literature review

Global Finance Journal 38 (2018) 45–64

Contents lists available at ScienceDirect

Global Finance Journal

journal homepage: www.elsevier.com/locate/gfj

ESG performance and firm value: The moderating role of
disclosure

Ali Fatemi a,⁎, Martin Glaum b, Stefanie Kaiser c
a Greenleaf Advisors LLC and DePaul University, 1 East Jackson Boulevard, Suite 5500, Chicago, IL 60614, USA
b WHU – Otto Beisheim School of Management, Burgplatz 2, 56179 Vallendar, Germany
c Independent. Part of this research was carried out while Stefanie Kaiser was affiliated to WHU – Otto Beisheim School of Management

a r t i c l e i n f o

⁎ Corresponding author.
E-mail addresses: [email protected] (A. Fatemi), m

1 In this paper, we use the term ESG interchangeably
corporate practice.

http://dx.doi.org/10.1016/j.gfj.2017.03.001
1044-0283/© 2017 Elsevier Inc. All rights reserved.

a b s t r a c t

Article history:
Received 26 May 2016
Received in revised form 24 February 2017
Accepted 4 March 2017
Available online 9 March 2017

This study investigates the effect of environmental, social, and governance (ESG) activities and
their disclosure on firm value. We find that ESG strengths increase firm value and that weak-
nesses decrease it. ESG disclosure, per se, decreases valuation. But more importantly, we find
that disclosure plays a crucial moderating role by mitigating the negative effect of weaknesses
and attenuating the positive effect of strengths.

© 2017 Elsevier Inc. All rights reserved.

JEL classification:
G30
G32
M41
Q51
Q56

Keywords:
ESG
CSR
Disclosure
Firm value

1. Introduction

In this paper, we investigate the interrelationship between a firm’s strengths and weaknesses with regard to environmental, social,
and governance (ESG) factors, its ESG-related disclosure, and its valuation.1 In recent years, numerous studies have attempted to measure
the performance and valuation impact of ESG factors. A stream of this literature has also addressed the determinants of the firm’s ESG
disclosure and the possible valuation effects of such disclosure. However, the question of how the value of the firm may be jointly affected
by ESG activities and the intensity of ESG-related disclosure remains largely unexplored. We hypothesize that the association between a
firm’s ESG activities and its valuation is moderated by its disclosures related to those activities. However, the impact of disclosure in this
context is not ex ante clear. One might expect a positive effect insofar as disclosure reduces information asymmetries and helps investors
better understand the firm’s ESG strengths or weaknesses. Alternatively, ESG disclosure may impair firm value if investors view such dis-
closure as “cheap talk” or “greenwashing”. Our findings indicate that disclosure effects differ for ESG strengths and for ESG concerns. More
precisely, while firms with ESG concerns benefit from ESG-related disclosure, firms with ESG strengths experience lower valuation effects
when they intensify their disclosure efforts.

[email protected] (M. Glaum), [email protected] (S. Kaiser).
with CSR (corporate social responsibility). Both terms are widely used in both the academic literature and in

46 A. Fatemi et al. / Global Finance Journal 38 (2018) 45–64

The question of how ESG factors may affect a firm’s financial performance and its value has been extensively investigated.
However, the resulting conclusions have been far from unanimous (see Clark & Viehs, 2014; Margolis, Elfenbein, & Walsh,
2009 for recent surveys of the literature). In early contributions, it was mostly taken as a given that environmental investments
or social responsibility activities that exceeded the legally binding minimum standards would entail additional costs and would
thus reduce firm value (e.g., Friedman, 1970). In fact, as Kim and Lyon (2015, p. 706) note, “the entire environmental regulatory
paradigm is built around the idea that firms must be forced to make environmental improvements, because they would otherwise
find them costly or unprofitable, and thus not undertake them on their own.” More recent contributions to the theory of the firm
regard ESG activities as having the potential to increase firm value (e.g., Fatemi, Fooladi, & Tehranian, 2015; Malik, 2015; Porter,
1991; Porter & Kramer, 2011; Porter & van der Linde, 1995; Roberts, 2004). For example, under the resource-based view of the
firm, environmentally or socially motivated activities can improve the management team’s capabilities and the firm’s potential
to attract qualified employees. Moreover, such activities can enhance the firm’s reputation and strengthen its interactions with
its stakeholders (Branco & Rodrigues, 2006). Jensen (2002) has synthesized the competing propositions and proposed that “en-
lightened [long-term] value maximization” requires the firm to take into account the interests of all of its important constituent
groups (see also Fatemi & Fooladi, 2013; Serafeim, 2014). The empirical literature dealing with ESG’s effects on financial perfor-
mance and on firm value does not produce unequivocal results either. However, while some studies have reported a negative as-
sociation or insignificant results (see Horvathova, 2010), a meta-analysis by Margolis et al. (2009) finds the overall effect to be
positive, though small and possibly decreasing over time (see also Orlitzky, Schmidt, & Rynes, 2003).

The academic debate notwithstanding, over the course of the last two decades many firms, especially large multinational ones,
have intensified their efforts to report on ESG matters in order to legitimate their behavior and improve their reputation. Accord-
ing to a study by KPMG (2011), in 1996 only 300 firms worldwide produced corporate social responsibility (CSR) reports. By
2014, this number had increased to N7000 worldwide (Khan, Serafeim, & Yoon, 2016). While many firms have adopted the Global
Reporting Initiative (GRI) guidelines and, more recently, the framework suggested by the International Integrated Reporting Coun-
cil (IIRC, 2013, 2014), the extent and quality of ESG disclosure remain heterogeneous (Ioannou & Serafeim, 2016).

To assess the impact of ESG disclosure on firm value, it is important to recognize that ESG reporting can reflect various motives
beyond the obvious wish to emphasize firm strengths and play down weaknesses. Disclosure may be used to explain changes in
ESG policies or to repair a reputation damaged by, for example, acute environmental harm (Brown & Deegan, 1998; Cho & Patten,
2007). It may be a mere façade. A firm might even understate its ESG activities for fear of alienating investors. As we explain in
detail below, extant empirical research has failed to document a consistent relationship between the extent of a firm’s ESG dis-
closure and its financial performance or valuation.

This paper focuses on the question of whether the association between a firm’s ESG activities and its valuation is moderated by
its ESG-related disclosure. Our empirical analysis is based on data for 1640 firm-year observations for publicly traded U.S. firms for
the years 2006 to 2011. We utilize data on ESG strengths and ESG concerns as compiled and reported by KLD Research and An-
alytics as proxies for a firm’s ESG performance, and we use Bloomberg’s ESG disclosure score (DISC) as an indicator of the extent
of a firm’s ESG disclosure. Taking into account the potential endogeneity of the firm’s disclosure strategy, we employ a two-stage
least squares (2SLS) model. The first stage of the model is composed of three equations that describe the determinants of a firm’s
disclosure of its ESG activities (DISC) and that describe the interactions of ESG disclosure with ESG strengths (STR*DISC) and ESG
concerns (CON*DISC). The second stage of the model is our main focus; it describes the relationship between firm value (TOBIN
Q), on the one hand, and ESG performance (STR, CON) and ESG disclosure (DISC), on the other hand. We also disaggregate the
relationship to examine the three dimensions of ESG scores: environmental, social, and governance factors.

Our main contribution to the literature is that we provide a link between two hitherto separate streams of research: studies
addressing the relationship between ESG performance and financial performance or firm value and those examining the determi-
nants and the value impact of ESG disclosure. We propose that the association between a firm’s ESG performance and value is
moderated by its ESG-related disclosure. Our empirical investigation provides evidence in support of the moderating role of dis-
closure. More precisely, our results show that ESG disclosure helps the firm alleviate the negative valuation effects of concerns
regarding its ESG performance. Furthermore, we find that for firms with ESG strengths, disclosure is negatively related to firm
value. Finally, we find that when evaluating the relevance of disclosure, investors differentiate among the three components of
ESG scores with regard to the nature of their informational content.

The remainder of the paper is structured as follows. In the next section, we review the literature relevant to our work and
develop our predictions. Section 3 explains the research design. Sections 4 and 5 describe the sample and data and present the
results. Section 6 offers a summary and concluding remarks.

2. Related literature and predictions

2.1. Environmental, social, and governance factors (ESG)

The question of how ESG factors affect a firm’s financial performance and, ultimately, its value has been the subject of conten-
tious debate. Rooted in neoclassical theory, the early understanding was that the relationship between ESG and financial perfor-
mance was uniformly negative (e.g., see Vance, 1975; Wright & Ferris, 1997). Friedman (1970) best summarizes this argument as
claiming that the maximization of owners’ profits is the firm’s only social responsibility. The underlying assumption is that the
payoffs of ESG activities do not exceed their costs. As Kim and Lyon (2015) note, a few recent papers (Fisher-Vanden &
Thorburn, 2011; Jacobs, Singhal, & Subramanian, 2010; Lyon, Lu, Shi, & Yin, 2013) continue to find that firms reporting

47A. Fatemi et al. / Global Finance Journal 38 (2018) 45–64

engagement in environmentally friendly activities or winning green awards experience negative abnormal returns. Such evidence
suggests that investors punish the firm for what they perceive as costly investments.

More recently, however, it has been argued that socially responsible behavior may have a net positive impact on performance
and firm value (Fatemi et al., 2015; Malik, 2015). Within the framework of the stakeholder theory (Freeman, 1984), it can be ar-
gued that socially responsible behavior better satisfies the interests of nonowner stakeholders (e.g., debtors, employees, cus-
tomers, and regulators), allowing for more efficient contracting (Jones, 1995) and opening new paths to further growth and
risk reduction (Fatemi & Fooladi, 2013). Evaluating the issue from a strategic management perspective, Porter and Kramer
(2006, p. 2) similarly posit that “CSR can be much more than a cost, a constraint, or a charitable deed—it can be a source of op-
portunity, innovation, and competitive advantage.”

On the empirical front, a large body of literature has also dealt with the effects of ESG (CSR) factors. Various studies document
a positive association between ESG and nonfinancial performance measures, including process efficiency and reduced material and
energy consumption (Aras & Crowther, 2008; Porter & van der Linde, 1995; Russo & Fouts, 1997); motivating employees,
attracting them to the firm, and creating a bonding mechanism for them (Bhattacharya, Sen, & Korschun, 2008; Greening &
Turban, 2000; Moskowitz, 1972); fostering customer loyalty (Albuquerque, Durnev, & Koskinen, 2015; Ramlugun & Raboute,
2015); advertising effectiveness and brand reputation (Cahan, Chen, Chen, & Nguyen, 2015; Hsu, 2012; McWilliams & Siegel,
2001; Reverte, 2009); reduction of regulatory burden (Freeman, 1984; Neiheisel, 1995); product differentiation and reductions
in price sensitivity (Boehe & Cruz, 2010; Flammer, 2015); and overall customer satisfaction (Pérez & del Bosque, 2015; Sen &
Bhattacharya, 2001; Walsh & Bartikowski, 2013; Xie, 2014).

The relationship between ESG performance and financial performance, as well as valuation, has also been extensively exam-
ined. A number of these studies have found either a negative (e.g., Boyle, Higgins, & Ghon Rhee, 1997; Vance, 1975; also see
Brammer, Brooks, & Pavelin, 2006) or a nonsignificant association between ESG performance and financial performance or firm
value (e.g., Alexander & Buchholz, 1978; Aupperle, Carroll, & Hatfield, 1985; Horvathova, 2010; McWilliams & Siegel, 2000;
Renneboog, Horst, & Zhang, 2008a, 2008b). Others have found a positive association (e.g., Bajic & Yurtoglu, 2017; Dimson,
Karakas, & Li, 2015; Eccles, Ioannou, & Serafeim, 2014; Edmans, 2011; Fatemi et al., 2015; Ge & Liu, 2015; Krüger, 2015). Al-
Tuwaijri, Christensen, and Hughes (2004) deploy a structural equations model within which economic performance is a function
of environmental performance and controls; environmental performance is, in turn, determined by economic performance and
controls. Although they find a positive influence of environmental performance on economic performance, they do not find a sig-
nificant impact of economic performance on environmental performance.2 El Ghoul, Guedhami, and Kim (2015) evaluate the re-
lationship between ESG performance and firm value in 53 countries and find ESG performance to be positively related to firm
value, especially in countries with weaker market-supporting institutions. They therefore conclude that ESG activities mitigate
market failures associated with institutional voids. Finally, a meta-analysis by Margolis et al. (2009) concludes that, on balance,
extant results point to a positive association between ESG activities and both financial performance and firm value. More specif-
ically, Margolis et al. (2009) aggregate 251 individual empirical studies (214 manuscripts) and find a small positive mean effect
size of r = 0.133 (median r = 0.09, weighted r = 0.11).3 For a subset of 106 studies published since 1998, the mean effect size is
only 0.090 (median r = 0.063, weighted r = 0.092), suggesting that the relationship between ESG performance and financial per-
formance may actually have weakened over time.

2.2. ESG disclosure

As Ioannou and Serafeim (2016) report, the form and intensity of ESG (or CSR) reporting differ across firms. Many firms ad-
here to the Global Reporting Initiative (GRI) guidelines in reporting their ESG performance (Vigneau, Humphreys, & Moon, 2015).
More recently, the Initiative for Integrated Reporting (IIR) has attempted to set a standard with its international framework,
which was published in 2013 (Cheng, Ioannou, & Serafeim, 2014; Eccles et al., 2014; Reuter & Messner, 2015). In addition to
the conventional methods of making such disclosures, firms have been increasingly using nontraditional methods, including
websites and social media (e.g., Eberle, Berens, & Li, 2013; Holder-Webb, Cohen, Nath, & Wood, 2009; Reilly & Hynan, 2014).

Research on the extent and the quality of ESG reporting has, for the most part, been based on ratings and checklists developed
by individual researchers who manually collected the data from annual reports or corporate websites (e.g., Aerts, Cormier, &
Magnan, 2008; Cho, Roberts, & Patten, 2010). More recently, ESG disclosure ratings have also come from specialized commercial
information providers. One such example is the ESG score database compiled by Bloomberg and used in the present study.

One prerequisite to understanding the impact of ESG reporting, or disclosure, on a firm’s financial performance and its valu-
ation is to recognize that reporting can reflect various motives. Under voluntary disclosure theory, developed by, among others,
Verrecchia (1983) and Dye (1985), it can be argued that a firm’s ESG engagement is a predictor of its ESG reporting practices:
firms with a positive ESG performance would choose to report extensively on their ESG activities, and those with a negative
ESG performance would choose to report minimally. According to this framework, firms signal their ESG performance to distin-
guish themselves from poorer performers and thus avoid the consequences of adverse selection (Akerlof, 1970). This argument

2 Like our study, that of Al-Tuwaijri et al. (2004) investigates the interrelations among economic performance, environmental performance, and environmental dis-
closure. However, their study differs from our work in the functional form of the assumed relationships. Most importantly, in their model, environmental disclosure is
one of the dependent variables, but it does not have a direct relationship with economic performance. In contrast, we suggest that the relationship between a firm’s ESG
activities and its valuation is moderated by its ESG-related disclosure.

3 For an older meta-analysis, see Orlitzky et al. (2003).

48 A. Fatemi et al. / Global Finance Journal 38 (2018) 45–64

is supported by Cahan et al. (2015), who find that good ESG performance generates favorable publicity, and that firms with good
ESG performance realize a higher firm value (or lower cost of capital) only if they also have favorable media coverage.

Alternatively, it is possible that the firm would use ESG reporting to manage the public’s perception by explaining changes in
its policies with regard to ESG matters. For example, it may intensify its disclosure in order to prevent, or alleviate, the negative
effects of acute environmental damages (or similar events) on its reputation and market value (Brown & Deegan, 1998; Cho &
Patten, 2007), or to restore its legitimacy (Campbell, Craven, & Shrives, 2003; Deegan, 2002). Companies could also use ESG dis-
closure as a mere façade (“cheap talk”), irrespectively of their true ESG performance (Cho, Laine, Roberts, & Rodrigue, 2015a). Fur-
thermore, a firm may attempt to seem more ESG conscious than it really is (“greenwashing”). It is also conceivable that managers
may have incentives not to publicize their environmental, charitable, or otherwise socially responsible investments if they fear
that investors may perceive these activities as unduly costly and detrimental to their interests. Consequently, a firm with a pos-
itive ESG performance may deliberately opt for a low level of ESG disclosure or even actively understate its ESG activities (“undue
modesty”,“brownwashing”; see Kim & Lyon, 2015).

Empirical research to date has produced mixed findings regarding the nature of the relationship between ESG performance
and ESG disclosure. Some earlier studies find no significant relationship between firms’ ESG performance and the intensity of
their ESG disclosure (Freedman & Wasley, 1990; Ingram & Frazier, 1980; Wiseman, 1982). Others find a negative relationship be-
tween environmental performance and environmental disclosure (Hughes, Anderson, & Golden, 2001; Patten, 2002). More re-
cently, studies by Gelb and Strawser (2001), Al-Tuwaijri et al. (2004), Clarkson, Li, Richardson, and Vasvari (2008, 2011),
Dhaliwal, Li, Tsang, and Yang (2011), Lyon and Maxwell (2011), and Gao, Dong, Ni, and Fu (2016) report positive associations.
The inconclusive results may be attributable to problems with the method employed, measurement problems (particularly the
extent of ESG disclosure), sample selection, or a failure to control for other relevant factors (Patten, 2002). For example, as
Clarkson et al. (2008) note, some of what is captured in the disclosure indices used by these studies might be nondiscretionary.
Therefore, the negative relationship reported in some of these studies between ESG performance and the extent of ESG disclosure
might be explained by the additional regulatory disclosure requirements arising from emerging environmental problems.

The empirical evidence on the value relevance of ESG disclosure is not fully conclusive either (see Gray, Kouhy, & Lavers, 1995
for a survey of the earlier literature). Some studies find the relationship to be negative (e.g., de Villiers & van Staden, 2011; Ho &
Taylor, 2007), while others report it as positive (e.g., Clarkson, Fang, Li, & Richardson, 2013; Gamerschlag, Möller, & Verbeeten,
2011; Gao et al., 2016; Middleton, 2015). Research by Brammer and Pavelin (2006) and by Bouten, Everaert, and Roberts
(2012) suggests that different forms of disclosure that emphasize “soft” or “hard” information may have different motives and
can have different effects on firm value. Dhaliwal, Li, Tsang, and Yang (2014) examine the relationship between ESG disclosure
and the equity cost of capital in an international setting that includes 31 countries. They divide these countries into two groups
that are either more or less stakeholder-oriented. They generally find a negative association between ESG disclosure and the cost
of equity capital, and this relationship is more pronounced in stakeholder-oriented countries. Finally, in a baseline model, Plumlee,
Brown, Hayes, and Marshall (2015) find no significant association between the overall level of voluntary ESG disclosure and the
value of the firm, its component cash flows, or its cost of capital. However, after controlling for ESG performance and differenti-
ating between the nature (positive, negative, neutral) and the type (soft, hard) of ESG disclosures, they find that high-quality soft
disclosure is significantly associated with both the cash flows and the cost of capital components of firm value. Building upon the
findings and the insights of this literature, we now proceed to develop our model.

3. The theoretical model: the moderating role of ESG disclosure

We hypothesize that the value of the firm is affected jointly by its ESG activities and the intensity of its ESG-related disclosure.
More precisely, we argue that the association between a firm’s ESG activities and its valuation is moderated by the disclosure re-
lated to such activities. Thus, our basic theoretical model is as follows:

Value of Firm ¼ f
ESG Performance;
ESG Disclosure;
ESG Performance � ESG Disclosure

0
@

1
A

Given extant findings, we expect a positive association between the firm’s ESG performance and its value. That is, holding ev-
erything else constant, we expect a positive relationship between ESG strengths and firm value and a negative relationship be-
tween ESG concerns and firm value. However, given that ESG disclosure can be driven by very different managerial motives,
and given the inconclusive findings of previous research, we do not form directional expectations regarding the first-order rela-
tionship between ESG disclosure and firm value or the interaction term(s) between ESG performance (strengths, weaknesses) and
ESG disclosure. Instead, we simply test the null hypotheses of no relationships.

Research design

Examining the impact of ESG performance, ESG disclosure, and the interaction of ESG performance and ESG disclosure on firm
value requires that we first address the possible endogeneity of ESG disclosure resulting from omitted variables or from simulta-
neity. If, for example, firm value affects ESG disclosure, then the latter will be correlated with the error term in a regression of

49A. Fatemi et al. / Global Finance Journal 38 (2018) 45–64

firm value on ESG disclosure, and the estimated coefficient will be biased and inconsistent. To take this potential endogeneity into
account, we use an instrumental variables approach (e.g., Eugster & Wagner, 2015; Hail, 2002; Jiao, 2010).

In our main analysis, we use three instrumental variables for our potentially endogenous variable of ESG disclosure (DISC).4 The first of
these is the existence of a CSR committee on the board of directors (CSRCOMM). Liao, Luo, and Tang (2015) and Peters and Romi (2014)
find that firms with a CSR committee are more likely to disclose their greenhouse gas emission information and have a higher quality of
such disclosure. Michelon and Parbonetti (2012) find that firms with a CSR committee disclose information on social issues more compre-
hensively. The evidence suggests that CSR committees play an important role in getting sustainability information to stakeholders.5 There-
fore, we expect our first instrumental variable to be highly correlated with ESG disclosure, thus satisfying the relevance requirement.
Peters and Romi (2015) find that the presence of a CSR committee does not affect firm value: for their sample, the market value of
firms with a CSR committee does not differ from the market value of those without a CSR committee. We therefore assume that our in-
strumental variable is unlikely to directly affect firm value, hence meeting the exogeneity condition (exclusion restriction).

Next, we use the dispersion of analysts’ earnings forecasts (DISP) as an instrument. The disclosure literature contains much ev-
idence indicating that the level of disclosure is negatively associated with the dispersion of analysts’ forecasts (e.g., Barron, Kile, &
O’Keefe, 1999; Hope, 2003; Lang & Lundholm, 1996). This is consistent with the intuition that disclosure reduces analysts’ uncer-
tainty about forecasted earnings and differences in information among analysts. Focusing on ESG disclosure, Dhaliwal et al.
(2011) and Dhaliwal, Radhakrishnan, Tsang, and Yang (2012) document that analysts’ forecast errors are lower for firms with better
CSR disclosure. Furthermore, Harjoto and Jo (2015) find that the dispersion of analysts’ earnings forecasts is negatively associated
with mandated ESG activities (“legal CSR activities”, in their terminology) and positively associated with “normative” ESG activities.6

Given these findings, we assume that it is likely that the dispersion of analysts’ earnings forecasts fulfills the instrument relevance
condition. Research on the association between analysts’ forecast dispersion and firm value (or future returns) does not provide con-
clusive evidence. A number of studies find that higher analyst forecast dispersion is associated with higher current stock prices and
lower future returns (e.g., Diether, Malloy, & Scherbina, 2002; Johnson, 2004), while others report that higher dispersions are asso-
ciated with lower current stock prices and higher future returns (e.g., Anderson, Ghysels, & Juergens, 2005; Barron, Stanford, & Yu,
2009). Some studies find a nonmonotonic association (e.g., Doukas, Kim, & Pantzalis, 2006; Li & Wu, 2014), but others fail to find
any significant association (e.g., Brennan, Chordia, & Subrahmanyam, 1998; Hwang & Li, 2008; Li & Wu, 2014). Given the mixed ev-
idence, we cannot assert ex ante whether it is likely that our second instrument satisfies the exogeneity condition.7 In this instance,
we rely on the results of several postestimation tests to assess the appropriateness of the instruments.

Finally, for our third instrument, we use the concentration of a firm’s stock ownership (OWNERCONC). Research has documented a
negative association between ownership concentration and the level of disclosure (for an overview, see, for example, Garcia-Meca &
Sanchez-Ballesta, 2010). Firms with few large shareholders (e.g., family-controlled firms) may have little motivation to disclose more in-
formation than the law requires because the demand for public disclosure is relatively weak. Large shareholders may obtain information
through means other than publicly disclosed reports; for example, they often have direct access to the board of directors. Focusing on CSR,
Brammer and Pavelin (2008), Reverte (2009), Bouten et al. (2012), Li, Luo, Wang, and Wu (2013) and Liao et al. (2015) find that firms
with more concentrated stock ownership have lower levels of disclosure. Hence, we expect the ownership concentration instrument to be
quite likely to meet the relevance condition. The evidence on the impact of ownership concentration on firm value is mixed. Some studies
document a positive association (e.g., Sraer & Thesmar, 2007; Villalonga & Amit, 2006), a few find a negative association (e.g., Anderson &
Reeb, 2004), others document a nonmonotonic association (e.g., Anderson & Reeb, 2003), and some find no significant association at all
(e.g., Perrini, Rossi, & Rovetta, 2008; Weiss & Hilger, 2012; Welch, 2003). Thus, we again rely on the results of the postestimation tests to
assess whether the exogeneity condition is satisfied for ownership concentration.

In specifying our model, we must further take into account the potential endogeneity of the interaction terms between ESG
disclosure and ESG performance. We differentiate between ESG strengths (STR) and ESG concerns (CON). Following
Wooldridge (2002), we use the interactions between ESG strengths (ESG concerns) and the instrumental variables for ESG disclo-
sure as instruments for the interaction terms between ESG strengths (ESG concerns) and ESG disclosure. Thus, our instrumental
variables approach consists of three first-stage regressions, one for each endogenous variable.8

4 In a
(DISC);

5 In a
surance

6 Leg
activitie

7 It is
conside

8 Plea
theory o
reading

DISC ¼ f CSRCOMM; DISP; OWNERCONCð Þ
STR�DISC ¼ f STR�CSRCOMM; STR�DISP; STR�OWNERCONC

� �
CON

�
DISC ¼ f CON�CSRCOMM; CON�DISP; CON�OWNERCONC

� �

dditional analyses, we also use an alternative approach with a different set of instrumental variables for the potentially endogenous variable ESG disclosure
see the section on robustness checks below.
ddition, Peters and Romi (2015) document that the existence of a CSR committee on the board of directors increases the likelihood of adopting CSR report as-
services.
al CSR activities are activities undertaken to comply with laws and regulations covering labor rights and product safety, etc. Normative CSR activities include
s undertaken in accordance with social norms, such as norms on charitable giving and work-life balance (see Harjoto & Jo, 2015).
often difficult to find variables that meet both the relevance and exogeneity conditions. Therefore, the search for and selection of instrumental variables is often
red “magic” (Chenhall & Moers, 2007; Larcker & Rusticus, 2010).
se note that when estimating our instrumental variables model using 2SLS estimation, we include all instruments as regressors in all first-stage regressions. The
f 2SLS estimation does not allow designations of instrumental variables to specific endogenous variables (Baum, 2006; Cameron & Trivedi, 2010). For ease of
, however, we do not show all instruments in all first-stage regression equations, either here or in the following section.

50 A. Fatemi et al. / Global Finance Journal 38 (2018) 45–64

To assess whether our instrumental variables satisfy the relevance condition and the exogeneity condition (exclusion restric-
tion), we rely on several postestimation test statistics. We report Angrist-Pischke’s Partial F-statistic (Angrist & Pischke, 2009) and
Shea’s Partial R2 (Shea, 1997) and the results for tests of underidentification, weak identification, and overidentification.

To complete our model, we add several control variables that the literature has identified as influencing ESG disclosure, ESG perfor-
mance, and firm value (e.g., Cho & Patten, 2007; Clarkson et al., 2008; Jiao, 2010; Jo & Harjoto, 2011; Peters & Romi, 2014). These variables
include proxies for firm profitability (return on assets, ROA) and growth of return on assets (ROAGROWTH), firm size (natural logarithm
of sales, LNSALES), asset intensity (ratio of assets to sales, ASSETSSALES), leverage (ratio of debt to equity, LEV), advertising intensity (ad-
vertising expenditures scaled by sales, ADVERT), research and development intensity (R & D expenditures scaled by sales, R&D), and asset
age (ratio of net property, plant, and equipment to gross property, plant, and equipment, NETGROSSPPE). Additionally, we include two
dummy variables that assume a value of 1 if figures for advertising expenditures or for research and development expenditures are not
available (ADVERTMISSING, R&DMISSING). Furthermore, we include industry- and year-fixed effects. Appendix A provides more detail
on the definitions of the variables and how they are calculated.

Thus, our two-stage model is as follows.
First stage9:

9 As w
10 The
sustaina
2010 to

DISCi;t ¼ α0 þ β1CSRCOMMi;t þ β2DISPi;t þ β3OWNERCONCi;t þ β4STRi;t þ β5CONi;t
þ β6ROAi;t þ β7ROAGROWTHi;t þ β8LNSALESi;t þ β9ASSETSSALESi;tþ
β10LEVi;t þ β11ADVERTi;t þ β12ADVERTMISSINGi;t þ β13R&Di;tþ
β14R&DMISSINGi;t þ β15NETGROSSPPEi;t þ INDUSTRYi;t þ YEARi;t þ εi;t

ð1Þ

STR�DISCi;t ¼ α0 þ β1STR
�CSRCOMMi;t þ β2STR

�DISPi;t þ β3STR
�OWNERCONCi;tþ

β4STRi;t þ β5CONi;t þ β6ROAi;t þ β7ROAGROWTHi;t þ β8LNSALESi;tþ
β9ASSETSSALESi;t þ β10LEVi;t þ β11ADVERTi;t þ β12ADVERTMISSINGi;tþ
β13R&Di;t þ β14R&DMISSINGi;t þ β15NETGROSSPPEi;tþ
INDUSTRYi;t þ YEARi;t þ εi;t

ð2Þ

CON�DISCi;t ¼ α0 þ β1CON
�CSRCOMMi;t þ β2CON

�DISPi;t þ β3CON
�OWNERCONCi;t

þ β4STRi;t þ β5CONi;t þ β6ROAi;t þ β7ROAGROWTHi;t þ β8LNSALESi;tþ
β9ASSETSSALESi;t þ β10LEVi;t þ β11ADVERTi;t þ β12ADVERTMISSINGi;t
þ β13R&Di;t þ β14R&DMISSINGi;t þ β15NETGROSSPPEi;tþ
INDUSTRYi;t þ YEARi;t þ εi;t

ð3Þ

Second stage:

TOBINQi;t ¼ α0 þ β1DISCi;t þ β2STRi;t þ β3STR
�DISCi;t þ β4CONi;t þ β5CON

�DISCi;tþ
β6ROAi;t þ β7ROAGROWTHi;t þ β8LNSALESi;t þ β9ASSETSSALESi;tþ
β10LEVi;t þ β11ADVERTi;t þ β12ADVERTMISSINGi;t þ β13R&Di;tþ
β14R&DMISSINGi;t þ β15NETGROSSPPEi;t þ INDUSTRYi;t þ YEARi;t þ εi;t

ð4Þ

4. Data and sample

To operationalize our model, in accord with the approach taken by other researchers (e.g., Aktas, De Bodt, & Cousin, 2011;
Cornett, Erhemjamts, & Tehranian, 2016; Harjoto & Jo, 2015; Plumlee et al., 2015), we use data compiled by KLD Research and
Analytics as a proxy for ESG activities. KLD divides a firm’s CSR activities into 13 categories: community, diversity, employment,
environment, human rights, product, alcohol, gaming, firearms, military, nuclear, tobacco, and corporate governance. It then com-
piles and reports the number of strengths and concerns for each category. In this study, we use these measures of strengths and
weaknesses as our proxy for a firm’s CSR activities.

We use Bloomberg’s measure of ESG disclosure, first made available in 2009, as a proxy for disclosure. To our knowledge, this
constitutes one of the first uses of Bloomberg’s index in an academic study. Bloomberg currently compiles approximately 300 data
points from each of approximately 11,000 companies in 63 countries. Screening publicly available sources, Bloomberg assesses the
extent of each firm’s disclosure of its environmental, social, and governance (ESG) activities. Bloomberg’s data points, which are
weighed by importance, come from company filings, such as CSR reports, annual reports, and corporate websites, and thus reflect
the universe of information publicly available to investors. Depending on the data points collected, and tailoring its reports to the
industry, Bloomberg then estimates disclosure scores ranging between 0.1 (lowest) and 100 (highest).10

e note above, for the sake of brevity, we do not show all instruments in all our first-stage regression equations, either here or in the following section (see footnote 8).
se ratings are widely used by investors and are perceived to be credible by sustainability professionals (see, for example, SustainAbility at http://www.
bility.com/company). Eccles, Krzus, Rogers, and Serafeim (2012) document investors’ use of Bloomberg’s ESG ratings over a six-month span ranging from
2011. They show that data on firms’ aggregate ESG disclosure scores have been widely consulted.

Table 1
Sample composition by year and industry.

Panel A: composition by year

2006 87
2007 227
2008 287
2009 326
2010 352
2011 361
Total 1640

Panel B: composition by industry

Agriculture & forestry 25
Mining & construction 165
Manufacturing 771
Transportation & utilities 318
Wholesale & retail 107
Services 254
Total 1640

51A. Fatemi et al. / Global Finance Journal 38 (2018) 45–64

We match the available data for all U.S. firms from KLD and Bloomberg, our two primary data sources. Because there are few
disclosure scores for years before 2006 and limited data from our supplemental sources (Worldscope, Thomson One and I/B/E/S;
see Appendix A for details), we net a total of 1640 firm-year observations for 403 U.S. listed companies for reporting periods be-
tween 2006 and 2011. Table 1 shows the composition of our sample by year (Panel A) and by industry (Panel B).

Table 2 presents the descriptive statistics of the variables. All continuous variables are winsorized at the 1% and 99% percen-
tiles. The mean value of Tobin’s q is 1.88 (median 1.56), with a standard deviation of 1.00. The mean ESG disclosure of the firms is
22.28 (median 16.12), with a standard deviation of 12.74. The mean value of KLD strengths is 3.36 (median 2.0) with a standard
deviation of 3.82, and concerns have a mean value of 3.66 (median 3.0) with a standard deviation of 2.46. These values appear
reasonable because they fall within the bounds of estimates reported in previous work on this topic. Among our instrumental var-
iables, 41.22% of our sample firms operate with standing sustainability committees. The mean value of the standard deviation of
analysts’ six-months-forward earnings forecasts is 13.46% (median 7%), with its own standard deviation at 20.5%. The mean value
for the largest block of shares held by a single shareholder is 11.8% (median 9.3%), with a standard deviation of close to 12%. Fi-
nally, the average values of our control variables also appear to be reasonable, with ROA at 8%, natural log of sales at 8.5, assets to
sales at 2.1, debt to equity at 45%, advertising expenses at 2.3% of sales, R&D expenditures at 2.85% of sales, and the ratio of net
property, plant, and equipment to gross PPE at 53%.

