Godfrey Et Al. 2009

Strategic Management Journal
Strat. Mgmt. J., 30: 425–445 (2009)

Published online 30 December 2008 in Wiley InterScience (www.interscience.wiley.com) DOI: 10.1002/smj.750
Received 30 October 2007; Final revision received 2 December 2008
THE RELATIONSHIP BETWEEN CORPORATE SOCIAL

RESPONSIBILITY AND SHAREHOLDER VALUE: AN
EMPIRICAL TEST OF THE RISK MANAGEMENT
HYPOTHESIS
PAUL C. GODFREY,1 * CRAIG B. MERRILL,2 and JARED M. HANSEN3
1
Marriott School of Management, Brigham Young University, Provo, Utah, U.S.A.
2
Marriott School of Management, Brigham Young University, Fellow, Wharton
Financial Institutions Center, Provo, Utah, U.S.A.
3
Belk College of Business, University of North Carolina-Charlotte, Charlotte, North
Carolina, U.S.A.
Do shareholders gain when managers disperse corporate resources through activities classified
as corporate social responsibility (CSR)? Strategy scholars have recently developed a theoretical
model that links such activities to shareholder value when a firm suffers a negative event; we
test key portions of this theory of the ‘insurance-like’ property of CSR activity. We posit that
such activity leads to positive attributions from stakeholders, who then temper their negative
judgments and sanctions toward firms because of this goodwill. We extend the risk management
model by theorizing that some types of CSR activities will be more likely to create goodwill
and offer insurance-like protection than other types. We delineate several firm and event specific
characteristics that we expect to influence the link between CSR activities and an insurance
effect. We then test our model using an event study of 178 negative legal/regulatory actions
against firms throughout the 11 years from 1993–2003. We find that participation in institutional
CSR activities—those aimed at a firm’s secondary stakeholders or society at large—provides
an ‘insurance-like’ benefit, while participation in technical CSRs—those activities targeting a
firm’s trading partners—yields no such benefits. We conclude by considering the implications
of our findings for future theorizing and research into the economic value of CSR engagement.
Copyright  2008 John Wiley & Sons, Ltd.
INTRODUCTION is not in the interest of his employers—Friedman

(1970 : 33)
What does it mean to say that the corporate execu-

tive has a ‘social responsibility’ in his capacity as In practice, corporate managers and fundraisers
businessman? If this statement is not pure rhetoric, agree that corporate transfers to charity represent
it must mean that he is to act in some way that a calculated purchase of advertising services or
goodwill.—(Knauer, 1994: 11)
Keywords: insurance; risk management; event study; cor-
porate social responsibility Do shareholders gain when a firm’s strategies
*Correspondence to: Paul C. Godfrey, Marriott School of Man-
agement, Brigham Young University, 789 TNRB, Provo, UT include disbursing corporate resources through
84602, USA. E-mail: Paul Godfrey@byu.edu participation in social initiatives, commonly
Copyright  2008 John Wiley & Sons, Ltd.

426 P. C. Godfrey, C. B. Merrill, and J. M. Hansen
referred to as corporate social responsibility (e.g., Blacconiere and Patten, 1994; Blacconiere
(CSR)? Since the original debate between Adolph and Northcutt, 1997; Freedman and Stagliano,
Berle (1931) and Merrick Dodd (1932) jump- 1991; Orlitzky and Benjamin, 2001; Schnietz and
started inquiry, scholars across several business Epstein, 2005; Williams and Barrett, 2000).
disciplines—accounting, finance, management, Richardson, Welker, and Hutchinson (1999) pro-
and marketing—have taken up this question vide a thorough review and thoughtful commentary
and offered both positive (Berle-sympathetic) and on much of this literature. While these studies have
normative (Dodd-sympathetic) accounts of CSR. contributed to our growing understanding of when
The depth of research is significant. For example, and how CSR activities preserve CFP, they invite
Margolis and Walsh (2001) review almost 100 further research that would address three signifi-
studies attempting to quantify a CSR-Corporate cant, related limitations.
Financial Performance (CFP) link. Likewise, the First, several of the studies lack a theoretical
Socially Responsible Investing Studies Web site explanation (e.g., Blacconiere and Patten, 1994)
(sristudies.org), lists over 225 studies discussing or offer underspecified and broad theoretical ratio-
elements of the CSR-CFP relationship.1 The nales to explain precise responses (e.g., Schnietz
overall orientation has been to argue and show and Epstein [2005] invoke the generalized notion
that CSR activities generate CFP. Most often of a reputation to explain why some firms suffer
the generation of financial performance occurs smaller losses during a crisis). We review, extend,
through some type of resource-based process, and test a detailed, theoretical rationale for why
such as the creation of stronger exchange CSR should preserve CFP. In particular, we pro-
relationships with customers (Brown and Dacin, pose that certain types of CSR activities can gener-
1997) or employees (Turban and Greening, 1997), ate moral capital or goodwill (Godfrey, 2005) that
the co-option or forestallment of government tempers punitive sanctions by stakeholders during
intervention (Neiheisel, 1995), or the enhancement a negative event (i.e., an insurance effect). We pro-
of future revenue growth (Lev, Petrovits, and pose that the resultant moral capital may have little
Radhakrishnan, 2006). to do with generating economic value, but plays a
Our work begins with the general question that substantial role in preserving economic value.
opened the article and moves to a more informa- Second, extant studies use relatively coarse-
tive question ‘when do shareholders gain if a firm’s grained measures for CSR (e.g., usually a single,
strategies include disbursing corporate resources monolithic measure of CSR or a single proxy
through participation in social initiatives?’ In par- such as disclosure or philanthropic giving). We
ticular, we investigate when participation in some continue a trend in the literature toward model-
types of CSR activities creates a form of goodwill ing CSR activities as a set of heterogeneous firm
or moral capital for the firm that acts as ‘insurance- actions (e.g., Godfrey, Hatch, and Hansen, 2008;
like’ protection when negative events occur (Gard- Hillman and Keim, 2001; Mattingly and Berman,
berg and Fombrun, 2006; Godfrey, 2005) that pre- 2006)—thereby avoiding the trap of aggregating
serves shareholder value (CFP). Thus, this article what may be different things into a single mono-
investigates a different aspect of the CSR-CFP lithic construct (Entine, 2003).
relationship; namely, that CSR activities can pro- Third, many of these earlier studies focused on
vide an insurance mechanism to preserve—rather industry- or economy-wide events as the dependent
than generate—CFP. ‘negative event.’ The role of firm-specific char-
Within the CSR-CFP empirical literature, a acteristics in the face of common events clearly
few empirical studies employed constructs and yields illumination; however, matching firm-
variables germane to the insurance perspective specific characteristics to firm-specific events may
allow us to deepen insight and understanding into
1
Given the number of excellent large-scale literature reviews the mechanisms whereby CSR can preserve CFP.
in this area, and in view of limited journal space, we suggest We match firm-level CSR activities with firm-
that interested readers consult the in-text citations for a sense
of this work, as well as the meta-analytic work of Griffin and specific, negative events—permitting insight into
Mahon (1997), Roman, Hayibor, and Agle (1999), or Orlitzky, different event types. Indeed, Schnietz and Epstein
Schmidt, and Rynes (2003). The studies by Griffin and Mahon (2005) in their investigation of the signal value
(1997) and Roman et al. (1999) provide a sense of the tension
in the literature as these authors use the same data to arrive at of CSR during the Seattle World Trade Organiza-
different conclusions. tion conference failure, conclude that their ‘result
Copyright  2008 John Wiley & Sons, Ltd. Strat. Mgmt. J., 30: 425–445 (2009)
DOI: 10.1002/smj
The Insurance Value of CSR 427
applies only to this specific event. More research Yet in the real world, firms invest in risk man-
is needed to test whether similar reactions have agement (such as fire insurance) even though these
occurred in response to other crises’ (Schnietz and investments come at a price in excess of expected
Epstein, 2005: 342). The current article addresses loss. Smith and Stulz (1985) and Stultz (2002)
this important call for research that investigates show that, because violations of the perfect mar-
more than one event. ket assumptions occur, risk reduction adds value
The article proceeds as follows. The first section to shareholders. One example is that risk reduction
describes the theoretical model and contains hypo- protects shareholders against the deadweight costs2
theses for empirical testing. The second section of severe financial distress in a way that investors
outlines the empirical method used to investi- could not accomplish in the market. Specifically, if
gate the hypotheses. The third section presents the managers can reduce the firm’s exposure to firm-
results of the empirical analysis. The fourth and specific risks that give rise to deadweight costs in
concluding section discusses the limitations and a way that investors cannot diversify away, then
interesting implications of the findings. value is added through risk management.
THEORY AND HYPOTHESIS CSR and moral capital

DEVELOPMENT CSR has been defined in the literature as volun-
tary corporate actions designed to improve social
We begin this section by describing the relation- conditions (Mackey, Mackey, and Barney, 2007),
ship between risk management activities and share- or as corporate actions not required by law that
holder value. Next, we discuss how CSR activities attempt to further some social good and extend
may create moral capital as the activities signal beyond the explicit transactional interests of the
the firm’s intentions to engage stakeholders in firm (McWilliams and Siegel, 2000). The volun-
an ‘other-regarding’ manner. Then, we consider tary nature of CSR means that these activities can
how this moral capital could provide insurance- be viewed, broadly, as gifts or grants from the cor-
like protection for the firm in the face of a negative poration to various stakeholder groups.3
event. We also outline three sets of characteris- That does not mean, however, that CSR repre-
tics that could mitigate the insurance-like value of sents a gift with no strings attached. Sociologists
CSR activities: (1) stakeholder characteristics that and anthropologists note that a gift’s ‘strings’ come
influence whether or not CSR engagement will cre- through the process of ‘intangible requitment,’
ate moral capital; (2) firm-specific characteristics or ‘unspecified reciprocity’ where gift giving is
that determine the potential loss during a negative seen as buying respect (Kennett, 1980; Tonkiss
event and may modify the insurance value of CSR and Passey, 1999). CSR activities also play a
activities, and (3) different event-specific charac- ‘rules maintenance’ role by sending information
teristics that should determine the insurance value about the firm to other social actors that reduces
of CSR. search and evaluation costs (Kennett, 1980). In this
case we argue that CSR signals a willingness to
act altruistically (other considering) as opposed
CSR and shareholder value to purely agonistically (self considering) (Sherry,
Insurance and shareholder value 1983).
In the theoretical perfect capital markets modeled 2

