Research Paper No.

1993
The Economics and Politics of Corporate
Social Performance

David P. Baron
Maretno A. Harjoto
Hoje Jo

July 2008

R E S E ARCH P AP E R S E RI E S

The Economics and Politics of Corporate Social Performance

David P. Baron, Maretno A. Harjoto, and Hoje Jo*

July 6, 2008

*Baron (corresponding author) is in the Graduate School of Business, Stanford University,

Stanford, CA 94305, (650) 723-3757, Email: dbaron@stanford.edu. Harjoto is in the
Graziadio School of Business and Management, Pepperdine University, 24255 Pacific
Coast Highway, Malibu, CA 90263, (310) 506-7352, (310) 506-4126 (fax), Email:
Maretno.A.Harjoto@Pepperdine.edu. Jo is in the Department of Finance, Leavey School
of Business, Santa Clara University, 500 El Camino Real, Santa Clara, CA 95053-0388,
(408) 554-4779, (408) 554-4029 (fax), Email: hjo@scu.edu. Jo acknowledges the Leavey
Research Grant and the Breetwor Fellowship for financial support.

The Economics and Politics of Corporate Social Performance

Abstract
This paper provides an empirical test of a theory that relates corporate financial
performance (CFP), corporate social performance (CSP), and social pressure from government
and social activist for improved social performance. A three-equation structural model is
estimated for a large number of firms for 1996-2004. The estimates are statistically and
economically significant and consistent with the theory. CFP as measured by Tobins q is
increasing in CSP, indicating that it is rewarded by consumers, employees, or investors, and
decreasing in social pressure. CSP is increasing in social pressure, indicating that social
performance is responsive to social pressure which mitigates some of the negative effect of social
pressure on CFP. CSP is also increasing in CFP, which is consistent with social performance
being a perquisite for management. Social pressure is decreasing in CFP and increasing in CSP,
which is consistent with social pressure being directed to soft targets that are likely to be
responsive. The measures of CSP and social pressure are also disaggregated, and the relations
among CFP, CSP, and social pressure are largely due to responsive CSP and social pressure
arising from private politics.

JEL Classifications: M14, L21

Keywords: corporate social responsibility, economics and politics

I. Introduction
Corporate social responsibility (CSR) has received increased attention from business, the
media, and researchers. A recent survey by the Economist Intelligence Unit found that 47 percent
of the firms responding agreed that corporate social responsibility (CSR) is a necessary cost of
doing business and 47 percent agreed that it gives us a distinctive position in the market.
(Economist, January 17, 2008) Only 4 percent of the respondents believed that corporate social
responsibility was a waste of time and money. The Economist observed, It is almost
unthinkable today for a big global corporation to be without [a CSR policy]. More than half of
the Fortune 1,000 companies regularly issue CSR reports, and a large number of firms around the
world are engaged in a serious effort to define and integrate CSR into various aspects of their
business (Tsoutsoura 2004). Despite the embrace by much of the business community, the
relations between social performance, financial performance, and social pressure remain as much
a matter of faith and speculation as of evidence, assessment, and calibration.
Baron (2006) distinguishes between corporate social performance (CSP) and CSR, where
the latter involves a moral duty to undertake social activities. In contrast, corporate social
performance (CSP) need not arise from moral considerations. CSP as considered here pertains to
social activities that satisfy two conditions. First, the social activities are beyond the
requirements of the law and regulations. Second, the social activities involve the private
provision of public goods or private redistribution. CSR implies CSP, but CSP need not be
morally motivated. 1 Most of the theoretical and empirical analyses pertain to CSP, since
identifying moral duties and their assignment is difficult. This paper provides empirical evidence
on the relations based on a theory of the underlying economics and politics of corporate social
performance.
CSP could be morally-motivated, but could be strategically chosen to serve the interests
of the firm. For example, strategic CSP could be applied locally to strengthen local community
relations and improve employee morale and productivity. Wal-Mart is the largest corporate
philanthropist, and most of its contributions are at the local level with employees participating in
the allocation of the contributions. CSP could also strengthen a brand as in the case of Starbucks
and Whole Foods. CSP could also be provided in response to social pressure or because it
appeases a stakeholder or pressure group. CSP could also be a perquisite for management in the
sense that managers like the accolades of the advocates of broadened social performance.
The Economist survey found that 23 percent of the firms agreed that corporate social responsibility is
meaningless if it includes things that companies would do anyway.

Agency theory then predicts that the consumption of perquisites should be increasing in the
availability of slack resources and in the discretion available to management.
Empirical studies have examined the relation between CSP and CFP, and while the
results are mixed, overall the research has found a positive but weak correlation. Interpretations
vary, however, and the direction of causality remains an open question. That is, good CSP could
cause good CFP, but good CFP could provide slack resources to spend on CSP. As the
Economist put it, Whether profitable companies feel rich enough to splash out on CSR, or CSR
brings profits.
This paper provides an empirical test of a positive theory of CSP and its relation to CFP
and to social pressure. The theory and empirical analysis view CFP and CSP as jointly
determined by a firm operating in product and capital markets and in the face of social pressure
generated by government, NGOs, and social activists. Social pressure reflects both current
pressure on the firm from government and activist challenges as well as potential future actions
such as legislation or regulation and private actions such as boycotts and media campaigns
against a firm. The theory predicts that CFP is decreasing in social pressure and increasing in
CSP if consumers, employees, or investors sufficiently reward it. It also predicts that CSP is
responsive to social pressure and in CFP as well if CSP is a perquisite for management. The
relation between social pressure and CFP and CSP depends on which firms are targeted for
pressure by government and social activists. The theory predicts that activists can prefer to target
firms that have high CSP because those firms are soft and more likely to concede to demands for
greater CSP.
The empirics are based on a three-equation, structural model in which financial
performance and social performance are chosen by a firm in the face of social pressure from
government and social activists, where the social pressure depends on the activities of the firm.
The empirical results are based on a large number (2,480) of firms for the years 1996-2004 and
are robust to several estimation approaches. The empirical results provide strong support for the
predictions of the theory. Greater CSP results in economically and statistically significant better
CFP, and social pressure reduces CFP, presumably because pressure can damage a brand or
reputation and may signal future actions against the firm. Higher (lagged) CFP leads to greater
CSP, which is consistent with the availability of slack resources that allow managers to consume
CSP as perquisites. CSP is increasing in (lagged) social pressure, so at least a portion of CSP is
responsive. Social pressure is greater for firms in controversial businesses such as nuclear power
and gambling and is greater the higher is CSP and is lower the better is CFP. These findings are

consistent with the soft target hypothesis and indicate that firms do not receive relief from
government and social activists for their CSP. 2
The magnitudes of these effects are also economically significant. For the full population
of firms a one standard deviation increase in (contemporaneous) social pressure results in a 2.7
percent decrease in CFP as measured by Tobins q, and a one standard deviation increase in
(contemporaneous) CSP results in an 8.5 percent increase in CFP. These magnitudes are large,
but the standard deviations of CFP, CSP, and social pressure are greater than their means, so one
standard deviation represents a large change. A one standard deviation increase in (lagged) CFP
results in an increase of 5.6 percent in CSP and a decrease of 4.7 percent in social pressure. CSP
and social pressure are positively associated with a one standard deviation increase in (lagged)
social pressure resulting in a 17.4 percent increase in CSP and a one standard deviation increase
in (lagged) CSP resulting in a 13.9 percent increase in social pressure.
To explore the relations among CFP, CSP, and social performance in more detail, social
pressure is disaggregated into public (government) politics and private (social activists) politics
components and CSP is disaggregated into strategic components likely to increase revenue or
productivity directly and components likely to be a response to social pressure. 3 A five-equation,
structural model is estimated with CFP, public politics pressure, private politics pressure,
strategic CSP, and responsive CSP as endogenous variables.
The pattern of results is consistent with those for the three equation model, but the
components of CSP and social pressure have quite different effects. The positive effect of CSP
on CFP is due to responsive CSP and not to strategic CSP, and the negative effect of social
pressure on CFP is due to private politics and not public politics. Moreover, responsive CSP is
increasing in (lagged) CFP, whereas strategic CSP is not. Responsive CSP is increasing in both
(lagged) public and private social pressure, whereas strategic CSP is increasing in neither.
Moreover, both private and public politics CSP are increasing in (lagged) responsive CSP, but
strategic CSP is increasing in neither. Disaggregation thus provides a deeper understanding of
the relations among CSP, CFP, and social pressure. The action is in responsive CSP and private
politics social pressure. They have direct effects on CFP consistent with the perquisites
hypothesis and is responsive to both components of social pressure. The soft-target hypothesis is
supported with respect to private politics pressure and CFP. Thus, firms are responsive to social

Sarah Connolly of the Freedom From Oil campaign explained the groups demonstrations against Toyota,
Building the Prius does not give Toyota license to mass-produce the Tundra. (The New York Times,
April 7, 2007.)
3
Baron (2001)(2003) introduces the concept of private politics.

pressure, and private politics actors such as social activists and NGOs select soft targets because
they are responsive.
The contributions of this paper are thus threefold. First, the paper estimates a theory in
which CFP, CSP, and social pressure are endogenous. The paper finds economically and
statistically significant relations among CFP, CSP, and social pressure, as predicted by the theory.
Second, in addition to estimating those relations, the paper finds support for four hypotheses
consumers, employees, or investors reward firms for their CSP and penalize them for incurring
social pressure, management may consume CSP as a perquisite, social activist and NGOs choose
soft targets to which to direct social pressure, and CSP is responsive to social pressure. Third,
disaggregating CSP and social pressure shows that the relations among CFP, CSP, and social
pressure are largely due to responsive CSP and social pressure from private rather than public
politics.
The next section reviews the literature on corporate social performance, and Section III
summarizes the theory and the empirical model. Section IV elaborates on the implications of the
theory and develops the principal hypotheses. Section V identifies the data, and Section VI
presents the empirical results and their interpretation. Conclusions are offered in the final section.

II. Literature
Vogel (2005) assessed the literature and concluded that CSR has a small impact on social
issues and on financial performance. Margolis and Walsh (2003) identified 127 empirical studies
and 13 surveys focusing on the relation between CSP and CFP. Although a number of studies
found no relation, they concluded that the overall weight of the studies showed a positive but
weak correlation between the two components of corporate performance. These studies,
however, generally did not explore causality. Since their survey, new theories have been
developed and additional empirical studies have been conducted.
Moon (2007) found no relation between CSP and CFP after controlling for unobserved
heterogeneity among firms. Distinguishing between positive and negative CSP, he found no
relation between negative CSP and CFP but found a negative relation between positive CSP and
CFP. 4 The results presented here differ from these findings and provide a more complete picture
of the relation between CSP, CFP, and social pressure.
Kotchen and Moon (2008) investigated the relation between positive CSP and negative
CSP controlling for firm size, financial structure, and return on assets. They found that higher
levels of negative CSP were associated with higher levels of positive CSP and that this effect was
4

What Moon refers to as negative CSP is used here to measure social pressure.

stronger in industries that received public scrutiny. They also found that concerns about corporate
governance led firms to undertake social activities on matters other than corporate governance.
They regressed positive CSP on negative CSP and found a positive coefficient that they
interpreted as a causal relation. If, however, they regressed negative CSP on lagged positive
CSP, they would also find a positive and significant coefficient, leaving the direction of causation
in doubt. They also found that financial performance measured as return on assets has no effect
on CSP, whereas we find that CSP is strongly increasing and social pressure is strongly
decreasing in CFP as measured by Tobins q.
By using a two-stage approach, a first-stage probit regression and a second-stage Heckman
regression, Harjoto and Jo (2007a) control for endogenous treatment effects, and find that CSR
activity enhances firm value. They also find evidence that the impact of external monitoring by
security analysts is more significant than those of other governance and monitoring mechanisms.
Harjoto and Jo (2007b) find that engagement in CSR is positively associated with governance
characteristics, including board independence, institutional ownership, and analyst following. In
addition, after correcting for endogeneity, they show that CSR engagement positively influences
operating performance and firm value. They conclude that neither a strategic-choice explanation
nor a product-signaling hypothesis is supported as a major motive of CSR activity.
Chatterji and Toffel (2007) examined the toxic releases of firms that were newly covered
by the social rating organization Kinder Lydenberg Domini (KLD). They concluded that firms
with prior good environmental performance did not change their toxic releases performance,
whereas firms with poor environmental performance improved their performance. This is
consistent with social pressure being directed to firms with poor environmental performance.
Becchetti, Ciciretti, and Hasan (2007) considered the effect on stock prices of the exit
from and entry into the Domini 400 Social Index. Using an event study methodology they found
that firms exiting the Index experienced a significant negative abnormal return that persisted.
They also found that the magnitude of the effects of exit and entry on abnormal returns increased
over time and tentatively concluded that the effects were due to the investment practices of
ethically screened funds rather than to information content. This study suggests that some
investors may be willing to pay a premium for CSP and that they are in sufficient numbers that
the premium persists. Heinkle, Kraus, and Zechner (2001) provided a theory in which some
investors shun certain stocks (e.g., non-green firms) and concluded that the proportion of such
investors needed to have a market effect is approximately 20 percent. Hong and Kacpercyzk
(2007) found that returns on sin stocks are higher than market returns and calibration revealed
magnitudes consistent with the theory of Heinkle, Kraus, and Zechner.

The findings of Becchetti, Ciciretti, and Hasan and Hong and Kappercyzk suggest that
investors provide a premium for shares of firms that have good social performance and penalize
firms with poor social performance. Our results are consistent with these results. We find that
social pressure depresses CFP directly, which could reflect an investor effect with some investors
shunning firms that face social pressure, and that firms engaged in controversial businesses face
greater social pressure. We also find that CSP increases CFP.
Fernndez-Kranz and Santal (2007) explored the relation between competition and CSP
and concluded that greater competition as measured by the Herfindahl-Hirschmann Index, import
competition, and other indices of competition are associated with greater CSP. They conclude
that this is consistent with the theory of strategic corporate social responsibility by Baron
(2001)(2006) in which firms engage in social activities because consumers, employees, or
investors are willing to reward firms for those activities. For example, CSP can provide product
differentiation as in Bagnoli and Watts (2003) and Baron (2008a)(2008b), and it may also
improve recruitment and motivate employees to be more productive or accept lower wages. 5
Siegel and Vitaliano (2007) conducted an empirical test of the motivation for corporate
social responsibility. They hypothesized that CSP is strategic and provides product
differentiation or signals high quality to consumers. They test whether firms producing
experience and credence goods are more likely to engage in CSP than firms that produce search
and non-durable experience goods. Their estimates support the hypothesis that CSP is used more
with experience and credence goods, which supports the concept of strategic CSP. They also find
that large firms are not more likely to engage in CSP, and the evidence is mixed about whether
more profitable firms are more likely to provide CSP.
Feddersen and Gilligan (2001) provide a signaling theory in which a social activist can
signal to consumers the attributes of a credence good thus allowing product differentiation even
though attributes are never observable. Fisman, Heal, and Nair (2006) provide a signaling theory
in which firms can either be self-interested or altruistic and can signal their altruism with CSP,
which then provides product differentiation. They hypothesize that product differentiation is
more important in more competitive, and hence less differentiated, industries, and thus CFP and
CSP should be more highly correlated in those industries. Measuring CSP by KLD measures of
corporate philanthropy, they find some support for their hypothesis.

