Corporate Social Responsibility and The Cost of Corporate Bonds

Corporate Social Responsibility and the Cost of Corporate Bonds
Wenxia Gea and Mingzhi Liua,*
a
Asper School of Business, University of Manitoba, Winnipeg, Canada
* Corresponding author
Abstract
This study examines how a firm’s corporate social responsibility (CSR) performance is
associated with the cost of its new bond issues. Using credit ratings as an ex ante cost of debt,
we find that better CSR performance is associated with better credit ratings. After controlling
for credit ratings, our results show that better CSR performance is associated with lower yield
spreads but some of the effect is absorbed by credit ratings. When we examine CSR strengths
and concerns separately, we find that a higher CSR strength (concern) score is associated with
lower (higher) yield spreads. Our results on the effect of firm performance on seven individual
CSR dimensions are generally consistent with our main findings. Our results indicate that firms
with better CSR performance are able to issue bonds at lower cost and that both CSR strengths
and concerns are considered by bondholders. Additional subsample test results suggest that the
association between CSR performance and bond yield spreads is more pronounced in
investment-grade and non-Rule 144a bonds, for financially healthier bond issuers, for issuers
with weaker corporate governance and higher information asymmetry, and for issuers operating
in environmentally sensitive industries.
Key words: Corporate Social Responsibility; Information Asymmetry; New Corporate Bond
Issue; Cost of Debt
1
1. Introduction
The World Bank Council for Sustainable Development defines corporate social
responsibility (CSR) as “the continuing commitment by business to behave ethically and
contribute to economic development while improving the quality of life of the workforce and
their families as well as of the local community and society at large.” CSR has received
increasing attention in the business and political world. A recent survey by Lacy et al. (2010)
finds that 93% of 766 chief executive officers (CEOs) from all over the world believe that issues
related to CSR are critical to the future success of their businesses. 1 Even the current economic
downturn has not weakened corporate commitment to CSR activities. About 74% of CEOs report
that the economic downturn has led their corporations to align CSR more closely with core
business. In the past two decades, a growing number of corporations have dedicated considerable
resources to CSR activities. According to the Forum for Sustainable and Responsible Investment
(formerly the Social Investment Forum), social responsibility-related investment in the United
States grew dramatically from 1995 to 2010, from $639 billion to $3.07 trillion. In 2010, social
responsibility investment accounted for 12% of total U.S. asset investment.
Trends in CSR activities have also attracted academic attention, particularly the rationale
behind CSR activities (Barnea and Rubin, 2010; Cespa and Cestone, 2007; Elfenbein et al., 2012;
Jensen, 2001). There are two opposing theories about CSR: the shareholder theory and the
stakeholder theory. According to Friedman’s (1962) shareholder theory, a corporation’s top
priority is to maximize firm value. Some shareholder theorists argue that CSR activities consume
corporate resources that should be allocated to generating profit. Thus, they predict a negative
relation between shareholder value maximization and CSR activities. In contrast, the stakeholder
1
Klaus Kleinfeld, chairman and CEO of Alcoa, one of the world’s largest aluminum producers, comments that CSR
issues now have to be on everyone’s agenda, and that represents a fundamental change (Lacy et al. 2010).
2
theory suggests that corporations should go beyond the interests of shareholders and consider the
interests of a broader group of stakeholders (Freeman, 1984). Jones (1995) extends stakeholder
theory by indicating that CSR is essential for corporations in obtaining necessary resources and
stakeholder support.
A KPMG survey finds that the most common reason for disclosing CSR is the belief that
social responsibility will provide potential economic benefits to the firm due to a positive
reputation effect (Verschoor, 2005). Recent studies provide mixed evidence of the effect of CSR
performance on the cost of financing. Dhaliwal et al. (2011) and Ghoul et al. (2011) show that
U.S. firms with superior CSR performance obtain cheaper equity financing. Using a sample of
loans to U.S. firms, however, Goss and Roberts (2011) find that banks do not reward the CSR
investments of borrowers. Until now, little evidence has been uncovered on how U.S. public
bondholders value CSR performance. To fill this void, this study examines the economic
consequences of CSR performance on the cost of public debt financing. Specifically, we
investigate the association between CSR performance (strengths and concerns) and the cost of
new corporate bond issues in the U.S. bond market.
We obtain CSR data from the RiskMetrics Group KLD STATS database. We collect
bond data from the Mergent Fixed Income Securities Database and accounting data from
Compustat. We assign the value of one for each item related to CSR strength or concern and then
define each firm’s overall CSR performance score as the difference between its total strength and
total concern scores. A higher CSR performance score indicates better CSR performance.
Our results indicate the following. First, we find that a higher overall CSR performance
score is associated with better credit ratings. After controlling for credit ratings, we find that a
higher overall CSR performance score is associated with lower bond yield spreads. This result is
3
robust to the use of instrumental variables generalized method of moments (GMM) estimators to
address potential endogeneity issues and a variety of other robustness tests.
Second, by examining CSR strengths and concerns separately, we find that a higher CSR
strength score is significantly associated with better credit ratings but the estimated coefficient of
the CSR concern score is not statistically significant. We also find that the CSR strength score is
negatively associated with bond yield spreads and the CSR concern score is positively associated
with bond yield spreads.
Third, focusing on firm performance in seven CSR dimensions, we find that higher firm
overall performance scores in four CSR dimensions (i.e., community, product, employee
relations, and governance) are significantly associated with lower bond yield spreads. Our results
also show that the strength scores in three CSR dimensions—environment, community, and
governance—are significantly associated with lower bond yield spreads, while the concern
scores in three CSR dimensions, including product, diversity, and employee relations, are
significantly associated with higher bond yield spreads.
Fourth, our subsample analyses show that the association between CSR performance and
the cost of debt is more pronounced in investment-grade bonds and non-Rule 144a bonds, for
financially healthier bond issuers, for issuers with weaker corporate governance and higher
information asymmetry, and for issuers operating in environmentally sensitive industries.
Last, we also examine whether CSR performance affects bond maturity and the intensity
of using covenants, two measures of the indirect cost of debt financing. We find that CSR
performance is not significantly associated with bond maturity but negatively associated with
covenant intensity. The results suggest that firms with better CSR performance face fewer
covenant restrictions. Taken together, our results suggest that bondholders perceive that firms
4
with better CSR performance are more creditworthy and that both CSR strengths and concerns
are considered. Our findings also imply that rating agencies are efficient at incorporating a firm’s
CSR performance information into credit ratings, but the intangible nature of CSR performance
valued by the market may not be fully captured by credit ratings.
Our study is related to but distinct from recent research on CSR and the cost of equity
capital (Chava, 2014; Dhaliwal et al., 2011; Ghoul et al., 2011; Sharfman and Fernando, 2008).
First, shareholders and debtholders have different claims on a firm’s net assets and thus tend to
have conflicting interests (Ahmed et al., 2002). Shareholders have residual claims on firm net
assets and thus the upside potential of their equity investments is unlimited, while debt providers
have no right to claim an extra payoff when the value of the borrowing firm’s assets exceeds its
contractual debt obligations. However, when firm assets are insufficient to cover contractual firm
obligations, lenders may receive less than the promised payments. Due to this asymmetric payoff
with respect to firm net assets, debt providers are naturally more interested in the downside risk
of their debt investments (Ge et al., 2012). Accordingly, the interests of shareholders and
debtholders can diverge. In addition, Merton’s (1974) theoretical bond pricing model indicates
that shareholders and bondholders value a firm’s operating volatility in different ways. On the
one hand, CSR activities consume resources that could otherwise help generate profits, so CSR-
related expenditure may be considered a poor use of shareholder money. The reduced profits also
mean reduced interest-paying ability and increased distress risk. On the other hand, CSR
activities can reduce environmental violation and litigation risk and help build firm reputation.
Given that debtholders are more sensitive to the downside risk, despite there is evidence that
shareholders view CSR performance positively, it is not clear how public bondholders trade off
the costs and benefits of CSR activities.
5
Second, corporate bond financing represents the major source of external funding for U.S.
corporations (Denis and Mihov, 2003). The total value of U.S. corporate bond issuance between
1992 and 2009 amounted to about $11.2 trillion, while the total value of equity issuance
(including common and preferred stock) for the same period was only about $3.1 trillion. 2
Therefore, understanding the bond market consequences of CSR is important and interesting in
its own merit.
Goss and Roberts (2011) investigate the impact of CSR performance on the cost of
private bank loans. Their findings suggest that banks charge more for loans to firms with social
responsibility concerns; however, banks do not reward firms for their CSR strengths. Our study
focuses on public debt financing, which differs substantially from private debt financing in terms
of monitoring efficiency (Diamond, 1984) and private information availability (Fama, 1985). As
delegated monitors, banks are given access to information about the borrowing firm that may not
be available to outsiders. Because banks are able to engage in more detailed monitoring as well
as tailor loan terms to strengthen monitoring effectiveness, they may be more finely attuned to
any impact of CSR than public lenders are (Goss and Roberts, 2011). On the other hand, since
public bondholders may not be able to obtain firm information through private channels as
private debtholders do, the information asymmetry problem is more severe in the public bond
markets than in the private debt markets. Bondholders may therefore require more voluntary
disclosure channels to overcome the limitations of mandatory disclosures. In this sense,
bondholders are more likely to use all publically available information, including CSR
disclosures, to assess the bond issuer’s litigation and credit risk. Therefore, whether the findings
in the loan markets are applicable to the bond markets is an empirical question. Thus, our study
complements that of Goss and Roberts (2011) and studies on the association between CSR
2
Data source: Securities Industry and Financial Markets Association (http://www.sifma.org).
6
performance and the cost of equity (e.g., Dhaliwal et al. 2011; Ghoul et al. 2011). The findings
from all these studies together will provide a comprehensive picture of the consequences of CSR
activities on the cost of debt and equity capital.
Our study is related to Menz (2010), which examines the relation between the valuation
of euro corporate bonds and the CSR standards of European companies; the author finds that
CSR has not yet been incorporated into the pricing of euro corporate bonds. Our study differs
from Menz (2010) in two important ways. First, Menz (2010) examines European firms, while
we study U.S. firms. Legal institutions in the United States differ from those of European
countries and U.S. firms are subject to considerably higher litigation risk related to information
disclosure (Fulbright and Jaworski, 2005; Seetharaman et al., 2002). Other studies suggest that
there are significant variations in CSR disclosure across countries and that institutional factors,
including culture, are significant in explaining cross-national differences in the level and quality
of CSR disclosure (van der Laan Smith et al., 2005). Maignan and Ferrell (2003) find significant
differences in consumer evaluation of corporate responsibility between the United States, France,
and Germany. Van der Laan Smith et al. (2010) also report systematic cross-national differences
in investment response to CSR disclosure. Thus, the significant differences in institutional
environment and culture between the United States and European countries suggest that the
results of Menz (2010) do not necessarily hold in the U.S. bond market. Second, Menz (2010)
focuses on the effect of CSR disclosure on the cost of corporate debt, while our study examines
the potential impact of CSR performance on U.S. corporate bond issues.
A concurrent study by Oikonomou et al. (2014) also examines the effects of CSR
strengths and concerns on the cost of U.S. corporate bonds and credit ratings. Although that
study’s findings are generally consistent with ours, the two have significant differences with
7
regard to sample construction and research design. First, Oikonomou et al. (2014) use data from
the primary and secondary bond markets, while our study focuses only on the primary bond
market, which provides a clean setting to investigate the effect of CSR performance on initial
bond offerings. Second, focusing on a smaller set of CSR dimensions, Oikonomou et al. (2014)
consider CSR strengths and concerns separately and remain silent on the effect of a firm’s
overall CSR performance. Prior literature indicates that the equity market observes overall CSR
performance, considering both strengths and concerns together, and thus the joint or net effect is
more important (Dhaliwal et al., 2011). In addition to examining the effect of CSR strengths and
concerns on a larger set of CSR dimensions, our study provide insights into the effect of a firm’s
overall CSR performance on its cost of public debt financing. Third, there is a concern about the
endogeneity between CSR performance and bond yield spreads and the joint determination of
bond yield spreads, maturity, and covenants. We use an instrumental variables approach to
address the potential endogeneity issue, which is not studied by Oikonomou et al. (2014).
Furthermore, we perform a battery of additional tests to illustrate the variations across different
subsamples (investment-grade vs. non–investment-grade bonds; Rule 144a vs. non-Rule 144a
bonds; issuers with high financial distress vs. those with low financial distress; issuers with high
information asymmetry vs. those with low information asymmetry; issuers with strong
governance vs. those with weak governance; and issuers operating in environmentally sensitive
industries vs. those operating in other industries), which enlarges the contributions of our study.
Our study contributes to the literature in the following ways. First, our study extends the
CSR literature by providing initial evidence of how a firm’s overall CSR performance is related
to its public debt financing in the U.S. primary bond market. Although there is no evidence that
CSR has been incorporated into the pricing of euro corporate bonds (Menz, 2010), our results
8
suggest that corporate commitment to CSR activities receives public bondholder support in the
United States. Our findings complement those of Goss and Roberts (2011) and suggest that
public debtholders and private lenders may value CSR activities differently. Our study also
complements that of Oikonomou et al. (2014) and suggests that public debtholders incorporate
information about a firm’s overall CSR performance into their initial bond pricing. Furthermore,
we provide evidence that CSR performance is also associated with the indirect cost of debt, that
is, bonds issued by firms with better CSR performance are subject to fewer covenant restrictions.
Second, this study extends the literature on corporate bond financing. Prior research
focuses largely on using financial attributes to explain corporate bond financing patterns
(Bharath et al., 2008; Jiang, 2008; Mansi et al., 2011). Our study focuses instead on the relation
between bond pricing and non-financial information, that is, corporate CSR performance. Our
findings suggest that the information of corporate CSR performance is value relevant to the
corporate bond market and that bondholders value good CSR performance positively. Therefore,
this study provides insight into how the U.S. bond market prices non-financial CSR information.
Finally, our findings have important practical implications. Prior studies show that CSR
performance reduces the cost of equity capital and our findings indicate that firms committed to
good CSR performance have lower costs of public debt financing. These findings suggest that
even though CSR activities consume corporate resources, they also bring benefits to corporations.
Therefore, CSR investments are beneficial not only to society, but also to the corporations
themselves because they reduce the cost of equity and debt capital. Managers make trade-offs in
the costs and benefits of CSR investment. Our evidence could be helpful in making managers
aware of the potential benefits of CSR activities.
9
The remainder of this study proceeds as follows. Section 2 reviews the related literature
and develops our hypotheses. Section 3 describes the sample and methodology. Section 4
presents the main results and subsample test results. Section 5 summarizes the results of
robustness tests. Section 6 sets forth our conclusions.
2. Literature review and hypothesis development
Until recently, there have been relatively few studies on the importance of non-financial
CSR information to the capital markets. The shareholder theory holds that a firm’s top priority is
to maximize firm value. Accordingly, CSR-related expenditure is considered a poor use of
shareholder money and this practice deviates from shareholder value maximization (Friedman,
1962). According to this theory, CSR activities consume resources that could otherwise help
generate profits for shareholders; hence a higher CSR performance could hurt shareholders’
interest. Reduced profits also mean reduced interest-paying ability and increased distress risk,
both of which can be related to higher costs of public debt financing.
On the other hand, the stakeholder theory suggests that corporations should go beyond
the interests of shareholders and consider the interests of a broader group of stakeholders
(Freeman, 1984). Jones (1995) extends the stakeholder theory by indicating that CSR is essential
for corporations in obtaining necessary resources and stakeholder support. From the stakeholder
theory perspective, CSR performance may reduce the cost of capital for two reasons. First, CSR-
related information captures the dimension of a corporation’s environmental risk and decreases
information asymmetry, which could eventually affect the cost of equity (e.g., Heinkel et al.,
2001; Hong and Kacperczyk, 2009; Jones, 1995; Orlitzky et al., 2003). Jensen and Meckling
(1976) indicate that any information that mitigates information asymmetries between contracting
parties, thereby reducing adverse selection and moral hazard problems, would be value relevant.
10
A large body of literature documents that financial disclosures provide value-relevant
information and thus reduce the cost of capital (e.g., Botosan, 1997; Diamond and Verrecchia,
1991; Graham et al., 2005; Healy and Palepu, 2001; Leuz and Verrecchia, 2000). Consistent with
this argument, Orlitzky et al. (2003) suggest that CSR-related information is value relevant.
Hong and Kacperczyk (2009) find that socially irresponsible firms may have higher litigation
risk, such as from lawsuits regarding environmental pollution, unsafe products, and employee
benefits. Litigation risk is likely to result in negative future cash flows and decrease a firm’s
interest-paying ability. Accordingly, investors, including bondholders, will discount the value of
irresponsible firms. In other words, the cost of bond financing may be lower for firms with good
CSR performance, compared to that for irresponsible firms.
Second, CSR performance can increase the size of the investor base. Heinkel et al. (2001)
argue that socially conscious investors prefer to exclude firms with low CSR performance from
their investment portfolios. In other words, firms with high CSR performance can increase the
relative size of their investor base. The larger the relative size of the investor base, the lower the
cost of capital and the higher the market valuation (Merton, 1987).
A few recent studies investigate the relation between CSR performance and the cost of
capital.3 Overall, their findings are consistent with the stakeholder theory. For example, Dhaliwal
et al. (2011) find that better CSR performers enjoy a significant reduction in the cost of equity
capital when they issue standalone CSR reports for the first time. Ghoul et al. (2011) and
Plumlee et al. (2014) find that firms with better environmental performance have lower costs of
3
Several studies examine the association between social disclosures and the cost of capital and the findings are
mixed. Using a sample of Canadian firms, Richardson and Welker (2001) find that social disclosure increases the
cost of equity capital. However, other studies do not find an association between social disclosure and the cost of
capital or firm reputation. For example, Clarkson et al. (2013) do not find evidence that voluntary environmental
disclosures affect the cost of equity capital. Linthicum et al. (2010) document that CSR did not mitigate negative
market reactions to the corporate reputations of Arthur Andersen clients following the Enron audit failure. Unlike
these studies, our study focuses on CSR performance.
11
equity capital. Focusing on private debt financing, Goss and Roberts (2011) find that the cost of
bank loans for firms with social responsibility concerns is higher than for more responsible firms;
however, banks do not reward the CSR investments of borrowers.
Public debt financing differs from private debt financing in terms of monitoring
efficiency (Diamond, 1984; Rajan, 1992) and private information availability (Bhattacharya and
Chiesa, 1995; Fama, 1985). Diamond (1984) indicates that a bank, as a financial intermediary,
can solve the information duplication and free-rider problems by monitoring borrowers. In
addition, Rajan (1992) argues that concentrated bank debt lenders have more incentive to
monitor borrowers compared to dispersed, “arm’s length” debtholders. Fama (1985) argues that
private debt lenders are more efficient and effective in obtaining private information about
borrowers than public bondholders are. Similarly, Bhattacharya and Chiesa (1995) develop an
analytical model that supports the view that borrowers share proprietary information with a
concentrated group of private lenders but not with diffuse public lenders. These findings imply
that private debt financing mitigates the information asymmetry between borrowers and lenders,
whereas public bondholders may not be able to obtain firm information through private channels.
Due to these monitoring and information disadvantages, information about corporate CSR
performance is likely more value relevant to public debtholders than to private lenders. More
specifically, from the stakeholder theory perspective, we expect bondholders to require lower
risk premiums for firms with better CSR performance, in general, and we expect CSR strengths
to be associated with lower bond yield spreads and CSR concerns to be associated with higher
bond yield spreads, in particular. We state our hypotheses in alternative form, as follows,
H1: Ceteris paribus, CSR performance is negatively associated with bond yield
spreads.
12
H2: Ceteris paribus, CSR strengths (concerns) are negatively (positively)
associated with bond yield spreads.
As mentioned earlier, if the shareholder theory holds, we will observe CSR performance
to be associated with higher bond yield spreads. If bondholders do not use CSR information in
their bond pricing, we will observe no association between CSR performance and bond yield
spreads.
3. Research design
3.1. Sample selection
We obtain CSR rating scores from the KLD STATS database, which provides ratings for
13 dimensions of corporate CSR performance: environment, community, product, diversity,
employee relations, human rights, governance, alcohol, gambling, firearms, military, nuclear
power, and tobacco. The KLD STATS database provides ratings for strengths and concerns in
qualitative issue areas (the first seven dimensions); controversial business issues (the last six
dimensions) are exclusionary screens, with only ratings for concerns available. Following prior
studies (e.g., Dhaliwal et al., 2011), our analysis focuses on the first seven dimensions. 4
Sharfman (1996) checks the construct validity of the KLD ratings and confirms that they
represent the core of the social performance construct. The KLD STATS database is recognized
as one of the largest and most influential databases for CSR measures and it has been used
extensively in academic research (e.g., Dhaliwal et al., 2011; Ghoul et al., 2011; Goss and
Roberts, 2011).5
4
We exclude firm’s strength and concern scores for compensation (in the governance dimension) because these
items are not related to CSR. The results are qualitatively similar when these items are included.
5
However, the KLD STATS database has limitations. For example, it does not cover detailed information about the
length of CSR reports.
13
We collect bond-specific information from the Mergent Fixed Income Securities
Database. We retrieve bond issuer financial information from the Compustat North America
database. We eliminate variable coupon and zero coupon bonds, as well as perpetual bonds,
because these bonds tend to be unique and behave more like equities (Bessembinder et al., 2009).
Since financial firms operate under different regulations and have different debt financing
characteristics than industrial firms (Jiang, 2008; Khurana and Raman, 2003), we exclude bonds
issued by public financial firms (Standard Industrial Classification codes 6000–6999). After
merging the data collected from the above three databases and eliminating observations with
missing values, we obtain 4,260 observations from 2,317 firms that disclosed CSR information
during the period 1992–2009 to test our hypotheses.6 To control for the effect of outliers, we
winsorize all continuous variables at the top and bottom 1%.
3.2. Empirical models
Based on prior research (Jiang, 2008; Khurana and Raman, 2003; Mansi et al., 2011;
Sengupta, 1998; Shi, 2003; Ziebart and Reiter, 1992), we use the following empirical models to
test the association between CSR performance and the yield spreads of new corporate bond
issues:
YieldSpreadijt = β0+ β1CSRTit + β2IssuerSizeit + β3ROAit + β4Leverageit +

