Plum Lee 2015

J. Account.
Public Policy xxx (2015) xxx–xxx
Contents lists available at ScienceDirect
J. Account. Public Policy

journal homepage: www.elsevier.com/locate/jaccpubpol
Voluntary environmental disclosure quality

and firm value: Further evidence
Marlene Plumlee a,⇑, Darrell Brown b, Rachel M. Hayes a, R. Scott Marshall b
a
University of Utah, United States
b
School of Business Administration, Portland State University, United States
a b s t r a c t
This study reexamines the relationship between the quality of a

firm’s voluntary environmental disclosures and firm value by
exploring the relationship between the components of firm value
(expected future cash flows and cost of equity) and voluntary
environmental disclosure quality. We measure voluntary environ-
mental disclosure quality using a disclosure index consistent with
the Global Reporting Initiative (GRI, 2006) disclosure framework
for a sample of US firms across five industries. In addition to overall
disclosure quality, we consider the type (i.e., hard/soft) and the
nature (i.e., positive/neutral/negative) of the disclosure in our anal-
ysis. Our analyses provide evidence that voluntary environmental
quality is associated with firm value through both the cash flow
and the cost of equity components, consistent with our expecta-
tions. More importantly, however, we demonstrate that both the
type and nature of the environmental disclosures is informative
in establishing the predicted relations. Thus, in addition to
providing evidence on the association between voluntary disclo-
sure quality and firm value, our results highlight the benefit of
parsing broader measures (e.g. voluntary environmental disclosure
quality) when examining complex relationships.
Ó 2015 Elsevier Inc. All rights reserved.
1. Introduction
In this study we reexamine the relation between voluntary environmental disclosure quality
(VEDQ) and firm value by exploring the association between the disclosure quality and the two com-
ponents of firm value – expected future cash flow and cost of equity capital (COEC). Our study builds
⇑ Corresponding author.
http://dx.doi.org/10.1016/j.jaccpubpol.2015.04.004
0278-4254/Ó 2015 Elsevier Inc. All rights reserved.
Please cite this article in press as: Plumlee, M., et al. Voluntary environmental disclosure quality and firm value:
Further evidence. J. Account. Public Policy (2015), http://dx.doi.org/10.1016/j.jaccpubpol.2015.04.004
2 M. Plumlee et al. / J. Account. Public Policy xxx (2015) xxx–xxx
on prior and concurrent research that presents inconsistent results surrounding the relation between
corporate social responsibility (CSR) and/or environmental disclosure and the COEC. For example,
Richardson and Welker (2001) document a significant positive association between CSR and the
COEC for a sample of Canadian firms during the early 1990s, inconsistent with theory. In contrast,
Dhaliwal et al. (2011) find that firms with CSR performance superior to that of their industry peers
enjoy a reduction in the COEC upon the initiation of standalone CSR reports, while Clarkson et al.
(2013) fail to document a significant relation between VEDQ and COEC but find that firms with higher
VEDQ report higher future return on assets. In light of these diverse results, we reexamine the relation
between VEDQ and firm value by examining the association between VEDQ and both the numerator
(expected future cash flows) and denominator (COEC) of firm value.
Our primary contributions to the literature lie in refining the measures employed in the analyses, in
terms of both the independent measures (i.e., VEDQ) and the dependent measures. Most of the prior
studies in this area use relatively crude measures of VEDQ. For example, Dhaliwal et al. (2011) use the
initiation of CSR reports, an indicator variable, in their analysis. Clarkson et al. (2013) rely on a single
value obtained from summing across a completed environmental disclosure index in their primary
regression, which implicitly presumes that each disclosed item measured has the same impact as
all other items.1 If this presumption is not descriptive, then the summed measure may not capture
the relevant cross-sectional variation. We consider this in our analysis by parsing the data from our
disclosure index (which is similar to the Clarkson et al., 2013 index) in the following manner. First,
consistent with Clarkson et al. (2013), we classify the items in the voluntary environmental disclosure
index into those that are related to hard/objective and soft/subjective disclosures.2 We also consider
the ‘‘nature’’ (positive, negative, or neutral3) of the items in the environmental disclosure index, similar
to Hutton et al. (2003), who document systematic differences in managers’ voluntary disclosures around
good and bad news earnings. We contend that decomposing total VEDQ into these more precise
measures allows us to capture variation in aspects of the disclosures expected to differentially affect firm
value.
In addition to providing finer measures of VEDQ, our analyses employ measures of environmental
performance that consider both positive (e.g., recycling) and negative (e.g. pollution) environmental
performance, rather than relying on a single negative aspect of environmental performance as in some
studies (e.g. Al-Tuwaijri et al., 2004; Clarkson et al., 2013).4 We provide analyses based on both a ‘‘net’’
measure of environmental performance (that captures positive and negative environmental performance
in a single proxy) and positive and negative environmental performance as two independent variables.
Doing so allows us to consider and control for a broader range of environmental performance.
We also decompose firm value into its COEC and cash flow components. We use two implied COEC
proxies that rely on stock price and expectations of future cash flows, consistent with a broad stream
of research that examines disclosure related issues using implied COEC (e.g. Clarkson et al., 2013;
Dhaliwal et al., 2011). Botosan et al. (2011) examine the construct validity of the various implied
COEC proxies available and provide strong support for the target price and PEG methods, with partic-
ular support for the target price method. Thus, we use the COEC based on the target price method for
our primary analysis. We also include the PEG method COEC in order to provide a comparison to con-
current studies that employ this measure (e.g. Clarkson et al., 2013). Finally, we explicitly include the
cash flow effects of disclosure in our analysis, which is generally not considered in prior research. We
estimate the expected cash flow component of firm value, again relying on analysts’ forecasts of future
expected cash flows and firm-specific COEC.
1
In supplementary analyses, the authors disaggregate their VEDQ measure, although doing so does not change the tenor of their
results.
2
An example of a hard/objective item would be item 1A in the index (Appendix A), which would have a ‘‘1’’ if a firm provides a
current period absolute amount of materials input into the production process and ‘‘0’’ otherwise. A soft/subjective item would be
item 36A, which would have a ‘‘1’’ if a firm indicates that Life Cycle Analysis (LCA) was identified as a corporate tool and ‘‘0’’
otherwise.
3
Examples of positive/negative/neutral items would include: positive = item 2A (materials input into the production process
from internally or externally supplied recycled materials); negative = item 10A (emission of green house gases); and neutral = item
1A (current period absolute amount of materials input into the production process).
4
We provide additional details about these data in the paper.
M. Plumlee et al. / J. Account. Public Policy xxx (2015) xxx–xxx 3
Our sample includes United States firms drawn from five industries – some considered sensitive to
environmental issues and some that are not. Much of the prior research focuses on sensitive industries
alone, although, as evidenced in our study, a significant proportion of firms within non-sensitive
industries provide voluntarily environmental disclosures. We measure VEDQ using the Plumlee
et al. (2009) (BMP) index, based on firm-specific voluntary environmental disclosures made within
stand-alone or annual reports over a six-year period (2000–2005).5 Based on these data, we construct
four sets of VEDQ measures that capture increasingly fine aspects of the disclosures. The first measure
captures overall variation in the VEDQ.6 Our second (third) measures are based on classifying the
VEDQ index items into hard or soft (positive, neutral, or negative) categories, based on whether the item
within the index captures hard/soft (positive/neutral/negative) aspects of disclosure. We contend that
hard disclosures are frequently viewed as more informative and credible than soft disclosures because
they are more objective, more verifiable at some later stage, and provide more precise data (Cormier
et al., 2011; Hutton et al., 2003). Parsing the data based on the nature of the disclosures is consistent with
prior work showing systematic differences in managers’ voluntary disclosures around good and bad
news earnings wherein managers provide ‘‘soft’’ disclosures in about equal measure for bad news fore-
casts, but provide significantly more ‘‘hard’’ disclosures for good news forecasts (Hutton et al., 2003).
We begin by establishing a positive relation between VEDQ and firm value after controlling for
environmental performance, consistent with Clarkson et al. (2013).7 We then explore whether parti-
tioning VEDQ by the disclosure type (e.g., Clarkson et al., 2008) and the disclosure nature (whether the
hard/soft disclosure is related to positive/neutral/negative environmental issues) improves our ability
to discern a relation with the components of firm value. Our results suggest that partitioning the disclo-
sures increases our ability to detect the associations between VEDQ and firm value, by allowing the asso-
ciations to differ across firm value components (e.g., cash flow and COEC) and across variation in VEDQ
for hard/soft and positive/neutral/negative environmental issues. In addition, unlike most prior studies,
we document a link between some components of VEDQ and the COEC, as predicted by theory. Our find-
ings complement and extend the work of Clarkson et al. (2013), who find environmental disclosures to
be incrementally informative over an alternative measure for environmental performance (toxic release
inventory (TRI) data) in predicting firm profitability and firm value. In contrast to our study, however,
they do not find an association between VEDQ and COEC. We argue that these differences are likely
due to our refined VEDQ measures, our use of the target price method of estimating COEC, and different
samples (they limit their analyses to environmentally sensitive industries in two years).
The remainder of the paper is organized as follows. In Section 2 we review the relevant literature
and develop hypotheses. Section 3 provides details of research design, variable measurement, and
sample selection, and presents descriptive statistics of the sample. We present our empirical results
in Section 4 and conclude in Section 5.
2. Literature review and development of hypotheses
2.1. Literature review
Clarkson et al. (2008) classify prior environmental accounting research into three broad categories:
studies that examine the valuation relevance of corporate environmental performance information
(environmental disclosures), studies that examine managerial decisions to disclose potential environ-
mental liabilities, and studies that explore the relation between environmental disclosures and envi-
ronmental performance. Our study contributes most directly to the first category, although we use
findings from the other areas to motivate our analyses. We provide a review of relevant environmental
accounting research in Section 2.1.1, with an emphasis on the first category as identified by Clarkson
et al. We also contribute to the literature that considers the relation between environmental perfor-
mance and COEC. We provide a review of the most relevant literature in that area in Section 2.1.2.
5
The index is provided in an Appendix A.
6
This measure is most similar to the measure employed in Clarkson et al. (2008, 2013).
7
In an experimental setting, Elliott et al. (2014) document that investors’ estimates of firm value are associated with CSR
performance. This evidence, along with evidence of an association between environmental disclosure and environmental
performance, suggests that controlling for environmental performance is essential in our model.
2.1.1. Firm value and environmental disclosures

