Received: 22 October 2017 Revised: 13 December 2017 Accepted: 12 January 2018

DOI: 10.1002/bse.2047

RESEARCH ARTICLE

Environmental, social and governance transparency and firm

value
Ellen Pei‐yi Yu1 | Christine Qian Guo1 | Bac Van Luu2

1
Birkbeck College, Department of
Management, University of London, UK Abstract
2
Russell Investments, London, UK We investigate whether environmental, social and governance (ESG) transparency, the extent of
Correspondence ESG disclosure, has an impact on firm value. Reducing investors’ information symmetry and
Ellen Pei‐yi Yu, Birkbeck College, Department agency costs is the mechanism by which better ESG transparency potentially impacts firm value.
of Management, University of London, Malet Using Bloomberg ESG disclosure scores to assess a firm's ESG transparency, we look at a sample
Street, Bloomsbury, London WC1E 7HX, UK.
of 1996 large cap companies across 47 developed and emerging countries and territories. Our
Email: [email protected]; peiyi_yu@ymail.
com empirical analyses suggest that the benefits from ESG disclosure outweigh their costs for the
average listed firm. We find supporting evidence for greater disclosure of ESG issues boosting
firm valuation measures, such as Tobin's Q. Furthermore, our results suggest that firms with
greater asset size, better liquidity, higher R&D intensity, fewer insider holdings and good past
financial performance will be more transparent in ESG issues.

KEY W ORDS

corporate governance, environmental policy, ESG disclosure, stakeholder engagement, sustainable

development

1 | I N T RO D U CT I O N & Phillips, 2001; King & Lenox, 2000; Margolis & Walsh, 2003; Ruf,
Muralidhar, & Paul, 1998). Some scholars study whether ESG criteria
There has been rapidly growing interest in environmental, social and can be viewed as a potential key factor for investment success
governance (ESG) issues from individual shareholders, institutional (Richardson, 2009), and whether shareholders prefer to invest in firms
investors, governments, local communities, employees and suppliers with a better corporate social response (CSR) image, which may result
over the past 10 years (Escrig‐Olmedo, Muñoz‐Torres, & Fernández‐ in better financial performances (De Bakker, Groenewegen, & Den
Izquierdo, 2013; Hill, Ainscough, Shank, & Manullang, 2007). The Hond, 2005; Margolis & Walsh, 2003). However, there is little research
Governance and Accountability Institute (2017) shows that 82% of that focuses on a firm's ESG transparency and the quantity of ESG
S&P 500 companies had embraced sustainability reporting in 2017, disclosure. To fill this gap, our analysis is based on data related to the
while this was the case for only 53% of S&P 500 companies in extent of ESG disclosure, rather than firms’ actual performance on
2012. An interesting example is provided by how investors are ahead ESG issues. We first examine how listed‐firms’ ESG transparency can
of ESG issues. FTSE Russell ruled out the addition of zero voting impact their performance. We then perform supplementary analyses
rights stocks because of concerns raised by shareholders. to model the determinants of ESG transparency.
Consequently, the investment management industry is starting to In this study, we use the Bloomberg ESG disclosure score to assess
accommodate ESG issues. Meanwhile, to respond to growing firms’ transparency. Bloomberg compiles ESG data on publicly listed
stakeholder interest in ESG data, rating agencies (i.e., MSCI, companies globally from published disclosure and news items, and
Thomson Reuters Asset4 rating agency), financial information pro- turns it into a single number. More precisely, the Bloomberg ESG dis-
viders (i.e., Bloomberg) and firms also report ESG data, which stands closure scores measure the amount of ESG disclosure data a company
for environmental (total greenhouse gas emissions, hazardous waste, reports publicly, but does not measure the company's actual ESG per-
environmental fines, etc.), social (the percentage of employee turn- formance. Zero to 100 is the range of the Bloomberg ESG disclosure
over, community spending, etc.) and governance data (board duration, score, which shows that the higher the disclosure score, the more
political donations, etc.). information is disclosed.
The ESG literature has focused on measures of corporate ESG per- We investigate ESG transparency over time and across countries.
formance and their link to financial performance (Eccles, Herz, Keegan, The sample components of this study are selected from the MSCI All

Bus Strat Env. 2018;27:987–1004. wileyonlinelibrary.com/journal/bse Copyright © 2018 John Wiley & Sons, Ltd and ERP Environment 987
988 YU ET AL.

Country World Index (ACWI). We show empirically that ESG transpar- economy by setting requirements for mandatory emission measure-
ency significantly influences firms’ Tobin's Q, and confirm that there is ment and reporting. Overall, the level of mandatory ESG disclosure is
a nonlinear relationship between ESG transparency and Tobin's Q. Our growing, which may be interpreted as a sign of the increasing impor-
empirical results can be interpreted as supporting evidence for pro- tance of ESG transparency.
moting ESG transparency. The implications of this study are significant.
We provide the following recommendations related to ESG transpar- 2.1.2 | ESG voluntary disclosure
ency. Firms are encouraged to report ESG data together with the Mandatory ESG disclosure dominates, but the growth in voluntary dis-
financial information that they are required to report to shareholders. closure is also strong. Companies in the European Union (EU) are
We recommend investors pay more attention to ESG transparency adapting to the EU Non‐Financial Reporting Directive issued in 2014,
along with traditional financial statements, and to support firms to although the directive does not specify standards that firms should fol-
increase the quantity of ESG disclosure. low in disclosing relevant information, such as environmental matters,
The remainder of our paper is organized as follows. In Section 2, human rights or board diversity. However, Germany is an exception.
we present the current development of ESG disclosure and the The German Sustainability Code, which was issued in 2011 as a volun-
responses to ESG transparency by relevant stakeholder parties. In Sec- tary guidance and can be furnished using a template, features 20 indi-
tion 3, we describe our data and research design. We present the cators of sustainability performance aligned with the Global Reporting
empirical results in Section 4. Finally, we summarize and conclude in Initiative (GRI) Guidelines, the United Nations Global Compact (UNGC)
Section 5. principles and the OECD guidelines for multinational companies. In
2011, the Institute of Company Secretaries of India proposed a guid-
ance note on nonfinancial disclosure to help firms to voluntarily make
2 | R E S P O N S E S T O T RA N S P A R E N C Y I N E S G appropriate disclosures beyond the narrow focus of financial informa-
ISSUES tion disclosure.

In response to growing pressure for ESG transparency and corporate

accountability, firms have started to report ESG data as a nonfinancial 2.2 | Global responses to ESG disclosure
report, in addition to traditional financial reports. However, the con- The quality and quantity of ESG disclosure data have increased dra-
tent of these nonfinancial reports varies widely due to the lack of reg- matically in the last two decades. However, the ESG data still lack com-
ulatory guidelines on how to report this information. In this section, we parability across firms and countries. Here, we discuss how different
outline the challenges of ESG transparency and discuss how relevant market participants have been influencing the latest development of
stakeholder parties influence this issue. ESG transparency.

2.2.1 | Stock exchanges and market regulators

2.1 | Current development of mandatory and
Stock exchanges can create listing guidelines around ESG disclosures,
voluntary ESG disclosure
while securities regulators can promote an improvement of the avail-
Previous studies (Bennear & Olmstead, 2008; Jin & Leslie, 2003) doc- ability of ESG data. For instance, the Ministry of Environmental Protec-
ument that mandatory disclosure regulations can improve operating tion in China, together with the China Securities Regulatory
performance with regard to water safety and the environment. KPMG Commission, launched the Green Security Law in 2008. The Green
(2016) identify around two‐thirds of sustainability reporting instru- Securities policy requires firms listed on the stock exchange in China
ments are mandatory and about one‐third voluntary. to disclose more information about their environmental record. Accord-
ing to The Company Act issued in 2006, quoted companies in the UK
2.1.1 | ESG mandatory disclosure should also disclose information in their annual review on environmen-
Government regulation is considered the most important of sustain- tal, employee, social and community matters, to provide an understand-
ability reporting instruments. In Organisation for Economic Co‐ ing of the performance and development of the company. By following
operation and Development (OECD) countries, new sustainability United Nations Sustainable Stock Exchange initiative (2017), stock
reporting requirements are introduced through accounting regulations exchanges can self‐regulate regarding ESG disclosure. There are 66
and company acts that address reporting with a particular focus on cer- partner exchanges with the UN Sustainable Stock Exchange in 2017.
tain areas, such as environmental pollutants and corporate governance. The UN Sustainable Stock Exchange initiative encourages exchanges
In many countries, increasing mandatory ESG disclosure requirements to make a voluntary commitment to promote improved ESG disclosure
are introduced through government regulations. For example, based and actual ESG performance among listed companies.
on the “Quoted companies GHG reporting” issued in 2013, the UK
requires corporations listed on the London Stock Exchange to report 2.2.2 | Policy‐makers and corporate reporting
their levels of greenhouse gas (GHG) emissions. Firms in China are organizations
required to disclose environmental information, according to the In their “Principles for Responsible Investments”, the United Nations
Environmental Information Disclosure Act issued by the Chinese (2006) encourage relevant stakeholder parties to incorporate ESG issues
Government in 2008. Mexico passed the Climate Change law in into their investment analyses. The United Nations Principles for
2012, which addresses climate change and the transition to a green Responsible Investment has about 1813 signatories in October 2017,
YU ET AL. 989