Consumer willingness to pay for social performance may be limited to certain market segments. In an
experiment Sen and Bhattacharya (2001) found that some subjects, those with little liking for CSP, were
less inclined to purchase a product associated with positive CSP. Negative CSP was found to reduce
subjects inclination to purchase a product, so consumers may have an asymmetric response to CSP.

Fisman, Heal, and Nair (2005) found that corporate social responsibility was greater for
consumer products companies that were advertising intensive, which is consistent with the view
that corporate social responsibility is undertaken to enhance a brand or product. Navarro (1988)
provided a model of corporate giving that increases profits and presented empirical evidence that
corporate giving is like advertising and that the profit motive drives giving.
Besley and Ghatak (2007) considered a model in which a subset of caring consumers has
a demand for public goods, where firms can provide those goods jointly with private goods.
Firms differentiate their offerings, and Bertrand competition leaves the surplus with consumers,
but the public good is undersupplied relative to the first-best. Besley and Ghatak also consider
the sustainable level of private provision when consumers have an imperfect monitoring
technology.
Some empirical studies have focused on a single dimension of social performance.
Dowell, Hart, and Yeung (2000) found that firms with a stringent global environmental policy
had better CFP as measured by Tobins q than did firms without such a policy. King and Lewis
(2001) found a positive relation between pollution reduction and Tobins q for a set of
manufacturing firms that reported toxic releases. Both studies estimated single equation models
and neither was able to reach a conclusion about the direction of causality.
A small empirical literature links social pressure and CSP. Maxwell, Lyon, and Hackett
(2001) found that the release of toxic substances by firms was lower the greater the Sierra Club
membership in the state. Hamilton (1993, p. 121) examined expansions of hazardous waste
facilities and concluded that firms took into account the potential for areas to mobilize and
engage in collective action in their selection of counties in which to add capacity He found
that a good proxy for the potential for collective action was voter turnout. Binder and Neumayer
(2005) studied emissions of SO2, smoke, and heavy particulates for a cross-section of 35
countries and found that emissions were lower the greater the presence of environmental NGOs in
a country.
Empirical research on the relation between CSP and CFP is typically silent about the
direction of causation. McGuire, Sundgren, and Schneeweis (1988) studied the relation between
CSP and CFP using Fortune magazines rankings of corporate reputation as an index of CSP and
using several market and accounting measures of CFP. They regressed CSP on CFP prior and
subsequent to the year in which CSP was measured and found that prior financial performance
was a better predictor than subsequent performance. They concluded (p. 869), it may be more
fruitful to consider financial performance as a variable influencing social responsibility than the
reverse. Waddock and Graves (1997) also found that CSP was positively related to prior

financial performance and concluded that their results supported the theory that slack resource
availability and CSP are positively related. The present paper unravels this simultaneity to
provide consistent estimates of both the effect of (lagged) CFP on CSP and the effect of CSP on
CFP.

III. A Theory of CFP, CSP, and Social Pressure

The empirical specification is based on a theory by Baron (2007)(2008a)(2008b) in which
CFP and CSP are jointly determined by a firm that may face social pressure from government or
private citizens and the organizations they form as considered by Baron (2001) and Baron and
Diermeier (2007). The theory also provides a framework for interpreting the empirical results.
Baron (2008a) distinguishes between moral and self-interested motivations for corporate social
activities. Moral motivations are independent of strategic considerations, but could depend on
firm and industry characteristics that determine whether firms encounter moral issues. For
example, a firm in the oil industry necessarily faces issues associated with the environment,
operating in developing countries, and safety concerns. Similarly, social pressure can accompany
the moral issues.
The positive theory tested is based on recent research that treats both CFP and CSP as
choices by firms that are embedded in product and capital markets as well as in a market for
social pressure. Graff Zivin and Small (2005) and Baron (2007)(2008a)(2008b) have provided
theories that yield an endogenous capital market value of CSP. The theory developed by Baron
includes a continuum of citizens with heterogeneous preferences for social causes, two firms, a
capital market and a product market, and an activist NGO that can put potentially harmful social
pressure on firms to provide more CSP. 6 Citizens allocate their endowments between savings,
personal giving to social causes, the purchase of shares of firms that do and do not have CSP, and
contributions to the activist to fund its production of social pressure. In the product market the
firms produce identical products but can use CSP to differentiate (vertically) their products, and
as in the basic theory of quality competition the firms separate with one providing CSP and the
other providing none. The activist chooses one firm to target with social pressure, and ex ante the
firms can provide CSP intended to induce the activist to target the other firm (Baron 2008b). The
amount of the social pressure depends on the contributions by citizens, which depend on their
expectations about the effectiveness of social pressure. Managers can also consume CSP as
perquisites, and shareholders can use managerial compensation contracts to structure the

Social pressure could also come from government.

incentives of management not only to consume perquisites but also to provide CSP that may be
valued by consumers, employees, or investors.
Citizens have warm glow preferences for personal giving to social causes and warm glow
preferences with varying intensities for the social performance of the firm. In the capital market
they trade shares and also can give personally to social causes. This allows the social activities of
a firm to be priced in the capital market. The equilibrium yields an expression for the market
value of the firm that is a linear function of the firms profits from operations and its social
performance, where the latter is valued at a social return determined in the equilibrium. The
social return is less than one unless corporate social performance is a perfect substitute for
personal giving. 7 This provides the basis for empirical specifications with the market value of a
firm a function of its social performance as well as its operating profits.
In the theory firms can choose CSP because it is rewarded. The rewards may come from
consumers, investors, or employees and other suppliers of factor inputs. 8 In addition, firms may
undertake social activities because managers have (warm glow) preferences for those activities.
That is, social activities can be perquisites for managers. Shareholders may use compensation
contracts to provide incentives for managers to engage in CSP, but with hidden actions and
uncertainty about performance those contracts are second best, leaving opportunities for
managers to consume CSP as perquisites. Baron (2008a) finds that the optimal compensation
contracts specify compensation as a weakly increasing function of the social performance of the
firm. Data on compensation contracts are not available, however, so measures of CFP,
management entrenchment, and external monitoring of management are used instead in the
empirical analysis.
Firms may also engage in social activities if those activities reduce potentially harmful
social pressure. Baron (2001)(2008b) and Baron and Diermeier (2007) show that a firm may
engage in social activities to make itself a less attractive target for social pressure from NGOs and
activists. In contrast, an activist seeking to increase aggregate CSP may target firms that are more
likely to respond. Baron (2008b) closes the theory by requiring that the social pressure generated
7

In the equilibrium the firm with CSP attracts a clientele of shareholders for whom CSP is a close
substitute for personal giving, whereas those citizens for whom it is a distant substitute do not hold shares
of the firm but instead support social causes through personal giving. Although there is no shareholder
unanimity in the theory, firms may be thought of as maximizing their market value, which provides an
explanation for strategic CSP.
8
Engaging in CSP could also have other effects on firm performance. For example, a sensitivity to social
concerns could affect the capabilities of the firm and result in improved operating performance. That is,
greater sensitivity of firms to their nonmarket environment may help identify changes that can impact their
operational performance. As the Economist (January 17, 2008) stated, If [CSP] helps business look
outwards more than they otherwise would and to think imaginatively about the risks and opportunities they
face, it is probably worth doing.

10

by the activist be funded by voluntary contributions from citizens, who can also allocate their
endowments directly to social causes and to shares of firms that do and do not provide CSP.
Social pressure should directly affect the market value of a firm through investors
perceptions about future social pressure, possible brand or reputation damage, or possibly
diverting some investors away from the firm. A firm facing potentially harmful social pressure
could undertake CSP in the hope of reducing that pressure or mitigating its effect. Some firms
are subject to social pressure directly from individual citizens, but most are subject to organized
social pressure led by government, NGOs and activists, or interest groups as in the case of labor
unions organizing pressure on Wal-Mart. Given that a firm is subject to social pressure and that
social pressure is harmful to the financial performance of the firm, responding by strengthening
CSP could mitigate harm in the future.
Social pressure arising from the concerns of citizens and government can affect all firms,
but, as indicated, much of social pressure targets selected firms. Some firms are selected for
social pressure because of their abusive actions; e.g., when a firm violates a regulation or law.
Alternatively, some firms incur social pressure due to private politics, as in the case of
environmental NGOs campaigning against oil companies on environmental issues. A firm could
also be the target of social pressure as a result of its economic impact on others. Much of the
social pressure against Wal-Mart was initiated and financed by organized labor that had failed to
organize Wal-Mart employees and that feared the loss of union jobs in the grocery industry.
Social pressure is thus a function of the operational and other activities of the firm and hence is a
jointly determined variable. The theory developed in Baron (2008b) predicts that (1) an activist
can have an incentive to direct private politics pressure to a soft firm, where soft is defined as
having weak incentives to resist pressure and hence be more likely to respond to demands, and
(2) sufficient CSP could lead an activist to select another firm with weaker CSP. A soft firm
could also be one with less ability to resist as a result of weak financial performance. Social
pressure, however, should not depend directly on the external monitoring of management by
analysts and institutional investors.
The theory that underpins the empirical work thus incorporates three decision makers: the
firm, the producers of social pressure, and investors who establish the market value of the firm.
Operations and CSP are jointly chosen by a firm, with the capital market valuing both operational
and social pressure. Social pressure is viewed as controlled by parties outside the firm. It can
directly affect CFP by leading some consumers or investors to shun a firm, and it could signal
factors that could affect financial performance in the future. Social pressure can be directed at a
firm because of its operations, the industry in which it operates, or the broader social

11

environment. That environment is characterized by a set of exogenous factors that reflect product
and capital market characteristics as well as firm descriptors that take into account heterogeneity
among firms.
To develop an empirical model from the theory, CFP will be measured by Tobin's ,
which is a better measure of financial performance that the market value of the firm. Tobins q is
defined as

where

is the market value of the firms securities and

is its total assets. The theory

finds market value to be a function of operational and social performance as

where

is the cash flow resulting from operations

expenditures

and the perquisites

from social pressure

of the firm, including the effects of CSP

consumed by managers,

as mitigated by (lagged)

social performance of the firm where

, and

is the harm to the firm

is the capital market premium for the

is the (endogenous) social return. Data are available

only to measure CSP and S, so the estimated effects correspond to

and

, respectively.
The three primary variables of interest, q, C, and S, have high serial correlation and hence

should be thought of as endogenous state variables that along with contemporaneous factors
influence the operational and social activities of the firm. That is, CSP and social pressure reflect
policies in place and increments to or decrements from those policies in the current period, rather
than being new choices each year. Evidence for this is the high correlation of 0.852 for CSP and
lagged CSP, 0.778 for social pressure and lagged social pressure, and 0.638 for q and lagged q.
The state variables are measured here by lagged values of CFP, CSP, and social pressure.
The empirical specification is:

The system of equations in (1)-(3) is identified from the exclusion of independent

variables. Five exclusions are incorporated in the specification. The first is that there are two
variables, the debt ratio and the dividend ratio, associated with the financial structure of the firm

12

that could affect Tobins q but should not directly affect either CSP or social pressure. The
second pertains to governance and monitoring that affect the discretion of management to engage
in CSP as perquisites. Governance pertains to characteristics of the firm such as entrenchment
that can affect discretion, whereas monitoring pertains to external factors, such as institutional
holdings and analyst coverage that can affect discretion. Social pressure is reasonably a function
of the governance structure of a firm but not the external monitoring by the investment
community. Consequently, monitoring controls, the percent of shares held by large investors,
percent of shares held by institutions, and number of analysts covering the firm, are excluded
from the social pressure equation. Third, KLD identifies exclusionary factors, such as cigarette
production, nuclear power, and gambling, that may be thought of as controversial. Controversy
should affect social pressure and not directly affect CFP or CSP, and hence it is excluded from
those equations. Fourth, KLD counts as CSP a pension program, but that is a part of a firms
compensation package and as such is a private good. That private good should affect CFP but
neither CSP nor social pressure. The fifth is the (lagged) values of the endogenous state variables,
which serve as predetermined variables in (1)-(3). The equations in (1)-(3) satisfy the rank and
order conditions and (over) identify the system.

IV. Elaborating on the Theory and Empirical Specification

CSP could have a number of motivations. In addition to morally-motivated CSP the
theory incorporates five self-interested explanations for CSP. Four focus on parties that could
value and reward or penalize the firm for its social performance. A firm could be rewarded by
consumers, investors, and employees and other suppliers of factor inputs, and CSP could deter
potentially harmful social pressure. Fifth, CSP could be a perquisite for management.
Consumer rewards: Consumers could value CSP and be willing to pay a premium for
the goods and services of a firm with social performance. Hiscox and Smyth (2006) and
Elfenbein and McManus (2007) present empirical studies indicating that some consumers are
willing to pay a premium for private goods that have social performance attached to them. 9 The
firm then has a private incentive to undertake the activities. Corporate social performance then is
9

Hiscox and Smyth (2006) conducted an experiment in which two identical products, towels and candles,
were sold with and without a Fair & Square label that identified the products as being produced under
good working conditions. Consumers were willing to pay a substantial premium for the labeled goods, but
as the authors caution the retailer was known for selling cause-related goods to high income people. The
authors state, it is safe to say that we were looking for a market for labor standards in a place where one
might expect to find it. Elfenbien and McManus (2007) compared the prices of identical items auctioned
on eBays non-charity and charity auction formats, where the latter involves designating a share of the
proceeds to go to a charity. They found an average 6 percent premium for items sold on the charity
auction.

13

a form of product differentiation and could be either a complement to or a substitute for

advertising, branding, and product quality. Navarro (1988) and Fisman, Heal, and Nair (2005)
view CSP as affecting sales in much the same way as advertising.
Employee and supplier rewards: Employees may be more productive for, or accept
lower wages from, a firm that provides social activities they value. Similarly, a firm with good
CSP may be able to attract more talented employees, or suppliers that embrace social
performance may give preference to the firm. Conversely, a firm such as Nike may require its
suppliers to abide by a code of conduct for social performance with respect to workers rights, so a
supplier that abides by the code is rewarded by Nike.
When the rewards exceed the cost of the CSP, these two explanations are referred to as
strategic CSP, since the activities are undertaken to increase profits. Any value maximizing firm
would conduct such social activities independently of any moral motivation. If CSP is strategic,
theories predict both that CSP is decreasing in competitiveness (Bagnoli and Watts 2003) and
increasing in competitiveness (Fisman, Heal, and Nair 2006). Our estimates provide weak
evidence that CSP is increasing in competitiveness.
Investor rewards: Investors may value the social activities of a firm and be willing to
pay a premium for its shares. The theory discussed above predicts that the capital market will
incorporate shareholders valuation (at the margin) of CSP into the market value of the firm.
This does not mean that CSP increases the market value of the firm. Instead, it could mean that
part of the cost of CSP can be offset by the premium investors pay for CSP. That is, if the cost of
CSP is C, the market value of the firm decreases by (1- )C. Unless investors value CSP more
highly than they value their personal giving to social causes, the social return is less than one.
Investors that value CSP could provide the premium by investing through socially responsible
investment funds.
As in Heinkle, Kraus, and Zechner, green investors can shun firms with poor CSP, which
yields an equilibrium premium for firms with good CSP. This can induce firms to improve their
CSP to attract green investors. The same logic implies that a firm that is abusive in the sense of
bad social performance could be penalized by investors in the capital market. Social pressure is a
reflection of allegedly abusive activities, and hence the market value of the firm should be
decreasing in social pressure. Social pressure today also can mean social pressure in the future.
The combined rewards hypothesis: Since the data do not allow the identification of the
individual effects of rewards by consumers, investors, and employees and other suppliers of
factor inputs, we estimate a combined rewards effect. CFP is positively related to CSP only if
consumers, employees, or investors reward the firm sufficiently for its social activities.