Early studies examining the link between environmental disclosures and firm value frequently
focus on the associations between specific environmental issues or events and stock price or stock
price changes. For example, Barth and McNichols (1994) document that the market assessed
Superfund liabilities in excess of the amounts disclosed by firms, consistent with these firms having
‘unrealized’ environmental liabilities. Blacconiere and Patten (1994) and Blacconiere and Northcutt
(1997) provide evidence of the benefits of improved environmental disclosures. Specifically,
Blacconiere and Patten (1994) document that, while chemical companies experienced negative share
price returns after a significant chemical leak (the Union Carbide Bhopal leak), the stock price reaction
was mitigated for firms with better environmental disclosures, and Blacconiere and Northcutt (1997)
find that chemical firms with more extensive environmental disclosures included in their 10-K reports
had a weaker negative reaction to environmental regulation than other firms. These studies provide
evidence of the link between environmental disclosures and firm value, although the focus is on
mandatory disclosures and environmental ‘‘events’’ or liabilities and there are no controls for environ-
mental performance.
In a study that considers disclosures absent specific events or liabilities, Richardson and Welker
(2001) examine the relation between social disclosures in annual reports (which include environmen-
tal disclosures) and the COEC for a sample of Canadian firms. They document an unexpected positive
relation between social disclosures and the COEC and explore whether that relation is due to ‘‘biases in
social disclosures’’.8 While their findings suggest that improved social disclosures increase COEC (reduce
firm value), the authors state this does not ‘‘imply that social disclosure has an overall negative effect on
the firm’’ (pg. 614) and suggest that further research consider other beneficial effects of social
disclosures.
In contrast to Richardson and Welker (2001), Aerts et al. (2008), and Cormier and Magnan (2007)
provide evidence that improved environmental disclosure is associated with a lower COEC, in some
settings. Aerts et al. (2008) show that enhanced environmental disclosures provided by European
and North American companies in 2002 are associated with more precise analysts’ earnings forecasts,
which might reduce COEC through a reduction in information risk. The authors consider mandatory
and voluntary, print and web-based environmental disclosures and find that the associations vary
by industry (weaker for environmentally sensitive industries), country (stronger for European than
North American companies), and disclosure venue (stronger for print for North American companies
and for web-based disclosures for European companies). Cormier and Magnan (2007) investigate how
environmental reporting affects the association between firm earnings and stock market value, using
environmental disclosures from Canada, France, and Germany from 1992–1998. Their findings docu-
ment an association for German firms, although they fail to find a consistent relation between the dis-
closures and stock valuation multiples for French and Canadian firms. While these studies provide a
link between environmental disclosures and firm value, unlike our study, they include both manda-
tory and voluntary disclosures.
Recent work more directly related to this study includes papers by Dhaliwal et al. (2011), Griffin
and Sun (2012), Clarkson et al. (2013), and Matsumura et al. (2014). For example, Griffin and Sun
use an event study to examine shareholders’ response to firms’ voluntary disclosures about green-
house gas emissions. They find that shareholders respond positively to the disclosures, and that the
responses are more positive for smaller companies with limited public information availability. In
addition, Matsumura et al. document that firm value is associated with carbon emissions and whether
firms voluntarily disclose that information. These findings suggest that investors view environmental
disclosures as relevant for firm value.
Dhaliwal et al. (2011) examine whether the initiation of voluntary disclosure of corporate social
responsibility9 (CSR) activities via a stand-alone CSR report is associated with a firm’s COEC. They find
that firms with a high COEC are more likely to release a stand-alone CSR report. Further, the COEC
decreases for firms that initiate reports with CSR performance above the industry mean. Similar to
8
The authors employ a proxy for COEC (the industry method) based on a method Botosan and Plumlee (2005) and Botosan et al.
(2011) suggest fails to capture variation in underlying firm risk. We refer the interested reader to those studies for details.
9
Environmental disclosures are a subset of CSR disclosures.
our study, Dhaliwal et al. examine the relation between voluntary disclosures and COEC, although their
focus is on how the issuance of a CSR is associated with a firm’s COEC, while we consider the association
between the type and nature of the disclosures within a firm’s environmental disclosures and both its
COEC and expected future cash flows.10
Clarkson et al. (2013) examine the impact of voluntary environmental disclosure on the COEC and
firm value. They capture variation in voluntary environmental disclosures presented via CSRs and cor-
porate web sites for firms within five environmentally sensitive industries and document a positive
association between disclosure quality and overall firm value, even after controlling for negative envi-
ronmental performance. When they examine whether this association is through the COEC, however,
they fail to document a significant relation. To explore whether the positive relation between volun-
tary environmental disclosure and firm value is due to a cash flow component, Clarkson et al. (2013)
regress future realized profitability (ROA) on their disclosure measures. In this setting, they document
a positive relation. This study is similar to ours in its direct examination of the associations between
VEDQ and both aspects of firm value. There are several significant differences, however. For example,
we consider firms that operate in both environmentally sensitive and non-sensitive industries over a
six-year period while Clarkson et al. (2013) focus on firms within environmentally sensitive industries
over two sample years. We also capture variation in both the type and nature of VEDQ and are able to
demonstrate that this partitioning and the use of a superior construct for COEC provide different
results. Finally, Clarkson et al. (2013) do not directly examine the association between VEDQ and
the expected cash flow component of firm value. Instead, they provide evidence that firms with higher
quality disclosures generate higher future ROA than their competitors.
In sum, prior research provides mixed results about the association between environmental disclo-
sures and firm value. Much of the focus of this research is on the denominator of firm value (COEC).
The findings in these studies suggest that the association varies depending on the type of disclosures
(voluntary or mandatory/ environmental or social), the disclosure venue, or the industry. The cash
flow component of firm value is examined less frequently, and these analyses are often related to
potential environmental liabilities.
2.1.2. Environmental disclosures and environmental performance

A second area of research related to our study is the link between environmental disclosures and
environmental performance (e.g. Al-Tuwaijri et al., 2004; Clarkson et al., 2008; Cormier et al., 2009,
2011; Patten, 2002). The findings in this area are also mixed. Al-Tuwaijri et al. (2004) limit their anal-
ysis to environmental disclosures related to specific pollution measures and occurrences (mandatory
disclosures) and document a positive association between these disclosures and environmental per-
formance. In contrast, Clarkson et al. (2008) examine the relation between environmental perfor-
mance and voluntary environmental disclosures and test competing economics-based and
socio-political theories of voluntary disclosure.11 Their findings suggest that environmental perfor-
mance and VEDQ are positively related, consistent with economics-based theories. However, they also
‘‘show that socio-political theories explain patterns in the data (‘‘legitimization’’) that cannot be
explained by economic disclosure theories’’ (pg. 303). In contrast, Cormier et al. (2011) and Cho and
Patten (2007) document that high polluting firms or poorer environmental performers disclose more
than low polluting/ high performers, consistent with socio-political theories.
10
In a footnote, Dhaliwal et al. (2011) discuss specifications that include indicator variables that proxy for firms’ effort and
commitment to better CSR disclosure – the length of the CSR report relative to industry average and whether the report is assured
by a third party. Conditional on the initiation of CSR disclosures, the length and assurance variables further reduce the COEC. Like
our type and nature measures, length and assurance capture elements of the quality of the disclosures (conditional on initiation of
a stand-alone CSR).
11
Economics-based theories predict that voluntary environmental disclosure will be used by ‘‘high-quality’’ companies to
differentiate themselves from ‘‘low-quality’’ companies to avoid an adverse selection problem (e.g. Wagenhofer, 1990). In this
setting, superior environmental performers have incentives to voluntarily disclose to inform investors of their quality, using
disclosures that cannot be easily mimicked by poorer environmental performers. This leads to a positive association between
environmental performance and VEDQ. In contrast, socio-political theories predict a negative association between environmental
performance and VEDQ, wherein poorer environmental performers have incentives to increase VEDQ to increase their
‘‘legitimacy’’. In this setting, firms have incentives to improve the perceptions that relevant stakeholders have about the firms’
environmental standings and do so by providing higher quality voluntary environmental to manage those perceptions (see Aerts
et al., 2008; Cho and Patten, 2007 for more detailed discussions).
In sum, the conclusions in this area are mixed. While studies consistently document a relation
between environmental performance and VEDQ, the sign of the relation might vary by
industry-type (e.g., Cho and Patten, 2007), the nature of the disclosure (Cormier et al., 2009;
Clarkson et al., 2008), and/or the venue within which the disclosure is provided (Aerts and Cormier,
2009; Cormier et al., 2009). Thus, we control for environmental performance and venue in our anal-
ysis. Additionally, we consider the type (e.g., hard or soft) and the nature (e.g., positive, negative, or
neutral) of those voluntary disclosures in our analysis.
2.2. Hypotheses development
Our goal is to gain a better understanding of how differences in the quality of firms’ voluntary envi-
ronmental disclosures relate to stock price. In our hypotheses development, we consider the two com-
ponents of firm value – the numerator (expected future cash flows) and denominator (COEC). We
consider the two components of firm value separately, as theories related to VEDQ and firm value
are based on these two components, which together yield firm value.
2.2.1. Hypotheses – Expected future cash flows