which cover asset owners, investment managers and service providers. conducted on ESG transparency and its impact on firm value. This sec-
Corporate reporting organizations that are independent and nonprofit tion describes the research design and hypotheses used in our investi-
also have an impact on ESG disclosure. For instance, the Sustainability gation along with the relevant literature. Our first task is to establish
Accounting Standards Board (SASB) develops and propagates sustain- whether a firm's transparency in ESG issues can influence its value.
ability accounting standards. The GRI is a similar independent organiza- We assume that all firms are concerned with maximizing firm value.
tion that helps relevant stakeholder parties to understand their impact Furthermore, we seek to model and examine the determinants of
on issues such as climate change, corruption and human rights. The SASB ESG disclosure.
and GRI Guidelines are often adopted for sustainability reports. Overall,
those corporate reporting organizations work toward making disclosure 3.1 | Firm's ESG transparency and firm value
comparable and decision‐useful for investors.
Will companies’ choice of ESG disclosure level influence firm value? To
answer this research question, we start with a brief review of the
2.2.3 | Independent sources
studies that focus on disclosure benefits and costs.
Several independent sources supply ESG data. Avetisyan and Hockerts
Eccles et al. (2014) find that high sustainability firms, which are
(2017) document how ESG rating agencies, such as EIRIS Foundation,
more long‐term oriented, have superior ESG measurement and better
Morgan Stanley Capital International, Vigeo and Sustainable Asset
disclosure practices. Other researchers (Galbreath, 2013; Margolis &
Management, provide data on the social performance of firms. Data
Walsh, 2003) document that firms with better ESG transparency are
on the impact of environmental performance of listed US firms is com-
more likely to obtain capital at a lower cost because of a better oper-
piled by investor research firms, such as KLD Research & Analytics,
ational reputation, resulting in lower reputational risk. Cheng, Ioannou,
Ceres, Trucost, the Standard and Poor's Corporation‐Newsweek.
and Serafeim (2014) document that firms with better ESG actual per-
Moreover, corporateregister.com provides a Reputation Score based
formance‐related scores can benefit from lower capital constraints.
on an opinion survey of CSR professionals, academics and other envi-
Serafeim and Grewal (2017) suggest that nonfinancial information
ronmental experts. Overall, there has been a trend toward specialized
can be used to predict expected future financial performance of the
ESG rating agencies (Delmas & Blass, 2010; Koellner, Weber, Fenchel,
firm. Bank of America Merrill Lynch (2017) find that ESG‐based
& Scholz, 2005; Lai, Melloni, & Stacchezzini, 2016).
investing would have helped investors avoid 90% of bankruptcies in
the time frame they examined. Conversely, some scholars argue that
2.2.4 | Capital market stakeholders
there is a significant cost associated with the levels of ESG disclosure.
Norges Bank Investment Management, who manage the world's larg- For instance, Aggarwal and Dow (2012) suggest that a firm's physical
est sovereign wealth fund, can be viewed as an example of how insti- assets can be treated as the direct costs associated with regulatory
tutional investors respond to ESG issues. The Norwegian fund set compliance to reduce GHG emissions. Hainmueller and Hiscox (2012)
investment criteria focusing on three areas: climate change, water document that customers are less willing to pay a premium for green
and children's rights. By doing so, investors can place pressure on their products and services in price‐sensitive market segments. Mattoo,
target investing firms—and incentivize the managers to improve their Subramanian, van der Mensbrugghe, and He (2009) suggest that inter-
ESG disclosure and actual ESG performance. national trade makes it possible to seek and take advantage of less
expensive climate change regulation regimes.
2.2.5 | Individual firms The relevant literature review leads to the development of our
Slager, Gond, and Moon (2012) document that more companies are research design. By modeling the following four estimating equations,
beginning to value their ESG ratings and communicate with interested we assess how value‐maximizing firms shape their responses in ESG
parties about ESG issues both internally and externally. Firms are transparency issues. We predict that rational managers aim to balance
becoming more sophisticated in their communications with the public the disclosure benefits and the disclosure costs by finding the optimal
regarding ESG issues (Hockerts & Moir, 2004; Vandekerckhove, Leys, ESG disclosure level. Our model below explains why an increase of
& Van Braeckel, 2008). Eccles, Ioannou, and Serafeim (2014) suggest ESG disclosure degree beyond a certain level may deteriorate firm per-
that because high sustainability firms are more long‐term oriented, formance rather than enhance it.
they are more likely to attract long‐term investors. There has also been We start with Equation (1), which posits that the firm's perfor-
a trend that firms report information to stakeholders beyond just mance TPt is a positive function of x, measured here at the disclosure
shareholders. Corporation operations affect not only communities level of ESG. We assume ESG disclosure level can enhance firm's total
but also the natural environments in which they operate. For instance, performance.
Shell was reported to be responsible for over 20 pollution accidents in
British waters in 2013. TPT ¼ α þ αxr : (1)

By assuming α > 0 and r > 0, a positive impact of the ESG disclo-

3 | RESEARCH DESIGN sure level is posited on the firm's performance through the magnitude
and the slope.
Having surveyed the state of ESG disclosure, we now move on to iden- Meanwhile, there is also a significant cost associated with the
tifying a gap in the literature that our paper is aiming to fill. While there levels of ESG disclosure (see Equation 2). This negative relationship is
are some studies on ESG performance, very little research has been written as:
990 YU ET AL.

TPEF ¼ −bxt ; (2) to assess a publicly listed company's transparency. The higher the dis-
closure score, the more nonfinancial information is disclosed.
where b > 0 and t > 0.
We are also interested in examining whether a firm's environ-
The overall impact of ESG disclosure on the firm's performance is
mental disclosure level has an impact on its Tobin's Q. Using a similar
the sum of these two equations, where axr represents the positive
methodology, the Bloomberg environmental disclosure score is
impact proposed by Equation (1) and −bxt represents the negative
compiled based on the extent of a company's environmental disclo-
impact proposed by Equation (2):
sure. The range of the Bloomberg environmental disclosure score is
between 0.1 and 100. Each data point is weighted according to its
TPðxÞ ¼ α þ TPT ðxÞ−TPEF ðxÞ ¼ α þ αxr −bxt (3)
importance.
Based on Equation (3), the values of a, b, r and t determine the
shape of the function of the firm's performance. A linear relationship 3.1.2 | R&D intensity – indicator of the agency and moni-
will only exist if r = t = 1. However, if r < t, an inverse U‐shaped toring costs
relationship will be formed. Otherwise, if r > t, the function of company If a firm is more transparent than its peers, its shareholders and
performance will be U‐shaped. The three equations discussed above stakeholders may have a greater ability to monitor the managerial
allow for potential nonlinearities in the relationship between ESG team. Smith and Watts (1992) state that agency costs and moral
disclosure and firm performance. hazard problems are likely to occur in firms with high growth
Therefore, we include a linear term for ESG disclosure, and the opportunities. Cheng et al. (2014) document that to increase ESG
quadratic term “ESG disclosure2” in Equation (4). Following previous disclosure can reduce information symmetry and agency costs by
studies (Aggarwal & Dow, 2012; Bebchuk, Cohen, & Ferrell, 2009; enhancing stakeholder engagement. Miller and Reisel (2012) and
Lang & Stulz, 1994; Lee, Lev, & Yeo, 2008; Shleifer & Vishny, 1997), Zhu and Cai (2014) also suggest that legal protection and accounting
we measure the firm's long‐term value by Tobin's Q. Equation (4) is disclosure requirements are likely to decrease information asymmetry
given as: between the principal and the agent. By following previous studies
(Himmelberg, Hubbard, & Palia, 1999; Lee et al., 2008), we measure
ðIndustry−adjusted Tobin QÞ ¼ a0 þ a1* ðadjusted ESG disclosureÞ (4)
the agency and monitoring costs by using the variable of R&D inten-
þa2* ðadjusted ESG disclosureÞ2 þ a3* logðfirm sizeÞ
sity. We include R&D intensity in Equation (4). We assume that R&D
þa4* ðadjusted leverage ratioÞ þ a5* ðliquidity ratioÞ
intensity is likely to be positively associated with the agency cost
þa6* ðGDP per capita based on PPPÞ þ a7* ðR&D intensityÞ related to managerial monitoring for firms that are difficult to
þa8* ðpercentage of independent directorsÞ monitor. Based on agency proxies, we can examine whether more
*
þa9 ðinstitutional ownershipÞ þ εðresidualÞ transparency in ESG issues can reduce agency costs associated with
moral hazard problems and information asymmetry between princi-
*where we include the key variables (ESG), (ESG)2, and control
pals and agents.
variables that have been shown to have an association with Tobin's
Finally, our discussion in 3.1 leads to Hypotheses 1(a) and 1(b):
Q. The definitions for all variables in Equation (4) are shown in
Table 1. Hypothesis 1 (a): We assume that the association
Table 1 presents the definition and our estimation methods for all between a firm's performance and its ESG disclosure is
variables in this study. conditional on agency costs and governance structures. A
publicly listed company's transparency in ESG issues can
3.1.1 |ESG disclosure and environmental disclosure – impact its Tobin's Q. We predict that the relationship
indicator of firm transparency between firm performance and ESG disclosure is not linear.
In this study, we focus on firm transparency rather than the firm's
Hypothesis 1 (b): We assume that the association
actual performance in ESG issues. We identify the Bloomberg ESG dis-
between a firm's performance and its environmental dis-
closure score as an appropriate indicator to measure firms’ transpar-
closure is conditional on agency costs and governance
ency. The Bloomberg ESG disclosure score is designed to measure
structures. A publicly listed company's transparency in
the amount of ESG data that firms report publicly, and does not
environmental issues can impact its Tobin's Q. We predict
measure the firm's performance. The score is realized based on the
that the relationship between firm performance and envi-
extent of a company's ESG disclosure. The score starts at 0.1 for firms
ronmental disclosure is not linear.
that disclose a minimum amount of ESG data to 100 for those that dis-
close every data point collected by Bloomberg. Firms that do not dis- We report our empirical results of Hypotheses 1(a) and 1(b) in
close anything are shown as N/A. Each data point is weighted Section 4.
regarding its importance (i.e., with data such as GHG emissions
carrying greater weight than other disclosures). The Bloomberg ESG
3.2 | Examining the determinants of a firm's ESG
disclosure score is also tailored to different industry sectors. Based
transparency: Country versus firm effects
on the methodology of the Bloomberg ESG score, we assume that this
disclosure score can be viewed as the reflection of a firm's voluntary Hebb (2006) documents that transparency not only aligns shareholders
and mandatory disclosures, which help shareholders and stakeholders and managers, it also allows other stakeholders to engage and to
YU ET AL. 991