Table 3 reports the cross-correlations of the variables. We note that Tobin’s q is positively correlated with ROA, advertising in-
tensity, and R&D intensity. It is negatively correlated with ESG concerns, natural log of sales, asset intensity, debt to equity ratio,
asset age, and each of the two dummy variables that signify missing values for advertising and R&D expenditures of the firm. ESG
strengths are positively correlated with ESG concerns, ROA, natural log of sales, advertising intensity, and R&D intensity. They are
negatively correlated with asset intensity, age of assets, debt to equity ratio, and each of the two dummy variables that signify
missing values for advertising and R&D expenditures. ESG concerns are positively correlated with natural log of sales, debt to eq-
uity ratio, and the dummy variable that signifies a missing value for advertising expenses. They are negatively correlated with

Table 2
Descriptive statistics.

Mean Standard deviation Percentile 25 Median Percentile 75

TOBINQ 1.8819 1.0010 1.2183 1.5639 2.1538
DISC 22.2769 12.7422 13.2200 16.1200 29.2050
CSRCOMM 0.4122 0.4924 0.0000 0.0000 1.0000
DISP 0.1346 0.2048 0.0400 0.0700 0.1400
OWNERCONC 11.7808 11.9586 6.9000 9.3000 12.4000
STR 3.3610 3.8158 1.0000 2.0000 5.0000
CON 3.6640 2.4577 2.0000 3.0000 5.0000
ROA 0.0838 0.0588 0.0436 0.0713 0.1086
ROAGROWTH -0.0002 1.4716 -0.2623 -0.0205 0.1655
LNSALES 8.5089 1.1905 7.6170 8.4852 9.3745
ASSETSSALES 2.1030 2.1440 0.9642 1.4156 2.4356
LEV 0.4516 0.5980 0.1030 0.4006 0.5915
ADVERT 0.0232 0.0673 0.0000 0.0000 0.0024
ADVERTMISSING 0.7250 0.4467 0.0000 1.0000 1.0000
R&D 0.0285 0.0594 0.0000 0.0000 0.0254
R&DMISSING 0.5604 0.4965 0.0000 1.0000 1.0000
NETGROSSPPE 0.5337 0.1559 0.4066 0.5316 0.6522

Note: Variables are defined as in Appendix A. The continuous variables are winsorized at the 1st and 99th percentiles.

Table 3
Correlations.

TOBINQ DISC STR CON ROA ROA GROWTH LN SALES ASSETS SALES LEV ADVERT ADVERTMISSING R&D R&D MISSING NET GROSS PPE

TOBINQ −0.0226 0.0107 −0.1250 0.6789 0.1342 −0.1310 −0.3384 −0.7184 0.2698 −0.2551 0.3135 −0.2692 −0.2362
DISC −0.0451 0.6124 0.3317 0.0241 0.0196 0.5136 −0.0896 0.0145 0.0719 −0.0791 0.1704 −0.1958 −0.0878
STR 0.0015 0.6351 0.2926 0.0630 0.0300 0.6147 −0.0874 −0.0163 0.1695 −0.1673 0.2511 −0.2480 −0.1496
CON −0.1311 0.3258 0.3617 −0.0375 0.0019 0.5358 −0.0383 0.1456 −0.1090 0.1088 −0.0568 0.0007 0.0669
ROA 0.7253 0.0293 0.0815 −0.0465 0.3040 0.0480 −0.4405 −0.5969 0.1707 −0.1635 0.2308 −0.2238 −0.2128
ROAGROWTH 0.0322 0.0207 0.0173 −0.0173 0.0903 0.0666 −0.0634 −0.0836 0.0154 −0.0140 0.0081 −0.0135 −0.0303
LNSALES −0.1640 0.4945 0.6062 0.5713 −0.0108 0.0269 −0.3020 0.0963 −0.0122 0.0033 0.0608 −0.1207 −0.0906
ASSETSSALES −0.2569 −0.1171 −0.1207 −0.0931 −0.3319 −0.0327 −0.2984 0.3610 −0.1476 0.1538 −0.1684 0.2453 0.4721
LEV −0.3977 −0.0538 −0.0678 0.0592 −0.3895 0.0248 0.0255 0.3889 −0.3123 0.3054 −0.4082 0.3373 0.2968
ADVERT 0.3030 −0.0012 0.0979 −0.0997 0.1761 0.0147 −0.0815 −0.0708 −0.1574 −0.9831 0.2908 −0.2428 −0.2508
ADVERTMISSING −0.2496 −0.1116 −0.1866 0.0781 −0.1682 −0.0004 −0.0094 0.0707 0.1830 −0.5612 −0.2750 0.2304 0.2468
R&D 0.2134 0.1028 0.1678 −0.1270 0.1052 0.0126 −0.1098 −0.0916 −0.2494 0.3004 −0.2450 −0.9470 −0.4050
R&DMISSING −0.1486 −0.1697 −0.2701 −0.0391 −0.1435 0.0014 −0.1250 0.2932 0.2775 −0.2093 0.2304 −0.5416 0.4299
NETGROSSPPE −0.1553 −0.0498 −0.1646 0.0345 −0.1647 −0.0108 −0.1104 0.4010 0.2670 −0.1842 0.2498 −0.1660 0.4197

Note: Pearson (Spearman) correlation coefficients are presented below (above) the diagonal line. Bold numbers denote correlations significant at least at the 5% level. Variables are defined as in Appendix A.

5
2

A
.Fatem

iet
al./

G
lobalFin

an
ce

Jou
rn
al38

(201
8)

4
5–

64

Table 4
Instrumental variable regression results.

First Stage
DISC

First Stage
STR*DISC

First Stage
CON*DISC

Second Stage
TOBINQ

CSRCOMM 5.5812***
(3.80)
(0.000)

−27.2802***
(−2.91)
(0.004)

−26.7078***
(−3.35)
(0.001)

DISP −5.7493
(−1.45)
(0.147)

−22.5013
(−1.10)
(0.274)

−62.8658**
(−2.47)
(0.014)

OWNERCONC −0.1574***
(−4.33)
(0.000)

−0.6066**
(−2.23)
(0.026)

−0.5226
(−1.53)
(0.128)

DISC −0.0182**
(−2.08)
(0.038)

STR 1.0246***
(3.77)
(0.000)

31.7633***
(8.76)
(0.000)

3.8882***
(3.71)
(0.000)

0.0739*
(1.91)
(0.056)

STR*DISC −0.0017*
(−1.84)
(0.066)

STR*CSRCOMM 0.1556
(0.76)
(0.448)

12.1597***
(5.38)
(0.000)

1.2461
(1.24)
(0.217)

STR*DISP 1.5990**
(2.44)
(0.015)

0.9630
(0.16)
(0.871)

8.8283***
(2.77)
(0.006)

STR*OWNERCONC 0.0240***
(3.61)
(0.000)

0.0920
(0.78)
(0.437)

0.1079***
(3.21)
(0.001)

CON −0.0602
(−0.20)
(0.845)

−2.4563
(−1.30)
(0.196)

19.5712***
(8.62)
(0.000)

−0.0810**
(−2.07)
(0.039)

CON*DISC 0.0031**
(2.25)
(0.024)

CON*CSRCOMM −0.5894*
(−1.84)
(0.067)

−3.1584
(−1.54)
(0.125)

9.0636***
(3.86)
(0.000)

CON*DISP 0.8600
(0.95)
(0.342)

7.8676
(1.46)
(0.145)

11.2842
(1.59)
(0.113)

CON*OWNERCONC 0.0201**
(2.17)
(0.030)

0.1216**
(2.17)
(0.030)

0.7292
(0.73)
(0.464)

ROA −0.5542
(−0.08)
(0.939)

27.5021
(0.51)
(0.609)

13.9996***
(3.71)
(0.000)

11.2647***
(15.10)
(0.000)

ROAGROWTH 0.1277
(0.98)
(0.329)

−0.0701
(−0.07)
(0.945)

0.6031
(1.09)
(0.278)

−0.0195*
(−1.66)
(0.097)

LNSALES 1.5235***
(2.87)
(0.004)

1.6365
(0.45)
(0.653)

3.9739
(1.58)
(0.115)

−0.1173***
(−3.11)
(0.002)

ASSETSSALES −0.3120*
(−1.88)
(0.061)

−1.2681
(−1.19)
(0.236)

−0.3564
(−0.42)
(0.677)

−0.0431***
(−2.99)
(0.003)

LEV −1.0520*
(−1.95)
(0.051)

−1.2066
(−0.37)
(0.715)

−6.7594*
(−1.94)
(0.053)

−0.1125***
(−2.63)
(0.009)

ADVERT −10.0060
(−1.56)
(0.120)

−79.1828
(−1.61)
(0.108)

−52.0662**
(−2.14)
(0.033)

1.0489*
(1.93)
(0.054)

ADVERTMISSING −0.9892
(−0.89)
(0.374)

−5.8204
(−0.72)
(0.470)

−4.6496
(−0.84)
(0.401)

−0.1214*
(−1.70)
(0.089)

R&D 13.6443
(0.98)
(0.327)

72.1738
(1.06)
(0.292)

69.4125*
(1.81)
(0.071)

1.3761**
(1.97)
(0.048)

R&DMISSING −0.2860 1.0478 −0.4935 −0.0987

(continued on next page)

53A. Fatemi et al. / Global Finance Journal 38 (2018) 45–64

Table 4 (continued)

First Stage
DISC

First Stage
STR*DISC

First Stage
CON*DISC

Second Stage
TOBINQ

(−0.20)
(0.840)

(0.11)
(0.909)

(−0.06)
(0.951)

(−1.31)
(0.192)

NETGROSSPPE 1.9493
(0.67)
(0.502)

15.1231
(0.82)
(0.410)

2.4867
(0.18)
(0.855)

0.0969
(0.53)
(0.594)

CONST 5.7930
(1.08)
(0.282)

−23.3059
(−0.71)
(0.479)

−30.3238
(−1.24)
(0.216)

2.5902***
(7.93)
(0.000)

INDUSTRY Included Included Included Included
YEAR Included Included Included Included
F-stat 22.55*** 55.48*** 45.65*** 26.94***
Adj. R2 51.36% 86.59% 79.72% 60.21%
Angrist-Pischke Partial F-stat 6.15*** 6.93*** 4.60***
Shea Partial R2 16.45% 14.06% 18.76%
Kleibergen-Paap rk LM statistic
χ2 (p-value)

25.5220***
(0.000)

Kleibergen-Paap rk Wald F statistic
(crit. value at 10% max. IV relative bias)

13.830
(9.37)

Anderson-Rubin Wald test χ2 (p-value) 17.25**
(0.0449)

Hansen J statistic χ2 (p-value) 7.6860
(0.2621)

N 1640 1640 1640 1640

Note: *** (**, *) denotes significance at the 1% (5%, 10%) level (two-sided test). t-statistics and p-values are given in parentheses. Variables are defined as in Ap-
pendix A.
Columns 1 to 3 present the estimation results for the three first-stage regressions for the three endogenous regressors, i.e., ESG disclosure (DISC), the interaction
between ESG strengths and ESG disclosure (STR*DISC), and the interaction between ESG concerns and ESG disclosure (CON*DISC). The existence of a CSR commit-
tee (CSRCOMM), analyst forecast dispersion (DISP), and ownership concentration (OWNERCONC) are used as instruments for the endogenous regressor ESG dis-
closure (DISC); the interactions among these three instruments and ESG strengths (ESG concerns) are used as instruments for the endogenous regressor STR*DISC
(CON*DISC). Column 4 presents the estimation results for the second-stage regression for the dependent variable, i.e., firm value (TOBINQ): TOBINQi,t = α0 + β1-
DISCi,t + β2STRi,t + β3STR*DISCi,t + β4CONi,t + β5CON*DISCi,t + Controls, where DISC is instrumented by CSRCOMM, DISP, and OWNERCONC; STR*DISC is instru-
mented by STR*CSRCOMM, STR*DISP, and STR*OWNERCONC; and CON*DISC is instrumented by CON*CSRCOMM, CON*DISP, and CON*OWNERCONC.

54 A. Fatemi et al. / Global Finance Journal 38 (2018) 45–64

asset intensity, advertising intensity, and R&D intensity. Finally, ESG disclosure scores are positively correlated with ESG strengths,
ESG concerns, natural log of sales, and R&D intensity. They are negatively correlated with asset intensity, debt to equity ratio, asset
age, and each of the two dummy variables that signify missing values of advertising and R&D expenses.

5. Empirical results and discussion

5.1. ESG performance, ESG disclosure, and firm value

Table 4 presents the estimated results for our 2SLS model. The first three columns of Table 4 show the results for the first-
stage regressions investigating the determinants of ESG disclosure and those of the interaction between ESG disclosure and
ESG strengths and ESG concerns. The last column of Table 4 shows the results for the second-stage regression, which examines
the influence of ESG performance, ESG disclosure, and the interaction between ESG performance and ESG disclosure on firm
value. These and the following analyses are based on panel data. Because our dataset contains more firms than years, following
Petersen (2009), we include dummy variables for each time period to capture the possible correlation of same-year observations
belonging to different firms, and we use standard errors clustered by firm to control for the possible correlation between same-
firm observations belonging to different years. These standard errors are robust to heteroskedasticity, according to White (1980).

We note that our model is not underidentified: the Kleibergen-Paap rk LM statistic (Kleibergen & Paap, 2006) is highly signif-
icant (p b 0.001). Additionally, the following test statistics indicate that our instrumental variables are relevant and strong. First,
the Kleibergen and Paap (2006) rk Wald F-statistic is higher (13.83) than the critical value (9.37) at a 10% maximal IV relative
bias. Second, the Angrist-Pischke Partial F-statistic is highly significant for all first-stage regressions (p b 0.001). Third, Shea’s Par-
tial R2 is between 14% and 19%, indicating that our instrumental variables explain a reasonable portion of the variation of ESG
disclosure and the variations of the interactions between ESG activities and ESG disclosure. Further, Hansen J, the overidentifica-
tion test statistic (Hansen, 1982), is insignificant (p = 0.2621). Thus, our instruments satisfy the exogeneity condition (exclusion
restriction).11 Furthermore, the Anderson-Rubin Wald test (Anderson & Rubin, 1949; Chernozhukov & Hansen, 2008) is significant
at the 5% level, suggesting that our endogenous variables influence firm value even in the presence of weak instruments.

11 Instead of the Anderson LM statistic, the Cragg-Donald Wald F-statistic, and the Sargan statistic, we report the Kleibergen-Paap rk LM statistic, the Kleibergen-Paap
rk Wald F-statistic, and the Hansen J statistic, as those statistics remain valid in the presence of heteroskedasticity.

55A. Fatemi et al. / Global Finance Journal 38 (2018) 45–64

The results reported in Table 4 indicate that the presence of a sustainability committee within the firm is a significant deter-
minant in all three first-stage regressions designed to explain disclosure (DISC), the interaction between disclosure and ESG
strengths (STR*DISC), and the interaction between disclosure and ESG concerns (CON*DISC). Ownership concentration is a signif-
icant determinant in two regressions (those of DISC and STR*DISC), and the dispersion of analysts’ earnings forecasts is a signif-
icant determinant only in one regression (that of CON*DISC). The results suggest that the existence of a sustainability committee
increases ESG disclosure (DISC), and ownership concentration decreases it. Sustainability committees increase disclosure in the
presence of both ESG strengths (STR*DISC) and ESG concerns (CON*DISC).

Finally, turning to the second-stage regression, our model explains approximately 60% of the variance of Tobin’s q. The results
indicate that ESG strengths significantly increase firm value and that ESG concerns significantly decrease it. ESG disclosure signif-
icantly decreases firm value.

Our main interest lies in the estimates of β3 and β5, the regression coefficients of the interaction terms STR*DISC and
CON*DISC, which measure the moderating effect of ESG disclosure on the association between ESG strengths and concerns and
firm value. The β3-estimate is significantly negative, albeit only at the 10% level (p = 0.066). This indicates that the generally neg-
ative valuation effect of ESG disclosure is exacerbated for firms with positive ESG performance. One possible explanation is that, if
the firm has ESG strengths, high disclosure may signal that the firm is overinvesting in ESG (see also Kim & Lyon, 2015 on this
point). An equally plausible explanation is that investors may perceive the firm as attempting to cover up for a lack of depth
in its ESG actions with “too much talk.” The extent of ESG-related disclosure is correlated more strongly with ESG strengths
than it is with ESG concerns. As Table 3 shows, the Pearson correlation coefficient for DISC and STR is 0.6351 (p b 0.0001), but
the correlation coefficient for DISC and CON is only 0.3258 (p b 0.0001). Therefore, firms might find it fruitful to exercise restraint
in their disclosures related to ESG strengths and devote more of their efforts to explaining and contextualizing ESG concerns.

Fig. 1a illustrates the interaction between ESG strengths and ESG disclosure. It shows the conditional effects revealed by our
instrumental variable regressions of Tobin’s q on ESG strengths and the control variables. We separate our sample into firms
with high (i.e., above-median) and low (i.e., below-median) ESG disclosure, and we estimate the instrumental variable model sep-
arately for each of the two subsamples. Conditional predictions are generated using the parameter estimates of the regression co-
efficients. To generate these predictions, we set the control variables to their relevant subsample medians, and we exclude both
the industry and year indicator variables. The resulting graph illustrates the moderating effect of the disclosure by showing that
the positive slope of the regression equation is higher for the low-disclosure subsample than it is for the high-disclosure subsam-
ple. Therefore, it appears that the positive association between ESG strengths and firm value is more pronounced for firms that
disclose less.

In Table 4, the β5-estimate is significantly positive (p = 0.024), indicating that in the presence of ESG concerns, higher ESG
disclosure increases valuation. Thus, it appears that more extensive disclosure tends to alleviate the valuation decrease associated
with ESG concerns. Fig. 1b illustrates this moderating effect. Through a process identical to that explained for Fig. 1a above, the
graph shows that the negative slope of the regression equation is much less steep for the high-disclosure subsample than it is
for the low-disclosure subsample. Therefore, it follows that the negative impact of ESG concerns on firm value is much less pro-
nounced for firms that disclose more. One possible explanation is that by properly framing the appropriateness of its operations
and its ESG policies, the firm succeeds in its efforts to legitimate its behavior and to affect investor expectations. Alternatively, the
firm may convince investors that it has made credible commitments to overcome ESG weaknesses in the future (e.g., it may
showcase convincing plans to reduce the externalities generated by its operations or to better comply with regulations).

Among the control variables, Tobin’s q is increased by ROA and R&D intensity (and less significantly by advertising intensity).
It is decreased by firm size (as proxied by natural log of sales), asset intensity, and financial leverage (and less significantly by
growth in ROA and the dummy variable signifying missing values of advertising expenses).

A closer examination indicates that for the average firm in our sample, ESG strengths and concerns exert economically mean-
ingful valuation effects. According to the KLD data, the median firm in our sample has two ESG strengths (mean: 3.3610) and
three ESG concerns (mean: 3.6640). Taken at face value, the estimation results indicate that, with everything else constant, and
in the absence of any ESG-related disclosure, adding another strength would, on average, increase Tobin’s q by 7.39%. Similarly,
an incremental concern would, on average, decrease it by 8.1%. To gain a complete picture of the valuation impact of ESG
strengths and concerns, we also need to take into account the moderating effects of ESG-related disclosure. The median firm in
our sample has a Bloomberg ESG disclosure rating of 16.12 (mean: 22.2769).12 Hence, considering the moderating effects of dis-
closure, a median firm that adds another ESG strength would increase valuation by just 4.65% (+0.0739 + 16.12*(−0.0017)). On
the other hand, an added ESG concern would decrease it by 3.1%. In other words, we find that ESG-related disclosure strongly
reduces the valuation effects of both ESG strengths and ESG concerns. In particular, with ESG concerns, the moderated valuation
effect is less than half what it would be without the disclosure.

5.2. Additional tests: ESG strengths, concerns, and their components

We perform additional tests to examine the differential influences of ESG strengths and ESG concerns on firm value and to
investigate whether the valuation consequences of ESG strengths and concerns and their disclosure hold uniformly for all three

12 In the absence of ESG strength and concerns, and with everything else held constant, each further ESG disclosure rating point is associated with a 1.82% decrease in
firm value. However, for the median firm, with two ESG strengths and three ESG concerns, a further ESG disclosure rating point is associated with a 1.23% decrease in
Tobin’s q (−0.0182 + 2 ∗ (−0.0017) + 3 ∗ 0.0031).

56 A. Fatemi et al. / Global Finance Journal 38 (2018) 45–64

components of ESG factors. The major findings of these additional tests are summarized in Table 5. For the sake of brevity, we
report the estimation results only for the coefficients of interest from the second-stage regressions of our 2SLS model and refrain
from reporting results for the first-stage regressions or for the control variables, which are generally very similar to those for our
main model reported in Table 4. In particular, the first-stage regressions are similar, and none of the five models suffers from
under-, weak, or overidentification. The postestimation tests indicate that our instrumental variables meet the relevance condition
and the exclusion restriction when used for the separate analyses of ESG strengths and concerns and for each of the three envi-
ronmental, social, and governance disclosure components.

The estimates of the coefficients for our variables of interest, DISC, STR, CON, and their interactions, STR*DISC and CON*DISC,
are also generally consistent with those for the main model. In all model variants, ESG-related disclosure significantly decreases
firm value. ESG strengths and ESG concerns, when examined independently of each other in models 1 and 2, also behave consis-
tently with the main model; that is, the strengths increase firm value and the concerns decrease it. Estimating the model sepa-
rately for each of the three components of ESG (models 3 to 5), we find again that concerns always decrease firm value, and
disclosure of those concerns always alleviates this reduction. Furthermore, in accord with the main results, for all three models,
the estimated coefficients for environmental, social, and governance strengths are positive, and the coefficients for the interaction
terms STR*DISC are negative. However, the estimated coefficients of STR and STR*DISC are statistically significant only in model 3
for environmental strengths and concerns, and only at the 10% level. In other words, our findings suggest that firms do not stand
to gain (or lose) significantly from investing in either social or governance strengths. At the same time, the results indicate that
investors react negatively to social and governance concerns. In fact, the estimated coefficient in model 5 for governance concerns

Fig. 1. a. Interaction between ESG disclosure and ESG strengths. b. Interaction between ESG disclosure and ESG concerns. Figure panels a and b are conditional
effects plots based on our instrumental variables regressions of Tobin’s q on ESG strengths (ESG concerns) and the control variables. We estimate the instrumental
variable model for ESG strengths (ESG concerns) for the subsamples of high (i.e., above-median) and low (i.e., below-median) ESG disclosure, generate conditional
predictions using the parameter estimates of the regression coefficients, and plot the association between Tobin’s q and ESG strengths and ESG concerns, respec-
tively. For the predictions, we set the control variables equal to their relevant subsample medians (indicator variables for industry and year are excluded when
generating the predictions).

Table 5
Additional tests—2SLS estimation results for subsamples (only second-stage regression shown).

(1) ESG strengths (2) ESG concerns (3) Environmental (4) Social (5) Governance

DISC −0.0220*
(−1.77)
(0.076)

−0.0174**
(−2.08)
(0.038)

−0.0183*
(−1.67)
(0.095)

−0.0290***
(−2.71)
(0.007)

−0.0626**
(−2.55)
(0.011)

STR 0.1048*
(1.80)
(0.072)

0.2340*
(1.75)
(0.079)

0.0253
(0.54)
(0.592)

1.6956
(1.28)
(0.202)

STR*DISC −0.0037**
(−2.24)
(0.025)

−0.0085*
(−1.69)
(0.091)

−0.0001
(−0.05)
(0.959)

−0.0329
(−1.42)
(0.157)

CON −0.0776**
(−2.03)
(0.043)

−0.5318**
(−2.30)
(0.022)

−0.1142*
(−1.85)
(0.065)

−1.2545**
(−2.31)
(0.021)

CON*DISC 0.0030**
(2.27)
(0.023)

0.0218**
(2.02)
(0.043)

0.0082***
(2.89)
(0.004)

0.0228**
(2.30)
(0.021)

Controls Included Included Included Included Included
Constant Included Included Included Included Included
INDUSTRY Included Included Included Included Included
YEAR Included Included Included Included Included
F-stat 13.49*** 28.45*** 9.07*** 14.92*** 19.68***
Adj. R2 45.33% 62.05% 30.15% 44.46% 51.68%
N 1239 1598 874 1506 1615

Note: *** (**, *) denotes significance at the 1% (5%, 10%) level (two-sided test). t-statistics and p-values are given in parentheses. Variables are defined as in Ap-
pendix A.
Table 5 presents the estimation results for the second-stage regression for the dependent variable, i.e., firm value (TOBINQ) for several subsamples. For the sake of
brevity, the estimation results for the three first-stage regressions for the three endogenous regressors, i.e., ESG disclosure (DISC), the interaction between ESG
strengths and ESG disclosure (STR*DISC), and the interaction between ESG concerns and ESG disclosure (CON*DISC), are not tabulated.
TOBINQi,t = α0 + β1DISCi,t + β2STRi,t + β3STR*DISCi,t + β4CONi,t + β5CON*DISCi,t + Controls, where DISC is instrumented by CSRCOMM, DISP, and
OWNERCONC; STR*DISC is instrumented by STR*CSRCOMM, STR*DISP, and STR*OWNERCONC; and CON*DISC is instrumented by CON*CSRCOMM, CON*DISP,
and CON*OWNERCONC.
Columns 1 and 2 present the estimation results for the ESG strengths and ESG concerns subsamples, respectively.
Columns 3 to 5 show the estimation results for the environmental, social, and governance subsamples, respectively.

57A. Fatemi et al. / Global Finance Journal 38 (2018) 45–64

is the largest among all model variants (β = −1.2545), much larger than the estimated coefficient for social concerns (β =
−0.1142) and that for environmental concerns (β = −0.5318).

In what follows, we again more closely examine the economic importance of the cumulative valuation effects of ESG strengths
and concerns on the one hand and ESG disclosure on the other hand. The median firm within our KLD environmental subsample
has a Bloomberg ESG disclosure rating of 13.9535. Thus, with all else constant, adding an environmental strength is associated
with an 11.54% increase in Tobin’s q (+0.234 + 13.9535 ∗ (−0.0085)). However, an incremental concern has an even stronger
impact; it is associated with a 22.76% decrease in Tobin’s q (−0.5318 + 13.9535 ∗ 0.0218). Again, for the KLD social and gover-
nance subsamples, only concerns and their interactions with their respective disclosures are value relevant. In the case of the KLD
social subsample, adding a further concern is, per se, associated with an 11.4% decrease in Tobin’s q. Once the moderating effect of
the related disclosure is taken into account, however, the decrease is only 4.23% (−0.1142 + 8.7719 ∗ 0.0082). For the KLD gov-
ernance subsample, an incremental concern in the absence of any related disclosure is associated with a massive 125.5% decrease
in Tobin’s q. However, once the strong mitigating effect of disclosure is factored in, the overall decrease is only 7.38% (−1.2545
+ 51.7857 ∗ 0.0228).

To sum up, while disclosure tends to reduce the negative valuation consequences of all three ESG components, this effect is much
more pronounced when those concerns are centered on governance issues rather than environmental or social concerns—perhaps be-
cause governance-related disclosures tend to be mandated and regulated by the SEC (see, for example, Holder-Webb, Cohen, Nath, &
Wood, 2008), so the markets can assess their veracity with relatively high ease and confidence. Disclosure related to social and environ-
mental concerns, in contrast, is usually voluntary, invariably more opaque, and more difficult to verify.

5.3. Robustness checks

Given the potential problems associated with the instrumental variables approach, especially the issue of selecting instrumen-
tal variables that satisfy the relevance and exogeneity conditions, Larcker and Rusticus (2010) recommend that the 2SLS/IV esti-
mates be compared to those obtained from a simple OLS. Accordingly, we repeat all our analyses using simple OLS estimation. The
results, presented in Appendix B,13 are in line with the 2SLS/IV estimation results: ESG strengths increase firm value and ESG con-
cerns decrease it. Furthermore, ESG disclosure per se decreases firm value. For ESG concerns, higher ESG disclosure increases

13 For brevity, the results for the separate subsamples of ESG strengths and concerns and for the separate subsamples of environmental, social, and governance activ-
ities and related disclosure are not tabulated. The results from OLS estimation are consistent with the results presented in the previous section from 2SLS/IV estimation.

58 A. Fatemi et al. / Global Finance Journal 38 (2018) 45–64

value, but for ESG strengths, higher ESG disclosure decreases it. The OLS estimated coefficients are comparable to those from 2SLS/
IV, and the significance levels are also generally comparable. As a further robustness test, we rerun all analyses using SEM estima-
tion. The results, presented in Appendix C, are again consistent with those from the 2SLS/IV method.

In line with the recommendations of Larcker and Rusticus (2010), we also establish whether the estimates of our 2SLS/IV are
robust with regard to the choice of instrumental variables. To that end, we first use the average level of disclosure for firms in the
same industry-year (where each firm is assigned to one of the 17 industry groups of Fama & French, 1997) while excluding the
current firm (e.g., Beekes, Brown, Zhan, & Zhang, 2016; Lang & Stice-Lawrence, 2015). The rationale behind this instrument is that
a firm’s ESG disclosure is influenced by the ESG disclosure of other firms within the same industry and the same year. Studies
have shown that a firm’s disclosure policy is influenced by the policies of its industry peers (e.g., Acharya, DeMarzo, & Kremer,
2011; Jorgensen & Kirschenheiter, 2012; Rogers, Schrand, & Zechman, 2014). Furthermore, ESG disclosure varies across industries
(see, e.g., Eccles et al., 2012; Gamerschlag et al., 2011) and over time as a result of laws and regulations (see, e.g., Cho, Michelon,
Patten, & Roberts, 2015b). Accordingly, we expect the industry-year average level of ESG disclosure to be quite likely to meet the
instrument relevance condition. Because each firm chooses its own value-maximizing disclosure policy, the average disclosure
level of the firm’s peers is unlikely to directly affect its market value. Therefore, we assume that our instrumental variable satisfies
the exogeneity condition (exclusion restriction).

Next, we use the level of institutional ownership as an instrumental variable. Research documents a positive association be-
tween institutional ownership and the level of disclosure (e.g., Boone & White, 2015; Bushee & Noe, 2000; Jennings &
Marques, 2011). Focusing on ESG, Saleh, Zulkifli, and Muhamad (2010) find that firms with greater institutional ownership
have greater ESG disclosure. Hence, we expect this instrument to be highly correlated with ESG disclosure, thus satisfying the rel-
evance requirement. The evidence regarding the impact of institutional ownership on firm value is mixed (e.g., Bhattacharya &
Graham, 2009; Elyasiani & Jia, 2010; Ruiz-Mallorqui & Santana-Martin, 2011). Therefore, we cannot assert ex ante whether our
second instrument is likely to meet the exogeneity condition. In this instance, we rely on the results of the postestimation
tests to assess the appropriateness of the instrument.

We reestimate all our 2SLS/IV models using our alternative set of instrumental variables (not tabulated). The postestimation
tests indicate that our model does not suffer from under-, weak, or overidentification with this set of instrumental variables.
Most importantly, the results from these regressions are consistent with the estimates reported in previous sections. That is,
once again, we find that ESG strengths increase firm value and ESG concerns decrease it. Furthermore, ESG disclosure per se de-
creases firm value, and in the presence of ESG concerns, higher ESG disclosure increases firm value. In the presence of ESG
strengths, higher ESG disclosure decreases firm value.

As a final test of robustness, we exclude industries and fiscal years one-by-one from our analyses. This does not lead to any
changes in any of our inferences, allowing us to conclude that our results are not driven by specific industries or time periods.

6. Conclusion

Our findings indicate that ESG strengths increase firm value and that ESG concerns decrease it. When isolated, ESG disclosure is also
found to decrease firm value. A more nuanced picture emerges once disclosure is interacted with ESG strengths or weaknesses. In the
presence of ESG strengths, high ESG disclosure weakens the positive valuation effect of the strengths. A possible explanation for this find-
ing is that the markets may interpret stepped-up disclosure as the firm’s attempt to justify an overinvestment in ESG activities. Disclosure
also weakens the negative valuation effects of ESG concerns, perhaps either because disclosures help firms legitimate their behavior by
explaining to investors the appropriateness of their operations and their ESG policies or because firms convince investors that they have
made credible commitments to change their operations and thus overcome ESG weaknesses.

When we repeat our analyses separately for ESG strengths and concerns, the results confirm the results for the general model. We
next estimate models for each of the three components that make up the firm’s ESG score: the environmental score, the social score,
and the governance score. These results indicate that environmental strengths increase the firm’s valuation and that weaknesses decrease
it; in both cases, disclosure wields a moderating influence. A slightly different conclusion obtains for social and governance factors: while
weaknesses in both areas again tend to decrease valuation, neither social nor governance strengths increase it. The valuation discounts
associated with social and governance weaknesses are again mitigated through the related disclosures.

Finally, the examinations at the level of the individual components reveal that investors discriminate strongly among the dif-
ferent dimensions of the ESG scores. Governance concerns lead to much steeper valuation discounts than social concerns or en-
vironmental concerns (in that order). At the same time, the moderating effects of governance-related disclosure are also much
stronger than those related to social or environmental concerns. We explain these effects in terms of differences in opacity. Gov-
ernance-related disclosures are often mandated and regulated by the SEC, and investors can assess their veracity with relatively
high ease and confidence. Disclosures related to social and environmental concerns, on the other hand, are mostly voluntary and
are therefore more opaque and more difficult to verify.

Acknowledgements

The authors are grateful to the journal’s guest editor and an anonymous referee for their invaluable suggestions and comments
on earlier drafts of this paper. The first author gratefully acknowledges the support received from the Development Bank of Japan
through its Shimomura Fellowship. However, this research did not receive any specific grant from funding agencies in the public,
commercial, or not-for-profit sectors.