14

Social pressure and the responsive CSP hypothesis: A firm could use CSP to reduce
potentially harmful social pressure originating from public politics or private politics. Relieving
social pressure is consistent with stakeholder theory in which firms undertake social activities to
balance the competing pressures from stakeholders. 10 If CSP is responsive to social pressure, the
coefficient of (lagged) social pressure in the CSP equation (2) should be positive. The responsive
CSP hypothesis is that CSP is increasing in social pressure.
The market value of the firm should be less than that of a firm not subject to social
pressure. This can depend on the social issue. For example, Epstein and Schnietz (2002)
conducted an event study of the share prices of firms designated by protestors as abusive in the
context of the 1999 demonstrations that disrupted the Seattle WTO meetings. They found that
firms designated as environmentally abusive experienced a statistically significant decrease in
their market value as a result of the demonstrations. In contrast, they found no significant change
in the value of firms that had been identified as abusive of worker rights.
Management perquisites hypothesis: Managers or board members could undertake
social activities because of their own personal interests. That is, social activities could be
perquisites for managers based on their own moral, warm glow, or self-interested preferences.
Managers could enjoy the accolades that can come from pressure groups and NGOs or receive
moral satisfaction from benefitting others. Social activities could also be payoffs to social
pressure groups in exchange for strengthening the job security of managers, as considered by
Cespa and Cestone (2007).
CSP as a perquisite should depend on the resources (CFP) available to managers and
board members and the discretion they have to serve their own interests. That discretion should
be an increasing function of management entrenchment and a decreasing function of the external
monitoring of management by the investment community. The management perquisites
hypothesis thus is that CSP is increasing in CFP and in management discretion.
Moral management and CSP The motivation for social activities is unobservable, and
corporate statements regarding mission and responsibilities can be motivated by self-interest as
well as by moral duties. If CSP is morally motivated, then it should be independent of financial
performance, whereas it could depend on the operations of the firm since they determine the
social issues it encounters. Morally-motivated CSP could be independent of social pressure, or
social pressure could be associated with the issues on which the firm acts morally. For example,

10

Tirole (2001) considers stakeholder theory from the perspective of corporate governance focusing on
incentive and control issues.

15

moral duty could result from a consensus in society about which party is best placed to deal with
a moral concern, and social pressure could accompany that consensus.
Moral management could be rewarded by consumers, employees, and investors, and they
could reward the firm both because of its motivation and because of its CSP. Fernndez-Kranz
and Santal (2007) argue that CSP should be independent of industry competitiveness if it is
morally motivated. They found that KLD strengths were greater and concerns less in more
competitive industries, suggesting that CSP is strategic rather than morally motivated. As
indicated below, our empirical results are consistent with their finding.
Soft target hypothesis: Social pressure originating from private politics can be directed
by activists and NGOs to selected firms. That pressure could be directed to firms with bad CSP,
but it could also be directed to soft targets. Soft targets could include those firms that have
provided CSP in the past and those that have the weakest incentives to resist the activist demands.
Baron (2008b) shows that firms with high CSP have a weaker incentive to resist social pressure
than do firms with low CSP. Suppose that CSP provides a measure of product differentiation in
the eyes of some consumers, and consider a social activist that can target either a firm with high
CSP or one with low CSP. Targeting consists of a demand for a high level of CSP and the threat
of harm from a campaign such as a boycott or activist generated adverse media coverage. If the
firm with low CSP is targeted and the campaign is successful in the sense that the firm concedes
to the demands, its CSP will be high which reduces product differentiation and intensifies price
competition. The target thus has a strong incentive to resist the campaign. In contrast if the firm
with high CSP is targeted and the campaign is successful, product differentiation increases which
lessens the intensity of price competition and offsets some of the additional cost from the higher
CSP. The incentive of the target with high CSP to resist the campaign is then weaker. Indeed,
high CSP may be a sign that the firm has succumbed to social pressure in the past. 11
Consequently, the activist can prefer to target the firm with the higher CSP. Social pressure thus
should be positively related to (lagged) CSP. Firms with weaker CFP should also be softer
targets.
An alternative to the soft target hypothesis is the pressure release hypothesis that greater
CSP reduces future social pressure because it responds to the expectations and demands of

11

As an example of social pressure not giving credit for CSP, Argenti (2004, pp. 110-111) explained the
decision by the activist organization Global Exchange to target Starbucks to sell Fair Trade Coffee: truly
socially responsible companies are actually more likely to be attacked by activist NGOs than those that are
not, Our interviews with Global Exchange suggested that Starbucks was a better target for the fair trade
issue because of its emphasis on social responsibility, as opposed to a larger company without a socially
responsible bent.

16

government, activists and NGOs, and the public. This hypothesis is also consistent with selection
of the worst offenders as targets rather than firms with high CSP.

V. Data and Measurement

A. Data
The data set is assembled from KLDs Socrates database, the IRRCs governance and director
database, CDA/Spectrum 13(f) filings, and the I/B/E/S database for the period from 1996 to 2004.
Kinder, Lydenberg, Domini Research & Analytics (KLD) is an independent research firm that
compiles data on the social performance of firms. Its Socrates database is the most comprehensive
and widely-used data on social performance and includes social ratings data for more than 3,000
companies. KLDs inclusive social criteria contain strength ratings and concern ratings for
community, diversity, employee relations, environment, governance, human rights, and product.
KLD also has exclusionary screens, such as alcohol, gambling, military, nuclear power, and
tobacco, which are viewed here as affecting social pressure. 12 The inclusive strength and concern
ratings are our main variables for measuring CSP and social pressure, respectively.13 Prior to 2001
KLD data included only approximately 650 firms listed on the S&P 500 or Domini 400 Social Index
as of August of each year. For 2001 and 2002 (2003 and 2004), the KLD data included
approximately 1,100 (3,100) firms listed on the S&P 500, the Domini 400 Social Indexes, or the
Russell 1,000 (Russell 3,000) Indexes as of December 31st of each year.
KLDs definition of corporate governance, which includes compensation, ownership, tax
disputes, and other issues, is quite different from the concept of corporate governance used in finance.
We include KLDs corporate governance dimension in CSP and social pressure, but we also use the
IRRC governance database, the IRRC director database, CDA/Spectrum 13(f) filings, and the
I/B/E/S database to obtain corporate governance and monitoring characteristics that include CEO
ownership, the proportion of outside independent directors, the proportion of institutional holdings,
the proportion of blockholdings, and the number of security analysts following the firm.
Specifically, (i) our sample firm must be available from the IRRC director database; (ii) CEO
ownership and insider blockholding data must be available; (iii) the data for outside institutional
holdings must be available from CDA/Spectrum 13(f) filings. These filings contain quarterly
information on common-stock positions greater than 10,000 shares or $200,000 for each institution
with more than $100 million in securities under management; and (iv) the number of analysts
12

Only 3.3 percent of the firm-year observations have an exclusionary screen.

While the KLD database reflects whether a company is engaged in CSP activities and includes a list of
the types of activities, it does not report how much each firm spends on CSP activities. We are not aware of
any systematic CSP expenditure data.

13

17

following a firm must be available from the I/B/E/S database. We also require that sufficient
COMPUSTAT and Center for Research in Security Prices (CRSP) data are available for our
estimation. This yields a panel of 2,480 firms.
The IRRC volumes are available only for the years of 1993, 1995, 1998, 2000, 2002, and
2004. Following Bebchuk and Cohen (2005) and Gompers, Ishii, and Metrick (2003, 2006), we
fill in the missing years by assuming that the governance provisions reported in any given year
are also in place in the year preceding the volumes publication. For instance, in the case of 1999
for which there is no IRRC volume in the subsequent year, we assume that the governance
provisions are the same as those reported in the IRRC volume published in 1998. To conduct the
robustness test, we also examine firms containing various CSP information from KLD,
governance characteristics from the IRRC, and analyst followings from the I/B/E/S from only the
IRRCs published years of 1993, 1995, 1998, 2000, 2002, and 2004. Our unreported results
suggest that overall results are essentially the same and the main results remain unchanged.

B. Measurement
CSP is assumed to be measured by the Strengths identified by KLD. These strengths
may or not correspond to actual improvements in social welfare but instead could reflect
measures that benefit a group favored by KLD but harms a less favored group. The strengths,
however, correspond to actions that appear to favor the public directly and seem to typically be
cast that way by the media. To assess the quality of the KLD data, Chatterji, Levine, and Toffel
(2007) used the KLD environment ratings to predict toxic releases reported in the governments
Toxic Releases Inventory as well as compliance with environmental regulations including the
number and value of penalties imposed. They concluded that the KLD ratings do not reflect all
the information available on environmental performance but are a good predictor of firms with
the worst environmental performance, although firm size is also a good predictor.
To investigate CSP in more detail the KLD strengths have been decomposed into those
(C1) more likely and those less likely (C2) to be directly rewarded by consumers or employees.
The former category corresponds to strategic CSP, and the latter category is viewed as responsive
CSP; i.e., likely to be a direct response to social pressure. For example, strengths such as
indigenous peoples relations, innovative giving, and diversity on board of directors seem
unlikely to have a direct impact on profits. In contrast, strategic CSP strengths such as protecting
the environment, philanthropy, and product quality, can be advertised to consumers and
emphasized to employees as part of corporate culture. Our assignment of individual KLD
strengths and concerns into categories is a matter of judgment.

18

Social pressure is measured by the KLD Concerns, some of which, such as Community
Other Concern reflecting strong community opposition, are direct measures of pressure,
whereas others, such as the production of ozone depleting chemicals, are indirect measures. To
investigate the source of social pressure, the social pressure measures are disaggregated into those
(Su) that clearly are associated with government such as civil fines and liabilities for hazardous
waste sites, those that are clearly independent of government (Sr) such as workplace reductions
and indigenous peoples relations, and those that may involve both such as the production of
agricultural chemicals. The first category corresponds to public (government) politics, the second
category corresponds to private (NGOs and the public) politics, and the third involves both. The
first two, public and private politics, measures are included in the estimation. Appendix A
identifies the strengths and concerns in each set.
KLD strengths and concerns are measured as 0-1 variables, and the number of measures
varies across the years, so an index is used to aggregate the measures over the categories in
Appendix A. Letting
Appendix A and

denote the indicator variable for firm i with strength j for year t from

the total number of KLD strengths in year t, the index

for firm-year

observation it is

A similar index is constructed for social pressure using the KLD concerns from Appendix A.
The KLD data collection system does not conform exactly to the definition of CSP used
here. KLD, for example, includes as Employee Relations strength an employee retirement
system, which is a private good for workers and a part of a compensation system with many
components among which there are tradeoffs. Consequently, only two strengths, strong union
relations and employee involvement, have been included in CSP, and the other measures in
Employee Relations have been taken out and incorporated in the empirics as a separate
independent variable (employee benefit index, Emp) affecting CFP.
CFP is measured by Tobins q, which as identified above is the market value of a firms
securities divided by its total assets. 14 This is subject to two types of variation across firms that
may be independent of the operational and social choices of a firm. The first includes factors that
can affect overall market values. These include macroeconomic performance, security issues,
and political risks. The second is industry-specific factors such as rising or falling prices due to
shifts in industry demand or restrictions on supply, as in the case of oil or other raw materials.
14

Following Chung and Pruitt (1994), Tobins q is calculated as: [Market value of common stock + Book
value of preferred stock + Book value of long-term debt + Book value of current liabilities Book value of
current assets Book value of Inventories] / Book value of total assets.

19

The first is taken into account using year dummy variables. The second is taken into account
using the 48 Fama and French (1997) industry dummy variables to capture differences across
industries. Some KLD concerns are much more likely in some industries than others. For
example, Investment Controversies pertains only to financial services firms, whereas
Hazardous Waste and Ozone Depleting Chemicals do not pertain to those firms. Similarly,
Indigenous Peoples Relations pertains to firms with international operations. To take into
account the competitiveness of the industry, the industry HHI is used.
The other controls may be thought of as being in three categories: variables that
characterize the operations of the firm (e.g., advertising, R&D, sales), those that characterize the
financial structure and risk of the firm (e.g., debt ratio, dividend ratio, variability of returns), and
those that pertain to governance and monitoring of the firm and its managers (e.g., entrenchment,
board independence, external monitoring). To measure management entrenchment, the Gompers,
Iishi, and Metrick (2003) index (Gindex) is used.
The SEC does not require firms to report advertising expenditures or research and
development expenditures, and 74 percent and 59 percent, respectively, of the firms do not do so.
To identify this non-reporting, a dummy variable with a value of 1 is included if advertising is not
reported, and a dummy variable with a value of 1 is included if R&D is not reported. Also, a
substantial number (1,065 of the 2,480) of firms representing 54 percent of the firm-year
observations have neither KLD strengths nor concerns. This could be because they faced no
social pressure and had no social performance, but it could also be that KLDs data collection
system failed to uncover social pressure or CSP. This is quite possible when KLD expanded its
data set in 2001 and 2003. Consequently, a dummy variable (Dummy KLD) has been used for
these firms. Another potential problem with the data is that the early years of the panel could
contain a selection effect. That is, the firms covered by KLD in the 1990s include those in the
S&P 500 plus those selected for the Domini 400 Social Index, where that selection is based on
CSP. This over represents the presence of CSP in the data set, and hence a dummy variable
(Domini 400) has been used to identify firms in the Domini 400 Index but not in either the S&P
500 or the Russell Indices. This variable is highly significant as expected.
Appendices B and C list the definitions and measures of the variables of interest.

VI. Empirical Results

A. Descriptive Statistics and Bivariate Correlations
Table 1 presents the means, standard deviations, minimums, and maximums of the variables
of interest. The mean of Tobins q is 1.663, whereas the mean of CSP (C1 + C2) is 0.081 and

20

social pressure (Su + Sr) is 0.069. Note that the standard deviations of these variables are larger
than their means, particularly for CSP and S. Table 2 presents the Spearman correlation matrix
for some of the main variables discussed in the previous section. Tobins q is positively
(inversely) related to CSP (social pressure), which may explain the overall (weak) positive
relation identified with empirical literature surveyed by Margolis and Walsh (2003). All CSP
variables are highly and positively correlated with social pressure variables. Firm size measured
by the log of sales or total assets is inversely related to CFP, but positively associated with CSP
and social pressure. As anticipated by the soft target hypothesis, the measure of social pressure, in
particular, the private politics measure is positively associated with lagged CSP measures and the
volatility of earnings while inversely related to lagged Tobins q. All of the above correlations are
statistically significant (p-values < 0.01).