TABLE 1 Definition and our estimation methods for all variables in this study
Variable Symbol Definition/estimation methods
Tobin's Q Tobin's Q In this study, firm value is estimated by Tobin's Q
Tobin's Q = (market capitalization + liabilities + preferred equity +
minority interests)/(total assets)
For each sample year, we subtract average Tobin's Q for the industry
from the firm‐level Tobin's Q. We have 10 GICS sectors in this
study. Data are from Bloomberg
ESG disclosure ESG This variable is an indicator of ESG transparency. ESG = (ESG
disclosure score/100)
For each sample year, we subtract the average ESG disclosure score
for the industry from the firm‐level ESG disclosure score. There
are 10 GICS sectors in this study. Bloomberg summarizes the ESG
disclosure score. Higher scores indicate more transparency on
ESG issues
ESG disclosure2 (ESG)2 We use the square of ESG disclosure
Environmental Environmental This variable is as an indicator of environmental transparency.
disclosure Environmental = (Environmental disclosure score/100)
We subtract the average Environmental disclosure score for the
industry from the firm‐level Environmental disclosure score.
There are 10 GICS sectors in this study. Bloomberg summarizes
the Environmental disclosure score. Higher scores indicate more
transparency in environmental issues
(Environmental (Environmental)2 We use the square of Environmental disclosure
disclosure)2
R&D intensity R&D intensity The sum of R&D costs divided by sales for the previous 3 years
log(Firm asset size) log(firm size) Firm size is natural logarithm of the book value of assets as reported
by Bloomberg
Leverage ratio Leverage Leverage is defined as the debt/total asset ratio as reported by
Bloomberg
For each sample year, we subtract average leverage of the industry
from the firm‐level leverage. There are 10 GICS sectors in this study
Liquidity ratio (current or quick) Current ratio or a quick ratio We adopt the current ratio and quick ratio reported by Bloomberg
as our liquidity indicators
Institutional ownership Institutional ownership The percentage of common equity owned by institutional shareholders
Percentages of independent Percentages of independent The proportion of independent directors, who are neither current nor
director director former managers of the firm
Insider holdings Insider holdings The percentage of common equity owned by officers and directors
log(Board size) log(Board size) The natural logarithm of the number of directors sitting on each firm's
board as of the annual general meeting date in the given year
Percentages of women in senior Percentages of women in The percentage of women employed in senior management positions
management senior management at the company
Operating margin Operating margin Earnings before interest, tax, depreciation and amortization (EBITDA)
is a measure of a company's operating performance
It is an essential way to evaluate a firm's performance without having
to factor in tax environments, financing decisions and accounting
decisions
Return on assets ROA Ratio of earnings before interests/total assets as reported by Bloomberg
3‐Year average return on equity ROE3Y Average return on equity for the last 3 years
5‐Year average return ROE5Y Average return on equity for the last 5 years
on equity
PB ratio PB ratio P/B ratio = (share price)/(book value per share)
log(GDP per capita based on log(GDP per capita based GDP per capita (PPP based) is gross domestic product converted to
PPP) on PPP) international dollars using purchasing power parity rates and
divided by total population
We adopt the data from the International Monetary Fund's World
Economic Outlook Database
Corruption index Corruption In this study, Corruption = (Corruption index/100)
We adopt the relevant data from Transparency International from 2012
to 2016. The more the corruption, the fewer points are awarded to
the country. However, no corruption index data are available for the
following five sample countries: Indonesia, Colombia, Thailand,
Philippines and Egypt. We can obtain the responding date for our
other 40 sample countries
992 YU ET AL.

control behavior within a firm. Here, we model the determinants of the financial expertise of inside directors rather than to the propor-
ESG transparency and group these possible determinants into two tion of outside directors on the board, which is unrelated to perfor-
categories: firm level and country level. mance. The role of institutional investment in promoting long‐term
environmental performance is ambiguous. By examining the 500
largest US firms, Aggarwal and Dow (2012) show that institutional
3.2.1 | Country level
ownership brings a significantly negative impact to firms’ environ-
Previous contributions to the literature (De Soto, 1989; Gnyawali,
mental policy. Trucost (2009) suggests that institutional investors
1996; Husted, 2005) suggest that economic development is the key
do not consider carbon exposure to be an essential criterion in firm
driver behind environment sustainability. For instance, Gnyawali
allocation decisions.
(1996) finds that people in richer countries make more demands on
Given the evidence from the corporate governance literature, we
firms for environmental and socially responsible performance because
propose to include corporate governance structure at the firm level
they are better informed. Therefore, we examine whether the level of
as a variable in our model. This leads to Hypothesis 3. We measure
economic development can help to explain why some countries have
the degree of direct monitoring using the following four factors: (i)
better ESG transparency than others.
insider holdings, (ii) institutional ownership, (iii) percentages of inde-
To represent the economic development of these 47 sample coun-
pendent directors and (iv) board size. In this hypothesis, we examine
tries and territories, we use the natural logarithm of per‐capita gross
whether a change in one of these four factors has any impact on
domestic product (GDP) converted to US dollars at purchasing power
ESG transparency.
parity (PPP) exchange rates. This is the measure we prefer when com-
paring living conditions or when looking at per‐capita welfare across
Hypothesis 3: (i) An increased percentage of insider
countries. A nation's GDP at PPP exchange rates is the sum value of
holdings is associated with a negative impact on ESG dis-
all the services and goods produced in the country, valued at prices
closure. (ii) An increased percentage of institutional own-
prevailing in the United States. Overall, the PPP exchange rates are rel-
ership will have a negative impact on ESG disclosure (iii)
atively stable over time. We use data from the International Monetary
An increased percentage of independent directors will
Fund's World Economic Outlook Database. This leads to Hypothesis 2.
have a positive impact on ESG disclosure. (iv) A greater
Hypothesis 2: ESG disclosure is high in countries where board size will have a positive impact on ESG disclosure.
the level of economic development is high.
Finally, the estimating equation for our Hypotheses 2 and 3 is
We also adopt the corruption index data as one of our control var- shown as follows:
iables, sourced from Transparency International. Augustine (2012) doc-
uments that corporate governance has both external and internal ðIndustry−adjusted ESG disclosureÞ ¼ b0 þ b1* logðfirm sizeÞ (5)
dimensions, which can complement each other. Therefore, from the þb2* ðadjusted ROAÞ þ b3* logðadjusted leverage ratioÞ
view of external dimension, we use the corruption index to view the þb4* ðliquidity ratioÞ þ b5* ðR&D intensityÞ
larger context in which these listed‐firms operate. Nevertheless, no þb6* ðInsider holdingsÞ þ b7* ðInstitutional ownershipÞ
corruption index data are available for five sample countries: Indonesia, þb8* ðpercentage of independent directorsÞ
Colombia, Thailand, Philippines and Egypt. þb9* ðpercentage of women in managementÞ
þb10* logðboard sizeÞ þ b11* ðGDP per capita based on PPPÞ
3.2.2 | Firm level þb12* ðcorruptionÞ þ εðresidualÞ
An effective governance framework (i.e., independent board directors,
institutional investor, insider holdings, board size, etc.) is likely to The definitions for all variables in Equation (5) are given in Table 1.
reduce the agency costs associated with the separation of ownership Finally, for robustness checks, we also replace ROA with the fol-
and control. Studies have examined the direct monitoring approach lowing firm performance indicators: operating margin, 3‐year average
as one of the effective governance mechanisms that can overcome return on equity, 5‐year average return on equity and P/B ratio. We
control problems (Dahya & McConnell, 2007; Lee et al., 2008; Lee & present our empirical results of Hypothesis 2 and 3 in Section 4.
Lee, 2009; Liu, Miletkov, Wei, & Yang, 2015; Palmberg, 2015). More-
over, Bebchuk and Weisbach (2010) state that many of these gover-
4 | DATA SOURCES AND EMPIRICAL
nance mechanisms can serve as substitutes for one another. Chen,
RESULTS
Chen, and Wei (2009) hold a similar view and suggest that in countries
with weak legal protection of investors, firm‐level corporate gover-
nance can supplement country‐level shareholder protection in reduc-
4.1 | Data sources
ing the cost of equity. We use a global dataset of 1996 firms, which are selected from MSCI
Palmberg (2015) documents that independent directors of All Country World Index (ACWI). This sample covers approximately
Swedish listed firms have a positive impact on firms’ investment per- 85% of the global investable equity by market value and includes coun-
formance. However, other studies have a conflicting view. Adams tries from 47 developed and emerging countries and territories. Our
and Jiang (2016) examine the UK's property‐casualty insurance sample period is 2012–2016. We group these 1996 sample firms into
industry and find that superior performance can be attributed to ten Global Industry Classification Standard (GISC) sectors (refer to
YU ET AL. 993

TABLE 2 Components obtained from the MSCI All Country World index from Transparency International for these sample countries.
Index (ACWI) for 2012–16 Table 3 presents descriptive statistics for all key variables. We include
Our sample includes sample statistics for firms for which data are available.
Our ten GICS Firm 47 countries and
sectors numbers territories

1 Consumer 352 Austria, Australia, Belgium, 4.2 | Econometric procedure

discretionary Brazil, Canada, Chile, China,
2 Consumer staple 196 Colombia, Czech Republic, We analyze our panel dataset by starting with the likelihood‐ratio test.
3 Energy 134 Denmark, Egypt, Finland, If we reject the null hypothesis, then a panel approach—random effects
4 Healthcare 160 France, Germany, Greece,
5 Industrials 357 Hong Kong, Hungary, India, model or the fixed effects model—must be used. We then apply the
6 Information 225 Indonesia, Ireland, Israel, Hausman test to decide which model suits our panel dataset better.
technology Italy, Japan, Macao,
7 Materials 207 Malaysia, Mexico, the We also adopt the White diagonal as our coefficient covariance
8 Real estate 149 Netherlands, New Zealand, method, which is robust to heteroscedasticity (Reed & Ye, 2011).
9 Telecommunication 87 Norway, Philippines,
services Poland, Portugal, Finally, we carry out normality tests of the residuals, which can exam-
10 Utilities 129 Qatar, Russia, Saudi ine whether our model is well specified. All the residual distributions of
Total 1996 Arabia, Singapore, South
these regressions we report in this study are normal, indicating that our
Africa, South Korea, Spain,
Sweden, Switzerland, estimating equations are well specified.
Taiwan, Thailand, Turkey, Correlations between variables are reported in Appendix Table A1.
UK, United Arab Emirates,
US We note that environmental disclosure is highly correlated to ESG dis-
closure (0.9618). Also, a firm with a better transparency in ESG issues
Source: We make this analysis. The sample companies in this study are
selected from ACWI with a sample period from 2012 to 2016, which is more likely to disclose more information on environmental issues. In
covers approximately 85% of the global investable equity opportunity set. this study, we do not mix any variables that are highly correlated (cor-
Our sample includes these 47 countries and territories. relation coefficient > 0.8), in the same estimating equation. This is
commonly adopted as a rule of thumb for avoiding a multicollinearity
Table 2), but we exclude financial services firms due to concerns that problem.
banking and financial regulations might affect transparency and its
impact on performance.
We collect ESG disclosure data, environmental disclosure data,
4.3 | ESG transparency and firm value
financial statement data and corporate governance data from How does ESG disclosure influence a firm's value? Using Equation (4),
Bloomberg, while the data of PPP‐based per‐capita GDP for our 47 we investigate whether a publicly listed company's transparency in
sample countries are from the International Monetary Fund's World ESG issues can impact its firm value as measured by Tobin's Q. Tobin's
Economic Outlook Database. We also adopt the annual corruption Q is estimated as the ratio of the enterprise value of the firm plus cash

