ArticlePDF Available

Corporate social responsibility and information flow

July 2020
Accounting and Finance 61(2)

Authors:

Northeastern Illinois University

We find that a firm’s greater commitment to corporate social responsibility (CSR) increases firm‐specific information incorporated into stock prices. We further show that information searches increase around major disclosure events for firms that are more socially responsible, as observed through requests for newly released annual (10‐K) filings on EDGAR and company ticker searches on Google around earnings announcements. Using alternative empirical specifications, we establish a robust and positive relation between CSR and stock price informativeness. Our results are consistent with the ethical and reputational view that a commitment to CSR encourages information acquisition and facilitates information flow into stock prices.

CSR and price informativeness

…

CSR and abnormal Google searches

…

CSR and future earnings captured in stock prices

…

Lead-lag regressions of CSR and price informativeness

…

Differential impact of CSR strengths and concerns

…

Figures - uploaded by Xiaohong Sara Wang

Content may be subject to copyright.

Content uploaded by Xiaohong Sara Wang

Content may be subject to copyright.

Corporate social responsibility and information ﬂow

Gary Chen

,Bin Wang

,Xiaohong Wang

College of Business Administration, University of Illinois at Chicago, Chicago, IL,

College of Business Administration, Marquette University, Milwaukee, WI,

College of Business and Management, Northeastern Illinois University, Chicago, IL, USA

Abstract

We ﬁnd that a ﬁrm’s greater commitment to corporate social responsibility

(CSR) increases ﬁrm-speciﬁc information incorporated into stock prices. We

further show that information searches increase around major disclosure events

for ﬁrms that are more socially responsible, as observed through requests for

newly released annual (10-K) ﬁlings on EDGAR and company ticker searches

on Google around earnings announcements. Using alternative empirical

speciﬁcations, we establish a robust and positive relation between CSR and

stock price informativeness. Our results are consistent with the ethical and

reputational view that a commitment to CSR encourages information

acquisition and facilitates information ﬂow into stock prices.

Key words: Corporate social performance; Corporate social responsibility;

Idiosyncratic volatility; Information acquisition; Probability of informed

trading; Stock price informativeness

JEL classiﬁcation: G14, M14, M19, M41

doi: 10.1111/acﬁ.12683

We owe particular thanks to those at the Securities and Exchange Commission who

have assisted us with the EDGAR search data. We also thank Michael Drake for

sharing data on Google searches, Stephen Brown for his data on the probability of

informed trade (PIN), and Eugene Soltes for his data on press coverage. We further

thank Li Cai, Alex Edmans, Feng Gao, Joel F. Houston, Kelly Huang, Chuck Kwok,

Brent Lao, Yihui Pan (discussant), Andrew Scott (discussant), and workshop/session

participants at Central Michigan University, Florida International University, North-

eastern Illinois University, the University of Illinois at Chicago, the 2014 Socially

Responsible Investing (SRI) Conference at DePaul University, the 2015 Financial

Management Association annual meeting, and the 2016 American Accounting Asso-

ciation annual meeting for comments and suggestions. All errors are our own.

Please address correspondence to Gary Chen via email: garychen@uic.edu

Accounting & Finance

1. Introduction

Over the past three decades, corporate social responsibility (CSR) has drawn

considerable interest from practitioners, regulators and academics as more

managers incorporate CSR activities into their business operations. Demands

from multiple stakeholders, including customers, investors, employees, suppli-

ers, community organisations and governments have inﬂuenced managers to

make greater commitments to CSR (McWilliams and Siegel, 2001).

As the

prevalence of CSR grows, there are calls for research into the measurable

economic consequences from CSR activities (HBS, 2011). In this paper, we

investigate whether and how CSR facilitates the incorporation of ﬁrm-speciﬁc

information into stock prices, or stock price informativeness. Understanding

the processes through which information is impounded into stock price is

fundamental to ﬁnancial economics. As the seminal work by Fama (1970)

states, stock prices can provide accurate signals about productivity and guide

eﬃcient capital allocation decisions in an eﬃcient market.

Ex ante, it is unclear whether a greater commitment to CSR can facilitate the

incorporation of more or less information into stock prices. On one hand, CSR

activities may encourage greater information acquisition and trading. A survey

of US chief ﬁnancial oﬃcers (CFO) shows that ﬁrms engage in CSR activities

primarily to build their reputations and to act as good corporate citizens.

These ethical and reputational concerns may restrain CSR ﬁrms from unethical

behaviour such as the manipulation of accounting numbers and the exploita-

tion of outside investors. Outside investors would have greater incentives to

acquire and trade on information due to lower information acquisition costs

and greater investor protection, leading to more informative stock prices. As a

result, the ethical and reputational view suggests a positive association between

CSR and price informativeness.

On the other hand, CSR activities may discourage information acquisition

and trading. A line of literature argues that CSR is a manifestation of agency

problems (Friedman, 1970). Self-serving managers may engage in CSR

activities to advance their personal interests, e.g., to achieve higher social

status, at the expense of shareholders (Barnard, 1997; Cespa and Cestone,

2007). In addition, managers may be more likely to commit agency behaviours,

Recently, the Business Roundtable, a group of CEOs representing nearly 200 major US

corporations, dropped the long-held notion that business decisions should ﬁrst and

foremost serve their shareholders, and stated that corporate leaders should take into

account all stakeholders including employees, customers, suppliers and society at large,

in their decision making beyond maximising proﬁts (Benoit, 2019).

Results are from the 2013 Global Business Outlook Survey conducted by Duke

University and CFO Magazine (http://www.reputationxchange.com/csr-through-the-eye

s-of-cfos).

2 G. Chen et al./Accounting & Finance

such as asset diversion for personal gain, when they serve many masters (all

stakeholders) and may not be held accountable (Jensen, 2001). Self-interested

managers may engage in complicated business transactions and accounting

manipulation to mask a ﬁrm’s true cash ﬂows in order to disguise opportunistic

behaviour (Desai and Dharmapala, 2009).

These actions can result in greater

information acquisition costs to investors. High information acquisition costs

(Grossman and Stiglitz, 1980) and concerns regarding insider exploitation

(Morck et al., 2000) can discourage outside investors from doing research and

trading on companies with greater CSR involvement, which results in less

informative stock prices. Consequently, the agency view predicts that CSR is

negatively associated with price informativeness.

Based on these arguments, we empirically examine the association between

CSR activities and stock price informativeness. We use idiosyncratic volatility,

the stock return variation unexplained by market movement, and the

probability of informed trading (PIN) as our main measures of price

informativeness. These two proxies are commonly used as measures of ﬁrm-

speciﬁc information captured in stock prices and have been validated in prior

literature (Easley et al., 1997a, 1997b; Morck et al., 2000). Following the

literature (Hong et al., 2012; Deng et al., 2013), we use the Kinder, Lydenberg,

Domini, and Co. (KLD) database to measure a ﬁrm’s commitment to CSR. We

show that the stock prices of more socially responsible ﬁrms have higher levels

of idiosyncratic volatility and PIN, consistent with the ethical and reputational

view that greater CSR commitment facilitates more ﬁrm-speciﬁc information

incorporated into stock prices.

We further provide direct evidence on the role of CSR in investors’ search

activities for ﬁrm-speciﬁc information. While it would be ideal to study the

association between CSR and investors’ search for private information, it is

inherently diﬃcult to do so because private information searches are usually

unobservable. To overcome this empirical challenge, we study the association

between CSR and investors’ searches on the SEC’s Electronic Data Gathering,

Analysis, and Retrieval (EDGAR) website for new 10-K ﬁnancial reports and

on Google for ﬁrms around their earnings announcements.

We believe that

For instance, former Enron CEO, Kenneth Lay, placed great importance on charity

and philanthropy. As a result, Enron became one of the largest corporate philan-

thropists in Texas donating millions to arts groups, scholarship funds and medical

facilities (Hemingway and Maclagan, 2004). At the same time, Enron used extensive,

complex tax shelters and special purpose entities to manipulate and mask its earnings

while preventing investors from understanding its sources (Desai and Dharmapala,

2009).

We use 10-K ﬁlings rather than 10-Q ﬁlings because 10-K ﬁlings are the most searched

ﬁlings on EDGAR (Drake et al., 2012) and are more informative to investors (Loughran

and McDonald, 2015). Furthermore, Li and Ramesh (2009) ﬁnd no signiﬁcant stock

price and volume reaction to quarterly ﬁlings after controlling for the concurrent release

of earnings news.

G. Chen et al./Accounting & Finance 3

this empirical design is suitable for our study for the following reasons. First,

both the 10-K reports and earnings announcements are important public

disclosures that can stimulate private information search in order to proﬁt from

the public disclosure events (McNichols and Trueman, 1994; Kim and

Verrecchia, 1997). Furthermore, recent literature shows that new information

contained in public disclosures is disseminated gradually due to investors’

limited attention and the signiﬁcant cost in information processing (Hirshleifer

and Teoh, 2003; Drake et al., 2012).

We ﬁnd that investors’ searches for

information around these major disclosure events is more pronounced for ﬁrms

that are more socially responsible.

We further examine how CSR is associated with the ability of current stock

prices to reﬂect future earnings. Stock prices are more informative when they

incorporate more value-relevant information including future earnings news

(Kothari and Sloan, 1992; Choi et al., 2011). If CSR incentivises (or

discourages) information acquisition and trading, then current stock prices

should be more (or less) predictive of future ﬁrm earnings. We use the future

earnings response coeﬃcient (or FERC) that relates current stock returns to

lead ﬁrm earnings as an alternative measure for price informativeness. We ﬁnd

that the stock prices of socially responsible ﬁrms contain more information

about ﬁrm future earnings.

We conduct a series of robustness checks to verify our main ﬁndings. We

conduct a lead-lag analysis and ﬁnd that the lagged (rather than lead or

contemporaneous) levels of CSR is positively associated with price informa-

tiveness. The results from the lead-lag analysis alleviate concerns about reverse

causality and simultaneity. We further ﬁnd results that are consistent with our

main ﬁnding when using a propensity score matched sample. Furthermore, we

ﬁnd that CSR strengths, rather than concerns, are the primary drivers of our

results. Finally, we investigate and show that our results are unlikely driven by

several alternative explanations.

This paper contributes to the growing body of literature on CSR. While an

extensive body of work examines the relation between CSR and equilibrium

ﬁrm value (Margolis et al., 2007; Flammer, 2015; Lys et al., 2015; Gregory

et al., 2016; Jones and Wright, 2016) and cost of capital (El Ghoul et al., 2011;

Goss and Roberts, 2011), few papers take a microstructure approach in

examining how CSR relates to the process of price discovery. Given the vital

role of stock prices in guiding eﬃcient capital allocation decisions, our paper

ﬁlls an important void in the literature. We provide a systematic study that

We deﬁne private information as the new information that has not yet been

incorporated into the stock price. Even though information in public disclosures is

available upon release, such information is not impounded into the stock price

instantaneously because investors need to digest the information over time, leading to

new information becoming gradually incorporated into the stock price. Therefore,

investigation of the dissemination of new information in public disclosures can shed

light on investors’ search for private information.

4 G. Chen et al./Accounting & Finance

examines CSR and stock price informativeness and ﬁnd a robust positive

relation.

This paper complements but diﬀers from prior studies that investigate the

relation between CSR and a ﬁrm’s information environment or CSR and

managerial opportunism (Gelb and Strawser, 2001; Petrovits, 2006; Chih et al.,

2008; Kim et al., 2012; Gao et al., 2014). While these studies investigate the role

of CSR on corporate policy and managerial behaviour, our paper extends the

literature by exploring how CSR relates to investor information acquisition. To

our knowledge, our work is the ﬁrst to directly address the relation between

CSR and investor information search activities. In addition, the role of CSR on

a ﬁrm’s information environment and managerial opportunism is still subject

to debate. Our ﬁnding that greater CSR commitment, and CSR strength in

particular, is associated with greater investor information acquisition and stock

price informativeness suggests that CSR contributes to a better information

and trading environment.

Our study also adds to the ongoing debate on the nature of CSR activities.

While some academics and practitioners criticise CSR activities as a type of

agency problem, our empirical results are consistent with the ethical and

reputational motives for CSR. Our ﬁndings may also have broader implications

for corporate policies on CSR. A growing line of research argues that stock

prices can impact corporate decisions and ﬁrm value through managers’

learning from information embedded in stock prices (Chen et al., 2007; Bond

et al., 2012; Zuo, 2016) or being disciplined by stock performance to make

value-maximising decisions (Jensen, 1986; Ferreira et al., 2011). If CSR

activities promote greater stock price informativeness, then managers of more

socially responsible ﬁrms may make better capital allocation decisions that

enhance ﬁrm value.

Lastly, our paper contributes to the literature on price informativeness. Prior

studies have examined a number of determinants of price informativeness

including corporate governance (Ferreira and Laux, 2007; Ferreira et al., 2011),

board composition (Gul et al., 2011), disclosure activities (Gelb and Zarowin,

2002), analyst coverage (Piotroski and Roulstone, 2004), institutional owner-

ship (Piotroski and Roulstone, 2004) and insider trading (Fernandes and

Ferreira, 2009). Nonetheless, research on whether CSR activities aﬀect stock

price informativeness remains scarce. Grewal et al. (2020) show that material

sustainability reports contain value-relevant ﬁrm-speciﬁc information and

therefore increase stock price informativeness. Complementing Grewal et al.

(2020), our study provides supportive evidence that holding CSR disclosure

constant, higher CSR ratings further attract more investors to search for ﬁrm-

speciﬁc information due to socially responsible ﬁrms’ commitment to a

trustworthy information and trading environment. Our study of the impact of

CSR on price informativeness is timely and important, as CSR has received

increasing attention from business, policymakers and academia.

G. Chen et al./Accounting & Finance 5

2. Literature review and hypothesis development

Economic theory suggests that price informativeness is determined by

frictions in information acquisition and trading. Traders pay a cost to become

informed. These informed traders then trade in the market and, in doing so,

incorporate their information into prices. In a frictionless market, a ﬁrm’s CSR

activities should have no impact on the informativeness of the ﬁrm’s stock

price. However, information acquisition costs (Grossman and Stiglitz, 1980;

Diamond and Verrecchia, 1981) and poor investor protection (Fishman and

Hagerty, 1992) discourage information collection and adversely aﬀect the

amount of ﬁrm-speciﬁc information incorporated in equilibrium stock prices.