B. Estimation of the Structural Equations

1. The 3-Equation Model
The system of equations in (1)-(3) is estimated for two panels using four procedures:
OLS, equation-by-equation estimation with Firm Fixed Effects (FFE), simultaneous equations
2SLS, and 3SLS. The first panel is the full data set of 11,784 firm-year observations. Because
lagged variables are used, the observation for the first year a firm appears in the data set is not
used as a firm-year observation. In addition, firms come and go as a result of missing data,
acquisitions and private buy-outs, and expansions of the KLD dataset. The second panel thus is
balanced with 484 firms for which data are available for all the years.
Estimates of the model in (1)-(3) for the two panels and four estimation procedures are
presented in Tables 3 and 4, where the t-statistics are adjusted for robust and clustered (by firm)
standard errors. For the unbalanced panel the three estimation procedures without FFE reported
in the left panel of Table 3A identify a strong and consistent set of relations among CFP, CSP,
and social pressure with all the coefficients of the same signs, and all except one are statistically
significant at the 0.01 level. Financial performance measured by Tobins q is strongly and
positively related to CSP, supporting the explanations for CSP based on the combined rewards
hypothesis. Financial performance is decreasing in social pressure, suggesting that social
pressure harms brands or corporate reputation or productivity or turns some investors away from
the firm. CSP is strongly increasing in (lagged) social pressure, so CSP is responsive to social
pressure and thus has an indirect as well as a direct effect on CFP. CSP is strongly increasing in
(lagged) CFP, which is consistent with CSP as management perquisites. Social pressure is
decreasing in CFP and increasing in (lagged) CSP, both of which are consistent with activists

21

having an incentive to target soft firms, as considered in more detail below. Social pressure is
greater for controversial businesses, as expected.
The relations among CFP, CSP, and social pressure could reflect unobserved firm
characteristics that are unchanged over the panel period. Firm fixed effects have thus been
included to check whether the basic relations among CFP, CSP, and social pressure might be due
to unchanging and unobserved firm characteristics. The equation-by-equation estimates are
presented in the right panel of Table 3A. Since the model estimated has three separate decision
makers; (1) firms (2) government and social activists, and (3) the capital market, including FFE
seems unlikely to negate the basic relations, although the coefficients of the controls, most of
which are firm-based could be affected. As indicated in the second panel of Table 3A, some of
the estimates of the basic relations are no longer significant.
Using FFE with the unbalanced panel, however, involves a considerable data and
estimation problem. Many of the firms are in the database only for two years. For example, all
the firms added when KLD added the Russell 2000 to its database in 2003 have only two years of
data. The same is true for other firms that appear for only two years because, for example, of
missing data for some years. For these firms the first data year is used to obtain the values of the
lagged variables, which leaves only one year of data for the estimation. With FFE the data for
these firms in effect clouds the estimation. That is, in a cross-section FFE cannot be used because
they would perfectly account for the residual. Similarly, including firms with one year of data in
the OLS-FFE procedures perfectly accounts for the residual for that firm. This can provide
unreliable estimates. Consequently, we focus on the balanced panel where this problem is not
present to identify the effects of FFE, the relations among CFP, CSP, and social pressure, and the
hypotheses.
Table 4 presents the four estimates for the balanced panel. Comparing the OLS and FFE
estimates indicates all the coefficients of primary interest have the same signs and are statistically
significant albeit mostly with lower (and one higher) significance levels. The basic relations
among CFP, CSP, and social pressure thus persist when FFE are included.
In addition, because our estimations are based on a three-equation, structural model in
which operations and CSP are chosen by a firm in the face of social pressure, we use
simultaneous equations estimation with 2SLS and 3SLS. Both the 2SLS and 3SLS estimates for
the relations among CFP, CSP, and social pressure have the same signs as the OLS and FFE
estimates. The 2SLS estimates have the advantage over OLS and FFE in that the estimated
coefficients should be more efficient, whereas the 3SLS is more efficient, but requires much
stronger assumptions about the error structure of the model. Consequently, in the following

22

discussion we focus on the 2SLS estimates for the balanced panel but only when the estimates are
consistent with the FFE estimates; i.e. their signs are the same and the estimates are statistically
significant for both. 15
The estimated model can be interpreted both as describing a pattern of conduct across a
broad set of firms and as a representation of an individual firm, but since CSP and social pressure
are best viewed as endogenous state variables that are incremented or decremented rather than
fresh choices each period, the former interpretation is better. The pattern then is that firms with
better financial performance have greater CSP and face less social pressure, firms with greater
CSP have better financial performance but face greater social pressure, and firms facing more
social pressure have higher CSP but worse financial performance. CFP is increasing in sales and
sales growth and decreasing in assets and the standard deviation of returns, as might be expected.
It is also increasing in the advertising ratio and the dividend ratio and decreasing in the debt ratio.
It is also increasing in external monitoring of the firm by institutional investors and analysts.
To assess if the estimates are economically significant, Table 5 presents the magnitude of
the effect on each endogenous variable of a one standard deviation increase in a right-side
variable. The economic magnitudes are uniformly greater with 3SLS than with the other three
estimation procedures, giving us less confidence in the 3SLS estimates, so only the results based
on the other three procedures are discussed here. Because of the data and estimation problems
the FFE estimates for the unbalanced panel are ignored. Firms with CSP and social pressure one
standard deviation above the panel means have from 5.2 to 9.5 percent higher and from 2.3 to 5.1
percent lower CFP, respectively. Almost all firms with positive CSP also faced social pressure
and vice versa, so for an individual firm the net effect of CSP and social pressure is the relevant
measure. Relative to the mean of Tobins q the net effect is between 2.5 and 6.4 percent. Firms
with (lagged) Tobins q one standard deviation above the mean encounter 4.7 to 9.3 percent less
social pressure and have 5.6 to 7.7 percent higher CSP. Similarly, firms with (lagged) CSP one
standard deviation above the mean face 7.5 to 13.9 percent greater social pressure, and firms with
(lagged) social pressure one standard deviation above its mean have 6.7 to 17.4 percent higher
CSP. In interpreting the magnitudes recall that the standard deviation of Tobins q is greater than

15

As a check on the plausibility of the estimates, consider two variables for which the signs of the
coefficients are identified by their definitions. The variable Dummy KLD identifies firms with neither
KLD strengths nor concerns, so the coefficients in the CSP and social pressure equations must be negative.
Similarly, the variable Domini 400 identifies firms selected for their social performance, so the coefficient
in the CSP equation should be positive. The estimated equations for all four estimation procedures and
both panels have the correct signs and are significant at the 0.01 level. This provides confidence in the data
and specifications.

23

its mean, and the standard deviations of CSR and social pressure are twice their means. A one
standard deviation difference thus is a large difference relative to the mean.
The estimates do not establish causation but have the same predictive features as causal
relations. That is, consider an exogenous shock that increases social pressure by, for example,
damaging the reputations of firms, as in the case of corporate scandals, media coverage of high
management compensation or backdating of stock options, or foolish investments by banks in
bonds backed by subprime mortgages. Under the framework of the theory the estimated
equations imply that the CFP of the firms contemporaneously decreases due to adverse consumer,
employee, or investor reactions. Firms respond to the increase in social pressure by increasing
CSP, which offsets to some extent the immediate impact of the shock to social pressure on CFP.
The offset could be from increased revenue from consumers who are willing to pay more for the
firms products, from investors who value the CSP, or employees who are motivated by the
greater social performance. Similarly, the immediate decrease in CFP resulting from the shock to
social pressure reduces CSP because slack is reduced and managers reduce their perquisites. The
increase in CSP resulting from the shock to social pressure then attracts more social pressure, and
the immediate decrease in CFP from the shock increases social pressure further. This increase
then has effects in the subsequent period.
Bailey and Moon (2008) identify a mechanism by which social pressure leads to greater
CSP. They studied S&P 500 firms that established public affairs/social responsibility board
committees and found that those receiving social pressure established these committees in an
attempt to mitigate the harm from social pressure. They interpret this result as evidence that
companies try to defend themselves against negative social outcomes through forming a
specialized Board level committee As discussed above the attempt to defend against social
pressure and mitigate its effect appears not to be unsuccessful, since social pressure is increasing
in (lagged) CSP. Perhaps the increase in social pressure would have been greater in the absence
of the committee.
To explore the hypothesis that social pressure is directed to softer targets, it is necessary
to define soft. Three concepts are used here. The first is a firm that has already engaged in CSP,
which can mean that the firm has been responsive to social pressure in the past and may again
respond to social pressure. The second is financial weakness, as discussed above. The third is a
management that is relatively unprotected from the market for control and competitive pressures
and hence is more likely to respond to social pressure to offset the effect on financial
performance. Social pressure is strongly increasing in (lagged) CSP, which is consistent with the
first concept. Social pressure is decreasing in CFP and increasing in the volatility of returns on a

24

firms shares, which is consistent with the second concept. 16 Social pressure is increasing in
industry HHI, which corresponds to a less competitive market, which is inconsistent with the
third concept. 17 The estimate of the coefficient of the Gindex representing management
entrenchment is negative with 2SLS but positive with FFE, so the effect of entrenchment is
unclear. Overall, the estimates provide mild support for the soft-target hypothesis. 18
Support for the perquisites hypothesis is provided by the finding that CSP is increasing in
(lagged) financial performance, which can be interpreted as greater financial slack that provides
more opportunities for managers to engage in CSP. This is supported by the positive and
statistically significant coefficient on the percent of shares owned by the CEO. As with social
pressure, the coefficients of Gindex are of opposite signs in the 2SLS and FFE estimates, so the
effect of entrenchment on CSP is unclear. CSP is decreasing in the HHI, so firms in more
competitive industries have higher CSP. This result is consistent with that of Fernndez-Kranz
and Santal and suggests that at least some portion of CSP is strategic. None of the other
coefficients are both consistent in sign and significant for the balanced panel. Overall, the results
generally, but weakly, favor the managerial perquisites hypothesis.
The estimates for the unbalanced panel are consistent with those for the balanced panel.
For example, comparing the 2SLS estimates for the two panels indicates that the signs of all
estimated coefficients for the relations among CFP, CSP, and social pressure are the same and all
are statistically significant although in some cases at different levels. As expected, the estimates
with FFE for the two panels are not as consistent because of the data and estimation problems
discussed above.

2. The five-equation model

The five-equation, structural model has the same form as the three-equation system, with
dependent variables q, C1, C2, Su, Sr. This system allows a more detailed investigation of the
relations among CFP, strategic CSP (C1), responsive CSP (C2), public politics social pressure
(Su), and private politics social pressure (Sr). The independent variables are the same as for the
three-equations system with both of the disaggregated variables substituted for the aggregated
16

CFP is decreasing in the standard deviation of returns indicating that higher volatility can be interpreted
as weaker financial performance.
17
The coefficient for the R&D ratio is positive and statistically significant, but the interpretation is unclear.
On the one hand, high R&D could reflect a strong firm. On the other hand, however, if a high R&D firm
entails high product and high market risk, then high R&D firm is not necessarily indicative of a strong firm.
18
Support for the soft target hypothesis is provided by King and Soule (2007) who studied which firms
were targeted by social and union protests. They estimated a Probit model and found that protestors tend
to target large, weakly performing firms. Firms that have been targeted by protestors in the past are more
likely to be protested against in the future. We also find that social pressure is directed to larger firms.

25

variables. The system is identified by the same exclusions. To save space Tables 6A and 6B
present only the 2SLS estimates for the unbalanced and balanced panels. The estimates are
generally consistent in sign and significance level, and the only coefficients to be discussed here
are those that have the same sign in both panels and are both statistically significant.
Most of the basic relations reported in Tables 6A and 6B are similar to those for the 3equation model, but the estimated coefficients for the disaggregated variables are informative.
CFP is increasing in responsive CSP (C2) but not in strategic CSP (C1) (for the unbalanced panel).
CFP is decreasing in private politics pressure (Sr), and the coefficients on public politics pressure
(Su) are negative but both are not statistically significant. Responsive CSP is increasing in both
components of social pressure, and the estimated coefficients for strategic CSP are positive but
not significant for the balanced panel. These estimates are thus consistent with the definitions of
responsive and strategic CSP. Responsive CSP is increasing in (lagged) CFP, indicating that
firms respond more the greater is the financial slack. This is also consistent with the perquisites
hypothesis. 19 Private and public social pressures are increasing in (lagged) responsive CSP, and
private politics social pressure is decreasing in (lagged) CFP, both of which are consistent with
the soft target hypothesis. Both private and public social pressures are increasing in KLDs
exclusionary criteria, which is consistent with more controversial lines of business attracting
social pressure.
Disaggregating CSP into strategic and responsive categories thus reveals that the effects
are primarily due to responsive CSP. A one standard deviation higher responsive CSP results in
economic effects on Tobins q that are very close to those in Table 4, indicating considerable
economic significance. 20 Similarly, a one standard deviation increase in private politics social
pressure reduces Tobins q by an estimated 0.9% for the unbalanced panel and 4.1% for the
balanced panel, which are also consistent with the effects estimated for the 3-equation model.
This supports the view that responsive CSP captures the effects of social performance on
financial performance and that private politics is the principal source of social pressure.
In contrast, strategic CSP has little effect on CFP. For the balanced panel CFP is
significantly increasing in strategic CSP but not with OLS and FFE (unreported). For the
unbalanced panel the coefficients are consistently negative although not statistically significant.
The absence of an effect could be due to the absence of a reward by consumers, employees, or
investors for these aspects of CSP. It could also reflect the cost associated with providing

19
20

The estimated coefficient for Gindex is insignificant, however.

For the unbalanced panel the economic effect is 7.3%, and for the balanced panel it is 9.5%.

26

strategic CSP, such as philanthropy and environmental programs. Similarly, the estimated
coefficients for public politics pressure are negative but not uniformly significant.

VII. Conclusions
Despite the frequently claimed causal impacts, the empirical evidence regarding the
relation among CFP, CSP, and social pressure has been mixed. This paper examines the
interrelation among CFP, CSP, and social pressure using a large data set of firms with CSP
engagement in the United States for 1996 to 2004. The underlying theory is confirmed by the
empirical estimates identifying robust relations among CFP, CSP, and social pressure. These
relations are both statistically and economically significant. Specifically, greater CSP is
associated with better financial performance, which supports the combined rewards hypothesis,
and better financial performance is associated with greater CSP, which supports the management
perquisites hypothesis. Greater CSP and weaker CFP result in greater social pressure, which
supports the soft target hypothesis. Greater social pressure results in weaker CFP, reflecting the
effects of pressure from government, social activists, and NGOs. Greater social pressure also
results in greater CSP, so social performance is responsive to social pressure.
To understand the relations among CSP, CFP, and social pressure in more detail, we
disaggregated CSP into two components, one judged to be strategic and consistent with combined
rewards hypothesis and the other judged likely to be a response to social pressure. We also
disaggregated social pressure into two components, one associated with public politics and the
other with private politics. The results show that CFP is increasing in responsive CSP but not in
strategic CSP and is decreasing in private politics social pressure but not in public politics asocial
pressure. Responsive CSP is increasing in both private and public politics social pressure, but
strategic CSP is not. Responsive CSP is increasing in CFP, which is consistent with the
perquisites hypothesis. Finally, both private and public politics social pressures are increasing in
responsive CSP, and private politics social pressure is decreasing in CFP, both of which are
consistent with the soft target hypothesis.