TABLE 3 Descriptive statistics of all variables (currency: US dollars)

Mean Maximum Minimum SD Observations
(ESG disclosure score/100) 0.3336 0.8678 0.0207 0.1605 7,818
(Environmental disclosure score/100) 0.3149 0.9380 0.0138 0.1771 6,321
log (Firm size) (measurement unit for firm size: million US dollars) 9.1805 13.7467 1.5261 1.3048 9,584
Leverage ratio (debt/assets) 0.2693 3.4680 0 0.1849 9,521
Quick ratio 0.0123 0.9281 0.0000 0.0209 9,192
Current ratio 0.0187 1.7181 0.0000 0.0288 9,293
Tobin's Q 0.0201 0.4339 0.0038 0.0182 9,409
ROA 0.0597 1.2081 −1.9867 0.0776 9,522
Operating margin 0.1271 25.2599 −150.7216 1.6629 9,573
ROE3Y 0.1547 8.0115 −1.6028 0.2357 9,045
PB ratio 0.0484 15.6820 0.0004 0.3199 9,201
Insider holdings (%) 0.0303 0.8439 0.0000 0.0873 9,412
Percentage of institutional investor holdings (%) 0.6180 1.5744 0.0000 0.2936 9,412
Percentage of women in management (%) 0.2179 0.7600 0.0000 0.1161 1870
log(Board size) 2.3128 4.1109 0.6931 0.2920 8,145
Percentage of independent board members (%) 0.5799 1.0000 0.0000 0.2697 7,679
R&D intensity (%) 0.0466 133.2702 0.0000 1.4161 8,928
log(GDP per capita based on PPP) 10.5370 11.8953 8.5166 0.5116 9,834
(Corruption/100) 0.6894 0.9200 0.2700 0.1397 9,415

This table reports descriptive statistics for the variables used in Equations (4) and (5). Refer to Table 1 for definitions of the variables. Our sample period is
from 2012 to 2016. For each variable, we present the full sample descriptive statistics.
 994

TABLE 4 Analyses of firm performance with “ESG disclosure” and “environmental disclosure,” 2012–16
Model (1) Equation (4) Model (2) Equation (4) Model (3) Equation (4) Model (4) Equation (4)

Hypotheses Hypothesis 1(a): We assume that the association Hypothesis 1(b): We assume that the association
between a firm's performance and ESG disclosure between a firm's performance and environmental
is conditional on agency costs and governance disclosure is conditional on agency costs and
structures. A publicly listed company's governance structures. A publicly listed company's
transparency in ESG issues can impact on its transparency in environmental issues can impact
Tobin's Q. We predict that the relationship on its Tobin's Q. We predict that the relationship
between firm performance and ESG disclosure is between firm performance and environmental
not linear disclosure is not linear
Dependent variable Firm Industry‐adjusted Industry‐adjusted Industry‐adjusted Industry‐adjusted
performance – Tobin's Q Tobin's Q Tobin's Q Tobin's Q Tobin's Q
Estimation method Panel EGLS Period Weights Panel EGLS Period Weights Panel EGLS Period Weights Panel EGLS Period Weights
Constant 0.0451*** (0.0049) 0.0453*** (0.0049) 0.0362*** (0.0047) 0.0359*** (0.0047)
ESG disclosure (industry‐adjusted) −0.0027** (0.0013) −0.0029** (0.0013)
(ESG disclosure)2 0.0130** (0.0062) 0.0136** (0.0063)
Environmental disclosure 0.0000 (0.0010) 0.0000 (0.0010)
(industry‐adjusted)
(Environmental disclosure)2 0.0026 (0.0042) 0.0022 (0.0042)
log(Firm size) −0.0046*** (0.0003) −0.0046*** (0.0003) −0.0035*** (0.0002) −0.0034*** (0.0002)
Leverage ratio (industry‐adjusted; −0.0029** (0.0012) −0.0024** (0.0012)
debt/assets)
Current ratio
Quick ratio 0.0386* (0.0204)
log(GDP per capita based on PPP) −0.0010** (0.0004) −0.0010** (0.0004) −0.0011*** (0.0004) −0.0012*** (0.0004)
R&D intensity 0.0270*** (0.0062) 0.0290*** (0.0062)
Percentage of independent 0.0080*** (0.0007) 0.0079*** (0.0007) 0.0084*** (0.0006) 0.0086*** (0.0006)
directord
Institutional ownership 0.0022*** (0.0007) 0.0023*** (0.0007) 0.0030*** (0.0007) 0.0029*** (0.0007)
Observations (firm number) 1996 1996 1444 1444
Regression residual Normally distributed Normally distributed Normally distributed Normally distributed
Adjusted R2 0.2037 0.2035 0.1541 0.1552

Asterisks denote significance at the

*10,
**5 and
***1% levels.
This table reports regression coefficients (standard deviations in parentheses) and diagnostic statistics for Equation (4). There are four specifications of the model. The sample comprises 1996 firms from the MSCI All‐
Share Index. Because the residuals of the regression are normally distributed, it indicates that our model is well specified. Our sample period is from 2012 to 2016. Due to lack of availability of environmental disclosure,
missing observations of environmental disclosure reduce our sample size from 1996 firms to 1444 firms. Table 4 summarizes our empirical results of Equation (4). Models (1) and (2) shows that a nonlinear relationship
YU

exists between firm performance and ESG disclosure, whereas this is not the case for environmental disclosure in Models (3) and (4). EGLS stands for the “Estimated Generalised Least Squares Method”.
ET AL.
YU ET AL. 995

FIGURE 1 Impact of ESG disclosure on

Tobin's Q. This figure shows the relationship
between Tobin's Q and the linear and
quadratic ESG disclosure variables because
the latter are statistically significant in the
performance regression in Equation 4. Our
sample group comprises 1996 firms from the
MSCI All share index from 2012 to 2016. The
group average of ESG disclosure is reported as
0.33 in this figure. The more information
disclosed, the higher the disclosure score. The
Bloomberg score ranges from 0 to 100. In this
study, we estimate ESG disclosure as the ratio
of Bloomberg ESG scores divided by 100.
Therefore, the maximum value of ESG
disclosure in this figure is 1 [Colour figure can
be viewed at wileyonlinelibrary.com]

to the book value of assets. We follow previous studies (Aggarwal & Tobin's Q. We predict that the relationship between the
Dow, 2012; Bebchuk et al., 2009) by using an industry‐adjusted mea- firm's performance and ESG disclosure is not linear.
sure of Tobin's Q, because Tobin's Q can be highly industry‐dependent.
Hypothesis 1 (b): We assume that the association
Meanwhile, we control for the firm's characteristics and two country‐
between the firm's performance and environmental dis-
level factors: level of economic development and corruption index.
closure is conditional on agency costs and governance
Hypothesis 1 (a): We assume that the association structures. A publicly listed company's transparency in
between the firm's performance and ESG disclosure is condi- environmental issues can impact its Tobin's Q. We predict
tional on agency costs and governance structures. A publicly that the relationship between the firm's performance and
listed company's transparency in ESG issues can impact its wnvironmental disclosure is not linear.

FIGURE 2 ESG disclosure vs. environmental disclosure. Source: This figure was constructed by the authors and the relevant data are collected
from Bloomberg. The sample firms are selected from the components of the MSCI All‐Share Index. Our sample period is from 2012 to 2016
[Colour figure can be viewed at wileyonlinelibrary.com]
TABLE 5 Analyses on ESG disclosure with the firm performance indicator ROA, 2012–16
996

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

Dependent variable ESG disclosure ESG disclosure ESG disclosure ESG disclosure Dependent variable ESG disclosure ESG disclosure
(not industry‐ (not industry‐ (not industry‐ (not industry‐ (industry‐adjusted) (industry‐adjusted) (industry‐adjusted)
adjusted) adjusted) adjusted) adjusted)
Estimation method Cross section Two‐stage least squares— Cross section Two‐stage least squares— Panel EGLS Period Panel EGLS Period
random effects Instrument with “the random effects Instrument with “the Weights Weights
average growth rate average growth rate of
of EPS in the last EPS in the last
three years.” three years.”
Constant −0.5416*** (0.1642) −0.4722** (0.1975) −0.5416*** (0.1642) −0.4764** (0.1993) Constant −0.8217*** (0.1130) −0.8190*** (0.1146)
log(Firm size) 0.0312*** (0.0035) 0.0277*** (0.0045) 0.0311*** (0.0035) 0.0275*** (0.0045) log(Firm size) 0.0304*** (0.0027) 0.0295*** (0.0026)
ROA (not industry‐adjusted) −0.0039 (0.0223) −0.1368 (0.0987) −0.0028 (0.0223) −0.1373 (0.0986) ROA (industry‐adjusted) 0.0992** (0.0411) 0.0999** (0.0404)
Current ratio −0.0122 (0.2523) −0.1250 (0.3323) Current ratio 1.2355*** (0.4210)
Quick ratio −0.1608 (0.2979) −0.2881 (0.3816) Quick ratio 0.5063 (0.4993)
Leverage (not industry‐ −0.0564** (0.0277) −0.0582** (0.0276) Leverage (industry‐adjusted)
adjusted)
Insider holdings Insider holdings
Institutional ownership Institutional ownership −0.0343*** (0.0101) −0.0318*** (0.0102)
Percentage of women in −0.0485* (0.0276) −0.0479* (0.0276) Percentage of women in −0.0428* (0.0255)
management management
log(Board size) 0.0213* (0.0111) 0.0209* (0.0118) 0.0212* (0.0111) 0.0208* (0.0118) log(Board size) 0.0566*** (0.0126) 0.0547*** (0.0126)
Percentage of independent Percentage of independent
directors directors
R&D intensity 0.0967* (0.0528) 0.1483** (0.0755) 0.0920* (0.0533) 0.1530** (0.0765) R&D intensity 0.2435*** (0.0540) 0.2772*** (0.0578)
log(GDP per capita based 0.0811*** (0.0176) 0.0764*** (0.0207) 0.0818*** (0.0176) 0.0771*** (0.0209) log(GDP per capita based 0.0670*** (0.0125) 0.0693*** (0.0127)
on PPP) on PPP)
Corruption −0.2583*** (0.0461) −0.2296*** (0.0535) −0.2584*** (0.0461) −0.2292*** (0.0537) Corruption −0.2559*** (0.0332) −0.2579*** (0.0334)
Observations (firm number) 1996 1996 1996 1996 Observations 1996 1996
Regression residual Normally distributed Normally distributed Normally distributed Normally distributed Regression residual Normally distributed Normally distributed
2 2
Adjusted R 0.1063 0.0726 0.1058 0.0712 Adjusted R 0.2462 0.2395