59A. Fatemi et al. / Global Finance Journal 38 (2018) 45–64

Appendix A. Variable definitions and data sources

Variable Description Source

TOBINQ Tobin’s q [(book value of assets − book value of equity − deferred taxes + market value of equity) / book value of assets]
(measured as in Servaes & Tamayo, 2013)

Datastream

DISC ESG disclosure score Bloomberg
CSRCOMM Existence of corporate social responsibility (CSR) committee (dichotomous variable; CSRCOMM = 1 if company has CSR

committee, CSRCOMM = 0 otherwise)
Datastream

DISP Standard deviation of analyst earnings forecast (6-month earnings forecast) I/B/E/S
OWNERCONC Ownership of single largest owner (in %) Datastream
STR ESG strengths KLD
CON ESG concerns KLD
ROA Return on assets Datastream
ROAGROWTH Growth of return on assets (in %) Datastream
LNSALES Natural logarithm of sales Datastream
ASSETSSALES Asset intensity (assets / sales) Datastream
LEV Leverage (total debt to market value of equity) Datastream
ADVERT Advertising intensity (advertising expenses / sales) Datastream
ADVERTMISSING Advertising intensity missing (dichotomous variable; ADVERTMISSING = 1 if information on advertising intensity is

missing, ADVERTMISSING = 0 otherwise)
Datastream

R&D Research and development (R&D) intensity (research and development expenses / sales) Datastream
R&DMISSING Research and development (R&D) intensity missing (dichotomous variable; R&DMISSING = 1 if information on R&D

intensity is missing, R&DMISSING = 0 otherwise)
Datastream

NETGROSSPPE Net to gross property, plant, and equipment Datastream

Appendix B. OLS regression results

TOBINQ

DISC −0.0200***
(−3.54)
(0.000)

STR 0.0379*
(1.71)
(0.089)

STR*DISC −0.0019***
(−2.95)
(0.003)

CON −0.0434*
(−1.94)
(0.053)

CON*DISC 0.0017**
(2.33)
(0.020)

ROA 4.7548***
(6.22)
(0.000)

ROAGROWTH −0.0284**
(−2.49)
(0.013)

LNSALES −0.1192***
(−2.88)
(0.004)

ASSETSSALES −0.0676***
(−4.08)
(0.000)

LEV −0.1925***
(−4.37)
(0.000)

ADVERT 0.4521
(0.82)
(0.415)

ADVERTMISSING −0.1543**
(−2.02)

(continued on next page)

(continued)

TOBINQ

(0.044)
R&D 2.5680***

(3.11)
(0.002)

R&DMISSING −0.1275
(−1.47)
(0.143)

NETGROSSPPE 0.1163
(0.55)
(0.580)

CONST 3.2284***
(9.82)
(0.000)

INDUSTRY Included
YEAR Included
F-stat 16.70***
Adj. R2 41.31%
N 1640

Note: *** (**, *) denotes significance at the 1% (5%, 10%) level (two-sided test). t-statistics and p-
values are given in parentheses. Variables are defined as in Appendix A.
The table presents results from an OLS estimation of the following regression equation: TOBINQi,t =α0 +
β1DISCi,t + β2STRi,t + β3STR*DISCi,t + β4CONi,t + β5CON*DISCi,t + Controls.

60 A. Fatemi et al. / Global Finance Journal 38 (2018) 45–64

Appendix C. SEM results

DISC STR*DISC CON*DISC TOBINQ

DISC −0.0193**
(−3.62)
(0.000)

CSRCOMM 3.6788***
(4.52)
(0.000)

DISP 1.2919
(0.69)
(0.489)

OWNERCONC −0.0850***
(−3.12)
(0.002)

STR 1.6514***
(8.08)
(0.000)

31.2016***
(10.93)
(0.000)

0.0373*
(1.74)
(0.082)

STR*DISC −0.0019***
(−2.96)
(0.003)

STR*CSRCOMM 6.7030***
(3.85)
(0.000)

STR*DISP 4.1780
(1.00)
(0.317)

STR*OWNERCONC −0.1814
(−1.64)
(0.101)

CON 0.0228
(0.11)
(0.916)

23.3024***
(12.39)
(0.000)

−0.0433**
(−2.03)
(0.043)

CON*DISC 0.0017**
(2.21)
(0.027)

CON*CSRCOMM 6.2544***
(4.24)
(0.000)

CON*DISP 3.5278
(0.88)

(continued)

DISC STR*DISC CON*DISC TOBINQ

(0.379)
CON*OWNERCONC −0.1283

(−1.57)
(0.117)

ROA 4.4486
(0.94)
(0.349)

57.0305*
(1.96)
(0.050)

37.1480
(1.38)
(0.166)

5.0127***
(6.30)
(0.000)

ROAGROWTH −0.0587
(−0.52)
(0.606)

−0.8847
(−1.43)
(0.152)

−0.8363
(−1.43)
(0.153)

−0.0327***
(−2.71)
(0.007)

LNSALES 1.6425***
(3.10)
(0.002)

−0.0293
(−0.01)
(0.991)

13.0150***
(5.83)
(0.000)

−0.1225***
(−3.19)
(0.001)

ASSETSSALES −0.1449
(−1.00)
(0.318)

−0.4211
(−0.45)
(0.652)

1.6941*
(1.92)
(0.055)

−0.0715***
(−4.56)
(0.000)

LEV −0.2955
(−0.57)
(0.568)

−1.9197
(−0.70)
(0.487)

−4.1877
(−1.17)
(0.240)

−0.1912***
(−4.37)
(0.009)

ADVERT −11.4772
(−1.55)
(0.120)

−60.8823
(−1.35)
(0.178)

−44.4276
(−1.37)
(0.171)

0.4679
(0.85)
(0.396)

ADVERTMISSING −1.6138
(−1.39)
(0.164)

−10.5294
(−1.52)
(0.129)

−11.0483*
(−1.76)
(0.078)

−0.1559**
(−2.06)
(0.039)

R&D 9.9365
(0.75)
(0.453)

7.1685
(0.15)
(0.882)

80.0841**
(2.19)
(0.029)

2.5320***
(3.13)
(0.002)

R&DMISSING −0.7781
(−0.51)
(0.614)

2.9738
(0.33)
(0.741)

−2.8015
(−0.26)
(0.792)

−0.1250
(−1.43)
(0.152)

NETGROSSPPE 4.0028
(1.40)
(0.160)

16.2611
(1.13)
(0.260)

4.6781
(0.32)
(0.752)

0.1057
(0.51)
(0.612)

CONST 1.5213
(0.29)
(0.774)

−21.8518
(−0.87)
(0.382)

−14.8339***
(−4.83)
(0.000)

3.2690***
(9.61)
(0.000)

INDUSTRY Included Included Included Included
YEAR Included Included Included Included
Wald test for equations χ2 392.15*** 1100.60*** 568.76*** 436.31***
Adj. R2 43.83% 81.84% 72.04% 44.78%
Root mean squared error of approximation (RMSEA) 0.140
Comparative fit index (CFI) 0.860
Standardized root mean squared residual (SRMR) 0.021
N 1640 1640 1640 1640

Note: *** (**, *) denotes significance at the 1% (5%, 10%) level (two-sided test). z-statistics and p-values are given in parentheses. Variables are defined as in
Appendix A.

61A. Fatemi et al. / Global Finance Journal 38 (2018) 45–64

The table presents results from SEM estimation. Columns 1 to 3 present the estimation results for the three regressions for the
three endogenous regressors, i.e., ESG disclosure (DISC), the interaction between ESG strengths and ESG disclosure (STR*DISC),
and the interaction between ESG concerns and ESG disclosure (CON*DISC). The existence of a CSR committee (CSRCOMM), ana-
lyst forecast dispersion (DISP), and ownership concentration (OWNERCONC) are used to explain the endogenous regressor ESG
disclosure (DISC); the interactions between these three variables and the ESG strengths (ESG concerns) are used to explain the
endogenous regressor STR*DISC (CON*DISC). Column 4 presents the estimation results for the regression for the dependent var-
iable, i.e., firm value (TOBINQ): TOBINQi,t = α0 + β1DISCi,t + β2STRi,t + β3STR*DISCi,t + β4CONi,t + β5CON*DISCi,t + Controls,
where DISC is explained by CSRCOMM, DISP, and OWNERCONC; STR*DISC is explained by STR*CSRCOMM, STR*DISP, and
STR*OWNERCONC; and CON*DISC is explained by CON*CSRCOMM, CON*DISP, and CON*OWNERCONC. In contrast to 2SLS/IV es-
timation, not all instruments are included as regressors in all regressions for the three endogenous regressors because SEM esti-
mation allows the designation of explanatory variables to specific endogenous variables.

References

Acharya, V. V., DeMarzo, P., & Kremer, I. (2011). Endogenous information flows and the clustering of announcements. American Economic Review, 101, 2955–2979.
Aerts, W., Cormier, D., & Magnan, M. (2008). Corporate environmental disclosure, financial markets and the media: An international perspective. Ecological Economics,

64, 643–659.

62 A. Fatemi et al. / Global Finance Journal 38 (2018) 45–64

Akerlof, G. A. (1970). The market for lemons: Quality uncertainty and the market mechanism. Quarterly Journal of Economics, 84, 488–500.
Aktas, N., De Bodt, E., & Cousin, J. -G. (2011). Do financial markets care about SRI? Evidence from mergers and acquisitions. Journal of Banking & Finance, 35, 1753–1761.
Albuquerque, R. A., Durnev, A., & Koskinen, Y. (2015). Corporate social responsibility and firm risk: Theory and empirical evidence. (Working paper) Boston University.
Alexander, G. J., & Buchholz, R. A. (1978). Corporate social responsibility and stock market performance. Academy of Management Journal, 21, 479–486.
Al-Tuwaijri, S. A., Christensen, T. E., & Hughes, K. E., II (2004). The relations among environmental disclosure, environmental performance, and economic performance:

A simultaneous equations approach. Accounting, Organizations and Society, 29, 447–471.
Anderson, R. C., & Reeb, D. M. (2003). Founding-family ownership and firm performance: Evidence from the S&P 500. Journal of Finance, 58, 1301–1328.
Anderson, R. C., & Reeb, D. M. (2004). Board composition: Balancing family influence in S&P 500 firms. Administrative Science Quarterly, 49, 209–237.
Anderson, T. W., & Rubin, H. (1949). Estimation of the parameters of a single equation in a complete system of stochastic equations. Annals of Mathematical Statistics,

20, 46–63.
Anderson, E. W., Ghysels, E., & Juergens, J. L. (2005). Do heterogeneous beliefs matter for asset pricing? Review of Financial Studies, 18, 875–924.
Angrist, J. D., & Pischke, J. -S. (2009). Mostly harmless econometrics: An empiricist’s companion. Princeton, NJ: Princeton University Press.
Aras, G., & Crowther, D. (2008). Evaluating sustainability: A need for standards. Issues in Social and Environmental Accounting, 2, 19–35.
Aupperle, K. E., Carroll, A. B., & Hatfield, J. D. (1985). An empirical examination of the relationship between corporate social responsibility and profitability. Academy of

Management Journal, 28, 446–463.
Bajic, S., & Yurtoglu, B. B. (2017). CSR, market value, and profitability: International evidence. In Sabri Boubaker, Douglas Cumming, & Duc Khuong Nguyen (Eds.),

Handbook of Finance and Sustainability, Edward Elgar (forthcoming).
Barron, O. E., Kile, C. O., & O’Keefe, T. B. (1999). MD&A quality as measured by the SEC and analysts’ earnings forecasts. Contemporary Accounting Research, 16, 75–109.
Barron, O. E., Stanford, M. H., & Yu, Y. (2009). Further evidence on the relation between analysts’ forecast dispersion and stock returns. Contemporary Accounting

Research, 26, 329–357.
Baum, C. F. (2006). An introduction to modern econometrics using Stata. College Station, TX: Stata Press.
Beekes, W., Brown, P., Zhan, W., & Zhang, Q. (2016). Corporate governance, companies’ disclosure practices, and market transparency: A cross-country study. Journal of

Business Finance and Accounting, 43, 263–297.
Bhattacharya, P. S., & Graham, M. A. (2009). On institutional ownership and firm performance: A disaggregated view. Journal of Multinational Financial Management, 19,

370–394.
Bhattacharya, C. B., Sen, S., & Korschun, D. (2008). Using corporate social responsibility to win the war for talent. MIT Sloan Management Review, 49, 37–44.
Boehe, D., & Cruz, L. B. (2010). Corporate social responsibility, product differentiation strategy and export performance. Journal of Business Ethics, 91, 325–346.
Boone, A. L., & White, J. T. (2015). The effect of institutional ownership on firm transparency and information production. Journal of Financial Economics, 117, 508–533.
Bouten, L., Everaert, P., & Roberts, R. W. (2012). How a two-step approach discloses different determinants of voluntary social and environmental reporting. Journal of

Business Finance and Accounting, 39, 567–605.
Boyle, E. D., Higgins, M. M., & Ghon Rhee, S. (1997). Stock market reaction to ethical initiatives of defense contractors: Theory and evidence. Critical Perspectives on

Accounting, 8, 541–561.
Brammer, S., & Pavelin, S. (2006). Voluntary environmental disclosures by large UK companies. Journal of Business Finance and Accounting, 33, 1168–1188.
Brammer, S., & Pavelin, S. (2008). Factors influencing the quality of corporate environmental disclosure. Business Strategy and the Environment, 17, 120–136.
Brammer, S., Brooks, C., & Pavelin, S. (2006). Corporate social performance and stock returns: UK evidence from disaggregate measures. Financial Management, 35,

97–116.
Branco, M., & Rodrigues, L. (2006). Corporate social responsibility and resource-based perspectives. Journal of Business Ethics, 69, 111–132.
Brennan, M. J., Chordia, T., & Subrahmanyam, A. (1998). Alternative factor specifications, security characteristics, and the cross-section of expected returns. Journal of

Financial Economics, 49, 345–373.
Brown, N., & Deegan, C. (1998). The public disclosure of environmental performance information—A dual test of media agenda setting theory and legitimacy theory.

Accounting and Business Research, 29, 21–41.
Bushee, B. J., & Noe, C. F. (2000). Corporate disclosure practices, institutional investors and stock return volatility. Journal of Accounting Research, 38, 171–202

(Supplement).
Cahan, S. F., Chen, C., Chen, L., & Nguyen, N. H. (2015). Corporate social responsibility and media coverage. Journal of Banking & Finance, 59, 409–422.
Cameron, A. C., & Trivedi, P. K. (2010). Microeconometrics using Stata. College Station, TX: Stata Press.
Campbell, D., Craven, B., & Shrives, P. (2003). Voluntary social reporting in three FTSE sectors: A comment on perception and legitimacy. Accounting, Auditing &

Accountability Journal, 16, 558–581.
Cheng, B., Ioannou, I., & Serafeim, G. (2014). Corporate social responsibility and access to finance. Strategic Management Journal, 35, 1–23.
Chenhall, R. H., & Moers, F. (2007). The issue of endogeneity within theory-based, quantitative management accounting research. The European Accounting Review, 16,

173–196.
Chernozhukov, V., & Hansen, C. (2008). The reduced form: A simple approach to inference with weak instruments. Economics Letters, 100, 68–71.
Cho, C. H., & Patten, D. M. (2007). The role of environmental disclosures as tools of legitimacy: A research note. Accounting, Organizations and Society, 32,

639–647.
Cho, C. H., Roberts, R. W., & Patten, D. M. (2010). The language of US corporate environmental disclosure. Accounting, Organizations and Society, 35, 431–443.
Cho, C. H., Laine, M., Roberts, R. W., & Rodrigue, M. (2015a). Organized hypocrisy, organizational façades, and sustainability reporting. Accounting, Organizations and

Society, 40, 78–94.
Cho, C. H., Michelon, G., Patten, D. M., & Roberts, R. W. (2015b). CSR disclosure: The more things change…? Accounting, Auditing & Accountability Journal, 28, 14–35.
Clark, G. L., & Viehs, M. (2014). The implications of corporate social responsibility for investors: An overview and evaluation of the existing CSR literature. (Working paper)

University of Oxford.
Clarkson, P. M., Li, Y., Richardson, G. D., & Vasvari, F. P. (2008). Revisiting the relation between environmental performance and environmental disclosure: An empirical

analysis. Accounting, Organizations and Society, 33, 303–327.
Clarkson, P. M., Li, Y., Richardson, G. D., & Vasvari, F. P. (2011). Does it really pay to be green? Determinants and consequences of proactive environmental strategies.

Journal of Accounting and Public Policy, 30, 122–144.
Clarkson, P. M., Fang, X., Li, Y., & Richardson, G. D. (2013). The relevance of environmental disclosures: Are such disclosures incrementally informative? Journal of

Accounting and Public Policy, 32, 410–431.
Cornett, M. M., Erhemjamts, O., & Tehranian, H. (2016). Greed or good deeds: An examination of the relation between corporate social responsibility and the financial

performance of U.S. commercial banks around the financial crisis. Journal of Banking & Finance, 70, 137–159.
Deegan, C. (2002). Introduction—the legitimizing effect of social and environmental disclosures—a theoretical foundation. Accounting, Auditing & Accountability Journal,

15, 282–311.
Dhaliwal, D. S., Li, O. Z., Tsang, A., & Yang, Y. G. (2011). Voluntary nonfinancial disclosure and the cost of equity capital: The initiation of corporate social responsibility

reporting. The Accounting Review, 86, 59–100.
Dhaliwal, D. S., Radhakrishnan, S., Tsang, A., & Yang, Y. G. (2012). Nonfinancial disclosure and analyst forecast accuracy: International evidence on corporate social re-

sponsibility disclosure. The Accounting Review, 87, 723–759.
Dhaliwal, D. S., Li, O. Z., Tsang, A., & Yang, Y. G. (2014). Corporate social responsibility disclosure and the cost of equity capital: The roles of stakeholder orientation and

financial transparency. Journal of Accounting and Public Policy, 33, 328–355.
Diether, K. B., Malloy, C. J., & Scherbina, A. (2002). Differences of opinion and the cross section of stock returns. Journal of Finance, 57, 2113–2141.
Dimson, E., Karakas, O., & Li, X. (2015). Active ownership. Review of Financial Studies, 28, 3225–3268.
Doukas, J. A., Kim, C., & Pantzalis, C. (2006). Divergence of opinion and equity returns under different states of earnings expectations. Journal of Financial Markets, 9,

310–331.
Dye, R. A. (1985). Disclosure of nonproprietary information. Journal of Accounting Research, 23, 123–145.

63A. Fatemi et al. / Global Finance Journal 38 (2018) 45–64

Eberle, D., Berens, G., & Li, T. (2013). The impact of interactive corporate social responsibility communication on corporate reputation. Journal of Business Ethics, 118,
731–746.

Eccles, R. G., Krzus, M., Rogers, J., & Serafeim, G. (2012). The need for sector-specific materiality and sustainability reporting standards. Journal of Applied Corporate
Finance, 24, 8–14.

Eccles, R. G., Ioannou, I., & Serafeim, G. (2014). The impact of corporate sustainability on organizational processes and performance. Management Science, 60,
2835–2857.

Edmans, A. (2011). Does the stock market fully value intangibles? Employee satisfaction and equity prices. Journal of Financial Economics, 101, 621–640.
El Ghoul, S., Guedhami, O., & Kim, Y. (2015). Country-level institutions, firm value and the role of corporate social responsibility initiatives. (Working paper) University of

Alberta.
Elyasiani, E., & Jia, J. (2010). Distribution of institutional ownership and corporate firm performance. Journal of Banking & Finance, 34, 606–620.
Eugster, F., & Wagner, A. F. (2015). Value reporting quality, operating performance, and stock market valuations. (Working paper) University of Zurich.
Fama, E. F., & French, K. R. (1997). Industry costs of equity. Journal of Financial Economics, 43, 153–193.
Fatemi, A. M., & Fooladi, I. J. (2013). Sustainable finance: A new paradigm. Global Finance Journal, 24, 101–113.
Fatemi, A. M., Fooladi, I. J., & Tehranian, H. (2015). Valuation effects of corporate social responsibility. Journal of Banking & Finance, 59, 182–192.
Fisher-Vanden, K., & Thorburn, K. S. (2011). Voluntary corporate environmental initiatives and shareholder wealth. Journal of Environmental Economics and

Management, 62, 430–445.
Flammer, C. (2015). Does product market competition foster corporate social responsibility? Evidence from trade liberalization. Strategic Management Journal, 36,

1469–1485.
Freedman, M., & Wasley, C. (1990). The association between environmental performance and environmental disclosure in annual reports and 10 Ks. Advances in Public

Interest Accounting, 3, 183–193.
Freeman, R. E. (1984). Stakeholder management. A strategic approach. Marchfield, MA: Pitman.
Friedman, M. (1970). The social responsibility of business is to increase its profits. New York Times Magazine (September 13, reprinted from (1962)).
Gamerschlag, R., Möller, K., & Verbeeten, F. (2011). Determinants of voluntary CSR disclosure: Empirical evidence from Germany. Review of Managerial Science, 5,

233–262.
Gao, F., Dong, Y., Ni, C., & Fu, R. (2016). Determinants and economic consequences of nonfinancial disclosure quality. The European Accounting Review, 25, 287–317.
Garcia-Meca, E., & Sanchez-Ballesta, J. P. (2010). The association of board independence and ownership concentration with voluntary disclosure: A meta-analysis. The

European Accounting Review, 19, 603–627.
Ge, W., & Liu, M. (2015). Corporate social responsibility and the cost of corporate bonds. Journal of Accounting and Public Policy, 34, 597–624.
Gelb, D. S., & Strawser, J. A. (2001). Corporate social responsibility and financial disclosures: An alternative explanation for increased disclosure. Journal of Business

Ethics, 33, 1–13.
Gray, R., Kouhy, R., & Lavers, S. (1995). Corporate social and environmental reporting: A review of the literature and a longitudinal study of UK disclosure. Accounting,

Auditing & Accountability Journal, 8, 47–77.
Greening, D. W., & Turban, D. B. (2000). Corporate social performance as a competitive advantage in attracting a quality workforce. Business & Society, 39, 254–280.
Hail, L. (2002). The impact of voluntary corporate disclosures on the ex-ante cost of capital for Swiss firms. The European Accounting Review, 11, 741–773.
Hansen, L. (1982). Large sample properties of generalized method of moments estimators. Econometrica, 50, 1029–1054.
Harjoto, M., & Jo, H. (2015). Legal vs. normative CSR: Differential impact on analyst dispersion, stock return volatility, cost of capital, and firm value. Journal of Business

Ethics, 128, 1–20.
Ho, L. -C. J., & Taylor, M. E. (2007). An empirical analysis of triple bottom-line reporting and its determinants: Evidence from the United States and Japan. Journal of

International Financial Management and Accounting, 18, 123–150.
Holder-Webb, L., Cohen, J., Nath, L., & Wood, D. (2008). A survey of governance disclosures among U.S. firms. Journal of Business Ethics, 83, 543–563.
Holder-Webb, L., Cohen, J., Nath, L., & Wood, D. (2009). The supply of corporate social responsibility disclosures among U.S. firms. Journal of Business Ethics, 84,

497–527.
Hope, O. -K. (2003). Accounting policy disclosures and analysts’ forecasts. Contemporary Accounting Research, 20, 295–321.
Horvathova, E. (2010). Does environmental performance affect financial performance? A meta-analysis. Ecological Economics, 70, 52–59.
Hsu, K. (2012). The advertising effects of corporate social responsibility on corporate reputation and brand equity: Evidence from the life insurance industry in Taiwan.

Journal of Business Ethics, 109, 189–201.
Hughes, S. B., Anderson, A., & Golden, S. (2001). Corporate environmental disclosures: Are they useful in determining environmental performance? Journal of

Accounting and Public Policy, 20, 217–240.
Hwang, C. -Y., & Li, Y. (2008). Analysts’ incentives and the dispersion effect. (Working paper) Nanyang Technical University.
Ingram, R. W., & Frazier, K. B. (1980). Environmental performance and corporate disclosure. Journal of Accounting Research, 18, 614–622.
International Integrated Reporting Council (IIRC) (2013). The International b IR N Framework. (London).
International Integrated Reporting Council (IIRC) (2014). bIR N Yearbook 2014. (London).
Ioannou, I., & Serafeim, G. (2016). The consequences of mandatory corporate sustainability reporting: Evidence from four countries. (Working paper) Harvard University.
Jacobs, B. W., Singhal, V. R., & Subramanian, R. (2010). An empirical investigation of environmental performance and the market value of the firm. Journal of Operations

Management, 28, 430–441.
Jennings, R., & Marques, A. (2011). The joint effects of corporate governance and regulation on the disclosure of manager-adjusted non-GAAP earnings in the US.

Journal of Business Finance and Accounting, 38, 364–394.
Jensen, M. C. (2002). Value maximization, stakeholder theory, and the corporate objective function. Business Ethics Quarterly, 12, 235–256.
Jiao, Y. (2010). Stakeholder welfare and firm value. Journal of Banking & Finance, 34, 2549–2561.
Jo, H., & Harjoto, M. (2011). Corporate governance and firm value: The impact of corporate social responsibility. Journal of Business Ethics, 103, 351–383.
Johnson, T. C. (2004). Forecast dispersion and the cross section of expected returns. Journal of Finance, 59, 1957–1978.
Jones, T. M. (1995). Instrumental stakeholder theory: A synthesis of ethics and economics. Academy of Management Review, 20, 404–437.
Jorgensen, B. N., & Kirschenheiter, M. T. (2012). Interactive discretionary disclosures. Contemporary Accounting Research, 29, 382–397.
Khan, M., Serafeim, G., & Yoon, A. (2016). Corporate sustainability: First evidence on materiality. The Accounting Review, 91, 1697–1724.
Kim, E. -H., & Lyon, T. P. (2015). Greenwash vs. brownwash: Exaggeration and undue modesty in corporate sustainability disclosure. Organization Science, 26, 705–723.
Kleibergen, F., & Paap, R. (2006). Generalized reduced rank tests using the singular value decomposition. Journal of Econometrics, 133, 97–126.
KPMG (2011). KPMG international survey of corporate responsibility reporting 2011.
Krüger, P. (2015). Corporate goodness and shareholder wealth. Journal of Financial Economics, 115, 304–329.
Lang, M., & Lundholm, R. (1996). Corporate disclosure policy and analyst behavior. The Accounting Review, 71, 467–492.
Lang, M., & Stice-Lawrence, L. (2015). Textual analysis and international financial reporting: Large sample evidence. Journal of Accounting and Economics, 60,

110–135.
Larcker, D. F., & Rusticus, T. O. (2010). On the use of instrumental variables in accounting research. Journal of Accounting and Economics, 49, 186–205.
Li, M. -Y., & Wu, J. -S. (2014). Analysts’ forecast dispersion and stock returns: A quantile regression approach. Journal of Behavioral Finance, 15, 175–183.
Li, Q., Luo, W., Wang, Y., & Wu, L. (2013). Firm performance, corporate ownership, and corporate social responsibility disclosure in China. Business Ethics: A European

Review, 22, 159–173.
Liao, L., Luo, L., & Tang, Q. (2015). Gender diversity, board independence, environmental committee and greenhouse gas disclosure. The British Accounting Review, 47,

409–424.
Lyon, T. P., & Maxwell, J. W. (2011). Greenwash: Corporate environmental disclosure under threat of audit. Journal of Economics and Management Strategy, 20, 3–41.
Lyon, T., Lu, Y., Shi, X., & Yin, Q. (2013). How do investors respond to green company awards in China? Ecological Economics, 94, 1–8.
Malik, M. (2015). Value-enhancing capabilities of CSR: A brief review of contemporary literature. Journal of Business Ethics, 127, 419–438.

64 A. Fatemi et al. / Global Finance Journal 38 (2018) 45–64

Margolis, J. D., Elfenbein, H. A., & Walsh, J. P. (2009). Does it pay to be good and does it matter? A meta-analysis of the relationship between corporate social and financial
performance. (Working paper) Harvard University.

McWilliams, A., & Siegel, D. (2000). Corporate social responsibility and financial performance: Correlation or misspecification? Strategic Management Journal, 21,
603–609.

McWilliams, A., & Siegel, D. (2001). Corporate social responsibility: A theory of the firm perspective. Academy of Management Review, 26, 117–127.
Michelon, G., & Parbonetti, A. (2012). The effect of corporate governance on sustainability disclosure. Journal of Management and Governance, 16, 477–509.
Middleton, A. (2015). Value relevance of firms’ integral environmental performance: Evidence from Russia. Journal of Accounting and Public Policy, 34, 204–211.
Moskowitz, M. R. (1972). Choosing socially responsible stocks. Business and Society Review, 71–75.
Neiheisel, S. R. (1995). Corporate strategy and the politics of goodwill: A political analysis of corporate philanthropy in America. New York: Peter Lang.
Orlitzky, M., Schmidt, F. L., & Rynes, S. L. (2003). Corporate social and financial performance: A meta-analysis. Organization Studies, 24, 403–441.
Patten, M. (2002). The relation between environmental performance and environmental disclosure: A research note. Accounting, Organizations and Society, 27,

763–773.
Pérez, A., & del Bosque, I. R. (2015). Corporate social responsibility and customer loyalty: Exploring the role of identification, satisfaction and type of company. Journal

of Services Marketing, 29, 15–25.
Perrini, F., Rossi, G., & Rovetta, B. (2008). Does ownership structure affect performance? Evidence from the Italian market. Corporate Governance: An International

Review, 16, 312–325.
Peters, G. F., & Romi, A. M. (2014). Does the voluntary adoption of corporate governance mechanisms improve environmental risk disclosures? Evidence from green-

house gas emission accounting. Journal of Business Ethics, 125, 637–666.
Peters, G. F., & Romi, A. M. (2015). The association between sustainability governance characteristics and the assurance of corporate sustainability reports. Auditing: A

Journal of Practice & Theory, 34, 163–198.
Petersen, M. A. (2009). Estimating standard errors in finance panel data sets: Computing approaches. Review of Financial Studies, 22, 435–480.
Plumlee, M., Brown, D., Hayes, R. M., & Marshall, R. S. (2015). Voluntary environmental disclosure quality and firm value: Further evidence. Journal of Accounting and

Public Policy, 34, 336–361.
Porter, M. E. (1991). Towards a dynamic theory of strategy. Strategic Management Journal, 12, 95–117.
Porter, M. E., & Kramer, M. R. (2006). Strategy and society: The link between competitive advantage and corporate social responsibility. Harvard Business Review, 84,

78–92.
Porter, M. E., & Kramer, M. R. (2011). Creating shared value. Harvard Business Review, 89, 62–77.
Porter, M. E., & van der Linde, C. (1995). Toward a new conception of the environment-competitiveness relationship. Journal of Economic Perspectives, 9, 97–118.
Ramlugun, V. G., & Raboute, W. G. (2015). Do CSR practices of banks in Mauritius lead to satisfaction and loyalty? Studies in Business and Economics, 10, 128–144.
Reilly, A. H., & Hynan, K. A. (2014). Corporate communication, sustainability, and social media: It’s not easy (really) being green. Business Horizons, 57, 747–758.
Renneboog, L., Horst, J. T., & Zhang, C. (2008a). Socially responsible investments: Institutional aspects, performance, and investor behavior. Journal of Banking & Finance,

32, 1723–1742.
Renneboog, L., Horst, J. T., & Zhang, C. (2008b). The price of ethics and stakeholder governance: The performance of socially responsible mutual funds. Journal of

Corporate Finance, 14, 302–322.
Reuter, M., & Messner, M. (2015). Lobbying on the Integrated Reporting Framework: An analysis of comment letters to the 2011 discussion paper of the IIRC.

Accounting, Auditing & Accountability Journal, 28, 365–402.
Reverte, C. (2009). Determinants of corporate social responsibility disclosure ratings by Spanish listed firms. Journal of Business Ethics, 88, 351–366.
Roberts, J. (2004). The modern firm: Organizational design for performance and growth. Oxford: Oxford University Press.
Rogers, J. L., Schrand, C., & Zechman, S. L. C. (2014). Do managers tacitly collude to withhold industry-wide bad news? (Working paper) University of Chicago.
Ruiz-Mallorqui, M. V., & Santana-Martin, D. J. (2011). Dominant institutional owners and firm value. Journal of Banking & Finance, 35, 118–129.
Russo, M. V., & Fouts, P. A. (1997). A resource-based perspective on corporate environmental performance and profitability. Academy of Management Journal, 40,

534–559.
Saleh, M., Zulkifli, N., & Muhamad, R. (2010). Corporate social responsibility disclosure and its relation on institutional ownership. Managerial Auditing Journal, 25,

591–613.
Sen, S., & Bhattacharya, C. B. (2001). Does doing good always lead to doing better? Consumer reactions to corporate social responsibility. Journal of Marketing Research,

38, 225–243.
Serafeim, G. (2014). The role of the corporation in society: An alternative view and opportunities for future research. (Working paper) Harvard University.
Servaes, H., & Tamayo, A. (2013). The impact of corporate social responsibility on firm value: The role of customer awareness. Management Science, 59, 1045–1061.
Shea, J. (1997). Instrument relevance in multivariate linear models: A simple measure. The Review of Economics and Statistics, 79, 348–352.
Sraer, D., & Thesmar, D. (2007). Performance and behavior of family firms: Evidence from the French stock market. Journal of the European Economic Association, 5,

709–751.
Vance, S. G. (1975). Are socially responsible corporations good investment risks? Management Review, 64, 18–24.
Verrecchia, R. E. (1983). Discretionary disclosure. Journal of Accounting and Economics, 5, 179–194.
Vigneau, L., Humphreys, M., & Moon, J. (2015). How do firms comply with international sustainability standards? Processes and consequences of adopting the Global

Reporting Initiative. Journal of Business Ethics, 131, 469–486.
Villalonga, B., & Amit, R. (2006). How do family ownership, control and management affect firm value? Journal of Financial Economics, 80, 385–417.
de Villiers, C., & van Staden, C. J. (2011). Where firms choose to disclose voluntary environmental information. Journal of Accounting and Public Policy, 30, 504–525.
Walsh, G., & Bartikowski, B. (2013). Exploring corporate ability and social responsibility associations as antecedents of customer satisfaction cross-culturally. Journal of

Business Research, 66, 989–995.
Weiss, C., & Hilger, S. (2012). Ownership concentration beyond good and evil: Is there an effect on corporate performance? Journal of Management and Governance, 16,

727–752.
Welch, E. (2003). The relationship between ownership structure and performance in listed Australian companies. Australian Journal of Management, 28, 287–305.
White, H. (1980). A heteroskedasticity-consistent covariance matrix estimator and a direct test for heteroskedasticity. Econometrica, 48, 817–838.
Wiseman, J. (1982). An evaluation of environmental disclosures made in corporate annual reports. Accounting, Organizations and Society, 7, 53–63.
Wooldridge, J. M. (2002). Econometric analysis of cross section and panel data. Cambridge, MA: MIT Press.
Wright, P., & Ferris, S. P. (1997). Agency conflict and corporate strategy: The effect of divestment on corporate value. Strategic Management Journal, 18, 77–83.
Xie, Y. (2014). The effects of corporate ability and corporate social responsibility on winning customer support: An integrative examination of the roles of satisfaction,

trust and identification. Global Economic Review, 43, 73–92.