Theories on costly information (Grossman and Stiglitz, 1980) predict that stock

prices cannot perfectly incorporate all available information in equilibrium so

that partially informed stock prices can provide investors enough compensa-

tion to oﬀset their information acquisition cost. In our study, we maintain the

same theoretical assumption. We argue that frictions related to information

acquisition and trading can lead CSR activities to have a direct consequence on

price informativeness and investor information acquisition, which we elaborate

below.

2.1. CSR and stock price informativeness

Two schools of thought generate contrasting predictions on the relation

between CSR and price informativeness. One line of research argues that CSR

signals a commitment to ethical behaviour and enhances a ﬁrm’s image and

reputation for being a good corporate citizen (Carroll, 1979; Fombrun and

Shanley, 1990; Jones, 1995). Advocates of CSR recognise that corporate

actions aﬀect all of the ﬁrm’s stakeholders including employees, customers,

shareholders, community, government and the society, and argue that

companies should consider their economic, legal, ethical and philanthropic

responsibilities in decision making (Carroll, 1979). Furthermore, some argue

that a ﬁrm’s CSR involvement helps build valuable trust and reputation among

its stakeholders (Fombrun and Shanley, 1990; Economist, 2005).

We focus our discussions on the costs of information acquisition and not the expected

beneﬁts of information acquisition because our hypotheses are motivated by the

theoretical predictions of Grossman and Stiglitz (1980). They posit that when

information acquisition is costly, price only partially reveals the private information

of the informed traders so that the expected beneﬁts of information acquisition are

enough to compensate for the costs of information acquisition. In equilibrium, when the

information acquisition costs are low (high), price is more (less) informative, leaving a

low (high) expected beneﬁt that is just enough to oﬀset the information acquisition costs.

Thus, the Grossman and Stiglitz (1980) model provides a clear analytical framework and

prediction that it is the costs of information acquisition that determine the informa-

tiveness of stock prices.

6 G. Chen et al./Accounting & Finance

The ethical and reputational motive for CSR can restrain managers from

engaging in activities that are socially unacceptable. Agency problems arise due

to the separation of ownership and control. An optimally designed compen-

sation contract can only partially mitigate the agency problem due to

unobserved managerial eﬀort and information, managerial wealth constraints

and managerial risk aversion (Murphy, 1999).

However, moral and reputa-

tional motives can incentivise management to advance the interest of outside

investors because self-seeking behaviour of management damages the valuable

social trust and reputation built up over time and impairs future economic

beneﬁts to the ﬁrm (Noreen, 1988; Diamond, 1989; Ensminger, 2001). The

previous literature documents that CSR activities can foster trust from ﬁrms’

stakeholders (Lins et al., 2017) and generate economic beneﬁts.

Therefore,

protecting a ﬁrm’s reputation built from CSR can curtail managerial

opportunism.

In particular, greater CSR involvement may discourage managerial informa-

tion manipulation and exploitation of outside investors. Accounting quality and

transparency in corporate dealings are considered important aspects of CSR by

investors (Economist, 2005). Several empirical studies show that ﬁrms with

greater CSR commitment have a better information environment. For instance,

Kim et al. (2012) show that socially responsible ﬁrms do less accrual and real

earnings management and are less likely to be the subject of Securities and

Exchange Commission (SEC) enforcement action for ﬁnancial misreporting.

Kim et al. (2014) document that CSR performance is negatively associated with

stock price crash risk, consistent with the view that CSR restrains ﬁrms from bad

news hoarding. Furthermore, studies ﬁnd that socially responsible ﬁrms provide

greater investor protection. Ferrell et al. (2016) provide evidence of fewer agency

problems, as reﬂected in a lower level of excess cash ﬂow and better corporate

governance among ﬁrms with higher CSR ratings. Gao et al. (2014) ﬁnd lower

insider trading proﬁts for managers of socially responsible ﬁrms, suggesting that

Murphy (1999) surveys the executive compensation literature and concludes that

‘Unfortunately, although there is a plethora of evidence on dysfunctional consequences

of poorly designed pay programs, there is surprisingly little direct evidence that higher

pay-performance sensitivities lead to higher stock-price performance.’

For example, it is argued in the literature that ﬁrms’ CSR commitment can build

stronger customer relationships (Brown and Dacin, 1997), attract and retain higher

quality workers (Greening and Turban, 2000) and reduce adverse political, regulatory

and social penalties from negative corporate events (Porter and Linde, 1995; Tran and

O’Sullivan, 2020).

G. Chen et al./Accounting & Finance 7

outside investors in high CSR ﬁrms have a lower probability of trading with

corporate insiders who have an information advantage.

The perceived trustworthy information and trading environment can

encourage outside investors to collect and trade on information of more

socially responsible ﬁrms, leading to more informative stock prices. Grossman

and Stiglitz (1980) argue that lower information acquisition costs encourage

more investors to acquire and trade on information and result in more

informative stock prices. If managers of socially responsible ﬁrms are more

truthful with company news, investors would incur lower costs to collect value-

relevant information. For example, investors can save on costs from verifying

the truthfulness of released information when they utilise the information to

forecast future ﬁrm performance. As predicted by Grossman and Stiglitz

(1980), investors may have greater incentive to acquire and trade on

information of socially responsible ﬁrms, resulting in more informative stock

prices.

In addition, Fishman and Hagerty (1992) and Morck et al. (2000)

contend that poor investor protection, such as opportunistic insider trading

and diversion of assets, can deter outside investors from acquiring and trading

on ﬁrm information due to fears of trading against and being exploited by

insiders. If socially responsible ﬁrms provide greater investor protection,

investors may devote more eﬀort and resources into acquiring and trading on

the information of these ﬁrms, leading to more information being impounded

into stock prices.

Another strand of literature advocates that CSR is a manifestation of agency

problems (Levitt, 1958; Friedman, 1970). Friedman (1970) argues that the only

social responsibility of business is to maximise proﬁts and any deviation from

this goal would hurt the foundations of a free society. Extending this view,

some argue that managers engage in CSR activities to derive private beneﬁts at

the expense of shareholders. For example, managers may spend corporate

resources to gain favour with board members through contributions to their

preferred causes, to achieve higher social status, or to push their own political/

Fishman and Hagerty (1992) theorise that insider trading discourages outside investors

from collecting and trading private information by limiting the gains of outside

investors. As a result, while informed trades by insiders might increase, overall informed

trading and stock price informativeness can decline because of reduced private

information acquisition and trading by outside investors. Empirically, Fernandes and

Ferreira (2009) show that the enforcement of laws that restrain insider trading improves

stock price informativeness.

While one can argue that information manipulation of socially irresponsible ﬁrms may

also attract investors to search for private information, the cost to discover a ﬁrm’s true

fundamental value can be prohibitively high due to complex accounting manoeuvres. As

argued by Grossman and Stiglitz (1980), higher information acquisition costs lead to

fewer traders acquiring information. In equilibrium, the price is less informative, with

the expected beneﬁts of acquiring information equal to the higher costs of information

acquisition.

8 G. Chen et al./Accounting & Finance

ideological agendas (Barnard, 1997). Furthermore, Jensen (2001) argues that if

a manager answers to many masters (i.e. all of the stakeholders in a ﬁrm instead

of only shareholders), the manager may not be held accountable for the misuse

of corporate resources and is more likely to engage in self-serving activities such

as asset diversion. To disguise their rent extraction behaviours, managers may

intentionally structure complicated business transactions and manipulate their

accounting numbers to obscure the true cash ﬂows of the ﬁrm (Desai and

Dharmapala, 2009).

Several studies show supportive evidence for the agency motive of CSR. In

line with the agency perspective, Barnea and Rubin (2010) show that managers

overinvest in CSR when they bear little of the cost. Brown et al. (2006) provide

evidence suggesting that ﬁrms with weak corporate governance or creditor

monitoring give more cash to charities and are more likely to establish

corporate foundations. Cespa and Cestone (2007) show that incumbent CEOs

may strategically engage in CSR activities in order to gain favour from social

and environmental activists and reduce their probability of future turnover.

Furthermore, ﬁndings from several studies suggest greater managerial oppor-

tunism through earnings management among more socially responsible ﬁrms

(Petrovits, 2006; Chih et al., 2008; Prior et al., 2008). For example, Petrovits

(2006) shows that ﬁrms use their charitable foundations as oﬀ-balance-sheet

reserves to manage earnings. In line with the agency view, Kruger (2015) ﬁnds

that investors respond negatively to positive CSR news.

If the agency view of CSR prevails, outside investors may be discouraged

from acquiring information about ﬁrms with greater CSR involvement due to

higher information acquisition costs and poorer investor protection. Even if the

agency problem triggers some sophisticated investors to search for ﬁrm-speciﬁc

information and short sell the stock, the substantial eﬀort and sophisticated

skills needed to uncover the true value-relevant information (masked through

complicated business transactions and accounting manipulation) and the high

transaction costs involved in short selling can deter a great majority of

investors from acquiring information.

According to Grossman and Stiglitz

(1980), such higher information acquisition costs lead to less information ﬂow

into stock prices. In addition, Morck et al. (2000) argue that agency problems

from a failure to protect outside investors from insiders can discourage ﬁrm-

speciﬁc information collection. For instance, insiders can shift income among a

The existing literature suggests that short-sales are generally executed by sophisticated

investors and require greater investing skills given that short sales are exposed to

unlimited losses and higher expenses than long positions (Diamond and Verrecchia,

1987; Dechow et al., 2001). Short-sellers are subject to lending fees in order to borrow

the stock. In addition, they face the risk of having an involuntarily closed short position

due to loan recalls and face constraints including government regulations or institutional

charter restrictions (Jones and Lamont, 2002). Both empirical and theoretical studies

show that short-sale constraints can hurt the eﬃciency of stock prices (Diamond and

Verrecchia, 1987; Jones and Lamont, 2002; Saﬃ and Sigurdsson, 2011).

G. Chen et al./Accounting & Finance 9

set of controlled companies through non-arm’s-length transactions for prod-

ucts, services or capital at artiﬁcial prices. Rational investors, knowing that

they cannot predict ﬁrms’ future cash ﬂow, invest fewer resources in acquiring

ﬁrm-speciﬁc information. Consequently, the agency view of CSR predicts that

more socially responsible ﬁrms have less informative stock prices.

To summarise, the ethical and reputational motive and the agency motive of

CSR leads to the following hypotheses, respectively:

H1a (Ethical and reputational motive): All else equal, CSR is positively related to

the informativeness of stock prices.

H1b (Agency motive): All else equal, CSR is negatively related to the informative-

ness of stock prices.

2.2. CSR and information acquisition

We further examine the role of CSR on investor information acquisition

decisions. Theories on costly information (Grossman and Stiglitz, 1980) predict

that lower information acquisition cost attracts more investors to acquire

information and become informed, leading to more informative stock prices.

Therefore, if CSR aﬀects stock price informativeness due to its impact on

information acquisition cost, we should see a direct bearing of CSR on investor’s

search for ﬁrm-speciﬁc information. The ethical and reputational motive of CSR

suggests that socially responsible ﬁrms are more truthful with company news and

provide greater investor protection, resulting in a lower cost of information

acquisition. Accordingly, this view predicts that socially responsible ﬁrms attract

more investors to acquire ﬁrm-speciﬁc information. On the other hand, the agency

motive of CSR argues that CSR is a manifestation of agency problems and may be

associated with greater information manipulation and exploitation of outside

investors, leading to a higher cost of information acquisition. Therefore, the agency

view predicts that CSR deters investors from collecting ﬁrm-speciﬁc information.

While it would be ideal to study the association between CSR and investors’

search for ﬁrm private information, it is inherently diﬃcult to do so because

private information searches are usually unobservable. To overcome this

empirical challenge, we study the association between CSR and investors’

searches for ﬁrm-speciﬁc information on the SEC’s EDGAR website for new 10-

K ﬁnancial reports and on Google around ﬁrms’ earnings announcements.

believe that this empirical design is suitable for our study for the following

reasons. First, prior research shows that public disclosure events can stimulate

private information search in anticipation of and in conjunction with the release

In equilibrium, the beneﬁt of information acquisition is low due to the more

informative stock price exactly oﬀsetting the low information acquisition cost.

See note 4.

10 G. Chen et al./Accounting & Finance

of public information (Kim and Verrecchia, 1997; Altschuler et al., 2015). In

addition, research shows that information in newly released public disclosures is

only gradually incorporated into stock prices due to investors’ limited attention

and the signiﬁcant eﬀort needed to process such information in a timely manner

(Hirshleifer and Teoh, 2003; Drake et al., 2012). Therefore, we directly test the

link between CSR and information acquisition by investigating whether CSR

involvement is associated with investor searches for new information around

ﬁling dates for 10-K ﬁnancial reports posted on the SEC’s EDGAR website and

Google searches for company information around earnings announcements.

Both the ﬁling of 10-K ﬁnancial reports and earnings announcements represent

signiﬁcant news events for investors and can dramatically drive investment

values. Thus, investors have greater incentives to acquire information around

these important events. These arguments lead to the following hypotheses:

H2a (Ethical and reputational motive): All else equal, CSR is positively related to

investors’ information acquisition around major disclosure events.

H2b (Agency motive): All else equal, CSR is negatively related investors’

information acquisition around major disclosure events.

2.3. CSR and future earnings captured in stock prices

To provide more evidence on the relation between CSR and stock price

informativeness, we further examine how CSR is associated with the ability of

current stock prices to reﬂect future earnings. Since current stock prices contain

the market expectation of value-relevant information (Kothari and Sloan,

1992; Choi et al., 2011), the extent to which current stock prices reveal future

earnings then indicates the informativeness of stock prices. To the extent that

CSR can encourage or discourage information acquisition and trading, we

argue that CSR can enhance (ethical and reputational view) or impair (agency

view) the incorporation of future earnings information in stock prices. These

predictions lead to the following hypotheses:

H3a (Ethical and reputational motive): All else equal, CSR enhances the

incorporation of future earnings news in contemporaneous stock prices.