Asterisks denote significance at the

*10,
**5 and
***1% levels.
This table reports regression coefficients (standard deviations in parentheses) and diagnostic statistics for the ESG disclosure regression in Equation 5, using return on assets (ROA) as the firm performance indicator. There
are six specifications of the model. The first four models report the regression without an industry‐adjusted in the following three variables: ESG disclosure, ROA and leverage ratio. For the last two models, Models (5) and
Model (6), we use an industry‐adjusted measure of these three variables. The sample comprises 1996 firms from the MSCI All‐Share Index. If the residuals of the regression are normally distributed, our model is well
specified. Our sample period is from 2012 to 2016. Moreover, we check for endogeneity issues that may be present in our analyses. We are concerned that the direction of causality between return on equity (ROA)
and ESG disclosure could run both ways. To ascertain whether this is the case, we use the panel least square estimation method supplemented by two‐stage least squares where appropriate. ROA is instrumented by
the following variable, the average growth rate of EPS in the last 3 years. We report results from the two‐stage least squares analyses in Models (2) and Model (4). Based on our empirical results shown in Models (1)
YU

to (4), we can confirm that ROA and ESG disclosure are not determined endogenously. However, in Models (5) and (6) with an industry‐adjusted in the following three variables: ESG disclosure, ROA and leverage ratio,
our empirical results show that the higher the ROA of the firms, the more the ESG transparency (disclosure) in firms.
ET AL.
TABLE 6 Analyses on ESG disclosure with the other four performance indicators: operating margin, 3‐year average return on equity, 5‐year average return on equity and PB ratio, with a sample period 2012–
YU

16
ET AL.

Model (1) Model (2) Model (3) Model (4) Model (5) Model (6) Model (7) Model (8)
Equation (5) Equation (5) Equation (5) Equation (5) Equation (5) Equation (5) Equation (5) Equation (5)
Dependent variable ESG disclosure ESG disclosure ESG disclosure ESG disclosure ESG disclosure ESG disclosure ESG disclosure ESG disclosure
(industry‐ (industry‐ (industry‐ (industry‐ (industry‐ (industry‐ (industry‐ (industry‐
adjusted) adjusted) adjusted) adjusted) adjusted) adjusted) adjusted) adjusted)
Estimation method Panel EGLS Period Panel EGLS Period Panel EGLS Period Panel EGLS Period Panel EGLS Period Panel EGLS Period Panel EGLS Period Panel EGLS Period
Weights Weights Weights Weights Weights Weights Weights Weights
Constant −0.7979*** (0.1113) −0.7939*** (0.1130) −0.8079*** (0.1157) −0.8048*** (0.1176) −0.8127*** (0.1171) −0.8096*** (0.1191) −0.8065*** (0.1113) −0.8000*** (0.1131)
log(Firm size) 0.0288*** (0.0026) 0.0279*** (0.0026) 0.0295*** (0.0027) 0.0284*** (0.0027) 0.0292*** (0.0027) 0.0281*** (0.0027) 0.0298*** (0.0026) 0.0287*** (0.0026)
Operating margin (industry‐
adjusted)
3‐Year average return on 0.0193** (0.0064) 0.0181** (0.0064)
equity (industry‐adjusted)
5‐Year average return on 0.0214*** (0.0056) 0.0200** (0.0056)
equity (industry‐adjusted)
PB ratio (industry‐adjusted)
Current ratio 1.2809*** (0.4146) 1.3509*** (0.4339) 1.3601*** (0.4445) 1.4008*** (0.4209)
Quick ratio 0.6084 (0.5035) 0.6376 (0.5129) 0.6446 (0.5288) 0.7531 (0.4934)
Leverage ratio (industry‐
adjusted)
Insider holdings −0.0936* (0.0558) −0.0953* (0.0567)
Institutional ownership −0.0331*** (0.0102) −0.0307*** (0.0102) −0.0377*** (0.0101) −0.0352*** (0.0101) −0.0399*** (0.0102) −0.0372*** (0.0102) −0.0289*** (0.0102) −0.0267*** (0.0102)
Percentage of women in −0.0430* (0.0260) −0.0505* (0.0257) −0.0524** (0.0260)
management
log(Board size) 0.0576*** (0.0126) 0.0559*** (0.0126) 0.0511*** (0.0126) 0.0496*** (0.0127) 0.0500*** (0.0127) 0.0485*** (0.0127) 0.0630*** (0.0126) 0.0616*** (0.0127)
Percentage of independent
directors
R&D intensity 0.2354*** (0.0541) 0.2648*** (0.0577) 0.2670*** (0.0568) 0.2983*** (0.0609) 0.2803*** (0.0590) 0.3120*** (0.0633) 0.2263*** (0.0531) 0.2557*** (0.0568)
log(GDP per capita based on 0.0658*** (0.0124) 0.0679*** (0.0125) 0.0682*** (0.0127) 0.0706*** (0.0129) 0.0695*** (0.0128) 0.0718*** (0.0130) 0.0646*** (0.0124) 0.0665*** (0.0126)
PPP)
Corruption −0.2585*** (0.0331) −0.2600*** (0.0333) −0.2656*** (0.0332) −0.2673*** (0.0334) −0.2680*** (0.0333) −0.2696*** (0.0336) −0.2662*** (0.0336) −0.2669*** (0.0339)
Observations (firm number) 1996 1996 1996 1996 1996 1996 1996 1996
Regression residual Normally distributed Normally distributed Normally distributed Normally distributed Normally distributed Normally distributed Normally distributed Normally distributed
Adjusted R2 0.2424 0.2357 0.2472 0.2394 0.2476 0.2397 0.2546 0.2468

Asterisks denote significance at the

*10,
**5 and
***1% levels.
This table reports regression coefficients (standard deviations in parentheses) and diagnostic statistics for the ESG disclosure regression in Equation 5, using the other four performance indicators: operating margin, 3‐year
average return on equity, 5‐year average return on equity and PB ratio. There are eight specifications of the model, and we use an industry‐adjusted measure of these three variables. The sample comprises 1996 firms
from the MSCI All‐Share Index. If the residuals of the regression are normally distributed, our model is well specified for the subsample. Our sample period is from 2012 to 2016. The empirical results (Models 3–6) show
997

that the higher the “3‐year average return on equity” and the higher the “5‐year average return on equity,” the more the ESG transparency and the environmental transparency (disclosure) in firms.
 998

TABLE 7 Analyses on environmental disclosure with the firm performance indicator ROA, 2012–16)
Model (1) Equation (2) Model (2) Equation (2) Model (3) Equation (2) Model (4) Equation (2) Model (5) Equation (2) Model (6) Equation (2)

Dependent variable Environmental disclosure Environmental disclosure Environmental disclosure Environmental disclosure Dependent variable Environmental disclosure Environmental disclosure
(not industry‐adjusted) (not industry‐adjusted) (not industry‐adjusted) (not industry‐adjusted) (industry‐adjusted) (industry‐adjusted) (industry‐adjusted)
Estimation method Cross section random Two‐stage least squares— Cross section random Two‐stage least squares— Panel EGLS Period Weights Panel EGLS Period
effects Instrument with “the effects Instrument with “the Weights
average growth rate of average growth rate of
EPS in the last 3 years” EPS in the last 3 years”
Constant −0.9212*** (0.2132) −0.8400** (0.2611) −0.9232*** (0.2137) −0.8437*** (0.2629) Constant −1.0180*** (0.1538) −1.0059*** (0.1559)
log(Firm size) 0.0400*** (0.0048) 0.0382*** (0.0060) 0.0399*** (0.0048) 0.0381*** (0.0061) log(Size) 0.0411*** (0.0035) 0.0397*** (0.0035)
ROA (not industry‐ −0.0133 (0.0323) −0.1124 (0.1096) −0.0129 (0.0324) −0.1130 (0.1089) ROA (industry‐adjusted) 0.1404*** (0.0541) 0.1388*** (0.0533)
adjusted)
Current ratio −0.1865 (0.3558) −0.1041 (0.4603) Current ratio 1.8766*** (0.5612)
Quick ratio 0.0589 (0.4208) −0.2946 (0.5310) Quick ratio 1.0652 (0.6525)
Leverage (not industry‐ −0.0815** (0.0360) −0.0495* (0.0291) −0.0838** (0.0358) Leverage (industry‐
adjusted) adjusted)
Insider holdings −0.1078* (0.0622) −0.1069* (0.0622) Insider holdings −0.1408* (0.0735) −0.1416* (0.0726)
Institutional ownership −0.0288* (0.0171) −0.0284* (0.0171) Institutional ownership −0.0805*** (0.0138) −0.0778*** (0.0138)
Percentage of women −0.1096*** (0.0367) −0.1052*** (0.0371) −0.1090*** (0.0367) −0.1036*** (0.0371) Percentage of women in −0.1527*** (0.0344) −0.1630*** (0.0340)
in management management
log(Board size) 0.0302* (0.0169) 0.0303* (0.0169) log(Board size) 0.0463*** (0.0171) 0.0436** (0.0171)
Percentage of Percentage of
independent independent director
director
R&D intensity 0.1827** (0.0788) 0.1962* (0.1059) 0.1762** (0.0795) 0.1991* (0.1072) R&D intensity 0.1487** (0.0644) 0.1837*** (0.0686)
log(GDP per capita 0.1082*** (0.0232) 0.0974*** (0.0277) 0.1088*** (0.0232) 0.0979*** (0.0278) log(GDP per capita based 0.0805*** (0.0166) 0.0829*** (0.0168)
based on PPP) on PPP)
Corruption −0.3112*** (0.0629) −0.2687*** (0.0739) −0.3110*** (0.0630) −0.2681*** (0.0742) Corruption −0.2783*** (0.0433) −0.2787*** (0.0436)
Observations (firm 1444 1444 1444 1444 Observations 1444 1444
number)
Regression residual Normally distributed Normally distributed Normally distributed Normally distributed Regression residual Normally distributed Normally distributed
Adjusted R2 0.1021 0.0741 0.1012 0.0726 Adjusted R2 0.2307 0.2229