Journal of

Risk and Financial
Management

Article

ESG Reporting and Analysts’ Recommendations in GCC:
The Moderation Role of Royal Family Directors

Abdulsamad Alazzani 1,* , Wan Nordin Wan-Hussin 2, Michael Jones 3 and Ahmed Al-hadi 4

����������
�������

Citation: Alazzani, Abdulsamad,

Wan Nordin Wan-Hussin, Michael

Jones, and Ahmed Al-hadi. 2021. ESG

Reporting and Analysts’

Recommendations in GCC: The

Moderation Role of Royal Family

Directors. Journal of Risk and Financial

Management 14: 72. https://doi.org/

10.3390/jrfm14020072

Academic Editor: Khaled Hussainey

Received: 19 January 2021

Accepted: 29 January 2021

Published: 7 February 2021

Publisher’s Note: MDPI stays neutral

with regard to jurisdictional claims in

published maps and institutional affil-

iations.

Copyright: © 2021 by the authors.

Licensee MDPI, Basel, Switzerland.

This article is an open access article

distributed under the terms and

conditions of the Creative Commons

Attribution (CC BY) license (https://

creativecommons.org/licenses/by/

4.0/).

1 College of Business and Economics, Qatar University, Doha 2713, Qatar
2 Othman Yeop Abdullah Graduate School of Business, Universiti Utara Malaysia, Kedah 06010, Malaysia;

[email protected]
3 School of Economics, Finance and Management, University of Bristol, Bristol BS8 1TU, UK;

[email protected]
4 Faculty of Business and Law, School of Accounting, Curtin University, Perth U1987, Australia;

[email protected]
* Correspondence: [email protected] or [email protected]

Abstract: This study examines whether financial analysts consider or incorporate the environmental,
social and governance disclosures (thereafter ESG) in their recommendations. We then test whether
royal family directors affect this relation. Using a dataset from six Gulf Cooperation Council (GCC)
countries, we find evidence that analysts’ recommendations are influenced by ESG information. Fur-
ther, we find the political connection negatively moderates the relationship between sell-side analysts’
recommendations and ESG. This suggests that financial analysts may assess the ESG disclosure in
those firms with the political connection of royalty, in GCC countries, as superficial compliance rather
than a genuine commitment. Our results are robust when subjected to endogeneity tests.

Keywords: environmental; GCC; investment recommendation; royal family directors; social and gov-
ernance

JEL Classification: M48; N15; M41; Q58

1. Introduction

Recent anecdotal evidence on the link between society, environment and business
continues to highlight the extent to which analysts pay heed to environmental, social and
governance (ESG) engagements by firms. This is not only in developed markets but also in
emerging markets, such as those in the Gulf Cooperation Council (GCC) countries. The
purpose of this study is to examine whether financial analysts react to and consider ESG
disclosures in their recommendations in GCC countries. We then test whether the social and
cultural players such as royal family directors that are politically connected, a prominent
feature of GCC boards, affect this relationship (Halawi et al. 2008; Al-Hadi et al. 2016).

Financial analysts1 are important in the international business context, where infor-
mation asymmetry between management and investors is high (Fung et al. 2016). The
increasing interest by international investors in GCC equity markets, which provide tax
haven opportunities and a strong return in capital markets (Bley and Saad 2012), signi-
fies the important role played by financial analysts who gather and process information
about companies and distribute this information through their stock recommendations.
As knowledgeable experts who conduct research and provide intelligence on firms they
follow, their stock recommendations help investors to make investment decisions through

1 Commonly, there are two types of analysts, sell-side and buy-side. The sell-side analysts are following a list of companies and provide regular
research reports to the firm’s clients and to financial platforms such as Capital IQ S&P, Bloomberg, and IBES. They mostly work for brokerage firms.
Buy-side analysts are the analysts who provide service for the purpose of fund management. In this study we used sell-side analysts.

J. Risk Financial Manag. 2021, 14, 72. https://doi.org/10.3390/jrfm14020072 https://www.mdpi.com/journal/jrfm

J. Risk Financial Manag. 2021, 14, 72 2 of 20

interpreting complex information and converting it to simple buy, hold and sell recom-
mendations (Jegadeesh et al. 2004; Brown et al. 2015). They also play significant roles in
mitigating agency problems and information asymmetry and have the ability to assess
ESG reporting. In addition, the increasing numbers of fund managers and investors who
allocate their investments to those firms with better ESG have encouraged financial ana-
lysts to consider ESG when issuing investment recommendations. Moreover, many big
investment institutions have sustainability indices which are essential for such investors.
We assume and theorize that the financial analysts with such expertise will serve investors
by incorporating ESG into their valuation models and recommendations.

The value of the analysts’ stock recommendations comes from at least two sources.
First, analysts are skilled in analyzing and synthesizing both private and public information
from management and other sources to investors (Newton 2019). Second, they can gather a
wide range of information unavailable to the investors, integrate the diverse information,
and professionally assess the prospects of firm future cash flows (Ivković and Jegadeesh
2004). The summary judgment recommending “buy/hold/sell” is the investment opinion
that analysts disseminate to investors regarding whether a given stock is worth buying or
selling. In essence, the recommendation captures forward-looking information that helps
investors gauge future cash flows and firm value (Luo et al. 2010).

Ioannou and Serafeim (2015) argue that the corporate social responsibility (CSR) activ-
ities taken by a firm may affect analysts’ recommendations through the following channels.
First, CSR enhances value by improving a firm’s long-term financial performance. Changes
in financial performance, therefore, may have direct impacts on analysts’ recommendations.
Second, the substantial amount of funds invested by socially responsible and environ-
mentally conscious investors in CSR-friendly firms might positively affect the stock prices
of those firms, thus also affecting analysts’ recommendations. For instance, Dhaliwal
et al. (2011) find that companies with higher CSR disclosures are more likely to attract
institutional investors and analysts’ coverage.

The GCC is a political and economic union of six Arabic monarchical countries. It
was established in 1981 and includes Saudi Arabia, Qatar, Oman, Kuwait, Bahrain and
UAE. GCC countries have many similarities. They share the same religion, culture and
political systems. They are monarchies and they have high oil and gas reservations, which
makes them among the richest countries in the world. In recent decades, an influx of
foreign capital and international investment institutions has been observed. The GCC
region provides a high return on investments and it is considered a tax haven.

Further, GCC countries are characterized by royal families that are involved in dif-
ferent aspects: social life, business, state ruler ministers in many key ministries, owners,
and directors on company boards (Al-Hadi et al. 2016; Kamrava et al. 2016). We theorize
that the huge overlap of those royal families, in many power positions in general and
as members of boards of directors in particular, may have a significant influence on the
decisions process of these companies, especially with regard to ESG. The equity analysts
may perceive this appearance pessimistically or optimistically. Thus, it will be interesting
to study the impact of the royal family directors, especially in a unique characteristic area
that cannot be found in any other area.

In this study, we introduce seven emerging-frontier markets from the GCC countries
as a new unique setting and context to re-examine this association. In particular, this study
investigates whether in developing countries the participants that monitor and channel
the flow of ESG information in capital markets perceive and assess this as important
information informing earnings expectations and valuing securities.

While a growing body of research examines the ESG issues in both developed and
developing countries, the social-political environment in developing countries has been
ignored. Our study responds to the call for more research looking into the contextual
motivations and challenges specifically faced by developing countries (Islam and Deegan
2008; Belal et al. 2013; Tilt 2016). Our study also responds to the calls of the studies of

J. Risk Financial Manag. 2021, 14, 72 3 of 20

Ioannou and Serafeim (2015); Eccles et al. (2011); and Yu (2011) for more research into ESG
and analysts’ recommendations.

We focus on GCC countries because foreign investors have been attracted to GCC
capital markets, especially after the economic reforms (Balli et al. 2011; Al-Hadi et al.
2015). For example, the listed companies in GCC capital markets have increased from 399
to 4682 between 2000 and 2017. Furthermore, many foreign investors’ entrance barriers
have been removed. For instance, the “ease of doing business index” has increased in
Saudi Arabia, from 15% in 2008 to 96% in 20152; in Oman, from 60% in 2008 to 77% in 2014;
and in Bahrain, from 18% in 2008 to 66% in 2015. Therefore, it is reasonable to expect that
investors will look for information intermediaries in the global capital markets such as
analysts to help them assess corporate performance.

This paper contributes to the prior literature as follows. First, by answering two
key questions: (1) Do the financial analysts incorporate and consider ESG in their recom-
mendations? (2) What is the key role of royal family directors in the above-mentioned
relationship? Second, we enrich the scant literature on non-financial information disclo-
sure and its implications on the stock market players overall and in developing countries
(i.e., the GCC markets). Most of the securities in the GCC capital markets rarely receive
analyst coverage and analyst recommendations (Al-Ajmi and Kim 2012; Al-Hadi et al.
2015). This may create a shortage in information disclosure, which, in turn, leads to higher
uncertainty regarding firm-specific information and information asymmetry. It is crucial to
study whether such securities obtain benefits in terms of favorable recommendations from
their ESG disclosures. There are very limited studies that investigate the impact of ESG
disclosures on analysts’ behavior, thus, we provide new evidence from the GCC emerging
markets.

We provide new evidence that the relationship between ESG and analysts’ recommen-
dations is positive, and the existence of the royal family on the board negatively moderates
the relationship between ESG and analysts’ recommendations. These results demonstrate
that the politically connected firms of the royal family exacerbate the information asymme-
try in the firms. These findings assume that the practical implication of emerging and GCC
markets must consider political issues when making decisions.

The rest of this paper is organized as follows. Sections 2 and 3 presents the background
and GCC setting, literature review, and hypotheses development. In Sections 4 and 5, we
discuss the research methodology, and Section 6 highlights the key findings and results
discussion. We present the conclusion of our study in Section 7.

2. Literature Review
2.1. Social Responsibility Policy, Environment, and Regulation in the GCC

Corporate social responsibility (CSR)3 regulation in GCC countries is considered in
its infancy stage and still voluntary. There is no clear CSR regulation or policies in GCC
governments and stock markets. There are, however, some initiatives released in some
GCC countries. For instance, in 2008, the Saudi Arabian General Investment Authority
(SAGIA) launched the Responsible Competitiveness Index (SARC) to rank more socially
responsible firms. However, this is a voluntary adoption. Oman is the first country in the
GCC that required firms to adopt a corporate governance code in 2003 (Al-Hadi et al.
2016), and the Omani Capital Market Authority (CMA) in several releases urges all joint
stock companies and investment funds to adopt the Oman Social Responsibility Initiative,
launched by the CMA Oman on November 2, 20094. “The company shall seek to exercise
its role as good citizen and to mitigate any adverse impact of its activities on the national
economy, community or environment at large”. The new code requires a CSR charter or
code, a CSR strategy, and an annual report on CSR activities.

2 https://tradingeconomics.com/saudi-arabia/ease-of-doing-business.
3 ESG and CSR terms are used interchangeably in this study.
4 http://www.gulfbase.com/news/cma-urges-listed-companies-to-take-csr-initiatives/221828.

J. Risk Financial Manag. 2021, 14, 72 4 of 20

The Kingdom of Bahrain, in order to diversify its economy from being reliant only
on oil, has promoted itself as an international banking center and encourages foreign
investments. Many multi-national corporations, specifically banking corporations includ-
ing banks and corporations, are encouraged to practice CSR and to adhere strictly to the
regulations. The Kingdom of Bahrain believes that in order for CSR to be effective, it must
be controlled, regulated, and standardized by the government. This is because, by nature,
a corporation will always look after its own interests. Thus, the government established
the Bahraini Association for Social Responsibility in 2011.

In July 2008, Qatar launched its 2030 national vision. The 2030 vision rests on four
pillars: human development, social development, economic development, and environ-
mental development, which are the crux of corporate social responsibility. Emir Tamim,
Amir of Qatar has demanded hard work to help accomplish the CSR goals and vision and
to advance the nation’s development5. To achieve this vision, the Ministry of Economy
and Commerce (MEC) has launched a Qatari CSR index that takes into account similar
international experiments based on relevant United Nations (UN)standards. In the UAE,
there are also trends to make CSR disclosures by listed companies compulsory.

2.2. Overview of ESG Disclosure Theories

Numerous theories explain voluntary disclosure of ESG information (or sustainability
reporting) including stakeholder theory, impression management theory, institutional the-
ory, discretionary disclosure theory, and legitimacy theory. For, example, legitimacy theory
provides some explanation of why firms adopt CSR. It asserts companies’ behavior in im-
plementing and developing voluntary social and environmental disclosure of information
in order to fulfill their social contract. This enables the recognition of their objectives and
survival in a jumpy and turbulent environment. Suchman (1995) considers that “legitimacy
is a generalized perception or assumption that the actions of an entity are desirable, proper,
or appropriate within some socially constructed system of norms, values, beliefs, and
definitions”. Legitimacy theory suggests that firms legitimate themselves through various
actions, including communication with relevant stakeholders. Sell-side analysts are of key
importance for shareholders who really pay attention to firms’ annual reports and interpret
the financial and non-financial information. This information is seen to be essential input
in their research reports.

According to Schlenker (1980), impression management is the means by which to
influence individuals. The success of influencing others through impression management
depends on the audience’s positive or negative perception (Gardner and Martinko 1998).
We assume that the board members from the royal families are more likely to manage
impression. Financial analysts are one of the main audiences in the stock market. Part of
this impression may be through influencing social responsibility activities, which may be
met negatively or positively by analysts.

3. Hypotheses Development

The financial analysts employed by brokerage firms, so-called “sell-side analysts,”
are commonly considered to be experts in investment analysis and security valuation. A
large body of literature documents the significant role of security analysts as information
intermediaries in capital markets (Bradshaw 2004; Healy and Palepu 2001; Ioannou and
Serafeim 2015; Luo et al. 2015) and in influencing stock prices (Givoly and Lakonishok 1984;
Stickel 1995; Givoly and Lakonishok 1984; Stickel 1995). As mentioned earlier, sell-side
analysts issue recommendations on securities, which are typically phrased as buy, sell, or
hold (Ioannou and Serafeim 2015).

Previous literature also investigates several determinants of analyst recommendations.
Luo et al. (2015) find international evidence that analysts’ recommendations mediate the
relationship between CSR and firms’ stock returns by reducing the strength of the effect

5 See http://www.mdps.gov.qa/en/qnv/Documents/QNV2030_English_v2.pdf.

J. Risk Financial Manag. 2021, 14, 72 5 of 20

between them. They explain that, as skilled industry experts, security analysts are able
to obtain access to private information that is not readily accessible to general investors
and so are better able to assess the value relevance of a firm’s corporate social performance
information. Thus, the analysts’ recommendations act as an informational pathway through
which corporate social performance affects corporate financial performance. Yu (2011) finds
that analysts tend to issue favourable recommendations for companies with better corporate
governance mechanisms in the emerging markets. Dhaliwal et al. (2012) investigate the
relationship between CSR and analysts’ forecast accuracy in several countries. They find
that the issuance of stand-alone CSR reports is associated with lower analyst forecast errors.
This relationship is stronger in more stakeholder-oriented countries. Garrido-Miralles et al.
(2016) also find a negative relationship between the earnings forecast error and the issue of
sustainability reports in Spain. Dhaliwal et al. (2011) find that companies with superior
CSR attract dedicated institutional investor and analyst coverage.

On the other hand, Adhikari (2016) finds that firms with greater analyst coverage
have lower corporate social responsibility scores, consistent with the view that spending
on CSR is a manifestation of an agency problem. Thus, if it is an agency problem then
better monitoring due to greater analyst coverage should force managers to cut back on
CSR activities. Garcia-Sanchez et al. (2020) show that firms which adopt sustainable devel-
opment goals received, immediately in the same or the next year, sell recommendations
from analysts, and long CSR strategies have less impact on recommendations. Zhang
and Wei (2019) show that analysts’ recommendations are positively and optimistically
associated with the firms with less information disclosure. They also find that analysts
who have private information have stronger recommendations. Hinze and Sump (2019), in
their review, find that many existing literature confirm the positive relationship between
CSR performance and optimistic recommendations, and Wang and Jiang (2019) show that
analysts’ recommendations mediate the relationship between brand equity and sustainable
performance in Chinese listed firms.

The literature review and interviews with financial analysts suggest that analysts
increasingly incorporate ESG information in their recommendations. According to Eccles
et al. (2011), there is a large and growing market interest in ESG information and policies.
Luo et al. (2015) conduct in-depth interviews with sell-side analysts who acknowledge
that they heed CSR disclosures and incorporate them into their investment recommenda-
tions. Moreover, Fieseler (2011) interview financial analysts, and the results suggest that
responsibility issues are increasingly becoming part of mainstream investment analysis.

Thus, we state the following hypothesis:

Hypothesis 1 (H1). There is a positive relationship between analysts’ recommendations and ESG
disclosure.

The Arabian Gulf communities have tribal origins. The tribal elder/Shaikh plays a
prominent and important role in the administration of tribal affairs. This role remained
prominent as these countries became monarchies and is now administered by royal families.
Recently, many royal families have become involved in business (Kamrava et al. 2016).
Many GCC-listed companies now have at least one royal family member on their board of
directors (Halawi et al. 2008). The effect of royal family directors on ESG disclosure can be
seen in two ways. First, a positive effect of royal family directors on the board will enhance
social responsibility activities. Alazzani et al. (2019) mention that to enhance the regimes’
legitimacy, the royal families may become more involved in CSR activities as a way to
minimize popular frustration with the increased concentration of wealth in the hands of
a few. Princes/Shaikhs must actively contribute to support the communities in which
they live and show that they care for the members of their community. From a political
and reputational aspect, royal family members have motives to adopt legal and ethical
business practices and CSR, to show their social support through sympathetic activities
(e.g., Eid Al-Thani in Qatar, the King Faisal Foundation and the Alwaleed Philanthropies in
Saudi Arabia; the Khalifa Foundation in the UAE; and the Alsobah Foundation in Kuwait).

J. Risk Financial Manag. 2021, 14, 72 6 of 20

Drawing on servant leadership theory, Alazzani et al. (2019) investigate whether the
presence of royal family members on GCC boards of directors influences CSR reporting.
They find a positive relationship between the presence of royal family directors and CSR
reporting.

On the other hand, a negative impact of the royal family on CSR disclosures is
grounded on the argument that royal family directors will maximize their self-interest.
For instance, Al-Hadi et al. (2016) find that ruling/royal families enjoy less risk reporting
pressures and more government rent-seeking, and more earnings management.

As mentioned earlier in the theories section, impression management is the activity of
controlling or regulating information which influences the impression formed by audiences
(Schlenker 1980). The success of the impact on others by impression management depends
on the perception of the audience. This influence may be positive or negative (Gardner and
Martinko 1998). As powerful people, royal family directors in our case might be expected
to perform more impression management than others. One of the key audiences in the
stock markets is financial analysts. So, we assume that impression management can explain
why royal family directors might moderate the relationship between ESG and analysts’
recommendations.

The key question is how financial analysts evaluate such an impact. We want to see if
they perceive it as a positive or negative impact. Thus, we state the following hypothesis:

Hypothesis 2 (H2). The existence of a royal family director will affect the relationship between
analysts’ recommendations and ESG disclosure.

4. Research Methodology and Design
4.1. Sample Selection and Data Source

There were 4386 total observations from 2010 to 2016 for companies in the six GCC
countries and the seven stock markets. We used the Bloomberg dataset6 for collecting
the data. We eliminated 3849 companies’ results for one of four reasons: (a) investment
recommendations were not available, (b) ESG data were not available via Bloomberg,
(c) accounting data were not available through Bloomberg and (d) companies only were
present for one year. The final sample consists of 537 firm-year observations.

Table 1 Panel A shows that Saudi Arabia represents about 31% of the total observations
(169), followed by the United Arab Emirates with 140 observations (26%) from the two
Dubai and Abu Dhabi Stock Markets. The other four countries, Qatar, Bahrain, Oman, and
Kuwait, constituted 43% of the sample.

Table 1. Sample distribution based on country and year.

Country 2010 2011 2012 2013 2014 2015 2016 Total %

Saudi Arabia 11 23 26 26 27 28 28 169 0.31
United Arab Emirates 11 21 22 24 24 25 13 140 0.26

Qatar 4 10 11 13 17 18 12 85 0.16
Kuwait 7 9 10 10 11 11 4 62 0.12
Oman 5 7 8 9 9 9 9 56 0.1

Bahrain 1 3 5 5 5 5 1 25 0.05

Total 39 73 82 87 93 96 67 537 100

4.2. Empirical Models and Variable Definitions

The following baseline ordinary least squares (OLS) and regression models are used
to examine the effect of ESG on analysts’ stock recommendations:

6 Bloomberg is an online database providing current and historical financial quotes, business newswires, and descriptive information, research and
statistics on over 52,000 companies worldwide. Bloomberg computes the ESG Disclosure score to quantify a company’s transparency in reporting
ESG information.

J. Risk Financial Manag. 2021, 14, 72 7 of 20

REC = β0 + β1ESG + β2ROA + β3LEVERAGE + β4RI + β5BIDASK + β6GOV_OWN + β7GDP + β8EXANALYST
+ β9BOARD + β10SIZE + β11TOBIN + Year-effect + Industry-effect + Country-effect + ε

(1)

The moderation effects of the royal family model are presented as;

REC = β0 + β1ESG + β2RFP + β3ESG * RFP + β4ROA + β5LEVERAGE + β6RI + β7BIDASK + β8GOV_OWN + β9GDP +
β10EXANALYST + β11BOARD + β12SIZE + β13TOBIN + Year-effect + Industry-effect + Country-effect + ε

(2)

The following is an explanation for these variables:

Variable Definition

REC
Indicates the analysts’ opinions on the stock performance. The rating is calculated by converting each of the
analysts’ recommendations into a number from 1–5 and taking the average. Originally, this measure is
reported with 5 = strong buy, 4 = buy, 3 = hold, 2 = underperform, and 1 = sell.

ESG
Proprietary Bloomberg scores based on the extent of a company’s Environmental, Social, and Governance
(ESG) disclosure.

RFP Percentage of royal family directors on the board.
ESG*RFP The interaction of ESG with RFP.
ROA Return on Assets.
LEVERAGE Total debt divided by total shareholders’ equity.

RI
The total excess return is defined as the return for the company above the market return. It is calculated as ri,t+
1 − rm,t+1, where i is the firm, and m is the market.
The Bloomberg ticker: TOT_RETURN_INDEX_GROSS_DVDS.

GOV_OWN Percentage of government ownership.
GDP Gross domestic product growth rate.
EXANALYST Log of (the number of directors on the board with experience in financial analysis plus 1)7.
BOARD The number of directors on the board.

BIDASK
A bid-ask spread is the amount or percentage by which the ask price (THE LOWEST) exceeds the bid price
(the highest) for security in the stock market and reflects the liquidity of the stocks. It is also a reflection of the
supply and demand for a particular security.

SIZE Log of total market value of a company’s shares.
TOBIN Ratio of the market value of a firm to the replacement cost of the firm’s assets.
SOCIAL Bloomberg score based on the extent of a company’s Social disclosure.
ENV Bloomberg score based on the extent of a company’s Environmental disclosure.
GOV Bloomberg score which measures the quality of corporate governance.

We also included year, country, and industry effects. Nearly one-third of sample
firms are involved in environmentally sensitive industries. Moreover, we used White
(1982) robustness test to obtain unbiased standard errors of OLS coefficients under het-
eroscedasticity to ensure that our estimator has the lowest variance among all unbiased
estimators.

5. Discussion of the Variables Used
5.1. Dependent Variable: Analysts’ Recommendations

Bloomberg rates the analysts’ recommendations by converting each of the analysts’
recommendations into a number from 1–5 and taking the average. A score of 3 means that
analysts believe that the stock should be held, less than 3 means it should be sold, and
greater than 3 means it should be bought. The company receives many recommendations
during the year from many employees of a brokerage or fund management house who
study companies and make buy-and-sell recommendations on stocks of these companies.
Bloomberg calculates the average of these investment recommendations (REC) during the
year. We use this average as a measurement of the analysts’ recommendations.

7 Plus 1 added because Zero once is logged will be missing.

J. Risk Financial Manag. 2021, 14, 72 8 of 20

5.2. Independent Variable
5.2.1. ESG

Recently, rating firms (e.g., Bloomberg, KLD, Thomson Reuter ’s ASSET4) have played
an important role in assessing areas ranging from sustainability to corporate (governance?).
Managers, investors, and scientists are increasingly relying on these ratings for strategic
decisions, investing trillions of dollars in capital, and studying CSR, guided by the implicit
assumption that the ratings are valid. ESG data have the potential to provide crucial market
transparency and a unique lens through which to assess the future company and investment
performance. It is crucial to acknowledge that there is a divergence between different ESG
ratings and organizations. Moreover, every organization can rank a particular company
differently (Chatterji et al. 2016). In this study, we use Bloomberg ESG rating. Bloomberg
LP focuses on a future in which environmental and social issues will have increasingly
critical implications for firms and investors. Increasing demand for sustainability analytics
has become an essential part of any investment decision. Bloomberg began collecting ESG
data in 2008.

The Bloomberg ESG Disclosure Score is based on Global Reporting Initiative (GRI’s)
guidelines and covers a total of 247 possible criteria across environmental, social and
governance dimensions (Eccles et al. 2011). This disclosure score out of 100 is based on
whether actual information is revealed (Mueller 2014) for each of the environmental, social
and governance categories (Wang and Sarkis 2017). This is not an assessment of a firm’s
strengths or concerns, as in the case of Kinder, Lydenberg, Domini Research and Analytics
(KLD) (Hillman and Keim 2001). These Bloomberg ESG Disclosure Scores are not precise
performance metrics. They specify the degree to which a firm reports ESG information
(Eccles et al. 2011).

Marquis et al. (2011) assert that the ESG database provided by Bloomberg is the most
comprehensive methodology to evaluate and assess firms’ ESG activities and outcomes.
Many previous studies have used ESG data from Bloomberg to measure ESG disclosure,
among them, Dorfleitner et al. (2015), Fatemi et al. (2017) and Halbritter and Dorfleitner
(2015). Thus, this study followed such papers and used the Bloomberg ESG data. We think
that it provides sufficient information to examine the relationships between ESG disclosure,
investment recommendations, and politically connected firms.

5.2.2. Royalty Political Connection

Following Al-Hadi et al. (2016), we measure the existence of royal family directors
using the percentage of royal family directors on the board of directors.

5.2.3. Control Variable

At the firm level, many control variables have been included in the model to control
for any potential confounding effects. We follow Ioannou and Serafeim (2015) and Luo et al.
(2015) by incorporating these variables in the model. Financial performance (ROA) is one
of the key factors in analysts’ decisions. We also control for LEVERAGE because we think
it is consistent with the idea that analysts mitigate information asymmetry, and so firms
with less analyst coverage may have higher debt ratios since they are unable to issue equity
regularly. Total returns index (RI) is also one of the capital market performance factors that
influences analysts’ decisions (see variable definition). We also control for the GOV_OWN,
as it is likely that government ownership in companies affects many inputs related to
investment recommendations. It is also expected that countries with high GDP may put
some effort into increasing disclosures and better organizing and monitoring of capital
market intermediaries. The directors with experience in financial analysis (EXANALYST)
might also affect the type of information disclosed by the firm.

Some previous studies found that board size (BOARD) affects financial analysts’
forecast accuracy (e.g., Byard et al. 2006). Liquidity of shares might also be another factor
that influences analysts’ recommendations Following Cumming et al. (2011) and Drake
et al. (2010), this study uses bid-ask spreads as a proxy for liquidity. Ioannou and Serafeim

J. Risk Financial Manag. 2021, 14, 72 9 of 20

(2010) use BIDASK as a control variable, and they argue that larger spreads characterize
more opaque companies. Analysts might find those companies harder to understand
and thus be less optimistic about them. Controlling for firm size (SIZE) is motivated by
Ioannou and Serafeim (2010), who argued that “the analysts might issue more optimistic
recommendations for large firms since trading in these firms generates more trading
commissions and these firms are more likely to generate investment banking business”.
Tobin’s q (TOBIN), the ratio of the market value of a company’s assets, has been used
widely in the literature as a firm value proxy. We conjecture that a firm with a low Tobin’s
q might be more likely to be undervalued than a firm with a high Tobin’s q. This will thus
affect analysts’ recommendations. We also control for year effects, industry effects and
country effects.

6. Findings and Result Discussion
6.1. Descriptive Statistics

The descriptive statistics of the variables are exhibited in Table 2. The average ESG
disclosure score is 12.3, which is quite low and consistent with the wide range of studies
conducted in the Middle East and GCC countries. In reality, in this region, the disclosure
of non-financial information is still in its infancy. The average REC is 3.7, which implies
that most companies gain favourable recommendations by analysts either to hold or buy. A
hold recommendation generally expects the security to perform at a market rate or the same
pace as comparable securities. A buy recommendation is given by analysts for a security
that is expected to outperform the average market return of comparable stocks in the same
sector or industry. Thus, the analysts believe that most securities in GCC stock markets are
performing well. In reality, this issue needs more in-depth investigation of why most of the
stocks gain favourable recommendations. The percentage of royal family directors (RFP) is
about 10.4%. On average, the sample companies have 92% LEVERAGE. This result implies
that most of the companies finance their operations from debt. The ROA of about 4.4%
implies that most of the firms are performing quite well. The average of BIDASK is 1.3. A
bid-ask spread is the amount or percentage by which the ask price exceeds the bid price for
a security in the stock market and reflects the liquidity of the stocks. It is also a reflection of
the supply and demand for a particular security. The spread percentage is normal where it is
assumed that the ask price is higher than the bid prices. The average BOARD size is around
nine members. On average, at least one of the directors has experience in financial analysis
(EXANALYST). The average government ownership in our sample is 12% (GOV_OWN).
The average of GDP is 4.4%. The average of TOBIN is 1.25.

Table 2. Descriptive statistics.

Variable n Mean S.D. Min 0.25 Med 0.75 Max

REC 537 3.7 1.03 1 3 4 4 5
ESG 537 12.29 10 2.07 6.61 10.33 10.97 47.11
RFP 537 10.44 14.57 0 0 0 17.65 57.14
ROA 537 4.35 5.35 −16.55 1.55 2.26 6.37 29.83

LEVERAGE 537 92.3 136.92 0 25.08 60.11 113.83 1692.94
RI 537 133.96 342.07 0.12 2.78 18.92 77.97 2452.89

BIDASK 537 1.33 1.6 0.07 0.48 1.01 1.33 18.18
GOV_OWN 537 0.12 0.23 0 0 0 0.1 0.84

GDP 537 4.43 2.96 −2.37 2.7 3.98 5.41 19.59
EXANALYST 537 0.33 0.56 0 0 0 1 2

BOARD 537 8.99 1.77 5 8 9 10 18
SIZE 537 8.9 1.82 0 7.84 9.24 10.17 12.72

TOBIN 537 1.25 0.44 0.51 1.02 1.09 1.33 3.82
All variables are defined in Empirical models and variable definitions section.

J. Risk Financial Manag. 2021, 14, 72 10 of 20

6.2. Univariate Analysis

Table 3 shows the t-test for all dependent and independent variables partitioned by
royally politically connected firms and unconnected firms. Our comments on the significant
results are set below. We find that the mean REC for politically connected firms is 3.84,
which is statistically larger than the mean REC for non-connected companies of 3.58 (two
tailed, p < 0.001). The results suggest that politically connected firms gain more optimistic
recommendations than non-politically connected firms. Based on our knowledge, this is
the first study which contributes to the politically connected firm literature by showing
that sell-side analysts tend to issue favourable recommendations to firms with political
connection, particularly, those firms with royal family directors. Importantly, we also find
that ESG for royal politically connected firms is higher than for non-connected firms (mean
of 13.18 compared to mean 11.6), which is significant at two tailed (p < 0.1). Similarly,
politically connected firms also have statistically higher SOCIAL and ENV scores than non-
politically connected firms. Further, the means of ROA, GOV_OWN, and TOBIN were also
found to be higher for politically connected companies, 5.4%, 15%, and 1.31%, compared
to non-connected companies, 3.5%, 9%, and 1.21%, (two tailed p < 0.001), respectively.
These results imply that politically connected companies perform well and attract higher
government investment than their counterparts.

Table 3. T-test between politically connected versus non-politically connected firms.

Non-Politically Connected Politically Connected
Difference t-stat

(n = 301) (n = 236)

REC 3.58 3.84 −0.26 −2.898 ***
ESG 11.59 13.18 −1.59 −1.836 *

SOCIAL 6.02 9.248 −3.23 −2.56 **
ENV 4.1 5.73 −1.63 −1.80 *
ROA 3.51 5.43 −1.92 −4.194 ***

LEVERAGE 93.63 90.59 3.04 0.255
RI 116.95 155.66 −38.71 −1.303

GOV_OWN 0.09 0.15 −0.06 −3.187 ***
GDP 4.23 4.69 −0.46 −1.779 *

EXANALYST 0.34 0.33 0.012 0.259
BOARD 8.98 9.02 −0.04 −0.262
BIDASK 1.31 1.35 −0.042 −0.299

SIZE 8.8 9.02 −0.225 −1.421
TOBIN 1.21 1.31 −0.106 −2.801 ***

* p < 0.10; ** p < 0.05; *** p < 0.01 indicate that the estimated coefficients are statistically significant at the 10 percent, 5
percent, and 1 percent levels, respectively. All variables are defined in Empirical models and variable definitions section.

We conduct further analysis to see which countries disclose more by comparing the
countries’ average ESG. In Table 4, we find the average of ESG disclosure in Oman is a
score of 17, which is the highest followed by United Arab Emirates. Kuwait has the lowest
disclosure with a score of 8.4.

