilcoxon rank sum test for differences in the
distributions. All p-values are two sided. # For the purpose of this analysis the RATINGS groups are collapsed into
five groups instead of the original seven used in the primary analysis due to the small number of firm having the
necessary data for the compensation analysis in the lowest (n=2) and highest (n=8) debt ratings groups. To calculate
the CEO’s portion of avoidable interest (CEOSHARE_INTEXP), we multiple 8% (spread between investment grade
and non-investment grade debt) times 0.65 (tax benefit to debt) times the CEO’s ownership stake in the firm times
the total debt outstanding. For CEO’s with zero ownership in the firm, we set the OC/ CEOSHARE_INTEXP to the
sample median for the respective compensation figure. See Table 3 for other variable definitions.

61
Table 10 Relation between Future Performance, Credit Ratings and Overcompensation

Panel A: Future performance and current credit rating (investment vs. non-investment grade)
Predicted Dependent variable
Variables Sign
NEXT YEAR’S ROA NEXT YEAR’S ROE
Firm Characteristics:
ROA + 0.436***
ROE + 0.470***
INVESTMENT GRADE ? 0.009*** 0.055***

Adjusted R-square 0.36 0.44

Sample Size 869 869

Panel B: Future performance and current period overcompensation

Predicted Dependent variable
Variables Sign NEXT YEAR’S NEXT YEAR’S NEXT YEAR’S
ROA ROA ROA

Firm Characteristics:
ROA + 0.498*** 0.480*** 0.496***
OC_ Salary % − -0.021***
OC Salary+Bonus% − -0.024***
OC_ Total Comp % − -0.023***

Adjusted R-square 0.41 0.42 0.43

Sample Size 717 717 717

Notes: Panel A presents the coefficient estimates from the following OLS regressions:
Model 1 ROAt +1 = β 0 + β 1 ROAt + β 2 INVESTMENTGRADEt + ε
Model 2 ROEt +1 = β 0 + β 1 ROEt + β 2 INVESTMENTGRADEt + ε
ROE is equal to net income before extraordinary items (Compustat #18) divided by book value of common equity
(Compustat #60). See Tables 3 and 9 for other variable definitions. Panel B presents the coefficient estimates form
the following OLS regressions
Model 1 ROAt +1 = β 0 + β 1 ROAt + β 2 OC _ Salary % t + ε
Model 2 ROAt +1 = β 0 + β 1 ROAt + β 2 OC _ Salary + Bonus% t + ε
Model 3 ROAt +1 = β 0 + β 1 ROAt + β 2 OC _ TotalComp % t + ε
***,**,* indicates significance at the 0.01, 0.05, and 0.10 level or better, respectively. See Table 9 for variable
definitions

62
 Table 11 Sensitivity Tests with Prior Period Performance

Variables Predicted 1-year Prior 3- year Prior 5-year Prior

Sign Performance Performance Performance
Firm Characteristics:

LEV − -2.154*** -1.794*** -1.763***

ROA + 0.107*** 0.089*** 0.095***
LOSS - -1.089*** -1.155*** -1.045***
INT_COV + 0.015*** 0.029*** 0.025***
SIZE + 0.643*** 0.634*** 0.628***
SUBORD − -0.569*** -0.544**** -0.580***
CAP_INTEN + 0.490*** 0.413** 0.244
FIN_UTILITY + 0.159 0.250 0.149
PP_RET ? -0.006*** -0.007** -0.009**

Governance Attributes
Ownership Structure and Influence
BLOCK ? -0.156*** -0.169*** -0.176***
%INST ? 0.159 0.303 0.133
%INSIDE − 0.497 0.461 0.503

Financial Stakeholder Rights &

Relations
G_SCORE ? 0.040* 0.036* 0.032

Financial Transparency
WCAQ + 7.736*** 6.738*** 10.147***
TIMELINESS + 3.382*** 3.363*** 3.034****
TOTFEES − 0.316 0.323 0.265
%AUD_IND + -0.063 -0.216 -0.637
FIN_EXPERT + 0.189 0.165 0.176

Board Structure and Processes

%BRD_IND + 1.107** 1.362** 1.337**
CEOPOWER − -0.167 -0.210* -0.245**
%BRD_EXPERT + 1.005*** 0.994** 1.214***
%BRD_STOCK + 0.915*** 1.127*** 1.133***
GOVERNANCE_POLICY + 0.116 0.179 0.191
%FINCOM_INSIDE + 0.276 0.713 0.952*

Generalized R-square 0.60 0.60 0.60

Likelihood ratio χ2 800.97*** 770.74*** 709.96***

Wald χ2 125.40*** 128.03*** 117.07***

Sample Size 870 832 783

See Table 3 for variable definitions.

63
 Notes: Table values are coefficient estimates from the following ordered logit models:

Model 1 RATING = f(Firm Characteristics, Prior 1 year Return, Governance Attributes)

Model 2 RATING = f(Firm Characteristics, Prior 3 year Return, Governance Attributes)
Model 3 RATING = f(Firm Characteristics, Prior 5 year Return, Governance Attributes)

***,**,* indicates significance at the 0.01, 0.05, and 0.10 level or better, respectively. The Wald χ2 statistic tests
whether the governance attributes added in each model as a whole explain a significant portion of the variation in
firms’ credit ratings. RATING= S&P Long Term Domestic Issuer Credit Rating (Compustat #280), see Table 1 for
numeric coding, PP_RET is equal to the prior period(s) return over the fiscal year, in the three and five year columns
it is set to the average return over the past three, five years. PP_RET is industry-adjusted, where industry groups are
defined by four, three, two, and one digit SIC codes with a minimum of 10 firms in each industry group. See Table
3 for other variable definitions.

64