H3b (Agency motive): All else equal, CSR impairs the incorporation of future

earnings news in contemporaneous stock prices.

3. Data, measures and descriptive statistics

3.1. Data sources and sample selection

We obtain the sample for our study from the KLD database on CSR. We

combine this sample with the Merged Compustat-CRSP database, the Investor

G. Chen et al./Accounting & Finance 11

Responsibility Research Center database, the Thomson-Reuters 13F institu-

tional holdings and insider trading databases, I/B/E/S analyst forecasts, Google

search traﬃc, and SEC EDGAR ﬁling request data. We exclude ﬁnancial

companies (SIC 6000–6999) and utilities (SIC 4900–4999) because the

operation of these ﬁrms may be subject to regulatory supervision. As in

Jegadeesh and Titman (2001), we further exclude stocks with a share price

below $5 as of the end of the ﬁscal year to ensure that results are not driven by

small, illiquid stocks or by the bid-ask bounce. Panel A of Table 1 shows the

sample selection procedure. The ﬁnal sample of our main price informativeness

regressions comes to 11,312 ﬁrm-year observations belonging to 2,414 ﬁrms

from 1995 to 2010.

Our resulting sample of ﬁrms varies over time as shown in

Panel B of Table 1, with a minimum of 224 ﬁrms in 1995 and a maximum of

1,330 ﬁrms in 2005 and 2006. Such variation is attributed to the fact that the

KLD database expanded coverage over the sample time period.

Panel C of

Table 1 presents the sample distribution by industry. As shown in the panel,

our sample covers a wide range of industries. Business services and retail make

up the largest proportion of ﬁrm-year observations at 12.92 and 8.58 percent,

respectively.

3.2. Measures

3.2.1. Stock price informativeness measures

We use idiosyncratic volatility (Ψ) and the probability of informed trading

(PIN) as our main proxies for price informativeness. French and Roll (1986)

and Roll (1988) state that idiosyncratic volatility, deﬁned as stock return

variation unexplained by market movement, measures the amount of ﬁrm-

speciﬁc information impounded into stock prices. This measure is further

supported and widely used as a measure of information ﬂow into stock prices

(Morck et al., 2000; Durnev et al., 2003; Chen et al., 2007; Ferreira and Laux,

2007; Ferreira et al., 2011; Gul et al., 2011).

As in Chen et al. (2007), we

The sample ends in 2010 to maintain consistency in the sample period across our

measures of stock price informativeness, since data on the probability of informed

trading (PIN) ends in 2010. Extending the sample to 2013 (the last year that KLD issued

CSR ratings as used in this paper) does not qualitatively change the results for

idiosyncratic volatility (Ψ), our ﬁrst measure of stock price informativeness.

KLD originally covered only ﬁrms in the S&P 500 Index and the Domini 400 Social

Management Index. KLD expanded its coverage to ﬁrms in the Russell 1000 in 2001 and

to ﬁrms in the Russell 3000 in 2003.

For instance, Durnev et al. (2003) show that the stock prices of ﬁrms with more

idiosyncratic volatility contain more information about future earnings.

12 G. Chen et al./Accounting & Finance

Table 1

Sample selection and descriptive statistics

Panel A: Sample selection

Firms Observations

KLD database 5,735 29,859

Intersection of observations in KLD, CRSP, and Compustat

databases

4,758 26,777

Less:

Firms in the ﬁnancial sector (SIC 6000–6999) and

utilities (SIC 4900–4999)

(1,190) (6,536)

Observations with missing price informativeness (Ψand PIN) (5,929)

Observations with insuﬃcient data for control variables or stock

price less than $5

(3,000)

Total 2,414 11,312

Panel B: Sample distribution by year

Year No. of observations Percent of total observations

1995 224 1.98

1996 280 2.48

1997 290 2.56

1998 294 2.60

1999 284 2.51

2000 272 2.40

2001 350 3.09

2002 486 4.30

2003 654 5.78

2004 1,318 11.65

(continued)

G. Chen et al./Accounting & Finance 13

Table 1 (continued)

Panel B: Sample distribution by year

Year No. of observations Percent of total observations

2005 1,330 11.76

2006 1,330 11.76

2007 1,216 10.75

2008 753 6.66

2009 1,215 10.74

2010 1,016 8.98

Total 11,312 100

Panel C: Sample distribution by industry

Fama-French

48-Industry Observations Percent Fama-French 48-Industry Observations Percent

Aircraft 112 0.99 Consumer Goods 260 2.30

Agriculture 44 0.39 Measuring and Control Equipment 312 2.76

Automobiles and Trucks 236 2.09 Machinery 602 5.32

Beer and Liquor 59 0.52 Restaurants, Hotels, Motels 237 2.10

Construction Materials 224 1.98 Medical Equipment 467 4.13

Printing and Publishing 177 1.56 Non-Metallic and Industrial Metal Mining 51 0.45

Shipping Containers 59 0.52 Petroleum and Natural Gas 611 5.40

Business Services 1,461 12.92 Business Supplies 263 2.32

Electronic Equipment 811 7.17 Personal Services 193 1.71

Apparel 209 1.85 Retail 971 8.58

Construction 141 1.25 Rubber and Plastic Products 74 0.65

Coal 46 0.41 Shipbuilding, Railroad Equipment 31 0.27

Computers 481 4.25 Tobacco Products 28 0.25

Pharmaceutical Products 716 6.33 Candy and Soda 21 0.19

(continued)

14 G. Chen et al./Accounting & Finance

Table 1 (continued)

Panel C: Sample distribution by industry

Fama-French

48-Industry Observations Percent Fama-French 48-Industry Observations Percent

Electrical Equipment 213 1.88 Steel Works Etc. 187 1.65

Fabricated Products 12 0.11 Communication 348 3.08

Food Products 266 2.35 Recreation 89 0.79

Entertainment 172 1.52 Transportation 387 3.42

Precious Metals 22 0.19 Textiles 32 0.28

Defense 35 0.31 Wholesale 440 3.89

Healthcare 212 1.87

Total 11,312 100

Panel D: Summary Statistics

Variables Mean Median SD P25 P75

Ψ0.949 0.922 1.155 0.240 1.605

PIN 0.114 0.107 0.050 0.083 0.137

CSR −0.126 −0.161 0.501 −0.375 0.125

CSR_STR 0.261 0.125 0.428 0.000 0.333

CSR_CON 0.387 0.333 0.403 0.000 0.533

LOG(EDGAR_1) 2.595 2.773 1.524 1.609 3.584

LOG(EDGAR_3) 3.293 3.555 1.693 2.639 4.317

LOG(EDGAR_1WK) 3.772 4.094 1.804 3.296 4.820

GSEARCH(−5, −1) 0.071 0.008 0.838 −0.059 0.089

GSEARCH(0) 0.321 0.035 1.588 −0.029 0.144

GSEARCH(1,5) 0.177 0.014 1.443 −0.053 0.105

SIZE 7.535 7.347 1.543 6.380 8.493

MB 1.062 0.995 0.767 0.567 1.472

(continued)

G. Chen et al./Accounting & Finance 15

Table 1 (continued)

Panel D: Summary Statistics

Variables Mean Median SD P25 P75

ROE 0.129 0.137 0.317 0.053 0.224

VROE 0.298 0.065 1.175 0.033 0.152

LEV 0.169 0.142 0.162 0.005 0.272

AGE 2.958 2.944 0.717 2.398 3.638

DIV 0.500 1.000 0.500 0.000 1.000

DIVER 0.917 1.000 0.276 1.000 1.000

ANALYSTS 1.809 1.833 0.678 1.386 2.277

BETA 1.189 1.130 0.509 0.834 1.481

EQ −0.045 −0.013 1.636 −0.152 0.059

INSTRADE −0.013 −0.003 0.037 −0.009 −0.001

CGSCORE −0.072 0.000 0.176 −0.167 0.000

INSTNUM 5.100 5.095 0.936 4.663 5.606

FREQ 1.087 1.099 0.982 0.000 1.946

LOGFILESIZE 2.084 1.540 2.286 1.067 2.307

LOGSUPPLY 2.137 2.197 0.493 1.946 2.485

SHORT 0.063 0.045 0.069 0.025 0.079

INSTOWN 0.632 0.721 0.293 0.518 0.848

|ROA|0.095 0.071 0.102 0.038 0.118

TURNOVER 0.207 0.163 0.175 0.098 0.258

NYSE 0.526 1.000 0.499 0.000 1.000

NEWS(−5,−1) 0.536 0.000 1.760 0.000 1.000

NEWS(0) 0.630 0.000 1.975 0.000 1.000

NEWS(1,5) 0.929 0.000 2.664 0.000 1.000

RET(m−3, m) 0.029 0.036 0.203 −0.069 0.130

FORECAST 0.556 1.000 0.497 0.000 1.000

LOSS 0.162 0.000 0.369 0.000 0.000

(continued)

16 G. Chen et al./Accounting & Finance

Table 1 (continued)

Panel D: Summary Statistics

Variables Mean Median SD P25 P75

|RET|0.014 0.010 0.017 0.004 0.018

BIDASK −7.406 −7.450 0.360 −7.665 −7.190

Log(ANM) 2.734 2.708 0.260 2.708 2.708

QTR4 0.319 0.000 0.466 0.000 1.000

RET

0.185 0.121 0.503 −0.133 0.404

EARN 0.030 0.045 0.093 0.019 0.068

EARN

t+1

0.035 0.046 0.095 0.017 0.074

RET

t+1

0.118 0.074 0.466 −0.174 0.341

ATGROWTH 0.126 0.073 0.261 −0.005 0.181

STDEARN 0.056 0.022 0.109 0.011 0.050

EARN_P 0.275 0.269 0.317 0.049 0.509

This table reports the sample selection procedure and descriptive statistics. The sample period is from 1995 to 2010. Panel A presents details of

the sample selection. Panel B shows the sample mean, median, standard deviation (SD), 25th percentile (P25) and 75th percentile (P75) of the key

variables. Variable deﬁnitions are provided in the Appendix. Financial and utilities industries (SIC 6000–6999 and 4900–4999) are excluded. All

variables are winsorised at the 1st and 99th percentiles.

G. Chen et al./Accounting & Finance 17

estimate Ψby regressing stock returns on an expanded market model including

the Fama-French 48 industry returns. Firm-speciﬁc information is estimated as

the variation of stock returns excluding market-wide and industry-wide

variation. For each ﬁrm iin year t, the ﬁrm-speciﬁc return variation is

estimated by computing 1 R2

i,tfrom the regression with daily returns in year t:

ri,j,d ¼β0þβ1rm,dþβ2rj,dþei,d (1)

where ri,j,dis the return of stock iin industry jon day d,r

m,d

is the value-

weighted market return on day d,r

j,d

is the return of industry jon day d.

Because 1 R2

i,tis skewed (Durnev et al., 2004), we take the logistic

transformation of 1 R2

i,tto ensure a normal distribution. Formally, idiosyn-

cratic volatility (ψi,t) is deﬁned as:

ψi,t¼ln 1R2

i,t

! (2)

Higher values of ψ

i,t

indicate that stock prices are more informative.

We also utilise PIN as an alternative measure to capture price informative-

ness. Proposed and developed by Easley et al. (1997a, 1997b, 2002), PIN is a

theoretically justiﬁed measure of private information in stock prices. The

measure is calculated based on microstructure trading models and gauges the

probability of informed trading. Given that informed traders place informed

bets based on their private information, PIN is conceptually compelling and is

widely used to measure private information in stock prices (Chen et al., 2007;

Ferreira and Laux, 2007; Ferreira et al., 2011). PIN is deﬁned as:

PIN ¼αi,dμi,d

αi,dμi,dþ2ɛi,d

 (3)

where α

i,d

equals the probability of an event on day dwhen informed investors

acquire private information regarding ﬁrm i,µ

i,d

is the arrival rate of buy (sell)

orders made by informed traders on day dfor ﬁrm i, and ϵ

i,d

is the daily arrival

rate of buy (sell) orders from uninformed traders. Intuitively, PIN is the

proportion of informed trades out of all (informed and uninformed) trades. We

utilise annual PIN estimates as computed in Brown and Hillegeist (2007).

PIN data are available at http://scholar.rhsmith.umd.edu/sbrown/pin-data

18 G. Chen et al./Accounting & Finance

Higher values of PIN indicate greater informed trading and suggest more

private information contained in stock prices.

3.2.2. Measure of CSR

We use the KLD database to construct our measure of CSR. The KLD

database is considered the most comprehensive and inﬂuential database on

CSR performance (Chatterji et al., 2009) and has been used extensively in the

literature (Hong et al., 2012; Deng et al., 2013). KLD evaluates ﬁrms on seven

major dimensions of CSR using various sources including surveys, ﬁnancial

reports, mainstream media and government documents. The seven major

dimensions are community, corporate governance, diversity, employee rela-

tions, environment, human rights, and product quality and safety. Within each

dimension, KLD assigns a binary indicator (0 or 1) to a set of strengths and

concerns.

As in Servaes and Tamayo (2013), we exclude the corporate governance

dimension of the KLD score to assess ﬁrms’ CSR performance. Servaes and

Tamayo (2013) contend that corporate governance is related to the mechanisms

that align the interests of managers to shareholders and is vastly diﬀerent from

the concept of CSR, which concerns corporate social objectives and relation-

ships with stakeholders. Furthermore, Harjoto and Jo (2011) ﬁnd that better

CSR performance is tied to stronger corporate governance. To the extent that

Ferreira and Laux (2007) show that ﬁrms with better corporate governance

have more informative stock prices, excluding the corporate governance

dimension can alleviate the confounding eﬀect of corporate governance on

price informativeness.

Following Deng et al. (2013), we scale each individual ﬁrm’s strength and

concern scores by the total number of strengths and concerns within each

dimension in each year and take the diﬀerence between the total strength and

It may appear contradictory that higher CSR is associated with a lower probability of

insider trading, yet is positively associated with the probability of informed trading

(PIN). The probability of informed trading consists of trades made by corporate insiders

(e.g., managers and board of directors) or trades made by outside investors that expense

resources to become informed. Thus, while CSR may restrain informed trades from

insiders (Gao et al. 2014), it may also encourage more outside investors to collect and

trade on private information for higher expected gains because outside investors are less

likely to trade with insiders (Fishman and Hagerty 1992; Fernandes and Ferreira 2009).