Table 4. Comparison between countries in terms of ESG.

Total Country mean_ESG

62 Kuwait 8.343037
85 Qatar 10.23224
25 Bahrain 10.73654

169 Saudi Arabia 10.90623
140 United Arab Emirates 15.2897
56 Oman 17.13645

Table 5 shows the Pearson Correlation Matrix for the main variables included in the
regressions. The REC is positively associated with ESG. This implies that the analysts in
our sample were favourable towards ESG disclosure. We also find a positive and significant
correlation between ESG and RFP. We also find that there is a positive and significant corre-
lation between REC and ROA, GDP, EXANALYST, SIZE, and environmentally sensitive
industries (IND).

J. Risk Financial Manag. 2021, 14, 72 11 of 20

Table 5. Pearson correlation matrix.

1 2 3 4 5 6 7 8 9 10 11 12 13 14

1 REC 1
2 ESG 0.11 * 1
3 RFP 0.06 0.11 ** 1
4 ROA 0.2 *** −0.05 0.05 −0.08 *
5 LEVERAGE −0.02 0.05 0.03 0.08 1
6 RI −0.05 −0.17 *** 0.22 *** −0.05 * −0.07 * 1
7 GOV_OWN 0.08 −0.01 −0.04 −0.06 0.16 *** −0.13 *** 1
8 GDP 0.14 *** −0.01 0.07 0.06 0.04 −0.16 *** 0.02 1
9 EXANALYST 0.17 *** 0.09 ** −0.13 *** −0.1 ** 0.17 *** −0.17 *** 0.58 *** 0.04 1

10 BOARD −0.07 0 −0.03 0.04 −0.11 *** −0.2 *** −0.01 0 0.07 1
11 BIDASK 0.02 0.06 0.01 −0.05 0.03 0.1 ** −0.07 0.01 0.05 −0.14 *** 1
12 SIZE 0.12 *** −0.02 0.04 0.12 *** 0 −0.25 *** 0.06 0.12 *** 0.01 0.15 *** −0.35 *** 1
13 TOBIN 0 −0.02 0 −0.05 −0.16 *** 0.01 0.17 *** 0.01 0.19 *** 0 −0.12 *** 0.19 *** 1
14 IND 0.1 ** 0.01 −0.2 *** −0.17 *** 0.17 *** −0.08 * 0.3 *** 0.02 0.33 *** −0.1 ** 0 −0.07 * 0.17 *** 1

* p < 0.10; ** p < 0.05; *** p< 0.01 indicate that the estimated coefficients are statistically significant at the 10 percent, 5 percent, and 1 percent levels, respectively. All variables are defined in Empirical models and
variable definitions section, except for IND, which takes a value of 1 if firm is in environmentally sensitive industries, and 0 otherwise.

J. Risk Financial Manag. 2021, 14, 72 12 of 20

6.3. Multivariate Analyses

In this section, we present the findings of estimation results after controlling for
several variables. Our dependent variable is investment recommendations (REC), and the
independent variables of interest are environmental, social and governance (ESG) and the
royal family directors (RFP). Table 6 shows the coefficients results of our base regression
for the association between ESG and REC.

Table 6. Baseline regression for the association between ESG and REC.

Variables Model 1

ESG 0.008 **
(0.004)

ROA 0.064 ***
(0.02)

LEVERAGE 0.018
(0.001)

RI 0.001 ***
(0.001)

GOV_OWN −0.667 ***
(0.23)

GDP 0.037 **
(0.02)

EXANALYST 0.299 ***
(0.09)

BOARD −0.066 ***
(0.02)

BIDASK 0.008
(0.03)

SIZE 0.145 ***
(0.05)

TOBIN −0.717 ***
(0.16)

Constant 3.904 ***
(0.52)

Industry effects Yes
Year effect Yes

Country effects Yes
Observations 537

R-squared 0.258

Robust standard errors in parentheses
** p < 0.05; *** p < 0.01 indicate that the estimated coefficients are statistically significant at the 10 percent, 5 percent,
and 1 percent levels, respectively. All variables are defined in Empirical models and variable definitions section.

In particular, the coefficient of REC and ESG of 0.008 is positive and statistically signif-
icant at (p < 0.05). This result is consistent with the study of Ioannou and Serafeim (2015),
which finds a significant positive relationship between CSR disclosure and investment
recommendation. The analysts favorably assess ESG disclosure. This result supports hy-
pothesis H1. Prior literature states that more informative disclosure attracts more financial
analysts because gathering information becomes less costly. The investors rely on analysts
when they make their investment decisions and might use ESG information. It might also
be perceived by market analysts as positively contributing towards a company’s long-term
profitability. They are much more likely to recommend potential shareholders to purchase
the shares of these firms with better ESG. Therefore, Table 6 provides results consistent with
our first hypothesis that analysts perceive ESG as a positive factor for a firm’s long-term
financial performance which improves the firm’s value. The rest of the control variables
have been found to influence the REC, except BIDASK.

J. Risk Financial Manag. 2021, 14, 72 13 of 20

6.4. Moderating Effects of Royalty Political Connection

Table 7 provides results of the moderation effects of royal family directors on the
relation between ESG and REC. The coefficients of ESG (.013) and RFP (.013) are found to
be positively and significantly associated with REC at p < 0.001 and p < 0.05, respectively.
Interestingly, the coefficient for ESG is higher in the recommendation model with moderator,
compared to the baseline regression. However, the interaction’s coefficient between ESG
and RFP is negative and significant at p < 0.05. One interpretation is that the analysts
perceive that those firms with both high ESG and high RFP may engage in window dressing
or greenwashing for the sake of appearing responsible. Several studies indicate that the
presence of royal family in the firm can attract several self-benefits such as lower cost of debt
(Al-Hadi et al. 2016), more government contracts, eliminating competitors from entering a
market (Bunkanwanicha and Wiwattanakantang 2009), and enhancing the performance
of family firms (Xu et al. 2015). Thus, analysts can recommend firms with royal family in
different circumstances. However, our result suggests that high ESG by RFP firms does
not necessarily lead analysts to issue more favourable recommendation for those firms.
Another interpretation stems from an impression management perspective. Those analysts’
experts see the ESG information disclosed by firms with royal family directors negatively,
thus they discount it because the information environment is opaque. In other words, they
do not really trust such information. As we mentioned earlier, impression management,
especially by powerful leaders, may affect audiences negatively.

Table 7. The moderating effect of royal family between ESG and REC.

Variables Model 2

ESG 0.013 ***
(0.005)

RFP 0.013 **
(0.006)

ESG * RFP −0.001 **
(0.001)

ROA 0.063 ***
(0.018)

LEVERAGE 0.001
(0.001)

RI 0.001 ***
(0.001)

GOV_OWN −0.682 ***
(0.221)

GDP 0.037 **
(0.016)

EXANALYST 0.292 ***
(0.089)

BOARD −0.066 ***
(0.022)

BIDASK 0.003
(0.035)

SIZE 0.156 ***
(0.054)

TOBIN −0.704 ***
(0.164)

J. Risk Financial Manag. 2021, 14, 72 14 of 20

Table 7. Cont.

Variables Model 2

Constant 2.059 ***
(0.525)

Industry effects YES
Year effect YES

Country effects YES
Observations 537

R-squared 0.23
Robust standard errors in parentheses: ** p < 0.05; *** p < 0.01 indicate that the estimated coefficients are statistically
significant at the 10 percent, 5 percent, and 1 percent levels, respectively. All variables are defined in Empirical
models and variable definitions section.

6.5. Endogeneity Check Using Two Stage Least Square Equation (2SLS)

The issue of endogeneity (or reverse causality) might be particularly problematic
when assessing the relation between analysts’ recommendations and ESG disclosures.
The sign, magnitude, or statistical significance of these estimates may be biased due to
endogeneity (e.g., causality effects, omitted control variables), that is, the ESG and the
error term being correlated. This may be because it is correlated with another determinant
that is excluded from or not fully controlled for in our regression models, even though
the OLS estimation suggests a positive and significant association between ESG and
recommendations. To address this concern, we adopt a two-stage instrumental variable
(2SLS) approach to re-examine the findings reported in Table 6. This method is suitable
only if the instrumental variable(s) is/are correlated with the endogenous regressor (here
ESG) but uncorrelated with the error term of the second-stage regression. To select our
instrumental variable, we follow León-Ledesma and Thirlwall (2002) and Wintoki et al.
(2012) and use a lagged variable (ESG_LG) of our main independent variable (ESG) in the
first stage of the Equation (3).

REC = β0 + β1ESG_LG + controls variables + e (3)

We then include the error term (ê) from Equation (3) into our second stage regression
Equation (4).

REC = β0 + β1ESG + controls variables + δê + e (4)

Table 8 shows the results of the second stage regression of the 2SLS. We find consistent
evidence with our main H1 that ESG is positively and significantly associated with REC. In
particular, the coefficient of ESG is 0.10 at p < 0.05. Furthermore, following prior literature
(e.g., Al-Hadi et al. 2016), we check the quality of using the instrumental variable. In
Table 8, we show a number of post-estimations tests using ESG_LG as our instrumental
variable. First, based on the Heckman endogeneity test, we find that our regressor in
the first stage is endogenous (p < 0.0433). Second, we run the under identification test
(Kleibergen–Paap rk LM statistic) and find that our p-value is significant at p < 0.001. Third,
we test for the weak identification test of our instrumental variable. We find that our
Kleibergen–Paap rk Wald F statistic value is 185.676, which is larger than the Stock–Yogo
weak ID test critical values: 10% maximal instrumental variables (IV) size of 16.38.

J. Risk Financial Manag. 2021, 14, 72 15 of 20

Table 8. Endogeneity check using Two Stage Least Square Equation (2SLS).

Variables Model 1

Second stage

ESG_LG 0.010 **
(0.01)

ROA 0.070 ***
(0.01)

LEVERAGE 0.001
(0.01)

RI 0.001 ***
(0.01)

GOV_OWN −0.551 **
(0.27)

GDP 0.009
(0.03)

EXANALYST 0.214 *
(0.11)

BOARD −0.068 **
(0.03)

BIDASK −0.007
(0.03)

SIZE 0.101 *
(0.06)

TOBIN −0.696 ***
(0.15)

Constant 2.569 ***
(0.55)

Industry effects Yes
Year effect Yes
Country effects Yes
Observations 441
R-squared 0.26

Second Stage:

1—Under identification test 0.001
2—Weak identification test 185.676
Stock–Yogo weak ID test critical values: 10%
maximal IV size

16.38

3—Endogeneity test 0.0433
Results of First stage with the post-estimations are available upon request. Robust standard errors in parentheses.
* p < 0.10; ** p < 0.05; *** p < 0.01 indicate that the estimated coefficients are statistically significant at the 10 percent,
5 percent, and 1 percent levels, respectively. All variables are defined in Empirical models and variable definitions
section.

6.6. Additional Analyses

In Table 9, we regress the disaggregated ESG measure into its three categories, ENV,
SOCIAL, and GOV, with the analysts’ recommendations (REC). These analyses will provide
us with a better understanding: first, which category of ESG (ENV, SOCIAL, or GOV)
affects the analysts’ recommendation of the firm, second, how the interplay between royal
family directors and the three categories of ESG (ENV, SOCIAL, or GOV) affects analysts’
recommendations.

J. Risk Financial Manag. 2021, 14, 72 16 of 20

Table 9. Sub-analyses with individual ESG index; ENV, SOCIAL, and GOV.

Variables (1) (2) (3) (4) (5) (6)

ENV ENV * RFP SOCIAL SOCIAL * RFP GOV GOV * REP

ENV 0.009 *** 0.015 ***
(0.003) (0.004)

RFP 0.009 ** 0.008 * 0.020 **
(0.004) (0.004) (0.009)

ENV * RFP −0.001 **
(0.000)

SOCIAL 0.007 *** 0.010 ***
(0.002) (0.003)

SOCIAL * RFP −0.0001 *
(0.000)

GOV −0.003 −0.001
(0.004) (0.005)

GOV * RFP −0.000
(0.000)

ROA 0.064 *** 0.062 *** 0.065 *** 0.063 *** 0.063 *** 0.063 ***
(0.017) (0.017) (0.018) (0.018) (0.017) (0.017)

LEVERAGE 0.000 0.000 0.000 0.000 0.000 0.000
(0.000) (0.000) (0.000) (0.000) (0.000) (0.000)

RI 0.001 *** 0.001 *** 0.001 *** 0.001 *** 0.001 *** 0.001 ***
(0.000) (0.000) (0.000) (0.000) (0.000) (0.000)

GOV_OWN −0.662 *** −0.659 *** −0.686 *** −0.694 *** −0.737 *** −0.783 ***
(0.225) (0.221) (0.225) (0.222) (0.226) (0.222)

GDP 0.036 ** 0.036 ** 0.036 ** 0.037 ** 0.037 ** 0.036 **
(0.016) (0.016) (0.016) (0.016) (0.016) (0.016)

EXANALYST 0.299 *** 0.292 *** 0.305 *** 0.302 *** 0.330 *** 0.337 ***
(0.090) (0.089) (0.090) (0.089) (0.093) (0.092)

BOARD −0.065 *** −0.066 *** −0.066 *** −0.069 *** −0.066 *** −0.068 ***
(0.022) (0.022) (0.022) (0.022) (0.022) (0.022)

BIDASK 0.009 0.004 0.009 0.005 0.008 0.008
(0.034) (0.034) (0.034) (0.034) (0.034) (0.034)

SIZE 0.136 ** 0.145 *** 0.139 ** 0.145 *** 0.149 *** 0.156 ***
(0.056) (0.056) (0.055) (0.054) (0.055) (0.055)

TOBIN −0.711 *** −0.682 *** −0.721 *** −0.712 *** −0.707 *** −0.705 ***
(0.164) (0.164) (0.165) (0.165) (0.162) (0.164)

Constant 2.241 *** 2.194 *** 2.210 *** 2.201 *** 2.238 *** 2.257 ***
(0.537) (0.533) (0.530) (0.528) (0.544) (0.549)

Industry effects Yes Yes Yes Yes Yes Yes
Year effect Yes Yes Yes Yes Yes Yes

Country effects Yes Yes Yes Yes Yes Yes
Observations 537 537 537 537 537 537

R-squared 0.261 0.269 0.261 0.267 0.255 0.262

Robust standard errors in parentheses. * p < 0.10, ** p < 0.05; *** p < 0.01 indicate that the estimated coefficients are statistically significant at
the 10 percent, 5 percent, and 1 percent levels, respectively. All variables are defined in Empirical models and variable definitions section,
except for SOCIAL, ENV and GOV: SOCIAL = Bloomberg score based on the extent of a company’s Social disclosure. ENV = Bloomberg
score based on the extent of a company’s Environmental disclosure. GOV = Bloomberg score which measures the quality of corporate
governance.

Table 9 shows six columns of regressions models: columns 1, 3, and 5 for ENV,
SOCIAL, and GOV individually, and columns 2, 4, and 6 for the interaction terms between
ENV, SOCIAL, and GOV and RFP. The results show that ENV and SOCIAL performance
are positively related to analysts’ recommendations, and their coefficients are 0.009 and
0.007 and significant at p < 0.05, respectively. However, we did not find that governance
positively related to analysts’ recommendations.

The interaction terms (ENV * RFP) and (SOCIAL * RFP) are both negative and signifi-
cant at p < 0.05. This might suggest that firms that have a higher presence of royal family
directors and report higher ENV and SOCIAL disclosures are likely to use the corporate

J. Risk Financial Manag. 2021, 14, 72 17 of 20

social responsibility engagement as a vehicle for impression management. If this is the case,
the analysts can see through the corporate spin and react with skepticism in giving their
recommendations. In fact, these results contribute to our findings that ENV and SOCIAL
activities are less structured and regulated in GCC countries. The royal families, therefore,
have more discretion to influence these types of practices in GCC countries.

On the other hand, the corporate governance disclosure (GOV) is applied in all
GCC countries with less skepticism since it is highly mandated and regulated as well as
being closely monitored by various authorities, such as Central Banks, auditors, stock
market regulators, and governmental ministries. Thus, we assume that the royal family
and analysts perceive GOV disclosure differently. In fact, we find in Table 9 that the
corporate governance category (GOV) of ESG and the interaction term (GOV * RFP) have
no relationship with analysts’ recommendations. This result is also consistent with prior
studies that provide evidence on the royal family having less discretion of corporate
governance in GCC countries (Al-Yahyaee and Al-Hadi 2016; Al-Hadi et al. 2015).

7. Conclusions

There has been a debate in academia regarding whether capital markets benefit from
positive investment recommendations arising from the social and environmental concerns
of the CSR practices. First, we are interested to see whether better ESG disclosure influences
sell-side analysts’ recommendations or not. Second, we want to see how financial analysts
view the firms with royalty political connections in terms of their ESG disclosure. There
are only a few prior studies that have focused on the link between ESG and investment
recommendations. This has created a gap in the literature. We fill this important lacuna
by bringing in new evidence from different cultures. We find that connected firms gain
more optimistic recommendations than non-politically connected firms. As far as we are
aware, this is the first time that a study that contributes to the politically connected firm
literature shows that sell-side analysts tend to issue favorable recommendations to firms
with political connection, particularly, those firms with royal family directors. Importantly,
we also find that ESG for royal politically connected firms is higher than for non-connected
firms (mean of 13.18 compared to mean 11.6), which is significant at two-tailed (p < 0.1).
We also find that there is a positive relationship between ESG disclosure and sell-side
analysts’ recommendations. These findings are in line with Ioannou and Serafeim (2015).
This study confirms the outcomes of the study of Eccles et al. (2011), which found that the
capital market shows a high interest in the level of CSR disclosures, and Luo et al. (2015),
who found that analysts factor CSR information into their investment recommendations.
It also found that ESG disclosures in companies with political connections are perceived
negatively by sell-side analysts.

Generally, analysts act as an important intermediary channel between the company
and the stakeholders. As industry experts, financial analysts reduce the information
asymmetry between firms and investors by incorporating firm CSR information into their
recommendations. We extend the literature from other capital market settings and cultures.

It is also noticeable that the ESG disclosure is very low. The pressure of stakeholders
and government bodies on firms to disclose ESG information is very weak. To improve
this, policymakers and stakeholders can be advised to play key roles by enforcing firms
and issuing some compulsory rules on ESG disclosure. The securities commissions can
also make some obligatory rules and follow some practices of some Western and Asian
securities commissions that obligate firms to disclose ESG information. Due to our small
sample size of 537 firm observations, it can be difficult to make inferences. Another
limitation is using the Bloomberg ESG rating as the proxy for ESG disclosure. There is a
divergence between different ESG ratings and every rating agency can rank a particular
company differently (Chatterji et al. 2016; Dorfleitner et al. 2015). Future research can
conduct surveys or interviews with financial analysts to study their perception about the
existence of royal family directors on the board of directors. It is also worth conducting
future research to study the relationship between foreign ownership and the performance

J. Risk Financial Manag. 2021, 14, 72 18 of 20

of firms by comparing firms with royal family directors and firms which have no royal
family directors.

Author Contributions: A.A. developed the idea and the theoretical framework, performed the
analysis, and writing the results. Both W.N.W.-H., M.J. contributed to the final version of the
manuscript. They also helped in editing and improving the paper. A.A.-h. contributed to robust
analyses. All authors have read and agreed to the published version of the manuscript.

Funding: This research received no external funding.

Informed Consent Statement: Not applicable.

Data Availability Statement: The data that support the findings of this study are available upon
request from the authors.

Acknowledgments: The authors would like to thank the editor of the journal and the participants
of the 23rd Annual Financial Reporting and Business Communication Conference, University of
Reading, and the participants of the Annual Meeting of the Decision Sciences Institute, Washington
D.C, for their valuable feedback.

Conflicts of Interest: The authors declare no conflict of interest.

References
Adhikari, Binay K. 2016. Causal Effect of Analyst Following on Corporate Social Responsibility. Journal of Corporate Finance 41: 201–16.

[CrossRef]
Al-Ajmi, Jasim, and J. H. Kim. 2012. Are Gulf Stock Markets Efficient? Evidence from New Multiple Variance Ratio Tests. Applied

Economics 44: 1737–47. [CrossRef]
Alazzani, Abdulsamad, Yaseen Aljanadi, and Obeid Shreim. 2019. The Impact of Existence of Royal Family Directors on Corporate

Social Responsibility Reporting: A Servant Leadership Perspective. Social Responsibility Journal 15: 120–36. [CrossRef]
Al-Hadi, Ahmed, Grantley Taylor, and Mahmud Hossain. 2015. Disaggregation, Auditor Conservatism and Implied Cost of Equity

Capital: An International Evidence from the GCC. Journal of Multinational Financial Management 29: 66–98. [CrossRef]
Al-Hadi, Ahmed, Grantley Taylor, and Khamis Hamed Al-Yahyaee. 2016. Ruling Family Political Connections and Risk Reporting:

Evidence from the GCC. International Journal of Accounting 51: 504–24. [CrossRef]
Al-Yahyaee, K., and Ahmed Al-Hadi. 2016. Ineffective Corporate Governance: Busyness of Internal Board Monitoring Committees.

Corporate Control and Ownership 13: 309–25. [CrossRef]
Balli, Faruk, Rosmy J. Louis, and Mohammad Osman. 2011. The Patterns of Cross-Border Portfolio Investments in the GCC Region: Do

Institutional Quality and the Number of Expatriates Play a Role? Journal of Economics and Finance 35: 434–55. [CrossRef]
Belal, Ataur Rahman, Stuart M. Cooper, and Robin W. Roberts. 2013. Vulnerable and Exploitable: The Need for Organisational

Accountability and Transparency in Emerging and Less Developed Economies. Accounting Forum 37: 81–91. [CrossRef]
Bley, Jorg, and Mohsen Saad. 2012. Idiosyncratic Risk and Expected Returns in Frontier Markets: Evidence from GCC. Journal of

International Financial Markets, Institutions and Money 22: 538–54. [CrossRef]
Bradshaw, Mark T. 2004. How Do Analysts Use Their Earnings Forecasts in Generating Stock Recommendations? Accounting Review 79:

25–50. [CrossRef]
Brown, Lawrence D., Andrew C. Call, Michael B. Clement, and Nathan Y. Sharp. 2015. Inside the ‘Black Box’ of Sell-Side Financial

Analysts. Journal of Accounting Research 53: 1–47. [CrossRef]
Bunkanwanicha, Pramuan, and Yupana Wiwattanakantang. 2009. Big business owners in politics. The Review of Financial Studies 22:

2133–2168. [CrossRef]
Byard, Donal, Ying Li, and Joseph Weintrop. 2006. Corporate Governance and the Quality of Financial Analysts’ Information. Journal of

Accounting and Public Policy 25: 609–25. [CrossRef]
Chatterji, Aaron K., Rodolphe Durand, David I. Levine, and Samuel Touboul. 2016. Do ratings of firms converge? Implications for

managers, investors and strategy researchers. Strategic Management Journal 37: 1597–1614. [CrossRef]
Cumming, Douglas, Sofia Johan, and Dan Li. 2011. Exchange Trading Rules and Stock Market Liquidity. Journal of Financial Economics

99: 651–71. [CrossRef]
Dhaliwal, Dan S., Oliver Zhen Li, Albert Tsang, and Yong George Yang. 2011. Voluntary Nonfinancial Disclosure and the Cost of

Equity Capital: The Initiation of Corporate Social Responsibility Reporting. Accounting Review 86: 59–100. [CrossRef]
Dhaliwal, Dan S., Suresh Radhakrishnan, Albert Tsang, and Yong George Yang. 2012. Nonfinancial Disclosure and Analyst Forecast

Accuracy: International Evidence on Corporate Social Responsibility Disclosure. The Accounting Review 87: 723–59. [CrossRef]
Dorfleitner, Gregor, Gerhard Halbritter, and Mai Nguyen. 2015. Measuring the Level and Risk of Corporate Responsibility—An

Empirical Comparison of Different ESG Rating Approaches. Journal of Asset Management 16: 450–66. [CrossRef]
Drake, Michael S., Linda A. Myers, and Lijie Yao. 2010. Are Liquidity Improvements around the Mandatory Adoption of IFRS

Attributable to Comparability Effects or to Quality Effects? Sarasota: American Accounting Association.

J. Risk Financial Manag. 2021, 14, 72 19 of 20

Eccles, Robert G., George Serafeim, and Michael P. Krzus. 2011. Market Interest in Nonfinancial Information. Journal of Applied Corporate
Finance 23: 113–27. [CrossRef]

Fatemi, Ali, Martin Glaum, and Stefanie Kaiser. 2017. ESG Performance and Firm Value: The Moderating Role of Disclosure. Global
Finance Journal 38: 45–64. [CrossRef]

Fieseler, Christian. 2011. On the Corporate Social Responsibility Perceptions of Equity Analysts. Business Ethics 20: 131–47. [CrossRef]
Fung, Simon Yu Kit, Gaoguang Stephen Zhou, and Xindong Kevin Zhu. 2016. Monitor Objectivity with Important Clients: Evidence

from Auditor Opinions around the World. Journal of International Business Studies. [CrossRef]
Garcia-Sanchez, Isabel-Maria, Beatriz Aibar-Guzman, Cristina Aibar-Guzman, and Lazaro Rodriguez-Ariza. 2020. “Sell” recommenda-

tions by analysts in response to business communication strategies concerning the Sustainable Development Goals and the SDG
compass. Journal of Cleaner Production 255: 120194. [CrossRef]

Gardner, W. L., and Mark J. Martinko. 1998. An organizational perspective of the effects of dysfunctional impression management.
Monographs in Organizational Behavior and Industrial Relations 23: 69–126.

Garrido-Miralles, Pascual, Ana Zorio-Grima, and María A. García-Benau. 2016. Sustainable Development, Stakeholder Engagement
and Analyst Forecasts’ Accuracy: Positive Evidence from the Spanish Setting. Sustainable Development 24: 77–88. [CrossRef]

Givoly, Dan, and Josef Lakonishok. 1984. Properties of Analysts’ Forecasts of Earnings: A Review and Analysis of the Research. Journal
of Accounting Literature 3: 117–52.

Halawi, A., B. Davidson, and National Investor. 2008. Power Matters: A Survey of GCC Boards. National Investor, Investment
Research. Available online: https://books.google.com.qa/books?idyz_zNwAACAAJ (accessed on 20 Decmber 2019).

Halbritter, Gerhard, and Gregor Dorfleitner. 2015. The Wages of Social Responsibility—Where Are They? A Critical Review of ESG
Investing. Review of Financial Economics 26: 25–35. [CrossRef]

Healy, Paul M., and Krishna G. Palepu. 2001. Information Asymmetry, Corporate Disclosure, and the Capital Markets: A Review of the
Empirical Disclosure Literature. Journal of Accounting and Economics. [CrossRef]

Hillman, Amy J., and Gerald D. Keim. 2001. Shareholder Value, Stakeholder Management, and Social Issues: What’s the Bottom Line?
Strategic Management Journal 22: 125–39. [CrossRef]

Hinze, Anne-Kathrin, and Franziska Sump. 2019. Corporate social responsibility and financial analysts: A review of the literature.
Sustainability Accounting, Management and Policy Journal 10: 183–207. [CrossRef]

Ioannou, Ioannis, and George Serafeim. 2010. THE IMPACT OF CORPORATE SOCIAL RESPONSIBILITY ON INVESTMENT
RECOMMENDATIONS. In Academy of Management Proceedings. Briarcliff Manor: Academy of Management, pp. 1–6.

Ioannou, Ioannis, and George Serafeim. 2015. The Impact of Corporate Social Responsibility on Investment Recommendations:
Analysts’ Perceptions and Shifting Institutional Logics. Strategic Management Journal 36: 1053–81. [CrossRef]

Islam, Azizul Muhammad, and Craig Deegan. 2008. Motivations for an Organisation within a Developing Country to Report Social
Responsibility Information: Evidence from Bangladesh. Accounting, Auditing and Accountability Journal 21: 850–74. [CrossRef]

Ivković, Zoran, and Narasimhan Jegadeesh. 2004. The timing and value of forecast and recommendation revisions. Journal of Financial
Economics 73: 433–63. [CrossRef]

Jegadeesh, Narasimhan, Joonghyuk Kim, Susan D Krische, and Charles M. C. Lee. 2004. Analyzing the Analysts: When Do
Recommendations Add Value? The Journal of Finance 59: 1083–124. [CrossRef]

Kamrava, Mehran, Gerd Nonneman, Anastasia Nosova, and Marc Valeri. 2016. Ruling Families and Business Elites in the Gulf
Monarchies: Ever Closer? Chatham House November: 1–14.

León-Ledesma, Miguel A., and A. P. Thirlwall. 2002. The Endogeneity of the Natural Rate of Growth. Cambridge Journal of Economics 26:
441–59. [CrossRef]

Luo, Xueming, Christian Homburg, and Jan Wieseke. 2010. Customer satisfaction, analyst stock recommendations, and firm value.
Journal of Marketing Research 47: 1041–58. [CrossRef]

Luo, Xueming, Heli Wang, Sascha Raithel, and Qinqin Zheng. 2015. Corporate Social Performance, Analyst Stock Recommendations,
and Firm Future Returns. Strategic Management Journal 36: 123–36. [CrossRef]

Marquis, Christopher, Daniel Beunza, Fabrizio Ferraro, and Bobbi Thomason. 2011. Driving Sustainability at Bloomberg LP. Boston:
Harvard Business School Organizational Behavior Unit Case, pp. 411–25.

Mueller, Florian. 2014. Portfolio Performance Implications of Environmental, Social and Governance Based Asset Selection, Working Paper.
Warsaw: University in Warsaw.

Newton, Nathan J. 2019. When Analysts Speak, Do Auditors Listen? Auditing 38: 221–45. [CrossRef]
Schlenker, Barry R. 1980. Impression Management. Monterey: Brooks/Cole, pp. 79–80.
Stickel, Scott E. 1995. The Anatomy of the Performance of Buy and Sell Recommendations. Financial Analysts Journal 51: 25–39.

[CrossRef]
Suchman, Mark C. 1995. Managing Legitimacy: Strategic and Institutional Approaches. Academy of Management Review. [CrossRef]
Tilt, Carol A. 2016. Corporate Social Responsibility Research: The Importance of Context. International Journal of Corporate Social

Responsibility 1: 1–9. [CrossRef]
Wang, Kui, and Wei Jiang. 2019. Brand equity and firm sustainable performance: The mediating role of analysts’ recommendations.

Sustainability 11: 1086. [CrossRef]
Wang, Zhihong, and Joseph Sarkis. 2017. Corporate Social Responsibility Governance, Outcomes, and Financial Performance. Journal

of Cleaner Production 162: 1607–16. [CrossRef]

J. Risk Financial Manag. 2021, 14, 72 20 of 20

White, Halbert. 1982. Maximum likelihood estimation of misspecified models. Econometrica: Journal of the Econometric Society 50: 1–25.
[CrossRef]

Wintoki, M. Babajide, James S. Linck, and Jeffry M. Netter. 2012. Endogeneity and the Dynamics of Internal Corporate Governance.
Journal of Financial Economics 105: 581–606. [CrossRef]

Xu, Nianhang, Qingbo Yuan, Xuanyu Jiang, and Kam C. Chan. 2015. Founder’s political connections, second generation involvement,
and family firm performance: Evidence from China. Journal of Corporate Finance 33: 243–59. [CrossRef]

Yu, Minna. 2011. Analyst Recommendations and Corporate Governance in Emerging Markets. International Journal of Accounting and
Information Management 19: 34–52. [CrossRef]

Zhang, Longwen, and Minghai Wei. 2019. How analyst recommendations respond to corporate uncertainty caused by investment
behavior: Currying favor with management or conflicts of interest from connections. China Finance Review International 10: 243–69.
[CrossRef]

Corporate ESG Profiles and Investor Horizons∗

Laura Starks†, Parth Venkat‡, and Qifei Zhu§

Abstract

Theories of Environmental, Social and Governance (ESG) investment assume group-

ings of investors who differ in their preferences or beliefs about ESG. We examine

whether investor horizon serves as a basis for these ESG groupings and find that longer-

horizon investors tilt their portfolios towards firms with high-ESG profiles. We provide

evidence that these results are plausibly causal through difference-in-differences tests

of shocks to firms’ ESG reputations. Further, consistent with implications of the im-

portance of investor horizon, long-term investors behave more patiently toward the

high-ESG firms in their portfolios, selling relatively less after negative earnings sur-

prises or poor stock returns.

∗The authors thank Adair Morse, Sadok El Ghoul, Raghavendra Rau, Doina Chichernea, Philipp Krueger,
Clemens Sialm, Sheridan Titman, participants at the WFA meeting, Darla Moore School, University of South
Carolina, Hitotsubashi University Corporate Finance Conference, FTSE-Russell World Investment Forum,
FMA meeting, PRI Academic Network Conference, University of Arizona, BI Norwegian Business School,
Cornell University, University of Oklahoma, University of Texas, and University of Virginia for helpful
comments. This paper won the 2018 PRI Best Quantitative Paper Award.
†McCombs School of Business, University of Texas at Austin
‡The Culverhouse College of Business, University of Alabama
§Nanyang Business School, Nanyang Technological University

Electronic copy available at: https://ssrn.com/abstract=3049943

I. Introduction

Increasingly, institutional investors consider aspects of firms’ Environmental, Social and

Governance (ESG) profiles when making investment decisions (e.g., Krueger, Sautner, and

Starks (2020)). However, considerable disagreement exists regarding how and why portfolio

decisions should be adjusted for ESG considerations. Some proponents of ESG investing cite

evidence that ESG practices—such as climate risk mitigation, enhancement of workplace

diversity and attention to supply-chain labor rights—have been associated with better long-

term returns and mitigated risks, particularly downside risks (e.g., Dunn, Fitzgibbons, and

Pomorski (2018), Hoepner, Oikonomou, Sautner, Starks, and Zhou (2019)).1 However, other

studies, such as those by Hong and Kacperczyk (2009) and Bolton and Kacperczyk (2020),

show evidence of superior financial returns for poor ESG firms (e.g., “sin” stocks or “brown”

stocks).