β5Big4it+ β6Z-scoreijt + β7lnMaturityijt + β8Covenantijt + β9IssueSizeijt +
β10lnRatingijt + β11BusiCycleijt + β12RedeemDijt + β13PutDijt + β14ConvertDijt +
β15415RegDijt + β16144aRegDijt + Industry & Year Indicators + εijt (1)
YieldSpreadijt = β0+ β1STRit + β2 CONit + β3IssuerSizeit + β4ROAit + β5Leverageit

+ β6Big4it + β7Z-scoreijt + β8lnMaturityijt + β9Covenantijt + β10IssueSizeijt +
β11lnRatingijt + β12BusiCycleijt + β13RedeemDijt + β14PutDijt + β15ConvertDijt +
β16415RegDijt + β17144aRegDijt + Industry & Year Indicators + εijt (2)
6
Relatively small sample sizes are common for studies pertaining to bond markets (Dhillon and Johnson, 1994; Shi,
2003). One possible explanation is that a large portion of bond issuers are non-public firms.
14
Dependent variable
Following previous studies, we measure the dependent variable YieldSpread as the
difference between the corporate bond yield at issuance and a Treasury bond yield with
comparable maturity 7 (Jiang, 2008; Shi, 2003; Wang and Zhang, 2009). The subscript ijt
indicates bond j for firm i in year t. Bond yield spreads capture the risk premiums that bond
issuers pay to bond investors to raise funds from the corporate bond market. Treasury bonds are
issued by national governments. Because government bonds are backed by the high-quality
credit and taxing power of a nation, they bear very little credit risk. Thus, the variable
YieldSpread is a direct and accurate measure of issuers’ incremental cost of a bond over a
comparable risk-free Treasury bond. By subtracting comparable Treasury bond yields from
corporate bond yields, we also control for the effect of economy-wide information.
Test variables
The test variables include proxies for CSR performance, strengths, and concerns. In
model (1), the variable CSRT is a corporation’s overall CSR performance score. It is measured as
the total CSR strength score (STR) minus the total CSR concern score (CON). We use model (1)
to test H1. Model (2) includes STR and CON as test variables. We use model (2) to test H2.
Firm-level control variables
IssuerSize: This is the natural logarithm of an issuer’s assets at the fiscal year-end
immediately prior to the corporate bond issuance date. Issuers with larger assets are perceived to
be less risky (lower default risk) than those with smaller assets. Hence, we expect IssuerSize to
be negatively related to the risk premium.
7
If the benchmark Treasury bond yield with a certain maturity is not available, we use an interpolation approach to
construct it. For example, suppose we have only a Treasury bond yield with maturities of five years (2.2%) and
seven years (2.4%) but we need a Treasury bond yield with a maturity of six years as a benchmark to calculate the
yield spreads of corporate bonds with a maturity of six years. In this case, we construct the benchmark Treasury
bond yield with a maturity of six years as follows: (2.2% + 2.4%)/2 = 2.3%.
15
ROA: This is an issuer’s return on assets, defined as net income divided by total assets at
the fiscal year-end immediately prior to the corporate bond issuance date. A higher return on
assets generally implies greater profitability. Thus, we expect ROA to be negatively related to the
risk premium.
Leverage: This is long-term debt divided by total assets at the fiscal year-end
immediately prior to the corporate bond issuance date. We predict that Leverage will be
positively related to the risk premium.
Big4: This is an indicator variable for auditor type. It takes the value of one for Big Four
auditors and zero otherwise. The variable Big4 is a common proxy for audit quality or earnings
quality; thus we expect its coefficient to be negative.
Z-score: This is the Altman (1968) Z-score, a measure of a corporation’s financial
strength. The Altman Z-score is calculated as follows: 1.2 × (working capital/total assets) + 1.4 ×
(retained earnings/total assets) + 3.3 × (earnings before interest and taxes/total assets) + 0.6 ×
(market value of equity/book value of total debt) + 1.0 × (sales/total assets). The higher Z-score
is, the lower the default risk. Hence, we expect Z-score to be negatively related to the risk
premium.
Bond-level control variables
lnRating: This is the natural logarithm of Standard & Poor’s bond ratings. We convert
Standard & Poor’s bond ratings sequentially to numbers, with one for AAA through 27 for no
rating. For issues without Standard & Poor’s ratings, we use Moody’s or Fitch ratings instead. A
bond rating indicates an issue’s creditworthiness; so we expect it to be positively related to the
risk premium (Jiang, 2008; Shi, 2003).
16
lnMaturity: This is the natural logarithm of the number of years to bond maturity. Usually,
bond issues with longer maturities are riskier than issues with shorter maturities (Khurana and
Raman, 2003; Shi, 2003). Thus, we expect the variable lnMaturity to be positively related to the
risk premium.
Covenant: This is a covenant index. Following Nikolaev (2010), we consider 36 covenant
restrictions (payout related, investment related, financing related, accounting related, and other
covenants) and count the number of bondholder protective covenants in a bond issue. Consistent
with Billett et al. (2007), we divide the number of bondholder protective covenants in a bond
issue by the maximum number of bondholder protective covenants in our sample to create an
index (Covenant) that varies from zero to one. Bondholders can charge higher yield spreads and
impose more restrictive covenants to bonds issued by less trustworthy firms; so we predict a
positive relation between YieldSpread and Covenant.
IssueSize: This is the natural logarithm of the par value of an initially issued bond, in
millions of dollars. A larger issue size can enjoy a lower risk premium due to economies of scale
in underwriting (Sengupta, 1998). However, Khurana and Raman (2003) point out that a large
issue size also increases underwriters’ difficulty in placing the issue with investors. On the basis
of cross-sectional observations spanning 20 years, Wang and Zhang (2009) find that issue size is
positively associated with bond yield spreads. Thus, we do not make predictions for the sign of
IssueSize.
BusiCycle: This is the business cycle variable, that is, the difference between the average
yield of Moody’s Aaa bonds and the average yield of 10-year U.S. Treasury bonds for the month
of issue. This variable controls for the time-series variation of risk premiums over the business
17
cycle. Prior studies predict that the variable BusiCycle will be positively related to the risk
premium (Jiang, 2008; Sengupta, 1998; Shi, 2003).
RedeemD: This is an indicator variable for a bond’s call feature. It equals one for bonds
that have a call option and zero otherwise. A redeemable bond offers issuers the option to
repurchase the bond before maturity. It increases the potential interest risk for bondholders, so
we expect the variable RedeemD to be positively related to the risk premium.
PutD: This is an indicator variable for a bond’s put feature. It equals one for bonds with a
put option and zero otherwise. Putable bonds offer bondholders the option to retire the bond
before maturity and thus we expect this variable to be negatively related to the risk premium.
ConvertD: This is an indicator variable for a bond’s convertible feature. It equals one for
convertible bonds and zero otherwise. Mayers (1998) suggests that reducing the interest rate is
an incentive for firms to issue convertible bonds. We expect this variable to be negatively related
to the risk premium.
415RegD: This is an indicator variable for U.S. Securities and Exchange Commission
(SEC) Rule 415 shelf registration bonds. We assign the value of one for bonds issued under a
shelf registration and zero otherwise. According to Rule 415, issuers are allowed to pre-register a
certain securities (e.g., equities and bonds). In the case of bonds, issuers have the option to take
bonds “off the shelf” and offer them to the public at a favorable time up to two years into the
future. Therefore, we expect the coefficient of this variable to be negative.
144aRegD: This is an indicator variable for U.S. SEC Rule 144a private placement bonds.
It equals one for bonds issued through private placements that are exempt from registration and
zero otherwise. Rule 144a issues are generally offered to a limited number of institutional
investors, known as qualified institutional buyers. Since institutional investors have stronger
18
negotiation power than public investors regarding the coupon rate, we expect this variable to be
positively related to the risk premium. On the other hand, similar to private lenders, institutional
investors could have some informational advantages; therefore, this variable can be negatively
associated with the risk premium. We do not make a prediction for this variable.
Industry and year indicator variables
In line with the literature, we include Fama and French’s 48 industry indicators and year
indicators in the above regression models to further control for potential differences in issuer and
issue features across industries and over time (e.g., Fama and Fench, 1997; Shi, 2003).
4. Empirical results
4.1. Univariate analysis
Table 1 provides descriptive statistics for our sample observations by year. The sample
period is from 1992 to 2009. Our sample contains 4,260 distinct new bond issues from 2,317
firms.
[Insert Table 1 about here]
Table 2 summarizes descriptive statistics of the key variables. Since our analyses require
firm- and bond-level variables, the number of observations used to calculate the descriptive
statistics differs between firm- and bond-level variables. The average yield spread is 154.13 basis
points (about 1.54%).8 On average, there are three covenants in a bond issue. The median credit
rating is 12, which means that more than half of new bond issues are non-investment grade, as
rated by Standard & Poor’s. About 7% and 14% of new bond issues have a put or convert option,
8
We have 502 observations with negative yield spreads. The negative yield spread could be justified by the putable,
convertible, or other features of the new bond issues. For example, putable bonds give bondholders the option to
retire the bond before maturity, an additional benefit for them. Thus, issuers can sometimes issue putable bonds by
offering yields that are lower than the comparable U.S. Treasury yield. In addition, Mayers (1998) suggests that
reducing the interest rate is one incentive for firms to issue convertible bonds. After removing putable and
convertible bonds from our sample, only 70 observations with negative yield spreads remain. The negative yield
spreads may be a result of some other bond features that are favorable to bondholders.
19
respectively. In addition, 70% of new bond issues are issued under shelf registration and 24% are
issued through private placement. The average scores for the three CSR performance proxies—
CSRT, STR, and CON—are 0.05, 2.28, and 2.23, respectively. The average ROA and leverage
ratio of the bond issuer are 3% and 26%, respectively. The majority of our sample firms (98%)
are audited by Big Four auditors.
Table 3 presents a correlation matrix of the key variables used in the regression analyses.
As expected, the variable YieldSpread is negatively correlated with the total CSR performance
score (CSRT) and strength score (STR) and positively correlated with the CSR concern score
(CON). Unexpectedly, the strength and concern scores are positively correlated. Many firms
have multidimensional performance goals. The use of a firm’s resources to engage in CSR
activities always has an opportunity cost. Implementing a strategy for social issue participation
may come at the cost of forgone opportunities to increase shareholder value (Hillman and Keim,
2001). Therefore, a firm may do better in some CSR dimensions than in others. A firm with more
CSR concerns may be more likely to commit to improving its CSR performance and developing
strengths in some dimensions but may not have enough resources to address concerns in other
dimensions. This could be the reason why we observe a positive correlation between STR and
CON. In terms of the other control variables, consistent with our predictions, YieldSpread
correlates negatively with ROA, Big4, Z-score, and lnMaturity and positively with Leverage.
Inconsistent with our predictions, YieldSpread correlates positively with lnRating, Covenant, and
IssuerSize.
20
4.2. Main results
In our model specifications, we use the initial bond yield spread as a proxy for the cost of
public debt. The bond yield spread represents the risk premium that a firm must pay for bond
financing and is thus a direct measure of the firm’s incremental cost of public debt. The cost of
debt is a function of default risk and a firm’s credit rating reflects its default risk and is therefore
used in prior studies as a proxy for the ex ante cost of public debt (e.g., Jiang, 2008; Ge and Kim,
2014). Ideally, credit rating agencies are efficient in incorporating the information of a firm’s
CSR performance. If credit ratings fully reflect all available and relevant information pertaining
to a bond issue’s default risk, then CSR performance should have no incremental explanatory
power after credit ratings are controlled for in our models. To investigate this issue, we carry out
our tests in the following steps. First, we examine the effect of CSR performance on bond
ratings; second, we examine our hypotheses by estimating our models excluding bond ratings;
and, third, we re-examine our hypotheses including bond ratings.
Table 4 summarizes the ordinary least squares (OLS) regression results. According to
Petersen (2009), the standard errors calculated by an OLS regression for panel data may be
biased due to residual correlations. Thus, we correct the standard errors of the OLS regression
coefficients for firm-level clustering, as well as for heteroskedasticity.
Panel A of Table 4 reports the results of testing the effect of overall CSR performance.
Column (1) reports the results when bond rating (lnRating) is the dependent variable. By
construction, the lower value of lnRating represents a better bond rating. The coefficient of the
overall CSR performance score (CSRT) is negative (-0.01) and significant at the 5% level,
suggesting that bond ratings incorporate information about a firm’s CSR performance. Column