As a result, CSR may be positively related to PIN.

For instance, in the environment dimension, a ﬁrm can receive a credit for the use of

clean energy and a concern for hazardous waste.

G. Chen et al./Accounting & Finance 19

concern scores.

Deng et al. (2013) point out that because the number of

subcategories of many CSR dimensions can change each year, a scaling scheme

can strengthen the comparability of CSR scores over time.

3.2.3. Information search measures

Our ﬁrst measure of investors’ information search is constructed from

recorded internet traﬃc on the SEC’s EDGAR website. We obtain the

EDGAR server logs from 2003 to 2010 and count the number of requests for

annual ﬁnancial statement (10-K ﬁlings) on the date that the ﬁling was posted,

up to 3 days after the ﬁling was posted, and up to one week after the post date.

We use event windows around the date that a new 10-K ﬁling is posted to

capture investors’ searches for new ﬁrm-speciﬁc information.

We remove

requests made by automated programs following Lee et al. (2015). Similar to

Loughran and McDonald (2015), we also remove requests for index pages or

ﬁle that was either relocated or not found. We take the natural logarithm of

these request counts to normalise the data.

We next measure investors’ search activities by the intensity of internet

searches for company information as captured by the Google Search Volume

Index (SVI). The Google SVI measures the frequency of searches for a

particular term or phrase entered into the Google search engine. Prior research

has used the Google SVI for measuring investors’ demand for information and

their attention (Da et al., 2011; Drake et al., 2012). Similar to Drake et al.

(2012), we use daily abnormal Google SVI in the 5-day period prior to earnings

announcements, on the announcement date, and the 5-day post earnings

announcement period to capture the intensity of investors’ information search.

We obtain the abnormal Google SVI data for S&P 500 ﬁrms during 2005–2008

from Michael Drake.

We focus on earnings announcements because they are

events in which new ﬁnancial information about a ﬁrm becomes available to

investors.

To illustrate, suppose that a hypothetical ﬁrm receives 1, 2, 1, 0, 1 and 1 KLD

strengths across the six dimensions, and the total numbers of strength for each

dimension are 5, 4, 3, 5, 6 and 4, respectively. According to our deﬁnition, the adjusted

total strength score for the ﬁrm is equal to 1/5+2/4+1/3+0/5+1/6+1/4=1.45.

Assuming that the adjusted concern score can be calculated similarly and is equal to

1.35, the adjusted CSR score is 1.45 −1.35 =0.1.

Consistent with limited investor attention, Drake et al. (2012) ﬁnd that ﬁlings may be

requested days or weeks after their initial release.

We thank Michael Drake for providing access to the Google search data. Abnormal

Google SVI is computed as the raw Google SVI score for ﬁrm ion day t, minus the

average Google SVI score for the same day of the week for that ﬁrm over the prior

10 weeks, scaled by this average Google SVI score. Please see Drake et al. (2012) for

further details on how to compute abnormal searches.

20 G. Chen et al./Accounting & Finance

3.3. Descriptive statistics

Panel D of Table 1 provides the summary statistics of the main variables

employed in our regression analyses. Variable deﬁnitions are provided in the

Appendix. All non-logarithmic, continuous variables are winsorised at the 1st

and 99th percentile to alleviate the inﬂuence of outliers.

The mean (median)

value of Ψis 0.949 (0.922) with a standard deviation of 1.155. These statistics

suggest that there is wide variation in idiosyncratic volatility. PIN has a mean

(median) value of 0.114 (0.107) and a standard deviation of 0.050, indicating

less variation compared to idiosyncratic volatility. The mean (median) value of

CSR is −0.126 (−0.161), suggesting that the average (median) ﬁrm in our

sample has more CSR concerns than strengths. However, the standard

deviation is large (0.501), indicating that our sample consists of a wide range of

ﬁrms on the CSR spectrum. The mean (median) values for the EDGAR search

variables LOG(EDGAR_1),LOG(EDGAR_3) and LOG(EDGAR_1WK) are

2.595 (2.773), 3.293 (3.555) and 3.772 (4.094) respectively, suggesting that

investors’ searches for information from a newly released 10-K ﬁling increase

over time although at a decreasing rate. Mean (median) values for GSEARCH

(−5, −1),GSEARCH(0) and GSEARCH(1, 5) are 0.071 (0.008), 0.321 (0.035)

and 0.177 (0.014) respectively. As for the other variables, our summary

statistics are largely consistent with the previous literature (Gul et al., 2011).

4. CSR and price informativeness

4.1. Baseline model

We estimate the following baseline empirical model to analyse the relation

between CSR and price informativeness as measured by idiosyncratic volatility

(Ψ) and the probability of informed trading (PIN):

As a robustness check, we also winsorise all non-logarithmic, continuous variables at

the 2nd and the 98th percentile and at the 5th and 95th percentile, respectively. CSR

remains economically and statistically signiﬁcant in the idiosyncratic volatility (Ψ) and

the probability of informed trading (PIN) regressions when we adopt these alternative

winsorisation thresholds.

Following the previous literature (Ferreira and Laux 2007; Ferreira et al., 2011; Gul

et al., 2011), we select the panel data ﬁxed eﬀect model for the price informativeness test

to keep our study comparable with prior studies. The unit-root test suggests that our

panel data is stationary and does not possess a unit root. We also rerun the baseline

model with random eﬀects and ﬁnd quantitively similar results.

G. Chen et al./Accounting & Finance 21

Ψor PINðÞ

i,t¼β0þβ1CSRi,t1þβ2SIZEi,t1þβ3MBi,t1þβ4ROEi,t1

þβ5VROEi,t1þβ6LEVi,t1þβ7AGEi,t1þβ8DIVi,t1

þβ9DIVERi,t1þβ10ANALYSTSi,t1þβ11 BETAi,t1

þβ12EQi,t1þβ13 INSTRADEi,t1þβ14 CGSCOREi,t1

þβ15INSTNUMi,t1þβ16 FREQi,t1þdtþdiþɛi,t

(4)

where iindexes the ﬁrm and tindexes the year. Year and ﬁrm ﬁxed eﬀects are

denoted by d

and d

, respectively.

To mitigate concerns of omitted variable bias, we control for an exhaustive

set of covariates that can potentially aﬀect stock price informativeness as

suggested in the literature (Gelb and Strawser, 2001; Piotroski and Roulstone,

2004; Jin and Myers, 2006; Ferreira and Laux, 2007; Fernandes and Ferreira,

2009; Gul et al., 2011; Li et al., 2014). These control variables include market

capitalisation (SIZE), market-to-book ratio (MB), return on equity (ROE),

volatility of return on equity (VROE), book leverage (LEV), ﬁrm age (AGE),

issuance of dividends (DIV), corporate diversiﬁcation in the form of multiple

operating segments (DIVER), analyst coverage (ANALYSTS), ﬁrm beta

(BETA), earnings quality (EQ), insider trading (INSTRADE), corporate

governance (CGSCORE), number of institutional investors (INSTNUM),

and voluntary disclosures measured by the frequency of management forecasts

(FREQ).

Detailed variable deﬁnitions are provided in the Appendix. Year

and ﬁrm ﬁxed eﬀects are included in the regressions to control for any

unobservable macroeconomic shocks or time-invariant ﬁrm characteristics that

could simultaneously drive CSR activity and our measures of price informa-

tiveness.

Table 2 presents the baseline regression results on the relation between CSR

and price informativeness based on the model speciﬁed in Equation (4).

Column (1) reports the results of regressing idiosyncratic volatility (Ψ)onCSR

after controlling for year and ﬁrm ﬁxed eﬀects. As shown in the column, the

estimated coeﬃcient of CSR is positive and statistically signiﬁcant. Column (3)

provides similar results when regressing the probability of informed trading,

PIN,onCSR. The results in both columns provide some preliminary support

that greater CSR commitment is associated with more informative stock prices.

Columns (2) and (4) present the results of the full model with additional

controls as in Equation (4). As column (2) shows, the coeﬃcient on CSR

In untabulated results, we also include the number of conference calls as an additional

measure of voluntary disclosure. Our main results remain qualitatively similar when we

control for conference calls. Including conference calls in the regression shortens our

sample period and reduces our sample size (we obtain conference call data from

Thomson Reuters Street Events, which covers few ﬁrms before 2001). As a result, we

only include management forecasts as a control in our main regressions.

22 G. Chen et al./Accounting & Finance

Table 2

CSR and price informativeness

(1) (2) (3) (4)

ΨΨPIN PIN

Coeﬃcient t-Statistic Coeﬃcient t-Statistic Coeﬃcient t-Statistic Coeﬃcient t-Statistic

CSR 0.107*** 3.33 0.074** 2.52 0.007*** 5.16 0.006*** 4.99

SIZE −0.223*** −6.92 −0.006*** −3.12

MB −0.036 −1.38 −0.002** −2.01

ROE −0.090** −2.50 0.001 0.80

VROE 0.023* 1.88 0.000 0.09

LEV 0.058 0.54 0.002 0.49

AGE −0.616*** −5.91 −0.006 −1.33

DIV 0.097** 2.43 −0.002 −0.81

DIVER 0.015 0.27 −0.001 −0.39

ANALYSTS −0.212*** −7.53 −0.003** −2.46

BETA −0.234*** −9.61 −0.013*** −11.12

EQ 0.013*** 3.08 0.000 0.83

INSTRADE −0.660** −2.28 −0.016* −1.70

CGSCORE 0.276*** 4.68 0.013*** 5.30

INSTNUM −0.378*** −5.79 −0.041*** −6.26

FREQ −0.012 −0.82 −0.001** −2.36

Year FE Yes Yes Yes Yes

Firm FE Yes Yes Yes Yes

Observations 11,312 11,312 11,312 11,312

Adj. R

0.676 0.714 0.692 0.744

This table reports the baseline regression results of CSR and stock price informativeness. The sample period is from 1995 to 2010. Coeﬃcients are

estimated based on the model presented in Equation (4). The dependent variables are idiosyncratic volatility (Ψ) in columns (1) and (2) and the

probability of informed trading (PIN) in columns (3) and (4). The main variable of interest is CSR score (CSR). Independent variables are lagged

by one year relative to the dependent variable as described in Equation (4). Detailed variable deﬁnitions, including control variables, are

described in the Appendix. Standard errors are adjusted for heteroskedasticity and clustered by ﬁrm.***, ** and * denote statistical signiﬁcance at

the 1, 5 and 10 percent levels, respectively.

G. Chen et al./Accounting & Finance 23

(coeﬃcient =0.074, t-statistic =2.52) remains positive and statistically signif-

icant after controlling for the other covariates of Ψ. The relation between CSR

and Ψis also economically signiﬁcant. A one standard deviation increase in

CSR is associated with a 3.21 percent standard deviation increase in Ψ.

The

economic magnitude of CSR is comparable to some of the important covariates

of price informativeness previously documented in the literature.

The control

variables generally have the expected signs and are consistent with the ﬁndings

of prior studies (Gul et al., 2011). Column (4) displays the results regressing

PIN on CSR and other covariates. The coeﬃcient on CSR in the regression of

PIN is positive (0.006) and is also statistically signiﬁcant at the 1 percent level

(t-statistic =4.99). A one standard deviation increase in CSR is associated with

a 6.01 percent standard deviation increase in PIN.

In summary, we show a

strong positive association between CSR and measures of price informativeness

after controlling for year and ﬁrm ﬁxed eﬀects and covariates documented in

the prior literature. The ﬁndings in Table 2 support hypothesis H1a and are

consistent with the prediction of the ethical and reputational view that greater

CSR encourages information acquisition and trading and enhances the ﬂow of

information into stock prices.

4.2. CSR and information acquisition

In this section, we analyse the relation between CSR and investor

information search. To investigate Hypotheses H2a and H2b, we ﬁrst

examine investors’ search activities for new 10-K ﬁlings using the following

model:

This number is calculated as 0.074*0.501/1.155. 0.074 is the coeﬃcient of CSR in

column (2) of Table 2. 0.501 and 1.155 are the standard deviations of CSR and

idiosyncratic volatility (Ψ) from Table 1.

For example, based on the estimates in Ferreira and Laux (2007), a one standard

deviation decrease in the G-index (better corporate governance) is associated with

1.66% standard deviation increase in idiosyncratic volatility. This number is calculated

as 0.0129*2.833/2.198. 0.0129 is the coeﬃcient of the G-index in column (2) of their table

3. 2.833 and 2.198 are the standard deviations of the G-index and idiosyncratic volatility

in their sample presented in their table 2.

This number is calculated as 0.006*0.501/0.05. 0.006 is the coeﬃcient of CSR in

column (4) of Table 2. 0.501 and 0.05 are the standard deviations of CSR and

probability of informed trading (PIN) from Table 1.

We select the panel data ﬁxed eﬀect model for the new 10-K ﬁling search test to keep

our study comparable with prior studies (Drake et al. 2015). The unit-root test suggests

that our panel data is stationary and does not possess a unit root. A regression model

with random eﬀects yields quantitively similar results.

24 G. Chen et al./Accounting & Finance

LOGðEDGAR Di,tÞ¼β0þβ1CSRi,t1þβ2LOGFILESIZEi,t

þβ3LOGSUPPLYi,tþβ4SHORTi,t1

þβ5SIZEi,t1þβ6ANALYSTSi,t1þβ7INSTOWNi,t1

þβ8INSTNUMi,t1þβ9ROA

i,t1þβ10MBi,t1

þβ11TURNOVERi,t1þβ12 LEVi,t1

þβ13NYSEi,t1þβ14 RET m 3,mðÞ

tþβ15LOSSt1

þdtþdmþdwþdiþɛi,t

(5)

where iindexes the ﬁrm and tindexes the year. EDGAR_D,D=1,3,and1WK

are the counts of individual requests for 10-K ﬁnancial reports on the date that

the ﬁling became publicly available (EDGAR_1), up to 3 days after the ﬁling

date (EDGAR_3) and up to 1 week after the ﬁling date (EDGAR_1WK).