Beyond the conceptual arguments regarding whether firms’ ESG practices result in long-

term value, a number of theories on sustainable investing characterize ESG investors accord-

ing to their non-pecuniary preferences or their beliefs about how ESG improves firm value.2

An important dimension for many of these theories is that the proportion of the ESG in-

vestors in the market affects the market equilibrium. In order to understand the potential

effects of these weightings of ESG investors as well as the effects of ESG investor demand

due to their risk-return motivations, we need a better understanding of ESG investor charac-

teristics. In this paper, we analyze portfolio holdings to test whether institutional investors’

portfolio tilts toward ESG are related to characteristics of the investors, with an emphasis

on investor horizons.

We find investment horizon to be a significant determinant in explaining investors’ hold-

1Theories suggest that better ESG practices increase firms’ fundamental value because such practices help
firms avoid value-destroying actions such as myopic decisions, strengthen a firm’s market position (Bénabou
and Tirole (2010)), better attract customers and provide employees with incentives for greater productivity
(Baron (2008), Baron (2001)), decrease firm risks (Albuquerque, Koskinen, and Zhang (2019)), and avoid
potential litigation (Eccles, Ioannou, and Serafeim (2014)).

2See, for example, Heinkel, Kraus, and Zechner (2001), Baker, Hollifield, and Osambela (2020), Oehmke
and Opp (2020),Pastor, Stambaugh, and Taylor (2020), Pedersen, Fitzgibbons, and Pomorski (2020).

1

Electronic copy available at: https://ssrn.com/abstract=3049943

ings of ESG stocks. Specifically, using various measures of investment horizon—reported

portfolio turnover for mutual funds as well as churn ratios derived from mutual fund and

13f institutional investor portfolio holdings—we document that longer horizon investors have

significantly stronger preferences for high-ESG firms as demonstrated by their portfolio tilts

towards stocks with high MSCI or Sustainalytics ESG scores.3 Correspondingly, we find that

these relationships also hold at the firm level—high-ESG firms have a shareholder base that

is, on average, more long-term oriented than that of low-ESG firms.

The findings that long-horizon investors have stronger preferences for high-ESG firms

than do short-horizon investors are consistent with anecdotal evidence from practitioners, a

number of whom have stated their belief that responsible investing pays off over the long

term, e.g., “We firmly believe that organizations that manage Environmental, Social and

Governance (ESG) factors effectively are more likely to endure and create more value over

the long-term than those which do not.” or “Improving ESG factors can improve the long-

term financial performance of a company.”4

The relationship between investment horizon and ESG preference is also consistent with

theories on investor myopia. Applied in an ESG context, these theories suggest differences

could exist between long-term and short-term investors. For example, Bolton, Scheinkman,

and Xiong (2006) argue that short-term earnings have a speculative component that can

induce price bubbles in the near term. Investors who benefit more from near-term stock price

appreciation would encourage managers to boost short-term earnings even at the expense

of long-term value, which suggests that ESG investments with longer payout periods may

not be attractive to such investors. Alternatively, long-term investors’ preferences for ESG

could arise from a model such as that of Froot, Perold, and Stein (1992), in which long-term

3We use the Gaspar, Massa, and Matos (2005) measure of churn ratio. In addition, we employ an
alternative horizon measure for the 13f investors through the classification of transient investors from Bushee
(1998).

4See Dimson, Kreutzer, Lake, Sjo, and Starks (2013). In addition, Larry Fink, the CEO of Blackrock,
the world’s largest asset manager, has written a number of letters over the years to the CEOs of Blackrock’s
portfolio firms regarding the relationship between long-term value and ESG or sustainability. See, for
example, https://www.blackrock.com/corporate/investor-relations/2016-larry-fink-ceo-letter.

2

Electronic copy available at: https://ssrn.com/abstract=3049943

and short-term investors have divergent evaluations of ESG projects due to asymmetric

information.

To better isolate a potential causal relationship between firms’ ESG profiles and institu-

tional investors’ portfolio decisions, we consider a shock that can cause investors to reassess

a firm’s ESG profile. The shock is the FTSE4Good US Select Index biannual rebalance in

which firms are added or eliminated from the Index due to evaluations by FTSE-Russell

regarding the firm’s ESG profile. Since FTSE4Good is an established index that specializes

in corporate ESG issues, we expect the inclusion and exclusion events to carry new infor-

mation regarding firms’ ESG reputations. Our hypothesis that ESG considerations carry

more weight for the long-term investors implies that these investors should have more pro-

nounced responses to the index rebalance news than other investors. Consistent with our

hypothesis, we find that stocks newly included into (excluded from) the index experience an

increase (decrease) in long-term investor ownership relative to other investors’ ownership,

both among mutual funds and 13f institutions. We further tighten the identification strategy

through a triple-differences analysis in which we compare included (excluded) stocks to a

matched control sample. The differential in trading behaviors of long-term investors versus

other investors remains robust under this analysis. These results are particularly striking

given that long-term investors have lower turnover in general, are less likely to adjust their

portfolios, and that the results are a lower bound on the actual portfolio adjustments as the

investors may have already conducted their own analyses on a firm’s ESG qualities.

We next consider an important implication of a systematic relationship between long-

term investor horizons and firms’ ESG profiles—that the investors should exhibit less short-

termism in this relationship and be more patient toward high-ESG firms’ management.

Specifically, if investors believe that high-ESG firm managers engage in business activities

that benefit shareholders in the long-run, the investors would tend to refrain from selling their

positions if the company encounters short-term under-performance. To test this hypothesis

regarding investors’ trading patterns around earnings surprises or poor returns, we first

3

Electronic copy available at: https://ssrn.com/abstract=3049943

confirm that the institutional investors in our sample tend to sell their portfolio holdings

following negative earnings surprises. We then conduct an analysis within each investor’s

portfolio to determine whether the selling sensitivity following negative earnings surprises

(or poor stock returns) varies systematically with the portfolio firm’s ESG score. When

comparing the positions of the same investor across the different portfolio firms, we find

that the selling sensitivity to recent underperformance is significantly reduced for the high-

ESG firms. Thus, consistent with our hypothesis, these results indicate that institutional

investors behave more patiently toward high-ESG firms, suggesting the investors believe that

these firms are in a position to create value in the long run.

Again for better identification, we use a shock to examine whether investors’ trading

patterns around negative earnings surprises change—we test for differences in these patterns

after a stock enters or exits the FTSE4Good Index. The results show that for firms newly

included into (or excluded from) the FTSE4Good US Select Index, the sensitivity of selling

trades with respect to negative earnings surprises decreases (or increases) significantly. Since

the comparison is made for the same firm shortly before and after the Index rebalance events,

the differences across the investors’ trading behaviors in their own portfolio can be attributed

to their changed perceptions about the ESG profile of their portfolio companies.

Finally, given the increasing interest in ESG for investment considerations, we examine

the related question of whether the increasing ESG focus is reflected in the aggregate ESG

profiles of institutional investor portfolios. We use the time-series of portfolios and both

the MSCI and Sustainaltyics ESG scores and we control for other firm characteristics that

have generated investor preferences in previous studies and for the changes in ESG scoring.

Our results show that, over the sample period, institutional investors have had a growing

preference for high-ESG firms and a decreasing preference for low-ESG firms.

Our paper is related to previous research on the relationship between institutional in-

vestors and firms’ ESG profiles as reflected in ESG or CSR scores. A number of studies have

examined this relationship with mixed results. Institutional ownership is positively related

4

Electronic copy available at: https://ssrn.com/abstract=3049943

to lower scores (Borghesi, Houston, and Naranjo (2014)); negatively related to firms with

improved scores (Gillan, Hartzell, Koch, and Starks (2010)); and negatively related to firms

with high environmental scores (Chava (2014); Fernando, Sharfman, and Uysal (2009)).

However, the relationship does not appear to be monotonic as Fernando et al. also find

that institutional ownership is negatively related to firms with low environmental scores.5

Our results on the generally growing preference for high-ESG firms in institutional investor

portfolios suggests that some of these earlier analyses regarding ESG could be different if

re-examined today.

Two principal differences exist between previous research on institutional investor pref-

erences related to firms’ ESG profiles and our paper. First, our focus is on the relationship

between investment horizons and investor preferences for corporate ESG. As discussed ear-

lier, both theory and practice suggest a relationship between investors’ long-term horizons

and desired ESG profiles for portfolio companies. Further, the theory focused on ESG in-

vestors, cited earlier, rests on assumptions that investor characteristics define them as ESG

investors or not. We provide one way in which these divisions can be viewed. For example,

investors more concerned with firm value in the long run, or investors with nonpecuniary

preferences for better environmental or social performance, would be more drawn to high-

ESG firms. An increase in these types of investors would drive the increasing institutional

ownership of high-ESG firms that we document.

A second important difference between our work and earlier work on the relationship

between institutional investors and firms’ ESG scores is that we provide analyses of exogenous

shocks to firms’ ESG reputations. In studying the latter, our paper is also related to that of

Krüger (2015), who examines market reactions to news that changes firms’ CSR profiles. He

finds strong negative reactions to negative events and weakly negative reactions to positive

5In addition, Nofsinger, Sulaeman, and Varma (2019) find that institutional ownership is negatively
related to low scores, but not positively related to high scores. Lins, Servaes, and Tamayo (2017) provide
evidence that firms with higher MSCI ES scores during the financial crisis were more successful, which they
interpret as resulting in part from investors having more social trust in these firms during a time of low
trust. Gibson and Krüger (2018) also document a relationship between institutional investors’ portfolio
performance and their sustainability footprints.

5

Electronic copy available at: https://ssrn.com/abstract=3049943

events. We take a different approach by analyzing investor reactions to FTSE4Good Index

rebalance events through the changes in portfolio holdings to the shocks.6

We contribute to the more general research on investor horizon, particularly short horizon

investors, and their relationship to corporate actions or asset markets such as Bushee (1998),

Cella, Ellul, and Giannetti (2013), Derrien, Kecskés, and Thesmar (2013), Massa, Yasuda,

and Zhang (2013), and Akbas, Jiang, and Koch (2020). Our research brings into focus the

interrelated themes of short-termism and ESG investing. That is, we contribute to the de-

bate on whether corporate managers as well as their institutional shareholders have become

too short-term oriented, engaging in myopic actions that hurt long-term shareholder value.

Our findings of a match between shareholder investment horizons and firm ESG activities,

as well as our findings of more investor patience toward high-ESG firms, not only support

the hypothesis that long-term investors prefer high-ESG profile firms, but also provides a

more nuanced view on the relationship between corporate short-termism and institutional

investors’ presence as firm shareholders. For example, it helps explain seemingly conflicting

evidence in the financial press. On the one hand, many allege that institutional investors ex-

ert pressure on corporate managers to beat their quarterly earnings expectations, hindering

the managers’ ability to implement long-term strategies (Pozen (2009)). On the other hand,

institutional investors can be quite vocal in encouraging the managers of their portfolio firms

to take a longer-term view. For example, a group of the world’s largest investors concerned

about the myopia of their portfolio firms’ managers reportedly met to develop “proposals

to improve company governance that would encourage longer-term investment and reduce

friction with shareholders” (Foley and McLannahan (2016)). Although we do not take an

explicit stance on the welfare implications of long-termism versus short-termism, we offer

empirical support for the perspective that shareholders’ heterogeneous horizons could influ-

6Our paper is related to the extensive literature on SRI and ESG investing. For example, Guerard (1997),
Benson and Humphrey (2008), Statman and Glushkov (2009), Renneboog, Ter Horst, and Zhang (2008),
and Riedl and Smeets (2017).In addition, better ESG ratings at the fund level are shown to attracthigher
investor flows (Hartzmark and Sussman (2019)).

6

Electronic copy available at: https://ssrn.com/abstract=3049943

ence corporate short-termism.7 Our results in this sense are consistent with those of other

researchers who have studied whether institutional investors cause changes in firms’ ESG

profiles either by examining engagements on ESG issues by specific institutional investors

(e.g., Dimson, Karakaş, and Li (2015); Dimson, Karakaş, and Li (2018); Hoepner et al.

(2019)) or by inferring the causal effects using other types of identification strategies (e.g.,

Kim, Kim, Kim, and Park (2019); Gloßner (2019); Dyck, Lins, Roth, and Wagner (2019).

II. Data and variable constructions

A. Data sources

We gather data on quarterly equity holdings from the Thomson Reuters s12 database for

mutual funds and from the Thomson Reuters s34 database for institutions that file 13f forms

with the SEC. We obtain information on mutual fund characteristics from the CRSP Mutual

Fund database, and include only actively-managed U.S. domestic equity mutual funds.8 We

remove ETFs and index funds by parsing fund names and removing funds that hold over

1,000 stocks. Our sample contains 98,252 mutual fund-years and 166,185 institution-years

observations and spans the period of 2000 to 2018.

We also employ data from the MSCI ESG STATs database, which contains an annual set

of positive and negative ESG indicators as assessed by MSCI and its predecessors who collect

data from company disclosures, academic and government databases, and other sources. Us-

ing a binary relative rating scale with a base of zero for neutral performance, their analysts

rate companies on a wide array of issues in the primary categories of environment, commu-

nity, diversity, employee relations, human rights, products, and corporate governance. The

scores on each criterion are 1 or 0 for positive performance indicators (“strengths”), and -1

7For a theoretical example under which long-termism is efficient, see Stein (1989). For an example under
which short-termism is efficient, see Thakor (2020)

8By confining our analysis to investments in U.S. firms, we do not encounter the influence of country
legal origin given that Liang and Renneboog (2017) provide evidence that country legal origin is important
in explaining firms’ social performance.

7

Electronic copy available at: https://ssrn.com/abstract=3049943

or 0 for negative performance indicators (“concerns”).9 We aggregate the data items for a

firm-year by summing all positive and negative indicators. In some of our specifications, we

separate out the positive and negative indicators. We also construct an ESG x/ Governance

measure that excludes governance strengths and concerns from the ESG measure, as well as

indicators associated with specific industries (e.g. firearms, nuclear, tobacco). MSCI ESG

data covers all firms in the MSCI USA IMI index. During our sample period of 2000 to 2018,

the average number of firms in each cross section is approximately 2,000 and our firm-level

sample includes 26,217 firm–years.

Researchers (e.g., Berg, Kolbel, and Rigobon (2020), Gibson, Krueger, and Schmidt

(2020)) have found differences across ESG rating agency rankings. To ascertain whether

our results could be driven by idiosyncrasies in the MSCI ESG scores, we repeat our main

analyses using Sustainalytics ESG scores. Because Sustainalytics has a shorter history of

scores and their coverage is more limited than that of the MSCI scores, this analysis is

conducted over a shorter sample period (2009 to 2017) and is confined to a subset of our

original sample of firms. The ESG scores from both Sustainalytics and MSCI take into

account a firm’s industry when assigning scores, thus, they have an embedded industry

adjustment.

Throughout our paper, we control for a number of firm characteristics, including each

firm’s market capitalization, past 12-month return, return volatility, stock turnover ratio,

and an S&P 500 Index constituency indicator (from the CRSP database). We also include

the firm’s book-to-market ratio, dividend yield, and gross profitability ratio, calculated from

the Compustat database.

B. Measurement of investment horizons

A natural way to estimate mutual fund investment horizons is by using their reported portfo-

lio turnover ratios, defined as the minimum of a mutual fund’s security sales and purchases,

9See “MSCI ESG KLD STATS: 1991-2014 Data Sets” data manual available on WRDS.

8

Electronic copy available at: https://ssrn.com/abstract=3049943

scaled by total net assets and obtainable from the CRSP database. Alternatively we can

measure a mutual fund’s revealed equity investment horizon by estimating the inter-quarter

changes in their portfolio holdings, available from Thomson Reuters. Investors who trade

more frequently have shorter holding periods, implying shorter investment horizons. Em-

ploying the procedure used by Gaspar et al. (2005), we calculate each fund’s churn ratios

(CR) with quarterly intervals. By definition the churn ratio can range from 0 to 2, with a

higher churn ratio indicating that the fund turns over its holdings faster:

CRj,t =

∑
i∈I |Sharesi,j,t ∗Pi,t −Sharesi,j,t−1 ∗Pi,t|∑

i∈I (Sharesi,j,t ∗Pi,t + Sharesi,j,t−1 ∗Pi,t−1)/2
, (1)

where Sharesi,j,t and Pi,t denote the number of shares and price of company i held by fund

j at quarter t. To smooth out measurement errors, we calculate the churn ratio for a given

fund-quarter as the moving-average churn ratio of the four trailing quarters.

Although the S.E.C. requires U.S. mutual funds to report their portfolio turnovers, 13f

institutions do not face such a requirement. Thus, we use churn ratios constructed from

their portfolio holdings as the primary investment horizon measure. As a supplementary

measure to check the robustness of the results, we employ the Bushee (1998) classification,

which uses estimates of portfolio turnover rates from holdings along with the institution’s

portfolio diversification to group institutional investors into three types: transient investors,

dedicated investors, and quasi indexers. The transient investors are those with the shortest

investment horizons.10

For our firm-level analyses, we calculate the weighted averages of the estimated investor

horizons for each firm’s mutual fund or 13f institutional shareholders. For example, we

compute the firm-level mutual fund turnover ratio by combining the turnover measures from

the funds that hold their shares,s according to the holdings’ weights. Specifically, for stock

i at time t, we take a weighted average of the turnover ratios of all its holding mutual funds

10We thank Professor Bushee for making his data available at
http://acct.wharton.upenn.edu/faculty/bushee/IIclass.html.

9

Electronic copy available at: https://ssrn.com/abstract=3049943

in set J, where we weight by the number of shares held by fund j:

Firm-level MF Turnoveri =

∑
j∈J (Turnover Ratioj,t ∗Sharesi,j,t)∑

j∈J Sharesi,j,t
. (2)

Similarly, we construct firm-level mutual fund (or 13f institution) churn ratios by taking

a weighted average of churn ratios across the mutual funds (13f institutions) invested in a

given firm, where the weighting is by the number of shares held:

Firm-level ChurnRatioi =

∑
j∈J (CRj,t ∗Sharesi,j,t)∑

j∈J Sharesi,j,t
. (3)

Summary statistics are shown in Table I. In Panel A, we tabulate fund-level characteris-

tics. The average annual reported portfolio turnover ratio for the active equity mutual funds

in our sample is 0.71. We also find a wide range of turnover rates across the funds given that

the standard deviation is 0.49. On a quarterly basis, the average mutual fund churn ratio is

0.37. We also include an indicator variable for whether a fund is an SRI fund according to

lists from Bloomberg and Morningstar. Only 2% of our funds (145) have such a designation.

Panel B tabulates the characteristics of 13f institutions and we find a similar churn ratio of

0.35.

Panel C reports the summary statistics at the firm-level. An average of 14.2% of the

firms’ shares in our sample is held by mutual funds and 69.6% is held by 13f institutions.11

Measuring turnover at the firm level, we find that the weighted-average turnover ratio of

a firm’s mutual fund holders is 42.9%. This is lower than the turnover ratio of an average

mutual fund, indicating that funds holding larger numbers of shares tend to have lower

portfolio turnover. Panel C also shows firm characteristics, such as market capitalization,

book-to-market ratio, and past stock return.

Panels A, B and C also report ESG scores at the fund-, 13f institution-, and firm-levels.

At the mutual fund level (Panel A) the weighted-average ESG for their portfolio holdings is

11The 13F institutional data includes mutual fund families or parent companies.

10

Electronic copy available at: https://ssrn.com/abstract=3049943

1.06. At the firm level (Panel C), the average ESG score of the sample firms is -0.03, with a

median of 0. The difference between the mean portfolio ESG score and the mean firm ESG

score reflects the fact that larger companies tend to have higher ESG scores. We further

decompose the ESG scores into strengths and concerns, as well as ESG scores excluding

governance.

III. Investor ESG preferences and their horizons

In this section, we analyze our hypotheses regarding the relation between investors’ prefer-

ences for corporate ESG and their investment horizons. As pointed out earlier, on a con-

ceptual basis, investor horizon is considered an important determinant of investor preference

for corporate ESG. To examine the relationship between investor horizon and firm’s ESG

profiles, we conduct tests from two separate perspectives because the implications of our

hypotheses should hold at both the investor level and the level of the underlying holdings,

i.e., the firms. Therefore, we first examine the relationship between the investors’ horizons

and the average ESG scores at the portfolio level. Second, we analyze the hypotheses at the

firm-level by aggregating the investor horizon of each firm’s shareholders and testing whether

the aggregate investor shareholder horizon, either mutual fund or 13f institution, is related

to the firm’s ESG profile.

A. Institutional investors’ revealed preference for Corporate ESG: Investor-

level evidence

We hypothesize that mutual funds and institutions with longer investment horizons hold

companies with higher ESG scores in their portfolios. To test whether the revealed prefer-

ences for corporate ESG derived from investor portfolio holdings are related to the investor’s

horizon, proxied by lower turnover ratios or lower churn ratios, we first sort mutual funds

or 13f institutions in each cross-section into five quintiles based on their portfolio turnover

11

Electronic copy available at: https://ssrn.com/abstract=3049943

ratios (for mutual funds) and churn ratios (for 13f institutions). Portfolio 1 consists of the

quintile with the lowest turnover or churn ratios, and is defined as the “long-horizon” group,

while Portfolio 5 consists of the quintile with the highest turnover or churn ratios, the ”short-

horizon” group. For each quintile, we then tabulate the average portfolio-level ESG score,

which is calculated as the value-weighted average ESG score of each fund’s or institution’s

underlying portfolio companies, averaged across the quintile. Panel (a) of Figure 1 shows

the results for actively managed equity mutual funds. The average turnover ratios for the

quintiles vary from 20% to 161%. The figure shows a clear monotonically decreasing re-

lationship between mutual fund turnover ratios and fund-level ESG scores. Long-horizon

funds (Portfolio 1) have the highest average ESG score of 0.91, while short-horizon funds

(Portfolio 5) have an average ESG score of 0.42. The spread of ESG scores across the groups

is economically meaningful, as it represents 28% of the standard deviation of the sample’s

fund-level ESG scores.

The same pattern between investors’ aggregate holdings’ ESG scores and investor hori-

zons is observed for the 13f institutions as shown in Panel (b) of Figure 1. Again we find the

relation between churn ratios and the portfolio holdings’ ESG scores to be monotonically

decreasing. Thus, the evidence from both mutual funds and 13f institutions suggests that

investors with long-term horizons tend to prefer firms with higher-rated ESG scores.

However, there may exist other potentially intervening factors that need to be considered

in examining the relation between portfolio horizons and portfolio-level ESG scores. For

example, funds and institutions with different horizons often follow different investment

styles, which may correlate with their portfolio companies’ ESG scores. In order to control

for other portfolio characteristics, we conduct the following regression:

FundESGi,t+1 = αt + β1Horizoni,t + β2Controlsi,t + �i,t, (4)

where FundESGi,t+1 is the weighted-average ESG score for fund i’s portfolio companies at

12

Electronic copy available at: https://ssrn.com/abstract=3049943

the end of the following quarter and horizon is either the fund’s annual reported portfolio

turnover ratio or the quarter’s churn ratio (CR).12 The control variables include the following

fund characteristics: the fund’s total net assets under management (TNA), the number

of holdings in the fund’s portfolio, and value-weighted measures of the fund’s underlying

portfolio firms’ characteristics: market capitalization, book-to-market ratio, and the previous

12-month return. The control variables also include time and Lipper objective category

fixed effects. Controlling for fund characteristics, combined with the fund’s Lipper objective

category, allows us to control for different investment strategies across funds and thus, helps

to isolate the variations in investment horizons. The standard errors are two-way clustered

at the fund and quarter level.

We report the regression results for the mutual fund sample in Columns (1) and (2) of

Table II, where we alternatively employ the turnover ratio in column (1) and the churn ratio

in column (2) to capture the inverse of the fund’s investment horizon. The results in column

(1) show, consistent with our hypothesis, that funds’ reported turnover ratios are negatively

related to their ESG scores—a relation that is both statistically and economically significant

as a one standard deviation increase in the turnover ratio (0.49) corresponds to a 0.08 point

decrease in the fund-level ESG score. Consistent with the simple relation depicted in Figure

1, the relation between a fund’s horizon and the weighted-average ESG score of its portfolio

holdings appears robust to controlling for fund investment style.

In column (2) of Table II, we find similar results using the fund’s churn ratio as an

alternative measure for investor horizon. A one standard deviation increase in a fund’s

churn ratio (0.23) corresponds to a 0.08 point decrease in the ESG score of the portfolio

holdings. Since the average mutual fund ESG score is 1.06 in the sample, the impact of

the churn ratio is both economically meaningful (8% change to the mean) and statistically

significant at the 1% level. Thus, longer-horizon mutual funds, which have lower turnover

or churn ratios, seem to prefer holding a portfolio more weighted toward high-ESG stocks.

12The quarter’s churn ratio is calculated according to Equation (1), using a four-quarter moving-average
to attenuate measurement errors.

13

Electronic copy available at: https://ssrn.com/abstract=3049943

One potential explanation for the observed relation between our measures of investor

horizon and fund ESG score is that it could be driven by a subset of mutual funds that have

investment mandates requiring them to invest in high-ESG stocks. Since the investment

universe for these ESG-only funds is more constrained, they may trade less frequently as a

result. To test the validity of this alternative explanation, in columns (3) and (4) of Table

II we report the results when we include an SRI indicator variable in the regression and

interact the SRI indicator with the respective investment horizon measures used in columns

(1) and (2).

The results show that the relation between investor horizon and portfolio company ESG

scores is not driven by the SRI funds in the sample. In both columns the coefficients on

the turnover and churn ratios are negative and significant, and after separating out SRI

funds, the magnitudes in columns (3) and (4) show little qualitative change from columns

(1) and (2). This result should not be surprising given that SRI funds represent such a small

percentage of the sample (2% as shown in Table I). Unsurprisingly, the coefficient on the

SRI fund dummy by itself is positive, indicating that SRI funds, on average, hold stocks

with higher ESG scores. Moreover, the coefficient on the interaction terms of the investor

horizon proxies and the SRI fund indicator shows that the negative relation between investor

horizon and fund ESG score appears more pronounced, suggesting that SRI funds’ horizons

also affect the extent of their ESG holdings.

Additionally, in column (5) of Table II, we use the 13f institutional investor sample

and examine the relation between their investment horizons and the ESG profiles of their

equity portfolio holdings. Consistent with the mutual fund results, institutions with longer

horizons (i.e., lower portfolio churn ratios) tend to hold stocks with higher ESG scores: The

coefficient on the churn ratio is negative and significant. Overall, the evidence in Table II for

both mutual funds and 13f institutions supports our hypothesis that investors with longer

investment horizons have greater preferences for high-ESG firms as compared to investors

with shorter horizons.

14

Electronic copy available at: https://ssrn.com/abstract=3049943

B. Institutional investors’ revealed preferences for higher corporate ESG

profiles: Firm-level evidence

Since long-term investors seem to prefer stocks with higher ESG profiles, a corollary hy-

pothesis is that firms with better ESG profiles should have shareholder bases with longer

investment horizons, on average. To test this hypothesis, we conduct empirical analyses at

the firm-year level: for each stock in our sample, we examine the relationship between the

firm’s ESG score and the value-weighted average investment horizon of its (mutual fund or

13f institution) shareholders.

InvestorHorizoni,t+1 = αt + β1ESGi,t + β2Controlsi,t + �i,t, (5)

where controls for firm i in time t include firm i’s market capitalization, book-to-market,

dividend yield, profitability, past 12-months return, return volatility and stock turnover

rates. We also control for industry (2-digit SIC) fixed effects to account for differences in

ESG profiles across industries. The results reported in Table III support our hypothesis.

In column (1), the investment horizon of a firm’s mutual fund shareholders is proxied by

the weighted-average turnover ratio of these funds. We find a significant negative relation

between this shareholder-average turnover ratio and a firm’s ESG score, which indicates

that firms with better ESG profiles tend to have investor bases with longer investment

horizons. A one standard deviation increase in a firm’s ESG score (2.32) is associated with a

decrease in the firm-level weighted-average fund shareholder turnover ratio of 0.51 percentage

point. Similarly, Column (2) reports that high-ESG firms, on average, have mutual fund

shareholders with significantly lower churn ratios, indicating a longer investment horizon.

One possible explanation for these results is that some mutual funds have relatively stable

fund flow patterns with less volatile inflows and outflows, giving them an appearance of

being long-term investors when in fact it simply derives from their underlying investor base.

To control for the influence of the mutual fund end investors’ purchases and redemptions,

15

Electronic copy available at: https://ssrn.com/abstract=3049943

in columns (1) and (2), we include a variable for the flow volatility of the firm’s mutual

fund shareholders. The inclusion of flow volatility supports the hypothesis that the relation

between investor horizon and corporate ESG profile is driven by fund managers’ trading

decisions, rather than the investment decisions of the funds’ end investors, which would be

reflected in the fund flows.

In column (3) of Table III, we regress the weighted-average churn ratio of a firm’s 13f

institution shareholders on corporate ESG scores. Consistent with our findings for mutual

fund shareholders, the coefficient indicates a significant (at the 1% level) and negative relation

between a firm’s ESG score and its 13f investors’ horizons.

We also employ the Bushee (1998) classification as an alternative measure of investor

horizon for 13f institutions. Both dedicated investors and quasi indexers have relatively

lower portfolio turnovers, while transient investors are characterized as short-term investors.

Hence, we employ the ratio between transient investors and all 13f institutions (TRA/13fOwn)

as the dependent variable. Our hypothesis predicts a negative relation between a firm’s ESG

score and the ratio of transient investors in the firm’s shareholder base. Consistent with our

hypothesis, column (4) of Table III shows a significantly negative relation exists between a

firm’s ESG score and the percentage ownership share of the transient investors. Thus, firms

with better ESG profiles have relatively more dedicated investors and quasi-indexers than

transient investors as shareholders, consistent with our results using institutional investor

churn ratios.

To conduct a deeper analysis of the firm-level results, we next examine several alterna-

tive measures of a firm’s ESG profile. First, we split the MSCI ESG measure into its two

major components, the firm’s ESG strength score and its ESG concern score.13 Columns

(1) through (4) of Table IV show the relation between the various shareholder horizon mea-

sures and firms’ ESG strengths and concerns. All four regressions indicate that the negative

13Interestingly, these two measures are positively correlated (31%), perhaps because many of the strengths
and concerns offset each other and firms seek good ESG performance in some indicators in order to com-
pensate for poor performance in others.

16

Electronic copy available at: https://ssrn.com/abstract=3049943

relation between ESG profiles and investor horizons is mostly driven by ESG strengths as

opposed to ESG weaknesses. Only in column (2), where we employ the average of the fund

shareholders’ churn ratios as the horizon proxy, do we also find a marginally significant

negative relation between shareholder horizon and ESG concerns. Thus, overall the results

suggest that ESG strengths better capture the considerations of long-term investors.

One concern that could arise about our baseline results with aggregate ESG scores is

that they are simply driven by the G (governance) quality of the company, and that we

are just documenting that certain institutions prefer better-governed firms. To address this

potential explanation, we remove all governance indicators as well as the business-function

related indicators that can result in firm exclusions (e.g., whether a firm operates in sin

industries). Columns (5) to (8) of Table IV show that our results remain both statistically

and economically significant for an ”ES” as opposed to an ESG measure. This result suggests

that the long-term institutional investors care about the environmental and social aspects

of corporate performance.

Finally, because some recent studies document considerable disagreement across different

ESG third-party ratings (Berg et al. (2020), Gibson et al. (2020)), we repeat our analyses

using Sustainalytics ESG scores in order to mitigate the concern that the relationship we

document is driven by idiosyncrasies of MSCI ESG scores. As mentioned previously, the

Sustainalytics scores have more limited coverage and a more limited sample period. Con-

sequently, in order to compare the results across the data from Sustainalytics and MSCI,

in this analysis we constrain our sample to the 2009-2017 period with only companies that

have both Sustainalytics and MSCI ESG scores. Table V shows the results using the two

types of scores side-by-side. Regardless of which ESG score we use, the negative relationship

between the average investment horizon of a firm’s investor base and its ESG profile is ro-

bust and significant. In terms of economic magnitude, a one standard deviation increase in

a firm’s Sustainanalytics score (10.4) is associated with a decrease of its average shareholder

turnover ratio of 0.38 percentage point. Moreover, this magnitude is quite similar to the

17

Electronic copy available at: https://ssrn.com/abstract=3049943

economic impact of the MSCI ESG scores measured earlier over the whole sample period at

0.51 percentage point.

In summary, our baseline empirical tests provide evidence that firms with higher ESG

scores attract mutual funds and 13f institutional investors with longer-term horizons. These

findings support the hypothesis that heterogeneity exists in investors’ preferences regarding

corporations’ ESG profiles and that the heterogeneity depends critically on investor horizons.

C. Identification from FTSE4Good US Index rebalances

Our finding of a significant positive association between a firm’s ESG profile and the in-

vestment horizon of the firm’s institutional investors is subject to an endogeneity concern

because there could exist unobserved firm characteristics that both attract long-term in-

vestors and correlate with a firm’s ESG performance. To better identify causality between

a firm’s corporate ESG profile and its investors’ horizons, we examine a set of shocks that

change only the (perceived) ESG standings of companies: the bi-annual FTSE4Good US

Select Index rebalances. Launched in 2001, the FTSE4Good Index Series is described as “a

series of benchmark and tradable indices for ESG (Environmental, Social and Governance)

investors.”14 FTSE-Russell develops proprietary criteria regarding the ESG performance

of companies using over 300 indicators. Firms are assigned ESG ratings from 0 (worst) to

5 (best) based on those criteria and those firms with an ESG rating of 3.1 or above are

included in the FTSE4Good Index, pending other requirements.15 A firm can be added to,

or deleted from, the Index either because its ESG profile is deemed by FTSE-Russell as suit-

able (unsuitable) for the Index, or because the firm is included or excluded from the wider

“universe” index, which is the FTSE US Index. We include only index rebalance events that

concern whether or not a firm meets the ESG-related criteria. This procedure results in 153

inclusion events and 115 exclusion events over the period of 2003 to 2015.

14FTSE4Good Index Series Factsheet. Available at
http://www.ftse.com/Analytics/FactSheets/temp/cfba7c47-6297-4c61-aa33-c550a79d0e93.pdf

15For a more complete description of FTSE4Good’s methodology, please visit
http://www.ftse.com/products/indices/FTSE4Good.