Deadweight costs associated with financial distress occur when
by the capital asset pricing model (Markowitz, other real, not merely opportunity, costs are imposed on the
1952; Sharpe, 1964), profit-maximizing managers firm as a result of the loss event. Such costs may include legal
do not invest in risk management. In the theory, costs associated with distress, refinancing costs, the diversion
of managerial time and attention, tighter supplier terms, loss of
expenditures in excess of expected losses made key employees, or the diminution of brand equity or reputation.
by the firm to reduce idiosyncratic risks reduce Stulz (2002) reports estimates of the expected value of these,
expected returns (the value of the firm), while and other, deadweight costs at around 3 percent of asset value.
3
offering no risk reduction beyond that attainable It is the largesse, or gift element, of CSR that so troubles
Friedman (1970) in the opening quote. The attached strings we
by investors holding a diversified portfolio of secu- describe below serve as the lens through which Knauer (1994)
rities. sees the reciprocal value.
DOI: 10.1002/smj
CSR activities will be only one among multi- as discontinuing products or services; some
ple signals that stakeholders use to determine the consequences may be local in their effect, such as
extent of a firm’s altruistic orientation (Goffman, a plant closure. Other negative events may have
1997) and there are very few purely altruistic gifts global ramifications, such as market-destabilizing
(Kennett, 1980). Further, the macro-institutional fraud or environmental disasters. When negative
norm for business is profit making (Friedland events occur, stakeholders respond by punishing
and Alford, 1991), which is known and accepted the firm with sanctions ranging from mild
by public actors in a capitalistic society. Profit (boycotts or badmouthing) to severe (revoking the
making requires a high degree of self-serving, right to do business). Such sanctions, Godfrey
self-dealing, and self-considering behaviors. These (2005) theorizes, follow a legalistic approach in
realities belie the notion that CSR activities can their derivation and application. Stakeholders mete
create a signal that a firm is completely or sub- out punishments based on both the negative effects
stantially altruistic in its orientation. of the act and the perceived state of mind and
We assert that the effect of CSR is not to send intentions of the offender. Punishments will be
a signal of complete altruism (lack of a profit more severe when bad acts are committed by bad
motive or a wholly other-regarding orientation) actors. The attribution of the state of mind is the
toward stakeholders, but rather CSR activities sig- mens rea determination in law (LaFave, 2000).
nal that the firm is not completely self-interested, Theorists argue that CSR engagement generates
that its leaders can, do, and will consider impacts economic value because the moral capital derived
on others or the social good in their decisions; in from CSR provides a mitigating factor in this mens
short, that managers and their firms possess an rea attribution process (Fombrun, Gardberg, and
‘other-considering’ disposition toward their vari- Barnett, 2000; Godfrey, 2005); the goodwill gen-
ous stakeholders. When such signals are received erated should reduce the overall severity of sanc-
and accepted by external stakeholders, firms accrue tions by encouraging stakeholders to give the firm
positive attributions or moral capital (Simon, 1995; ‘the benefit of the doubt’ when ambiguity over
Van Herpen, Pennings, and Meulenberg, 2003) to motive exists (Uzzi, 1997). Put another way, CSR-
the extent these outsiders see the firm engaging in based moral capital creates value if it helps stake-
activities they deem socially or morally desirable holders attribute the negative event to managerial
(Godfrey, 2005). maladroitness rather than malevolence, and tem-
Two features of CSR activity work to deter- per their reactions accordingly. Measuring stake-
mine the strength, and hence potential value, of this holders’ mental processes proves difficult, perhaps
other-considering signal. First, the activity must be even impossible; however, we can observe whether
public knowledge, be it through firm self-reporting stakeholder groups behave in a manner consistent
or the reports and analysis of others. Second, CSR with a theorized attribution process. Such con-
engagement must be substantial enough to create sistency would imply that CSR activity provides
a credible and reasonable declaration of unselfish ‘insurance-like’ protection (Friedman, 1953; God-
intention. CSR activity that captures the atten- frey, 2005).
tion of media or outside evaluators (e.g., industry
groups such as The Conference Board or invest-
ment rating analysts) represents an investment sub- Theory testing: the effects of CSR engagement
stantial enough to be noticed and seen as a credible
commitment. CSR activity meeting these two crite- Based on the preceding logic, it is reasonable to
ria could be referred to as substantial or noteworthy propose that the economic value of CSR activity
CSR. We use the simple term CSR engagement for is contingent: engagement sends a signal of a non-
ease of use in what follows. self-serving orientation that argues for a lack of
mens rea only in the context of a negative event.
In the face of a negative event, investors must
Moral capital as insurance
anticipate the potential actions of other stakeholder
Even under the best of circumstances, business groups. How will affected stakeholders (customers,
activity sometimes creates negative impacts suppliers, regulators, employees, etc.) react? How
among important stakeholder groups. Some will they assess the action and the mens rea condi-
negative impacts may be relatively benign, such tion? What sanctions may occur and at what level
DOI: 10.1002/smj
of severity?4 Negative events, to the extent they create flexibility for managers to pursue strate-
are unanticipated or partially anticipated (Malat- gies that may not be in the direct interests of
esta, 1985), should generate negative stock price those primary stakeholders (e.g., normative power
reactions as investors anticipate negative stake- within the nexus of contracts; see Fama and Miller,
holder reactions; we posit that CSR activity will 1972). CSR activities targeting primary stakehold-
signal to investors the presence of moral capital ers should produce exchange capital among those
that may temper potential sanctions. Firms with no groups—the potential to create more advantageous
CSR activity lack this form of buffering goodwill exchanges between the firm and its primary stake-
and stand exposed to potentially greater impacts. holders. Such CSR activities, however, are less
In this case, CSR activity serves as a signal to likely to produce moral capital ; indeed, precisely
investors about the probable reactions of other key because these actions can be viewed through a
stakeholders. Based on this logic we assert power-exchange lens they may be seen as wholly
consistent with the firm’s profit-making interest
Hypothesis 1: In the context of a negative event, and viewed as merely self-serving, rather than
the decline in shareholder value is smaller for other-regarding, behaviors.
firms that engage in CSR activities than for firms CSR activities directed toward secondary stake-
that do not. holders have the opposite profile. Because these
stakeholders lack both urgency and power to
press their claims on the firm, it seems implausi-
Extending theory: stakeholder characteristics
ble that such activities will be viewed as purely
Scholars writing in both the academic (Mitchell, self-interested actions by managers designed to
Agle, and Wood, 1997) and practitioner (Freeman, enhance the exchange prospects with primary
Harrison, and Wicks, 2008) literature seek to dif- stakeholders. These acts are more likely to be
ferentiate the stakeholder groups that firms interact viewed as voluntary acts of social beneficence,
with. Freeman et al. (2008) create a parsimonious based on normative or pragmatic appeals, and thus
classification: primary stakeholders—those who provide evidence of an ‘other-regarding’ orienta-
are essential to the operation of the business—and tion by the firm’s mangers when compared to CSR
secondary stakeholders—those who can influence activities targeting primary stakeholders.
the firm’s primary stakeholders. Primary stake- Support for this theoretical distinction is found
holders make legitimate claims on the firm and its empirically; Mattingly and Berman (2006) per-
managers and have both urgency and power (util- formed exploratory factor analysis on the Kinder
itarian, coercive, or normative) to enforce those Lydenburg Domini (KLD) investment firm social
claims. In contrast, secondary stakeholders have ratings dataset. They uncover a pattern in the data
legitimate claims on the firm, but lack both urgency similar to the above distinction regarding CSR
and power to enforce those claims (Mitchell et al., activities that target the firm’s primary stakehold-
1997). ers, which they refer to as technical CSR (TCSR),
Because primary stakeholders possess both and those activities that target the firm’s secondary
power and urgency to press their claims for CSR stakeholders, which they refer to as institutional
activities, these activities are highly likely to CSR (ICSR). We maintain that CSR activities are
induce these stakeholders to engage in increased not inherently different but that because stake-
exchanges with the firm (e.g., utilitarian power), holder recipients of CSR activities differ in critical
to placate stakeholders and forestall negative eco- ways, the perception of other-regarding motiva-
nomic consequences (e.g., coercive power), or tions for the behaviors should also differ according
to the preceding logic. Based on this distinction
4
We thank an anonymous reviewer for bringing to our attention and logic we present
the different ways in which stakeholder sanctions may affect
shareholder value. Sanctions may reduce expected cash flows to
the firm, reducing the numerator in the share price equation, or
by increasing the level of risk—the discount rate or denominator Hypothesis 2: In the context of a negative event,
in the equation. We do not seek to differentiate, theoretically or the shareholder value-loss mitigating property
empirically, between the ultimate drivers of a loss in shareholder
value, but rather focus on the role of CSR activities in mitigating of CSR engagement is greater for institutional
the aggregate price reaction. CSR activities than for technical CSR activities.
DOI: 10.1002/smj
Extending theory: firm characteristics operate. Rindova, Pollock, and Hayward (2006)
note that larger firms face greater scrutiny from
Intangible assets
media, special interests, and stakeholders than
The link between risk management activities and their smaller counterparts; larger firms have higher
shareholder value outlined above suggests that the profiles than their smaller counterparts. Godfrey
level of intangible assets held by the firm will (2005) further notes that such a high profile should
impact the value of its CSR activity. Stulz (2002) increase the risk of negative actions toward the
reminds us that insurance—as a form of risk man- firm by outside constituents. If larger firms are
agement—becomes more valuable the higher the more likely to experience negative events, either
costs of financial distress. Negative events that through chance or targeting from constituents, and
reduce a firm’s cash flows or raise its risk level if those negative events are more likely to attract
may create greater levels of financial distress to attention from media, regulators, or other third
the extent that they cause stakeholders to react in parties, then the firm’s CSR engagements will
ways that destroy intangible assets and future cash be considered by more individuals and groups as
flows (or raise risk levels) beyond the direct impact they determine the firm’s mens rea condition. CSR
of the negative event (e.g., customers devalue the engagement by larger firms will be more valuable
brand, key employees leave or work less produc- because it is likely to be used more frequently
tively, suppliers tighten terms, increased costs of in generating mens rea evidence than for smaller
capital preclude otherwise profitable investments, firms, thus
etc.). CSR activities that dissuade stronger punitive
responses should prove more valuable for firms Hypothesis 4: In the context of a negative event,
when stakeholder relationships and the resulting the shareholder value-loss mitigating property
intangible assets play a larger role in creating of CSR engagement is greater for larger firms
value. Given this logic, we offer than for smaller firms.
Hypothesis 3: In the context of a negative event,
the shareholder value-loss mitigating property Negative social impacts
of CSR engagement is greater for firms with
higher levels of intangible assets than for firms Negative social impacts may arise from indus-
with lower levels of intangible assets. try membership (e.g., tobacco, nuclear power, or
gaming) or from a firm’s own actions over time
(e.g., the history of negative union relations at
Size General Motors). The moral capital arising from
We theorize that size plays a role in the value of CSR activities comes from the signal of non-self-
CSR activities in that firm size influences the risk serving intentions; engagement in activities with
level faced by a firm. Simply put, firms with a negative effects on stakeholders sends a clear sig-
larger market presence incur more risk than smaller nal of an intention to act self-interestedly rather
firms. Kimberly (1976) notes that size can be con- than considering and accommodating the needs
sidered as either a dimension, or feature, of an of others or society at large. The juxtaposition
organization or as a context within which managers of other-considering with self-serving behaviors
operate. When viewed as a feature of the orga- should weaken the perceived intensity of commit-
nization, larger market presence (size) translates ment to the former in the presence of strong evi-
into more transactions (both internal and external) dence of the latter. For firms with negative social
and more transactions, ipso facto, lead to a higher impacts, engagement in CSR may be perceived as
probability of negative events; there are simply ‘blood money’ to either atone for past sins (e.g.,
more opportunities for negative outcomes. As a Denny’s support of African-American issues) or it
consequence, larger firms should be more willing may be a substitute/complement for other negative
to engage in CSR to cover the increased risk than practices (e.g., tobacco companies that try to offset
smaller firms. their negative product image through generous phi-
But will CSR engagement be more valuable for lanthropy). In these cases, the signal value of CSR
a larger firm? We assert that it will, when size may be diminished at best, or seen as ingratiation
is viewed as a context within which managers at worst (Jones, 1964). We advance
DOI: 10.1002/smj
Hypothesis 5: In the context of a negative event, Stakeholder-based negative events