Because of data constraints, the time period is limited to the years between 2003

and 2010. Similar to Drake et al. (2015), we further control for 10-K ﬁle size

(LOGFILESIZE), the supply of ﬁrm ﬁlings on EDGAR (LOGSUPPLY), the

number of shorted shares (SHORT), market capitalisation (SIZE), analyst

coverage (ANALYSTS), institutional ownership (INSTOWN), the number of

institutional investors (INSTNUM), the magnitude of ﬁrm performance (|

ROA|), market-to-book ratio (MB), trading volume (TURNOVER), book

leverage (LEV), and whether the ﬁrm’s stock is traded on the New York Stock

Exchange (NYSE), cumulative returns over the past three months (RET(m-3,

m)), and whether the ﬁrm had an earnings loss (LOSS). We further control for

year, month, weekday and ﬁrm ﬁxed eﬀects, denoted by d

and d

respectively. Similar to Hirshleifer et al. (2009), we cluster the standard errors

by ﬁrm and ﬁling date to account for correlation of the residuals across ﬁrm

and time.

Table 3 presents the results. Column (1) shows that greater CSR is associated

with more requests on the date that a 10-K ﬁling is available on EDGAR, as

evidenced by a coeﬃcient of 0.096 at the 10 percent level for CSR. A one

standard deviation increase in CSR is associated with a 3.16 percent standard

deviation increase in searches for new 10-K ﬁlings available through EDGAR

on the ﬁling date (LOG(EDGAR_1)).

Columns (2) and (3) both provide

evidence that is consistent with the ﬁndings in column (1). Column (2), using a

three-day ﬁling request window, shows a statistically signiﬁcant and positive

coeﬃcient for CSR (coeﬃcient =0.088, t-statistic =2.05), while column (3),

The economic magnitudes for the results of EDGAR new 10-K ﬁling searches and

abnormal Google searches are calculated in a similar manner as stated in footnotes 23

and 25. To save space, we do not specify the calculations of these economic magnitudes.

G. Chen et al./Accounting & Finance 25

using a one-week ﬁling request window, also shows a positive and signiﬁcant

coeﬃcient (coeﬃcient =0.078, t-statistic =2.43) for CSR.

These results

support hypothesis H2a and provide evidence that more socially responsible

Table 3

CSR and new 10-K ﬁling requests

(1) (2) (3)

LOG(EDGAR_1) LOG(EDGAR_3) LOG(EDGAR_1WK)

Coeﬃcient t-Statistic Coeﬃcient t-Statistic Coeﬃcient t-Statistic

CSR 0.096* 1.80 0.088* 2.05 0.078** 2.43

LOGFILESIZE 0.027*** 7.84 0.025*** 6.29 0.024*** 4.61

LOGSUPPLY −0.179 −1.22 −0.271** −2.69 −0.247** −2.69

SHORT 0.258 1.06 0.183 0.83 0.135 0.61

SIZE 0.117*** 5.43 0.144*** 3.53 0.165*** 4.29

ANALYSTS 0.030 1.01 0.019 0.59 0.013 0.32

INSTOWN −0.092 −0.60 −0.100 −0.80 −0.126 −0.94

INSTNUM 0.080 1.52 0.123* 1.81 0.140* 1.87

|ROA|0.304** 2.29 0.193 1.42 0.212 1.60

MB 0.009 0.19 −0.017 −0.22 −0.011 −0.15

TURNOVER 0.221 1.60 0.316** 2.55 0.360** 2.78

LEV 0.014 0.13 0.154 1.15 0.223 1.36

NYSE −0.163** −2.84 −0.138 −1.62 −0.123 −1.27

CRET(m-3, m) 0.037 0.54 −0.020 −0.17 −0.066 −0.53

LOSS 0.006 0.34 0.044** 2.28 0.054** 2.90

Year FE Yes Yes Yes

Month FE Yes Yes Yes

Weekday FE Yes Yes Yes

Firm FE Yes Yes Yes

Observations 9,517 9,517 9,517

Adj. R

0.602 0.713 0.751

This table reports the regression results of CSR and requests for new 10-K ﬁlings. The sample

period is from 2003 to 2010. Coeﬃcients are estimated based on the model presented in

Equation (5). The dependent variable is 10-K requests on the ﬁling date (LOG(EDGAR_1)),

in a 3-day window beginning from the ﬁling date (LOG(EDGAR_3)), and in a 1-week

window starting from the ﬁling date (LOG(EDGAR_1WK)). The main variable of interest is

CSR score (CSR). Independent variables generally lag dependent variables by one year as

described in Equation (5). Detailed variable deﬁnitions, including control variables, are

described in the Appendix. Standard errors are adjusted for heteroskedasticity and clustered

by ﬁrm and ﬁling date. ***, ** and * denote statistical signiﬁcance at the 1, 5 and 10 percent

levels, respectively.

In terms of economic signiﬁcance, a one standard deviation increase in CSR is

associated with a 2.60 percent standard deviation increase in three-day new 10-K ﬁling

searches (LOG(EDGAR_3)) and a 2.17 percent standard deviation increase in one-week

new 10-K ﬁling searches (LOG(EDGAR_1WK)) on EDGAR, respectively.

26 G. Chen et al./Accounting & Finance

ﬁrms experience greater searches by investors as evidenced by more requests for

new 10-K ﬁlings.

For our second test of the hypotheses on information search (Hypotheses

H2a and H2b), we estimate the following empirical model to examine investors’

search activities on Google for ﬁrm-speciﬁc information around earnings

announcements:

GSEARCHi,d,t¼β0

þβ1EARNINGS ANNOUNCEMENT 5, 1½

i,d,t

CSRi,t1þβ2EARNINGS ANNOUNCEMENT 0½

i,d,t

CSRi,t1þβ3EARNINGS ANNOUNCEMENT 1,5½

i,d,t

CSRi,t1þβ4CSRi,t1

þβ5EARNINGS ANNOUNCEMENT 5, 1½

i,d,t

þβ6EARNINGS ANNOUNCEMENT 0½

i,d,t

þβ7EARNINGS ANNOUNCEMENT 1,5½

i,d,t

þβ8NEWSi,t1þβ9RET

i,t1þβ10TURNOVERi,t1

þβ11BIDASKi,t1þβ12 ANMi,t1

þβ13SIZEi,t1þβ14 MBi,t1þβ15ANALYSTSi,t1

þβ16INSTOWNi,t1þβ17 QTR4i,d,t

þβ18LOSSi,t1þβ19 ROA

i,t1þβ20CRET m 3, mðÞ

i,d,t

þβ21FORECASTi,d,tþdtþdmþdwþdiþɛi,d,t

(6)

where iindexes the ﬁrm, dindexes the day and tindexes the year. GSEARCH is

the daily abnormal Google SVI, EARNINGS_ANNOUNCEMENT[−5,-1] is

an indicator set to one if the day is in the window from 5 days to 1 day prior to

the earnings announcement, EARNINGS_ANNOUNCEMENT[0] is an indi-

cator for the date of the announcement and EARNINGS_ANNOUNCEMENT

We acknowledge that investors can obtain annual ﬁnancial statement data from

sources other than EDGAR, such as the ﬁrm’s own website. Thus, there is potential

measurement error because we only observe requests for ﬁlings on EDGAR. This

measurement error adds noise to the information search measure and biases against

ﬁnding any signiﬁcant results. In addition, our Google search results are consistent with

the EDGAR ﬁling request results, providing supportive evidence of our inference.

We select the panel data ﬁxed eﬀect model for the Google search test to keep our study

comparable with prior studies (Drake et al. 2012). The unit-root test suggests that our

panel data is stationary and does not possess a unit root. We further ﬁnd quantitively

similar results with a random eﬀects model.

G. Chen et al./Accounting & Finance 27

[1, 5] is an indicator set to one if the day is in the period from 1 to 5 days after

the earnings announcement. Because of data constraints, the time period is

limited to the years between 2005 and 2008. Year, month, weekday and ﬁrm

ﬁxed eﬀects are denoted by d

and d

, respectively. Following the

literature (Drake et al., 2012), we control for the number of news articles that

mention the ﬁrm (NEWS), the absolute value of raw stock returns (|RET|),

trading volume (TURNOVER), the bid-ask spread (BIDASK), the number of

ﬁrms announcing earnings on the same day (ANM), market capitalisation

(SIZE), market-to-book ratio (MB), analyst coverage (ANALYSTS), institu-

tional ownership (INSTOWN), an indicator for earnings announcements made

in the fourth quarter (QTR4), an indicator for earnings loss (LOSS),

performance (|ROA|), and cumulative returns over the past three months

(RET(m-3, m)). We also include FORECAST, an indicator set to one if the

ﬁrm issues a management forecast concurrently with the earnings announce-

ment to control for additional voluntary disclosures. As in Hirshleifer et al.

(2009), we cluster the standard errors by ﬁrm and earnings announcement date

to account for correlation in the residuals across ﬁrm and time.

Table 4 presents the results. Consistent with Drake et al. (2012), we ﬁnd

greater investor demand for company information around earnings announce-

ments in the periods during and after an earnings announcement. Furthermore,

the results suggest that more socially responsible ﬁrms have incrementally more

Google searches in the period around an earnings announcement. On the date

of the earnings announcement, ﬁrms that are more socially responsible

experience greater search activity for company information as evidenced by the

positive and statistically signiﬁcant coeﬃcient of EARNINGS_ANNOUNCE-

MENT[0]×CSR. Subsequently, in the ﬁve-day period after an earnings

announcement, the interaction term EARNINGS_ANNOUNCEMENT

[1,5] ×CSR is positive and statistically signiﬁcant at the 5 percent level. In

terms of economic signiﬁcance, a one standard deviation increase in CSR is

associated with a 5.14 percent (5.07 percent) standard deviation increase in the

abnormal Google searches on the date of (in the ﬁve-day period after) an

earnings announcement. Thus, ﬁrms with greater CSR experience higher

abnormal Google searches surrounding an earnings announcement, suggesting

that there is more search activity by investors for information regarding ﬁrms

that are more socially responsible.

Overall, the results in Tables 3 and 4 support hypothesis H2a and suggest

that more socially responsible ﬁrms are associated with greater investor search

activity for ﬁrm-speciﬁc information around major disclosure events. These

results provide direct evidence that greater CSR commitment encourages

investors to actively acquire information and lends further support to the

ethical and reputational view of CSR.

28 G. Chen et al./Accounting & Finance

Table 4

CSR and abnormal Google searches

GSEARCH

Coeﬃcient t-Statistic

EARNINGS_ANNOUNCEMENT[−5, −1]×CSR 0.021 1.06

EARNINGS_ANNOUNCEMENT[0]×CSR 0.163** 2.11

EARNINGS_ANNOUNCEMENT[1,5]×CSR 0.146** 2.41

CSR −0.053 −1.47

EARNINGS_ANNOUNCEMENT[−5,-1] 0.021* 1.68

EARNINGS_ANNOUNCEMENT[0] 0.324*** 3.21

EARNINGS_ANNOUNCEMENT[1,5] 0.180*** 5.13

NEWS 0.012*** 2.97

|RET|2.179*** 4.09

TURNOVER 0.096 1.42

BIDASK −0.002 −0.05

ANM −0.022 −0.50

SIZE −0.046* −1.89

MB 0.012 0.70

ANALYSTS −0.042 −1.29

INSTOWN 0.073 0.60

QTR4 −0.004 −0.38

LOSS 0.009 0.27

|ROA|−0.045 −0.52

CRET(m-3, m) 0.062** 2.05

FORECAST −0.013 −0.79

Year FE Yes

Month FE Yes

Weekday FE Yes

Firm FE Yes

Observations 197,403

Adj. R

0.103

This table reports the regression results of CSR and abnormal Google searches. Our sample

includes S&P 500 ﬁrms between 2005 and 2008. Coeﬃcients are estimated based on the model

presented in Equation (6). The dependent variable is daily abnormal Google search

(GSEARCH). The main variables of interest are the interaction of the pre-earnings

announcement date indicator and CSR score (EARNINGS_ANNOUNCEMENT[−5,-

1] ×CSR), the interaction of the earnings announcement date indicator and CSR score

(EARNINGS_ANNOUNCEMENT[0] ×CSR), and the interaction of the post earnings

announcement date indicator and CSR score (EARNINGS_ANNOUNCEMENT

[1,5] ×CSR). Independent variables other than the (pre- and post) earnings announcement

date indicator generally lag dependent variables by one year as described in Equation (6).

Detailed variable deﬁnitions, including control variables, are described in the Appendix.

Standard errors are adjusted for heteroskedasticity and clustered by ﬁrm and earnings

announcement day. ***, ** and * denote statistical signiﬁcance at the 1, 5 and 10 percent

levels, respectively.

G. Chen et al./Accounting & Finance 29

4.3. CSR and future earnings captured in stock prices

We use an alternative measure of price informativeness as a robustness check

and to examine Hypotheses H3a and H3b.

Namely, we examine how CSR

aﬀects the extent to which current stock prices reﬂect future ﬁrm-speciﬁc

information as captured in FERC. The intuition is that if CSR incentivises (or

discourages) information acquisition and trading, then current stock returns

should be more (or less) predictive of future ﬁrm earnings. We follow prior

literature (Kothari and Sloan, 1992; Choi et al., 2011) and estimate the

following model:

RETi,t¼β0þβ1CSRi,tEARNi,t1þβ2CSRi,tEARNi,tþβ3CSRi,t

EARNi,tþ1þβ4CSRi,tRETi,tþ1þβ5CSRi,tEARN Pi,t1

þβ6EARN Pi,t1þβ7EARNi,t1þβ8EARNi,tþβ9EARNi,tþ1

þβ10RETi,tþ1þβ11 CSRi,tþβ12 SIZEi,tþβ13ATGROWTHi,t

þβ14LOSSi,tþβ15 INSTOWNi,tþβ16 STDEARNi,t

þβ17BETAi,tþdtþdiþɛi,t

(7)

In Equation (7), RET

i,t

is the annual return for ﬁrm iat the end of ﬁscal year

t,EARN

i,t−1

,EARN

i,t

and EARN

i,t+1

represent earnings per share for ﬁscal

year t−1, t, and t+1 deﬂated by the stock price at the beginning of the ﬁscal

year t. The coeﬃcient of interest is β

. If CSR increases the extent to which

future earnings is reﬂected in current period stock returns, then we would

expect to observe a signiﬁcantly positive β

. Following prior research, we

controls for future ﬁrm-level stock returns (RET

t+1

), ﬁrm size (SIZE), total

ﬁrm asset growth (ATGROWTH), an indicator set to 1 (and 0 otherwise) if the

ﬁrm experienced a loss in the current year (LOSS), percentage of shares held by

institutional investors (INSTOWN), future earnings volatility (STDEARN)

To provide more evidence, we adopt price delay as an additional measure of price

eﬃciency as in Dong et al. (2016). Price delay measures the average delay with which

stock prices respond to information. Greater investor information search and trading

should facilitate faster incorporation of information into stock prices. We ﬁnd that CSR

is negatively associated with price delay, suggesting that CSR attracts more investor

information search and trade and facilitates information ﬂow into stock prices more

quickly.