(2) of Panel A reports the results of testing the effect of CSR performance on yield spreads,
21
without controlling for bond ratings. The coefficient of CSRT is 3.67, significant at the 1% level,
consistent with H1, that firms with better CSR performance can issue bonds at lower yield
spreads. Column (3) of Panel A presents the results of testing H1 including bond ratings, as
specified in model (1). The coefficient of CSRT is negative and significant at the 1% level, but its
magnitude becomes smaller (-3.20) than -3.67 in column (2). The difference is statistically
significant at the 5% level (Chi-square = 3.77). The results indicate that good CSR performance
is valued by bondholders, but some effect of CSR performance on yield spreads is absorbed by
bond ratings. The results also imply that the intangible nature of good CSR performance is
appreciated by the markets but may not be fully captured in credit ratings. The result is
economically significant as well. Table 2 shows that mean offering amount is $432.80 million
and average maturity is about 12 years. Thus, holding other variables unchanged, on average, a
one-point increase in the overall CSR performance score is associated with a decrease of 3.20
basis points in the yield spread, which can save a firm a total interest cost of $1.66 million per
bond issue (432.80 × 3.20 ÷10000 × 12 = 1.66).
Furthermore, we follow Ashbaugh-Skaife et al. (2006) and collapse bond ratings into a
seven-point scale, with one for a AAA rating; two for AA+, AA, and AA-; three for A+, A, and
A-; four for BBB+, BBB, and BBB-; five for BB+, BB, and BB-; six for B+, B, and B-; and
seven for CCC+, CCC, CC, C, D, and SD. We perform an ordered logistic regression of bond
rating groups (RatingGroup) on the same set of explanatory variables. The results for CSRT
(-0.09, p < 0.01; not tabulated here for brevity) are consistent with the OLS results as reported in
column (1). We also re-estimate model (1) by replacing lnRating with RatingGroup (column 3 in
Panel A). After controlling for RatingGroup, the coefficient of CSRT is -2.91 (p = 0.01) and the
22
difference between this coefficient without and with RatingGroup in the model (-3.67 vs. -2.91)
is statistically significant (Chi-square = 5.73; p < 0.05).
Panel B of Table 4 summarizes the results of testing the effect of CSR strengths and
concerns separately (H2). As shown in column (1) of Panel B, the results indicate that a higher
CSR strength score (STR) is associated with a better bond rating, but there is no significant
association between the CSR concern score (CON) and bond ratings. In column (2), the
coefficient of STR is negative and significant and the coefficient of CON is positive and
significant. Column (3) presents the results after controlling for bond ratings. The coefficient of
STR becomes smaller and the coefficient of CON remains significantly positive. By comparing
the results in columns (2) and (3), the difference in the coefficient of STR is statistically
significant (Chi-square = 6.93; p < 0.01), while the difference in the coefficients of CON is not
significant (Chi-square = 0.83; p = 0.36). The results are robust when an ordered logit regression
model is used and when we replace lnRating in column (3) with RatingGroup. The results in
column (3) of Panel B suggest that, after bond ratings and other control variables are controlled
for, on average, a one-point increase in the total CSR strength score is associated with a decrease
of 2.42 basis points in the yield spread and a one-point increase in the total CSR concern score is
associated with an increase of 4.45 basis points in the yield spread.
In Table 4, most firm- and bond-level control variables have the expected signs for their
coefficients. For brevity, our discussion focuses on the coefficients in column (3) of Panel A.
The coefficients of IssuerSize, ROA, Big4, and Z-score are negative and significant at the 1%
level, suggesting that larger firms, more profitable firms, clients of Big Four auditors, and firms
with lower default risk can issue bonds at lower cost. The coefficient of Leverage is positive and
23
statistically significant. Overall, the results for these control variables are consistent with prior
studies (e.g., Jiang, 2008; Khurana and Raman, 2003; Shi, 2003).
As for the bond-level controls, the coefficient of credit rating (lnRating) is significantly
positive (46.71; p < 0.01). This finding reinforces Jiang’s (2008) argument that the credit rating
captures the creditworthiness of an issue. Unexpectedly, the coefficient of IssueSize is negative
but not statistically significant. Consistent with prior studies (e.g., Jiang, 2008; Sengupta, 1998),
the coefficient of the business cycle variable (BusiCycle) is positive and significant, suggesting
that issuers pay higher risk premiums for new bond issues when there is a larger difference
between Moody’s Aaa bond yields and the 10-year U.S. Treasury bond yield for the same month.
As predicted, the coefficient of the redeemable indicator variable (RedeemD) is significantly
positive, while the coefficients of the putable (PutD), convertible (ConvertD), and 415 regulation
(415RegD) indicators are significantly negative. These results suggest that bond issuers pay
higher risk premiums for bonds that have redeemable features but pay lower risk premiums for
bonds that have putable and convertible features and for bonds that are issued under shelf
registration. The positive coefficient of 144aRegD suggests that the bonds issued through private
placements have a higher coupon rate.
Overall, the independent variables in these two model specifications (column 3 of Panels
A and B in Table 4) explain about 62% of the variance of the dependent variable (YieldSpread),
and both models are significant at the 1% level.
4.3. Relation between firm scores in individual CSR dimensions and yield spreads
As mentioned in Section 3.1, the KLD STATS database provides ratings for strengths
and concerns in seven qualitative issue areas: environment (pollution prevention, recycling, clean
24
energy, etc.), the community (charitable giving, innovative giving, community engagement,
widespread or egregious community impacts due to company operations, etc.), the product
(product quality and safety, customer relations, etc.), diversity (women and minority contracting,
gay and lesbian policies, the employment of underrepresented groups, etc.), employee relations
(union relations, cash profit sharing, employee involvement, health and safety, etc.), human
rights (human rights policies and initiatives, operations in Burma or Sudan, etc.), and governance
(reporting quality, public policy, governance structure, business ethics, etc.).
To provide more insight on how firm performance in individual CSR dimensions relates
to bond yield spreads, we replace CSRT in model (1) with the firm score in each of these seven
CSR dimensions. The firm score in each individual CSR dimension is equal to the difference
between the total strength score and the total concern score on the items included in each
dimension. The results, reported in Panel A of Table 5, show that six out of seven dimension
scores (the exception being human rights) are negatively associated with bond yield spreads, but
only four of these associations (Community, Product, Employee Relations, and Governance) are
statistically significant at the 5% or 10% level. These results suggest that the association between
overall CSR performance and yield spreads as reported in Table 4 is primarily driven by a firm’s
overall performance in four dimensions, namely, community, product, employee relations, and
governance.
Next, we test the association between strengths and concerns by subcategory and yield
spreads. Panel B of Table 5 summarizes the results. We find that the estimated coefficients of the
strength scores in all dimensions are negative but only three strength scores on environment,
community, and governance are significant at the 5% or 10% level. The estimated coefficients of
the concern scores (except for human rights) are positive and the concern scores in three
25
dimensions—product, diversity, and employee relations—are significant at the 5% or 10% level.
For human rights, neither the coefficient of the strength score nor the coefficient of the concern
score is statistically significant. The descriptive statistics (not tabulated here) show that 99% of
our sample has no human right strengths and more than 87% of our sample has no human right
concerns. Thus, our sample may lack the power to detect the effect of human rights strengths and
concerns on the cost of debt.
Hillman and Keim (2001) suggest that building better relations with primary
stakeholders—such as government, employees, customers, suppliers, and communities—could
help firms develop intangible, valuable assets that can be sources of competitive advantage.
Consistent with this proposition, our results suggest that bondholders reward bond issuers that
demonstrate their legal responsibility to the government (captured by the environment
dimension), discretionary responsibility to the community, and their commitment to improving
corporate governance and penalize bond issuers that have employment concerns (captured by
two dimensions, diversity and employee relations) and product safety and quality concerns.
Bondholders may view product quality and safety, good customer relations, unbiased
employment, and good employee relations as a “must-have” of any corporation and thus they are
not surprised by an issuer’s strengths in these dimensions. This could explain why the strength
scores on product, diversity, and employee relations are not statistically significant. However, it
is surprising that the coefficients of environment and community concerns are not statistically
significant, since environment and community concerns could imply some litigation risk.
26
The results reported in Tables 4 and 5 are generally consistent with the stakeholder
theory.9 Taken together, our results indicate that both CSR strengths and concerns are considered
by bondholders and overall CSR performance matters in initial bond pricing.
4.4. Issues related to corporate governance
Ghoul et al. (2011) argue that CSR performance measures should not include conflicts of
interest between insiders and shareholders, so they exclude CSR strength and concern scores that
relate to corporate governance to construct CSR measures. As a robustness check, we reconstruct
the CSR performance measures (CSRT, STR, and CON) after excluding the governance
dimension. We repeat our analyses using these new measures. Overall, the results (not tabulated
here for brevity) are consistent with those reported in Table 4, except that the negative
coefficient of the strength score becomes significant at the 10% level.
Several studies find that a firm’s CSR performance incorporates its governance quality
(e.g., Coffey and Wang, 1998; Haniffa and Cooke, 2005; Stephen et al., 2010), which is one of
the determinants of the cost of debt (e.g., Anderson et al., 2004; Ashbaugh-Skaife et al., 2006;
Klock et al., 2005). As an additional test, we examine whether the effect of CSR performance on
yield spreads still holds after controlling for conflicts of interest between insiders and
shareholders. Gompers et al. (2003) developed a governance index (G-index) of agency conflicts
between managers and shareholders using various anti-takeover–related activities that restrict
shareholder rights. We obtain G-index data from Andrew Metrick’s website.10 Since G-index is
available only for some of our sample firms in certain years (1990, 1993, 1995, 1998, 2000, 2002,
2004, and 2006), we assume that the G-index remains unchanged until it is updated; adding G-
9
Untabulated results show that CSR performance is positively associated with ROA, cash flows from operations,
and the Z-score in year t + 1, suggesting that better CSR performance in year t is associated with higher profitability
and fewer financial risks in the future. This could explain why CSR performance is associated with lower yield
spreads.
10
See http://faculty.som.yale.edu/andrewmetrick/data.html, last accessed April 2, 2014.
27
index data reduces our sample size from 4,260 to 3,797 observations. We re-estimate models (1)
and (2) for the reduced sample. The results (not tabulated for brevity) show that the coefficient of
G-index is not statistically significant and, after controlling for G-index, the results on all our
CSR performance measures are consistent with our main findings. When we use alternative
measures of CSR (excluding the governance dimension), the results are consistent, except that
the negative coefficient of the strength score is significant at the 10% level.
After the passage of the Sarbanes–Oxley Act (SOX), a firm’s governance quality has
received more attention and many U.S. public firms have improved their corporate governance to
comply with SOX. Firms with weaker corporate governance have a worse information
environment than those with stronger corporate governance and thus creditors are more likely to
use all publically available information, including CSR disclosures, to assess bond issuers’
litigation and credit risk. Next, we examine whether the effect of CSR performance is more
pronounced in firms with weaker governance in the pre-SOX (1992–2001) and post-SOX (2002–
2009) periods.
We partition our samples based on the median value of G-index in the pre- and post-SOX
periods. A higher G-index value indicates weaker shareholder rights, that is, poor corporate
governance; so, the subsample with weak (strong) governance contains firms whose G-index is
above (below) the median value. We focus on overall CSR performance and re-estimate model
(1) for these paired subsamples. As shown in Table 6, in the pre-SOX period, the coefficient of
CSRT is negative and significant in both subsamples, but the result is stronger in the subsample
with weak governance. In the post-SOX period, the coefficient of CSRT is negative and
significant in the subsample with weak governance only. The results are robust when we use
alternative measure of CSRT that excludes the governance dimension (not tabulated here for
28
brevity). Our results indicate that bondholders are more likely to incorporate CSR performance
into their assessment of default risk of bonds issued by firms with weaker corporate governance,
especially since the passage of SOX.
4.5. Additional subsample analyses