Asterisks denote significance at the

*10,
**5 and
***1% levels.
This table reports regression coefficients (standard deviations in parentheses) and diagnostic statistics for the environmental disclosure regression in Equation 5, using return on asset (ROA) as the firm's performance
indicator. There are six specifications of the model. The first four models report the regression without any industry adjustment in the following three variables: environmental disclosure, ROA and leverage ratio. For
last two models, Models (5) and Model (6), we use an industry‐adjusted measure of these three variables. The sample comprises 1444 firms from the MSCI All‐Share Index. If the residuals of the regression are normally
distributed, our model is well specified. Our sample period is from 2012 to 2016. Moreover, we check for endogeneity issues that may be present in our analyses. We are concerned that the direction of causality between
ROA and the environmental disclosure could run both ways. To ascertain whether this is the case, we use the panel least square estimation method supplemented by two‐stage least squares where appropriate. ROA is
instrumented by the following variable, the average growth rate of EPS in the last 3 years. We report results from the two‐stage least squares analyses in Models (2) and (4). Based on our empirical results shown in Models
YU

(1) to (4), we can confirm that ROA and environmental disclosure are not determined endogenously. However, in Models (5) and (6) we use industry‐adjusted measures in the following three variables: environmental
disclosure, ROA and leverage ratio; our empirical results show that the higher the ROA of the firms, the more the environmental transparency (disclosure) in firms.
ET AL.
YU ET AL. 999

We report our empirical results for Hypothesis 1(a) and 1(b) in can reduce the agency costs associated with moral hazard problems.
Table 4. Our finding confirms that increasing ESG disclosure can reduce
Our empirical results reveal a nonlinear relationship between ESG investors’ information symmetry and agency costs, which is consistent
transparency and a firm's performance. The linear term and the qua- with previous findings (Cheng et al., 2014; Miller & Reisel, 2012; Zhu &
dratic term of ESG disclosure are statistically significant. Based on Cai, 2014).
Model (1) and Model (2) shown in Table 4, we interpret our empirical Regarding the control variables, a few of interesting findings
results as supporting evidence for Hypothesis 1(a). We show the rela- emerge. We obtain consistent results for the variable of log(size) in
tionship between ESG transparency and Tobin's Q in Figure 1. Further- Table 4, which has a negative and statistically significant impact on
more, Table 3 shows that the average ESG transparency of our Tobin's Q. This may be explained by the way we select our sample
observations is 0.33. Therefore, the average score of ESG transparency firms. Companies are selected from the MSCI ACWI, which captures
is greater than a local minimum point at 0.2077, placing it on the right‐ large and midcap stocks across 23 developed markets and 24 emerging
hand side of the U‐shaped curve. Based on our empirical results markets. The negative sign of log(size) may imply that there are
(Models 1 and 2 in Table 4), we learn that ESG disclosure benefits diseconomies of scale.
exceed disclosure costs as soon as firm transparency in ESG issues
based on the disclosure score rises above 0.2077. This value is equal
4.4 | Determinants of ESG transparency: Country
to a 20.77 ESG disclosure score out of a maximum score of 100.
Overall, because the average ESG transparency in our sample is
versus firm‐level factors
0.3336 (Table 3, Figure 1), most of our sample firms could obtain net The explanatory variables that we believe will influence ESG disclosure
benefits from greater ESG disclosure. The impact of ESG disclosure can be grouped into two categories: country level and firm level. In this
on Tobin's Q is also economically significant. We find that a one‐ section, we use Equation (5) to examine Hypotheses 2 and 3.
standard deviation increase in ESG transparency can positively Firstly, we include a firm performance indicator, return on assets
enhance Tobin's Q by around 4.77% of the mean, all else being equal. (ROA), as our key control variable in Equation (5). We then check for
For each of these 10 GICS sectors in this study, we observe that all endogeneity issues that may be present in our regression analyses.
have an average ESG disclosure score greater than 20.77 points out We suspect that higher ROA would lead to increased ESG disclosure.
of a maximum of 100 (Figure 2). Meanwhile, the impact on ESG disclosure may also significantly
Our finding is similar to that of the Bank of America Merrill Lynch enhance ROA because of a possible reduction in firms’ reputational
(2017) and Eccles et al. (2001). Eccles et al. (2001) document that if a risk. We are concerned that the direction of causality between ROA
firm's market value is over its book value, additional nonfinancial infor- and disclosure could run both ways. To ascertain whether this is the
mation can provide insights into the firm's intangible assets that are case, we use the panel least square estimation method supplemented
not captured in traditional financial statements. In this study, we also by two‐stage least squares estimates. We investigate this by
find that ESG data are value‐relevant. Figure 2 indicates that better instrumenting our ESG disclosure with the average growth rate of
ESG transparency is beneficial to Tobin's Q. This may imply that ESG EPS (earnings per share) in the last 3 years (EPS3Y).
transparency can provide insightful information to investors and that Table 5 reports regression coefficients (standard deviations in
ESG disclosure can be used as one of the methods to improve a firm's parentheses) and diagnostic statistics for the ESG disclosure regression
corporate governance. For instance, as stakeholders’ expectations may in Equation 5, using ROA as the firm performance indicator. There are
shape the image of a company, firms are likely to have an interest in six specifications for this model. The first four models report the
adapting their management methods to environmental and social stan- regression without industry adjustment in the following three vari-
dards if they wish to attract investors. ables: ESG disclosure, ROA and the leverage ratio. Based on the results
Furthermore, we discuss how environmental disclosure can influ- shown in Models (1) to (4), we verify that ROA and ESG disclosure are
ence Tobin's Q. Missing observations of environmental disclosure not determined endogenously. Therefore, there is no two‐way effect
reduce our sample size from 1996 to 1444 firms. Based on Models between ROA and ESG disclosure. Furthermore, in Models (5) and
(3) and (4) shown in Table 4, Hypothesis 1(b) is rejected. Our empirical (6), we use an industry‐adjusted measure of these three variables:
results show that neither the linear term nor the quadratic term of ESG disclosure, ROA and the leverage ratio. The regression results
environmental disclosure is statistically significant to the performance shown in the last two models (Table 5) suggest that the higher ROA,
indicator, Tobin's Q. We suggest that a publicly listed company's envi- the more ESG transparency (disclosure) in firms.
ronmental disclosure does not impact on its Tobin's Q. We continue by examining transparency in environmental issues.
In this study, we define R&D intensity as the sum of research and The results reported in Models (5) and (6) in Table 7 imply that a firm
development (R&D) costs divided by sales for the previous 3 years. It with a higher ROA will be more transparent in environmental issues.
needs innovation activities to generate an impact on firm performance. The influence of ROA on environmental disclosure is positive and statis-
We use R&D intensity as firms’ agency and monitoring costs. We find tically significant. We also confirm that ROA is not endogenously deter-
that the variable of “R&D intensity” has a statistically positive impact mined by the environmental disclosure (refer to Models 1–4 in Table 7.
on a firm's performance (Models 1 and 2 in Table 4. Our results can The relevant 2SLS results are shown in Models (2) and (4) in Table 7.
be interpreted as the supporting evidence for Hypothesis 1(a). For Furthermore, for consistency and robustness, we use four other
firms with greater R&D intensity, which imply that their assets or activ- firm performance indicators in place of ROA in Equation 5. These firm
ities are difficult for shareholders to monitor, better ESG transparency performance indicators are operating margin, the 3‐year average
 1000

TABLE 8 Analyses on environmental disclosure with the other four performance indicators: operating margin, 3‐year average return on equity, 5‐year average return on equity and PB ratio, with a sample
period 2012–16
Model (1) Model (2) Model (3) Model (4) Model (5) Model (6) Model (7) Model (8)
Equation (5) Equation (5) Equation (5) Equation (5) Equation (5) Equation (5) Equation (2) Equation (2)
Dependent Environmental Environmental Environmental Environmental Environmental Environmental Environmental Environmental
variable disclosure disclosure disclosure disclosure disclosure disclosure disclosure disclosure
(industry‐ (industry‐ (industry‐ (industry‐ (industry‐ (industry‐ (industry‐ (industry‐
adjusted) adjusted) adjusted) adjusted) adjusted) adjusted) adjusted) adjusted)
Estimation Panel EGLS Period Panel EGLS Period Panel EGLS Period Panel EGLS Period Panel EGLS Panel EGLS Period Panel EGLS Period Panel EGLS Period
method Weights Weights Weights Weights Period Weights Weights Weights Weights
Constant −0.9882*** (0.1513) −0.9745*** (0.1535) −0.9892*** (0.1575) −0.9766*** (0.1600) −0.9864*** (0.1593) −0.9734*** (0.1620) −1.0045*** (0.1522) −0.9873*** (0.1545)
Log(Firm size) 0.0390*** (0.0035) 0.0377*** (0.0034) 0.0406*** (0.0035) 0.0389*** (0.0035) 0.0409*** (0.0036) 0.0392*** (0.0036) 0.0404*** (0.0035) 0.0388*** (0.0035)
Operating margin
(industry‐
adjusted)
3‐Year average 0.0280*** (0.0088) 0.0263*** (0.0088)
return on equity
(industry‐
adjusted)
5‐Year average 0.0317*** (0.0076) 0.0297*** (0.0077)
return on equity
(industry‐
adjusted)
PB ratio (industry‐
adjusted)
Current ratio 1.9391*** (0.5529) 2.0545*** (0.5739) 2.0952*** (0.5830) 2.1042*** (0.5668)
Quick ratio 1.2063* (0.6583) 1.2520* (0.6670) 1.2708* (0.6832) 1.3800** (0.6532)
Leverage
(industry‐
adjusted)
Insider holdings −0.1494** (0.0733) −0.1499** (0.0724) −0.1389* (0.0752) −0.1411* (0.0741) −0.1268* (0.0766)
Institutional −0.0792*** (0.0138) −0.0767*** (0.0138) −0.0832*** (0.0138) −0.0804*** (0.0138) −0.0839*** (0.0139) −0.0808*** (0.0140) −0.0743*** (0.0139) −0.0719*** (0.0139)
ownership
Percentage of −0.1448*** (0.0346) −0.1559*** (0.0343) −0.1613*** (0.0348) −0.1721*** (0.0344) −0.1629*** (0.0351) −0.1735*** (0.0347) −0.1273*** (0.0350) −0.1388*** (0.0345)
women in
management
log(Board size) 0.0472*** (0.0171) 0.0448*** (0.0171) 0.0388** (0.0172) 0.0366** (0.0172) 0.0369** (0.0174) 0.0348** (0.0174) 0.0527*** (0.0173) 0.0506*** (0.0173)
Percentage of
independent
director
R&D intensity 0.1368** (0.0642) 0.1666** (0.0683) 0.1665** (0.0662) 0.1988*** (0.0706) 0.1746** (0.0677) 0.2083*** (0.0723) 0.1305** (0.0644) 0.1608** (0.0685)

(Continues)
YU
ET AL.
 YU ET AL. 1001

return on equity (ROE3Y), the 5‐year average return on equity

This table reports regression coefficients (standard deviations in parentheses) and diagnostic statistics for the environmental disclosure regression in Equation 5, using the other four performance indicators: operating

disclosure, return on assets (ROA) and leverage ratio. The sample comprises 1444 firms from the MSCI All‐Share Index. If the residuals of the regression are normally distributed, our model is well specified. Our sample
period is from 2012 to 2016. The empirical results (Models 3–6) show that the higher the “3‐year average return on equity” and the higher the “5‐year average return on equity,” the more the environmental transparency
margin, 3‐year average return on equity, 5‐year average return on equity and PB ratio. There are eight specifications of the model, and we use an industry‐adjusted measure in the following three variables: environmental
Normally distributed
−0.2939*** (0.0444)
0.0805*** (0.0167)
(ROE5Y) and the price‐to‐book ratio (PB). We report the relevant
Equation (2) empirical results in Tables 6 and 8. Overall, our results (Tables 5–8) sug-
gest that ROA, ROE3Y and ROE5Y have a positive influence on ESG
Model (8)

0.2248
and environmental disclosure. Based on our empirical results, we con-

1444
clude that a firm with good past financial performance is more likely to
be more transparent in ESG issues.