We choose the panel data ﬁxed eﬀect model for the FERC test to keep our study

comparable with prior studies (Choi et al., 2011). The unit-root test suggests that our

panel data is stationary and does not possess a unit root. Our results remain

quantitatively similar when we use a random eﬀects model.

30 G. Chen et al./Accounting & Finance

and ﬁrm beta (BETA). We also control for earnings persistence (EARN_P

t−1

)

and its interaction with CSR to account for the potential diﬀerences in earnings

persistence among ﬁrms with diﬀerent CSR involvement and its eﬀect on

FERC. We further include year and ﬁrm ﬁxed eﬀects (d

The regression results on FERC are reported in Table 5. As the table shows,

the coeﬃcient for the interaction term CSR

×EARN

t+1

is positive and

statistically signiﬁcant (coeﬃcient =0.319, t-statistic =2.00).

The results

suggest that after controlling for a host of variables suggested by prior

literature, ﬁrms with greater CSR commitment are associated with a higher

FERC. That is, stock prices of socially responsible ﬁrms reﬂect more

information about future earnings. This table provides further evidence to

support our main ﬁndings in Table 2.

5. Interpretation of the relation between CSR and price informativeness

5.1. Lead-lag analysis

Prior studies suggest that the information embedded in stock prices can

impact corporate decisions such as investments (Chen et al., 2007), corporate

governance (Ferreira et al., 2011) and management forecasts (Zuo, 2016). It is

possible that managers learn of information in the stock price and make CSR

investments accordingly. Similar to Ramalingegowda and Yu (2012), we

conduct a lead-lag analysis by regressing contemporaneous Ψand PIN on CSR

levels in the lagged, current and lead year. If greater CSR commitment leads to

more informative stock prices, only lagged values of CSR should impact

contemporaneous levels of price informativeness. In contrast, if there is reverse

causality and (or) simultaneity, we would observe signiﬁcant associations

between lead and (or) current levels of CSR and contemporaneous price

informativeness.

Table 6 presents the lead-lag regression results. In column (1), when we

regress Ψon CSR

t+1

,CSR

and CSR

t−1

, only the coeﬃcient of CSR

t−1

statistically signiﬁcant (coeﬃcient =0.144, t-statistic =3.21). Similarly, in

column (2), when the dependent variable is PIN, the only statistically

signiﬁcant coeﬃcient is CSR

t−1

. Thus, in both columns, only past year CSR

Some studies use three-year future earnings instead of next year’s earnings in FERC

studies (Tucker and Zarowin, 2006). Therefore, we replicate our results using three-year

future earnings instead. Our results show that the interaction term of CSR and three-

year future earnings is statistically signiﬁcant at the 10% level. Our result could suggest

that the enhanced information acquisition from better CSR performance incorporates

more near-term future earnings into the stock price than longer-term future earnings. It

is also possible that the reduced signiﬁcance is due to measurement error in future

earnings (Collins et al., 1994).

G. Chen et al./Accounting & Finance 31

is positively associated with measures of stock price informativeness.

These

results further alleviate the concerns about reverse causality and simultaneity.

5.2. Propensity score matched sample

To address potential endogeneity concerns of the observed positive associ-

ation between CSR and informative stock prices, we use a propensity score

Table 5

CSR and future earnings captured in stock prices

RET

Coeﬃcient t-Statistic

CSR

×EARN

t−1

−0.146 −0.99

CSR

×EARN

0.017 0.09

CSR

×EARN

t+1

0.319** 2.00

CSR

×RET

t+1

−1.227*** −15.82

CSR

×EARN_P

t−1

0.061 1.51

EARN_P

t−1

−0.048* −1.85

EARN

t−1

0.481*** 4.02

EARN

1.156*** 11.51

EARN

t+1

−0.038** −2.37

RET

t+1

−0.022 −0.87

CSR

0.001 0.03

SIZE 0.354*** 20.84

ATGROWTH 0.114*** 4.87

LOSS 0.021 0.95

INSTOWN 0.128* 1.92

STDEARN 0.500*** 6.53

BETA 0.090*** 5.24

Year FE Yes

Firm FE Yes

Observations 11,141

Adj. R

0.381

This table reports the regression results of CSR and an alternative measure of price

informativeness, future earnings response coeﬃcient (or FERC). The sample period is from

1995 to 2010. The dependent variable is the current period stock returns (RET

). The main

variable of interest is the interaction term of CSR and earnings per share in the following year

(CSR

×EARN

t+1

). Contemporaneous control variables are used as described in Equa-

tion (8). Detailed variable deﬁnitions, including control variables, are described in the

Appendix. Standard errors are clustered by ﬁrm. ***, ** and * denote statistical signiﬁcance

at the 1, 5 and 10 percent levels respectively.

In untabulated results, we also include lagged Ψand PIN as additional controls and

ﬁnd that only past year CSR is signiﬁcantly positively associated with stock price

informativeness.

32 G. Chen et al./Accounting & Finance

matched (PSM) sample. We begin by estimating the likelihood of a ﬁrm having

high CSR using a probit model:

Prob High CSR ¼1ðÞ

i,t¼β0þβ1ATOi,tþβ2PMi,tþβ3CASHi,t

þβ4CFOi,tþβ5LEVi,tþβ6MBi,tþβ7SIZEi,t

þβ8RDi,tþβ9ADVERTISINGi,tþβ10LITi,t

þβ11CGSCOREi,tþdtþdjþɛi,t

(8)

Table 6

Lead-lag regressions of CSR and price informativeness

(1) ΔPIN (2)

Coeﬃcient t-Statistic Coeﬃcient t-Statistic

CSR

t+1

0.045 1.20 0.002 1.51

CSR

0.021 0.63 0.001 1.06

CSR

t−1

0.144*** 3.21 0.006*** 2.96

SIZE −0.270*** −7.53 −0.006** −2.43

MB −0.060* −1.96 −0.002 −1.55

ROE −0.076* −1.73 −0.001 −0.43

VROE 0.014 0.98 −0.000 −0.21

LEV 0.068 0.55 −0.002 −0.45

AGE −0.721*** −5.91 0.002 0.29

DD 0.098** 2.12 −0.001 −0.56

DIVER −0.012 −0.21 0.001 0.23

ANALYSTS −0.208*** −6.41 −0.003** −2.30

BETA −0.231*** −8.51 −0.011*** −8.15

EQ −0.010** −2.19 −0.000 −0.68

INSTRADE −0.441 −1.27 −0.018* −1.68

CGSCORE 0.288*** 3.65 0.014*** 4.43

INSTNUM −0.385*** −5.02 −0.044*** −5.05

FREQ −0.003 −0.18 −0.001 −1.34

Year FE Yes Yes

Firm FE Yes Yes

Observations 8,552 8,552

Adj. R

0.715 0.733

This table reports the lead-lag regression results of CSR and price informativeness (Ψand PIN).

The sample period is from 1995 to 2010. Column (1) reports the results with contemporaneous Ψ

as the dependent variable. Column (2) displays the results with contemporaneous PIN as the

dependent variable. CSR

t+1

,CSR

and CSR

t−1

are the lead, current and lagged CSR scores,

respectively. Control variables generally lag dependent variables by 1 year as described in

Equation (5). Detailed variable deﬁnitions, including control variables, are described in the

Appendix. Standard errors are adjusted for heteroskedasticity and clustered by ﬁrm. ***, ** and

* denote statistical signiﬁcance at the 1, 5 and 10 percent levels.

G. Chen et al./Accounting & Finance 33

where iindexes the ﬁrm and tindexes the year. High CSR is coded as one if

CSR is greater than the median and zero otherwise. Our prediction model for

CSR is based on the model developed by Lys et al. (2015), which synthesises the

determinants of CSR documented in the previously literature. Speciﬁcally,

these determinants are sales to assets (ATO), proﬁt margin (PM), cash holdings

(CASH), operating cash ﬂows (CFO), book leverage (LEV), market-to-book

(MB), ﬁrm size (SIZE), research and development expenses (RD), advertising

(ADVERTISING), litigation expenses (LIT) and corporate governance

(CGSCORE). Year (d

) and industry (d

) ﬁxed eﬀects are included because

certain macroeconomic and industry factors may inﬂuence ﬁrm commitment to

CSR. We then match ﬁrms with high CSR to ﬁrms with low CSR based on the

estimated propensity score.

Panel A of Table 7 reports the results of estimating Equation (8). In the pre-

match sample, several characteristics diﬀerentiate high from low CSR ﬁrms as

shown in column (1). These variables are CFO,LEV,MB,SIZE and

CGSCORE. In the post-match sample, none of those characteristics are

statistically signiﬁcant as shown in column (2), suggesting that the diﬀerences in

observed ﬁrm characteristics largely disappear in the matched sample. Panel B

of Table 6 reports the estimation results after re-running the baseline model,

following Equation (4), using the matched sample. There remains a signiﬁcant,

positive and signiﬁcant association between CSR and price informativeness,

measured by Ψand PIN. The economic and statistical signiﬁcance of the

coeﬃcient for CSR is comparable to the coeﬃcient of CSR in the baseline

regression in Table 2. Overall, this set of results mitigates the endogeneity

concern that factors which contribute to a ﬁrm’s choice of CSR involvement

explain the diﬀerential in price informativeness.

5.3. CSR strengths and concerns

Our measure of CSR follows prior literature (Deng et al., 2013) where we

compute a net score, subtracting adjusted CSR concerns from strengths. CSR

strengths are reﬂective of proactive ﬁrm policies such as pollution prevention

and a diverse labour force, while concerns are often associated with public

incidents such as workplace accidents, lawsuits and regulatory problems. While

a commitment to CSR can incentivise information acquisition, CSR concerns,

such as the 2010 British Petroleum oil spill or other major public incidents,

might also draw attention and trigger greater investor search for information.

To investigate the diﬀerential relation between CSR strengths and weaknesses

on price informativeness, we separate our CSR score into a score for CSR

strengths (CSR_STR) and a score for CSR concerns (CSR_CON) and rerun

our baseline model in Equation (4).

Table 8 presents the results. We ﬁrst examine the inclusion of CSR_STR and

CSR_CON separately. As shown in columns (1) and (4), the coeﬃcient of

CSR_STR is positive and statistically signiﬁcant at the 1 percent level for both

34 G. Chen et al./Accounting & Finance

Table 7

Propensity score matched sample

Panel A: Probit regressions before and after propensity score matching

Prob(High CSR =1)

(1) (2)

Pre-match Post-match

Coeﬃcient t-Statistic Coeﬃcient t-Statistic

ATO 0.026 0.38 0.079 0.99

PM −0.019 −0.17 −0.010 −0.05

CASH −0.291 −1.06 −0.446 −1.07

CFO 1.139*** 3.32 0.450 0.99

LEV −0.719*** −3.23 0.318 1.19

MB 0.213*** 4.46 −0.056 −0.95

SIZE 0.149*** 5.04 0.049 1.44

RD 0.157 0.88 0.035 0.11

ADVERTISING 0.377 0.57 −0.387 −0.28

LIT 1.750 0.69 1.636 0.51

CGSCORE 0.467*** 2.75 −0.019 −0.09

CONSTANT −1.774* −1.72 −0.720 −0.71

Year FE Yes Yes

Industry FE Yes Yes

Observations 11,259 9,649

Adj. R

0.057 0.047

Panel B: Regressions of CSR and price informativeness using propensity score matched sample

(1)

ΔPIN (2)

ΨPIN

Coeﬃcient t-Statistic Coeﬃcient t-Statistic

CSR 0.065** 1.98 0.005*** 4.05

SIZE −0.225*** −5.95 −0.008*** −2.96

MB 0.003 0.11 −0.000 −0.26

ROE −0.107** −2.33 −0.001 −0.49

VROE 0.018 1.08 0.000 0.50

LEV 0.111 0.93 −0.004 −0.85

AGE −0.383*** −3.34 −0.007 −1.27

DIV 0.033 0.64 −0.002 −1.11

DIVER −0.070 −1.13 −0.001 −0.47

ANALYSTS −0.182*** −5.43 −0.003* −1.75

BETA −0.213*** −6.96 −0.012*** −8.25

EQ 0.012** 2.48 0.000 0.70

(continued)

G. Chen et al./Accounting & Finance 35

Ψand PIN. However, in columns (2) and (5) the coeﬃcient of CSR_CON is

statistically insigniﬁcant for Ψand signiﬁcantly negative for PIN. Similar

results are presented in column (3) and (6) when both CSR_STR and

CSR_CON are included in the same regression. Overall, this table provides

evidence suggesting that ﬁrms’ proactive involvement in CSR activities, as

measured by CSR strengths, draws more trading and searches for information

and promotes more informative stock prices.