In this subsection, we perform further analyses to provide corroborative evidence
regarding the cross-sectional variation of the association between a firm’s CSR performance and
bond yield spreads. All the subsample analyses focus on overall CSR performance.
First, we examine whether there is a variation in the association between CSR
performance and bond yield spreads between investment-grade bonds (BBB or higher) and non–
investment-grade bonds (BB or lower). Specifically, we estimate model (1) for these two
subsamples excluding bond ratings first and then re-estimate model (1) including bond ratings.
As shown in columns (1) and (2) of Table 7, the coefficient of CSRT is negative and significant
for investment-grade bonds, which is consistent with the findings in Oikonomou et al. (2014),
while the negative coefficient of CSRT is not significant for non–investment-grade bonds. The
results, after controlling for bond ratings, are consistent but not tabulated here for brevity.
Oikonomou et al. (2014) argue that the issuers of low-rated bonds can benefit most from the
significant reductions in the cost of debt that can materialize mainly through proactive
involvement in CSR practices. Their findings on very low-rated bonds are consistent with this
argument. As an additional analysis, we run the regression for highly speculative-grade bonds
(B+ or lower). The results indicate that better CSR performance is associated with lower yield
spreads when bond ratings are excluded from the model; however, such an association becomes
insignificant after controlling for bond ratings (not tabulated for brevity). Our results suggest that
29
CSR performance is especially important for investment-grade bonds and credit ratings absorb
the effect of CSR performance for highly speculative-grade bonds.
Second, we separate bond issuers based on their level of financial distress as proxied by
Altman’s (1968) Z-score and re-estimate model (1) (excluding Z-score as a control variable) for
subsamples with Z-score below the median value versus subsamples with Z-score above the
median value, respectively. The variable Z-score captures a firm’s financial strength. The higher
the Z-score, the lower the financial distress. As reported in columns (3) and (4) of Table 7, we
find that the effect of CSRT on the bond yield spread is slightly stronger in the subsample with
high Z-score (-3.43, p < 0.01) than in the subsample with low Z-score (-2.89, p < 0.10). CSR
activities consume a corporation’s economic resources. Since CSR-related expenditures may
increase the default risk of firms in greater financial distress (lower Z-score), bondholders would
value CSR performance less for such bond issuers. Financially healthier firms usually have fewer
constraints when participating in CSR-related activities and the market may attach a higher value
to their CSR commitments. Consistent with this argument, our findings indicate that bondholders
pay more attention to the CSR performance of financially healthier bond issuers.
Third, we run the regression for Rule 144a and non-Rule 144a bonds, respectively. Rule
144a bond issues are generally offered to a limited number of institutional investors. Institutional
investors have informational advantages over other public bondholders, so a firm’s CSR
performance could be less value relevant to institutional investors. We generate two subsamples,
one with Rule 144a bonds and the other one without Rule 144a bonds, and run model (1)
(excluding the indicator variable 144aRegD) for each subsample. As shown in columns (5) and
(6) of Table 7, we find that the coefficient of CSRT for non-Rule 144a bonds is consistent with
our primary results. For Rule 144a bonds, the coefficient of CSRT is not statistically significant.
30
The results indicate that, compared with institutional investors investing in Rule 144a bonds,
bondholders investing in other bonds are more likely to use information about a firm’s CSR
performance in their bond pricing to overcome their informational disadvantages.
Fourth, we analyze whether the association between CSR performance and the cost of
debt is more pronounced for issuers with higher information asymmetry. Similar in spirit to Gu
and Li (2007) and Huddart and Ke (2007), we measure firm-level information asymmetry using
the amount of intangible assets scaled by total assets (Intangible). The cost of developing
intangible assets is not capitalized but, rather, expensed, according to accounting standards,
which prevents firms from communicating the value of intangible assets through selective
capitalization. We expect that firms with many intangible assets are growth firms and use the
variable Intangible as a proxy for information asymmetry. We split our sample based on the
median value of Intangible. A higher Intangible value represents informational opaqueness and a
higher level of information asymmetry; so firms in the high-information (low-information)
asymmetry subsample are those whose Intangible is above (below) the median. Columns (7) and
(8) of Table 7 present the results. We find that the coefficient of CSRT is negative and significant
in the subsample with high information asymmetry and not significant in the subsample with low
information asymmetry. The results suggest that when a firm has worse information environment,
bondholders are more likely to incorporate information about its CSR performance into the
assessment of its credit risk.
Last, we separate bond issuers based on whether they are operating in environmentally
sensitive industries or in non-environmentally sensitive industries and perform the regression
analysis for each subsample. Firms operating in environmentally sensitive industries face greater
environment-related exposure to and scrutiny by the markets than firms from non-
31
environmentally sensitive industries (Cho and Patten 2007). According to Clarkson et al. (2004),
firms from environmentally sensitive industries are more likely to have latent environmental
liabilities due to future capital spending obligations to comply with more stringent environmental
policies. Accordingly, we expect that positive CSR performance would be more value relevant to
offsetting the negative valuation of environmentally sensitive industries by the markets.
Following prior literature, we categorize a bond issuer as being in the environmentally sensitive
industry subgroup if it operates in one of the following industries: oil and gas; forestry, pulp, and
paper; energy; chemicals and drugs; mining and resources; and utilities; otherwise, we categorize
bond issuers as being in the non-environmentally sensitive industry subgroup (e.g., Aerts et al.
2006; Cho and Patten 2007). As summarized in columns (9) and (10) of Table 7, the coefficient
of CSRT for the environmentally sensitive industry subsample is consistent with our primary
results, while the coefficient of CSRT for the non-environmentally sensitive industry subsample
is not statistically significant. The results indicate that bondholders pay more attention to the
CSR performance of bond issuers operating in environmentally sensitive industries and are more
likely to use this information in their bond pricing.
5. Robustness checks
5.1. Instrumental variables approach
Our test models have a potential endogeneity problem. For example, if firms with better
performance are more active in CSR activities, the association between CSR performance and
bond yields may be a simple manifestation of a latent variable’s effect. It is possible that other
omitted variables may be driving the observed associations. Another concern is that bond yield,
32
bond maturity, and covenants could be simultaneously determined. We use instrumental variable
regression GMM estimators to address these concerns.
Our choice of instrumental variables for CSR performance follows Deng et al. (2013).
Angelidis and Ibrahim (2004) find that the degree of religiousness is positively correlated with
attitudes toward CSR. Rubin (2008) finds that firms with high CSR ratings tend to be located in
Democratic or blue states. Consistent with Deng et al. (2013), we use a religion rank
(ReligionRank) and a blue state indicator (Blue) as instrumental variables for the overall CSR
performance score (CSRT). The variable ReligionRank measures the religion ranking of the state
in which the acquirer’s headquarters is located, which ranges between one and 50. The ranking is
based on the ratio of the number of religious adherents in the acquirer’s state to the total
population in that state in 2000. A higher ranking indicates greater religiosity. The variable Blue
is a dummy variable that equals one if a firm’s headquarters are located in a blue or Democratic
state and zero otherwise.11
Our choice of instrumental variables for yield spread and bond maturity follows Bharath
et al. (2011). Specifically, given that contemporaneous default spreads (DefaultSpread) should
affect yield spreads at the time of pricing and the average yield spread of bonds issued over the
previous six months (AvgYieldSpread) captures the recent evolution in bond pricing and is a
significant factor in pricing of new bonds, we use DefaultSpread and AvgYieldSpread as two
instruments for the observed yield spreads. Prior studies show that asset maturity (Barclay and
Smith, 1995; Hart and Moore, 1994) and term spread (Brick and Ravid, 1985, 1991) are two key
factors that determine debt maturity, so we choose asset maturity (AssetMaturity) and term
spread (TermSpread) as instruments for bond maturity (the definitions of these two variables are
11
Since CSR strengths and concerns are more likely jointly determined, it is difficult to find appropriate
instrumental variables for strengths and concerns, respectively. Therefore, we perform the instrumental variable test
for model (1) only.
33
provided in the Appendix). We use the average number of covenants of bonds issued over the
previous six months scaled by the maximum number of covenants in our sample (AvgCovenant)
as an instrumental variable for the covenant index.
Greene (2002) and Kennedy (2003) observe that the GMM estimates of instrumental
variables regression models are efficient when the regression errors are heteroskedastic and/or
autocorrelated. Prior studies (e.g., Billett et al., 2007; Brockman et al., 2010) use GMM to
address the endogeneity or simultaneity problem. Thus, we estimate model (1) using
instrumental variables GMM estimators, with statistics robust to heteroskedasticity and
clustering on issuer ID.12 We find that when YieldSpread is the dependent variable, the estimated
coefficient of CSRT is negative and significant, consistent with our findings reported in Panel A
of Table 4. As for two measures of the indirect cost of bonds—maturity and covenants—we find
that CSR performance is not significantly associated with maturity but is negatively associated
with covenant intensity. This result suggests that bonds issued by firms with better CSR
performance contain fewer covenant restrictions. Given that the average number of covenants in
a bond issue is 2.93 as shown in Table 2, this result lacks economic significance (the
instrumental variables regression results are not tabulated here for brevity).
5.2. Hierarchical linear models
Our data sets contain multiple new bond issues for a single firm in the same fiscal year.
Since multi-level observations violate the assumption of residual independence at the lower bond
12
Because some instrumental variables are missing, our sample reduces to 4,205 observations using the instrumental
variables regression approach. Given that there are so many indicator variables in the model, when we perform the
instrumental variables GMM regression, the estimated covariance matrix of moment conditions is not of full rank,
so the optimal weighting matrix for the GMM estimation cannot be calculated. Therefore, we use the Fama–French
10-industry classification in the instrumental variables GMM approach. The main results reported in Table 4 are
robust when we replace the 48 Fama–French industry indicators with the 10 Fama–French industry indicators. Since
the variables BusiCycle and DefaultSpread capture a similar concept and are highly correlated (the correlation is
0.90), we use instrumental variables GMM to estimate model (1) without BusiCycle. We also use instrumental
variables GMM to estimate model (1) with BusiCycle. The results are consistent.
34
level, the standard errors from the OLS regression may be biased. Hierarchical linear models,
which use maximum likelihood estimation, are widely used in social science research to address
potential multi-level observation problems (Ang et al., 2002; Seibert et al., 2004). We repeat our
analyses of models (1) and (2) using hierarchical linear models regression. Overall, the results
(not tabulated here for brevity) are qualitatively similar to our main test results.
As an additional robustness test, for firms with multiple bond issues during a fiscal year,
we keep only the bond issue with the largest offering amount. The regression results (not
tabulated for brevity) for this reduced sample are consistent with our main test results. The
magnitudes of the coefficients of CSRT, STR, and CON are larger.
5.3. Excluding convertible bonds
A small percentage of new corporate bond issues have convertible features. Convertible
bonds differ from straight bonds and linear regression models may not be appropriate to explore
the relation between yield spreads and convertible features (Khurana and Raman, 2003). As
another robustness check, we delete new bond issues with convertible features and re-run models
(1) and (2). The main findings still hold (the results are not tabulated).
6. Conclusion
This study complements the bond financing literature by using non-financial attributes
encompassed in CSR performance to explain the price and non-price terms of corporate bond
financing. The shareholder theory suggests that CSR activities consume cash and thus increase a
firm’s distress risk, while the stakeholder theory suggests that CSR performance benefits capital
markets through the reduction of information asymmetry between contracting parties and a
decreased perceived litigation risk on the part of issuers. In this study, we investigate how CSR
performance is related to the cost of new bond issues.
35
Our sample comprises 4,260 new public bond issues in the U.S. market in the period