Normally distributed
−0.2956*** (0.0439)
0.0786*** (0.0165)

4.4.1 | Country level

Equation (2)
Model (7)

0.2336 Hypothesis 2: ESG/environmental disclosure is high in

1444

countries where the level of economic development is high.

The results in Tables 6 and 8 suggest that ESG and environmental

Normally distributed
−0.2938*** (0.0438)
0.0830*** (0.0173)

disclosure is high in countries where the level of economic develop-

ment is high. This should come as no surprise as previous studies also
Equation (5)

show that environmental degradation is attributed to low economic

Model (6)

0.2213

development (Gnyawali, 1996; Husted, 2005).

1444

With regard to the corruption index, our results show that a coun-
try with less corruption will be less forthcoming in ESG/environmental
Normally distributed
−0.2941*** (0.0434)
0.0805*** (0.0171)

disclosure.
Equation (5)
Model (5)

4.4.2 | Firm level

0.2307
1444

Hypothesis 3: (i) An increased percentage of insider

holdings is associated with a negative impact on ESG dis-
Normally distributed
−0.2895*** (0.0436)
0.0829*** (0.0172)

closure. (ii) An increased percentage of institutional own-

ership will have a negative impact on ESG disclosure. (iii)
Equation (5)

An increased percentage of independent directors will

Model (4)

0.2217

have a positive impact on ESG disclosure. (iv) A greater

1444

board size will have a positive impact on ESG disclosure.

We present the relevant empirical results of Hypothesis 3. A

Normally distributed
−0.2897*** (0.0432)
0.0804*** (0.0169)

higher percentage of insider holdings is detrimental to a firm's ESG

(Table 6) and environmental transparency (Table 8). Our results sug-
Equation (5)
Model (3)

gest that ESG/environmental disclosure is lower in firms with a higher

0.2307
1444

percentage of insider holdings. This is similar to findings of Serafeim

and Grewal (2017), which suggests that firms that are larger and less
closely held tend to disclose more.
Normally distributed
−0.2821*** (0.0435)
0.0813*** (0.0166)

The results we present also show that the percentage of indepen-

dent directors on the board does not significantly affect ESG and envi-
Equation (5)
Model (2)

ronmental disclosure. This result suggests that independent board

0.2193

members are not necessarily more interested in ESG transparency than

1444

internal board members.

We find supporting evidence that ESG disclosure is better in firms
Normally distributed
−0.2827*** (0.0431)
0.0793*** (0.0163)

with larger boards. Finally, institutional ownership has a negative

impact on ESG and environmental disclosure (Table 8). Our finding is
Equation (5)

Asterisks denote significance at the

similar to the previous two studies (Aggarwal & Dow, 2012; Trucost,
Model (1)

0.2271

2009). Examining the 500 largest US firms, Aggarwal and Dow

1444

(2012) show that institutional ownership has a significantly negative

(Continued)

impact on a firm's environmental policy. Trucost (2009) suggests that

(disclosure) in firms.
Observations (firm
capita based on

institutional investors do not consider carbon exposure as an essential

Log(GDP per

criterion for firm allocation decision.

Adjusted R2

***1% levels.
Corruption

Regression
number)

residual

We also report the findings for our control variables. The effects
TABLE 8

PPP)

**5 and

of firm size, liquidity (current ratio) and R&D intensity across the mod-
*10,

ule specifications shown in Tables 6 and 8 are consistent. Our results

1002 YU ET AL.

suggest that these three factors exert a significant positive influence Business and Society, 51(4), 659–676. https://doi.org/10.1177/
on a firm's ESG/environmental transparency. We conclude that a firm 0007650312448623

with greater size, fewer insider holdings, lower percentage of institu- Avetisyan, E., & Hockerts, K. (2017). The consolidation of the ESG rating
industry as an enactment of institutional retrogression. Business
tional investors, better liquidity (current ratio) and higher R&D inten- Strategy and the Environment, 26(3), 316–330. https://doi.org/
sity will be more transparent in ESG issues. 10.1002/bse.1919
Bank of America Merrill Lynch (2017). Equity strategy focus point, ESG part
II: A deeper dive. New York, NY: Bank of America Merrill Lynch.
5 | C O N CL U S I O N
Bebchuk, L., Cohen, A., & Ferrell, A. (2009). What matters in corporate
governance? Review of Financial Studies, 22(2), 783–827. https://doi.
Previous literature has concentrated on assessing the best practice for org/10.1093/rfs/hhn099
each E, S and G dimension. In this study, our contribution is to focus on Bebchuk, L. A., & Weisbach, M. S. (2010). The state of corporate
examining a publicly listed company's ESG transparency and the quan- governance research. Review of Financial Studies, 23(3), 939–961.
https://doi.org/10.1093/rfs/hhp121
tity of ESG disclosure data. First, we evaluate the relationship between
Bennear, L. S., & Olmstead, S. M. (2008). The impacts of the ‘right to know’:
a firm's value (measured as Tobin's Q) and ESG transparency. We adopt
Information disclosure and the violation of drinking water standards.
the Bloomberg ESG disclosure score as an indicator of a company's Journal of Environmental Economics and Management, 56(2), 117–130.
ESG transparency. In addition, we model the determinants of ESG https://doi.org/10.1016/j.jeem.2008.03.002
transparency. Chen, K. C. W., Chen, Z., & Wei, K. C. J. (2009). Legal protection of
Our findings are similar to those of Bank of America Merrill investors, corporate governance, and the cost of equity capital. Journal
of Corporate Finance, 15(3), 273–289. https://doi.org/10.1016/j.
Lynch (2017) and Eccles et al. (2001), who suggest that ESG data jcorpfin.2009.01.001
are value‐relevant. We find that more ESG transparency is beneficial Cheng, B., Ioannou, I., & Serafeim, G. (2014). Corporate social responsibility
to value as measured by Tobin's Q, and there is a nonlinear relation- and access to finance. Strategic Management Journal, 35(1), 1–23.
ship between ESG and Tobin's Q. Our results suggest that ESG trans- https://doi.org/10.1002/smj.2131

parency can be viewed as additional nonfinancial information that Dahya, J., & McConnell, J. J. (2007). Board composition, corporate perfor-
mance, and the Cadbury Committee recommendation. Journal of
provides insight to investors. They also confirm that an increase in
Financial and Quantitative Analysis, 42(03), 535–564. https://doi.org/
ESG disclosure can reduce investors’ information symmetry and 10.1017/S0022109000004099
agency costs, which is consistent with the findings of Cheng et al. De Bakker, F. G. A., Groenewegen, P., & Den Hond, F. (2005). A bibliometric
(2014). With regard to the determinants of ESG transparency, our analysis of 30 years of research and theory on corporate social respon-
sibility and corporate social performance. Business and Society, 44(3),
analysis suggests that firms with greater size, fewer insider holdings,
283–317. https://doi.org/10.1177/0007650305278086
a lower percentage of institutional investors, better liquidity (current
De Soto, H. (1989). The other path. New York, NY: Harper & Row.
ratio) and higher R&D intensity will disclose more on ESG and
Delmas, M., & Blass, V. D. (2010). Measuring corporate environmental per-
environmental issues. formance: The trade‐offs of sustainability ratings. Business Strategy and
Finally, we propose that policy‐makers and regulators set the Environment, 19(4), 245–260. https://doi.org/10.1002/bse.676
mandatory or voluntary requirements to encourage firms to disclose Eccles, R. G., Herz, R. H., Keegan, E. M., & Phillips, D. M. H. (2001). The value
extensively. Better ESG transparency can only be achieved by a collab- reporting revolution: Moving beyond the earnings game. New York, NY:
John Wiley & Sons.
orative effort between companies, stock exchanges, security regula-
Eccles, R. G., Ioannou, I., & Serafeim, G. (2014). The impact of corporate
tors, investors and corporate reporting organizations, such as SASB
sustainability on organizational processes and performance. Manage-
and GRI. This study has limitations that could give rise to future ment Science, 60(11), 2835–2857. https://doi.org/10.1287/
research. We only examine the quantity of ESG disclosure data, but mnsc.2014.1984
the quality of ESG disclosure remains of interest. As firms go towards Escrig‐Olmedo, E., Muñoz‐Torres, M. J., & Fernández‐Izquierdo, M. Á.
(2013). Sustainable development and the financial system: Society's
providing sufficient ESG disclosure to the public in the future,
perceptions about socially responsible investing. Business Strategy and
researchers should focus on making ESG disclosure data comparable the Environment, 22(6), 410–428. https://doi.org/10.1002/bse.1755
across firms and countries. Galbreath, J. (2013). ESG in focus: The Australian evidence. Journal of
Business Ethics, 118(3), 529–541. https://doi.org/10.1007/s10551‐
ORCID 012‐1607‐9
Gnyawali, D. F. (1996). Corporate social performance: An international
Ellen Pei‐yi Yu http://orcid.org/0000-0002-9427-1201 perspective. Advances in International Comparative Management, 11,
251–273.
RE FE R ENC E S Governance and Accountability Institute (2017). SP500 companies pub-
Adams, M., & Jiang, W. (2016). Do outside directors influence the financial lished corporate‐sustainability reports 2016. New York, NY:
performance of risk‐trading firms? Evidence from the United Kingdom Governance and Accounting Institute, Inc.
(UK) insurance industry. Journal of Banking & Finance, 64, 36–51. Hainmueller, J., & Hiscox, M. J. (2012). Buying green? Field experimental
https://doi.org/10.1016/j.jbankfin.2015.11.018 tests of consumer support for environmentalism. SSRN Electronic
Aggarwal, R., & Dow, S. (2012). Corporate governance and business Journal. https://doi.org/10.2139/ssrn.2062429
strategies for climate change and environmental mitigation. The Hebb, T. (2006). The economic inefficiency of secrecy: Pension fund inves-
European Journal of Finance, 18(3–4), 311–331. https://doi.org/ tors’ corporate transparency concerns. Journal of Business Ethics, 63(4),
10.1080/1351847X.2011.579745 385–405. https://doi.org/10.1007/s10551‐005‐3968‐9
Augustine, D. (2012). Good practice in corporate governance: Hill, R. P., Ainscough, T., Shank, T., & Manullang, D. (2007). Corporate social
Transparency, trust, and performance in the microfinance industry. responsibility and socially responsible investing: A global perspective.
YU ET AL. 1003