18

Electronic copy available at: https://ssrn.com/abstract=3049943

Using the FTSE4Good Index semi-annual rebalance events as shocks to corporate ESG

profiles, we test our hypotheses regarding investor horizons and ESG preferences and examine

how long-term investors adjust their portfolios in comparison to other investors. If long-term

investors indeed have a stronger preference for firms with better ESG profiles, as our earlier

results suggest, we expect these investors to respond more strongly to the Index rebalances

than other investors. Thus, we test whether investors react to a shock to firms’ public ESG

profiles by examining how they change their portfolio holdings following a stock’s inclusion to,

or exclusion from, the Index, and whether the responses vary across investors with different

horizons. To be clear, FTSE4Good Index inclusions and exclusions may or may not represent

shocks to a firm’s actual ESG profile. If new revelations of a firm’s ESG conduct promptly

changes the firm’s membership in the FTSE4Good Index, then rebalances represent shocks

to the firm’s actual ESG profile. Alternatively, the rebalances may represent shocks to

investors’ perceptions of a firm’s ESG profile, to the extent that the investors may give more

credence to a firm’s ESG profile when it is certified by a third party such as FTSE-Russell.

If FTSE4Good Index rebalances only result in a shock to perceptions of a firm’s ESG profile,

the investor responses that we estimate in our tests may be a lower bound for the impact of

actual ESG shocks.

We extract the holdings of mutual funds and 13f institutions for firms that are newly

included into, or excluded from, the Index. For both mutual funds and 13f institutions, we

define long-term investors as investors whose trailing four-quarter portfolio churn ratios fall

below the 30th percentile at the end of quarter t-2 relative to the rebalance events, where

we set the quarters so that the rebalances take place between the end of quarter t-1 and

the end of quarter t. We then examine the portfolio ownership of long-term investors and

other investors during the event quarters t-2, t-1, t, and t+1. The first two quarters are the

“before” quarters and the latter two quarters are the “after” quarters.

We then conduct difference-in-differences analyses both at the investor-level and at the

firm-level. At the investor-level analysis, the following specification is examined for investor

19

Electronic copy available at: https://ssrn.com/abstract=3049943

i of firm j:

Holdingsi,j,t
SharesOutj

= αi,t + β1LongTermi,t ∗Afterj,t + β2LongTermi,t + β3Afterj,t + �i,j,t, (6)

where LongTerm is a dummy variable indicating that investor i is classified as a long-

term investor, or zero otherwise, and After is a dummy variable indicating the two quarters

following the inclusion or exclusion event. To absorb heterogeneity at the investor level (e.g.,

differences in portfolio size), investor-by-event fixed effects are included, so the base-level of

LongTerm is subsumed.

To assess the aggregate effect of long-term investors’ ESG preferences, we further ag-

gregate the portfolio holdings across investors of the same type (long-term versus other

investors). To this end, we combine the observations such that for each stock-quarter, there

are only two observations: the aggregate ownership of long-term investors for firm j and the

aggregate ownership of other investors for firm j. The following equation is then estimated:

∑
i∈Itypekj

Holdingsi,j,t

SharesOutj
= α + β1LongTermk,t ∗Afterj,t + β2LongTermk,t + β3Afterj,t + �i,j,t,

(7)

where k ∈ {LongTerm,Other}. LongTerm takes the value of one if k = LongTerm, or

zero otherwise, and After is a dummy variable indicating the two quarters following the

inclusion or exclusion event.

In both models, standard errors are clustered at the event level. The coefficient of interest

is β1, which under our hypothesis should be positive for inclusion events and negative for

exclusion events. The identifying assumption is that, other than firms’ ESG reputations,

FTSE4Good Index inclusion and exclusion decisions do not correlate with firm characteristics

for which long-term and other investors may have different preferences.

We run regressions separately for index inclusions and exclusions. The results for inclu-

sion events are shown in Panel A of Table VI. At the investor-level, for the index inclusions,

column (1) shows that an average long-term mutual fund increases its ownership in stock j

20

Electronic copy available at: https://ssrn.com/abstract=3049943

by 0.002 percentage point after the inclusion, relative to other mutual funds. Column (2)

shows that an average long-term 13f institution increases its ownership by 0.024 percentage

point after index inclusions, relative to other 13f institutions. Both increases are marginally

significant at the 10% level.

The increased ownership of individual long-term investors also changes the aggregate

ownership structure of the event stocks. In columns (3) and (4) of Table VI, we find that at

the stock-level, the aggregate long-term mutual fund ownership increases by 0.23 percentage

point as compared to the ownership by other mutual funds. This represents a 3% increase

from the sample-average long-term mutual fund ownership. The long-term institutional

ownership increases by 1.33 percentage points, a 4.1% increase from the sample-average long-

term institutional ownership of 31.8 percentage points. These results are consistent with our

prediction that shocks that increase a stock’s ESG reputation make it more attractive to

investors with long-term horizons.

In Panel B of Table VI we report the results from an examination of events in which

stocks are excluded from the FTSE4Good US Select Index because of deterioration in their

ESG profiles. Parallel to our findings for index inclusions, long-term investors decrease their

holdings in newly excluded stocks relative to other investors. This is particularly true for

mutual fund investors, both at the investor-level and at the stock level (columns (1) and

(3)). For example, on aggregate, long-term mutual funds shed 0.358 percentage point of

ownership after an exclusion event, relative to other investors. On the other hand, long-

term 13f institutions do not respond to index exclusions, as their change in ownership is not

significantly different from zero (columns (2) and (4)).

Our identification assumption for using FTSE4Good Index rebalances as a laboratory

is that, other than firms’ ESG reputations, the index inclusion and exclusion decisions do

not correlate with firm covariates on which long-term investors and other investors have

different preferences. While there are no obvious alternative characteristics other than ESG

that FTSE-Russell takes into account in making these ESG-related rebalance decisions,

21

Electronic copy available at: https://ssrn.com/abstract=3049943

we conduct a triple-difference test with a matched control sample to further tighten our

identification strategy. The idea is to construct a sample of control stocks that have similar

observables as the inclusion (exclusion) stocks. If the long-term investor ownership does not

change around rebalances for these control stocks, it implies that the changes in long-term

ownership for inclusion or exclusion stocks are induced by shocks to a firm’s ESG profile.

The control sample is constructed as follows. For each inclusion stock, we first match it

with stocks that are not in the FTSE4Good Index before the rebalance event. For an exclu-

sion stock, we match it with stocks that are already in the FTSE4Good Index. Candidate

matched stocks must have the same two-digit SIC industry code as the inclusion or exclu-

sion stock (called the “treated” stock). We then sort the candidate stocks on the differences

between their market capitalization and the treated stock’s market capitalization. This gen-

erates a “market cap rank,” where the candidate stock with rank = 1 has the closest market

capitalization with the treated stock. We conduct the same ranking practice with respect to

the past 12-month stock returns prior to the rebalance for candidate matches, and generate

a “return rank.” The stock with smallest sum of market cap rank and return rank for each

treated stock enters into the control sample.

Panel A of Table VII displays the characteristics of the treated stocks and control stocks.

Due to the lack of availability of sufficient matches, 148 inclusions and 112 exclusions are

included in this analysis (as compared to the original 153 and 115 events). The market cap-

italization and past 12-month returns are statistically indistinguishable between the treated

group and the control group.

In Panel B of Table VII, we report the results from the triple-difference test conducted

to delineate the changes from the FTSE4Good rebalances on long-term investor ownership

22

Electronic copy available at: https://ssrn.com/abstract=3049943

between treated stocks and control stocks.

∑
i∈Itypekj

Holdingsi,j,t

SharesOutj
= α + β1LongTermk,t ∗Afterj,t ∗Treatedj + β2LongTermk,t

∗Afterj,t + β3Afterj,t ∗Treatedj + β4LongTermk,t ∗Treatedj

+ β5LongTermk,t + β6Afterj,t + β7Treatedj + �i,j,t, (8)

where k ∈ {LongTerm,Other}. The prediction is that β should be positive for index

inclusions and negative for index exclusions.

Columns (1) and (2) of Table VII, Panel B show that, after an inclusion event, the

ownership of long-term investors increases significantly relative to the ownership of other

investors, even after taking into account how their ownership changes for control stocks with

similar characteristics. For example, in column (1), long-term mutual fund investors increase

their ownership by 2.04 percentage points and β is significant at the 1% level. Similarly,

after a stock is excluded from the FTSE4Good Index, the aggregate ownership of long-term

investors drops, using the ownership of control stocks as a benchmark. Given the similarity

between treated stocks and control stocks, we are relatively confident that the responses by

long-term investors in changing their portfolio holdings is attributed to long-term investors’

consideration for firms’ ESG profiles.

Taken together, the analysis in this subsection shows that long-term investors respond

differently from other investors following shocks to firms’ ESG profiles. The discrete changes

in corporate ESG profiles as captured by FTSE4Good Index rebalances allow us to obtain

sharper results about whether investors’ horizons affect their preferences for corporate ESG.

In general, we find that as compared to other investors, long-term investors tend to increase

their holdings after a firm’s ESG profile improves and decrease their holdings after a firm’s

ESG profile deteriorates.

23

Electronic copy available at: https://ssrn.com/abstract=3049943

IV. Corporate ESG profiles and investor

short-termism

If our findings that high-ESG firms tend to attract more long-term-oriented investors are

due to these investors’ beliefs in the long-run value of high ESG firms, then an important

corollary implication is that we should observe less investor short-termism with respect to

these firms. That is, for a given mutual fund with a long-term orientation, does the manager

act more patiently toward the portfolio’s high-ESG firms relative to its low-ESG firms?

A. Long-term investor patience towards high-ESG firms

To test the hypothesis that long-term investors are more patient with the high-ESG firms

in their portfolios as compared to the low-ESG firms, we examine fund managers’ selling

decisions and whether the sensitivity of their selling with respect to firm performance varies

with the ESG profiles of their portfolio companies. That is, we test whether the relation

between firm underperformance and fund managers’ trading decisions becomes more muted

in the presence of a high corporate ESG profile. We test this hypothesis using two alternative

measures of firm performance, stock returns and earnings surprises.

The first set of regressions we run associate the funds’ selling decisions with the firms’

past 12-month excess returns. The observations are at the fund-stock-quarter level and the

standard errors are clustered by stock. The dependent variable is a dummy indicating that

fund i decreases its holding in firm j between period t− 1 and t. Since the emphasis of this

section focuses on the different levels of “patience” toward different companies held by the

same fund, all specifications include fund-quarter fixed-effects.

Dummy(Sell)i,j,t = αi,t + β1Min(0,ExcessReturnj,t)+

β2ESGj,t ∗Min(0,ExcessReturnj,t) + β3Controlsj,t + �i,j,t
(9)

where ESGj,t is the MSCI ESG score of stock j at time t and the control variables include

24

Electronic copy available at: https://ssrn.com/abstract=3049943

the firm’s return volatility, stock turnover, ln(market capitalization), book-to-market ratio,

and dividend yield. Since we are interested in mutual funds’ selling decisions following poor

stock performance, Min(0,ExcessReturnj,t) measures the negative component of a stock’s

12-month excess return relative to the market return, where the raw returns are summed

over the previous 12 months. The intercept, αi,t, indicates fund-by-time fixed effects.

The results from this regression are reported in Table VIII. As expected, the coefficient

β1 in column (1) shows that a strong inverse relation exists between recent negative stock

returns and the probability of selling a position: a 10 percentage point decrease in the stock

excess returns raises the probability of a position being sold by 1.25 percentage points. The

coefficient of primary interest in this regression is β2, which captures the differential effect of

past return on selling decisions between high and low ESG firms. According to our hypothesis

β2 should have the opposite sign to β1, which would indicate that a high-ESG profile reduces

the probability that the mutual fund manager will sell the poorly performing stock. Results

from Table VIII support this claim, as β2 has the opposite sign of β1. The result suggests that,

given poor recent stock performance, a firm with a higher ESG score is less likely to be sold by

a mutual fund manager as compared to another firm in that manager’s portfolio with a lower

ESG score. If a firm’s ESG score increases by one standard deviation (2.32), the sensitivity

between past poor returns and the mutual-fund selling decision is attenuated by 8% relative

to the baseline level (0.0044*2.32/0.125). This result suggests that fund managers are more

patient towards firms with well-established ESG profiles, i.e., they exhibit less short-termism.

The finding further strengthens when controlling for unobserved firm characteristics using

stock fixed effects (column (2)).

We also employ a variable that captures a fund’s selling through a more continuous

methodology. In columns (3) and (4) of Table VIII, the dependent variable is (the negative

of) changes in fund i’s fractional trade of stock j in quarter t (−∆Holdingsi,j,t
Holdingsi,j,t−1

). A similar

pattern is found to the previous dichotomous method: mutual funds’ selling trades are

positively related to the past return of the stocks (β1 < 0). For example, a 10 percentage

25

Electronic copy available at: https://ssrn.com/abstract=3049943

points decrease in excess stock returns is associated with a 4 percentage points decrease

in the fund’s holdings of a particular stock. However, when a firm has a high-ESG score,

this relation is weakened (β2 > 0). The interaction effect is statistically significant. Again,

mutual fund managers seem to treat high-ESG firms with more patience.

Another widely-followed measure of a company’s short-term performance is the quarterly

earnings outcome, especially relative to analysts’ expectation. Evidence suggests that U.S.

public company managers face pressures to meet or beat the firm’s earnings expectations

(Graham, Harvey, and Rajgopal (2005)). Such emphasis on meeting short-term goals al-

legedly causes managers to expend energy and time managing earnings, hence distracting

them from investing in long-term, valuable projects. Having a good ESG profile, however,

may convince investors that the firm has a plan to create shareholder value in the long-

run. The investors are then more likely to tolerate earnings shortfalls and stick with their

investments.

Using data on earnings and analyst forecasts from the IBES dataset, we measure earnings

surprise in two ways. The first measure is the firm’s seasonal-adjusted earnings growth, de-

rived from the Compustat earnings per share item, scaled by current stock prices (
Xj,t−Xj,t−4

Pj,t
,

where Xj,t is the Compustat-based earnings per share before extraordinary items). Livnat

and Mendenhall (2006) conclude that since EPS follows a seasonal random walk, the seasonal-

adjusted earnings growth is the best proxy for the “surprise” component. To construct the

second measure of earnings surprise, we calculate the difference between the firm’s actual

earnings and the median analyst forecast of those earnings, scaled by the firm’s share price.

We run a set of regressions with fund-quarter fixed-effects (to isolate the selling decisions

for a given mutual fund) as follows:

Dummy(Sell)i,j,t = αi,t + β1EarningsShortfallj,t+

β2ESGj,t ∗EarningsShortfallj,t + β3Controlsi,t + �i,j,t,
(10)

where EarningsShortfall is based on the shortfall of announced earnings. We use two

26

Electronic copy available at: https://ssrn.com/abstract=3049943

measures for EarningsShortfall, either: (1) a dummy variable that indicates a negative

earnings surprise, that is, Dummy(EarningsSurprise < 0); or (2) the actual magnitude of

a negative earnings surprise, i.e., Max(−EarningsSurprise, 0). ESGj,t is the MSCI ESG

score of stock j at time t and the control variables include the firm’s return volatility, stock

turnover, ln(market capitalization), book-to-market ratio, and dividend yield.

Similar to the expectation for poor stock performance, the expectation regarding recent

earnings is that although fund managers are more likely to sell their holdings following

earnings shortfalls (β1 > 0), if a firm has a high-ESG profile, the relation between earnings

shortfall and funds’ selling is weakened (β2 < 0). The results in Table IX support this

hypothesis. Column (1) of Table IX shows, conditional on a negative earnings surprise, a

given mutual fund is more likely to sell its position in the company by 1.09 percentage points.

When a company has a higher ESG score, however, that same mutual fund is significantly

less likely to sell following an earnings shortfall. In fact, if a company increases its ESG score

by one standard deviation (2.32), the sensitivity between the earnings shortfall and mutual-

fund selling will decrease by 17% relative to the baseline level (2.32*0.000789/0.0109). When

the earnings shortfall is measured by the actual magnitude of the earnings surprise, the ESG

score similarly has a mitigating effect on the sensitivity of trading on earnings (column

(2)). If we change the dependent variable to the fractional trading of mutual fund positions

(columns (3) and (4)), the results are again consistent in showing that mutual fund managers

appear to be more patient in holding their positions in a high-ESG firm following adverse

earnings announcements.

Columns (5)-(8) of Table IX repeat the analyses using deviations from the median analyst

forecast as proxies for earnings surprises. The results are qualitatively and quantitatively

similar to the results in the other columns: mutual fund managers tend to sell stocks following

negative earnings surprises; having a high-ESG score mitigates the relation between earnings

shortfalls and selling. Fund managers seem to base their trading decisions less on recent

negative earnings announcements if the firm has a strong ESG score. To corporate executives

27

Electronic copy available at: https://ssrn.com/abstract=3049943

concerned about pressure from their investors’ myopia, the dampening effect of the corporate

ESG profile on the earnings sensitivity of trading may present improving ESG profile as an

appealing approach in maintaining their investor relationships.

To summarize, the findings in this section complement what we document in the previous

section: not only do high-ESG firms attract the type of institutional investors that are long-

term focused, investors also act more patiently towards high-ESG firms relative to other firms

in their portfolios. In particular, mutual funds are less likely to sell a firm following poor

stock performance or earnings shortfall if the firm has a high-ESG profile. To the extent that

institutional investors’ emphasis on firms meeting or beating earnings expectations drives the

alleged corporate “myopia”, better ESG profiles may free corporate managers to undertake

long-term projects.

B. Investors’ patience before and after FTSE4Good US Index rebalances

The investor patience hypothesis also should be reflected in changes around the FTSE4Good

Index rebalance events. In particular, the hypothesis implies that institutional investors

should become more patient after a stock is included into the FTSE4Good Index or less

patient for one that is excluded from the Index. Since we can compare the same stock

in a relatively short time window, four quarters before and after the rebalance event, this

setting gives us further identification for tests in which we attribute the changes of investors’

trading patterns to the shock of changes in (perceived) corporate ESG profiles. Thus, we test

whether selling sensitivities to negative earnings surprises change for a stock newly included

into, or excluded from, the Index. We adopt the following empirical framework

Dummy(Sell)i,j,t = αj + β1EarningsShortj,t+

β2FTSEj,t ∗EarningsShortj,t + β3FTSEj,t + �i,j,t,
(11)

where EarningsShort is a dummy indicating firm j’s earnings surprise is negative during

quarter t and FTSEj,t is an indicator variable for whether firm j is newly included in the

28

Electronic copy available at: https://ssrn.com/abstract=3049943

FTSE4Good Index during quarter t. The specification includes event-stock fixed effects.

This ensures that we compare the same stock before and after its inclusion or exclusion

event.

Since both inclusion and exclusion events are in the sample, the dummy variable FTSE

is defined in the following way: For stocks that are added to the FTSE4Good US Index at

time t, quarters t− 4 to t− 1 are defined as non-FTSE periods, and quarters t + 1 to t + 4

are defined as FTSE periods. For stocks that are newly excluded from the Index at time t,

quarters t + 4 to t + 1 are defined as non-FTSE periods, while quarters t − 1 to t − 4 are

defined as FTSE periods. By defining FTSE periods in this way, there is a clear prediction

on the sign of the coefficient β2 from the above equation. Since FTSE periods are associated

with times in which the firm’s ESG reputation is expected to be relatively higher, one would

expect mutual funds’ selling to be less sensitive to earnings shortfalls during these periods.

Hence, β2 is predicted to be negative. Note that we exclude quarter t, the quarter when

the changes are announced, in order to avoid trading activities directly associated with the

initial information about index inclusions and exclusions.

Table X displays the results. In columns (1) and (2), earnings surprises are measured by

the firms’ seasonal-adjusted earnings growth rates. If the dependent variable (column (2))

is the fraction trading from the last quarter, a negative earnings surprise is associated with

an average selling of 3.15 percentage points of last-quarter holdings. After rebalance events,

however, this sensitivity is decreased by 0.75 percentage points. This difference is marginally

significant at the 10% level.

In columns (3) and (4) of Table X we measure earnings surprise using the difference be-

tween actual announcement earnings and the median analyst forecast. In this case, a negative

earnings surprise increases mutual funds’ propensity to sell by 3.24 percentage points. After

the rebalance events, however, the selling propensity drops by 1.01 percentage points, which

is statistically significant at a marginal level, but highly economically significant. Crucially,

even though fund managers, on average, still decide to sell some of the stock, the amount

29

Electronic copy available at: https://ssrn.com/abstract=3049943

sold is significantly less. An ESG-enhancing FTSE rebalance event decreases the average

fraction of shares sold following negative earnings surprises from 4.2 percentage points to

2.75 percentage points (0.042−0.0145). This differential in trading sensitivity is statistically

significant beyond the .001 level.

These findings suggest a well-identified relation between companies’ ESG profiles and

mutual fund managers’ propensity to sell following an earnings shortfall. Consistent with

our hypothesis, mutual fund managers appear to be more (less) patient towards firms whose

ESG profiles improve (deteriorate).

V. Changes in investor preferences for corporate ESG

over time

Thus far, we have demonstrated that, on average, long-term investors tend to prefer high-

ESG stocks. We next consider whether aggregate institutional ownership in firms in high

or low ESG firms has changed over time given the recent emphasis on using ESG as an

investment criterion in recent years. For example, since its founding in 2006, the UN-backed

Principles for Responsible Investment (PRI) has grown from 100 signatories to 3,038 and

$103.4 trillion in assets under management by 2020.16

As a preliminary analysis of aggregate relationships, we first sort the sample firms into

quartile groups at the end of 2003 by their total net MSCI ESG score.17 We then track the

mutual fund or 13f institutional ownership level for the highest and lowest ESG score cohorts

over time. To account for the heterogeneity in firm characteristics, we further adjust the

16https://www.unpri.org/about The PRI ”works to understand the investment implications of environ-
mental, social and governance (ESG) factors and to supports its international network of investor signatories
in incorporating these factors into their investment and ownership decisions. The PRI acts in the long-term
interests of its signatories, of the financial markets and economies in which they operate and ultimately of
the environment and society as a whole.”

17We choose the beginning year to be 2003 because that was the first year that the MSCI ESG coverage
was expanded to all MSCI USA IMI Index firms. Prior to that time, the analysis was limited to the 1,000
largest US companies.

30

Electronic copy available at: https://ssrn.com/abstract=3049943

percentage (mutual fund or 13f institution) ownership of a firm by the average percentage

ownership in the 125 Daniel, Grinblatt, Titman, and Wermers (1997) (DGTW) benchmark

portfolios for that quarter.18

The upper panel of Figure 2 shows the DGTW-adjusted mutual fund ownership of the

top ESG-quartile firms and the bottom ESG-quartile firms. Over time, consistent with the

increasing public focus on ESG activities as investment criteria, we find an increasing mutual

fund preference for the higher ESG-profile firms and a decreasing preference for the lower

ESG-profile firms. At the beginning of the sample period the DGTW-adjusted mutual fund

ownership is considerably lower for high-ESG firms than for low ESG firms, but the relative

preference completely reverses later in the sample period. This result appears to reflect the

changing investment landscape during the past ten to fifteen years in which ESG investing

has attracted increasing mainstream attention and has become more widely accepted and

practiced. For example, Bialkowski and Starks (2018) show increasing media attention to

ESG issues over time and increasing relative flows to SRI funds. Hartzmark and Sussman

(2019) show that when Morningstar started rating mutual funds by the sustainability of

their holdings, the funds with the highest sustainability rankings received the most flows. In

the lower panel of Figure 2, we plot the average DGTW-adjusted 13f institutional ownership

for the top and bottom ESG-quartile portfolios. The general pattern is similar: high-ESG

firms have a lower 13f institutional ownership at the beginning of our sample; the gap in

institutional ownership gradually diminishes and reverses as time goes by.

To more systematically test whether investors’ preferences for ESG have significantly

changed over time, we conduct firm-level panel regressions, controlling for other firm char-

acteristics that may affect ownership:

InvestorOwnershipj,t+1 = αt + β1ESGj,t + β2Controlsj,t + �j,t, (12)

18In each quarter, we sort eligible stocks into 5 by 5 by 5 groups sorted by market capitalization, book-to-
market ratio, and past 12-month return momentum. We calculate the average percentage ownership within
each bucket, and then subtract the bucket-mean from the corresponding percentage ownership of a firm.
(See Daniel, et. al., (1997) for details on their benchmark portfolios.)

31

Electronic copy available at: https://ssrn.com/abstract=3049943

where InvestorOwnership represents several different measures of investor ownership. The

first measure we use, Ownershipj,t+1, is the percentage ownership held by a mutual fund or

13f institution in firm j at time t + 1, adjusted for the DGTW benchmark portfolios. We

also use the change in ownership for firm j, and ownership breadth, defined as the number of

institutions that hold any shares of a given firm divided by the total number of institutions

that hold shares in the firm (Chen, Hong, and Stein (2002)). ESGj,t is the MSCI ESG score

of firm j at time t and the control variables include the firm’s ln(market capitalization), book-

to-market ratio, dividend yield, profitability ratio, past 12-month return, return volatility,

and stock turnover.

The results are tabulated in Table XI. The findings suggest that ESG preferences among

institutional investors are a relatively new phenomenon. Considering the average relation-

ships over the entire sample period, the results show that firms with higher ESG scores have

lower levels of mutual fund and 13f institutional ownership (columns (1) and (5)). The neg-

ative preference could be attributable to unobserved firm characteristics not captured by the

control variables or it could be attributable to different interpretations by investors during

the early years of our sample as to the benefits and costs of corporate social responsibility.19

This negative preference for ESG, however, has been changing rapidly over time. When

we focus on investor holdings before and after 2010 (columns (2) and (6)) the difference

in actively managed mutual fund and 13f institutional investor preferences for low versus

high-ESG firms has all but disappeared. In column (3), consistent with the trends shown

in the previous figure, we document that the year-over-year change in active mutual fund

ownership of high-ESG firms is positively associated with the firm’s ESG score. This is

also consistent with the graphical evidence we present in Figure 2, where the ownership gap

between the high- and low-ESG profile of the underlying firms diminishes over time and, in

the case of mutual fund ownership, reverses.

19Ioannou and Serafeim (2015) argue that equity analysts viewed corporate social responsibility (CSR) as
an agency cost in earlier years. Similar to our findings regarding the institutional ownership gap, the authors
find that the analysts’ pessimistic perception gradually reverses.

32

Electronic copy available at: https://ssrn.com/abstract=3049943

Finally, we show that for both mutual funds and 13f institutions, ownership breadth is

positively associated with a firm’s ESG score (columns (4) and (8)). It suggests that firms

with higher ESG performances tend to attract a wider set of investors.

VI. Conclusions

Investors have expressed growing interest over time in firms’ actions with regard to ESG

issues. This interest has been spurred at least in part by concerns about the long-term

effects of firm actions and an emphasis to reject a focus on short-term profits at the expense

of longer-term wealth. In this paper, we provide comprehensive empirical evidence regarding

the revealed preferences of mutual funds and institutional investors for high-ESG firms and

how these preferences are related to the investors’ horizons.

When corporate ESG policies are designed to improve shareholder value in the long-run,

perhaps at the expense of current earnings, long-term oriented investors would be more

likely to place higher value on the firms with better ESG profiles. In both investor-level

analyses and firm-level analyses, we find a positive relation between investors’ horizons and

their preferences for high-ESG stocks. In the cross-section, firms with stronger ESG profiles

attract shareholder bases that are on average more long-term. These long-term investors

turn over their portfolio more slowly, which would conceivably allow them to capture any

value created by corporate ESG policies. Our results imply a match between the horizon of

investors and the horizons of their portfolio firms.

By examining shocks to the perceptions of firms’ ESG profiles, we provide evidence of

causality in the relation between firms’ ESG profiles and their institutional shareholders’

preferences. The inclusion into (exclusion from) the FTSE4Good Index induces differen-

tial responses from long-term investors and other investors, supporting the importance of

investment horizon in ESG investing.

An implication of our results is a potential increasing pressure on firms to consider and

33

Electronic copy available at: https://ssrn.com/abstract=3049943

improve their ESG profiles. That is, companies who would like to attract investors with

longer-term orientations could do so by improving their ESG performance. Since having

a longer-term oriented shareholder base is often claimed to be desirable, companies may

have strong incentives to do “good” to have the “right” investors. Our results also have

important implications for the future economy if one believes that the short-termism of

some institutional investors holds back corporate innovations and investments and has social

costs for financial markets and the economy as argued by a number of commentators (e.g.,

Keynes (1935); Lipton (1979); Kay (2012)).

34

Electronic copy available at: https://ssrn.com/abstract=3049943

References

Akbas, Ferhat, Chao Jiang, and Paul D. Koch, 2020, Insider investment horizon, Journal of

Finance 75, 1579–1627.

Albuquerque, Rui, Yrjo Koskinen, and Chendi Zhang, 2019, Corporate social responsibility

and firm risk: Theory and empirical evidence, Management Science 65, 4451–4469.

Baker, Steven D., Burton Hollifield, and Emilio Osambela, 2020, Asset prices and portfolios

with externalities, Unpublished working paper.

Baron, David P, 2001, Private politics, corporate social responsibility, and integrated strat-

egy, Journal of Economics & Management Strategy 10, 7–45.

Baron, David P, 2008, Managerial contracting and corporate social responsibility, Journal

of Public Economics 92, 268–288.

Bénabou, Roland, and Jean Tirole, 2010, Individual and corporate social responsibility,

Economica 77, 1–19.

Benson, Karen L, and Jacquelyn E Humphrey, 2008, Socially responsible investment funds:

Investor reaction to current and past returns, Journal of Banking and Finance 32, 1850–

1859.

Berg, Florian, Julian F. Kolbel, and Roberto Rigobon, 2020, Aggregate confusion: The

divergence of ESG ratings, Working paper, Massachusetts Institute of Technology.

Bialkowski, Jedrzej, and Laura T Starks, 2018, SRI funds: Investor demand, exogenous

shocks and ESG profiles, Working paper, University of Canterbury and University of

Texas at Austin.

Bolton, Patrick, and Marcin Kacperczyk, 2020, Do investors care about carbon risk?, Na-

tional Bureau of Economic Research working paper.

Bolton, Patrick, Jose Scheinkman, and Wei Xiong, 2006, Executive compensation and short-

termist behaviour in speculative markets, The Review of Economic Studies 73, 577–610.

Borghesi, Richard, Joel F Houston, and Andy Naranjo, 2014, Corporate socially responsible

investments: CEO altruism, reputation, and shareholder interests, Journal of Corporate

35

Electronic copy available at: https://ssrn.com/abstract=3049943

Finance 26, 164–181.

Bushee, Brian J, 1998, The influence of institutional investors on myopic R&D investment

behavior, Accounting Review 32, 305–333.

Cella, Cristina, Andrew Ellul, and Mariassunta Giannetti, 2013, Investors’ horizons and the

amplification of market shocks, Review of Financial Studies 26, 1607–1648.

Chava, Sudheer, 2014, Environmental externalities and cost of capital, Management Science

60, 2223–2247.

Chen, Joseph, Harrison Hong, and Jeremy C Stein, 2002, Breadth of ownership and stock

returns, Journal of Financial Economics 66, 171–205.

Daniel, Kent, Mark Grinblatt, Sheridan Titman, and Russ Wermers, 1997, Measuring mutual

fund performance with characteristic-based benchmarks, The Journal of Finance 52, 1035–

1058.

Derrien, François, Ambrus Kecskés, and David Thesmar, 2013, Investor horizons and corpo-

rate policies, The Journal of Financial and Quantitative Analysis 48, 1755–1780.

Dimson, Elroy, Oğuzhan Karakaş, and Xi Li, 2015, Active ownership, Review of Financial

Studies 28, 3225–3268.

Dimson, Elroy, Oğuzhan Karakaş, and Xi Li, 2018, Coordinated engagements, Working

paper, London Business School.

Dimson, Elroy, Idar Kreutzer, Rob Lake, Hege Sjo, and Laura Starks, 2013, Responsible in-

vestment and the Norwegian Government Pension Fund Global, Oslo: Norwegian Ministry

of Finance .

Dunn, Jeff, Shaun Fitzgibbons, and Lukasz Pomorski, 2018, Assessing risk through environ-

mental, social and governance exposures, Journal of Investment Management 16.

Dyck, Alexander, Karl V. Lins, Lukas Roth, and Hannes F. Wagner, 2019, Do institutional

investors drive corporate social responsibility? International evidence, Journal of Finan-

cial Economics 131, 693 – 714.

Eccles, Robert G, Ioannis Ioannou, and George Serafeim, 2014, The impact of corporate sus-

36

Electronic copy available at: https://ssrn.com/abstract=3049943

tainability on organizational processes and performance, Management Science 60, 2835–

2857.

Fernando, Chitru S, Mark P Sharfman, and Vahap B Uysal, 2009, Do investors want firms

to be green? Environmental performance, ownership and stock market liquidity, Academy

of Management Proceedings 2009, 1–6.

Foley, Stephen, and Ben McLannahan, 2016, Top US financial groups hold secret summits

on long-termism, Financial Times .

Froot, Kenneth A, Andre F Perold, and Jeremy C Stein, 1992, Shareholder trading prac-

tices and corporate investment horizons, Working Paper, National Bureau of Economic

Research.

Gaspar, José-Miguel, Massimo Massa, and Pedro Matos, 2005, Shareholder investment hori-

zons and the market for corporate control, Journal of Financial Economics 76, 135–165.

Gibson, Rajna, Philipp Krueger, and Peter Schmidt, 2020, ESG rating disagreement and

stock returns, ECGI Working paper.

Gibson, Rajna, and Philipp Krüger, 2018, The sustainability footprint of institutional in-

vestors, Swiss Finance Institute Research Paper and ECGI Working Paper .

Gillan, Stuart, Jay C Hartzell, Andrew Koch, and Laura Starks, 2010, Firms’ environmental,

social and governance (ESG) choices, performance and managerial motivation, Working

paper, University of Georgia, University of Pittsburgh and University of Texas at Austin.

Gloßner, Simon, 2019, Investor horizons, long-term blockholders, and corporate social re-

sponsibility, Journal of Banking & Finance 103, 78–97.

Graham, John R., Campbell R. Harvey, and Shiva Rajgopal, 2005, The economic implica-

tions of corporate financial reporting, Journal of Accounting and Economics 40, 3–73.

Guerard, John B, 1997, Is there a cost to being socially responsible in investing?, The Journal

of Investing 6, 11–18.