The first means by which VEDQ might be associated with firm value is through a numerator effect.
Some prior research suggests that environmental disclosure quality serves as a signal of a firm’s environ-
mental practices, which affect financial performance and, ultimately, firm value (e.g., Al-Tuwaijri et al.,
2004). These disclosures provide information about practices related to protecting the environment that
could reduce government regulation and the resulting compliance costs, potential litigation, and/or pol-
lution remediation costs. Furthermore, prior research suggests that disclosures related to these issues
inform stakeholders (beyond investors) who opt to partner with, patronize, or work for environmentally
responsible firms, which leads to superior sales and financial performance (Lev et al., 2010). These studies
suggest a positive relation between VEDQ and expected future cash flows. In contrast, a number of stud-
ies suggest that environmental disclosures are used as a means to gain legitimacy, where disclosures are
provided as a means of improving stakeholders’ perceptions of a firm’s environmental standing (e.g., Cho
and Patten, 2007). These studies question the credibility of those disclosures in terms of informing stake-
holders about a firm’s true environmental performance. If this is descriptive of the voluntary disclosures,
we would expect no association between voluntary disclosure quality and future expected cash flows,
after controlling for environmental performance. This leads to our first hypothesis.
Hypothesis 1. VEDQ is related to firm value through an association with the cash flow component
(expected future cash flows).
Research that examines environmental disclosures documents systematic differences across those
disclosures, based on the disclosure venue, industry, and environmental performance. Thus, in our
research design we include controls for these factors. Furthermore, prior research suggests that the
type of the disclosure, frequently classified as hard or soft (Campbell et al., 2003; Cho and Patten,
2007; Clarkson et al., 2008, 2013), will be differentially informative and/or have differential impacts
(Cormier et al., 2009; Hutton et al., 2003). These studies contend that hard disclosures are more infor-
mative and credible than soft disclosures because they are more objective, more verifiable at some
later stage, and provide more precise data. On the other hand, given the imprecise nature of voluntary
environmental information, soft disclosures might provide the most effective means of providing rel-
evant information. Based on these findings, we predict that disclosure type – hard or soft – affects the
association between VEDQ and expected future cash flows, although we do not predict a sign.
Hypothesis 1A. The association between VEDQ and expected future cash flows differs by disclosure type.
We also consider the underlying ‘‘nature’’ of the voluntary disclosures in our analysis. Clarkson
et al. (2008, 2013) provide evidence that hard and soft disclosures vary systematically with environ-
mental performance and across their disclosure categories (e.g. credibility and environmental spend-
ing). Richardson and Welker (2001) argue that there might be a positive association between CSR and
COEC, predicated on firms with greater environmental risk or poorer environmental performance
consistently providing more environmental disclosure to legitimize their existence (Aerts and
Cormier, 2009; Cho and Patten, 2007). Furthermore, Hutton et al. (2003) document systematic differ-
ences in managers’ voluntary disclosures conditional on whether those disclosures are around good or
bad earnings news. Specifically, they find that managers provide soft disclosures in about equal
measure for bad news forecasts, but provide significantly more hard disclosures for good news fore-
casts. In our setting we are unable to classify whether the disclosure is about good/bad news, so we
classify the voluntary disclosures into those that are related to positive/neutral/negative environmen-
tal issues. We label this categorization as the nature of the information.
Hypothesis 1B. The association between VEDQ and expected future cash flows differs by the
disclosure nature.
Finally, we use both the type (hard/soft) and nature (positive/neutral/negative) classifications to
further refine our VEDQ. Given the findings in Hutton et al. (2003), we predict that the association
between VEDQ and expected future cash flows is a function of the type_nature of those disclosures.
This leads to our next hypothesis.
Hypothesis 1C. The association between VEDQ and expected future cash flows differs by the
disclosure type_nature.
2.2.2. Hypotheses development – COEC

The second means by which VEDQ might be associated with firm value is through the denominator
of the firm value. Prior theoretical and empirical disclosure research provides considerable support for
a negative association between disclosure quality and the COEC. Traditional economic theory predicts
that increased voluntary disclosures will be associated with a decrease in the COEC through a reduc-
tion in information asymmetry or estimation risk (Barry and Brown, 1985; Coles et al., 1995),
increased awareness of a firm’s existence and enlarging the investor base (Merton, 1987), or increased
liquidity (Amihud and Mendelson, 1986). Empirical research provides substantial support for this rela-
tion (e.g. Botosan and Plumlee, 2002; Cormier et al., 2009; Leuz and Verrecchia, 2000), although much
of this evidence is associated with financial disclosures.
In contrast, as discussed earlier, Richardson and Welker (2001) argue that there might be a positive
association between VEDQ and COEC due to a consistent bias in the type of firm that provides higher
quality voluntary disclosures. In this setting, higher quality disclosures might be associated with an
increased COEC. There is some empirical support for this relation (e.g., Richardson and Welker,
2001; Botosan and Plumlee, 2002). This leads to the following hypothesis:
Hypothesis 2. VEDQ is related to firm value through an association with the COEC.
As discussed earlier, prior research documents systematic differences across environmental disclo-
sures, based on the venue employed, industry, and environmental performance, which we incorporate
in our research design. We also consider the potential differential impact of disclosure type in our pre-
dicted association with COEC. If investors view one type of disclosure as differentially informative and
credible (e.g., Clarkson et al., 2008, 2013), we predict that the association between COEC and VEDQ
will differ across disclosure type, as stated below.
Hypothesis 2A. The association between VEDQ and COEC differs by disclosure type.
Finally, prior research that examines the association between VEDQ and COEC does not consider
the nature of the disclosures, since all voluntary disclosures are considered equally informative and
credible. Given the multiple roles suggested for VEDQ and concerns with selection bias regarding
the provision of these disclosures, however, we argue that the predicted links between disclosure
quality and COEC differ by the nature and the type_nature of the disclosures. This leads to our final
hypotheses.
Hypothesis 2B. The association between VEDQ and COEC differs by the nature of the disclosures.
Hypothesis 2C. The association between VEDQ and COEC differs by the type_nature of the disclosures.
3. Research design, variable measurement, and sample selection
3.1. Research design
To test our hypotheses, we employ three sets of regression models with firm value,12 expected
future cash flows, and COEC as the dependent variables and the increasingly refined proxies for VEDQ
as the independent variables. For each of the models we include a relevant set of control variables (based
on the dependent measure) along with our variables of interest.
3.1.1. Regression model – Firm value

Our first model focuses on the association between VEDQ and firm value. We employ a modified
Ohlson (1995) model found in the prior literature (e.g., Clarkson et al., 2004, 2008, 2013) to examine
this relation:
P ¼ a þ b1 BV þ b2 AE þ b3 MFInd þ b4 MFDisc þ b5 NetEP þ b6 CER þ b7 R DScoren þ e ð1Þ
P is the stock price as of the date COEC is estimated. Consistent with the prior literature, we include
two model-specific control variables: BV (book value per share at the beginning of the estimation
quarter) and AE (abnormal earnings per share). AE is calculated as the difference between actual earn-
ings per share at the end of fiscal period less the required rate of return on beginning BV. We expect
positive coefficients on these variables.
We include controls for voluntary disclosure and environmental performance in all our regression
models. To control for voluntary disclosure, we include two related proxies drawn from prior research.
MFInd equals one if a firm issued management guidance during each of the years t 1, t, and t + 1, and
zero otherwise (Clarkson et al., 2013) and controls for firm-level financial disclosure propensity.
MFDisc is the product of Supplier, Frequency, and Precision consistent with Baginski and Rakow
(2012). Supplier equals one if a firm issued at least one annual or quarterly management forecast
(MF) during the year, zero otherwise. Frequency is the sum of the annual and quarterly MF issued
by the firm over the period. Precision is the average precision of those forecasts, where qualitative
(open-ended, range, point) forecasts are assigned a Precision of one (two, three, four). Consistent with
Baginski and Rakow (2012), this variable controls for a firm’s voluntary disclosure policy. Including
both these proxies controls for a firm’s general voluntary disclosure policy and allows us to explore
the incremental impact of voluntary environmental disclosures. We expect MFInd and MFDisc to enter
the COEC models with a negative sign, based on prior research (e.g., Baginski and Rakow, 2012;
Botosan and Plumlee, 2002; Clarkson et al., 2013) but do not provide directional predictions for the
firm value or expected future cash flow regression models. We control for environmental performance
using NetEP, the firm-specific difference between KLD reported strengths (GdEP) and concerns (BdEP)
(described in more detail below). We expect this variable to be positively associated with firm value
and the numerator (expected future cash flow) and to be negatively associated with the denominator
(COEC). We also include GdEP and BdEP in place of NetEP to allow the weights on good and bad envi-
ronmental performance to vary in the regressions.
Our interest lies in understanding the role of voluntary environmental disclosures. Thus, we
include the following variables. To capture whether a firm reports its voluntary environmental disclo-
sures via a standalone report or not, we include CER, which equals one if a firm provides its voluntary
environmental disclosures through a stand-alone report and zero otherwise. Our primary variables of
interest (DScore) are based on a completed voluntary environmental disclosure index, which captures
firm-specific variation in voluntary environmental disclosures.13 We use these data to construct an
increasingly fine set of DScore measures for use in our analysis. Specifics of these variables are discussed
more fully in the ‘‘Variable measurement’’ section below.
12
While we do not present formal hypotheses related to overall firm value, we include firm value as a dependent measure in our
analyses.
13
We use the within-industry/year ranked value of the DScore measures, denoted R_DScore, in our tests.
3.1.2. Regression model – Expected future cash flows

Our second regression model focuses on the relation between the numerator of firm value
(expected future cash flow) and VEDQ. We employ the regression model below, which is similar to
the previous model in terms of the independent variables of interest:
EFCF ¼ v þ g1 CF 1 þ g2 LSale þ g3 MFInd þ g4 MFDisc þ g5 NetEP þ g6 CER þ gn R DScoren þ m ð2Þ
EFCF is the cash flow component of firm value, which we capture using Value Line’s discounted forecasts
of future firm price (target price) (we discuss this measure in more detail below). To isolate the incre-
mental effect of VEDQ, we include forecasted current period cash flows per share (CF_1) and the log of
total sales (LSale) as control variables (Doyle et al., 2003). All other variables are as described above.
3.1.3. Regression – COEC

Our final regression model focuses on the relation between the denominator of firm value (COEC)
and VEDQ. We employ the model detailed below, drawn from prior research (e.g., Botosan and
Plumlee, 2002; Richardson and Welker, 2001; Clarkson et al., 2013):
COEC ¼ / þ c1 BETA þ c2 LSIZE þ c3 LBTM þ c4 LEV þ c5 LTG þ c6 MFInd þ c7 MFDisc
þ c6 NetEP þ c7 CER þ cn R DScoren þ j ð3Þ
COEC is one of two implied cost of equity capital proxies, based on either the target price or the PEG
method. We also include several control variables. Specifically, we include capital market beta (BETA)
calculated from a regression of the monthly returns (with a minimum of 12 out of 60 monthly returns)
and a market index return equal to the value-weighted NYSE/AMEX return with data from CRSP; firm
size (LSIZE) calculated as the log of the firm’s market value of equity as of June 30th of the year COEC is
estimated; book-to-market (LBTM) calculated as the log of BV scaled by P, leverage (LEV) calculated as
long term debt scaled by total common equity, and long term growth (LTG) calculated as change in
long-range forecasted earnings scaled by earnings. The other variables are as previously defined.
3.2. Variable measurement
3.2.1. Dependent variables