5.4. Alternative explanations

We further address several alternative explanations of our results in this

subsection. First, it is possible that CSR is implemented strategically to give

ﬁrms comparative advantages, such as fostering stronger customer relation-

ships (Brown and Dacin, 1997) or attracting and retaining higher quality

workers (Greening and Turban, 2000). The distinct investments in CSR may

Table 7 (continued)

Panel B: Regressions of CSR and price informativeness using propensity score matched sample

(1)

ΔPIN (2)

ΨPIN

Coeﬃcient t-Statistic Coeﬃcient t-Statistic

INSTRADE −0.715** −2.37 −0.014 −1.18

CGSCORE 0.302*** 4.31 0.014*** 4.26

INSTNUM −0.386*** −4.72 −0.040*** −4.95

FREQ −0.006 −0.38 −0.001 −1.25

Year FE Yes Yes

Firm FE Yes Yes

Observations 9,649 9,649

Adj. R

0.751 0.775

This table reports estimation results from a propensity score matched (PSM) sample. The

sample period is from 1995 to 2010. A probit model is ﬁrst used to estimate the likelihood of

ﬁrms exhibiting high levels of CSR. Contemporaneous independent variables are used as

described in Equation (7). The estimation result is reported in column (1) of Panel A, with the

dependent variable being a dichotomous variable that equals 1 if a ﬁrm has a high level of

CSR. We then match ﬁrms based on the estimated propensity scores. Column (2) of Panel A

reports the post-match regression results. Panel B reports the estimation results when we

regress measures of price informativeness on CSR using the matched sample. Independent

variables lag dependent variables by 1 year as described in Equation (4). Detailed variable

deﬁnitions are given in the Appendix. Standard errors are adjusted for heteroskedasticity and

ﬁrm-level clustering. *, ** and *** indicate a signiﬁcance level of less than 0.10, 0.05 and 0.01,

respectively, based on a two-tailed test.

36 G. Chen et al./Accounting & Finance

Table 8

Diﬀerential impact of CSR strengths and concerns

ΨPIN

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

CSR_STR 0.122*** 0.120*** 0.006*** 0.007***

(3.32) (3.29) (5.23) (5.36)

CSR_CON 0.049 0.040 −0.003* −0.003**

(1.26) (1.06) (−1.91) (−2.23)

SIZE −0.226*** −0.225*** −0.227*** −0.007*** −0.006*** −0.006***

(−7.01) (−7.00) (−7.04) (−3.19) (−3.11) (−3.15)

MB −0.033 −0.037 −0.033 −0.002* −0.002** −0.002*

(−1.26) (−1.41) (−1.24) (−1.86) (−2.16) (−1.91)

ROE −0.090** −0.088** −0.090** 0.001 0.001 0.001

(−2.51) (−2.45) (−2.51) (0.82) (0.87) (0.79)

VROE 0.023* 0.022* 0.022* 0.000 0.000 0.000

(1.85) (1.86) (1.84) (0.03) (0.07) (0.07)

LEV 0.038 0.063 0.036 0.001 0.003 0.001

(0.35) (0.59) (0.33) (0.27) (0.69) (0.31)

AGE −0.582*** −0.604*** −0.573*** −0.004 −0.007 −0.005

(−5.50) (−5.72) (−5.36) (−0.91) (−1.45) (−1.08)

DIV 0.096** 0.096** 0.095** −0.002 −0.002 −0.002

(2.38) (2.38) (2.36) (−0.87) (−0.80) (−0.84)

DIVER 0.018 0.019 0.019 −0.001 −0.001 −0.001

(0.32) (0.34) (0.35) (−0.27) (−0.33) (−0.32)

ANALYSTS −0.210*** −0.212*** −0.210*** −0.003** −0.003** −0.003**

(−7.46) (−7.53) (−7.45) (−2.38) (−2.51) (−2.40)

BETA −0.234*** −0.239*** −0.235*** −0.013*** −0.013*** −0.013***

(−9.70) (−9.76) (−9.71) (−11.11) (−11.26) (−11.09)

EQ 0.012*** 0.014*** 0.012*** 0.000 0.000 0.000

(2.88) (3.23) (2.91) (0.68) (0.99) (0.63)

INSTRADE −0.667** −0.668** −0.669** −0.016* −0.016* −0.016*

(−2.30) (−2.30) (−2.31) (−1.75) (−1.72) (−1.73)

CGSCORE 0.240*** 0.294*** 0.241*** 0.011*** 0.014*** 0.011***

(3.76) (5.26) (3.78) (4.78) (5.95) (4.74)

INSTNUM −0.381*** −0.370*** −0.379*** −0.041*** −0.041*** −0.042***

(−5.81) (−5.67) (−5.78) (−6.23) (−6.22) (−6.25)

FREQ −0.011 −0.010 −0.010 −0.001** −0.001** −0.001**

(−0.75) (−0.70) (−0.71) (−2.23) (−2.29) (−2.31)

Year FE Yes Yes Yes Yes Yes Yes

Firm FE Yes Yes Yes Yes Yes Yes

Observations 11,312 11,312 11,312 11,312 11,312 11,312

Adj. R

0.714 0.714 0.714 0.744 0.743 0.744

This table reports the regression results of CSR strengths (CSR_STR) and concerns

(CSR_CON) and price informativeness. The sample period is from 1995 to 2010. The

dependent variables are idiosyncratic volatility (Ψ) in columns (1)–(3) and the probability of

informed trading (PIN) in columns (4)–(6). The main variables of interest are CSR strengths

(CSR_STR) and concerns (CSR_CON). Independent variables generally lag dependent

variables by 1 year as described in Equation (4). Detailed variable deﬁnitions, including

control variables, are described in the Appendix. Standard errors are adjusted for

heteroskedasticity and clustered by ﬁrm. T-statistics are reported in parentheses. ***, **

and * denote statistical signiﬁcance at the 1, 5 and 10 percent levels.

G. Chen et al./Accounting & Finance 37

result in lower systematic risk exposures (B´

enabou and Tirole, 2010) and

thereby mechanically drive higher idiosyncratic volatilities among ﬁrms with

higher CSR. To further explore this strategic view of CSR, we directly control

for the systematic risk, BETA, and continue to ﬁnd a positive and signiﬁcant

relation between CSR and price informativeness. Our main results continue to

hold when we exclude the ﬁnancial crisis period during which high CSR ﬁrms

may display a diﬀerent systematic risk due to their resilience (B´

enabou and

Tirole, 2010). While we cannot completely rule out the strategic view of CSR,

our results suggest that the better information and trading environment of

higher CSR ﬁrms plays a role in facilitating information ﬂow into stock prices.

Second, greater CSR expenditures may signal better future ﬁnancial

performance (Lys et al., 2015) and attract more information search. Foremost,

it is still under debate whether CSR is linked to ﬁnancial performance. An

analysis of 167 studies by Margolis et al. (2007) ﬁnd only a small positive

correlation between CSR and ﬁnancial performance. Furthermore, even if CSR

can signal future ﬁnancial performance, it is not clear as to why only CSR

strength can attract more information search. Indeed, if CSR concerns indicate

poorer future ﬁnancial performance, they can also attract greater information

search by investors, who actively exploit arbitrage opportunities. However,

Table 8 shows that greater CSR concern is not associated with more

information search, which does not support the CSR signalling view.

Third, because prior and concurrent research suggests that the disclosures of

CSR activities can potentially attract more dedicated institutional investors and

analyst coverage (Dhaliwal et al., 2011) and enhance price informativeness

(Grewal et al., 2020), we further control for CSR disclosures to verify that our

results are not driven by CSR disclosures. We measure CSR disclosure by a

dummy variable for whether the ﬁrm issued a stand-alone CSR report, as

deﬁned by Thomson Reuters Asset4, in our regressions of price informative-

ness. In untabulated results, we ﬁnd that CSR ratings remain a positive and

signiﬁcant factor when controlling for the dissemination of CSR activities.

further alleviate this concern, we examine investors’ requests for new 10-K

ﬁlings that rarely contain discussions of CSR issues. A survey by KPMG shows

that only 1 percent of the top 100 ﬁrms in the United States integrated CSR

reports into their 10-K ﬁlings during 2005–2008. Therefore, requests for new

10-K ﬁlings up to 2008 should mostly consist of searches for non-CSR

disclosures. We rerun the regression of 10-K ﬁling requests using a sample up to

year 2008. In untabulated results, we continue to ﬁnd that higher CSR ratings

are associated with more requests for new 10-K ﬁlings on EDGAR. This test

further alleviates the concern that CSR disclosures drive our main results.

To save space, results of these alternative explanation tests are not reported but are

available upon request.

38 G. Chen et al./Accounting & Finance

6. Conclusion

We examine the role of CSR on the ﬂow of information into stock prices. The

ethical and reputational view of CSR hypothesises that ﬁrms participate in

CSR activities to fulﬁl their expected social responsibilities and to establish a

reputation as good corporate citizens. Therefore, a commitment to CSR can

incentivise information acquisition by indicating lower information acquisition

costs and greater protection for outside investors. However, the agency view of

CSR argues that CSR activities are a manifestation of agency problems in the

ﬁrm. Self-serving managers may engage in CSR activities to derive private

beneﬁts at the expense of shareholders and may manipulate information to

disguise their opportunism, which discourages investor information acquisition

and leads to less informed stock prices. We empirically examine the association

between CSR and stock price informativeness as measured by idiosyncratic

volatility and PIN. Our results consistently show a positive association between

CSR and ﬁrm-speciﬁc information contained in stock prices.

Using data on investors’ search activity on EDGAR for new 10-K ﬁlings and

on Google for company news around earnings announcements, our ﬁndings

suggest that investors collect more information from ﬁrms that are more

socially responsible around these major disclosure events. These results provide

direct evidence that CSR engagement encourages more information search

activity.

We conduct a number of robustness checks using alternative empirical

speciﬁcations. The lead-lag analyses suggest a positive relation between lagged

CSR (rather than lead or contemporaneous) and price informativeness. Results

from a propensity score matched sample are consistent with our main ﬁndings

and suggest that CSR is not an aggregate proxy of other ﬁrm characteristics.

Furthermore, results using FERC as an alternative measure of price informa-

tiveness provide further support for our main ﬁndings. In addition, we show

that our results are unlikely to be driven by several alternative explanations.

While these robustness checks help alleviate the potential problem of

endogeneity, to the extent that these procedures are not exhaustive, one should

be cautious in making casual inferences from the ﬁndings in this paper.

Our paper adds to the growing literature on price discovery and CSR. We

examine whether and how CSR is associated with the ﬂow of information to

stock prices –a fundamental question in ﬁnancial economics. To the best of our

knowledge, no study has investigated this topic in-depth. Overall, our results

are consistent with the ethical and reputational view that ﬁrms’ active

engagement in CSR facilitates information ﬂow into stock prices.

Our study further has some potential policy and research implications. Our

results suggest that a commitment to CSR beneﬁts ﬁrms with more informative

stock prices, which may provide ﬁrms more accurate signals to eﬃciently

allocate their capital. The existing literature documents mixed evidence on the

relation between CSR and investment eﬃciency (Bhandari and Javakhadze,

G. Chen et al./Accounting & Finance 39

2017; Benlemlih and Bitar, 2018). It can be promising for future research to

examine the relation between CSR and corporate capital allocation decisions,

particularly how stock price informativeness plays a role in this relation. Our

study further informs policy makers. While various policy tools are used to

promote capital market eﬃciency, our study implies that CSR activities can

help ﬁrms to self-regulate their information and trading environment and

enhance market eﬃciency.

References

Altschuler, D., G. Chen, and J. Zhou, 2015, Anticipation of management forecasts and

analysts’ private information search, Review of Accounting Studies 20, 803–838.

Barnard, J. W., 1997, Corporate philanthropy, executives’ pet charities and the agency

problem, New York Law School Review 41, 1147–1178.

Barnea, A., and A. Rubin, 2010, Corporate social responsibility as a conﬂict between

shareholders, Journal of Business Ethics 97, 71–86.

Benabou, R., and J. Tirole, 2010, Individual and corporate social responsibility,

Economica 77, 1–19.

Benlemlih, M., and M. Bitar, 2018, Corporate social responsibility and investment

eﬃciency, Journal of Business Ethics 148, 647–671.

Benoit, D., 2019, Move over, shareholders: Top CEOs say companies have obligations to

society, Wall Street Journal (19 August).

Bhandari, A., and D. Javakhadze, 2017, Corporate social responsibility and capital

allocation eﬃciency, Journal of Corporate Finance 43, 354–377.

Bond, P., A. Edmans, and I. Goldstein, 2012, The real eﬀects of ﬁnancial markets,

Annual Review of Financial Economics 4, 339–360.

Brown, T. J., and P. A. Dacin, 1997, The company and the product: Corporate

associations and consumer product responses, Journal of Marketing 6, 68–84.

Brown, S., and S. A. Hillegeist, 2007, How disclosure quality aﬀects the level of

information asymmetry, Review of Accounting Studies 12, 443–477.

Brown, W. O., E. Helland, and J. K. Smith, 2006, Corporate philanthropic practices,

Journal of Corporate Finance 12, 855–877.

Carroll, A. B., 1979, A three-dimensional conceptual model of corporate performance,

The Academy of Management Review 4, 497–505.

Cespa, G., and G. Cestone, 2007, Corporate social responsibility and managerial

entrenchment, Journal of Economic and Management Strategy 16, 741–771.

Chatterji, A. K., D. I. Levine, and M. W. Toﬀel, 2009, How well do social ratings

actually measure corporate social responsibility?, Journal of Economics and Manage-

ment Strategy 18, 125–169.

Chen, Q., I. Goldstein, and W. Jiang, 2007, Price informativeness and investment

sensitivity to stock price, Review of Financial Studies 20, 619–650.

Chih, H. L., C. H. Shen, and F. C. Kang, 2008, Corporate social responsibility, investor

protection, and earnings management: Some international evidence, Journal of

Business Ethics 79, 179–198.

Choi, J. H., L. Myers, Y. Zang, and D. Ziebart, 2011, Do management EPS forecasts

allow returns to reﬂect future earnings? Implications for the continuation of

management’s quarterly earnings guidance, Review of Accounting Studies 16, 143–182.

Collins, D., S. Kothari, J. Shanken, and R. Sloan, 1994, Lack of timeliness versus noise

as explanations for low contemporaneous return-earnings association, Journal of

Accounting and Economics 18, 289–324.

40 G. Chen et al./Accounting & Finance

Da, Z., J. Engelberg, and P. Gao, 2011, In search of attention, Journal of Finance 66,

1461–1499.

Dechow, P. M., A. P. Hutton, L. Meulbroek, and R. G. Sloan, 2001, Short-sellers,

fundamental analysis, and stock returns, Journal of Financial Economics 61, 77–106.

Dechow, P., R. Sloan, and A. Sweeney, 1995, Detecting earnings management, The

Accounting Review 70, 193–225.