the shareholder value-loss mitigating property
When negative events arise because the well-being
of CSR engagement is less for firms with more
of stakeholders is jeopardized in some way (such
negative social impacts than for firms with fewer
as product safety issues or health and safety viola-
negative social impacts.
tions in the workplace), the important elements of
the mens rea attribution process differ from com-
Extending theory: event characteristics petitive events. In these events, there is little ambi-
guity about the badness of the act; the evidence
Competitive events suggests, and may be compelling, that the firm
caused either actual or potential negative outcomes
The institutional environment of business and for key stakeholder groups. What remain ambigu-
hypernorms of competition legitimate aggressive ous, however, are the underlying motivations of the
competitive moves, as long as those moves remain actors involved and the question becomes the clas-
within the bounds of the law or custom (Don- sic mens rea question outlined above: was the act
aldson and Dunfee, 1999). The definition of vig- the deliberate result of a malevolent, self-serving
orous and aggressive actions becomes fuzzy at management, or was the act the result of mal-
the margin, where interpretations may differ (one adapted systems or the maladroit handling of a
firm’s aggressive action is another’s unfair com- situation? In the former case, bad actors caused
petition), or boundary conditions may be unclear a bad act; in the latter, good actors got caught
(whether an action actually constitutes a breach of in a bad situation. In stakeholder-related negative
contract, whether product/process similarity actu- events, the determination of the intentions of the
ally constitute patent infringement). When nega- actors becomes a relevant question in determining
tive events occur that involve interpretations of the appropriate punishment; we expect the moral
acceptable competitive behavior and its bound- capital generated by CSR engagement to play an
aries, we believe stakeholders should expect firms important role in estimates of negative stakeholder
to be both vigorous and aggressive in defense of reactions, and we propose
their competitive actions throughout the litigation
process. Competitive success, necessary for cus- Hypothesis 7: In the context of a stakeholder-
tomers to have good products, employees to retain based negative event, the decline in shareholder
or enhance their jobs, governments and commu- value is smaller for firms that engage in CSR
nities to garner tax revenue, and investors to earn activities than for firms that do not.
returns, depends on a firm’s commitment to pur-
sue its interests and act in a decidedly self-serving
manner in these cases. In these competition-based Integrity-related negative events
negative events, the question turns on whether the
We consider a final set of negative events: those
act itself was a bad act or simply a misinterpre-
where the integrity or moral character of the firm
tation of rule or policy. The issue of intention
is in question. Such events would include, but not
or motivation becomes framed in terms of com-
be limited to, actions by management that vio-
petitive realities: to be a good actor in this sit-
lated well-accepted principles of ethical behavior,
uation means to be self-serving and defend the
such as promise keeping, or resulted in the unfair
firm’s explicit interests. The other-regarding sig-
treatment of employees or other key stakeholders.
nal value of CSR activity—a key element in the
As with stakeholder-based events, the status of the
mens rea attribution process—should be irrelevant
action is relatively unambiguous and negative; the
in this case, as the question does not turn on the
question turns on the, in this case, moral charac-
other-regarding dispositions of the actors involved.
ter/intentions of the actors involved.
Given this logic, we posit
The ambiguity here surrounds management’s
core commitment to abide by well-accepted social
Hypothesis 6: In the context of a competitively principles and norms of ethical behavior; in short,
based negative event, the decline in shareholder stakeholders may doubt the firm’s fundamental
value is unrelated to engagement in CSR activi- integrity. Engagement in CSR activities presents
ties. stakeholders with two choices for resolving the
DOI: 10.1002/smj
ambiguity: (1) they can view the CSR engagement The Socrates dataset has been used by a num-
as evidence of good character and the lack of a ber of researchers in strategy (e.g., Waddock and
‘bad mind’; or (2) they can view CSR as evidence Graves, 1997), human resources (e.g., Turban and
of hypocrisy and a ‘very bad mind.’ We assert Greening, 1997), business and society (e.g., Mat-
that stakeholders should opt for the former inter- tingly and Berman, 2006), and finance (e.g., Fis-
pretation, as Uzzi (1997) finds that one benefit of man, Heal, and Nair, 2005). Researchers have
ongoing relations and goodwill between stakehold- shown that the dataset exhibits robust construct
ers and a firm is the willingness of stakeholders to validity around its underlying measures (e.g., Mat-
give the firm the benefit of the doubt in situations tingly and Berman, 2006; Scharfman, 1996; Szwa-
of ambiguity of intention. We expect that in this jkowski and Figlewicz, 1999). Use of the KLD
case, similar to stakeholder-based events, the moral dataset, however, is not without critics, such as
capital arising from CSR engagement should play a Entine (2003), who notes several credulity prob-
significant role in assessments of managerial inten- lems as the data are stretched to meet the objectives
tions, and we propose of individual researchers.6
Hypothesis 8: In the context of an integrity-

based negative event, declines in shareholder Dependent variable
value are smaller for firms that engage in CSR Our variable of interest is the change in share-
activities than for firms that do not. holder value, as represented by stock price, sur-
rounding a negative event; we began our research
We have now advanced the theoretical argu- by identifying negative legal or regulatory actions
ments and articulated hypotheses about the against the firms in our sample. We reviewed Wall
expected relationships between a firm’s engage- Street Journal articles published between 1992 and
ment in CSR activities and the impact on share- 2003 to match the time frame of available inde-
holder value in the context of a negative event. We pendent variable information. This search yielded
now describe our empirical test of these hypothe- 254 possible events, with 99 firms experiencing
ses. an event and 61 firms not experiencing an event.7
Negative events could be the initiation of a lawsuit
against the firm by a customer, third party, or com-
METHOD petitor, or the announcement of regulatory action
Sample construction 6
Entine’s (2003) primary criticism is that when individual item
We began our analysis with the assumption that scores are combined across dimensions into the larger constructs
legal or regulatory actions against firms may be (such as CSP or corporate social performance) often used by
researchers, the data become meaningless as agglomeration often
rare, and that investors may rely on past CSR per- destroys the information value contained in finer-grained mea-
formance data in factoring in any insurance effect. sures. Similar concerns are found in Hillman and Keim (2001),
For practical research reasons, we also wanted to and Mattingly and Berman (2006). We agree with their rea-
soning. We strike a balance between measuring an individual
limit the number of firms for which we needed to item, or category performance and the creation of an overarch-
search out potential events. We settled on a sam- ing, monolithic measure by collapsing individual CSR activities
ple consisting of the 160 firms that appear from into theoretically relevant, yet mid-range, variables.
7
We performed a number of sensitivity tests on the 96 firms
1991–2002 in the Socrates dataset constructed and in our final sample. We compared these firms against the 61
used by the KLD investment firm; this satisfied our with no event (three firms were dropped for confounding events
assumptions around having a large window to cap- or because they skewed regression results) and found no sta-
tistically significant differences in firm size or CSR activity
ture events, and having a panel of data to use while participation. We also tested these 96 against the S&P 500, the
limiting the number of firms we researched.5 relevant population, for each time period and, again, found no
significant differences in either size or CSR activity participa-
tion. We feel our sample presents a representative sample of
5
We grant that our initial sample was opportunistically drawn; large, publicly held companies. The fact that CSR activity did
however, the core issue with a sample should not be randomness, not differ between the 96 firms sued and the 61 not sued indicates
but representativeness. As we report below, we took several that CSR may have value in insuring against risk but little value
steps to insure that our sample is representative of the larger in mitigating risk; that is, firms that engage in CSR are neither
population, relatively large (Standard & Poor’s [S&P] 500), more, nor less, likely than their nonparticipating competitors to
publicly traded firms. create these types of negative events.
DOI: 10.1002/smj
(e.g., investigation, fines, penalties, etc.) by a gov- Using stock price data from the Center for
ernment entity. We also included the announced the Research on Stock Prices, we calculated the
resolution of regulatory action (e.g., settling an abnormal return over the sixteen days (eight days
adverse equal employment opportunity claim) only before and eight days after) surrounding the event
if data on the initiation of such an event could not interest—legal or regulatory action against a firm.
be found. The cumulative abnormal return (CAR) is the
We screened each of our 254 potential events summed difference of actual return versus expected
for: (1) action against the corporate entity and return
not one of its directors, officers, or employees N
individually; and (2) the presence of any mate- CARi = εit