We use three dependent variables in our analysis: stock price (P), expected future cash flows (EFCF),
and cost of equity capital (COEC). The calculation of these variables is described below.
P is firm stock price on the Value Line publication date during the third calendar quarter of the year
following the release of the voluntary environmental disclosures.14 We rely on Value Line forecasts made
in the third calendar quarter of the year following the release of the voluntary environmental disclosures
to construct proxies for the two components of firm value (EFCF and COEC). Firm-specific Value Line fore-
casts of future stock price (target price) discounted back to the current period are our explicit measure of
expected future cash flows. In addition to making earnings and other types of short and long-term fore-
casts, Value Line forecasts a ‘‘high’’ and ‘‘low’’ firm stock price 3–5 years in the future (we use the mean of
these values). These forecasts have been used to provide long-range terminal values for use in calculating
the implied COEC (e.g., Botosan and Plumlee, 2002; Francis et al., 2004). As our goal is to empirically sep-
arate firm value into a cash flow (numerator) and COEC (denominator) component, we rely on target
prices to provide our proxy for expected future cash flows and in our calculation of implied COEC.
COEC is the implied cost of equity, calculated using Value Line forecasts of target prices (the target
price method) or based on changes in long-range earnings forecasts scaled by current stock price (the
PEG method). These measures have been widely used in prior research (e.g., Botosan and Plumlee,
2002; Francis et al., 2004). Furthermore, Botosan and Plumlee (2005) and Botosan et al. (2011) com-
pare the validity of various proxies for the expected cost of equity and conclude that the target price
method consistently outperforms the other proxies.15 The target price method (COECTP), which utilizes
14
All analyst forecast data are drawn from Value Line forecasts made in the third calendar quarter of the year after the voluntary
environmental disclosures are made public.
15
For a more detailed discussion of the calculation of these proxies and evidence supporting their construct validity, please see
either of these studies.
the mean Value Line forecasts of target prices as the terminal value along with forecasts of dividend pay-
outs and current stock price to derive an implied COEC, is our primary measure. To provide a comparison
to related research in this area, we also use the PEG method (COECPEG), which is calculated as the square
root of the change in long-range earnings forecasts scaled by current stock price.
3.2.2. Independent variables