1992–2009. Consistent with the stakeholder theory, we find that overall CSR performance is
associated with better credit ratings and lower yield spreads in new corporate bond issues. We
also find that the CSR strength score is associated with lower bond yield spreads, while the CSR
concern score is associated with higher bond yield spreads. In addition, using the number of
covenants as a measure of the indirect cost of debt, we find that overall CSR performance is
associated with less covenant intensity. Taken together, our results suggest that firms with better
CSR performance can raise public debt at lower cost. We further find that the association
between CSR performance and the cost of debt is more pronounced in investment-grade bonds
and non-Rule 144a bonds. Our results also indicate that bondholders value CSR performance
more for financially healthy firms than for financially distressed firms and that bondholders are
more likely to use CSR performance information to assess the creditworthiness of issuers with
weaker corporate governance and worse information environments and those operating in
environmentally sensitive industries.
Our findings have several practical implications. First, from standard-setting and
regulatory perspectives, the finding that the U.S. bond market values CSR performance further
supports policies emphasizing CSR commitment. Second, from the bond issuer perspective, our
findings can help issuers understand the association between the non-financial aspects of firm
performance and the cost of debt. Our findings may encourage firms to disclose non-financial
information voluntarily, thus providing a better information environment for external users.
The literature provides evidence of the effect of CSR disclosure and performance on the
cost of equity capital (e.g., Chava, 2014; Dhaliwal et al., 2011; Ghoul et al., 2011; Sharfman and
Fernando, 2008) and the cost of private debt (Goss and Roberts, 2011). Our study, along with the
36
concurrent study of Oikonomou et al. (2014), complements this line of research by providing
empirical evidence on the association between CSR performance and the cost of public debt.
Future research could examine how the costs of these three sources of capital are affected by
CSR performance and how they are influenced by each other. If data allow, future research could
also focus on CSR initiators and examine whether they are more likely to issue debt in the future
and, conditional on issuing debt, whether their cost of debt financing is lower.
37
References
Aerts, W., Cormier, D., Magnan, M., 2006. Intra-industry imitation in corporate environmental
reporting: An international perspective. J. Account. Public Policy 25(3), 299-331.
Ahmed, A., Billings, B., Morton, R., Stanford-Harris, M., 2002. The role of accounting
conservatism in mitigating bondholder-shareholder conflicts over dividend policy and in
reducing debt costs. Account. Rev. 77(4), 867-890.
Altman, R., 1968. Financial ratios, discriminant analysis and the prediction of corporate
bankruptcy. J. Finan. 23(4), 589-609.
Anderson, R.C., Mansi, S., Reeb, D.M., 2004. Board characteristics, accounting report integrity
and the cost of debt. J. Account. Econ. 68, 315–342.
Ang, S., Slaughter, S., Ng, K., 2002. Human capital and institutional determinants of information
technology compensation: Modeling multilevel and cross-level interactions. Manage. Sci.
48(11), 1427-1445.
Angelidis, J., Ibrahim, N., 2004. An exploratory study of the impact of degree of religiousness
upon an individual's corporate social responsiveness orientation. J. Bus. Ethics 51, 119–128.
Ashbaugh-Skaife, H., Collins, D.W., LaFond, R., 2006. The effects of corporate governance on
firms’ credit ratings. J. Account. Econ. 42, 203-243.
Barclay, M.J., Smith, C.W., 1995. The maturity structure of corporate debt. J. Finan. 50, 609-631.
Barnea, A., Rubin, A., 2010. Corporate social responsibility as a conflict between shareholders. J.
Bus. Ethics 97, 71-86.
Bessembinder, H., Kahle, K., Maxwell, W., Xu, D., 2009. Measuring abnormal bond
performance. Rev. Financ. Stud. 22(10), 4219-4258.
Bharath, S., Dahija, S., Saunders, A., Srinivasan, A., 2011. Lending relationships and loan
contract terms. Rev. Financ. Stud. 24(4), 1141-1203.
Bharath, S., Sunder, J., Sunder, S., 2008. Accounting quality and debt contracting. Account. Rev.
83(1), 1-28.
Bhattacharya, S., Chiesa, G., 1995. Proprietary information, financial intermediation, and
research incentives. J. Financ. Intermed. 4(4), 328-357.
Billett, M., King, D., Mauer, D., 2007. Growth opportunities and the choice of leverage, debt
maturity, and covenants. J. Finan. 62 (2), 697-730.
38
Botosan, C.A., 1997. Disclosure level and the cost of equity capital. Acc. Rev. 72, 323-349.
Brick, I., Ravid, A., 1985. On the relevance of debt maturity structure. J. Finan. 40, 1423-1437.
Brick, I., Ravid, A., 1991. Interest rate uncertainty and the optimal debt maturity structure. J.
Financ. Quant. Anal. 26, 363-381.
Brockman, P., Martin, X., Unlu, E., 2010. Executive compensation and the maturity structure of
corporate debt. J. Finan. 65 (3), 1123-1161.
Cespa, G., Cestone, G., 2007. Corporate social responsibility and managerial entrenchment. J.
Econ. Manage. Strat. 16, 741-771.
Chava, S., 2014. Environmental externalities and cost of capital. Forthcoming in Manage. Sci.
Cho, C.H., Patten, D.M., 2007. The role of environmental disclosures as tools of legitimacy: A
research note. Account. Org. Soc. 32, 639-647.
Clarkson, P.M., Fang, X.H., Li Y., Richardson, G., 2013. The relevance of environmental
disclosures: Are such disclosures incrementally informative? J. Account. Public Policy 32(5),
410-431.
Clarkson, P.M., Li, Y., Richardson, G., 2004. The market valuation of environmental capital
expenditures by pulp and paper companies. Acc. Rev. 79(2), 329-354.
Coffey B.S., Wang, J., 1998. Board diversity and managerial control as predictors of corporate
social performance. J. Bus. Ethics 17(4), 1595-603.
Deng, X., Kang, J., Low, B., 2013. Corporate social responsibility and stakeholder value
maximization: Evidence from mergers. J. Financ. Econ. 110(1), 87-109.
Denis, D., Mihov, V., 2003. The choice among bank debt, non-bank private debt, and public debt:
Evidence from new corporate borrowings. J. Financ. Econ. 70(1), 3-28.
Dhaliwal, D.S., Li, O.Z., Tsang, A., Yang Y.G., 2011. Voluntary nonfinancial disclosure and the
cost of equity capital: The initiation of corporate social responsibility reporting. Acc. Rev.
86(1), 59-100.
Dhillon, U., Johnson, H., 1994. The effect of dividend changes on stock and bond prices. J.
Finan. 64(1), 281-289.
Diamond, D., 1984. Financial intermediation and delegated monitoring. Rev. Econ. Stud. 51(3),
393-414.
Diamond, D., Verrecchia, R., 1991. Disclosure, liquidity and the cost of equity capital. J. Finan.
46, 1325-1360.
39
Elfenbein, D.W., Fisman, R., Mcmanus B., 2012. Charity as a substitute for reputation: Evidence
from an online marketplace. Rev. Econ. Stud. 79(4), 1441-1468.
Fama, E.F., 1985. What’s different about banks? J. Monetary Econ. 15(1), 29-39.
Fama, E.F., French, K.R., 1997. Industry costs of equity. J. Financ. Econ. 43, 153-193.
Freeman, R.E., 1984. Strategic management: A stakeholder approach. Boston: Pitman.
Friedman, M., 1962. Capitalism and Freedom. Chicago: University of Chicago Press.
Fulbright & Jaworski L.L.P. 2005. Second Annual Litigation Trends Survey findings. Available
at http://www.fulbright.com/mediaroom/files/FJ0536-UK-V13.pdf.
Ge, W., Kim, J.-B., 2014. Real earnings management and the cost of new corporate bonds. J. Bus.
Res. 67(4), 641-647.
Ge, W., Kim, J.-B., Song, B., 2012. Internal governance, legal institutions and bank loan
contracting around the world. J. Corp. Financ. 18 (3), 413-432.
Ghoul, S.E., Guedhami, O., Kwok, C.C.Y., Mishra, D., 2011. Does corporate social
responsibility affect the cost of capital? J. Bank. Financ. 35(9), 2388-2406.
Gompers, P., Ishii, J., Metrick, A., 2003. Corporate governance and equity prices. Q. J. Econ.
118, 107-156.
Goss, A., Roberts, G., 2011. The impact of corporate social responsibility on the cost of bank
loans. J. Bank. Financ. 35 (7), 1794-1810.
Graham, J.R., Harvey, C.R., Rajgopal, S., 2005. The economic implications of corporate
financial reporting. J. Account. Econ. 40, 3-73.
Greene, W.H., 2002. Econometric Analysis, Prentice Hall, Englewood Cliffs, NJ.
Gu, F., Li, J.Q., 2007. The credibility of voluntary disclosure and insider stock transactions. J.
Account. Res. 45, 771-810.
Haniffa, R.M., Cooke, T.E., 2005. The impact of culture and governance on corporate social
reporting. J. Account. Public Policy 24 (5), 391-430.
Hart, O., Moore, J., 1994. A theory of debt based on the inalienability of human capital. Q. J.
Econ. 109, 841-879.
Healy, P., Palepu, K., 2001. Information asymmetry, corporate disclosure, and the capital
markets: A review of the empirical disclosure literature. J. Account. Econ. 31, 405-440.
40
Heinkel, R., Kraus, A., Zechner, J., 2001. The effect of green investment on corporate behaviour.
J. Financ. Quant. Anal. 36,431-449.
Hillman, A.J., Keim, G.D., 2001. Shareholder value, stakeholder management, and social issues:
What’s the bottom line? Strateg. Manage. J. 22(2), 125–139.
Hong, H., Kacperczyk, M., 2009. The price of sin: The effects of social norms on markets, J.
Financ. Econ. 93, 15-36.
Huddart, S.J., Ke, B., 2007. Information asymmetry and cross-sectional variation in insider
trading. Contemp. Account. Res. 24, 195-232.
Jensen, M., 2001. Value maximization, stakeholder theory, and the corporate objective function.
J. Appl. Corp. Financ. 14(3), 8-21.
Jensen, M., Meckling, W., 1976. Theory of the firm: Managerial behavior, agency costs and
ownership structure. J. Financ. Econ. 3(4), 305-360.
Jiang, J., 2008. Beating earnings benchmarks and the cost of debt. Account. Rev. 83(2), 377-416.
Jones, T.M., 1995. Instrumental stakeholder theory: A synthesis of ethics and economics. Acad.
Manage. Rev. 20, 404-437.
Kennedy, P., 2003. A Guide to Econometrics, MIT Press, Cambridge, MA.
Khurana, I., Raman, K., 2003. Are fundamentals priced in the bond market? Contemp. Account.
Res. 20(3), 465-494.
Klock, M.S., Mansi, S.A., Maxwell, W.F., 2005. Does corporate governance matter to
bondholders? J. Financ. Quant. Anal. 40(4), 693-719.
Lacy, P., Cooper, T., Hayward, R., Neuberger, L., 2010. A new era of sustainability. UN global
compact—Accenture CEO study.
Leuz, C., Verrecchia, R., 2000. The economic consequences of increased disclosure. Account.
Res. 38, 91-124.
Linthicum, C., Reitenga, A.L., Sanchez, J.M., 2010. Social responsibility and corporate
reputation: The case of the Arthur Andersen Enron audit failure. J. Account. Public Policy 29,
160-176.
Maignan, I., Ferrell, O.C., 2003. Nature of corporate responsibilities: Perspectives from
American, French, and German consumers. J. Bus. Res. 56 (1), 55-67.
Mansi, S., Maxwell, W., Miller, D., 2011. Analyst forecast characteristics and the cost of debt.
Rev. Account. Stud. 16(1), 116-142.
41
Mayers, D., 1998. Why firms issue convertible bonds: The matching of financial and real
investment options. J. Financ. Econ. 47(1), 83-102.
Menz, K.M., 2010. Corporate social responsibility: Is it rewarded by the corporate bond market?
A critical note. J. Bus. Ethics 96, 117-134.
Merton, R., 1974. On the pricing of corporate debt: The risk structure of interest rates. J. Finan.
29(2), 449-470.
Merton, R., 1987. A simple model of capital market equilibrium with incomplete information. J.
Finan. 42, 483-510.
Nikolaev, V., 2010. Debt covenants and accounting conservatism. J. Account. Res. 48(1), 137-
175.
Oikonomou, I., Brooks, C., Pavelin, S., 2014. The effects of corporate social performance on the
cost of corporate debt and credit ratings. Financ. Rev. 49, 49-75.
Orlitzky, M., Schmidt, F., Rynes, S., 2003. Corporate social and financial performance: A meta-
analysis. Organ. Stud. 24, 403-441.
Petersen, M., 2009. Estimating standard errors in finance panel data sets: Comparing approaches.
Rev. Financ. Stud. 22(1), 435-480.
Plumlee, M., Brown, D., Marshall, R.S., 2014. Voluntary environmental disclosure quality and
firm value: Further evidence. Forthcoming in J. Account. Public Policy.
Rajan, R.G., 1992. Insiders and outsiders: The choice between informed and arm’s-length debt. J.
Finan. 47(4), 1367-1400.
Richardson, A.J., Welker, M., 2001. Social disclosure, financial disclosure and the cost of equity
capital. Account. Org. Soc. 26, 597-616.
Rubin, A., 2008. Political views and corporate decision making: the case of corporate social
responsibility. Financ. Rev. 43, 337–360.
Seetharaman, A., Gul, F.A., Lynn, S.G., 2002. Litigation risk and audit fees: evidence from UK
firms cross-listed on US markets. J. Acc. Econ. 33, 91-115.
Seibert, S., Silver, S., Randolph, A., 2004. Taking empowerment to the next level: A multiple-
level model of empowerment, performance, and satisfaction. Acad. Manage. J. 47(3), 332-
349.
Sengupta, P., 1998. Corporate disclosure quality and the cost of debt. Account. Rev. 73(4), 459-
474.
42
Sharfman, M.P., 1996. The construct validity of the Kinder, Lydenberg & Domini social
performance ratings data. J. Bus. Ethics 15(3), 287-296.
Sharfman, M.P., Fernando, C.S., 2008. Environmental risk management and the cost of capital.
Strateg. Manage. J. 29, 569-592.
Shi, C., 2003. On the trade-off between the future benefits and riskiness of R&D. J. Account.
Econ. 35(2), 227-254.
Stephen, B, Noushi, R., Corinne, P., 2010. The impact of board diversity and gender composition
on corporate social responsibility and firm reputation. J. Bus. Ethics 97(2), 207-221.
Van der Laan Smith, J., Adhikari, A., Tondkar, R.H., 2005. Exploring differences in social
disclosures internationally: a stakeholder perspective. J. Account. Public Policy 24, 123-151.
Van der Laan Smith, J., Adhikari, A., Tondkar, R.H., Andrews, R.L., 2010. The impact of
corporate social disclosure on investment behavior: A cross-national study. J Account. Public
Policy 29(2), 177-192.
Verschoor, C., 2005. Is there financial value in corporate values? Strateg. Finan. 87(1), 17-18.
Wang, A., Zhang, G., 2009. Institutional ownership and credit spreads: An information
asymmetry perspective. J. Empir. Financ. 16(4), 597-612.
Ziebart, D., Reiter, S., 1992. Bond ratings, bond yields and financial information. Contemp.
Account. Res. 9(1), 252-282.
43
Table 1. Descriptive statistics of number of new corporate bond issues by year
Year Frequency Percentage