27

References
Argenti, Paul A. 2004. Collaborating with Activists: How Starbucks Works with NGOs.
California Management Review. 47: 91-114.
Bailey, Elizabeth E. and Jon J. Moon. 2008. Board Committees and Integrated Strategy: How
Firms Integrate Market and Non-market Strategies for Mitigation. Working paper, University of
Pennsylvania.
Bagnoli, Mark and Susan G. Watts. 2003. Selling to Socially Responsible Consumers:
Competition and the Private Provision of Public Goods. Journal of Economics and
Management Strategy. 12: 419-445.
Baron, David P. 2001. Private Politics, Corporate Social Responsibility, and Integrated
Strategy, Journal of Economics and Management Strategy. 10 (Spring): 7-45.
Baron, David P. 2003. Private Politics. Journal of Economics and Management Strategy. 21:
31-66.
Baron, David P. 2006. Business and Its Environment, 5th ed. Upper Saddle River, New Jersey:
Prentice Hall.
Baron, David P. 2007. Corporate Social Responsibility and Social Entrepreneurship. Journal
of Economics and Management Strategy. 16: 683-717.
Baron, David P. 2008a. Managerial Contracting and Corporate Social Responsibility. Journal
of Public Economics. 92: 268-288.
Baron, David P. 2008b. A Positive Theory of Moral Management, Social Pressure, and
Corporate Social Performance. Journal of Economics and Management Strategy. (forthcoming)
Baron, David P. and Daniel Diermeier. 2007. Strategic Activism and Nonmarket Strategy.
Journal of Economics and Management Strategy. 16: 599-634.
Becchetti, Leonardo, Rocco Ciciretti, and Iftekhar Hasan. 2007. Corporate Social
Responsibility and Shareholders Value: An Event Study Analysis. Working Paper 2007-6,
Federal Reserve Bank of Atlanta.
Bebchuk, L. and A. Cohen. 2005. The costs of entrenched boards. Journal of Financial
Economics. 78: 409-433.
Besley, Timothy and Maitreeshh Ghatak. 2007. Retailing public goods: The economics of
corporate social responsibility. Journal of Public Economics. 91: 1645-1663.
Binder, Seth and Eric Neumayer. 2005. Environmental pressure group strength and air
pollution: An empirical analysis. Ecological Economics. 55: 527-538.

28

Cespa, Giovanni and Giancinta Cestone. 2007. Corporate Social Responsibility and Managerial
Entrenchment. Journal of Economics & Management Strategy. 16: 741-771.
Chatterji, Aaron K., David I. Levine, and Michael W. Toffel. 2007. How Well Do Social Ratings
Actually Measure Corporate Social Responsibility? Journal of Economics and Management
Strategy. (forthcoming).
Chatterji, Aaron K. and Michael W. Toffel. 2007. Shamed and Able: How Firms Respond to
Information Disclosure. Working paper, Duke University.
Chung, K. and S.A. Pruitt. 1994. A simple approximation of Tobins q. Financial Management
23: 70-74.
Dowell, G., S. Hart and B. Yeung 2000. Do Corporate Global Environmental Standards Create
or Destroy Market Value? Management Science. 46: 1059-1074.
Elfenbein, Daniel W. and Brian McManus. 2007. A Greater Price for a Greater Good?
Evidence that Consumers Pay More for Charity-Linked Products. Working paper, Washington
University.
Epstein, Marc J. and Karen E. Schnietz. 2002. Measuring the Cost of Environmental and Labor
Protests to Globalization: An Event Study of the Failed 1999 Seattle WTO Talks. International
Trade Journal. 41:129-160.
Fama, Eugene F. and Kenneth R. French. 1997. Industry Costs of Equity. Journal of Financial
Economics. 43: 153-197.
Feddersen, Timothy J. and Thomas W. Gilligan. 2001. Saints and Markets: Activists and the
Supply of Credence Goods. Journal of Economics and Management Strategy. 10: 149-171.
Fernndez-Kranz, Daniel and Juan Santal. 2007. When Necessity Becomes a Virtue: The
Effect of Product Market Competition on Corporate Social responsibility. Working paper, Saint
Louis University, Madrid.
Fisman, Ray, Geoffrey Heal, and Vinay B. Nair. 2005. Corporate Social Responsibility: Doing
Well by Doing Good? Working paper, Columbia University.
Fisman, Ray, Geoffrey Heal, and Vinay B. Nair. 2006. A Model of Corporate Philanthropy.
Working paper, Columbia University.
Gompers, Paul, Joy Ishii, and A. Metrick. 2003. Corporate governance and equity prices.
Quarterly Journal of Economics. 118: 107-155.
Gompers, Paul, Joy Ishii, and A. Metrick. 2006. Extreme governance: an analysis of dual-class
firms in United States. Working paper, Harvard University.
Graff Zivin, Joshua and Arthur Small. 2005. A Modigliani-Miller Theory of Altruistic Corporate
Social Responsibility. Topics in Economic Analysis & Policy. 5: Article 10.
Hamilton, James T. 1993. Politics and Social Costs: Estimating the Impact of Collective Action
on Hazardous Waste Facilities. RAND Journal of Economics. 24: 101-125.

29

Harjoto, Maretno A. and Hoje Jo. 2007a. Corporate Governance and Firm value: The Impact of
Corporate Social Responsibility. Working paper, Santa Clara University.
Harjoto, Maretno A. and Hoje Jo. 2007b. Why Do Firms Engage in Corporate Social
Responsibility? Working paper, Santa Clara University.
Heinkle, Robert, Alan Kraus, and Josef Zechner. 2001. The Effect of Green Investment on
Corporate Behavior. Journal of Financial and Quantitative Analysis. 36: 431-449.
Hiscox, Michael J. and Nicholas F.B. Smyth. 2006. Is There Consumer Demand for Improved
Labor Standards? Evidence from Field Experiments in Social Product Labeling. Working
paper, Harvard University.
Hong, Harrison and Marcin Kacperczyk. 2007. The Price of Sin: The Effects of Social Norms
on Markets. Working paper, Princeton University.
King, Andrew A. and Michael J. Lewis. 2001. Does It Pay to Be Green? Journal of Industrial
Ecology. 5: 105-116.
King, Brayden G. and Sarah A. Soule. 2007. Social Movements as Extra-institutional
Entrepreneurs: The Effect of Protests on Stock Price Returns. Administrative Science Quarterly.
52: 413-442.
Kotchen, Matthew J. and Jon Jungbien Moon. 2008. Corporate Social Responsibility for
Irresponsibility. Working paper, University of California-Santa Barbara.
Margolis, Joshua D. and James P. Walsh. 2003. Misery Loves Companies: Rethinking Social
Initiatives by Business. Administrative Sciences Quarterly. 48: 268-305.
Maxwell, John W., Thomas P. Lyon, and Steven C. Hackett. 2000. Self-Regulation and Social
Welfare: The Political Economy of Corporate Environmentalism. Journal of Law and
Economics. 43: 583-618.
McGuire, Jean B., Alison Sundgren, and Thomas Schneeweis. 1988. Corporate Social
Responsibility and Firm Financial Performance. Academy of Management Journal. 31: 854872.
Moon, Jon Jungbien. 2007. In Good Companies? A Critical Evaluation of the Corporate Social
Performance-Corporate Financial Performance Link. Working paper, University of
Pennsylvania.
Navarro, Peter. 1988. Why Do Corporations Give to Charity? Journal of Business. 61: 65-93.
Sen, Sankar and C.B. Bhattacharya. 2001. Does Doing Good Always Lead to Doing Better?
Consumer Reactions to Corporate Social Responsibility. Journal of Marketing Research. 38:
225-243.
Siegel, Donald S. and Donald F. Vitaliano. 2007. An Empirical Analysis of the Strategic Use of
Corporate Social Responsibility. Journal of Economics & Management Strategy. 16: 773-792.

30

Tirole, Jean. 2001. Corporate Governance. Econometrica. 69: 1-35.

Tsoutsoura, M., 2004. Corporate social responsibility and financial performance. Working
paper, Center for Responsible Business, Haas School of Business, University of California at
Berkeley.
Vogel, David. 2005. The Market for Virtue: The Potential and Limits of Corporate Social
Responsibility. Washington, DC: Brookings Institute.
Waddock, Sandra A. and Samual B. Graves. 1997. The Corporate Social PerformanceFinancial Performance Link. Strategic Management Journal. 18: 303-319.

31

Appendix A: KLD Strengths and Concerns Categorization

Panel A: KLD Strengths (Corporate Social Performance, C) Categorization
Strategic CSR (C1)
Community
Generous giving
Non-US charitable giving
Other strength

Responses to social pressure (C2)

Community
Innovative giving
Support for housing
Support for education
Indigenous peoples relations

Environment
Beneficial products
Pollution prevention
Alternative fuels
Property, plant, and equipment

Environment
Recycling
Communications
Other strength

Diversity
Promotion
Family benefits

Diversity
CEO
Board of directors
Women/minority contracting
Employment of the disabled
Progressive gay & lesbian policies
Other strengths

Employee relations (Emp)

Cash profit sharing
Strong retirement benefits
Health and safety strength
Other strength

Employee relations
No layoff policy
Employee involvement
Strong union relations

Human rights
Labor rights strength
Other strength

Human rights
Positive record in South Africa
Indigenous peoples relations

Product
Quality
R&D/Innovation

Product
Benefits to economically disadvantaged

Corporate governance c1

Corporate governance c2
Limited compensation
Ownership strength

32

Panel B: KLD Concerns (Social Pressure, S) Categorization

ConcernsPublic Politics su

ConcernsPrivate Politics sr

ConcernsBoth (su and sr)

Community
Investment controversies
Negative economic impact
Indigenous peoples relations
Other concerns

Environment
Regulatory problems

Ozone depleting chemicals

Substantial emissions
Climate change

Hazardous waste
Agricultural chemicals
Other concern

Diversity
Controversies

Non-representation

Other concern

Employee relations
Poor union relations
Health safety concern

Workplace reductions
Pension/benefits

Other concerns

South Africa
Northern Ireland
Burma
Mexico
International labor
Indigenous peoples relations

Other concerns

Human rights

Product
Product safety

Marketing/contracting
controversy

Antitrust
Other concerns

Corporate governance
Tax disputes

High compensation
Ownership concerns

Source: The Kinder, Lydenberg, and Dominis (KLD) Socrates database

33

Other concern

Appendix B: Strategic choice (C1), reaction to social pressure (C2), employee benefits index
(Emp), public pressure (Su), private pressure (Sr), and KLD Exclusionary indices

Strategic CSP (C1): it is calculated from the sum of all strategic choice criteria (c1) defined in Panel A of
Appendix A for each firm in year t divided by the maximum sum of all strategic choice criteria for all firms
in year t.
Responsive CSP (C2): it is calculated from the sum of all reactions to social pressure (c2) criteria defined in
Panel A of Appendix A for each firm in year t divided by the maximum sum of all reactions to social
pressure criteria for all firms in year t.
Public pressure index (Su): it is calculated from the sum of all public pressure criteria (Su) defined in Panel
B of Appendix A for each firm in each year t divided by the maximum sum of all public pressure criteria
for all firms in year t.
Private pressure index (Sr): it is calculated from the sum of all private pressure criteria (Sr) defined in Panel
B of Appendix A for each firm in each year t divided by the maximum sum of all private pressure criteria
for all firms in year t.
KLD Exclusionary index (KLD Exc): it is calculated from the sum of all KLD exclusionary screens
(Alcohol, Gambling, Firearms, Military, Nuclear Power, and Tobacco) defined in KLD Socrates database
for each firm in each year t divided by the maximum sum of all KLD exclusionary screens for all firms in
year t.
Employee index (Emp): it is calculated from the sum of Strong union relationships, Cash profit sharing,
Strong retirement benefits, Health and safety benefits, and Other strengths in KLD Employee Relations
Strengths criteria from Panel A of Appendix A for each firm in each year t divided by the maximum sum
of Strong union relationships, Cash profit sharing, Strong retirement benefits, Health and safety benefits,
and Other strengths in KLD Employee Relations Strengths criteria for all firms in year t.

34

Appendix C: Variable Definitions and Measures

Variable
Tobin Q

Definition
(Market value of common equity + Preferred Stock + Total Debt)/Total Assets as a
measure of firms Corporate Financial Performance (CFP)
Lg(Tobin Q)
One year lag of Tobin Q
Dummy KLD
Dummy variable = 1 if firms do not have any KLD Strengths or Concerns Scores
Corporate Social Performance (CSP) Index
(C1+C2)
Lg(C1+C2)
One year lag of Corporate Social Performance (CSP) Index
C1
Strategic Choice Index
C2
Reaction to Social Pressure Index
Lg(C1)
One year lag of C1
Lg(C2)
One year lag of C2
Emp
Employee Index, consists of Strong union relationships, Cash profit sharing, Strong
retirement benefits, Health and safety benefits, and Other strengths in KLD Employee
Relations Strengths criteria.
Social Pressure Index from Public (su) and Private (sr) Pressures
(Su+Sr)
Lg(Su+Sr)
One year lag of Social Pressure Index
Su
Public Pressure Index
Sr
Private Pressure Index
Lg(Su)
One year lag of Public Pressure Index
Lg(Sr)
One year lag of Private Pressure Index
Domini400
Dummy variable = 1 if firms are in Domini400 but not in SP500 or Russell 1000/2000
Firms with Domini400 = 1 is also known as the Subsample of Domini400
KLD Exc
KLD exclusionary criteria index from the KLD Exclusionary Screens including Alcohol,
Gambling, Firearms, Military, Nuclear Power, and Tobacco
Ln(Sale)
Natural log of firms annual net sales
Ln(Asset)
Natural log of firms annual total assets
Debtr
Long term debt divided by total asset
Rndr
Research and development expense divided by total sales
Advr
Advertising expense divided by total sales
Rndumy
Dummy variable = 1 if firms do not have reported Research and development expense
Advdumy
Dummy variable = 1 if firms do not have reported advertising expense
IndusHHI
Industry Herfindahl-Hirschman Index calculated based on firms annual sales using the
Fama-French 48 Industries
Capxr
Capital expenditure expense divided by total sales
Salegrw
Sales growth rate from previous year to current year
Divr
Dividend divided by book value of equity
Stdret
Standard deviation of monthly stock returns three years prior to current year
Gindex
Gompers, Ishii and Metrick index
Pctdirshr
Percentage of director shares ownership
Pctceown
Percentage of CEO shares ownership
Pctindep
Number of independent outside directors/Number of total directors
Ln(Block)
Natural log of sum of total blockholdings (5% or more)
Pctinsti
Percentage of institutional share ownerships
Loganal
Natural log of (number of analysts + 1 )
Note: Strategic choice (C1), reaction to social pressure (C2), employee index (Emp), public pressure (Su),
and private pressure (Sr) indices are calculated based on the sum of KLD criteria for each of these measures
indicated in Appendix A for each firm divided by the maximum sum of KLD criteria for each of these
measures year by year since KLD criteria and availability of KLD scores in each criteria changes year by
year. Appendix B provides description to construct these indices.