DScore. We form a series of related DScore measures to test our hypotheses. These measures cap-
ture variation in VEDQ across firms based on a line-by-line analysis of voluntary environmental dis-
closures. To categorize and collect these data we use the index designed by and employed in
Marshall et al. (2007) (similar to the index used in Clarkson et al., 2008, 2013). The index was based
on the Global Reporting Initiative (GRI, 2006) framework and was designed using the advice of an
industry expert, similar to Clarkson et al. (2008).16 For each firm-year, we complete the index using data
hand-collected from firms’ voluntary environmental disclosures presented within a stand-alone report
(frequently labeled a corporate environmental report or a corporate sustainability report), if the firm
issued one. If not, we complete the index using voluntary environmental disclosures provided within
the firm’s annual report. If a firm does not issue either a stand-alone environmental report or an annual
report, we exclude them from the sample. Two independent coders completed the index for each
firm-year observation; the initial rank correlation between the completed indices for the two coders
was greater than 0.87. The completed indices were compared and differences across the coders were
reconciled by a third, independent coder. The voluntary environmental disclosure measures dis-
cussed below were based on the scores obtained from the completed indices. Given the importance of
industry in environmental issues and to control for potential time-specific factors, we use the
within-industry/year ranked value of each of the DScore measures.
Total Disclosure. TScore is the total score obtained from a completed index for each firm-year.
Consistent with prior literature that relies on disclosure indices to capture disclosure quality, we
employ intra-industry/year ranks of the total score (R_TScore) in our analyses.17
Disclosure Type. To examine the impact of disclosure type in explaining the relation between VEDQ
and firm value, we classify each item in our index as Hard (objective) or Soft (subjective). Within our
index, the vast majority of the items (276 of the 351) were classified as Hard and the remaining as Soft,
consistent in many respects with the Global Reporting Initiative framework and Clarkson et al.
(2008).18 R_HardD and R_SoftD employed in our analysis are the intra-industry/year ranks of the scores
obtained from items included in the index classified as objective or subjective, respectively.
Disclosure Nature. Our next set of voluntary environmental disclosure measures classifies the items
in the disclosure index based on the nature of the disclosure issue. Specifically, we classify the index
items into positive/neutral/negative based on the nature of the environmental issue captured by the
item. Our classification is similar to prior studies that classify disclosures based on the content of the
information (e.g., Hutton et al., 2003; Skinner, 1994). It differs, however, in that those studies typically
classify disclosures as good/neutral/bad based on whether the disclosure suggests increases/no chan-
ge/decreases relative to prior earnings. In our setting, however, it is difficult to determine how the
information disclosed should be classified without a baseline to use as comparison. Thus, we focus
on the general nature of the information instead of the classification of the firm response.19 Within
our index almost half of the items are classified as positive (172), almost one-third as neutral (112),
and 67 items are classified as negative. These are the potential items firms might disclose based on
16
The index includes 62 individual indicators, with two to eight possible aspects of each indicator. For example, the index allows
for differentiation as to whether the disclosed information was quantitative or qualitative, in absolute or relative amounts, etc.
Each indicator and aspect was identified as either present or absent; indicators were not evaluated as being positive or negative. A
more detailed discussion of the index can be found in Marshall et al. (2007) or Plumlee et al. (2009).
17
Consistent with prior research that employs disclosure scores in the analysis (e.g., Lang and Lundholm, 1993, 1996; Bushee and
Noe, 2000; Botosan and Plumlee, 2002; Lundholm and Myers, 2002; Drake et al., 2009; Clarkson et al., 2013), we rank firms
according to their score within each industry-year for use in the regression models.
18
Classification of the index items into Hard/Soft and Positive/Neutral/Negative was done independently by two co-authors.
Differences between the two, which were minimal, were discussed and reconciled.
19
For example, disclosures related to ‘‘Materials input into the production process from internally or externally supplied recycled
materials’’ are classified as Positive; disclosures related to ‘‘Use of water’’ are classified as Neutral; and disclosures related to
‘‘Emission of green house gases’’ are classified as Negative.
GRI standards. R_PosD, R_NeuD, and R_NegD employed in our analysis are the intra-industry/year ranks of
the scores obtained from items included in the index classified as positive, neutral or negative,
respectively.
Disclosure Type_Nature. Our final set of VEDQ measures incorporates both the type and nature clas-
sifications. Within our index, 109 (112, 35, 63, 32) of the 351 items are classified as Hard/Positive
(Hard/Neutral, Hard/Negative, Soft/Positive, Soft/Negative). R_HdPos, R_HdNeu, R_HdNeg, R_SfPos,
and R_SfNeg employed in our analysis are the intra-industry/year ranks of the scores obtained from
items included in the index classified as hard/positive, hard/neutral, hard/negative, soft/positive or
soft/negative, respectively.20
Environmental Performance. We capture variance in firms’ environmental performance using data
from KLD Research and Analytics, Inc. Prior research typically relies on one of two data sources to cap-
ture environmental performance: the TRI data or the KLD data. Several studies (e.g. Al-Tuwaijri et al.,
2004; Clarkson et al., 2004, 2008; Patten, 2002) employ the TRI data, which is available from the US
Environmental Protection Agency’s toxic release inventory database and includes information on toxic
chemical releases and other waste management activities reported annually by manufacturing facility.
Generally, firm-specific proxies are based on summing the toxic releases or toxic waste recycled across
facilities and scaling by a value that varies by firm size. The second data source used to capture envi-
ronmental performance is KLD (e.g., Cho and Patten, 2007; Cho et al., 2010; Dhaliwal et al., 2012).21
These data are based on an extensive assessment by KLD of each company’s environmental management,
planning and impact assessment, utilization of resources, compliance with applicable laws and regula-
tions, and emissions. Based on these assessments, KLD provides indicators of both environmental
strengths and concerns.22 We elect to employ the KLD data over the TRI data to capture environmental
performance for two reasons. First, the KLD data identify positive as well as negative environmental per-
formance, while the TRI data focus on a single aspect of negative environmental performance (pollution)
and are bounded at zero. In addition, the KLD data cover a broader range of environmental issues beyond
pollution, including recycling and climate change. We rely on R_NetEP, the intra-industry/year rank of
the net of the KLD strength and concern environmental performance measures in our analyses, which
allows us to assess a firm’s environmental performance relative to its industry peers. We also estimate
our models with both R_GdEP and R_BdEP, the intra-industry/year ranks of the strength and concern KLD
environmental performance measures, respectively, which allows us to provide evidence on the impact
of including each aspect of environmental performance separately. Table 1 provides a detailed discussion
of the variables included in our study.
3.3. Sample
Our sample includes US listed firms drawn from five industries (oil & gas, chemical, food/beverage,
pharmaceutical, and electric utilities) over a six-year period (2000–2005). Unlike some studies (e.g.,
Clarkson et al., 2008, 2013), we include both environmentally sensitive and non-sensitive industries
in our sample. In addition we do not require that firms have standalone environmental reports nor
do we include web-based items. Prior research documents systematic differences in environmental
disclosure quality across industries and suggests that the role for these disclosures may also vary
by the environmental sensitivity of the industry. Thus, including firms from both sensitive and
non-sensitive industries has the potential to enhance our understanding of the role of voluntary envi-
ronmental disclosures. We limit our analysis to five industries, given the need to hand-collect data.
20
There are only 5 classifications, as there were no neutral disclosures classified as soft.
21
KLD Research and Analytics, Inc. is an independent ratings company that evaluates and makes available the social performance
of over 650 companies every year, comprising mainly all firms in the S&P 500 and Domini 400 Social SMIndex. During 2001–2002,
KLD expanded its coverage to include the largest 1000 U.S. companies by market capitalization. Since 2003, it has covered the
largest 3000 U.S. companies based on market capitalization. One component of the social performance included in the database is
related to environmental performance. This database provides a quantifiable, enhanced, independent ratings system and has been
used as a proxy for environmental and social performance for a large body of research.
22
Strength ratings are related to (1) clean energy; (2) beneficial products and services; (3) pollution prevention; (4) recycling;
and (5) management systems. Concern ratings are related to (1) climate change; (2) ozone depleting or agricultural chemicals; (3)
hazardous waster; (4) regulatory problems; and (5) substantial emissions. (KLD Research and Analytics, Inc., 2003).
We identify firms within our five industries for which we have the information required to form
our proxies for COEC, expected future cash flows, and control variables, after eliminating firms that
report mergers.23 We limit our analysis to the set of firms for which we have the necessary KLD envi-
ronmental performance data. Our final sample includes 474 firm-years over a six-year period.
Table 2 provides descriptive statistics for the entire sample and by industry. Panel A provides
statistics regarding the disclosure measures used in our analyses, while panel B provides details for
the common control variables and the dependent measures used in the analysis. Consistent with prior
research, we document that the use of CERs and the level of voluntary environmental disclosures vary
systematically by industry. Just over a third of our sample firms use a CER to disclose their environ-
mental information; firms within environmentally sensitive industries (e.g., Chem) employ CERs more
frequently (almost 40 percent of the time), while firms within non-sensitive industries (e.g., Food) use
them less often (22 percent of the time). For the full sample the average TScore is 8.74; the mean val-
ues across industries range from a low of 5.45 in the Food industry to a high of 11.74 in the Electric
industry.24,25 A similar pattern holds when the voluntary environmental scores are classified by type,
nature, and by type_nature. Industries classified as environmentally sensitive (Chem, Oil, and Elec) have
higher mean disclosures scores than other industries. In untabulated analysis, we find the mean TScore
and other voluntary environmental scores vary across the sample years, although the relation across
industries is consistent across years – firms in environmentally sensitive industries consistently provide
greater environmental disclosures than firms in non-sensitive industries.
We document that, on average, firms provide more objective than subjective disclosures, for the
full sample (6.3 versus 2.3) and across all industries. Furthermore, we find that on average firms more
frequently provide environmental disclosures related to positive items than neutral or negative items.
However, the average firms in environmentally sensitive industries more frequently provide negative
than neutral disclosures while the opposite is true for firms in non-sensitive industries.
Table 2 panel B documents that environmental performance varies by industry, although not
strictly by environmental sensitivity. The mean environmental performance for firms in the Oil and
Elec industries is the lowest across the sample (1.4 and 1.6, respectively) and the mean environ-
mental performance for firms in the Chem and Food industries is virtually the same (0.62 and
0.70, respectively). It is interesting to note that the average (median) net environmental performance
for the full sample is negative and is either negative or zero for each industry. Our control for overall
disclosure quality, MFInd, varies across industries as well, with firms within the Oil and Elec industries
most frequently issuing forecasts, while firms in Chem and Pharm do so much less frequently.
Table 3 presents Spearman correlations for the primary variables included in our study. The first
eleven variables relate directly to environmental disclosures. Nine of them are directly inter-related
(e.g., R_HardD and R_SoftD are a result of systematically partitioning R_TScore), which is reflected in
23
The set of firms included in our analysis is limited to firms that met our criteria during the first collection period (2000–2002).
We updated our sample by collecting additional data for the set of firms included in our initial sample.
24
As noted earlier, to address concerns with the inherent subjectivity of coding these disclosures we had two independent coders
for all our firms. In addition, we compare the raw scores obtained from our disclosure index with analogous scores presented in
Clarkson et al. (2008, 2013). Specifically, we compared the Clarkson et al. (2013) data for the first year in their sample (2003) with
our data for the same year for the three environmentally sensitive industries. Clarkson et al. document that the mean/median total
score for their sample firms is 14.32/10.00; for that sample year we report mean/median total scores of 17.1/9.5 (untabulated).
When we compare hard/soft scores, Clarkson et al. report mean values of 9.68/4.4 while we obtain values of 7.4/2.6. These values
appear consistent with the scores obtained by Clarkson et al. In addition, we calculated the correlation between the scores
produced using our measure and the scores produced by the Clarkson et al. (2008) index for the year 2003 for the overlapping set
of firms. The correlation between their raw scores and ours is 0.68; the correlation between their ‘‘hard’’ disclosure items and ours
is 0.68. These values are based on the same data, using two similar indices and means of classifying hard/soft items. Clarkson et al.
(2013) do not classify disclosures into positive/neutral/negative, so we are unable to compare those values. We contend that the
high correlation provides support for the validity of our measure. We thank Peter Clarkson and Xiaohua Feng for kindly sharing
their data with us.
25
The maximum disclosure score obtained by any firm during our sample period is 108. Our index has 351 available points to
allow us to capture a variety of disclosures from one sustainability report to the next, and across industries (as industries tend to
provide disclosures across specific areas). As noted by Clarkson et al. (2008), who report a maximum score of 68 out of a possible
95 in their sample, ‘‘the attainable score is lower than this (the total number of available points), even for an excellent
sustainability report’’ due to the index design. Nonetheless, we acknowledge that our low average scores suggest room for
improvement in sustainability reporting, consistent with other studies.
Table 1
Variable descriptions.
Price= Firm stock price on the publication date of the Value Line forecasts
COEC= Firm specific COEC calculated as (1) the internal rate of return that equates current stock price with the Value
Line target price (5 years in the future) and Value Line forecasted dividends (TP method) or as the square root
of the difference between analysts forecast of 5 year ahead and 4 year ahead earnings, scaled by stock price
(PEG method)
EFCF= Value Line forecasted target firm price, calculated as the mean of the forecasted high and low stock price 3–
5 years in the future discounted back to the current period using the firm-specific COEC
CER= 1 if a firm issues a stand-alone corporate environmental report; 0 otherwise
R_TScore= The intra-industry/year rank of the annual score from a completed VEDQ index
R_HardD= The intra-industry/year rank of the annual score from a completed VEDQ index classified as hard
R_SoftD= The intra-industry/year rank of the annual score from a completed VEDQ index classified as soft
R_PosD= The intra-industry/year rank of the annual score from a completed VEDQ index classified as positive nature
R_NeuD= The intra-industry/year rank of the annual score from a completed VEDQ index classified as neutral nature
R_NegD= The intra-industry/year rank of the annual score from a completed VEDQ index classified as negative nature
R_HdPos= The intra-industry/year rank of the annual score from a completed VEDQ index classified as hard and positive
R_HdNeg= The intra-industry/year rank of the annual score from a completed VEDQ index classified as hard and negative
R_SfPos= The intra-industry/year rank of the annual score from a completed VEDQ index classified as soft and positive
R_SfNeg= The intra-industry/year rank of the annual score from a completed VEDQ index classified as soft and negative
NetEP= The net of environmental performance strengths and concerns (both from KLD)
GdEP= The sum of environmental performance strengths (from KLD)
BdEP= The sum of environmental performance concerns (from KLD)
BV= Book value per share of total common equity
AE= Abnormal earnings to common equity, defined as forecasted earnings to common equity less COEC times
beginning-of-period book value of common equity
MFInd= 1 if a firm issued management guidance during each of the years t 1, t and t + 1, and 0 otherwise
MFDisc= The product of Supplier, Frequency, and Precision consistent with Baginski and Rakow (2012). Supplier equals
one if a firm issued at least one annual or quarterly management forecast during the year, zero otherwise.
Frequency is the sum of the annual and quarterly MF issued by the firm over the period. Precision is the
average precision of those forecasts, where qualitative (open-ended, range, point) are assigned a Precision of
one (two, three, four)
LSize= Log of market value
Beta= Capital market beta estimated using the market model with a minimum of 12 out of 60 monthly returns and
a market index return equal to the value-weighted NYSE/AMEX return with data from CRSP
LBTM= Log of BV scaled by stock price
CF_1= Value Line forecasted current period firm-specific cash flows
LSale= log of total sales
LEV= Long term debt scaled by total common equity
LTG= Change in long-range earnings forecasts scaled by beginning earnings
the high positive correlations across those variables. It is interesting to note, however, that even given
these high correlations, the correlations between the disclosure quality variables and other variables
differ. For example, R_TScore is highly correlated with R_HardD (q = 0.975), which is in turn highly
correlated with R_HdPos and R_HdNeg (q = 0.879 and 0.924, respectively). Even so, we find the
correlations between these variables and EFCF to be different from each other: the correlation between
R_TScore and EFCF is 0.074, while the correlations between R_HdPos and EFCF is 0.08, between
R_HdNeu and EFCF is 0.125 and between R_HdNeg and EFCF is insignificant. We also document that
the correlations between the disclosure variables and our two COEC proxies differ: the measures
are not significantly related to the COECTP, while almost all are negatively correlated with COECPEG.
The correlation between the two cost of equity measures (COECTP and COECPEG) is high by traditional
standards (q = 0.549), although it appears that the shared variance across these measures is not
common to our disclosure measures. Environmental performance is correlated with all of the environ-
mental disclosure variables; good and bad performance (GdEP and BdEP) are both positively correlated
with disclosure, while net environmental performance (NetEP) is negatively correlated with all the
measures of environmental disclosure. NetEP is also negatively correlated with one of the disclosure
control variables, MFInd. Interestingly, neither of the disclosure control variables (MFInd and
MFDisc) is significantly correlated with any of the environmental disclosure variables. Note that these
correlations are based on the sample as a whole, while we employ intra-industry/year ranks in our
analyses.
Table 2
Descriptive statistics.
Full sample Chem Elec Oil Food Pharm