rial confounding corporate events (e.g., earnings t=1
adjustments or forecasts, announcement of major
sales, mergers) within a seven-day period preced- for the i th firm over the event window centered
ing the focal event. Management (McWilliams and on t (t = o represents the event day) and bounded
Seigel, 1997) and finance (Peterson, 1989) schol- by n (the window on either side). CAR captures
ars reviewing event study protocols caution that the sum of the day-to-day differences between the
material events coincident to the event of interfirm’s actual and expected returns. We relied on the
est, such as earnings announcements, new product interpretation of the Patell z test statistic to deter-
announcements, major sales, or a merger/acquisi- mine the appropriate event window. The Patell z,
tion announcement, confound the ability of the a restrictive parametric test, examines the likeli-
event study methodology to assign abnormal share hood, for each day of the sample period, that the
price movements to the focal event. We eliminated difference between the observed sample mean and
events with potentially confounding events within the expected sample mean values (for each firm
seven days of the event of interest, resulting in a individually) differs from 0 (Patell, 1976). The
sample of 185 events. Patell z values suggest a two-day (t−1 , t0 , the day
Change in shareholder value (our dependent before and the day of the announcement) window
variable) is the unexpected percentage change in where observed returns differed significantly from
the stock price surrounding the event, or the abnor- expected returns.9 Our primary dependent vari-
mal return. Stock prices change daily, due to mar- able is cumulative abnormal return, the cumulative
abnormal stock return (as a percentage) accruing
ket conditions as well as firm specific events. Fol-
over the two-day event window.
lowing Campbell, Lo, and MacKinlay (1997) we
regressed stock returns for the firms in our sam-
ple against returns on a broad market portfolio for Independent variables
a period of 128 to eight trading days prior to the CSR variables
event window, using
The Socrates dataset contains the raw social initia-
tive participation measures we used in the study.
Rit = αi + βi Rmt + εit
The Socrates ratings are publicly available and
known to all interested market participants. While
In this model α captures the average return for not a reputation rating per se, the Socrates data
the i th firm’s stock with no market movement and provide inputs for individual stakeholder groups
β measures that stock’s performance relative to (investors and others) in their formulation of a
the market.8 Rit captures the expected return for firm’s reputation for social involvement or social
the i th firm’s share price on that day controlling responsibility, an important indicator of other-
for market-based fluctuations. εit is the abnormal considering conduct (Fombrun, 1996; Orlitzky
return as represented in and Benjamin, 2001). Inclusion in the Socrates
database also means that a firm’s activities meet
εit = Rit − (αi + βi Rmt )
9
We also used the generalized sign z test, a nonparametric test
with less stringent assumptions about independence and variance
8
We used the Eventus package offered with the Wharton than the Patell test. This test corroborates the results of the Patell
Research Data Service. We used the equally weighted market z scores that the appropriate measurement window is the event
index of stock returns to create the benchmark market returns. day and the day preceding (Cowan, 1992).
DOI: 10.1002/smj
the criteria for material, noteworthy, or substan- by summing the total negative individual item
tial; activity becomes publicly and widely known scores across the six major dimensions.9
and carries an assessment of being substantial in
nature (judged either by quantity or quality). Other variables
To better understand the nature and timing of the
data, we spoke with KLD staff members about how We collected data on firm size, measured as the
these scores are calculated. KLD staff indicated natural logarithm of sales for the fiscal year before
that the rating criteria are robust across firms and the event, and the natural logarithm of the market-
that firms are given an opportunity to respond to a to-book ratio of each firm the day before the
potential rating before it is made public. The data event of the event, from the Compustat data tapes.
collection process for each company follows no Market-to-book has been shown in the literature
rigid annual schedule, but the calendar year-end to correlate strongly with Tobin’s q, the theoreti-
data for any year represents all ratings collected cal standard for measuring intangible assets (Vil-
during the year. To control for the possibility that lalonga, 2004). We parsed the sample according
a KLD rating in any year chronologically followed to event specific characteristics as well. To define
a focal event, we lagged our measures by one the characteristics, we reviewed the original Wall
year to insure that the ratings for each firm were Street Journal articles for each event and coded
public knowledge at the time of the event (Brealey, the events according to keywords in the articles
Myers, and Marcus, 1995). describing the nature of the event, which resulted
The Socrates data contain 41 separate binary in an initial classification of six event categories.
item measures of firm engagement along six social We then collapsed the six into three larger cate-
dimensions (community involvement, corporate gories, legal/regulatory actions arising from com-
governance, employee relations, environmental petitive behaviors, health/safety violations, and
stewardship, diversity, and product quality), with a integrity-based events. Moving from six clear cat-
firm scoring one for the observed presence of the egories to three broader categories was necessary
measure, zero for its absence. Socrates captures to have samples large enough for meaningful anal-
data on activities seen as both positive and neg- ysis. Details on these event types appear in the
ative. Since our theoretical interest surrounds the Appendix.
qualitative choice of engagement in CSR activi- Our theoretical model contains four constructs:
ties, we coded an overall CSR participation vari- (1) CSR activity, (2) the moral capital or good-
able one if, for any one of the positive items, a will arising from that activity, (3) attribution of
firm scores greater than zero, zero otherwise. Fol- the mens rea during a negative event, and (4)
lowing Mattingly and Berman (2006), we coded shareholder value. CSR activity and shareholder
ICSR participation one if the firm scored greater value easily map onto observable variables, and
than zero on any of the positive items under the we assert that a defensible link between CSR
community or diversity dimensions, zero other- activity and goodwill has been established both
wise. We coded TCSR participation one if the firm theoretically (Gardberg and Fombrun, 2006; God-
scored greater than zero on any of the positive frey, 2005) and empirically (Turban and Green-
items under the governance, employee relations, ing, 1997; Simon, 1995; Van Herpen et al., 2003;
or product relations, zero otherwise (Mattingly and Williams and Barrett, 2000). The mens rea attribu-
Berman, 2006).8 We calculated CSR negative level tion process remains as an unobserved, and most
likely unobservable, variable (Godfrey and Hill,
1995); unobservable because it is intrapsychic and
8
Over the 11-year period we investigated, KLD added two new may be a tacit or semiconscious process (Gladwell,
item measures that affected our institutional CSR participation 2005; Winter, 1987). Investors or other stakehold-
variable: community support of education (1993), and the provi-
sion of benefits to gay/lesbian partners (1995). We included these ers may have difficulty explaining why they make
measures in our counts in the relevant years. To test whether the rapid judgments giving a firm ‘the benefit of the
inclusion of these new items skewed the data, we examined the
mean values of all of our CSR variables, year by year. Over the
9
11 years, the mean values of each variable increased; however, We would have preferred a participation variable in negative
for each year-over-year period, the means do not differ signif- CSRs to create a symmetrical variable with our ICSR and TCSR
icantly. The growth in CSR participation does not seem to be variables; however, 94 percent of the firms in the sample engaged
an artifact of including more measures, but rather an increase in in some negative CSR activity, and a participation variable
underlying participation rates by firms. would create no variance for analysis.
DOI: 10.1002/smj
doubt’ (Uzzi, 1997). In light of this, our empirical bootstrap estimation procedure in STATA 9/SE
work cannot provide a definitive, but at best a con- (Chernick, 1999, StataCorp., 2005) to provide
sistent and plausible, test of our theoretical model us with a better estimate of the underlying
(Godfrey and Hill, 1995; Popper, 1962). standard errors in our analysis. The events we
collected occurred over several years, and we
regressed abnormal return on dummy variables
Model specification
representing each year to test for any time-
Based on our theoretical arguments—the ability specific effect on abnormal returns across the
of CSR engagement to protect shareholder value in sample (Hsiao, 2003). None of these regressions
the face a negative event—we employed the event yielded significant parameter estimates, providing
study method, well established in the strategy lit- corroboration that any effects we find are time
erature (e.g., Hoskisson, Harrison, and Dubofsky, independent. We ran regression diagnostics to look
1991; Kumar, 2005; Lee, 2001; Mahoney and for outliers and removed seven observations that
Mahoney, 1993; Reuer, 2001) as well as the finan- substantially skewed regression results, consistent
cial economics literature (see Peterson, 1989, for with normal practice.11 Thus, our final sample
a review). Event studies provide a systematic pro- includes 178 observations. Analysis also indicated
cedure for examining the impact of a variety of that the dependent variable, abnormal return,
business events on the value of the firm (share- was not normally distributed. To test for any
holder value) as all losses in a stock’s price have adverse effects, we also performed median quantile
a permanent effect and represent a tangible loss regression. The median (50-percentile quantile)
in the earnings value of a security to investors and mean regression results are consistent with
(Orlitzky, Schmidt, and Rynes, 2003). While event the ordinary least squares results reported here and
studies utilize efficient markets hypothesis (EMH) available from the authors upon request.
assumptions, the notion that the stock prices adjust The discrete variables we used to measure
to reflect new information (Peterson, 1989) about a CSR engagement provided us with the ability to
firm and its prospects is well accepted even by crit- further analyze the data based on a ‘treatment
ics of the EMH such as Shleifer (2000), who notes groups’ approach. Using the ICSR and TCSR
that the basic assumptions guiding event studies engagement variables as a basis, we classified
produce robust results, and that these studies have firms into one of four ‘treatment’ categories:
significantly enhanced understanding of stock mar- (1) participation in neither type of CSR activ-
ket reactions to events. ity (neither), (2) participation in ICSR activity,
Well-established methodological rules exist for (3) participation in TCSR activity, or (4) participa-
event studies (Peterson, 1989). The procedures for tion in both types of CSR activity (both). Using
event studies are straightforward. First, identify the this approach we compared the abnormal returns
event of interest, and its timing. Second, control for the entire sample and the sample split by the
for other announcements or events that may cause mean values of market-to-book, firm size, negative
investors to alter their valuations of the firm in social impacts, and the three event-specific cate-
addition to the focal event of interest. Third, pre- gories.
dict the stock return in the absence of the event of We used the Mann-Whitney test to analyze the
interest. Fourth, observe how the actual return dif- data in this treatment group regime. The test is
fered from the predicted return. Fifth, use regres- nonparametric and useful for estimating small sam-
sion analysis to test whether or not the variables of ples (Rice, 1988). The Mann-Whitney test cal-
interest are related to the changes in stock price. culates a test statistic based on the rank-ordered
Consistent with the event study protocol, we
used regression analysis to test our hypotheses.
Because we had multiple events from the same 11
We chose to remove the few outliers rather than Winsorize
companies,10 a number of discrete variables, the data for two reasons. First, the very small number of events
lost did not significantly affect the power of our study (Cohen,
and concerns about the underlying distribution 1988). Second, one advantage of dummy variable regression is
structure of our dependent variable, we used the the ability to use the parameter estimates to calculate average
predicted losses (gains). This ability would have been lost by
Winsorizing the variables (Hardy, 1983). Thus, given the focus
10
The average was just under two (1.86) with a maximum of of this study on predicting the loss/gain, we followed the practice
eight events by one firm. of outlier removal.
DOI: 10.1002/smj
measurements of the variable of interest, abnor-
Integrity
0.0000† 1.0
mal return, by treatment group membership. The
test compares an expected rank distribution with
the actual distribution of the ranks. Because the
H/S
test focuses on the ranks rather than the values of
1.0
the variable of interest, it is nonparametric (dis-
Competitive
tribution free) and can be used with small (and
0.0000†
0.0000†
unbalanced) test samples. The relevant test statistic
1.0
is normally distributed and the resulting z-scores
can be used to assess the probability that the values
participation participation
of the variable of interest differs across groups.
0.2363†
0.0918†
−0.1419†
TCSR
1.0
RESULTS
0.3299∗∗∗ †
Table 1 presents the Pearson product moment and
0.2177†
−0.1743†
−0.0575†
ICSR
tetrachoric correlations for the measures used in
the study. Tetrachoric correlations are the appropri-
1.0
ate method for measuring the pairwise correlations
between the discrete (dummy) variables used in
0.1981∗∗∗
negative
the study (Stata, 2005). The correlations indicate
CSR
0.0492
−0.0986
−0.0919
0.1426
that the discrete variables have been properly con-
1.0
structed (e.g., ICSR participation correlates per-
p < 0.01, † = tetrachoric correlation, controlling for categorical variables