35

Table 1. Descriptive Statistics

This table presents the means, standard deviation, minimum, and maximum of the variables of interest. See
Appendix C for variable definitions.
Variable
Tobin Q
Lg(Tobin Q)
(C1+C2)
Lg(C1+C2)
C1
C2
Lg(C1)
Lg(C2)
(Su+Sr)
Lg(Su+Sr)
Su
Sr
Lg(Su)
Lg(Sr)
KLD Exc
Domini400
Emp
Dummy KLD
Ln(Sale)
Ln(Asset)
Debtr
Rndr
Advr
Rndumy
Advdumy
IndusHHI
Capxr
Salegrw
Divr
Stdret
Gindex
Pctdirshr
Pctceown
Pctindep
Ln(blks)
Pctinsti
Loganal

Obs
11,784
11,784
11,784
11,784
11,784
11,784
11,784
11,784
11,784
11,784
11,784
11,784
11,784
11,784
11,784
11,784
11,784
11,784
11,784
11,784
11,784
11,784
11,784
11,784
11,784
11,784
11,784
11,784
11,784
11,784
11,784
11,784
11,784
11,784
11,784
11,784
11,784

36

Mean
1.663
1.762
0.081
0.065
0.036
0.044
0.029
0.035
0.069
0.049
0.031
0.037
0.026
0.023
0.033
0.057
0.053
0.539
7.300
7.606
0.240
0.035
0.009
0.589
0.741
0.116
0.071
0.121
0.037
11.818
9.226
0.077
1.576
0.636
13.907
61.226
2.232

Std. Dev
1.808
2.253
0.152
0.141
0.074
0.096
0.069
0.088
0.133
0.119
0.094
0.077
0.089
0.062
0.130
0.232
0.153
0.499
1.452
1.664
0.193
0.082
0.026
0.492
0.438
0.108
0.104
0.302
0.360
6.333
2.706
0.209
5.203
0.183
5.405
19.311
0.699

Min
0.043
0.058
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1.078
0.506
0
0
0
0
0
0.019
0
-0.967
6.900
0
1
0
0
0
0
0.00002
0.693

Max
78.423
78.423
1
1
0.583
0.8
0.583
0.778
1
1
1
0.545
1
0.5
1
1
1
1
12.484
14.210
5.464
0.992
0.367
1
1
1
0.995
7.110
35.978
122.987
19
12.290
83.131
1
20.146
99.979
3.800

Table 2. Bivariate Correlation Coefficients

This table reports Spearman correlation coefficients among variables for the 11,784 firm-year observations from 1996 to 2004. See Appendix C for variable
definitions. * indicate statistical significance at 5% level or less.
No

Tobin Q

0.638*

Lg(TobinQ)
Dummy
KLD

-0.045*

-0.031*

(C1+C2)

0.053*

0.038*

-0.576*

Lg(C1+C2)

0.049*

0.029*

-0.488*

0.852*

C1

0.043*

0.028*

-0.533*

0.862*

0.725*

C2

0.050*

0.038*

-0.502*

0.920*

0.791*

0.594*

Lg(C1)

0.040*

0.021*

-0.457*

0.734*

0.868*

0.803*

0.544*

10

11

12

13

14

15

16

Lg(C2)

0.048*

0.031*

-0.425*

0.790*

0.922*

0.533*

0.841*

0.608*

10

Emp

0.037*

0.033*

-0.375*

0.407*

0.364*

0.383*

0.350*

0.335*

0.321*

11

(Su+Sr)

-0.033*

-0.037*

-0.558*

0.428*

0.399*

0.371*

0.392*

0.347*

0.367*

0.266*

12

Lg(Su+Sr)

-0.028*

-0.038*

-0.427*

0.394*

0.454*

0.339*

0.363*

0.394*

0.418*

0.246*

0.778*

13

Su

-0.025*

-0.036*

-0.357*

0.362*

0.305*

0.322*

0.325*

0.270*

0.277*

0.188*

0.821*

0.647*

14

Sr

-0.027*

-0.021*

-0.524*

0.295*

0.314*

0.246*

0.278*

0.268*

0.293*

0.228*

0.719*

0.550*

0.195*

15

Lg(Su)

-0.017

-0.025*

-0.305*

0.331*

0.381*

0.288*

0.303*

0.337*

0.346*

0.176*

0.648*

0.861*

0.736*

0.219*

16

Lg(Sr)

-0.029*

-0.036*

-0.381*

0.280*

0.323*

0.236*

0.262*

0.272*

0.305*

0.219*

0.562*

0.682*

0.184*

0.741*

0.216*

17

Domini400

-0.027*

-0.026*

-0.199*

0.102*

0.077*

0.154*

0.044*

0.125*

0.026*

0.155*

-0.046*

-0.044*

-0.045*

-0.025*

-0.044*

-0.020*

18

KLD Exc

-0.019*

-0.028*

-0.207*

0.193*

0.164*

0.166*

0.178*

0.136*

0.156*

0.133*

0.278*

0.239*

0.256*

0.168*

0.217*

0.148*

19

Ln(Sale)

-0.090*

-0.107*

-0.448*

0.504*

0.467*

0.419*

0.476*

0.387*

0.445*

0.273*

0.483*

0.447*

0.429*

0.308*

0.391*

0.295*

20

Ln(Asset)

-0.176*

-0.157*

-0.439*

0.488*

0.445*

0.386*

0.476*

0.353*

0.435*

0.249*

0.463*

0.429*

0.392*

0.319*

0.357*

0.309*

21

Debtr

-0.170*

-0.167*

-0.016

0.045*

0.051*

0.039*

0.042*

0.044*

0.047*

0.006

0.091*

0.094*

0.102*

0.031*

0.097*

0.040*

22

Rndr

0.293*

0.311*

0.001

0.010

0.004

0.012

0.007

0.007

0.0003

0.032*

-0.003

-0.012

-0.043*

0.047*

-0.035*

0.027*

23

Advr

0.128*

0.108*

-0.071*

0.119*

0.115*

0.084*

0.124*

0.079*

0.121*

0.029*

0.012

0.013

0.017

0.001

0.018*

-0.002

24

Rndumy

-0.244*

-0.227*

0.048*

-0.097*

-0.091*

-0.118*

-0.062*

-0.109*

-0.061*

-0.096*

-0.059*

-0.055*

-0.033*

-0.063*

-0.036*

-0.054*

25

Advdumy

-0.118*

-0.116*

0.109*

-0.116*

-0.111*

-0.073*

-0.129*

-0.067*

-0.125*

-0.049*

-0.048*

-0.035*

-0.017

-0.061*

-0.018*

-0.041*

37

Correlation Coefficients (continued)

No

10

11

12

13

14

15

16

26

IndusHHI

0.029*

0.014

0.014

-0.009

-0.008

-0.013

-0.004

-0.015

-0.001

0.022*

0.064*

0.057*

0.063*

0.033*

0.052*

0.034*

27

Capxr

0.030*

0.057*

0.011

-0.026*

-0.031*

-0.004

-0.039*

-0.012

-0.041*

0.053*

0.029*

0.025*

0.032*

0.011

0.024*

0.014

28

Salegrw

0.185*

0.227*

0.039*

-0.057*

-0.070*

-0.048*

-0.054*

-0.059*

-0.066*

-0.021*

-0.042*

-0.058*

-0.056*

-0.005

-0.057*

-0.029*

29

Divr

0.021*

0.007

-0.038*

0.039*

0.039*

0.036*

0.034*

0.035*

0.034*

0.019*

0.038*

0.029*

0.038*

0.019*

0.027*

0.018

30

Stdret

0.156*

0.239*

0.087*

-0.144*

-0.127*

-0.141*

-0.119*

-0.124*

-0.106*

-0.070*

-0.072*

-0.082*

-0.165*

0.077*

-0.143*

0.049*

31

Gindex

-0.121*

-0.124*

-0.164*

0.089*

0.093*

0.114*

0.053*

0.117*

0.058*

0.039*

0.084*

0.081*

0.077*

0.051*

0.071*

0.054*

32

Pctdirshr

0.034*

0.042*

0.095*

-0.072*

-0.051*

-0.065*

-0.065*

-0.039*

-0.052*

-0.059*

-0.098*

-0.081*

-0.075*

-0.077*

-0.063*

-0.064*

33

Pctceown

0.053*

0.064*

0.102*

-0.079*

-0.066*

-0.075*

-0.069*

-0.064*

-0.056*

-0.063*

-0.098*

-0.089*

-0.076*

-0.076*

-0.071*

-0.068*

34

Pctindep

-0.069*

-0.061*

-0.203*

0.165*

0.166*

0.159*

0.139*

0.158*

0.142*

0.095*

0.175*

0.1645*

0.108*

0.169*

0.115*

0.151*

35

Ln(Block)

-0.004

0.003

-0.039*

-0.049*

-0.029*

-0.039*

-0.047*

-0.019*

-0.032*

-0.025*

0.032*

0.034*

-0.014

0.0723*

-0.001

0.066*

36

Pctinsti

0.083*

0.044*

-0.185*

0.039*

0.048*

0.049*

0.024*

0.059*

0.031*

0.006

0.116*

0.083*

0.024*

0.169*

0.031*

0.114*

37

Loganal

0.176*

0.167*

-0.352*

0.370*

0.321*

0.307*

0.350*

0.264*

0.307*

0.216*

0.314*

0.279*

0.272*

0.209*

0.240*

0.191*

38

Table 2 Bivariate Correlation Coefficients (continued)

No

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

17

Domini400

18

KLD Exc

-0.053*

19

Ln(Sale)

-0.075*

0.256*

20

Ln(Asset)

-0.121*

0.228*

0.834*

21

Debtr

-0.008

0.077*

0.170*

0.211*

22

Rndr

-0.058*

-0.022*

-0.283*

-0.209*

-0.193*

23

Advr

0.006

-0.008

0.046*

-0.031*

-0.022*

0.025*

24

Rndumy

0.020*

-0.089*

0.104*

0.183*

0.160*

-0.513*

-0.033*

25

Advdumy

0.0002

0.029*

-0.078*

0.061*

0.061*

-0.064*

-0.585*

0.069*

26

IndusHHI

-0.030*

0.169*

-0.012

-0.061*

0.060*

-0.013

0.050*

-0.155*

-0.047*

27

Capxr

-0.003

0.015

-0.110*

-0.015

0.185*

0.033*

-0.042*

0.077*

0.067*

0.066*

28

Salegrw

-0.039*

-0.017

0.009

0.012

-0.006

0.056*

-0.013

0.015

0.022*

-0.014

0.082*

29

Divr

-0.001

0.063*

0.039*

0.038*

0.039*

-0.026*

0.038*

0.005

-0.022*

0.052*

-0.010

-0.015

30

Stdret

-0.095*

-0.102*

-0.347*

-0.354*

-0.160*

0.443*

0.041*

-0.232*

-0.139*

0.006

0.038*

0.072*

-0.059*

31

Gindex

0.037*

0.040*

0.187*

0.167*

0.079*

-0.127*

-0.058*

-0.001

0.059*

-0.038*

-0.045*

-0.078*

0.033*

-0.216*

32

Pctdirshr

0.022*

-0.056*

-0.104*

-0.116*

0.017

-0.036*

0.079*

0.058*

-0.048*

0.058*

-0.009

0.009

-0.01

0.0856*

-0.159*

33

Pctceown

0.018

-0.048*

-0.143*

-0.155*

-0.049*

-0.005

0.076*

0.052*

-0.052*

0.043*

-0.003

0.016

-0.016

0.116*

-0.156*

0.401*

34

Pctindep

-0.028*

0.085*

0.162*

0.184*

0.015

0.027*

-0.056*

-0.118*

0.034*

-0.061*

-0.039*

-0.080*

0.032*

-0.082*

0.270*

-0.237*

35

Ln(Block)

-0.035*

-0.008

-0.024*

-0.088*

0.014

0.049*

-0.017

-0.061*

-0.037*

0.032*

-0.001

0.003

-0.019*

0.153*

-0.007

-0.031*

36

Pctinsti

-0.073*

0.019*

0.109*

0.028*

-0.032*

0.035*

-0.021*

-0.093*

-0.065*

0.054*

0.002

0.045*

-0.018*

0.119*

0.055*

-0.155*

37

Loganal

-0.133*

0.145*

0.577*

0.594*

0.011

0.057*

0.041*

0.004

-0.048*

-0.083*

0.133*

0.111*

0.014

-0.076*

0.084*

-0.144*

Correlation Coefficients (continued)

No

33

34

35

36

33

Pctceown

34

Pctindep

-0.204*

35

Ln(Block)

0.025*

0.031*

36

Pctinsti

-0.091*

0.181*

0.508*

37

Loganal

-0.148*

0.122*

-0.012

0.175*

39

Table 3A OLS and Fixed Effects Regressions of CFP, CSP, and Social Pressures for
Unbalanced Panel Data
This table shows the results from the three stage estimation method in which the dependent variables are CFP measured
by Tobins Q, CSP measured by C = (C1+C2), and S = (Su+Sr) as social pressure. The unbalanced panel includes all firms
and Domini 400 Firms includes only firms in Domini 400 Index. T-statistics are adjusted for robust and clustered (by
firm) standard errors and reported in parentheses. F-F industry dummy is based on Fama and French (1997) industry
classification. See Appendix C for variable definitions. ***, ** and * statistically significant at the 1%, 5% and 10%
levels, respectively.