Hartzmark, Samuel M, and Abigail B Sussman, 2019, Do investors value sustainability? a

natural experiment examining ranking and fund flows, The Journal of Finance 74, 2789–

37

Electronic copy available at: https://ssrn.com/abstract=3049943

2837.

Heinkel, Robert, Alan Kraus, and Josef Zechner, 2001, The effect of green investment on

corporate behavior, Journal of Financial and Quantitative Analysis 36, 431–439.

Hoepner, Andreas G. F., Ioannis Oikonomou, Zacharias Sautner, Laura T Starks, and Xiao Y

Zhou, 2019, ESG shareholder engagement and downside risk, Unpublished working paper

.

Hong, Harrison, and Marcin Kacperczyk, 2009, The price of sin: The effects of social norms

on markets, Journal of Financial Economics 93, 15–36.

Ioannou, Ioannis, and George Serafeim, 2015, The impact of corporate social responsibility

on investment recommendations: Analysts’ perceptions and shifting institutional logics,

Strategic Management Journal 36, 1053–1081.

Kay, John, 2012, The Kay review of UK equity markets and long-term decision making,

Final Report 9.

Kim, Hyun-Dong, Taeyeon Kim, Yura Kim, and Kwangwoo Park, 2019, Do long-term insti-

tutional investors promote corporate social responsibility activities?, Journal of Banking

& Finance 101, 256–269.

Krueger, Philipp, Zacharias Sautner, and Laura T Starks, 2020, The importance of climate

risks for institutional investors, The Review of Financial Studies 33, 1067–1111.

Krüger, Philipp, 2015, Corporate goodness and shareholder wealth, Journal of Financial

Economics 115, 304–329.

Liang, Hao, and Luc Renneboog, 2017, On the foundations of corporate social responsibility,

The Journal of Finance 72, 853–910.

Lins, Karl V., Henri Servaes, and Ane Tamayo, 2017, Social capital, trust, and firm perfor-

mance: The value of corporate social responsibility during the financial crisis, The Journal

of Finance 72, 1785–1824.

Lipton, Martin, 1979, Takeover bids in the target’s boardroom, The Business Lawyer 101–

134.

38

Electronic copy available at: https://ssrn.com/abstract=3049943

Livnat, Joshua, and Richard R Mendenhall, 2006, Comparing the post–earnings announce-

ment drift for surprises calculated from analyst and time series forecasts, Journal of Ac-

counting Research 44, 177–205.

Massa, Massimo, Ayako Yasuda, and Lei Zhang, 2013, Supply uncertainty of the bond

investor base and the leverage of the firm, Journal of Financial Economics 110, 185–214.

Nofsinger, John R, Johan Sulaeman, and Abhishek Varma, 2019, Institutional investors and

corporate social responsibility, Journal of Corporate Finance 58, 700–725.

Oehmke, Martin, and Marcus Opp, 2020, A theory of socially responsible investing, Working

paper.

Pastor, Lubos, Robert F. Stambaugh, and Lucian A. Taylor, 2020, Sustainable investing in

equilibrium, Journal of Financial Economics, forthcoming .

Pedersen, Lasse Heje, Shaun Fitzgibbons, and Lukasz Pomorski, 2020, Responsible investing:

The ESG-Efficient frontier, Unpublished working paper.

Pozen, Robert C., 2009, Can we break the tyranny of quarterly results?, Harvard Business

Review .

Renneboog, Luc, Jenke Ter Horst, and Chendi Zhang, 2008, Socially responsible investments:

Institutional aspects, performance, and investor behavior, Journal of Banking & Finance

32, 1723–1742.

Riedl, Arno, and Paul Smeets, 2017, Why do investors hold socially responsible mutual

funds?, The Journal of Finance 72, 2505–2550.

Statman, Meir, and Denys Glushkov, 2009, The wages of social responsibility, Financial

Analysts Journal 65, 33–46.

Stein, Jeremy C, 1989, Efficient capital markets, inefficient firms: A model of myopic corpo-

rate behavior, The Quarterly Journal of Economics 104, 655–669.

Thakor, Richard, 2020, Short-termism, managerial talent, and firm value, Review of Corpo-

rate Finance Studies .

39

Electronic copy available at: https://ssrn.com/abstract=3049943

Figure 1: Portfolio Holdings’ ESG Scores and Investor Investment Horizons

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

Long Horizon
Quintile,

Turnover = 20%
Quintile 2,

Turnover = 41%
Quintile 3,

Turover = 65%
Quintile 4,

Turnover = 98%

Short Horizon
Quintile,

Turnover = 161%

W
ei

gh
te

d-
av

er
ag

e
ES

G
S

co
re

e

Portfolio Turnover Quintiles
Average Turnover

(a) Mutual funds sorted by turnover ratios

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

Long Horizon Quintile,
Churn Ratio = 10%

Quintile 2,
Churn Ratio = 19%

Quintile 3,
Churn Ratio = 25%

Quintile 4,
Churn Ratio = 42%

Short Horizon Quintile,
Churn Ratio = 92%

W
ei

gh
te

d-
av

er
ag

e
ES

G
s

co
re

Portfolio Turnover Quintiles
Average Churn Ratio

(b) 13F institutions sorted by churn ratios

This figure shows the average ESG scores of investors’ portfolio holdings with relation to their horizons. Investor
horizons are measured by the portfolio turnover ratio for mutual funds and the portfolio churn ratios for 13f
institutions. Investors with the lowest portfolio turnover ratios (churn ratios) are defined as ‘long-horizon”. Panel
(a) shows mutual funds and Panel (b) shows 13f institutions. Each mutual fund’s (13f institution’s) ESG score is
calculated as the value-weighted average ESG score of its portfolio companies.

40

Electronic copy available at: https://ssrn.com/abstract=3049943

Figure 2: Aggregate Ownership for ESG-Sorted Stock Portfolios Adjusted for DGTW Benchmark
Groups

-1
-.5

0
.5

1
M

ut
ua

l F
un

d
O

w
ne

rs
hi

p
– D

G
TW

a
dj

us
te

d
(%

)

2004 2006 2008 2010 2012 2014 2016 2018

Lowest ESG Quartile Highest ESG Quartile

(a) Mutual Fund Ownership

-2
-1

0
1

2
13

F
In

st
itu

tio
ns

O
w

ne
rs

hi
p

– D
G

TW
a

dj
us

te
d

(%
)

2004 2006 2008 2010 2012 2014 2016 2018

Lowest ESG Quartile Highest ESG Quartile

(b) 13f Institution Ownership

At the end of 2003, we sort stocks into quartile portfolios based on their MSCI ESG score, and track their fu-
ture average mutual fund ownership (Panel (a)) and 13f institutional ownership (Panel (b)) over the 2004–2018
period. The mutual fund/institutional ownership of a stock is adjusted by the average ownership within its DGTW
benchmark group each year.

41

Electronic copy available at: https://ssrn.com/abstract=3049943

Table I: Summary Statistics

Panel A: Mutual Fund-Level Variables
N=98,252 Mean Stdev P25 Median P75
Porfolio Turnover (Annual) 0.71 0.49 0.31 0.57 0.98
Churn Ratio (Quarterly) 0.37 0.23 0.20 0.32 0.48
SRI Fund Indicator 0.02 0.15 0.00 0.00 0.00
ln(Fund TNA) 5.78 1.90 4.43 5.79 7.11

Weighted-Average Values for Portfolio Stocks:
MSCI ESG Score 1.06 1.85 -0.37 0.63 2.40
ln(Market Cap) 9.87 1.65 8.40 10.59 11.28
Book-to-Market 0.50 0.20 0.36 0.47 0.59
12-Month Return 0.19 0.23 0.06 0.18 0.30

Panel B: 13f Institution-Level Variables
N=166,185 Mean Stdev P25 Median P75
Churn Ratio (Quarterly) 0.35 0.32 0.13 0.24 0.45
ln(Total Holdings value) 19.73 1.82 18.47 19.43 20.82

Weighted-average Values for Portfolio Stocks:
MSCI ESG score 1.53 1.88 0.07 1.24 3.06
ln(Market Cap) 10.78 1.16 10.28 11.16 11.59
Book-to-Market 0.52 0.30 0.37 0.45 0.58
12-Month Return 0.18 0.22 0.06 0.16 0.27

42

Electronic copy available at: https://ssrn.com/abstract=3049943

Table I: Summary Statistics (continued)

Panel C: Firm-level Variables
N=26,217 Mean Stdev P25 Median P75
MSCI ESG Score -0.03 2.32 -1 0 1
MSCI ESG Strengths 1.55 2.35 0 1 2
MSCI ESG Concerns -1.57 1.88 -2 -1 0
MSCI ESG Score Excluding Governance 0.22 2.29 -1 0 1
ln(Market Cap) 7.26 1.59 6.06 7.05 8.23
Book-to-Market 0.61 0.47 0.29 0.50 0.78
12-Month Return (%) 14.57 47.25 -13.04 9.84 34.11
Return Volatility (%) 10.77 6.22 6.40 9.17 13.32
Dividend Yield (%) 1.28 2.01 0.00 0.15 2.00
Profitability (%) 9.19 14.19 3.61 10.46 16.17
Daily Turnover (%) 0.24 0.52 0.07 0.13 0.22
Mutual Fund Ownership(%) 14.21 8.53 7.64 12.68 20.01
Mutual Fund Breadth(%) 3.53 3.86 1.11 2.33 4.57
13f Institution Ownership(%) 69.60 22.04 56.67 74.79 87.06
13f Institution Breadth(%) 7.60 7.78 3.01 4.87 9.10

Weighted-Average Values for the Firms’ Shareholders:
Mutual Fund Shareholder Turnover (%) 42.98 13.99 31.77 39.84 51.87
Mutual Fund Shareholder Churn Ratio (%) 26.77 8.87 20.24 25.35 31.95
13f Institution Shareholder Churn Ratio (%) 23.56 6.74 18.52 22.76 27.56
Transient/Total 13f Institution Ownership (%) 17.65 14.68 4.70 16.27 27.25

Panel (A) reports the summary statistics for the full sample of fund-quarters used in the mutual fund-level analyses.
Panel (B) shows the summary statistics for 13f institutions. Panel (C) shows the summary statistics for firm-years
used in the firm-level analyses. The sample period covers 2000-2018.

43

Electronic copy available at: https://ssrn.com/abstract=3049943

Table II: Investor-Level Analysis: Portfolio ESG Score and Investor Horizon

Dependent variable: Investor-level ESG Score
Sample Mutual Funds 13f Institutions

(1) (2) (3) (4) (5)
Fund Turnover Ratio -0.163∗∗∗ -0.150∗∗∗

(0.0396) (0.0389)

Fund or 13f Churn Ratio -0.322∗∗∗ -0.298∗∗∗ -0.337∗∗∗

(0.0765) (0.0754) (0.0299)

Fund Turnover Ratio * SRI Fund Indicator -0.443∗∗

(0.210)

Fund Churn Ratio * SRI Fund Indicator -0.703∗

(0.370)

Holdings ln(Market Cap) 0.652∗∗∗ 0.649∗∗∗ 0.656∗∗∗ 0.653∗∗∗ 0.665∗∗∗

(0.0385) (0.0384) (0.0383) (0.0382) (0.0466)

Holdings Book-to-Market 0.508∗ 0.519∗ 0.520∗ 0.531∗ 0.0584
(0.276) (0.276) (0.276) (0.276) (0.0648)

Holdings 12-Month Return 0.645∗∗ 0.655∗∗ 0.648∗∗ 0.659∗∗ -0.170
(0.282) (0.285) (0.282) (0.285) (0.115)

ln(Fund TNA) -0.000617 0.000274 0.00187 0.00268
(0.00743) (0.00740) (0.00740) (0.00737)

Number of Stocks in Holdings -0.000761∗∗∗ -0.000827∗∗∗ -0.000758∗∗∗ -0.000820∗∗∗ -0.000138∗∗∗

(0.000146) (0.000149) (0.000142) (0.000146) (0.0000283)

SRI Fund Indicator 0.794∗∗∗ 0.772∗∗∗

(0.189) (0.190)

Ln(Total Holdings Value) 0.00124
(0.00728)

Observations 98252 98252 98252 98252 166185
Adjusted R2 0.609 0.609 0.612 0.611 0.721
Quarter FE Y Y Y Y Y
Investment Objective FE Y Y Y Y N

This table reports the results from an investor-level regression of the average ESG score for the portfolio holdings
on the investor’s horizon: InvestorESGi,t = α+β1Horizoni,t +β2Xi,t +�i,t. InvestorESG is the weighted-average
ESG score of an investor’s portfolio holdings. Mutual funds’ investment horizons are proxied by either estimated
fund churn ratios or reported fund turnover ratios. 13f investors’ horizons are proxied by churn ratios. Fund and 13f
churn ratios are estimated over the previous four quarters. SRIFund is an indicator for whether a fund is classified
as a Socially Responsible Investment fund. Observations are at fund-quarter level. All specifications include quarter
fixed-effects. Standard errors are two-way clustered by fund and by quarter, and shown in parentheses. *, **, and
*** and indicate 10%, 5%, and 1% significance respectively. The sample period covers 2000-2018.

44

Electronic copy available at: https://ssrn.com/abstract=3049943

Table III: Firm-level analysis: ESG Score and Shareholder Investment Horizon

Shareholder Horizon Proxied by: MF Turnover MF Churn 13f Churn TRAt+1/13fOwn
(1) (2) (3) (4)

MSCI ESG Score -0.222∗∗ -0.145∗∗∗ -0.0984∗∗∗ -0.0964∗∗

(0.102) (0.0461) (0.0319) (0.0423)

Ln(Market Cap) 0.555∗ -0.571∗∗∗ -0.593∗∗∗ -0.683∗∗∗

(0.288) (0.161) (0.0714) (0.124)

Book-to-Market -2.252∗∗∗ -0.896∗∗∗ -0.0996 -0.785∗∗∗

(0.334) (0.228) (0.187) (0.266)

Dividend Yield -0.814∗∗∗ -0.364∗∗∗ -0.367∗∗∗ -0.248∗∗∗

(0.125) (0.0486) (0.0477) (0.0807)

Profitability Ratio -0.00657 0.000692 -0.0204∗∗∗ -0.0105
(0.0140) (0.00782) (0.00566) (0.00749)

Past 12-month Return 0.0608∗∗∗ 0.0289∗∗∗ 0.0115∗∗∗ 0.0207∗∗∗

(0.00601) (0.00295) (0.00131) (0.00343)

Return Volatility 0.265∗∗∗ 0.198∗∗∗ 0.200∗∗∗ 0.192∗∗∗

(0.0460) (0.0251) (0.0282) (0.0301)

Stock Turnover 1.186∗∗∗ 0.709∗∗∗ 1.102∗∗∗ 0.574∗

(0.308) (0.221) (0.154) (0.327)

Underlying Fund Flow Volatility 1.474∗∗∗ 1.839∗∗∗

(0.251) (0.253)
Observations 26217 26217 26217 26217
Adjusted R2 0.451 0.442 0.500 0.684
Industry (SIC2) Fixed-Effects Y Y Y Y
Year Fixed-Effects Y Y Y Y

This table reports the results of regressions of shareholders’ investment horizons on firms’ MSCI ESG scores and
control variables. The observations are at the firm-year level. In Column (1), the dependent variable is the
weighted-average turnover ratio of a firm’s mutual fund shareholders. In Column (2), the dependent variable is
the weighted-average Churn ratio of a firm’s mutual fund shareholders. In Column (3), the dependent variable is
the weighted-average Churn ratio of a firm’s 13f institution shareholders. In Column (4), the dependent variable
is the ratio between transient (TRA) institution ownership and total 13f institution ownership in the firm. The
classification of 13f institutions in this specification follows Bushee (1998). Year fixed-effects and industry fixed-
effects are included across specifications. Standard errors are double clustered at the stock and year level, and
shown in parentheses. *, **, and *** and indicate 10%, 5%, and 1% significance respectively. The sample period
covers 2000-2018.

45

Electronic copy available at: https://ssrn.com/abstract=3049943

T
a
b
le

IV
:

F
ir

m
-L

ev
el

A
n
a
ly

si
s:

R
o
b
u
st

n
es

s

D
ep

en
d
en

t
V

a
ri

a
b
le

(%
)

M
F

T
u
rn

o
v
er

M
F
C
h
u
r
n

1
3
f
C
h
u
r
n

T
R
A

t+
1
/
1
3
f
O
w
n

M
F

T
u
rn

o
v
er

M
F
C
h
u
r
n

1
3
f
C
h
u
r
n

T
R
A

t+
1
/
1
3
f
O
w
n

(1
)

(2
)

(3
)

(4
)

(5
)

(6
)

(7
)

(8
)

M
S
C

I
E

S
G

S
tr

en
g
th

S
co

re
-0

.4
1
5
∗∗
∗

-0
.1

4
5
∗∗

-0
.1

0
6
∗∗
∗

-0
.1

5
1
∗∗
∗

(0
.1

1
2
)

(0
.0

5
2
9
)

(0
.0

3
5
3
)

(0
.0

3
8
8
)

M
S
C

I
E

S
G

C
o
n
ce

rn
S
co

re
0
.1

7
1

-0
.1

0
8
∗

-0
.0

5
5
1

0
.0

2
5
3

(0
.1

3
9
)

(0
.0

5
4
8
)

(0
.0

3
8
5
)

(0
.0

5
9
0
)

M
S
C

I
E

S
G

x
/
G

o
v
er

n
a
n
ce

-0
.1

7
6
∗

-0
.1

2
7
∗∗
∗

-0
.0

7
9
3
∗∗

-0
.0

9
3
9
∗∗

(0
.0

8
8
9
)

(0
.0

4
3
1
)

(0
.0

2
8
3
)

(0
.0

4
0
5
)

ln
(M

a
rk

et
C

a
p
)

0
.9

9
4
∗∗
∗

-0
.5

4
6
∗∗
∗

-0
.5

5
6
∗∗
∗

-0
.5

5
3
∗∗
∗

0
.5

6
9
∗

-0
.5

5
5
∗∗
∗

-0
.5

8
5
∗∗
∗

-0
.6

0
4
∗∗
∗

(0
.3

0
8
)

(0
.1

8
3
)

(0
.0

7
8
1
)

(0
.1

3
0
)

(0
.2

9
4
)

(0
.1

6
3
)

(0
.0

7
1
5
)

(0
.1

2
2
)

B
o
o
k
-t

o
-M

a
rk

et
-1

.9
5
6
∗∗
∗

-0
.8

7
7
∗∗
∗

-0
.0

7
3
5

-0
.6

9
6
∗∗

-2
.2

2
1
∗∗
∗

-0
.8

7
5
∗∗
∗

-0
.0

8
5
9

-0
.7

5
9
∗∗

(0
.3

3
9
)

(0
.2

3
5
)

(0
.1

8
6
)

(0
.2

6
3
)

(0
.3

2
9
)

(0
.2

2
6
)

(0
.1

8
3
)

(0
.2

8
1
)

D
iv

id
en

d
Y

ie
ld

-0
.7

7
5
∗∗
∗

-0
.3

6
2
∗∗
∗

-0
.3

6
4
∗∗
∗

-0
.2

3
6
∗∗
∗

-0
.8

1
7
∗∗
∗

-0
.3

6
5
∗∗
∗

-0
.3

6
9
∗∗
∗

-0
.2

7
2
∗∗

(0
.1

1
7
)

(0
.0

4
8
0
)

(0
.0

4
8
3
)

(0
.0

8
0
2
)

(0
.1

2
4
)

(0
.0

4
8
5
)

(0
.0

4
7
6
)

(0
.0

9
4
2
)

P
ro

fi
ta

b
il
it

y
R

a
ti

o
-0

.0
0
9
9
1

0
.0

0
0
5
7
8

-0
.0

2
0
6
∗∗
∗

-0
.0

1
1
5

-0
.0

0
6
5
3

0
.0

0
0
6
8
2

-0
.0

2
0
4
∗∗
∗

0
.0

0
0
3
1
0

(0
.0

1
3
7
)

(0
.0

0
7
8
6
)

(0
.0

0
5
6
7
)

(0
.0

0
7
4
4
)

(0
.0

1
4
0
)

(0
.0

0
7
8
3
)

(0
.0

0
5
6
6
)

(0
.0

0
7
9
1
)

P
a
st

1
2
-m

o
n
th

R
et

u
rn

0
.0

6
1
7
∗∗
∗

0
.0

2
8
9
∗∗
∗

0
.0

1
1
6
∗∗
∗

0
.0

2
0
9
∗∗
∗

0
.0

6
0
8
∗∗
∗

0
.0

2
8
9
∗∗
∗

0
.0

1
1
5
∗∗
∗

0
.0

2
1
6
∗∗
∗

(0
.0

0
6
2
4
)

(0
.0

0
2
9
8
)

(0
.0

0
1
3
3
)

(0
.0

0
3
4
9
)

(0
.0

0
6
0
1
)

(0
.0

0
2
9
5
)

(0
.0

0
1
3
2
)

(0
.0

0
3
6
8
)

R
et

u
rn

V
o
la

ti
li
ty

0
.2

7
3
∗∗
∗

0
.1

9
8
∗∗
∗

0
.2

0
0
∗∗
∗

0
.1

9
4
∗∗
∗

0
.2

6
6
∗∗
∗

0
.1

9
8
∗∗
∗

0
.2

0
0
∗∗
∗

0
.2

1
7
∗∗
∗

(0
.0

4
6
3
)

(0
.0

2
5
3
)

(0
.0

2
8
3
)

(0
.0

3
0
2
)

(0
.0

4
6
0
)

(0
.0

2
5
2
)

(0
.0

2
8
2
)

(0
.0

4
0
3
)

S
to

ck
T

u
rn

o
v
er

1
.1

5
1
∗∗
∗

0
.7

0
7
∗∗
∗

1
.1

0
0
∗∗
∗

0
.5

6
4

1
.1

9
2
∗∗
∗

0
.7

1
3
∗∗
∗

1
.1

0
5
∗∗
∗

0
.7

4
0
∗∗

(0
.3

0
6
)

(0
.2

2
1
)

(0
.1

5
4
)

(0
.3

2
6
)

(0
.3

0
7
)

(0
.2

2
1
)

(0
.1

5
4
)

(0
.3

3
4
)

U
n
d
er

ly
in

g
F

u
n
d

F
lo

w
V

o
la

ti
li
ty

1
.4

9
3
∗∗
∗

1
.8

4
0
∗∗
∗

1
.4

6
3
∗∗
∗

1
.8

3
3
∗∗
∗

(0
.2

4
3
)

(0
.2

5
3
)

(0
.2

5
1
)

(0
.2

5
4
)

O
b
se

rv
a
ti

o
n
s

2
6
2
0
7

2
6
2
1
7

2
6
2
1
7

2
6
2
1
7

2
6
2
0
7

2
6
2
1
7

2
6
2
1
7

2
6
2
1
7

A
d
ju

st
ed

R
2

0
.4

5
3

0
.4

4
2

0
.5

0
0

0
.6

8
4

0
.4

5
0

0
.4

4
1

0
.4

9
9

0
.6

7
9

In
d
u
st

ry
(S

IC
2
)

F
ix

ed
-E

ff
ec

ts
Y

Y
Y

Y
Y

Y
Y

Y
Y

ea
r

F
ix

ed
-E

ff
ec

ts
Y

Y
Y

Y
Y

Y
Y

Y

T
h

is
ta

b
le

re
p

o
rt

s
th

e
re

su
lt

s
o
f

re
g
re

ss
io

n
s

o
f

sh
a
re

h
o
ld

er
s’

in
v
es

tm
en

t
h

o
ri

zo
n

s
o
n

se
v
er

a
l

co
m

p
o
n

en
ts

o
f

fi
rm

s’
M

S
C

I
E

S
G

sc
o
re

s.
T

h
e

o
b

se
rv

a
ti

o
n

s
a
re

a
t

th
e

fi
rm

-y
ea

r
le

v
el

.
In

C
o
lu

m
n

s
(1

)
to

(4
),

th
e

fi
rm

’s
E

S
G

sc
o
re

is
d

ec
o
m

p
o
se

d
in

to
“
st

re
n

g
th

s”
a
n

d
“
co

n
ce

rn
s”

,
w

h
er

e
st

re
n

g
th

sc
o
re

s
a
re

p
o
si

ti
v
e

a
n
d

co
n

ce
rn

sc
o
re

s
a
re

n
eg

a
ti

v
e.

In
C

o
lu

m
n

s
(5

)
to

(8
),

w
e

ex
cl

u
d

e
co

m
p

o
n

en
ts

re
g
a
rd

in
g

co
rp

o
ra

te
g
o
v
er

n
a
n

ce
w

h
en

ca
lc

u
la

ti
n

g
co

m
p

a
n
y

E
S

G
sc

o
re

s.
Y

ea
r

fi
x
ed

-e
ff

ec
ts

a
n

d
in

d
u

st
ry

fi
x
ed

-e
ff

ec
ts

a
re

in
cl

u
d

ed
a
cr

o
ss

sp
ec

ifi
ca

ti
o
n

s.
S

ta
n

d
a
rd

er
ro

rs
a
re

d
o
u

b
le

cl
u

st
er

ed
a
t

th
e

st
o
ck

a
n

d
y
ea

r
le

v
el

,
a
n

d
sh

o
w

n
in

p
a
re

n
th

es
es

.
*
,

*
*
,

a
n

d
*
*
*

a
n

d
in

d
ic

a
te

1
0
%

,
5
%

,
a
n

d
1
%

si
g
n

ifi
ca

n
ce

re
sp

ec
ti

v
el

y.

46

Electronic copy available at: https://ssrn.com/abstract=3049943

T
a
b
le

V
:

F
ir

m
-L

ev
el

A
n
a
ly

si
s:

M
S
C

I
a
n
d

S
u
st

a
in

a
ly

ti
cs

E
S
G

S
co

re
s;

2
0
0
9

–
2
0
1
7

D
ep

en
d
en

t
v
a
ri

a
b
le

(%
)

M
F

T
u
rn

o
v
er

M
F
C
h
u
r
n

1
3
f
C
h
u
r
n

T
R
A

t+
1
/
1
3
f
O
w
n

M
F

T
u
rn

o
v
er

M
F
C
h
u
r
n

1
3
f
C
h
u
r
n

T
R
A

t+
1
/
1
3
f
O
w
n

(1
)

(2
)

(3
)

(4
)

(5
)

(6
)

(7
)

(8
)

M
S
C

I
E

S
G

S
co

re
-0

.2
0
1
∗∗

-0
.1

2
6
∗∗
∗

-0
.0

8
8
7
∗

-0
.1

4
2
∗∗

(0
.0

6
4
0
)

(0
.0

3
5
3
)

(0
.0

3
9
0
)

(0
.0

4
9
9
)

S
u
st

a
in

a
ly

ti
cs

E
S
G

S
co

re
-0

.0
3
6
3
∗

-0
.0

2
2
6
∗∗

-0
.0

2
5
2
∗∗

-0
.0

1
5
2

(0
.0

1
9
6
)

(0
.0

0
9
8
2
)

(0
.0

1
0
3
)

(0
.0

1
2
9
)

L
n
(M

a
rk

et
C

a
p
)

-0
.3

4
2

-1
.0

2
8
∗∗
∗

-1
.1

4
0
∗∗
∗

-1
.0

0
8
∗∗

-0
.4

6
2
∗∗

-1
.1

0
4
∗∗
∗

-1
.1

7
3
∗∗
∗

-1
.1

1
7
∗∗
∗

(0
.1

9
2
)

(0
.1

7
8
)

(0
.0

9
7
4
)

(0
.3

4
8
)

(0
.1

8
2
)

(0
.0

9
5
4
)

(0
.1

0
2
)

(0
.1

1
0
)

B
o
o
k
-t

o
-M

a
rk

et
R

a
ti

o
-0

.7
3
7

0
.1

5
9

0
.1

3
4

-0
.2

1
2

-0
.7

6
9

0
.1

3
9

0
.1

3
9

-0
.2

5
8

(0
.6

2
1
)

(0
.2

5
0
)

(0
.3

3
0
)

(0
.4

6
4
)

(0
.5

4
6
)

(0
.2

8
6
)

(0
.3

1
5
)

(0
.4

2
9
)

D
iv

id
en

d
Y

ie
ld

-0
.8

4
8
∗∗
∗

-0
.4

4
5
∗∗
∗

-0
.4

6
2
∗∗
∗

-0
.3

1
2
∗∗

-0
.8

5
0
∗∗
∗

-0
.4

4
7
∗∗
∗

-0
.4

5
9
∗∗
∗

-0
.3

2
0
∗∗
∗

(0
.1

6
3
)

(0
.0

6
8
0
)

(0
.0

6
3
5
)

(0
.1

3
3
)

(0
.1

0
1
)

(0
.0

5
3
2
)

(0
.0

5
4
5
)

(0
.0

6
4
8
)

P
ro

fi
ta

b
il
it

y
R

a
ti

o
-0

.0
5
6
0
∗∗

-0
.0

1
2
7

-0
.0

1
6
9

0
.0

1
2
4

-0
.0

5
6
5
∗∗
∗

-0
.0

1
2
9

-0
.0

1
7
4
∗

0
.0

1
2
4

(0
.0

2
0
2
)

(0
.0

0
8
6
1
)

(0
.0

1
0
6
)

(0
.0

1
4
7
)

(0
.0

1
7
6
)

(0
.0

0
9
3
6
)

(0
.0

1
0
1
)

(0
.0

1
2
8
)

P
a
st

1
2
-m

o
n
th

R
et

u
rn

0
.0

2
6
1
∗∗
∗

0
.0

1
5
7
∗∗
∗

0
.0

0
6
4
0

0
.0

1
5
6
∗∗
∗

0
.0

2
5
8
∗∗
∗

0
.0

1
5
5
∗∗
∗

0
.0

0
6
2
7
∗∗
∗

0
.0

1
5
3
∗∗
∗

(0
.0

0
4
7
0
)

(0
.0

0
2
5
3
)

(0
.0

0
3
7
4
)

(0
.0

0
3
8
2
)

(0
.0

0
3
8
4
)

(0
.0

0
2
2
7
)

(0
.0

0
2
3
2
)

(0
.0

0
3
1
3
)

R
et

u
rn

V
o
la

ti
li
ty

0
.1

7
1
∗∗

0
.0

9
4
0
∗

0
.1

5
3
∗∗
∗

0
.0

7
3
5

0
.1

6
9
∗∗
∗

0
.0

9
2
6
∗∗
∗

0
.1

5
1
∗∗
∗

0
.0

7
2
5
∗∗

(0
.0

6
4
9
)

(0
.0

4
1
5
)

(0
.0

3
8
4
)

(0
.0

5
8
6
)

(0
.0

5
8
6
)

(0
.0

2
7
0
)

(0
.0

2
5
0
)

(0
.0

3
6
4
)

S
to

ck
T

u
rn

o
v
er

1
.7

1
7

0
.9

4
1
∗

1
.8

6
8
∗∗
∗

1
.0

5
8

1
.6

8
3
∗

0
.9

1
9
∗∗

1
.8

4
8
∗∗
∗

1
.0

4
0

(1
.1

3
1
)

(0
.4

5
8
)

(0
.4

6
4
)

(0
.6

0
3
)

(0
.9

2
9
)

(0
.3

6
7
)

(0
.4

1
9
)

(0
.6

5
6
)

U
n
d
er

ly
in

g
F

u
n
d

F
lo

w
V

o
la

ti
li
ty

2
.7

0
2
∗∗
∗

2
.4

3
4
∗∗
∗

2
.7

0
5
∗∗
∗

2
.4

3
6
∗∗
∗

(0
.3

8
4
)

(0
.3

9
8
)

(0
.3

4
9
)

(0
.2

5
2
)

O
b
se

rv
a
ti

o
n
s

5
8
9
7

5
8
9
8

5
8
9
8

5
8
9
8

5
8
9
7

5
8
9
8

5
8
9
8

5
8
9
8

A
d
ju

st
ed

R
2

0
.4

2
7

0
.5

3
3

0
.4

9
3

0
.7

7
9

0
.4

2
5

0
.5

3
2

0
.4

9
3

0
.7

7
8

In
d
u
st

ry
(s

ic
2
)

F
ix

ed
-E

ff
ec

ts
Y

Y
Y

Y
Y

Y
Y

Y
Y

ea
r

F
ix

ed
-E

ff
ec

ts
Y

Y
Y

Y
Y

Y
Y

Y

T
h

is
ta

b
le

re
p

o
rt

s
th

e
re

su
lt

s
o
f
re

g
re

ss
io

n
s

o
f
sh

a
re

h
o
ld

er
s’

in
v
es

tm
en

t
h

o
ri

zo
n

s
o
n

fi
rm

E
S

G
p

ro
fi

le
,
a
s

m
ea

su
re

d
b
y

b
o
th

M
S

C
I

E
S

G
sc

o
re

s
a
n

d
S

u
st

a
in

a
ly

ti
cs

sc
o
re

s.
T

h
e

sa
m

p
le

is
co

n
st

ra
in

ed
to

th
e

p
er

io
d

o
f

2
0
0
9

–
2
0
1
7

fo
r

th
e

se
t

o
f

fi
rm

s
w

it
h

b
o
th

M
S
C

I
E

S
G

sc
o
re

a
n

d
S

u
st

a
in

a
ly

ti
cs

sc
o
re

a
v
a
il

a
b

le
.

T
h

e
o
b

se
rv

a
ti

o
n

s
a
re

a
t

th
e

fi
rm

-y
ea

r
le

v
el

.
Y

ea
r

fi
x
ed

-e
ff

ec
ts

a
n

d
in

d
u

st
ry

fi
x
ed

-e
ff

ec
ts

a
re

in
cl

u
d

ed
a
cr

o
ss

sp
ec

ifi
ca

ti
o
n

s.
S

ta
n

d
a
rd

er
ro

rs
a
re

d
o
u

b
le

cl
u

st
er

ed
a
t

st
o
ck

a
n

d
y
ea

r
le

v
el

,
a
n

d
sh

o
w

n
in

p
a
re

n
th

es
es

.
*
,

*
*
,

a
n

d
*
*
*

a
n

d
in

d
ic

a
te

1
0
%

,
5
%

,
a
n

d
1
%

si
g
n

ifi
ca

n
ce

re
sp

ec
ti

v
el

y.

47

Electronic copy available at: https://ssrn.com/abstract=3049943