1992 137 3.22
1993 199 4.67
1994 147 3.45
1995 159 3.73
1996 161 3.78
1997 192 4.51
1998 229 5.38
1999 152 3.57
2000 228 5.35
2001 345 8.1
2002 252 5.92
2003 309 7.25
2004 285 6.69
2005 212 4.98
2006 237 5.56
2007 352 8.26
2008 221 5.19
2009 443 10.39
Total 4260 100
This table presents the annual distributions of the number of new corporate bond issues in our study. Our sample
contains 4,260 new bond issues from 2,317 firms during the period 1992–2009.
44
Table 2. Descriptive statistics of key variables
Variable N Mean Std. Dev. 25th Percentile Median 75th Percentile

Bond-level variables
YieldSpread 4260 154.13 211.31 62.61 131.64 247.16
Rating 4260 11.67 7.39 6 9 14
lnRating 4260 2.27 0.62 1.79 2.20 2.64
RatingGroup 4260 3.56 1.56 2 4 5
Number of covenants 4260 2.93 3.28 0 2 5
Covenant 4260 0.14 0.16 0 0.10 0.24
Maturity in years 4260 11.81 10.02 5 10 11
lnMaturity 4260 2.22 0.69 1.61 2.30 2.40
Offering amount 4260 432.80 2681.95 100 250 500
IssueSize 4260 11.80 2.05 11.51 12.43 13.12
RedeemD 4260 0.66 0.47 0 1 1
PutD 4260 0.07 0.25 0 0 0
ConvertD 4260 0.14 0.35 0 0 0
415RegD 4260 0.70 0.46 0 1 1
144aRegD 4260 0.24 0.43 0 0 0
Firm-level variables
CSRT 2323 0.05 2.97 -2 0 1
STR 2323 2.28 2.83 0 1 3
CON 2323 2.23 2.27 1 2 3
IssuerSize 2323 8.46 1.37 7.55 8.45 9.47
ROA 2323 0.03 0.09 0.01 0.05 0.08
Leverage 2323 0.26 0.17 0.15 0.24 0.35
Big4 2323 0.98 0.14 1 1 1
Z-score 2323 3.24 2.54 1.61 2.70 4.16
This table reports the descriptive statistics of key variables. The definitions of all the variables are provided in the Appendix.
45
Table 3. Correlation matrix
YieldSpread CSRT STR CON IssuerSize ROA Leverage Big4 Z-score lnMaturity IssueSize Rating
CSRT -0.14***
STR -0.02 0.69***
CON 0.16*** -0.47*** 0.31***
IssuerSize 0.13*** 0.02 0.46*** 0.54***
ROA -0.11*** 0.15*** 0.18*** 0.03* 0.14***
Leverage 0.21*** -0.04*** -0.18*** -0.17*** -0.03** -0.30***
Big4 -0.05*** 0.03** 0.07*** 0.04** 0.17*** 0.03* 0.02
Z-score -0.26*** 0.25*** 0.12*** -0.18*** -0.22*** 0.39*** -0.29*** -0.04***
lnMaturity -0.15*** 0.02 0.01 -0.02 -0.07*** 0.06*** -0.10*** 0.03* 0.07***
IssueSize 0.00 0.06*** 0.26*** 0.24*** 0.18*** 0.13*** -0.51*** -0.01 0.08*** 0.14***
lnRating 0.04*** -0.17*** -0.38*** -0.24*** -0.46*** -0.37*** 0.28*** -0.12*** -0.14*** -0.05*** -0.17***
Covenant 0.14*** 0.04** 0.13*** 0.11*** 0.13*** 0.15*** -0.19*** 0.05*** 0.03** 0.10*** 0.39*** -0.16***
This table reports the correlation matrix of some key variables used in our primary tests. The definitions of all variables are provided in the Appendix. The superscripts
*, **, and *** denote significance at the 10%, 5%, and 1% levels, respectively.
46
Table 4. Association between CSR performance and the cost of corporate bonds
Panel A: Association between overall CSR performance and the cost of corporate bonds
(1) (2) (3) (2) vs. (3)