fectly with CSR participation because a one on
participation
1.0000∗∗∗ †
1.0000∗∗∗ †
the first variable by definition leads to a one on
0.2097†
−0.0657†
−0.1279†
CSR
0.1051
the second). The results of both tables show that
Pairwise Pearson product moment and tetrachoric correlations (N = 178)
multicolinearity among the variables is unlikely to

1.0
present problems. We note the positive correlation
between CSR participation and ICSR participation
0.2187∗∗
Market/
0.1270∗
0.1002
−0.0867
−0.0133
0.0755
−0.0078
book
with abnormal return, but the lack of correlation

1.0
between size, market-to-book, CSR negative level

with abnormal return.
0.2945∗∗∗
0.4274∗∗∗
0.4005∗∗∗
0.2713∗∗∗
0.0126
−0.0178
−0.1002
0.0867
Size
CSR participation (Hypotheses 1 and 2)

1.0
Table 2 presents the results of our regression tests,

while Table 3 displays the results of the Mann-
Abnormal
.02190∗∗∗
Whitney tests. Table 3 reads similar to a correla-

0.6292 0.4844 0.1734∗∗
3.5786 2.6333 0.0172
0.5000 0.5014 0.0490
0.3146 0.0349 0.0369

0.2977 0.0344 −0.0019
0.0824
Market-to-book 1.2655 0.8348 −0.0082
0.3876 0.0366 −0.0334
tion matrix for each set of firm-specific characteris-

−0.0121 0.0432 1.0
tics we examined. The initial columns contain the

moment data for each treatment group, followed
0.8146 0.0292
by the sample size. The next four columns present

9.2669 1.099
dev.
St.
the key test statistic (the z-score) comparing the

CAR values for each treatment groups.
∗∗∗
Mean
Column 1 of Table 2 shows that the control

p < 0.025,
variables have no independent effects on abnor-

mal return, consistent with the EMH hypothesis.
participation
participation
participation
CSR negatives
Column 2 indicates a positive and statistically sig-

Health/safety
Competitive
∗∗
nificant value for the CSR parameter estimate, pro-

Abnormal
p < 0.05,
Integrity
viding support for Hypothesis 1: the presence of

return
Variable
Table 1.
TCSR
ICSR
an insurance effect for firms that engage in CSR.

CSR
Size
Column 3 decomposes CSR participation into its

∗
DOI: 10.1002/smj
Table 2. Overall model and firm-level effects (standard errors in parentheses)
Regression 1 2 3 4 5 6 7
Constant −0.0437∗∗ −0.0379∗∗ −0.0241 −0.0222∗∗∗∗ −0.0196∗∗∗∗ −0.0266∗ −0.0218∗∗∗

(0.0228) (0.0214) (0.0270) (0.0059) (0.0075) (0.0176) (0.0107)
CSR 0.0241∗∗
participation
(0.0127)
Institutional 0.0158∗∗ 0.0152∗∗∗∗ 0.0161∗∗∗∗ 0.0148∗∗∗ 0.0153∗∗∗∗
CSR
participation
(0.0088) (0.0062) (0.0063) (0.0071) (0.0059)
Technical CSR 0.0010 0.0011 0.0013 0.0009 0.0011
participation
(0.0064) (0.0049) 0.0067) (0.0049) (0.0063)
Market to book −0.0005 −0.0019 −0.0025 −0.0025
(0.0032) (0.0024) (0.0026) (0.0027)
Size 0.0036 0.0010 0.0006 0.0005
(0.0028) (0.0031) (0.0039) (0.0020)
CSR negative −0.0004 −0.0003 −0.0003 −0.0001
level
(0.0014) (0.0017) (0.0019) (0.0014)
R2 0.0073 0.0498 0.0328 0.0462 0.0325 0.0304 0.0303
Wald chi2 1.95 11.67∗∗∗ 10.08∗∗ 6.15∗∗ 6.68∗ 7.40∗ 7.92∗∗
ICSR > TCSR 3.79∗∗ 3.10∗∗ 3.09∗∗ 4.77∗∗∗ 4.24∗∗∗
(χ 2 )
N 178 178 178 178 178 178 178
∗ ∗∗ ∗∗∗ ∗∗∗∗
p < 0.10, p < 0.05, p < 0.025, p < 0.01, one-tailed test
Table 3. CAR means by theoretical treatment groups, firm-specific characteristics
Group Treatment Mean St. dev. N |z| value for meancol < meanrow (meancol > meanrow )
Neither ICSR TCSR Both

Whole sample Neither −0.0266 0.0648 42 —
N = 178 ICSR −0.0031 0.0327 47 2.346∗∗∗∗ —
TCSR −0.0134 0.0275 24 1.093 (1.763)∗∗ —
Both −0.0086 0.0349 65 1.850∗∗ 0.961 0.869 —
Hi M2B firms Neither −0.0432 0.0488 8 —
N = 74 ICSR −0.0056 0.0200 17 1.922∗∗ —
TCSR −0.0189 0.0369 8 1.260 (1.282)∗ —
Both −0.00084 0.0349 38 2.014∗∗∗ 0.291 0.667 —
LoM2B firms Neither −0.0221 0.0681 34 —
N = 104 ICSR −0.0017 0.0384 30 1.655∗∗ —
TCSR −0.0107 0.0224 16 0.707 (1.395)∗ —
Both −0.0091 0.0364 27 0.864 1.039 0.452 —
Large firms Neither −0.0516 0.0436 7 —
N = 91 ICSR −0.0047 0.032 27 2.385∗∗∗∗ —
TCSR −0.0101 0.0037 10 2.342∗∗∗∗ 0.342 —
Both −0.0065 0.0355 47 2.614∗∗∗∗ 0.101 0.010 —
Small firms Neither −0.0215 0.0676 35 —
N = 87 ICSR −0.0009 0.0340 20 1.837∗∗ —
TCSR −0.0158 0.0194 14 0.122 2.345∗∗∗∗ —
DOI: 10.1002/smj
Table 3. (Continued )
Group Treatment Mean St. dev. N |z| value for meancol < meanrow (meancol > meanrow )
Both −0.0143 0.0339 18 0.310 1.462∗ 0.779 —