Panel A: Environmental disclosures
N 474 79 142 88 88 77
CER Mean 0.34 0.38 0.39 0.36 0.22 0.31
TScore
Mean 8.74 9.44 11.74 7.59 5.45 7.53
50th 2.00 4.00 6.5 3.00 0.00 0.00
25th 0.00 0.00 0.00 0.00 0.00 0.00
75th 11.00 13.00 17.00 11.00 4.00 8.00
SD 14.12 14.64 15.66 10.73 11.76 15.56
HardD
Mean 6.39 7.04 8.70 5.49 3.51 5.77
50th 2.00 3.00 4.00 2.50 0.00 0.00
25th 0.00 0.00 0.00 0.00 0.00 0.00
75th 8.00 11.00 10.00 9.00 3.00 4.00
SD 10.67 11.19 12.20 7.48 8.13 11.98
SoftD
Mean 2.35 2.41 3.04 2.10 1.94 1.77
50th 0.00 1.00 1.00 0.00 0.00 0.00
25th 0.00 0.00 0.00 0.00 0.00 0.00
75th 3.00 2.00 4.00 2.00 2.00 1.00
SD 4.29 4.32 4.57 3.58 3.99 4.71
Positive
Mean 4.07 4.15 5.70 3.31 3.03 3.01
50th 0.50 1.00 2.00 1.00 0.00 0.00
25th 0.00 0.00 0.00 0.00 0.00 0.00
75th 5.00 6.00 9.00 4.00 3.00 3.00
SD 6.77 6.53 7.51 4.95 6.81 6.88
Neutral
Mean 1.80 2.24 1.84 0.92 1.49 2.61
50th 0.00 0.00 0.00 0.00 0.00 0.00
25th 0.00 0.00 0.00 0.00 0.00 0.00
75th 1.00 4.00 2.00 0.00 0.00 0.00
SD 4.25 4.39 3.92 2.42 3.88 6.19
Negative
Mean 2.88 3.05 4.20 3.36 0.93 1.91
50th 1.00 2.00 3.00 2.00 0.00 0.00
25th 0.00 0.00 0.00 0.00 0.00 0.00
75th 4.00 4.00 6.00 4.50 1.00 2.00
SD 4.58 5.19 5.38 4.38 1.60 3.93
Panel B – dependent measures and the common control variables
N 474 79 142 88 88 77
NetEP
Mean 1.00 0.62 1.40 1.64 0.70 0.29
50th 1.00 0.00 2.00 2.00 0.00 0.00
25th 2.00 2.00 2.00 3.00 1.00 0.00
75th 0.00 1.00 0.00 0.00 0.00 0.00
SD 1.46 1.68 1.34 1.76 0.95 0.93
GdEP
Mean 0.37 0.75 0.38 0.39 0.10 0.26
50th 0.00 0.00 0.00 0.00 0.00 0.00
25th 1.00 0.00 0.00 0.00 0.00 0.00
75th 1.00 1.00 1.00 1.00 0.00 1.00
SD 0.59 0.85 0.51 0.56 0.36 0.44
BdEP
Mean 1.38 1.37 1.78 2.08 0.79 0.53
50th 1.00 1.00 2.00 2.00 0.00 0.00
Table 2 (continued)
Full sample Chem Elec Oil Food Pharm

25th 0.00 0.00 1.00 0.00 0.00 0.00
75th 2.00 3.00 3.00 4.00 1.00 1.00
SD 1.40 1.64 1.14 1.60 0.97 0.99
MFInd
Mean 0.53 0.31 0.70 0.74 0.41 0.35
MFDisc
Mean 1.75 1.79 2.01 0.72 1.72 2.40
50th 1.10 1.45 1.79 0.69 1.09 2.98
25th 0.00 0.69 0.69 0.00 0.00 0.67
75th 3.37 3.40 3.64 0.72 3.30 3.91
SD 1.58 1.52 1.61 0.91 1.62 1.64
COECtp
Mean 0.12 0.13 0.11 0.09 0.14 0.12
50th 0.11 0.12 0.09 0.09 0.14 0.11
25th 0.08 0.09 0.07 0.05 0.11 0.07
75th 0.15 0.17 0.13 0.13 0.17 0.17
SD 0.07 0.06 0.07 0.07 0.05 0.08
COECpeg
Mean 0.08 0.10 0.07 0.07 0.08 0.09
50th 0.08 0.10 0.07 0.07 0.08 0.09
25th 0.06 0.08 0.05 0.06 0.07 0.08
75th 0.10 0.12 0.09 0.10 0.10 0.10
SD 0.03 0.03 0.03 0.05 0.02 0.02
EFCF
Mean 27.89 28.62 22.82 29.42 26.59 36.22
50th 25.79 27.58 22.52 28.05 23.63 37.20
25th 17.35 22.52 15.86 19.12 16.24 20.61
75th 35.34 35.33 28.74 38.32 36.20 49.05
SD 13.47 10.69 9.51 13.24 12.15 18.75
This table reports descriptive statistics for the sample and by industry. Panel A includes statistics for the VEDQ partitions and
Panel B includes the common control variables and the dependent measures. Column 1 provides statistics for the full sample,
the next five columns report values based on industry. All variables are defined in Table 1. TScore = score from a completed
VEDQ index. HardD (SoftD, Positive, Neutral, Negative) = the score from a completed VEDQ index classified as hard (soft, positive,
neutral, negative). All other variables are defined in Table 1.
4. Empirical results
We use panel data in our analysis, which may result in residuals that are correlated across firms or
time leading to bias in OLS standard errors (Petersen, 2009). To control for this, we do the following.
First, we use intra-industry, within-year ranks to capture cross-sectional variation in the voluntary
disclosure measures and environmental performance. We also include controls for whether a firm
operates in a sensitive or non-sensitive industry. Finally, we control for time-series correlation in
the error terms and provide t-statistics based on the corrected errors in our estimation process.26
We present results across Tables 4–8. Table 4 begins with the basic model, which includes the
model-specific and other control variables along with the broadest disclosure quality measures (CER
and R_TScore). To highlight the importance of controlling for environmental performance in our anal-
yses, Table 5 presents a model that incorporates R_NetEP or its two components R_GdEP and R_BdEP
along with the broad disclosure quality measures. The remaining tables include the systematically
more refined VEDQ measures, formed by partitioning the voluntary environmental disclosure index
items based on whether they provide hard/soft data or positive/neutral/negative content and provides
results based on using either R_NetEP or its two components R_GdEP and R_BdEP to control for
26
We also re-estimated all our models and calculated t-statistics and significance levels based on two-way clustered standard
errors, with substantively similar results.
16
Table 3
Spearman correlation statistics.
M. Plumlee et al. / J. Account. Public Policy xxx (2015) xxx–xxx

2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
1: CER 0.650 0.644 0.569 0.637 0.621 0.572 0.666 0.533 0.493 0.388 0.140 0.290 0.264 0.005 0.076 0.212 0.035 0.166
2: TScore – 0.975 0.866 0.939 0.727 0.922 0.863 0.888 0.810 0.413 0.182 0.175 0.197 0.033 0.001 0.074 0.066 0.096
3: HardD – 0.769 0.897 0.751 0.937 0.879 0.924 0.725 0.373 0.176 0.186 0.190 0.041 0.011 0.072 0.053 0.092
4: SoftD – 0.891 0.608 0.741 0.713 0.670 0.936 0.462 0.185 0.116 0.178 0.004 0.031 0.085 0.070 0.114
5: PosD – 0.641 0.813 0.905 0.806 0.880 0.347 0.166 0.181 0.190 0.020 0.009 0.093 0.059 0.103
6: NeuD – 0.635 0.655 0.582 0.544 0.371 0.112 0.144 0.075 0.084 0.011 0.164 0.014 0.110
7: NegD – 0.780 0.962 0.660 0.452 0.203 0.141 0.259 0.007 0.000 0.023 0.039 0.115
8: HdPos – 0.764 0.665 0.398 0.142 0.205 0.197 0.001 0.024 0.125 0.011 0.131
9: HdNeg – 0.658 0.262 0.176 0.150 0.232 0.018 0.003 0.008 0.069 0.063
10: SfPos – 0.211 0.136 0.123 0.144 0.032 0.035 0.060 0.098 0.066
11: SfNeg – 0.213 0.016 0.220 0.143 0.017 0.067 0.110 0.211
12: NetEP – 0.297 0.914 0.102 0.106 0.130 0.016 0.070
13: GdEP – 0.116 0.057 0.054 0.109 0.030 0.012
14: BdEP – 0.082 0.087 0.079 0.029 0.088
15: MFInd – 0.231 0.115 0.166 0.160
16: MFDisc – 0.105 0.021 0.063
17: EFCF – 0.312 0.233
18: COECTP – 0.549
19: COECPEG –
All variables are defined in Tables 1 and 2. Correlations significant at <0.05 are bolded.
Table 4
Regression models with general disclosure and environmental disclosure.

Dependent variables are firm stock price (PRICE), Value Line forecasted target discounted back to the current period using the firm-specific COEC (EFCF), and firm-specific implied COEC
based on the Target Price or the PEG method (COEC). All variables are defined in Table 1. N = 474. Int/Ind indicates whether the intercept/industry controls are statistically significant. Two-
tailed t-statistics and significance levels are based a linear regression model with a correction for auto-correlated error terms.
⁄
Significant at the 0.10 level.
⁄⁄
⁄⁄⁄
17
18
Table 5
Regression models with combined proxy for environmental disclosure quality.

All variables are defined in Table 1. N = 474. Int/Ind indicates whether the intercept/industry controls are statistically significant. Two-tailed t-statistics and significance levels are based a
linear regression model with a correction for auto-correlated error terms.
⁄
⁄⁄
⁄⁄⁄
Table 6
Regression models with environmental disclosure by type.

⁄
⁄⁄
⁄⁄⁄
19
Table 7
20
Regression models with environmental disclosure by nature.

⁄
⁄⁄
⁄⁄⁄
Table 8
Regression model with disclosure type_nature.