UNBALANCED
PANEL DATA
Lag(C)
C

Tobin Q

0.00202
(4.10)***
0.11861
(6.51)***

Lag(S)

Emp
Dummy KLD
Ln(Sale)
Ln(Asset)
Debtr
Rndr
Advr
Rndumy
Advdumy
IndusHHI
Capxr
Salegrw
Divr

-0.00145
(3.27)***

-0.33969
(2.77)***

KLD Exc
Ddomini400

S
0.06826
(5.23)***

0.56692
(5.12)***

Lag(TobinQ)

OLS
C

-0.31545
(6.02)***
0.16632
(1.76)*
-0.29823
(7.00)***
0.16652
(4.26)***
-0.51175
(12.13)***
-0.45744
(2.32)**
2.00357
(4.81)***
3.72050
(2.11)**
-0.35478
(6.33)***
0.10332
(2.23)**
0.46119
(1.74)*
0.39880
(2.69)***
0.84448
(5.02)***
0.11732
(3.34)***

0.03447
(6.27)***

0.09723
(7.20)***
-0.04004
(10.08)***

-0.11499
(38.83)***
0.02403
(11.56)***
0.01284
(6.40)***

-0.09218
(31.88)***
0.00801
(4.50)***
0.02057
(12.37)***

0.04548
(3.36)***
0.19550
(3.44)***
-0.02916
(8.98)***
-0.00481
(1.52)
-0.01911
(0.69)
0.03872
(3.81)***
-0.02206
(7.91)***

0.05412
(4.30)***
-0.09547
(2.44)**
0.00376
(1.19)
-0.00090
(0.33)
0.06533
(2.64)***
-0.05304
(4.85)***
-0.01559
(4.09)***

40

FIXED EFFECTS
Tobin Q
C
S
0.07356
(7.88)***
0.35581
(1.75)*
-0.00011
-0.00103
(0.23)
(1.80)*
0.06728
(8.22)***
-0.19179
(1.08)
0.03413
(3.56)***
-0.27885
0.02828
-0.01319
(3.21)***
(6.40)***
(2.60)***
0.13263
(0.91)
-0.01496
-0.07265
-0.08942
(0.30)
(32.17)*** (34.38)***
0.29230
-0.00493
0.01422
(4.03)***
(1.34)
(3.37)***
-1.10048
0.02116
-0.00102
(15.51)*** (5.94)***
(0.25)
-1.13959
(7.03)***
-1.95107
-0.01137
0.06136
(4.42)***
(0.50)
(2.37)**
6.58798
0.00377
0.06210
(3.97)***
(0.04)
(0.64)
-0.05228
0.00906
0.01266
(0.37)
(1.25)
(1.52)
0.34898
0.00610
-0.00157
(4.69)***
(1.60)
(0.36)
0.40439
-0.03387
0.09332
(1.12)
(1.84)*
(4.44)***
0.42186
0.00095
-0.02471
(1.59)
(0.07)
(1.58)
0.53537
-0.00356
-0.00623
(10.67)***
(1.37)
(2.11)**
0.04087
(1.27)

Stdret
Gindex
Pctdirshr
Pctceown
Pctindep
Ln(Block)
Pctinsti
Loganal
Intercept
F-F industry
dummy
Year dummy
Observations
Adjusted R2
Number of Firms

-0.02499
(4.85)***
-0.03590
(6.35)***
0.10385
(1.30)
0.00945
(1.88)*
-0.33570
(2.67)***
-0.01921
(4.82)***
0.00443
(3.71)***
0.79069
(19.45)***
3.77775
(9.45)***

-0.00216
(5.06)***
-0.00126
(0.14)
0.00060
(2.79)***
0.03650
(6.09)***
-0.00066
(2.20)**
-0.00027
(3.78)***
-0.00240
(1.17)
-0.10498
(4.01)***

Yes
Yes
11,784
0.3021
2,480

Yes
Yes
11,784
0.4791
2,480

0.00132
(7.38)***
-0.00219
(5.83)***
-0.00070
(0.16)
0.00018
(1.47)
0.00353
(0.71)

-0.14561
(5.50)***

-0.02909
(5.26)***
-0.02143
(1.28)
0.01498
(0.17)
0.00070
(0.19)
-0.19148
(1.40)
-0.01509
(4.45)***
0.01390
(7.99)***
0.09048
(1.78)*
7.31873
(16.59)***

0.00195
(2.28)**
-0.00990
(2.26)**
0.00028
(1.47)
-0.00713
(1.02)
0.00052
(3.02)***
-0.00057
(6.40)***
0.00027
(0.10)
-0.00086
(0.04)

Yes
Yes
11,784
0.4769
2,480

Dropped
Yes
11,784
0.1131
2,480

Dropped
Yes
11,784
0.2145
2,480

41

0.00110
(3.39)***
0.00123
(1.25)
0.00327
(0.65)
0.00028
(1.27)
0.00413
(0.52)

-0.03537
(1.39)
Dropped
Yes
11,784
0.3409
2,480

Table 3B Two-Stage Least Square and Three-Stage Least Square Regressions of

CFP, CSP, and Social Pressures for Unbalanced Panel
This table shows the results from the three stage estimation method in which the dependent variables are CFP measured
by Tobins Q, CSP measured by C = (C1+C2), and S = (Su+Sr) as social pressure. The unbalanced sample includes all
firms and Domini 400 Firms includes only firms in Domini 400 Index. T-statistics are adjusted for robust and clustered
(by firm) standard errors and reported in parentheses. F-F industry dummy is based on Fama and French (1997)
industry classification. See Appendix C for variable definitions. ***, ** and * statistically significant at the 1%, 5%
and 10% levels, respectively.

UNBALANCED
PANEL DATA
Lag(C)
C

Tobin Q

Dummy KLD
Ln(Sale)
Ln(Asset)
Debtr
Rndr
Advr
Rndumy
Advdumy
IndusHHI
Capxr
Salegrw
Divr

3SLS
C

S
0.11332
(14.16)***

0.00336
(6.36)***
0.16827
(16.05)***

-0.00192
(4.24)***

-0.67865
(3.11)***

KLD Exc

Emp

-0.00145
(3.19)***

-0.33791
(1.74)*

-0.32308
(4.68)***
0.10775
(1.00)
-0.25949
(5.26)***
0.15832
(5.04)***
-0.51716
(17.20)***
-0.45812
(5.39)***
1.98793
(7.88)***
3.64685
(5.15)***
-0.34417
(6.93)***
0.10407
(2.40)**
0.46470
(1.23)
0.38956
(2.14)**
0.85230
(17.49)***
0.11582

Tobin Q

1.55338
(9.25)***
0.00202
(3.80)***
0.11861
(11.25)***

Lag(S)

Ddomini400

S
0.06826
(8.48)***

0.93134
(5.52)***

Lag(TobinQ)

2SLS
C

0.03447
(7.01)***

0.09723
(12.34)***
-0.04004
(9.30)***

-0.11499
(39.67)***
0.02403
(10.72)***
0.01284
(6.03)***

-0.09218
(35.88)***
0.00801
(4.11)***
0.02057
(11.15)***

0.04548
(2.55)**
0.19550
(3.81)***
-0.02916
(8.15)***
-0.00481
(1.53)
-0.01911
(0.70)
0.03872
(2.96)***
-0.02206
(6.14)***

0.05412
(3.40)***
-0.09547
(2.12)**
0.00376
(1.19)
-0.00090
(0.33)
0.06533
(2.72)***
-0.05304
(4.66)***
-0.01559
(4.95)***

42

-0.35702
(5.19)***
0.10706
(1.02)
-0.21820
(4.44)***
0.14637
(4.68)***
-0.51860
(17.31)***
-0.45615
(5.39)***
1.97233
(7.84)***
3.49520
(4.95)***
-0.32519
(6.57)***
0.10700
(2.48)**
0.49557
(1.32)
0.34950
(1.93)*
0.85950
(17.69)***
0.11607

0.03555
(7.25)***

0.09755
(12.46)***
-0.04183
(9.74)***

-0.11209
(38.80)***
0.02362
(10.57)***
0.01228
(5.79)***

-0.08901
(34.76)***
0.00699
(3.60)***
0.01982
(10.78)***

0.03969
(2.23)**
0.19275
(3.77)***
-0.02863
(8.03)***
-0.00489
(1.57)
-0.02278
(0.84)
0.03946
(3.03)***
-0.02269
(6.34)***

0.05324
(3.36)***
-0.10372
(2.31)**
0.00466
(1.49)
-0.00085
(0.31)
0.06436
(2.69)***
-0.05364
(4.73)***
-0.01357
(4.32)***

Stdret
Gindex
Pctdirshr
Pctceown
Pctindep
Ln(Block)
Pctinsti
Loganal
Intercept
F-F industry
dummy
Year dummy
Observations
Adjusted R2
Number of Firms

(2.91)***
-0.02502
(7.89)***
-0.03517
(6.14)***
0.10383
(1.33)
0.00922
(3.03)***
-0.34847
(3.91)***
-0.01903
(6.13)***
0.00452
(4.52)***
0.79111
(25.44)***
3.82461
(7.92)***

-0.00216
(5.26)***
-0.00126
(0.22)
0.00060
(2.74)***
0.03650
(5.67)***
-0.00066
(2.94)***
-0.00027
(3.71)***
-0.00240
(1.06)
-0.10498
(3.02)***

Yes
Yes
11,784
0.3016
2,480

Yes
Yes
11,784
0.4791
2,480

-0.14561
(4.76)***

(2.93)***
-0.02469
(7.82)***
-0.03447
(6.04)***
0.10405
(1.33)
0.00890
(2.94)***
-0.36976
(4.17)***
-0.01860
(6.02)***
0.00468
(4.69)***
0.79040
(25.51)***
3.84467
(7.99)***

-0.00203
(4.96)***
-0.00125
(0.22)
0.00058
(2.66)***
0.03624
(5.65)***
-0.00064
(2.88)***
-0.00026
(3.64)***
-0.00411
(1.83)*
-0.09726
(2.81)***

Yes
Yes
11,784
0.4769
2,480

Yes
Yes
11,784
0.2983
2,480

Yes
Yes
11,784
0.4778
2,480

0.00132
(6.69)***
-0.00219
(6.06)***
-0.00070
(0.14)
0.00018
(0.93)
0.00353
(0.63)

43

0.00133
(6.78)***
-0.00215
(5.96)***
-0.00095
(0.19)
0.00016
(0.85)
0.00211
(0.38)

-0.13173
(4.32)***
Yes
Yes
11,784
0.4755
2,480

Table 4A OLS and Fixed Effects Regressions of CFP, CSP, and Social Pressures for
Balanced Panel Data
This table shows the results from the three stage estimation method in which the dependent variables are CFP measured
by Tobins Q, CSP measured by C = (C1+C2), and S = (Su+Sr) as social pressure. The balanced panel includes all firms
and Domini 400 Firms includes only firms in Domini 400 Index. T-statistics are adjusted for robust and clustered (by
firm) standard errors and reported in parentheses. F-F industry dummy is based on Fama and French (1997) industry
classification. See Appendix C for variable definitions. ***, ** and * statistically significant at the 1%, 5% and 10%
levels, respectively.

BALANCED
PANEL DATA
Lag(C)
C

Tobin Q

0.00644
(3.37)***
0.06383
(2.74)***

Lag(S)

Emp
Dummy KLD
Ln(Sale)
Ln(Asset)
Debtr
Rndr
Advr
Rndumy
Advdumy
IndusHHI
Capxr
Salegrw
Divr

-0.00618
(4.68)***

-0.27571
(1.83)*

KLD Exc
Ddomini400

S
0.04440
(2.47)***

0.64483
(4.58)***

Lag(TobinQ)

OLS
C

-0.22508
(3.40)***
0.25214
(2.17)**
-0.23335
(4.72)***
0.10590
(2.44)**
-0.46830
(12.07)***
-1.55519
(9.54)***
4.65631
(4.83)***
6.18306
(4.58)***
-0.16484
(2.56)**
0.24675
(3.71)***
0.07358
(0.25)
0.64985
(2.47)**
0.70714
(5.46)***
0.07613

0.05288
(6.52)***

0.08577
(4.81)***
-0.00989
(1.64)

-0.11206
(24.09)***
0.03623
(7.73)***
0.02522
(5.34)***

-0.08138
(17.60)***
0.01253
(2.88)***
0.03427
(8.76)***

0.10597
(2.06)**
0.46999
(3.40)***
-0.03265
(4.70)***
-0.01765
(2.56)**
-0.09133
(2.19)**
0.05873
(1.85)*
-0.03420
(4.42)***

0.09975
(1.67)*
-0.21727
(2.57)**
0.03173
(4.67)***
0.00440
(0.79)
0.10269
(2.81)***
-0.02801
(0.90)
-0.01877
(1.59)

44

FIXED EFFECTS
Tobin Q
C
S
0.06023
(3.86)***
0.80592
(2.47)***
0.00232
-0.00426
(1.69)*
(2.48)**
0.05740
(4.51)***
-0.24306
(1.65)*
0.02940
(2.03)**
-0.26879
0.03074
-0.00627
(3.73)***
(4.60)***
(0.80)
0.17028
(1.71)*
-0.16414
-0.08286
-0.08607
(3.37)*** (19.96)*** (17.55)***
0.53189
0.00050
0.03331
(6.43)***
(0.07)
(3.70)***
-1.03230
0.02309
-0.00435
(13.34)*** (3.26)***
(0.53)
-1.18326
(6.41)***
-2.98196
-0.01760
0.17627
(4.06)***
(0.26)
(2.20)**
6.82364
0.13818
0.16155
(3.52)***
(0.76)
(0.76)
0.16718
0.01108
0.03882
(1.22)
(0.86)
(2.58)***
0.14386
0.00783
-0.00689
(1.97)**
(1.15)
(0.86)
-0.26643
-0.06028
0.12096
(0.85)
(2.07)**
(3.55)***
0.34987
-0.01568
-0.02627
(0.99)
(0.48)
(0.68)
0.22091
-0.00894
-0.01438
(3.45)***
(1.49)
(2.05)**
0.04364

Stdret
Gindex
Pctdirshr
Pctceown
Pctindep
Ln(Block)
Pctinsti
Loganal
Intercept
F-F industry
dummy
Year dummy
Observations
Adjusted R2
Number of Firms

(2.74)***
-0.05122
(6.23)***
-0.03947
(5.05)***
0.27662
(1.17)
0.00113
(0.15)
-0.58617
(4.21)***
-0.01436
(3.22)***
0.00401
(2.60)***
0.77119
(17.44)***
3.52101
(11.62)***

-0.00054
(0.66)
-0.01561
(0.91)
0.00321
(4.33)***
0.05039
(3.77)***
-0.00092
(1.98)**
-0.00027
(1.64)
-0.01966
(4.33)***
-0.33110
(8.96)***

Yes
Yes
4,356
0.4676
484

Yes
Yes
4,356
0.5031
484

-0.35027
(10.87)***

(2.05)**
-0.02573
(3.22)***
-0.01271
(0.72)
-0.15589
(1.05)
0.00162
(0.27)
-0.21483
(1.49)
-0.00331
(1.08)
0.01335
(6.04)***
0.26040
(4.60)***
4.94263
(9.27)***

0.00590
(3.58)***
-0.00945
(0.68)
0.00105
(1.90)*
-0.02063
(1.53)
0.00064
(2.26)**
-0.00106
(5.17)***
0.00615
(1.16)
-0.04781
(0.99)

-0.18543
(3.26)***

Yes
Yes
4,356
0.4883
484

Dropped
Yes
4,356
0.1527
484

Dropped
Yes
4,356
0.2151
484

Dropped
Yes
4,356
0.3057
484

0.00209
(3.25)***
-0.00324
(4.06)***
0.00842
(0.56)
-0.00094
(1.91)*
-0.00323
(0.28)

45

0.00326
(3.73)***
0.00323
(1.66)*
0.01668
(1.03)
0.00057
(0.88)
-0.01728
(1.10)

Table 4B Two-Stage Least Square and Three-Stage Least Square Regressions of

CFP, CSP, and Social Pressures for Balanced Panel Data
This table shows the results from the three stage estimation method in which the dependent variables are CFP measured
by Tobins Q, CSP measured by C = (C1+C2), and S = (Su+Sr) as social pressure. The Full Sample includes all firms and
Domini 400 Firms includes only firms in Domini 400 Index. T-statistics are adjusted for robust and clustered (by firm)
standard errors and reported in parentheses. F-F industry dummy is based on Fama and French (1997) industry
classification. See Appendix C for variable definitions. ***, ** and * statistically significant at the 1%, 5% and 10%
levels, respectively.