High CSR negative Neither −0.0253 0.0424 17 —
N = 74 ICSR −0.0097 0.0318 18 0.957 —
TCSR −0.0122 0.0148 6 1.120 (0.200) —
Both −0.0115 0.0404 33 1.239 0.197 —
Low CSR negative Neither −0.0274 0.0773 25 —
N = 104 ICSR 0.0010 0.0331 29 2.221∗∗∗ —
TCSR −0.0142 0.0309 18 0.480 (2.145)∗∗∗ —
Both −0.0057 0.0287 32 1.415∗ 1.415∗ —
∗
p < 0.10, ∗∗ p < 0.05, ∗∗∗ p < 0.025, ∗∗∗∗ p < 0.01
Note: A smaller mean value indicates a positive effect on the CAR.
two theoretically relevant components, ICSR and perspective. Examining Table 3, we see that par-
TCSR participation. The parameter estimate for ticipation in ICSR activities appears to help both
ICSR is positive and significant, while the esti- firms with higher and lower market-to-book val-
mate for TCSR is positive but not significant. We ues. The pattern we see in the whole sample
ran a χ 2 test on the two CSR parameter estimates repeats itself. ICSR activity, as a treatment, leads to
(reported on the second line from the bottom), and statistically significantly lower losses when com-
the result indicates that the parameter estimates are pared to both the neither and TCSR treatment
statistically different. Because the constant term groups. To find further evidence to test Hypoth-
lacked significance, we present Models 4–7 (con- esis 3, we calculated Cohen’s d (the standardized
sidering only the effects of ICSR and TCSR, and mean difference), a measure of the effect size, for
these two variables run with each of the controls the ICSR-neither treatment group comparison. For
separately), which indicate that both the parameter the high market-to-book subsample Cohen’s d is
estimate for the constant and for the CSR variables 1.01—a strong effect (Cohen, 1988)—while the
are consistent and significant across models. small subsample Cohen’s d is 0.37—a moderate
The first group of Table 3, the whole sample, effect. This statistic implies that firms with greater
confirms the results of the regression analysis. levels of intangible assets benefit more by engag-
The CAR for firms with ICSR participation (only) ing in ICSR activities. Based on the collective
differs significantly from both the neither treat- evidence, we find support for Hypothesis 3.
ment group and the TCSR participation (only) Again from Table 3, the large firm subsample
group. ICSR does not differ from the both treat- displays marked differences between the neither
ment group; however, examining the mean values treatment group and any of the CSR treatment
leads us to conclude that the primary insurance- groups. Firms engaging in ICSR activities had
like effect is driven by ICSR participation. The the smallest negative abnormal return and the dif-
evidence in Tables 2 and 3 substantiates the theory ference between the means for ICSR and TCSR
underpinning Hypothesis 2. In short, ICSR partic- attains statistical significance. For the small firm
ipation appears to have an insurance-like effect, subsample, participation in ICSR activities also
while TCSR participation does not. resulted in the smallest loss of shareholder value
and is significantly different from both the neither
treatment group and the TCSR treatment groups.
Firm specific characteristics (Hypotheses 3, 4,
The Cohen’s d statistic for the larger firms indi-
and 5)
cates a strong effect for any CSR treatment group
The data in Table 1, the correlation matrix, indi- (1.34 for ICSR versus neither, 1.34 for TCSR
cate that firm size is positively and significantly v ersus neither, and 1.13 for both versus neither).
associated with all types of positive CSR activi- Cohen’s d for the ICSR-TCSR comparison, 0.21,
ties, supporting one implication of the insurance implies a small effect difference between the two.
DOI: 10.1002/smj
For larger firms, engagement in any CSR activ- ICSR parameter estimate was positive and signif-
ity proved beneficial, while for small firms only icant, the overall goodness of fit for the model
engagement in ICSR activities appears to have an was poor, and the ICSR estimate did not differ
insurance effect. Participation in ICSR activities from the TCSR estimate). The health/safety data
benefited both large and small firms, although the contained in Table 5 reveal a similar story, albeit
effects for larger firms were more pronounced, and somewhat more supportive of Hypothesis 7. The
so we find support for Hypothesis 4. ICSR treatment group outperformed the neither
The results in Table 3 for firms with high and treatment group by a statistically significant mar-
low negative CSR levels show that any insurance gin; however, the health/safety subsample exhib-
effect appears limited to firms with lower levels ited no difference from the TCSR group. The data
of negative social impact. The CAR losses for all provide very tentative support, at best, in regard to
treatment groups in the high negative subsample do the theoretical prediction in Hypothesis 7.
not differ, while in the low negative subsample the Column 3 of Table 4 sets out the regression
ICSR treatment group displays the pattern found results for the integrity-based negative events. The
in other firm-specific characteristics—this group ICSR participation variable is positive, statistically
experienced statistically significant lower losses significant from both zero and the TCSR estimate,
than the neither, TCSR, and both treatment groups. and the model exhibits strong goodness of fit;
Thus, the data supports Hypothesis 5. indeed, this model has the highest R2 of any model
we estimated. The integrity section of Table 5 cor-
roborates the regression results. The ICSR treat-
Event-specific characteristics (Hypotheses 6, 7,
ment group outperformed all other groups (neither,
and 8)
TCSR, and both) by a statistically significant mar-
Tables 4 and 5 present the results of, respectively, gin. The data provide a strong warrant for Hypoth-
the regression and Mann-Whitney analyses for the esis 8; it appears that when a firm’s character is in
data subdivided by event characteristics. Colum 1 question, character evidence has the greatest value.
of Table 4 and the first group in Table 5 show that In summary the data provide evidence consis-
our analysis failed to find a statistically significant tent with the theory that drives Hypotheses 1 (the
effect for either ICSR or TCSR participation when insurance-like role of CSR engagement), 2 (the
the negative event centered on competitive issues, distinction between types of CSR engagement), 3
providing a warrant to not reject Hypothesis 6. (a greater insurance effect for firms with larger val-
The regression analysis also fails to find strong ues of intangible assets), 4 (a greater effect for
evidence of an effect for either type of CSR larger firms), 5 (an insurance effect for firms with
when the negative event arose from the stake- fewer negative social impacts to contend with),
holder concerns of health and safety (while the 6 (CSR activities provide no insurance protection
Table 4. Event-specific characteristics (standard errors in parentheses)
Regression 1 2 3
Competitive Health/safety Integrity
Constant −0.0121∗∗ −0.0236∗∗∗∗ −0.0332∗∗

(0.0072) (0.0088) (0.0189)
Institutional participation 0.0011 0.0182∗∗ 0.0299∗∗
(0.0085) (0.0096) (0.0159)
Technical CSR participation 0.0030 0.0011 −0.0042
(0.0084) (0.0088) (0.0092)
R2 0.0026 0.0602 0.0595
Wald chi2 0.18 3.59 6.10∗∗
ICSR > TCSR (χ2) 0.02 1.91 6.05∗∗∗
N 56 53 69
∗ ∗∗ ∗∗∗ ∗∗∗∗
p < 0.10, p < 0.05, p < 0.025, p < 0.01
DOI: 10.1002/smj
Table 5. Mann Whitney test values (significance levels) for CAR means tests, event specific effects
Group Treatment Mean St. dev N |z| value for meancol < meanrow (meancol > meanrow )