Variables Prediction
DV: lnRating DV: YieldSpread DV: YieldSpread
Coefficient t-Value Coefficient t-Value Coefficient t-Value Chi-square p-value
CSRT - -0.01 -2.18** -3.67 -2.97*** -3.20 -2.62*** 3.77** 0.05
IssuerSize - -0.13 -8.75*** -18.76 -5.51*** -13.97 -3.88***
ROA - -0.81 -5.98*** -287.11 -5.98*** -258.44 -5.35***
Leverage + 0.28 3.36*** 136.89 5.81*** 126.84 5.24***
Big4 - -0.10 -1.64* -93.71 -2.86*** -90.31 -2.75***
Z-score - -0.02 -3.12*** -6.68 -3.08*** -5.92 -2.74***
lnMaturity + -0.07 -4.78*** -4.29 -1.04 -1.74 -0.43
Covenant + 0.30 3.15*** 145.19 5.46*** 134.61 5.17***
IssueSize -0.05 -4.54*** -2.36 -1.04 -0.75 -0.32
lnRating + - - - - 35.52 3.84***
BusiCycle + 0.00 0.10 1.32 10.48*** 1.33 10.62***
RedeemD + 0.06 2.61*** 21.77 3.10*** 19.68 2.85***
PutD - 0.06 1.31* -137.45 -9.55*** -139.43 -9.75***
ConvertD - 0.48 13.08*** -281.74 -17.68*** -298.64 -17.90***
415RegD - -0.06 -1.89** -46.40 -2.81*** -44.11 -2.67***
144aRegD 0.12 2.82*** 46.24 2.49** 41.88 2.22**
Industry indicators Yes Yes Yes
Year indicators Yes Yes Yes
No. of observations 4260 4260 4260
Model fit (F-value) 98.11*** 107.61*** 130.95***
Adjusted R2 56.99% 61.76% 62.23%
47
Table 4 (continued)
Panel B: Association between CSR strengths and concerns and the cost of corporate bonds
(1) (2) (3) (2) vs. (3)
Variables Prediction
DV: lnRating DV: YieldSpread DV: YieldSpread
Coefficient t-Value Coefficient t-Value Coefficient t-Value Chi-square p-value
STR - -0.03 -3.16*** -3.34 -2.21** -2.42 -1.60* 6.93*** 0.01
CON + -0.01 -0.89 4.22 2.03** 4.45 2.18** 0.83 0.36
IssuerSize - -0.10 -7.35*** -19.60 -4.73*** -15.83 -3.78***
ROA - -0.85 -6.26*** -285.94 -5.94*** -255.19 -5.26***
Leverage + 0.27 3.21*** 137.31 5.82*** 127.60 5.28***
Big4 - -0.10 -1.72** -93.57 -2.85*** -89.93 -2.74***
Z-score - -0.02 -2.92*** -6.72 -3.08*** -6.00 -2.76***
lnMaturity + -0.07 -4.71*** -4.33 -1.05 -1.79 -0.44
Covenant + 0.25 2.77*** 146.38 5.42*** 137.16 5.18***
IssueSize -0.04 -4.32*** -2.47 -1.08 -0.96 -0.42
lnRating + - - - - 36.20 3.89***
BusiCycle + 0.00 0.29 1.33 10.44*** 1.32 10.56***
RedeemD + 0.06 2.68*** 21.72 3.09*** 19.53 2.82***
PutD - 0.05 1.24 -137.37 -9.54*** -139.28 -9.74***
ConvertD - 0.47 13.15*** -281.60 -17.64*** -298.65 -17.90***
415RegD - -0.06 -1.85** -46.43 -2.80*** -44.13 -2.66***
144aRegD 0.11 2.54** 46.58 2.51** 42.58 2.27**
Industry indicators Yes Yes Yes
Year indicators Yes Yes Yes
No. of observations 4260 4260 4260
Model fit (F-value) 140.32*** 105.42*** 126.79***
Adjusted R2 57.44% 61.77% 62.25%
This table reports the OLS regression results of the association between CSR performance and the cost of corporate bonds. The definitions of all the variables are provided in the
Appendix. The standard errors of the estimated coefficients are corrected for firm-level clustering and heteroskedasticity. The superscripts *, **, and *** denote significance at
the 10%, 5%, and 1% levels, respectively (two-tailed test if the sign is not predicted).
48
Table 5. Association between firm scores on individual CSR dimensions and bond yield spreads
Panel A: Association between subcategories of overall CSR performance and bond yield spreads
Variables Prediction (1) (2) (3) (4) (5) (6) (7)
Environment - -3.58
Community - -9.38**
Product - -9.17**
Diversity - -2.55
Employee Relations - -5.77**
Human Rights - 0.26
Governance - -8.60*
IssuerSize - -14.49*** -13.33*** -15.25*** -12.84*** -13.43*** -13.64*** -13.61***
ROA - -263.17*** -263.14*** -266.20*** -265.39*** -260.57*** -265.29*** -261.14***
Leverage + 123.22*** 122.62*** 121.82*** 119.91*** 126.91*** 121.06*** 122.42***
Big4 - -91.10*** -91.91*** -90.16*** -90.43*** -90.79*** -90.97*** -91.28***
Z-score - -6.36*** -6.30*** -6.13*** -6.34*** -5.94*** -6.39*** -6.38***
lnMaturity + -1.75 -1.64 -1.96 -1.68 -1.52 -1.67 -1.66
Covenant + 138.32*** 136.68*** 139.87*** 137.20*** 138.18*** 139.24*** 137.57***
IssueSize -1.03 -0.74 -1.68 -1.06 -0.97 -1.21 -1.17
lnRating + 36.62*** 35.67*** 38.11*** 36.61*** 36.85*** 37.19*** 37.09***
BusiCycle + 1.32*** 1.32*** 1.32*** 1.32*** 1.32*** 1.32*** 1.32***
RedeemD + 19.95*** 19.35*** 19.31*** 19.67*** 19.74*** 19.66*** 19.65***
PutD - -139.72*** -139.66*** -139.71*** -139.58*** -140.24*** -139.85*** -139.74***
ConvertD - -298.54*** -298.54*** -297.68*** -298.66*** -297.34*** -298.30*** -298.72***
415RegD - -44.62*** -44.28*** -44.92*** -44.21*** -44.18*** -44.48*** -44.43***
144aRegD 43.28*** 43.67** 42.84** 43.51** 44.03*** 44.08*** -43.60**
Industry indicators Yes Yes Yes Yes Yes Yes Yes
Year indicators Yes Yes Yes Yes Yes Yes Yes
No. of observations 4260 4260 4260 4260 4260 4260 4260
Model fit (F-value) 118.12*** 140.72*** 104.90*** 113.80*** 121.24*** 110.52*** 110.86***
Adjusted R2 62.11% 62.17% 62.18% 62.10% 62.14% 62.08% 62.11%
49
Table 5 (continued)
Panel B: Association between subcategories of CSR strengths and concerns and bond yield spreads
Subcategories STR & CON (1) (2) (3) (4) (5) (6) (7)
Strengths -6.38*
Environment
Concerns 1.81
Strengths -9.96**
Community
Concerns 7.76
Strengths -2.10
Product
Concerns 11.11**
Strengths -0.05
Diversity
Concerns 16.28**
Strengths -3.97
Employee Relations
Concerns 7.87*
Strengths -23.18
Human Rights
Concerns -1.44
Strengths -15.10*
Governance
Concerns 2.11
Control variables Yes Yes Yes Yes Yes Yes Yes
Industry indicators Yes Yes Yes Yes Yes Yes Yes
Year indicators Yes Yes Yes Yes Yes Yes Yes
No. of observations 4260 4260 4260 4260 4260 4260 4260
Model fit (F-value) 119.16*** 138.53*** 101.63*** 114.37*** 116.35*** 108.20*** 108.65***
Adjusted R2 62.12% 62.17% 62.19% 62.18% 62.15% 62.09% 62.12%
This table reports the OLS regression results of the tests related to firm scores on individual CSR dimensions. Specifically, Panel A reports the results of the
association between overall CSR performance for each category and bond yield spreads. Panel B reports the results of the association between subcategories of CSR
strengths and concerns and bond yield spreads. The definitions of all the variables are provided in the Appendix. The standard errors of the estimated coefficients are
corrected for firm-level clustering and heteroskedasticity. The superscripts *, **, and *** denote significance at the 10%, 5%, and 1% levels, respectively (two-tailed
test if the sign is not predicted).
50
Table 6. Association between overall CSR performance and bond yield spreads: Weak and
strong governance subsamples
Pre-SOX Post-SOX
Variables Prediction Strong Weak Strong Weak
governance governance governance governance
CSRT - -3.90*** -5.05** 0.84 -5.38**
IssuerSize - -10.82** -27.74*** -7.39 -14.60*
ROA - -366.17*** 35.00 -290.01*** -553.99***
Leverage + 7.83 229.66*** 165.73 46.72
Big4 - 56.96* -77.41** -98.16** 13.23
Z-score - -3.27* -5.01 -3.32 -9.77**
lnMaturity + 21.85*** 18.56*** -13.19* -16.84**
Covenant + 73.89* 190.79*** 126.77*** 148.92***
IssueSize -5.16 -1.82 -2.16 4.68
lnRating + 37.91*** 16.66 48.05*** 70.10***
BusiCycle + 0.62*** 0.45** 1.72*** 1.57***
RedeemD + 26.53*** 7.48 16.59 42.13***
PutD - -120.05*** -102.11*** -157.29 -150.44***
ConvertD - -244.68*** -195.15*** -292.25 -234.78***
415RegD - -6.66 42.00* -139.08*** -67.44***
144aRegD 59.80** 64.65* 39.38 -1.14
Industry indicators Yes Yes Yes Yes
Year indicators Yes Yes Yes Yes
No. of observations 1028 861 1190 718
Model fit (F-value) 624.44*** 721.33*** 115.86*** 53.01***
Adjusted R2 66.63% 49.90% 66.97% 63.38%
This table reports the results of the association between CSR performance and bond yield spreads for strong
corporate governance and weak corporate governance subsamples in the pre- and post-SOX periods. The
observations in the strong (weak) governance subsample are those whose G-index is below (above) the median. The
definitions of all the variables are provided in the Appendix. The standard errors of the estimated coefficients are
corrected for firm-level clustering and heteroskedasticity. The superscripts *, **, and *** denote significance at the
10%, 5%, and 1% levels, respectively (two-tailed test if the sign is not predicted).
51
Table 7. Association between overall CSR performance and bond yield spreads: Subsample analyses
Environmentally sensitive
Bond rating Z-score Rule 144a bond Information asymmetry
industry
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)
Variables
Non-
Investment Rule 144a Non-rule Non-
investment High Low High Low Sensitive
grade bonds 144a bonds sensitive
grade
CSRT -3.31*** -2.08 -3.43*** -2.89* 1.61 -2.54** -4.28*** -1.82 -7.60*** -0.97
IssuerSize -20.21*** -16.17*** -24.70*** -7.90** -3.46 -3.34 -22.94*** -6.23 -13.25** -19.64***
ROA -305.51*** -249.31*** -260.40*** -242.98*** -297.47 -120.14** -462.87*** -97.69* -22.09 -420.82***
Leverage -51.95* 125.11*** 98.66*** 129.01*** 77.26** 121.19*** 121.20*** 147.79*** 131.54** 117.07***
Big4 -30.94* -76.36*** -80.55** -87.69** -65.90** -53.01 -6.6 -179.62*** -49.78 -87.90**
Z-score -9.74 -3.3 - - -3.42 -5.91** -3.22 -7.72*** -6.59** -6.37***
lnMaturity 16.53*** 55.33*** 5.50 -5.83 -22.73** 8.30** 2.98 -4.82 -11.78** 4.54
Covenant 131.70*** 61.38* 83.08** 141.28*** 34.53 270.72*** 90.21*** 173.60*** 199.72*** 92.26***
IssueSize -3.07 10.28** -0.50 -0.45 -5.21 -9.06*** 2.55 -1.97 -3.74 3.88
lnRating - - 12.82* 92.49*** 133.58*** 22.18*** 37.36*** 29.48** 21.93* 43.64***
BusiCycle 1.35*** 1.59*** 1.30*** 1.57*** 1.54*** 1.36*** 1.68*** 0.93*** 1.31*** 1.35***
RedeemD -0.91 49.33*** 25.99*** 14.26 58.38*** 0.95 15.87* 22.91** 8.37 23.74***
PutD -150.15*** -98.34*** -123.38*** -147.10*** -72.56*** -184.22*** -148.68*** -144.07*** -182.92*** -127.41***
ConvertD -150.47*** -348.31*** -317.33*** -301.11*** -468.58*** -110.29 -335.12*** -249.62*** -208.90*** -335.60***
415RegD -7.29 -71.31** 3.56 -63.82*** 67.08 -15.76 -36.81** -36.06 -31.66 -46.91***
144aRegD 19.01 -6.1 51.06*** 33.26 - - 35.97* 42.47* 83.34** 25.10*
Industry indicators Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes
Year indicators Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes
No. of observations 2456 1804 2128 2132 1030 3230 2006 2005 1561 2699
Model fit (F-value) 35.88*** 130.34*** 482.63*** 244.59*** 330.38*** 60.42*** 461.67*** 132.98*** 55.49*** 93.71***
Adjusted R2 58.25% 72.50% 63.90% 62.95% 80.39% 55.77% 68.25% 61.41% 61.94% 64.52%
This table reports the results of the association between overall CSR performance and bond yield spreads for the following five subgroups: investment-grade bonds versus
non–investment-grade bonds; issuers with a high and a low Z-score; Rule 144a bonds versus non-Rule 144a bonds; bond issuers with high and low information asymmetry;
and bond issuers operating in environmentally sensitive and non-environmentally sensitive industries. Bond issuers with high (low) information asymmetry are those whose
intangible ratio (Intangible) is above (below) the median. Firms in environmentally sensitive industries operate in the following industries: oil and gas; forestry, pulp, and
paper; energy; chemicals and drugs; mining and resources; and utilities. The definitions of all the variables are provided in the Appendix. The standard errors of the
estimated coefficients are corrected for firm-level clustering and heteroskedasticity. The superscripts *, **, and *** denote significance at the 10%, 5%, and 1% levels,
respectively (two-tailed test if the sign is not predicted).
52
Appendix. Variable definitions
Variable Definition
YieldSpread Initial corporate bond yield minus Treasury bond yield with comparable maturity.
A corporation’s total CSR strength score minus the total CSR concern score (proxy
CSRT
for overall CSR performance).
STR A corporation’s total CSR strength score (proxy for CSR strengths).
CON A corporation’s total CSR concern score (proxy for CSR concerns).
The natural logarithm of the issuer’s assets at the fiscal year-end immediately prior
IssuerSize
to the new corporate bond issuance date.
Return on assets of the issuer, defined as net income divided by total assets at the
ROA
fiscal year-end immediately prior to the new corporate bond issuance date.
Long-term debt divided by total assets of the issuer at the fiscal year-end
Leverage
immediately prior to the new corporate bond issuance date.
An indicator variable that equals one if a firm is audited by a Big Four auditor and
Big4
zero otherwise.
Altman’s Z-score is calculated as follows: 1.2 × (working capital/total assets) + 1.4
× (retained earnings/total assets) + 3.3 × (earnings before interest and taxes/total
Z-score
assets) + 0.6 × (market value of equity/book value of total debt) + 1.0 × (sales/total
assets).
Difference between the average yield of Moody's Aaa bonds and the average yield
BusiCycle
of 10-year U.S. Treasury bonds for the month of issue.
lnMaturity The natural logarithm of the number of years until the bond matures.
The natural logarithm of the par value of the bond initially issued, in millions of
IssueSize
dollars.
Bond rating by Standard & Poor’s, sequentially converted to numbers, with one
Rating for AAA through 27 for no rating. For issues without a Standard & Poor’s rating,
Moody’s and Fitch ratings are used instead.
lnRating The natural logarithm of Rating.
Indicator variable that equals 1 for a AAA rating; 2 for AA+, AA, and AA-; 3 for
RatingGroup A+, A, and A-; 4 for BBB+, BBB, and BBB-; 5 for BB+, BB, and BB-; 6 for B+, B,
and B-; and 7 for CCC+, CCC, CC, C, D, and SD.
The total number of bondholder protective covenants in a bond issue divided by the
Covenant
maximum number of bondholder protective covenants in our sample.
An indicator variable for the call feature of a bond. It equals one for bonds that have
RedeemD
an embedded call option and zero otherwise.
An indicator variable for the put feature of a bond. It equals one for bonds that have
PutD
an embedded put option and zero otherwise.
An indicator variable for the convertible feature of a bond. It equals one for bonds
ConvertD
that have an embedded convertible option and zero otherwise.
53
An indicator variable for the SEC Rule 415 shelf registration feature of a bond. It
415RegD
equals one for bonds issued under a shelf registration and zero otherwise.
An indicator variable for the SEC Rule 144a private placement feature of a bond. It
144aRegD equals one for bonds issued through private placement exempt from registration and
zero otherwise.
An instrumental variable for YieldSpread. It is the difference between the yields on
DefaultSpread Moody’s seasoned Baa-rated corporate bonds and 10-year U.S. government bonds
in the month corporate bonds are issued.
An instrumental variable for YieldSpread. It is the average yield spread on bond
AvgYieldSpread
issues completed over the previous six months.
An instrumental variable for lnMaturity. It is measured as the weighted average of
maturity of current assets (CA) and net property, plant, and equipment (NPPE):
AssetMaturity
.
An instrumental variable for lnMaturity. It is the difference between the yields on