Deng, X., J. Kang, and B. S. Low, 2013, Corporate social responsibility and stakeholder

value maximization: evidence from mergers, Journal of Financial Economics 110,

87–109.

Desai, M., and D. Dharmapala, 2009, Earnings management, corporate tax shelters,

and book-tax alignment, National Tax Journal 62, 169–186.

Dhaliwal, D. S., O. Z. Li, A. Tsang, and Y. G. Yang, 2011, Voluntary nonﬁnancial

disclosure and the cost of equity capital: The initiation of corporate social

responsibility reporting, The Accounting Review 86, 59–100.

Diamond, D. W., 1989, Reputation acquisition in debt markets, Journal of Political

Economy 97, 828–862.

Diamond, D. W., and R. E. Verrecchia, 1981, Information aggregation in a noisy

rational expectations economy, Journal of Financial Economics 9, 221–235.

Diamond, D., and R. E. Verrecchia, 1987, Constraints on short-selling and asset price

adjustment to private information, Journal of Financial Economics 18, 277–311.

Dong, Y., O. Z. Li, Y. Lin, and C. Ni, 2016, Does information-processing cost aﬀect

ﬁrm-speciﬁc information acquisition? Evidence from XBRL adoption, Journal of

Financial and Quantitative Analysis 51, 435–462.

Drake, M. S., D. T. Roulstone, and J. R. Thornock, 2012a, Investor information

demand: evidence from Google searches around earnings announcements, Journal of

Accounting Research 50, 1001–1040.

Drake, M. S., D. T. Roulstone, and J. R. Thornock. 2015, The determinants and

consequences of information acquisition via EDGAR, Contemporary Accounting

Research 32, 1128–1161.

Durnev, A., R. Morck, B. Yeung, and P. Zarowin, 2003, Does greater ﬁrm-speciﬁc

return variation mean more or less informed stock pricing?, Journal of Accounting

Research 41, 797–836.

Durnev, A., R. Morck, and B. Yeung, 2004, Value-enhancing capital budgeting and

ﬁrm-speciﬁc stock return variation, Journal of Finance 59, 65–105.

Easley, D., M. K. Nicholas, and M. O’Hara, 1997a, The information content of the

trading process, Journal of Empirical Finance 4, 159–186.

Easley, D., M. K. Nicholas, and M. O’Hara, 1997b, One day in the life of a very

common stock, Review of Financial Studies 10, 805–835.

Easley, D., S. Hvidkjaer, and M. O’Hara, 2002, Is information risk a determinant of

asset returns?, Journal of Finance 57, 2185–2221.

Economist, 2005, The importance of corporate responsibility. (Economist Intelligence

Unit, London, UK).

Ensminger, J., 2001, Reputation, trust, and the principal agent problem, in: K. S. Cook,

ed., Trust and Society (Russell Sage Foundation, New York, NY), 185–201.

Fama, E. F., 1970, Eﬃcient capital markets: a review of theory and empirical work,

Journal of Finance 25, 28–30.

Fernandes, N., and M. Ferreira, 2009, Insider trading laws and stock price informa-

tiveness, Review of Financial Studies 22, 1845–1887.

Ferreira, M., and P. Laux, 2007, Corporate governance, idiosyncratic risk, and

information ﬂow, Journal of Finance 62, 951–989.

Ferreira, D., M. Ferreira, and C. Raposo, 2011, Board structure and price informa-

tiveness, Journal of Financial Economics 99, 523–545.

G. Chen et al./Accounting & Finance 41

Ferrell, A., H. Liang, and L. Renneboog, 2016, Socially responsible ﬁrms, Journal of

Financial Economics 122, 585–606.

Fishman, M. J., and K. M. Hagerty, 1992, Insider trading and the eﬃciency of stock

prices, The RAND Journal of Economics 23, 106–122.

Flammer, C., 2015, Does corporate social responsibility lead to superior ﬁnancial

performance?, A regression discontinuity approach, Management Science 61,

2549–2568.

Fombrun, C., and M. Shanley, 1990, What’s in a name? Reputation building and

corporate strategy, Academy of Management Journal 33, 233–258.

French, K., and R. Roll, 1986, Stock return variances: the arrival of information and the

reaction of traders, Journal of Financial Economics 17, 5–26.

Friedman, M. 1970, The social responsibility of business is to increase its proﬁts, New

York Times Magazine (13 September), 122–126.

Gao, F., L. L. Lisic, and I. Zhang, 2014, Commitment to social good and insider

trading, Journal of Accounting and Economics 57, 149–175.

Gelb, D. S., and J. A. Strawser, 2001, Corporate social responsibility and ﬁnancial

disclosures: An alternative explanation for increased disclosure, Journal of Business

Ethics 33, 1–13.

Gelb, D., and P. Zarowin, 2002, Corporate disclosure policy and the informativeness of

stock prices, Review of Accounting Studies 7, 33–52.

Ghoul, S. E., O. Guedhami, C. C. Y. Kwok, and D. R. Mishra, 2011, Does corporate

social responsibility aﬀect the cost of capital?, Journal of Banking and Finance 35,

2388–2406.

Goss, A., and G. S. Roberts, 2011, The impact of corporate social responsibility on the

cost of bank loans, Journal of Banking and Finance 35, 1794–1810.

Greening, D. W., and D. B. Turban, 2000, Corporate social performance as a

competitive advantage in attracting a quality workforce, Business and Society 39,

254–280.

Gregory, A., J. Whittaker, and X. Yan, 2016, Corporate social performance,

competitive advantage, earnings persistence and ﬁrm value, Journal of Business

Finance and Accounting 43, 3–30.

Grewal, J., C. Hauptmann, and G. Serafeim, 2017, Material sustainability and stock

price informativeness, Journal of Business Ethics. https://doi.org/10.1007/s10551-020-

04451-2

Grossman, S., and J. Stiglitz, 1980, On the impossibility of informationally eﬃcient

markets, American Economic Review 70, 393–408.

Gul, F. A., B. Srinidhi, and A. C. Ng, 2011, Does board gender diversity improve the

informativeness of stock prices?, Journal of Accounting and Economics 51, 314–338.

Harjoto, H., and H. Jo, 2011, Corporate governance and CSR nexus, Journal of Business

Ethics 100, 45–67.

Harvard Business School (HBS), 2011, Conference on ‘‘Research in Corporate

Accountability Reporting’. Available at: http://www.hbs.edu/faculty/conferences/

2013-corporate-accountability-reporting/Pages/default.aspx

Hemingway, C. A., and P. W. Maclagan, 2004, Managers’ personal values as drivers of

corporate social responsibility, Journal of Business Ethic 50, 33–44.

Hirshleifer, D., S. Lim, and S. H. Teoh, 2009, Driven to distraction: extraneous events

and underreaction to earnings news, Journal of Finance 64, 2289–2325.

Hirshleifer, D., and S. H. Teoh, 2003, Limited attention, information disclosure, and

ﬁnancial reporting, Journal of Accounting and Economics 36, 337–386.

Hong, H., J. D. Kubik, and J. A. Scheinkman, 2012, Financial constraints on corporate

goodness, NBER Working paper No. 18476. Available at: http://www.nber.org/pape

rs/w18476.pdf

42 G. Chen et al./Accounting & Finance

Jegadeesh, N., and S. Titman, 2001, Proﬁtability of momentum strategies: an evaluation

of alternative explanations, Journal of Finance 56, 699–720.

Jensen, M. C., 1986, Agency costs of free cash ﬂow, corporate ﬁnance, and takeovers,

The American Economic Review 76, 323–329.

Jensen, M. C., 2001, Value maximization, stakeholder theory, and the corporate

objective function, Journal of Applied Corporate Finance 14, 8–21.

Jin, L., and S. C. Myers, 2006, R

around the world: New theory and new tests, Journal

of Financial Economics 79, 257–292.

Jones, J. J., 1991, Earnings management during import relief investigations, Journal of

Accounting Research 29, 193–228.

Jones, T., 1995, Instrumental stakeholder theory: A synthesis of ethics and economics,

The Academy of Management Review 20, 404–437.

Jones, C. M., and O. A. Lamont, 2002, Short-sale constraints and stock returns, Journal

of Financial Economics 66, 207–239.

Jones, S., and C. Wright, 2016, Fashion or future: does creating shared value pay?,

Accounting and Finance 58, 1111–1139.

Kim, O., and R. E. Verrecchia, 1997, Pre-announcement and event-period private

information, Journal of Accounting and Economics 24, 395–419.

Kim, Y., M. S. Park, and B. Wier, 2012, Is earnings quality associated with corporate

social responsibility?, The Accounting Review 87, 761–796.

Kim, Y., H. Li, and S. Li, 2014, Corporate social responsibility and stock price crash

risk, Journal of Banking and Finance 43, 1–13.

Kothari, S. P., and R. G. Sloan, 1992, Information in prices about future earnings:

Implications for earnings response coeﬃcients, Journal of Accounting and Economics

15, 143–171.

Kruger, P., 2015, Corporate goodness and shareholder wealth, Journal of Financial

Economics 115, 304–329.

Lee, C. M. C., P. Ma, and C. C. Y. Wang, 2015, Search based peer ﬁrms: Aggregating

investor perceptions through internet co-searches, Journal of Financial Economics 116,

410–431.

Levitt, T., 1958, The Dangers of Social Responsibility. (Harvard Business Review),

41–50.

Li, E. X., and K. Ramesh, 2009, Market reaction surrounding the ﬁling of periodic SEC

reports, The Accounting Review 84, 1171–1208.

Li, B., S. Rajgopal, and M. Venkatachalam, 2014, R2 and idiosyncratic risk are not

interchangeable, The Accounting Review 89, 2261–2295.

Lins, K. V., H. Servaes, and A. N. E. Tamayo, 2017, Social capital, trust, and ﬁrm

performance: The value of corporate social responsibility during the ﬁnancial crisis,

Journal of Finance 72, 1785–1824.

Loughran, T., and B. McDonald, 2015, Information decay and ﬁnancial disclosures,

Working paper (University of Notre Dame).

Lys, T., J. P. Naughton, and C. Wang, 2015, Signaling through corporate accountability

reporting, Journal of Accounting and Economics 60, 56–72.

Margolis, J. D., H. A. Elfenbein, and J. P. Walsh, 2007, Does it pay to be good? A meta-

analysis and redirection of research on the relationship between corporate social and

ﬁnancial performance, Working paper (Harvard Business School).

McNichols, M., and B. Trueman, 1994, Public disclosure, private information

collection, and short-term trading, Journal of Accounting and Economics 17, 69–94.

McWilliams, A., and D. Siegel, 2001, Corporate social responsibility: a theory of the

ﬁrm perspective, The Academy of Management Review 26, 117–127.

G. Chen et al./Accounting & Finance 43

Morck, R., B. Yeung, and W. Yu, 2000, The information content of stock markets: Why

do emerging markets have synchronous stock price movements?, Journal of Financial

Economics 58, 215–260.

Murphy, K. J., 1999, Executive compensation, in: O. C. Ashenfelter, D. Card, eds.,

Handbook of Labor Economics, vol. 3B (Elsevier, Amsterdam), 2485–2563.

Noreen, E., 1988, The economics of ethics: a new perspective on agency theory,

Accounting, Organizations and Society 13, 359–369.

Petrovits, C. M., 2006, Corporate-sponsored foundations and earnings management,

Journal of Accounting and Economics 41, 335–362.

Piotroski, J. D., and D. T. Roulstone, 2004, The inﬂuence of analysts, institutional

investors, and insiders on the incorporation of market, industry, and ﬁrm-speciﬁc

information into stock prices, The Accounting Review 79, 1119–1151.

Porter, M. E., and C. Linde, 1995, Green and competitive: Ending the stalemate,

Harvard Business Review 73, 120–134.

Prior, D., J. Surroca, and J. A. Tribo, 2008, Are socially responsible managers really

ethical? Exploring the relationship between earnings management and corporate

social responsibility, Corporate Governance: An International Review 16, 160–177.

Ramalingegowda, S., and Y. Yu, 2012, Institutional ownership and conservatism,

Journal of Accounting and Economics 53, 98–114.

Roll, R. 1988, R2, The Journal of Finance 43, 541–566.

Saﬃ, P. A. C., and K. Sigurdsson, 2011, Price eﬃciency and short selling, Review of

Financial Studies 24, 821–852.

Servaes, H., and A. Tamayo, 2013, The impact of corporate social responsibility on ﬁrm

value: The role of customer awareness, Management Science 59, 1045–1061.

Soltes, E., 2009, News dissemination and the impact of the business press, Working paper

(Harvard Business School).

Tran, N., and D. O’Sullivan, 2020, The relationship between corporate social

responsibility, ﬁnancial misstatements and SEC enforcement actions, Accounting

and Finance 60, 1111–1147.

Tucker, J. W., and P. A. Zarowin, 2006, Does income smoothing improve earnings

informativeness?, The Accounting Review 81, 251–270.

Zuo, L., 2016, The informational feedback eﬀect of stock prices on corporate disclosure,

Journal of Accounting and Economics 61, 391–413.

44 G. Chen et al./Accounting & Finance

Appendix

Variable deﬁnitions

Variables Deﬁnition

Price informativeness

Idiosyncratic

volatility

ΨAnnual logistic transformed relative

volatility estimated from an expanded

market model including the Fama-

French 48 industry returns

Probability of

information-based

trading

PIN Annual probability of information-

based trading as in Easley et al. (2002)

Future earnings

response coeﬃcient

FERC The coeﬃcient of future earnings

captured in stock prices as in Choi et al.

(2011)

Corporate social responsibility (CSR)

Corporate social

responsibility

CSR The sum of yearly adjusted community,

diversity, employee relations,

environment, human rights, and

product quality and safety KLD CSR

scores. Adjusted CSR is estimated by

scaling the raw strength and concern

scores of each category by the number

of items of the strengths and concerns

of that category in the year and then

taking the net diﬀerence between the

strength and concern scores for that

Corporate social responsibility and information flow

Abstract and Figures

Recommended publications

The effect of profitability and liquidity on CSR disclosure and its implication to economic conseque...

Corporate Social Responsibility and Financial Fraud: The Moderating Effects of Governance and Religi...

Is Institutional Ownership Related to Corporate Social Responsibility? The Non-Linear Relation and i...