⁄
⁄⁄
⁄⁄⁄
21
environmental performance. Given the strong industry effect of environmental performance and
potentially unique environmental disclosure items, we use intra-industry, within-year ranks of our
environmental performance and disclosure measures in the analyses. In each table, the first two col-
umns report coefficients and t-statistics when firm value is the dependent measure. The next columns
include coefficient estimates and t-statistics when expected future cash flow (the numerator of firm
value) is the dependent measure. The final four columns include coefficient estimates and t-statistics
based on our two COEC proxies (the denominator of firm value). We present two panels in Tables 5–8;
panel A includes a single control for environmental performance (R_NetEP) while in Panel B we sep-
arate environmental performance into positive and negative aspects (R_GdEP and R_BdEP).
Each regression also includes the voluntary disclosure control variables MFInd and MFDisc. Finally,
we include industry type indicator variables in all models to control for whether firms are classified as
either environmentally sensitive or not. For ease of exposition we exclude the intercept and the
industry-type indicators from the table, and simply note when they are statistically significant.
4.1. Overall disclosure
Table 4 presents our baseline models, where we include our environmental disclosure related vari-
ables (CER and R_TScore) as independent measures. Across the three models (Price, EFCF and COEC),
we find little evidence that variance in voluntary environmental quality, as measured by CER and
R_TScore, is associated with firm value through either EFCF or COEC. The issuance of a CER is associated
with higher EFCF, although the lack of a control for environmental performance raises the question of
whether that association is due to environmental performance or the issuance of a CER. In each COEC
regression, one of the voluntary disclosure variables MFInd and MFDisc is significantly negative, as
expected. We find similar relations between COEC and the voluntary disclosure control variables
throughout the remaining regressions.
In Table 5, we re-estimate our baseline models along with our proxies for firm environmental perfor-
mance. Panel A incorporates a single measure of environmental performance (R_NetEP), while panel B
separates environmental performance into its positive and negative components – R_GdEP and
R_BdEP.27 R_NetEP is the intra-industry within-year rank of NetEP, such that firms that have better relative
environmental performance are ranked higher than firms with worse relative environmental performance.
We document that firms with higher relative environmental performance have higher EFCF, consistent with
better performers generating more profit than poorer performers. We also find – using both COEC proxies –
that firms with better relative environmental performance are less risky than poorer environmental per-
formers. This result is consistent with our expectation and prior research (e.g., Ghoul et al., 2011). Adding
environmental performance to the models has a limited impact on associations between the dependent
and independent variables across the models, however. Thus, the significantly positive coefficient on CER
in the EFCF model can be interpreted as firms with CERs being more profitable than firms that do not.
To allow us to explore whether the positive and negative aspects of environmental performance
have differential impacts in our models, we employ the intra-industry, within-year ranks of GdEP
and BdEP in place of R_NetEP in Panel B. The associations between our dependent measures and envi-
ronmental variables are generally unaffected by this change, although we do document a positive
association between R_TScore and stock price in Panel B but not Panel A. Separating environmental
performance into its positive and negative aspects does reduce the overall explanatory of the models
and, in two of our models, neither R_GdEP nor R_BdEP is statistically significant. These results provide
limited support for H1, in that there is there is a positive incremental impact of providing voluntary
environmental disclosures via a CER. Consistent with prior research, our basic model provides no sup-
port for H2.
4.2. Partitioned VEDQ
Table 6 builds on the specifications in Table 5, which includes environmental performance, CER,
and R_TScore as independent variables. Specifically, we substitute R_HardD and R_SoftD for R_TScore
27
We present results based on positive/negative aspects of environmental performance to examine whether good/bad
environmental performance is differentially weighted in the analyses.
to test H1A and H2A, which predict that the association between disclosures and EFCF and COEC differ
across disclosure type, respectively. Our results provide little support for these hypotheses, however.
In general, substituting R_HardD and R_SoftD for R_TScore has little impact across the models; the
coefficients on R_HardD or R_SoftD are not statistically different from zero in any of the models.
Incorporating a single measure for environmental performance (R_NetEP in Panel A) generally
increases the overall explanatory power relative to including R_GdEP and R_BdEP (Panel B) in the
models.
Similar to Table 6, the results presented in Table 7 build on the specifications in Table 5. In this case,
we partition our disclosure score based on the nature of the disclosures, instead of the type: R_PosD,
R_NeuD, and R_NegD are substituted for R_TScore. This substitution allows us to test H1B and H2B,
which predict that the nature of the environmental disclosures impacts the association between dis-
closures and EFCF and COEC. Consistent with prior research that considers the content of disclosures,
we find the nature of the disclosed items is associated with the relation between voluntary disclosure
quality and both EFCF and our primary proxy for COEC. Specifically, higher quality positive disclosures
are positively associated with EFCF (coefficient = 7.840) while higher quality neutral disclosures are
positively associated with COECTP (coefficient = 0.050). These findings support H1B, in that the relation
between EFCF and disclosure quality for disclosures classified as positive differs from the relation
between EFCF and disclosure quality for disclosures classified as neutral and those classified as
negative. We also find support for H2B, in that the relation between COECTP and disclosure quality
for disclosures classified as neutral differs from the relation between COECTP and disclosure quality
for disclosures classified as positive or negative. The results suggest that higher quality positive
disclosures are positively related to EFCF, while higher quality neutral disclosures are associated with
higher COEC.
Consistent with the earlier models, incorporating a single measure of environmental performance
in the regressions versus a measure for positive and one for negative aspects of environmental seems
to suggest that the combined measure provides greater explanatory power.
Our final table (Table 8) presents results based on classifying voluntary environmental disclosures
based on both type (hard/soft) and nature (positive/neutral/negative) of the disclosures. Again, we
substitute ranked variables based on the parsed voluntary environmental data for R_TScore. This
substitution allows us to test H1C and H2C, which predict that the association between disclosures
and EFCF and COEC differs by the type_nature of the environmental disclosures. Our results are consis-
tent with our predictions: higher quality disclosures that are classified as soft and positive are posi-
tively associated with EFCF (coefficient = 9.983) while the other type_nature disclosures are not
significant in this model. This finding may reflect leading edge performers identifying good things that
they are planning to do and best practice environmental behaviors. When we examine the
associations with COECTP, we find that higher quality disclosures that are classified as soft/positive
(soft/negative) are negatively (positively) associated with COEC (coefficients = 0.061 and 0.069,
respectively) and that the magnitudes of these coefficients are statistically different from each other.
These findings support H1C and H2C. More importantly, our findings provide support for soft/positive
environmental disclosures leading to lower COEC, while soft/negative environmental disclosure
quality is positively related to COEC, consistent with these disclosures being related to firm riskiness.
The results in Panel A, based on a single measure of environmental performance, and those in Panel B,
based on two measures, are consistent in terms of the environmental disclosure measures.
4.3. Discussion
Our analysis examines the relations between the two components of firm value and VEDQ. We
provide insight into findings from prior research and economic theory surrounding voluntary environ-
mental disclosures. Our basic models, included in Tables 4 and 5, replicate findings from prior litera-
ture, showing that overall environmental disclosure quality is not significantly associated with either
EFCF or COEC. However, as predicted, we find that classifying the voluntary environmental disclosures
based on type and nature increases our ability to identify significant relations. Specifically, we find
that the coefficients in the models differ according to the content of the disclosures and that using
the parsed data in our analysis increases the overall explanatory of our regression models. It also
suggests that understanding the complex relations may require more a precise understanding and
measurement of the various proxies, rather than treating all disclosures as though they are equal
and have similar relations with the dependent measures.
In our analysis we include two proxies for the cost of equity capital, one based on the target price
method and one based on the PEG method. As discussed earlier, our primary analysis relies on COECTP,
while we include COECPEG to allow us to compare to concurrent research in this area. Our reliance on
COECTP is based on prior studies that strongly support the use of this variable (e.g., Botosan et al.,
2011). Furthermore, the target price method of estimating COEC, which is not subject to many of
the concerns raised in the literature (Lambert, 2009), provides different outcomes from studies that
employ other methods of estimating COEC (e.g. Ogneva et al., 2007; Ashbaugh-Skaife et al., 2009).
Our results are consistent with that prior research in that we are unable to document a relation
between COECPEG and our various specifications of voluntary environmental disclosures. At the same
time, however, we find support for H2 and H3 using COECTP. We argue that this finding is consistent
with one measure (COECTP) providing a superior measure of COEC.
5. Conclusions
In this study, we reexamine the relation between firm value and VEDQ value through its associa-
tions with both the cash flow and cost of equity components. Our study builds on prior and concurrent
research that provides inconsistent results surrounding the relation between corporate social respon-
sibility and/or environmental disclosure and the COEC. In addition, we contribute to an understanding
of the link between environmental performance and COEC. In contrast with many other studies, we
include environmental performance measures based on both positive and negative aspects of
environmental performance – the KLD strengths and concerns – instead of a single proxy that is based
on negative aspects of environmental performance, the TRI.
We also contribute to research that focuses on the numerator of firm value. Overall, we document
that positive disclosure quality is significantly positively associated with firm value. We also find that
the numerator (EFCF) component of firm value is significantly associated with VEDQ, but only when
VEDQ is classified by the nature and the type_nature of the disclosures. Finally, we document that firm
COEC measured using the target price method is, as suggested by theory, negatively associated with
improved VEDQ related to soft_positive items and positively associated with VEDQ related to soft_neg-
ative items. These findings support separating firm value into its numerator and denominator and the
need to consider finer measures of VEDQ in the analysis. In summary, our study extends prior research
by considering the relation between voluntary environmental disclosure quality and the cash flow and
COEC components of firm value independent of each other and by classifying voluntary environmental
disclosures by type and type_nature.
Our findings complement and extend those provided in Clarkson et al. (2013), who find environ-
mental disclosures to be incrementally informative over an alternative measure for environmental
performance (TRI data) in predicting firm profitability and firm value but, in contrast to our study, fail
to document an association with cost of equity. These differences are most likely due to differences in
our environmental performance and COEC measures and samples (their analysis focuses on environ-
mentally sensitive industries over two years), as well as our refinements of the environmental
disclosures. Our sample period covers the 2000–2005 time period, which might limit our ability to
generalize the documented results to the current time period.
Acknowledgements
The authors gratefully acknowledge generous funding from the KPMG and University of Illinois’s
Business Measurement Research Program. We have benefited from the comments of two anonymous
reviewers, Teri Lombardi Yohn, and participants at Georgetown University, Portland State University,
University of Oklahoma, the 2009 International Conference on Business and Sustainability. All remain-
ing errors are our own.
Appendix A. Supplementary data
Supplementary data associated with this article can be found, in the online version, at http://dx.doi.
org/10.1016/j.jaccpubpol.2015.04.004.
References
Aerts, W., Cormier, D., 2009. Media legitimacy and corporate environmental communication. Acc. Organ. Soc. 34 (1), 1–27.
Aerts, W., Cormier, D., Magnan, M., 2008. Corporate environmental disclosure, financial markets and the media: an international
perspective. Ecol. Econ. 64, 643–659.
Al-Tuwaijri, S., Christensen, T., Hughes II, K., 2004. The relations among environmental disclosure, environmental performance,
and economic performance: a simultaneous equations approach. Acc. Organ. Soc. 29, 447–471.
Amihud, Y., Mendelson, H., 1986. Asset pricing and the bid-ask spread. J. Financ. Econ. 17, 223–249.
Ashbaugh-Skaife, H., Collins, D., Kinney, W., Lafond, R., 2009. The effect of internal control deficiencies on firm risk and the cost
of equity capital. J. Acc. Res. 47, 1–43.
Baginski, S., Rakow, K., 2012. Management earnings forecast disclosure policy and the cost of equity capital. Rev. Acc. Stud. 17
(2), 279–321.
Barry, C.B., Brown, S.J., 1985. Differential information and security market equilibrium. J. Financ. Quant. Anal. 20 (4), 407–422.
Barth, M., McNichols, M., 1994. Estimation and market valuation of environmental liabilities relating to superfund sites. J. Acc.
Res. 32, 177–209, Supplement.
Blacconiere, W.G., Northcutt, D., 1997. Environmental information and market reactions to environmental legislation. J. Acc.,
Audit. Financ. 12 (2), 149–178.
Blacconiere, W.G., Patten, D., 1994. Environmental disclosures, regulatory costs and changes in share value. J. Acc. Econ. 18,
357–377.
Botosan, C., Plumlee, M., 2002. A re-examination of disclosure level and the expected cost of equity capital. J. Acc. Res. 40, 21–40.
Botosan, C., Plumlee, M., 2005. Assessing alternative proxies for the expected risk premium. Acc. Rev. 80 (1), 21–53.
Botosan, C., Plumlee, M., Wen, J., 2011. The relation between expected returns, realized returns, and firm risk characteristics.
Contemp. Acc. Res. 28 (4), 1085–1122.
Bushee, B., Noe, C., 2000. Corporate disclosure practices, institutional investors, and stock return volatility. J. Acc. Res. 38
(Supplement), 171–202.
Campbell, K., Sefcik, S., Soderstrom, N., 2003. Disclosure of private information and reduction of uncertainty: environmental
liabilities in the chemical industry. Rev. Quant. Financ. Acc. 21 (4), 349–378.
Cho, C., Patten, D., 2007. The role of environmental disclosures as tools of legitimacy: a research note. Acc. Organ. Soc. 32 (7/8),
639–647.
Cho, C., Roberts, R., Patten, D., 2010. The language of US corporate environmental disclosure. Acc. Organ. Soc. 35, 431–443.
Clarkson, P., Li, Y., Richardson, G., 2004. The market valuation of environmental capital expenditures by pulp and paper
companies. Acc. Rev. 79 (2), 329–353.
Clarkson, P., Li, Y., Richardson, G., Vasvari, F., 2008. Revisiting the relation between environmental performance and
environmental disclosure: an empirical analysis. Acc. Organ. Soc. 33, 303–327.
Clarkson, P., Fang, X., Li, Y., Richardson, G., 2013. The relevance of environmental disclosure: are such disclosures incrementally
informative? J. Acc. Public Policy 32, 410–431.
Coles, J., Loewenstein, U., Suay, J., 1995. On equilibrium pricing under parameter uncertainty. J. Financ. Quant. Anal. 30 (3), 347–
364.
Cormier, D., Magnan, M., 2007. The revisited contribution of environmental reporting to investors’ valuation of a firm’s
earnings: an international perspective. Ecol. Econ. 62, 613–626.
Cormier, D., Ledoux, M., Magnan, M., 2009. The use of web sites as a disclosure platform for corporate performance. Int. J. Acc.
Inf. Syst. 10 (1), 1–24.
Cormier, D., Ledoux, M., Magnan, M., 2011. The informational contribution of social and environmental disclosures for investors.
Manage. Decis. 49 (8), 1276–1304.
Dhaliwal, D., Li, O., Tsang, A., Yang, Y., 2011. Voluntary nonfinancial disclosure and the cost of equity capital: the initiation of
corporate social responsibility reporting. Acc. Rev. 86 (1), 59–72.
Dhaliwal, D., Radhakrishnan, S., Tsang, A., Yang, Y., 2012. Nonfinancial disclosures and analyst forecast accuracy: international
evidence on corporate social responsibility disclosure. Acc. Rev. 87 (3), 723–759.
Doyle, J., Lundholm, R., Soliman, M., 2003. The predictive value of expenses excluded from pro forma earnings. Rev. Acc. Stud. 8,
145–174.
Drake, M., Myers, J., Myers, L., 2009. Disclosure quality and the mispricing of accruals and cash flow. J. Acc., Audit., Financ. 24 (1),
1–34.
Elliott, W., Jackson, K., Peecher, M., White, B., 2014. The unintended effect of corporate social responsibility performance on
investors’ estimates of fundamental value. Acc. Rev. 89 (1), 275–302.
Francis, J., LaFond, R., Olsson, P., Schipper, K., 2004. Costs of equity and earnings attributes. Acc. Rev. 79, 967–1010.
Ghoul, S., Guedhami, O., Kwok, C., Mishra, D., 2011. Does corporate social responsibility affect the cost of capital? J. Bank. Financ.
35 (9), 2388–2406.
GRI, 2006. Sustainability Reporting Guidelines. Global Reporting Initiative. Boston, MA.
Griffin, P., Sun, Y., 2012. Going green: market reaction to CSR newswire releases. J. Acc. Public Policy 32, 93–113.
Hutton, A., Miller, G., Skinner, D., 2003. The role of supplementary statements with management earnings forecasts. J. Acc. Res.
41 (5), 867–890.
KLD Research and Analytics Inc., 2003. KLD ratings data: Inclusive social rating criteria. KLD, Boston, MA.
Lambert, R., 2009. Discussion of ‘‘on the relation between expected returns and implied cost of capital’’. Rev. Acc. Stud. 14 (2–3),
260–268.
Lang, M., Lundholm, R., 1993. Cross-sectional determinants of analysts’ ratings of corporate disclosures. J. Acc. Res. 31 (2), 246–
271.
Lang, M., Lundholm, R., 1996. Corporate disclosure policy and analyst behavior. Acc. Rev. 71 (4), 467–493.
Leuz, C., Verrecchia, R., 2000. The economic consequences of increased disclosure. J. Acc. Res. 38 (Supplement), 91–136.
Lev, B., Petrovits, C., Radhakrishnan, S., 2010. Is doing good good for you? How corporate charitable contributions enhance
revenue growth. Strateg. Manage. J. 31 (February), 182–200.
Lundholm, R., Myers, L., 2002. Bringing the future forward: the effect of disclosure on the returns-earnings relation. J. Acc. Res.
40 (3), 809–839.
Marshall, R.S., Brown, D., Plumlee, M., 2007. Negotiated transparency? Corporate citizenship engagement and environmental
disclosure. J. Corp. Citizenship 28 (December).
Matsumura, E., Prakash, R., Vera-Munoz, S., 2014. Firm-value effects of carbon emissions and carbon disclosures. Acc. Rev. 89
(2), 695–724.
Merton, R.C., 1987. A simple model of capital market equilibrium with incomplete information. J. Financ. 42, 483–510.
Ogneva, M., Raghunandan, K., Subramanyam, K., 2007. Internal control weakness and cost of equity: evidence from SOX section
404 disclosures. Acc. Rev. 82 (5), 1255–1297.
Ohlson, J., 1995. Earnings, book value, and dividends in security valuation. Contemp. Acc. Res. 11 (2), 661–687.
Patten, D., 2002. The relation between environmental performance and environmental disclosure: a research note. Acc. Organ.
Soc. 27, 763–773.
Petersen, M., 2009. Estimating standard errors in finance panel data sets: comparing approaches. Rev. Financ. Stud. 22 (1), 435–
480.
Plumlee, M., Brown, D., Marshall, R., 2009. Voluntary Environmental Disclosure Quality and Firm Value: Roles of Venue and
Industry Type. Portland State University Working Paper.
Richardson, A., Welker, M., 2001. Social disclosure, financial disclosure and the cost of equity capital. Acc. Organ. Soc. 26, 597–
616.
Skinner, D., 1994. Why firms voluntarily disclose bad news. J. Acc. Econ. 32 (1), 38–60.
Wagenhofer, A., 1990. Voluntary disclosure with a strategic opponent. J. Acc. Econ. 12 (4), 341–363.