TermSpread 10- and one-year U.S. Treasury bonds measured in the month corporate bonds are
issued.
An instrumental variable for Covenant. It is the average number of covenants of
AvgCovenant bonds issued over the previous six months divided by the maximum number of
bondholder protective covenants for our sample.
An instrumental variable for CSRT. It equals one if the bond issuer’s headquarters is
located in a blue or Democratic state and zero otherwise. A state is considered a
Blue blue state if it is listed as a blue state at both of the following websites:
http://en.wikipedia.org/wiki/File:Red_state,_blue_state.svg and
http://azpundit.com/list-of-the-most- democratic-republican-states/
An instrumental variable for CSRT. It is the ranking of the state in which the bond
issuer’s headquarters are located and ranges between one and 51. The ranking is
ReligionRank based on the ratio of the number of religious adherents in the issuer’s state over the
total population in that state in 2000. The data on religiosity were obtained from the
Association of Religion Data Archive.
G-index Governance index developed by Gompers et al. (2003).
Intangible Intangible ratio, defined as intangible assets divided by total assets.
54

Corporate Social Responsibility and The Cost of Corporate Bonds

Uploaded by

Copyright:

Available Formats

Corporate Social Responsibility and The Cost of Corporate Bonds

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Corporate Social Responsibility and The Cost of Corporate Bonds

Uploaded by

Copyright:

Available Formats

Corporate Social Responsibility and the Cost of Corporate Bonds

Wenxia Gea and Mingzhi Liua,*

Issue; Cost of Debt

responsibility (CSR) as “the continuing commitment by business to behave ethically and

responsibility investment accounted for 12% of total U.S. asset investment.

stakeholder theory. According to Friedman’s (1962) shareholder theory, a corporation’s top

consequences of CSR performance on the cost of public debt financing. Specifically, we

new corporate bond issues in the U.S. bond market.

address potential endogeneity issues and a variety of other robustness tests.

with bond yield spreads.

significantly associated with higher bond yield spreads.

information asymmetry, and for issuers operating in environmentally sensitive industries.

valued by the market may not be fully captured by credit ratings.

the costs and benefits of CSR activities.

its own merit.

disclosure channels to overcome the limitations of mandatory disclosures. In this sense,

activities on the cost of debt and equity capital.

in investment response to CSR disclosure. Thus, the significant differences in institutional

the potential impact of CSR performance on U.S. corporate bond issues.

aware of the potential benefits of CSR activities.

robustness tests. Section 6 sets forth our conclusions.

2. Literature review and hypothesis development

to maximize firm value. Accordingly, CSR-related expenditure is considered a poor use of

both of which can be related to higher costs of public debt financing.

CSR performance, compared to that for irresponsible firms.

however, banks do not reward the CSR investments of borrowers.

associated with bond yield spreads.

3.1. Sample selection

13 dimensions of corporate CSR performance: environment, community, product, diversity,

winsorize all continuous variables at the top and bottom 1%.

3.2. Empirical models

YieldSpreadijt = β0+ β1CSRTit + β2IssuerSizeit + β3ROAit + β4Leverageit +

YieldSpreadijt = β0+ β1STRit + β2 CONit + β3IssuerSizeit + β4ROAit + β5Leverageit

Following previous studies, we measure the dependent variable YieldSpread as the

Firm-level control variables

be negatively related to the risk premium.

positively related to the risk premium.

quality; thus we expect its coefficient to be negative.

Z-score: This is the Altman (1968) Z-score, a measure of a corporation’s financial

Bond-level control variables

bond rating indicates an issue’s creditworthiness; so we expect it to be positively related to the

risk premium (Jiang, 2008; Shi, 2003).

Covenant: This is a covenant index. Following Nikolaev (2010), we consider 36 covenant

positive relation between YieldSpread and Covenant.

premium (Jiang, 2008; Sengupta, 1998; Shi, 2003).

we expect the variable RedeemD to be positively related to the risk premium.

to the risk premium.

future. Therefore, we expect the coefficient of this variable to be negative.

Industry and year indicator variables

4.1. Univariate analysis

[Insert Table 1 about here]

are audited by Big Four auditors.

[Insert Table 2 about here]

[Insert Table 3 about here]

and, third, we re-examine our hypotheses including bond ratings.

coefficients for firm-level clustering, as well as for heteroskedasticity.

bond issue (432.80 × 3.20 ÷10000 × 12 = 1.66).

is statistically significant (Chi-square = 5.73; p < 0.05).