Journal of Business Ethics, 70(2), 165–174. https://doi.org/10.1007/ Miller, D. P., & Reisel, N. (2012). Do country‐level investor protections
s10551‐006‐9103‐8 affect security‐level contract design? Evidence from foreign bond cov-
Himmelberg, C. P., Hubbard, R. G., & Palia, D. (1999). Understanding the enants. Review of Financial Studies, 25(2), 408–438. https://doi.org/
determinants of managerial ownership and the link between ownership 10.1093/rfs/hhr097
and performance. Journal of Financial Economics, 53(3), 353–384. Palmberg, J. (2015). The performance effect of corporate board of direc-
https://doi.org/10.1016/S0304‐405X(99)00025‐2 tors. European Journal of Law and Economics, 40(2), 273–292. https://
Hockerts, K., & Moir, L. (2004). Communicating corporate responsibility to doi.org/10.1007/s10657‐012‐9369‐5
investors: The changing role of the investor relations function. Journal Reed, W. R., & Ye, H. (2011). Which panel data estimator should I use?
of Business Ethics, 52(1), 85–98. https://doi.org/10.1023/B: Applied Economics, 43(8), 985–1000. https://doi.org/10.1080/
BUSI.0000033109.35980.16 00036840802600087
Husted, B. W. (2005). Culture and ecology: A cross‐national study of the Richardson, B. J. (2009). Keeping ethical investment ethical: Regulatory
determinants of environmental sustainability. Management International issues for investing for sustainability. Journal of Business Ethics, 87(4),
Review, 45(3), 349–370. 555–572. https://doi.org/10.1007/s10551‐008‐9958‐y
Jin, G. Z., & Leslie, P. (2003). The effect of information on product quality: Ruf, B. M., Muralidhar, K., & Paul, K. (1998). The development of a system-
Evidence from restaurant hygiene grade cards. The Quarterly Journal atic, aggregate measure of corporate social performance. Journal of
of Economics, 118(2), 409–451. https://doi.org/10.1162/ Management, 24(1), 119–133. https://doi.org/10.1177/
003355303321675428 014920639802400101
King, A., & Lenox, M. (2000). Does it really pay to be green? Accounting for Serafeim, G., & Grewal, J. (2017). The value relevance of corporate sustain-
strategy selection on the relationship between environment and financial ability disclosures: An analysis of a dataset from one large asset owner.
performance. New York, NY: New York University. SSRN Electronic Journal.. https://doi.org/10.2139/ssrn.2966767
Koellner, T., Weber, O., Fenchel, M., & Scholz, R. (2005). Principles for Shleifer, A., & Vishny, R. W. (1997). A survey of corporate governance. Jour-
sustainability rating of investment funds. Business Strategy and the nal of Finance, 52(2), 737–783. https://doi.org/10.1111/j.1540‐
Environment, 14(1), 54–70. https://doi.org/10.1002/bse.423 6261.1997.tb04820.x
KPMG (2016). Carrots and sticks: Global trends in sustainability reporting Slager, R., Gond, J.‐P., & Moon, J. (2012). Standardization as institutional
regulation and policy. New York, NY: United Nations Environment work: The regulatory power of a responsible investment standard.
Programme and KPMG International. Organization Studies, 33(5–6), 763–790. https://doi.org/10.1177/
Lai, A., Melloni, G., & Stacchezzini, R. (2016). Corporate sustainable 0170840612443628
development: Is “integrated reporting” a legitimation strategy? Business Smith, C. W., & Watts, R. L. (1992). The investment opportunity set and
Strategy and the Environment, 25(3), 165–177. https://doi.org/ corporate financing, dividend, and compensation policies. Journal of
10.1002/bse.1863 Financial Economics, 32(3), 263–292. https://doi.org/10.1016/0304‐
405X(92)90029‐W
Lang, L., & Stulz, R. (1994). Tobin's q, corporate diversification, and firm
performance. Journal of Political Economy, 102(6), 1248–1280. https:// Trucost (2009). Carbon risks in UK equity funds. Report commissioned by the
doi.org/10.1086/261970 WWF. July
Lee, J., & Lee, C. (2009). IT governance‐based IT strategy and management: United Nations (2006). United Nations principles for responsible investment.
Literature review and future research directions. In Information technol- London, UK: PRI Association.
ogy governance and service management: Frameworks and adaptations Vandekerckhove, W., Leys, J., & Van Braeckel, D. (2008). A speech‐act
(pp. 44–62). Hershey, PA: IGI Global. model for talking to management. Building a framework for evaluating
Lee, K. W., Lev, B., & Yeo, G. H. H. (2008). Executive pay dispersion, corpo- communication within the SRI engagement process. Journal of Business
rate governance, and firm performance. Review of Quantitative Finance Ethics, 82(1), 77–91. https://doi.org/10.1007/s10551‐007‐9563‐5
and Accounting, 30(3), 315–338. https://doi.org/10.1007/s11156‐ Zhu, H., & Cai, K. (2014). Cultural distance and bond pricing: Evidence in
007‐0053‐8 the Yankee and rule 144A bond markets. Journal of Financial Research,
Liu, Y., Miletkov, M. K., Wei, Z., & Yang, T. (2015). Board independence and 37(3), 357–384. https://doi.org/10.1111/jfir.12040
firm performance in China. Journal of Corporate Finance, 30, 223–244.
https://doi.org/10.1016/j.jcorpfin.2014.12.004
Margolis, J. D., & Walsh, J. P. (2003). Misery loves companies: Rethinking How to cite this article: Yu EP, Guo CQ, Luu BV. Environmen-
social initiatives by business. Administrative Science Quarterly, 48(2), tal, social and governance transparency and firm value. Bus
268–305. https://doi.org/10.2307/3556659 Strat Env. 2018;27:987–1004. https://doi.org/10.1002/
Mattoo, A., Subramanian, A., van der Mensbrugghe, D., & He, J. (2009). bse.2047
Reconciling climate change and trade policy. Center for Global Devel-
opment Working Paper no. 189.
APPENDIX
1004

This table provides a summary of the variables used in this study.

TABLE A1 Correlation matrix

log Operating Return3 Return5 Tobin's Environmental log R&D

Variable Leverage ROA ESG (Size) Insider Institutional Quick Current Women margin average average Q PB log(GDP) Corruption disclosure (board) Independent intensity
Leverage 1
ROA −0.1904 1
ESG 0.0151 −0.0967 1
log(Size) 0.0747 −0.3764 0.4119 1
Insider −0.0389 −0.0011 −0.0827 −0.2051 1
Institutional 0.0003 0.1164 −0.1211 −0.0644 −0.2719 1
Quick −0.2563 0.1291 0.0192 −0.0827 0.0194 0.0679 1
Current −0.3097 0.1250 0.0280 −0.1369 0.0126 0.0930 0.8912 1
Women −0.0247 0.1937 −0.1530 −0.1068 0.0473 0.0886 0.0068 −0.0119 1
Operating 0.2494 0.3543 −0.0087 −0.0192 −0.1010 0.0257 0.1552 0.0892 0.0676 1
margin
Return3 0.1050 0.4829 −0.0369 −0.2003 −0.0259 0.0757 −0.0603 −0.0862 0.1548 0.1635 1
average
Return5 0.1142 0.4738 −0.0316 −0.2063 −0.0186 0.0690 −0.0706 −0.0968 0.1519 0.1521 0.9495 1
average
Tobin's Q −0.1111 0.7972 −0.1469 −0.4293 0.0529 0.0956 0.1656 0.1477 0.2038 0.2382 0.4277 0.4174 1
PB 0.1056 0.4478 −0.0444 −0.2287 0.0154 0.0852 −0.0390 −0.0748 0.1733 0.1002 0.7181 0.6604 0.6043 1
log(GDP) −0.1043 −0.0845 0.0294 0.1799 −0.0143 0.0907 0.1350 0.1211 0.1769 0.0014 0.0356 0.0435 −0.0646 −0.0062 1
Corruption −0.1314 −0.0694 −0.1404 0.0811 −0.0873 0.1185 0.0209 0.0587 0.1610 −0.0552 −0.0016 −0.0004 −0.0352 −0.0275 0.7514 1
Environmental −0.0009 −0.0822 0.9618 0.3963 −0.0672 −0.1571 0.0581 0.0661 −0.1848 −0.0137 −0.0368 −0.0317 −0.1292 −0.0412 0.0361 −0.1250 1
disclosure
log(board) 0.1149 −0.1614 0.3008 0.4310 −0.0662 −0.0175 −0.1219 −0.1472 −0.0837 −0.0768 0.0114 0.0178 −0.1712 −0.0560 −0.0231 −0.1143 0.2674 1
Independent 0.0464 0.0014 0.0365 0.2345 −0.2199 0.3464 0.0302 0.0252 0.2412 0.1723 −0.0399 −0.0481 −0.0149 −0.0318 0.3015 0.3409 −0.0153 −0.0999 1
R&D intensity −0.1886 0.0712 0.0928 0.0565 0.0012 0.0970 0.3994 0.3373 0.0765 0.0480 −0.0424 −0.0434 0.1872 0.0242 0.1767 0.1122 0.1264 −0.0071 0.1349 1
YU
ET AL.