BALANCED
PANEL DATA
Lag(C)
C

Tobin Q

Dummy KLD
Ln(Sale)
Ln(Asset)
Debtr
Rndr
Advr
Rndumy
Advdumy
IndusHHI
Capxr
Salegrw
Divr

3SLS
C

S
0.07293
(5.40)***

0.01210
(7.22)***
0.09425
(5.67)***

-0.01108
(7.58)***

-1.22598
(5.61)***

KLD Exc

Emp

-0.00618
(4.18)***

-0.52863
(2.39)**

-0.23839
(3.21)***
0.22485
(2.12)**
-0.23231
(4.23)***
0.10283
(2.28)**
-0.46379
(10.72)***
-1.55514
(10.91)***
4.65019
(8.55)***
6.00515
(6.75)***
-0.14903
(2.31)**
0.25043
(4.50)***
0.11836
(0.27)
0.63407
(2.03)**
0.70753
(8.75)***
0.07556

Tobin Q

1.48904
(8.77)***
0.00644
(3.80)***
0.06383
(3.80)***

Lag(S)

Ddomini400

S
0.04440
(3.26)***

0.87312
(5.08)***

Lag(TobinQ)

2SLS
C

0.05288
(6.28)***

0.08577
(6.63)***
-0.00989
(1.31)

-0.11206
(20.68)***
0.03623
(7.10)***
0.02522
(5.21)***

-0.08138
(16.51)***
0.01253
(2.74)***
0.03427
(8.02)***

0.10597
(1.77)*
0.46999
(4.60)***
-0.03265
(4.45)***
-0.01765
(2.77)***
-0.09133
(1.84)*
0.05873
(1.65)*
-0.03420
(3.65)***

0.09975
(1.78)*
-0.21727
(2.35)**
0.03173
(4.79)***
0.00440
(0.76)
0.10269
(2.30)**
-0.02801
(0.88)
-0.01877
(2.22)**

46

-0.27739
(3.77)***
0.21987
(2.10)**
-0.22174
(4.08)***
0.09097
(2.03)**
-0.45180
(10.54)***
-1.53555
(10.93)***
4.59736
(8.53)***
5.52483
(6.27)***
-0.10879
(1.70)*
0.26230
(4.76)***
0.24632
(0.57)
0.57179
(1.85)*
0.70752
(8.83)***
0.07534

0.05324
(6.38)***

0.08333
(6.51)***
-0.01316
(1.76)*

-0.10892
(20.28)***
0.03534
(6.98)***
0.02656
(5.54)***

-0.08146
(16.68)***
0.01279
(2.82)***
0.03195
(7.54)***

0.07593
(1.28)
0.43571
(4.30)***
-0.03240
(4.46)***
-0.01922
(3.04)***
-0.09441
(1.92)*
0.05465
(1.55)
-0.03812
(4.11)***

0.13756
(2.47)**
-0.19162
(2.09)**
0.03163
(4.82)***
0.00573
(1.01)
0.10148
(2.29)**
-0.02371
(0.75)
-0.01320
(1.58)

Stdret
Gindex
Pctdirshr
Pctceown
Pctindep
Ln(Block)
Pctinsti
Loganal
Intercept
F-F industry
dummy
Year dummy
Observations
Adjusted R2
Number of Firms

(2.50)**
-0.05038
(7.62)***
-0.04005
(5.48)***
0.27816
(1.59)
0.00019
(0.03)
-0.59686
(4.88)***
-0.01395
(4.14)***
0.00394
(2.65)***
0.76959
(17.23)***
3.49875
(7.34)***

-0.00054
(0.65)
-0.01561
(0.78)
0.00321
(4.44)***
0.05039
(3.61)***
-0.00092
(2.38)**
-0.00027
(1.61)
-0.01966
(3.78)***
-0.33110
(6.20)***

Yes
Yes
4,356
0.4669
484

Yes
Yes
4,356
0.5031
484

-0.26517
(4.35)***

(2.53)**
-0.04823
(7.38)***
-0.04152
(5.73)***
0.29190
(1.68)*
-0.00233
(0.37)
-0.62669
(5.17)***
-0.01328
(3.99)***
0.00409
(2.78)***
0.77129
(17.47)***
3.42214
(7.25)***

-0.00021
(0.25)
-0.01632
(0.82)
0.00322
(4.50)***
0.05274
(3.82)***
-0.00083
(2.17)**
-0.00026
(1.55)
-0.02427
(4.71)***
-0.33200
(6.28)***

-0.21361
(3.54)***

Yes
Yes
4,356
0.4883
484

Yes
Yes
4,356
0.4577
484

Yes
Yes
4,356
0.5015
484

Yes
Yes
4,356
0.4866
484

0.00209
(3.12)***
-0.00324
(4.32)***
0.00842
(0.47)
-0.00094
(1.44)
-0.00323
(0.26)

47

0.00185
(2.80)***
-0.00343
(4.61)***
0.00861
(0.49)
-0.00098
(1.52)
-0.00765
(0.62)

Table 5. The percentage change in a dependent variable from a one standard

deviation increase in an independent variable
This Table reports the percentage change in a dependent variable (evaluated at the mean)
from a one standard deviation increase in a right-side variable.
* coefficient is not significant.
Tobin Q Tobin Q
C
C
S
S
equation

equation

equation

equation

equation

equation

lag Q

lag S

lag Q

lag C

OLS

5.182

-2.717

5.619

17.425

-4.735

13.949

OLS-FFE

3.252

9.884

-3.363

15.032

2SLS

8.513

-2.702

5.619

17.425

-4.735

13.949

3SLS

14.198

-5.428

9.345

24.721

-6.269

19.544

7.713

7.480

-9.288

7.487

Right-side
variable
Unbalanced panel

Balanced panel
OLS

6.984

-2.654

OLS-FFE

8.728

-2.340

2.779

6.727

-6.403

10.156

2SLS

9.456

-5.089

7.713

7.480

-9.288

7.487

3SLS

16.127

-11.803

14.492

11.045

-16.653

12.230

48

Table 6A. Two-Stage Least Square (2SLS) Regressions of CFP, CSP, and Social
Pressures with 5-equations system for unbalanced panel data
This table shows the results from the three stage estimation method in which the dependent variables are CFP measured
by Tobins Q, CSP measured by C1 and C2, and social pressures from public (Su) and private (Sr) politics. The data is
from the Kinder, Lydenberg, and Dominis (KLD) Socrates database. T-statistics are adjusted for robust and clustered (by
firm) standard errors and reported in parentheses. F-F industry dummy is based on Fama and French (1997) industry
classification. See Appendix C for variable definitions. ***, ** and * statistically significant at the 1%, 5% and 10%
levels, respectively.

UNBALANCED
PANEL DATA
Lg(C1)

Tobin Q

C1

2SLS
C2

Lg(C2)
C1
C2

0.00048
(1.71)*
0.05012
(7.21)***
0.03451
(3.32)***

Lg(Su)
Lg(Sr)

Sr

0.00153
(4.28)***
0.07865
(8.83)***
0.06285
(4.72)***

Emp
KLD Dummy
Ln(Sale)
Ln(Asset)
Debtr
Rndr
Advr
Rndumy
Advdumy

-0.00071
(1.95)*

-0.00075
(2.76)***

0.01507
(3.19)***
-0.00541
(2.09)**

-0.42207
(1.84)*
-0.22042
(1.86)*

KLD Exc
Domini400

Sr
-0.00304
(0.29)
0.04838
(5.81)***

-0.17500
(0.45)
1.65394
(5.91)***

Lg(TobinQ)

Su

Su
0.04457
(3.13)***
0.03906
(3.50)***

-0.30242
(4.35)***
0.12563
(1.16)
-0.27584
(5.50)***
0.16134
(5.14)***
-0.52182
(17.34)***
-0.45744
(5.38)***
1.95643
(7.74)***
3.67949
(5.19)***
-0.35145
(7.07)***
0.11115

0.02711
(10.46)***

0.00784
(2.36)**

0.08192
(12.94)***
-0.03417
(9.85)***

-0.05618
(36.86)***
0.01108
(9.41)***
0.00238
(2.13)**

-0.05871
(30.05)***
0.01291
(8.55)***
0.01051
(7.33)***

-0.04624
(22.33)***
0.00634
(4.04)***
0.01286
(8.67)***

-0.04626
(29.94)***
0.00170
(1.45)
0.00761
(6.88)***

0.00363
(0.39)
0.09509
(3.53)***
-0.01564
(8.32)***
0.00177

0.04252
(3.54)***
0.10036
(2.91)***
-0.01355
(5.63)***
-0.00660

0.00692
(0.54)
-0.05443
(1.50)
0.00233
(0.92)
-0.00012

0.04670
(4.89)***
-0.04051
(1.50)
0.00129
(0.68)
-0.00060

49

IndusHHI
Capxr
Salegrw
Divr
Stdret
Gindex
Pctdirshr
Pctceown
Pctindep
Ln(Block)
Pctinsti
Loganal
Intercept
F-F Industry dummy
Year dummy
Adjusted R2
Observations
Number of firms

(2.56)**
0.44645
(1.18)
0.40439
(2.22)**
0.85184
(17.42)***
0.11685
(2.94)***
-0.02519
(7.93)***
-0.03381
(5.89)***
0.11005
(1.41)
0.00930
(3.06)***
-0.34229
(3.84)***
-0.01931
(6.21)***
0.00464
(4.63)***
0.79076
(25.40)***
3.86688
(8.01)***
Yes
Yes
0.3018
11,784
2,480

(1.08)
-0.01280
(0.89)
0.02379
(3.46)***
-0.00763
(4.04)***

-0.00018
(0.83)
0.00261
(0.88)
0.00027
(2.32)**
0.01714
(5.07)***
-0.00038
(3.22)***
-0.00006
(1.54)
-0.00049
(0.41)
-0.01870
(1.02)
Yes
Yes
0.3928
11,784
2,480

50

(3.12)***
-0.00456
(0.25)
0.01480
(1.68)*
-0.01435
(5.93)***

-0.00199
(7.20)***
-0.00380
(1.00)
0.00033
(2.26)**
0.01915
(4.42)***
-0.00026
(1.73)*
-0.00021
(4.40)***
-0.00178
(1.17)
-0.08747
(3.73)***
Yes
Yes
0.4064
11,784
2,480

(0.05)
0.00479
(0.25)
-0.03008
(3.29)***
-0.01846
(7.29)***

(0.36)
0.06073
(4.22)***
-0.02266
(3.32)***
0.00294
(1.56)

0.00055
(3.47)***
-0.00088
(3.03)***
0.00141
(0.36)
0.00025
(1.64)
0.00984
(2.18)**

0.00077
(6.47)***
-0.00128
(5.88)***
-0.00188
(0.64)
-0.00008
(0.68)
-0.00626
(1.86)*

-0.09024
(3.67)***
Yes
Yes
0.3259
11,784
2,480

-0.05490
(2.99)***
Yes
Yes
0.4436
11,784
2,480

Table 6B. Two-Stage Least Square (2SLS) Regressions of CFP, CSP, and Social
Pressures with 5-equations system for balanced panel data
This table shows the results from the three stage estimation method in which the dependent variables are CFP measured
by Tobins Q, CSP measured by C1 and C2, and social pressures from public (Su) and private (Sr) politics. The data is
from the Kinder, Lydenberg, and Dominis (KLD) Socrates database. T-statistics are adjusted for robust and clustered (by
firm) standard errors and reported in parentheses. F-F industry dummy is based on Fama and French (1997) industry
classification. See Appendix C for variable definitions. ***, ** and * statistically significant at the 1%, 5% and 10%
levels, respectively.

BALANCED
PANEL DATA
Lg(C1)

Tobin Q

C1

2SLS
C2

Lg(C2)
C1
C2

0.00044
(0.47)
0.01786
(1.57)
0.01347
(0.76)

Lg(Su)
Lg(Sr)

Sr

0.00598
(5.33)***
0.05151
(3.78)***
0.03765
(1.77)*

Emp
KLD Dummy
Ln(Sale)
Ln(Asset)
Debtr
Rndr
Advr
Rndumy
Advdumy

-0.00390
(3.25)***

-0.00240
(2.84)***

0.02598
(3.51)***
0.00971
(2.25)**

-0.39404
(1.36)
-0.89300
(2.23)**

KLD Exc
Domini400

Sr
-0.00326
(0.21)
0.04711
(3.60)***

0.89104
(2.45)**
2.14025
(7.49)***

Lg(TobinQ)

Su

Su
0.01021
(0.46)
0.02618
(1.81)*

-0.18655
(2.49)**
0.26731
(2.51)**
-0.26854
(4.85)***
0.10291
(2.27)**
-0.47087
(10.87)***
-1.56559
(10.98)***
4.43157
(8.13)***
6.13849
(6.88)***
-0.17566
(2.71)***
0.27852

0.03799
(8.11)***

0.01522
(2.72)***

0.05940
(5.65)***
-0.01899
(3.09)***

-0.05966
(19.88)***
0.01701
(6.01)***
0.00652
(2.44)**

-0.05235
(14.60)***
0.01912
(5.66)***
0.01874
(5.85)***

-0.04713
(11.72)***
0.01383
(3.71)***
0.01923
(5.53)***

-0.03497
(12.37)***
-0.00121
(0.46)
0.01471
(6.02)***

-0.01416
(0.43)
0.24032
(4.24)***
-0.02126
(5.24)***
0.00184

0.12094
(3.05)***
0.23133
(3.42)***
-0.01148
(2.37)**
-0.01946

0.04579
(1.00)
-0.24864
(3.31)***
0.02860
(5.31)***
0.00167

0.05129
(1.60)
0.03363
(0.64)
0.00270
(0.71)
0.00332

51

IndusHHI
Capxr
Salegrw
Divr
Stdret
Gindex
Pctdirshr
Pctceown
Pctindep
Ln(Block)
Pctinsti
Loganal
Intercept
F-F Industry dummy
Year dummy
Adjusted R2
Observations
Number of firms

(4.99)***
0.10022
(0.23)
0.61551
(1.97)**
0.70564
(8.68)***
0.07693
(2.55)**
-0.05012
(7.52)***
-0.03773
(5.15)***
0.30213
(1.73)*
-0.00114
(0.18)
-0.59225
(4.83)***
-0.01453
(4.31)***
0.00462
(3.09)***
0.77002
(17.16)***
3.52302
(7.39)***
Yes
Yes
0.4671
4,356
484

(0.52)
-0.05576
(2.03)**
0.02124
(1.08)
-0.01551
(2.99)***

0.00047
(1.02)
0.00163
(0.15)
0.00096
(2.40)**
0.02381
(3.08)***
-0.00065
(3.04)***
0.00009
(1.01)
-0.00738
(2.56)**
-0.10866
(2.95)***
Yes
Yes
0.3958
4,356
484

52

(4.61)***
-0.03423
(1.04)
0.03749
(1.59)
-0.01865
(3.01)***

-0.00102
(1.85)*
-0.01714
(1.29)
0.00225
(4.70)***
0.02627
(2.84)***
-0.00026
(1.02)
-0.00037
(3.30)***
-0.01211
(3.51)***
-0.21259
(4.83)***
Yes
Yes
0.4521
4,356
484

(0.36)
0.00812
(0.22)
-0.01994
(0.77)
-0.02773
(4.04)***

(1.01)
0.09546
(3.74)***
-0.00758
(0.42)
0.00908
(1.88)*

-0.00039
(0.72)
-0.00170
(2.79)***
0.00740
(0.51)
-0.00023
(0.43)
0.01214
(1.19)

0.00246
(6.43)***
-0.00148
(3.45)***
0.00104
(0.10)
-0.00074
(1.99)**
-0.01545
(2.16)**

-0.08367
(1.69)*
Yes
Yes
0.3848
4,356
484

-0.18027
(5.18)***
Yes
Yes
0.4645
4,356
484