Plum Lee 2015

Diunggah oleh

Informasi Dokumen

Hak Cipta

Format Tersedia

Bagikan dokumen Ini

Bagikan atau Tanam Dokumen

Opsi Berbagi

Apakah menurut Anda dokumen ini bermanfaat?

Apakah konten ini tidak pantas?

Hak Cipta:

Format Tersedia

Plum Lee 2015

Diunggah oleh

Hak Cipta:

Format Tersedia

J. Account.

Public Policy xxx (2015) xxx–xxx

Contents lists available at ScienceDirect

J. Account. Public Policy

Voluntary environmental disclosure quality

This study reexamines the relationship between the quality of a

2. Literature review and development of hypotheses

2.1. Literature review

2.1.1. Firm value and environmental disclosures

2.1.2. Environmental disclosures and environmental performance

2.2. Hypotheses development

2.2.1. Hypotheses – Expected future cash ﬂows

2.2.2. Hypotheses development – COEC

3. Research design, variable measurement, and sample selection

3.1. Research design

3.1.1. Regression model – Firm value

P ¼ a þ b1 BV þ b2 AE þ b3 MFInd þ b4 MFDisc þ b5 NetEP þ b6 CER þ b7 R DScoren þ e ð1Þ

3.1.2. Regression model – Expected future cash ﬂows

3.1.3. Regression – COEC

3.2. Variable measurement

3.2.1. Dependent variables

3.2.2. Independent variables

Full sample Chem Elec Oil Food Pharm

Full sample Chem Elec Oil Food Pharm

M. Plumlee et al. / J. Account. Public Policy xxx (2015) xxx–xxx

M. Plumlee et al. / J. Account. Public Policy xxx (2015) xxx–xxx

M. Plumlee et al. / J. Account. Public Policy xxx (2015) xxx–xxx

Regression models with environmental disclosure by type.

M. Plumlee et al. / J. Account. Public Policy xxx (2015) xxx–xxx

Regression models with environmental disclosure by nature.

M. Plumlee et al. / J. Account. Public Policy xxx (2015) xxx–xxx

Regression model with disclosure type_nature.

M. Plumlee et al. / J. Account. Public Policy xxx (2015) xxx–xxx

4.2. Partitioned VEDQ

Appendix A. Supplementary data

Anda mungkin juga menyukai