Competitive actions Neither −0.0118 0.0277 12 —
N = 56 ICSR −0.0114 0.0289 10 0.066 —
TCSR −0.0094 0.0348 13 0.598 (0.124) —
Both −0.0079 0.0364 21 0.430 0.254 0.106 —
Health/safety actions Neither −0.0266 0.0419 15 —
N = 53 ICSR −0.0024 0.0329 15 1.659∗∗ —
TCSR −0.0150 0.0198 6 0.856 1.012 —
Both −0.0070 0.0397 17 1.605∗ 0.132 1.260 —
Integrity-based actions Neither −0.0383 0.0981 15 —
N = 69 ICSR 0.0002 0.0349 22 1.949∗∗ —
TCSR −0.0221 0.0087 5 0.218 2.185∗∗∗ —
Both −0.0103 0.0317 27 0.932 1.689∗∗ 1.168 —
∗ ∗∗ ∗∗∗ ∗∗∗∗
p < 0.10, p < 0.05, p < 0.025, p < 0.01
for firms during competitively based events), and assess penalties would facilitate a quantum leap
8 (an insurance-like effect when events challenge in researchers’ ability to understand the insurance
the integrity of the firm). The evidence does not effects observed in this study. Second, the study
support Hypothesis 7. We now consider our find- examined general legal/regulatory actions initiated
ings within the broader context of the CSR-CFP against a firm. This data collection strategy was
relationship literature. consistent with the theoretical model and helped
avoid biasing results ex ante by categorizing some
types of negative events as appropriate (such as
DISCUSSION negative health/safety outcomes) to include, and
some events (such as competitor suits over patent
Before discussing the findings of our study and infringement) as inappropriate to include. It is rea-
considering its broader implications, we recognize sonable to believe that social initiatives targeting a
the limitations of our work and consider a promi- firm’s primary stakeholders, such as TCSR activ-
nent alternative explanation for our findings: that ity promoting positive labor relations, may yield
CSR engagement stands merely as a proxy for insurance-like benefits when the negative event
‘quality management,’ the underlying driver of any affects employees. Future work is needed that
observed relationship between CSR activities and looks at more targeted and specialized types of
an insurance-like effect. We then highlight our key events to determine the breadth of the insurance
findings and raise some questions to guide future effects.
inquiry. Third, different metrics of social initiatives may
yield different results. Again, our theoretical inter-
est lie primarily with the value of engagement
Limitations
in CSR activities at a qualitatively noteworthy
As with all empirical studies, this study has lim- level. The Socrates dataset works well because it
itations that potentially constrain the generaliz- measures a qualitative, ‘stock’ variable—whether
ability of the results. First, while the findings firms participate in activities or not—and facil-
are consistent with the proposed theoretical argu- itates a counting of initiatives. Measures such as
ment, a direct test of the proposed theory is the actual dollar amount invested in CSR activities
not feasible because a key construct—the mens may yield different results. For example, continu-
rea attribution process—is unobservable. Devel- ous data on actual spending for social initiatives
oping methods to uncover and understand the would help provide a finer-grained analysis of the
actual attribution processes stakeholders use to insurance-like value of CSR activities. Although
DOI: 10.1002/smj
such data is hard to get and may bias any results affected stakeholders (thus mitigating cash flow
by including only firms that voluntarily disclose losses/increased risk) and/or (2) the ability to effi-
data, the results of such a study would provide ciently and effectively negotiate a solution to the
an important replication, generalization, and exten- issue that minimizes exposure to the firm’s share-
sion to the results of this research. holders. Given this direct relationship between
Another limitation—and area of interesting, management quality and efficacy of resolution, it
potential future research—concerns the role of seems reasonable to assert that investors would
managerial intentions. It was not possible to ascer- always reward the quality of management, or prox-
tain whether the firms in our study intended to use ies thereof if unobservable, with less negative
CSR as a strategic, insurance-based initiative, a downward revisions of share price.
mechanism to build intangible resources, or as a Our results, particularly the event-specific
manifestation of corporate citizenship. Indeed, this characteristics, provide one situation where CSR
study’s purpose was to investigate whether CSR seems to impact shareholder value (the abnormal
can provide an insurance effect and not whether return)—the integrity-centered events—and two
managers seek such an effect in their decision where there is no observed relationship. If quality
and allocation processes. While difficult to mea- of management should always matter, then the
sure (e.g., if a firm admits it is engaging in CSR lack of an observed relationship between CSR and
for insurance purposes, that admission may or may CFP in these settings casts doubt on the logical
not dissipate any attributions of an altruistic dis- structure of the quality of management hypothesis.
position), future work should focus on revealing Our results suggest two theoretical possibilities:
the underlying managerial motivations for involve- (1) Proponents of the quality of management
ment in CSR activities. explanation will need to establish why the value
of management quality—an arguably generalized
skill—should be contingent on the type of event
Quality of management
experienced, or (2) CSR activities signal more
Waddock and Graves (1997) invoked a quality than merely the quality of a firm’s management
of management construct to provide a theoretical and create their own economic value through the
explanation for the presence of the CSR-CFP rela- processes akin to those we outline previously. The
tionship in their work. Over the last decade, the second option seems to us both more parsimonious
quality of management hypothesis has been used and reasonable.
as an alternative explanation for the presence of
observed positive CSR-CFP relationships (see Lev
Key findings
et al., 2006). In its strongest form, the quality of
management thesis argues that CSR has no inher- The event study corroborates the theoretical argu-
ent value; the apparent value comes as investors ments underpinning this research. In short, par-
interpret CSR engagement as a signal of the quality ticipation in CSR activities does seem to yield
of the underlying management team. Any positive insurance-like protection to a number of firms, sub-
relationship between CSR and CFP is spurious, ject to the constraints we identified. In relation
because it only represents two other, hypothesized to the important, growing research stream on the
but unobservable, positive relationships between value of CSR in preserving CFP, this article pro-
quality of management/CSR and the quality of vides a test of a new, detailed explanation of such
management/CFP. The logic can be summarized phenomena (Godfrey, 2005).
as ‘if quality management drives both CSR and Perhaps a more significant finding than the over-
CFP, then an observed relationship between CSR all support of the risk management view, how-
and CFP will be positive.’ ever, is the finding that the insurance effect holds
Our setting offers a solid test of such an argu- for ICSR activities, aimed at secondary stakehold-
ment. In assessing the potential damage due to ers, but not for TCSR activities focused on the
a negative event, and any subsequent downward firm’s primary transaction partners. Engagement in
revision in share price, it seems logical to us that TCSRs benefited only the large firms in the sample
investors would indeed factor in an assessment while engagement in ICSRs benefited all groups.
of the quality of management, as it pertains to The findings prove consistent with the proposed
either (1) the ability to do damage control with logic that TCSR activities, because they can be
DOI: 10.1002/smj
viewed through the lens of power and exchange question the character of the firm and its man-
value, do not create the same type of moral capital agers. We hope this finding (on firm character
and insurance-like protection as ICSR activities. and events) will spur researchers to investigate the
These results inform future research in two details and dynamics involved in this process. One
ways. First, studies based on monolithic construc- such strategy would be to do multiple, firm-event
tions of CSR—those that collapse different CSR case studies. There is still much to learn from
activities into a single construct or those using research examining firm character and character
only a single proxy—may fail to capture signif- events.
icant differential effects. For example, it would be
interesting to reinterpret the Schnietz and Epstein
(2005) study in this light; does the distinction CONCLUSION
between ICSR and TCSR activities play a role
in an economy-wide crisis? Second, new theoret- CSR activity is often either impugned as prima
ical opportunities now exist to extend the current facie evidence of managerial malfeasance or held
study to both finer-grained CSR activities (how up as a shining example of enlightened self-
do philanthropic CSR activities create value as interest. This study’s findings indicate that CSR,
opposed to diversity-focused CSRs?) and to the particularly investment aimed at secondary stake-
larger theoretical argument (CSR is one driver of a holders, represents a potential method of creating
firm’s reputation, which is in turn a driver of firm- value for shareholders in the face of certain types
specific resources). Several researchers continue to of negative events. In short, good deeds appear to
argue that more is always better (a stronger reputa- earn chits. The results indicate that managers of
tion leads to greater competitive advantages); our firms who engage in CSR activity can create value
findings imply that only more of certain types of at times for their shareholders through the creation
resources (e.g., moral capital that comes only from of insurance-like protection (i.e., they appear to be
ICSR activities) may provide greater advantages. acting as wise fiduciaries and agents of their cor-
That is, the findings suggest that a Mertonian mid- porate owners), which suggests a rapprochement
range conception that bounds the value of CSR between the positions espoused in the opening
may be more appropriate (Merton, 1967). Such comments by Friedman (1970) and Knaeur (1994).
an argument has deep implications for both the-
orists and managers because each must discover
the limits of value-producing investments in these ACKNOWLEDGEMENTS
activities.
Third, we find it compelling that the strongest We thank Jim Walsh and the two anonymous
effect in our analysis surrounds those events that reviewers at SMJ for their pointed and insight-
cast doubt on a firm’s fundamental character as ful feedback in improving the manuscript. We also
an honest, promise keeping entity. In such cases, thank Brad Agle, Troy Carpenter, Jennifer Griffin,
and only in such cases, does evidence that the firm Jeff Harrison, Shelby Hunt, Robb Jensen, Grant
acts in other-regarding ways provide insurance like McQueen, Gerry Sanders, Dave Whetten, mem-
protection. The mean market capitalization for the bers of the BYU Strategy group, and the University
day preceding these integrity-centered events was of Utah Finance faculty—with special thanks to
$32.6 billion. Firms not engaging in ICSR activ- Cal Boardman, Mike Lemmon, and Scott Schaefer
ities lost, on average, $72.4 million, while firms for their helpful feedback or research contributions
that did engage in these activities only lost only to our work.
$22.8 million, less than one-third of their counter-
parts. Almost $50 million worth of insurance pro-
tection from investing in ICSR activities appears REFERENCES
as a wise investment to us, and the size of the gap
between participants and nonparticipants is strik- Berle AA. 1931. Corporate powers as powers in trust.
ing. Harvard Law Review 44: 1049–1074.
Blacconiere WG, Northcutt WD. 1997. Environmental
These numbers suggest a straightforward and information and market reactions to environmental
powerful story; character evidence appears most legislation. Journal of Accounting, Auditing, and
valuable when it speaks directly to events that Finance 12(2): 149–178.
DOI: 10.1002/smj
Blacconiere WG, Patten DM. 1994. Environmental dis- Godfrey PC. 2005. The relationship between corporate
closures, regulatory costs, and changes in firm value. philanthropy and shareholder wealth: a risk manage-
Journal of Accounting and Economics 18: 357–377. ment perspective. Academy of Management Review
Brealey RA, Myers SC, Marcus AJ. 1995. Fundamentals 30(4): 777–798.
of Corporate Finance. McGraw-Hill: New York. Godfrey PC, Hatch NW, Hansen JM. 2008. Toward
Brown TJ, Dacin PA. 1997. The company and the a general theory of CSRs: the roles of benefi-
product: corporate associations and consumer product cence, profitability, insurance, and industry hetero-
responses. Journal of Marketing 61(1): 68–84. geneity. Business & Society, April 24, 2008 DOI:
Campbell JY, Lo AW, MacKinlay AC. 1997. The Econo- 10.1177/0007650308315494.
metrics of Financial Markets. Princeton University Godfrey PC, Hill CWL. 1995. The problem of unob-
Press: Princeton, NJ. servables in strategic management research. Strategic
Chernick MR. 1999. Bootstrap Methods: A Practitioner’s Management Journal 16: 519–534.
Guide. New York: John Wiley and Sons, New York. Goffman E. 1997. Self-presentation. In The Goffman
Cohen J. 1988. Statistical Power for the Behavioral Reader, Lemert C, Branaman A (eds). Blackwell
Sciences (2nd edn). Lawrence Erlbaum: Hillsdale, NJ. Publishers Ltd.: Oxford, UK; 21–25.
Cowan AR. 1992. Nonparametric event study tests. Griffin JJ, Mahon JF. 1997. The corporate social
Review of Quantitative Finance and Accounting 2(4): performance and corporate financial performance
343–358. debate: twenty-five years of incomparable research.
Dodd EM. 1932. For whom are corporate managers Business and Society 36(1): 5–31.
trustees. Harvard Law Review 45(7): 1145–1163. Hardy MA. 1983. Regression with Dummy Variables.
Donaldson T, Dunfee TW. 1999. Ties That Bind . Harvard Sage: Thousand Oaks, CA.
Business School Press: Boston, MA. Hillman AJ, Keim GD. 2001. Shareholder value, stake-
Entine J. 2003. The myth of social investing: a critique holder management, and social issues: what’s the
of its practice and consequences for corporate social bottom line? Strategic Management Journal 22:
performance research. Organization Environment 16: 125–140.
352–368. Hoskisson RE, Harrison JS, Dubofsky DA. 1991. Capital
Fama EF, Miller MH. 1972. The Theory of Finance. market evaluation of m-form implementation and
Dryden Press: Hinsdale, IL. diversification strategy. Strategic Management Journal
Fisman R, Heal F, Nair VB. 2005. Corporate social 12: 271–279.
responsibility: doing well by doing good? Work- Hsiao C. 2003. Analysis of Panel Data (2nd edn).
ing paper, Columbia University Business School, Cambridge University Press: Cambridge, UK.
New York. Available at SSRN: http://ssrn.com/ Jones EE. 1964. Ingratiation: A Social Psychological
abstract=813286. Analyis. Appleton-Century-Crofts: New York.
Fombrun C. 1996. Reputation. Harvard Business School Kennett DA. 1980. Altruism and economic behavior II:
Press: Boston, MA. Private charity and public policy. American Journal of
Fombrun C, Gardberg NA, Barnett ML. 2000. Opportu- Economics and Sociology 39(4): 337–353.
nity platforms and safety nets: corporate citizenship Kimberly Jr. 1976. Organizational size and the struc-
and reputational risk. Business and Society Review turalist perspective: a review, critique, and proposal.
105(1): 85–106. Administrative Science Quarterly 21(4): 571–597.
Freedman M, Stagliano AJ. 1991. Differences in social- Knauer NJ. 1994. The paradox of corporate giving: tax
cost disclosures: a market test of investor reactions. expenditures, the nature of the corporation, and the
Accounting, Auditing, and Accountability Journal social construction of charity. DePaul Law Review
4(1): 68–83. 44(1): 1–97.
Freeman RE, Harrison JS, Wicks A. 2008. Managing for Kumar MVS. 2005. The value from acquiring and
Stakeholders: Survival, Reputation, and Success. Yale divesting a joint venture: a real options approach.
University Press: New Haven, CT. Strategic Management Journal 26(4): 321–331.
Friedland R, Alford RR. 1991. Bringing society back in: LaFave WR. 2000. Criminal Law (3rd edn). West Group:
symbols, practices, and institutional contradictions. In St. Paul, MN.
The New Institutionalism in Organizational Analysis, Lee PM. 2001. What’s in a name.com? The effects
Powell WW, DiMaggio PJ (eds). The University of of ‘.com’ name changes on stock prices and
Chicago Press: Chicago, IL; 232–266. trading activity. Strategic Management Journal 22(8):
Friedman M. 1953. Essays in Positive Economics. 793–804.
University of Chicago Press: Chicago, IL. Lev B, Petrovits C, Radhakrishnan S. 2006. Is doing
Friedman M. 1970. The social responsibility of busi- good good for you: yes, charitable contributions
ness is to increase profit. New York Times Maga- enhance revenue growth. Working paper New York
zine(September 13): , 122, 126 32–33. University Stern School of Business, New York.
Gardberg NA, Fombrun C. 2006. Corporate citizenship: http://archive.nyu.edu/fda/bitstream/2451/27464/2/
creating intangible assets across institutional envi- SSRN-id920502.pdf.
ronments. Academy of Management Review 31(2): Mackey A, Mackey TB, Barney JB. 2007. Corporate
329–346. social responsibility and firm performance: investor
Gladwell M. 2005. Blink: The Power of Thinking Without preferences and corporate strategies. Academy of
Thinking. Back Bay Books: New York. Management Review 32(3): 817–835.
DOI: 10.1002/smj
Mahoney JM, Mahoney JT. 1993. An empirical investi- Roman RM, Hayibor S, Agle BR. 1999. The
gation of the effect of corporate charter antitakeover relationship between social and financial performance:
amendments on stockholder wealth. Strategic Man- repainting a portrait. Business and Society 38(1):
agement Journal 14(1): 17–31. 109–125.
Malatesta P. 1985. Partially anticipated events: a model of Scharfman M. 1996. The construct validity of the Kinder,
stock price reactions with an application to corporate Lydenburg, and Domini social performance ratings
acquisitions. Journal of Financial Economics 14(2): data. Journal of Business Ethics 15(3): 287–296.
237–251. Schnietz KE, Epstein MJ. 2005. Exploring the financial
Margolis JD, Walsh JP. 2001. People and Profits: The value of a reputation for corporate social responsibility
Search for a Link Between a Company’s Social and during a crisis. Corporate Reputation Review 7(4):
Financial Performance. Lawrence Erlbaum: Mahweh, 327–345.
NJ. Sharpe WF. 1964. Capital asset prices: a theory of
Markowitz H. 1952. Portfolio selection. Journal of
market equilibrium under conditions of risk. Journal
Finance 7: 77–91.
of Finance 19(3): 425–442.
Mattingly JE, Berman S. 2006. Measurement of corpo-
Sherry JF. 1983. Gift giving in anthropological perspec-
rate social action: discovering taxonomy in the Kinder
Lydenburg Domini ratings data. Business and Society tive. Journal of Consumer Research 10(2): 157–168.
45(1): 20–46. Shleifer A. 2000. Inefficient Markets: An Introduction
McWilliams A, Siegel D. 1997. Event studies in to Behavioral Finance (Clarendon Lectures in
management research: theoretical and empirical issues. Economics). Oxford University Press: Oxford, UK.
Academy of Management Journal 40(3): 626–657. Simon FL. 1995. Global corporate philanthropy: a
McWilliams A, Siegel D. 2000. Corporate social respon- strategic framework. International Marketing Review
sibility and financial performance: correlation or mis- 12(4): 20–37.
specification. Strategic Management Journal 21(5): Smith CW, Stultz R. 1985. The determinants of
603–609. firms’ hedging policies. Journal of Financial and
Merton RK. 1967. On Theoretical Sociology. Free Press: Quantitative Analysis 20(4): 391–405.
New York. StataCorp. 2005. Stata Statistical Software, Release 9 .
Mitchell RK, Agle BR, Wood DJ. 1997. Toward a theory StataCorp LP: College Station, TX.
of stakeholder identification and salience: defining the Stultz RM. 2002. Risk management and derivatives.
principle of who and what really counts. Academy of Southwestern College Publications: New York.
Management Review 22(4): 853–886. Szwajkowski E, Figlewicz RE. 1999. Evaluating corpo-
Neiheisel SR. 1995. Corporate Strategy and the Politics rate performance: a comparison of the Fortune Repu-
of Goodwill: A Political Analysis of Corporate tation survey and the Socrates social rating database.
Philanthropy in America (American University Studies Journal of Managerial Issues 11(2): 137–154.
Series X, Political Science). Peter Lang: New York. Tonkiss F, Passey A. 1999. Trust, confidence and
Orlitzky M, Benjamin JD. 2001. Corporate social perfor- voluntary organisations(sic): between values and
mance and firm risk: a meta-analytic review. Business insitutions. Sociology 33(2): 257–274.
and Society 40(4): 369–397. Turban DB, Greeening DW. 1997. Corporate social
Orlitzky M, Schmidt FL, Rynes SL. 2003. Corporate performance and organizational attractiveness to
social and financial performance: a meta-analysis. prospective employees. Academy of Management
Organization Studies 24(3): 403–441.
Journal 40(3): 658–672.
Patell JM. 1976. Corporate forecasts of earnings per share
Uzzi B. 1997. Social structure and competition in
and stock price behavior: empirical tests. Journal of
Accounting Research 14(2): 246–276. interfirm networks: the paradox of embeddedness.
Peterson PP. 1989. Event studies: a review of issues Administrative Science Quarterly 42: 35–67.
and methodology. Quarterly Journal of Business and Van Herpen HWI, Pennings JME, Meulenberg MTG.
Economics 28(3): 36–66. 2003. Consumers’ evaluation of socially responsible
Popper KR (ed). 1962. Conjectures and Refutations: The activities in retailing. Working paper, Maastricht
Growth of Scientific Knowledge. Basic Books: New University, Maastricht, The Netherlands.
York. Villalonga B. 2004. Intangible resources, Tobin’s q, and
Reuer JJ. 2001. From hybrids to hierarchies: shareholder sustainability of performance differences. Journal of
wealth effects of joint venture partner buyouts. Economic Behavior & Organization 54: 205–230.
Strategic Management Journal 22(1): 27–44. Waddock SA, Graves SB. 1997. The corporate social
Rice JA. 1988. Mathematical Statistics and Data performance-financial performance link. Strategic
Analysis. Wadsworth & Brooks: Pacific Grove, CA. Management Journal 18(4): 303–319.
Richardson A, Welker M, Hutchinson I. 1999. Managing Williams RJ, Barrett JD. 2000. Corporate philanthropy,
capital market reactions to corporate social responsi- criminal activity, and firm reputation: is there a link?
bility. International Journal of Management Reviews Journal of Business Ethics 26(4): 341–350.
1(1): 17–44. Winter SG. 1987. Knowledge and competence as
Rindova VP, Pollock TG, Hayward M : A. 2006. Celebrity strategic assets. In The Competitive Challenge,
firms: the social construction of market popularity. Teece DJ (ed). Ballinger Publishing Company:
Academy of Management Review 31(1): 50–71. Cambridge, MA; 159–184.
DOI: 10.1002/smj
Appendix: Types of Events, Sorted by Final Category
Category General types of events N
Competitive actions 56
Competition conspiracy
Anti-trust claims
Patent infringements
Price fixing accusations
Health/safety actions 53
Consumer medical/injury issues
Product safety problems
Quality control issues
Environment/pollution events
Integrity-based actions 69
Discrimination claims
Fraud accusations
False claims/dishonesty
Pension or Investor obligation claims
Bribery allegations
DOI: 10.1002/smj

Godfrey Et Al. 2009

Uploaded by

Copyright:

Available Formats

Godfrey Et Al. 2009

Uploaded by

Document Information

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

Godfrey Et Al. 2009

Uploaded by

Copyright:

Available Formats

Strategic Management Journal

Strat. Mgmt. J., 30: 425–445 (2009)

THE RELATIONSHIP BETWEEN CORPORATE SOCIAL

INTRODUCTION is not in the interest of his employers—Friedman

What does it mean to say that the corporate execu-

Copyright  2008 John Wiley & Sons, Ltd.

THEORY AND HYPOTHESIS CSR and moral capital

In the theoretical perfect capital markets modeled 2

Hypothesis 5: In the context of a negative event, Stakeholder-based negative events

Hypothesis 8: In the context of an integrity-

individually; and (2) the presence of any mate- CARi = εit

measurements of the variable of interest, abnor-

p < 0.01, † = tetrachoric correlation, controlling for categorical variables

multicolinearity among the variables is unlikely to

with abnormal return, but the lack of correlation

between size, market-to-book, CSR negative level

CSR participation (Hypotheses 1 and 2)

Table 2 presents the results of our regression tests,

Whitney tests. Table 3 reads similar to a correla-

0.5000 0.5014 0.0490

0.3146 0.0349 0.0369

0.3876 0.0366 −0.0334

tion matrix for each set of firm-specific characteris-

tics we examined. The initial columns contain the

by the sample size. The next four columns present

the key test statistic (the z-score) comparing the

Column 1 of Table 2 shows that the control

variables have no independent effects on abnor-

Column 2 indicates a positive and statistically sig-

nificant value for the CSR parameter estimate, pro-

viding support for Hypothesis 1: the presence of

an insurance effect for firms that engage in CSR.

Column 3 decomposes CSR participation into its

Constant −0.0437∗∗ −0.0379∗∗ −0.0241 −0.0222∗∗∗∗ −0.0196∗∗∗∗ −0.0266∗ −0.0218∗∗∗

Table 3. CAR means by theoretical treatment groups, firm-specific characteristics

Neither ICSR TCSR Both

Both −0.0143 0.0339 18 0.310 1.462∗ 0.779 —

Table 4. Event-specific characteristics (standard errors in parentheses)

Constant −0.0121∗∗ −0.0236∗∗∗∗ −0.0332∗∗

Neither ICSR TCSR Both

Category General types of events N

You might also like

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.