Description
Financial report environmental disclosure has been widely criticized because the extent of disclosure both varies
in response to exposures facing the firm and because it is not an accurate measure of firm environmental performance.
This study suggests there are at least two potential problems that need to be addressed before this disclosure
can be completely condemned as meaningless. A first potential problem is the earlier studies’ focus on
broad measures of environmental disclosure. Hidden within those broader measures may be pieces of meaningful
information. Second, while the disclosure examined previously may not correspond well with past environmental
performance, it, or at least parts of it, may still provide meaningful information for assessing future environmental
action. This study attempts to address these shortcomings by examining one specific, but potentially
useful, category of environmental disclosure: projections of future spending for pollution abatement and control
equipment.
The accuracy of ?nancial report projections of
future environmental capital expenditures: a research note
Dennis M. Patten
*
Department of Accounting, Illinois State University, 2200 North & School Streets, Normal, IL 61761-5520, USA
Abstract
Financial report environmental disclosure has been widely criticized because the extent of disclosure both varies
in response to exposures facing the ?rm and because it is not an accurate measure of ?rm environmental perfor-
mance. This study suggests there are at least two potential problems that need to be addressed before this dis-
closure can be completely condemned as meaningless. A ?rst potential problem is the earlier studies’ focus on
broad measures of environmental disclosure. Hidden within those broader measures may be pieces of meaningful
information. Second, while the disclosure examined previously may not correspond well with past environ-
mental performance, it, or at least parts of it, may still provide meaningful information for assessing future envi-
ronmental action. This study attempts to address these shortcomings by examining one speci?c, but potentially
useful, category of environmental disclosure: projections of future spending for pollution abatement and control
equipment.
Based on 355 sets of projected/actual spending drawn from 10K reports ?led with the US’s Security and Exchange
Commission between 1993 and 2002, inclusive, results suggest that the projections may be more misleading than
meaningful. Actual spending was lower than the projected amount for more than 75% of the observations, and the
mean projection error (the di?erence adjusted for the size of the projection) was a negative 16.4%. Analysis of
projection errors for a 2-year, rather than a 1-year window revealed a similar distribution. In contrast to the accuracy
of the environmental capital expenditure projections, the actual/projection di?erence for total capital expenditures
was very small, suggesting that it is not a di?culty in estimating future capital spending that is driving the error
results. The projection errors also did not appear to be a function of changes in company revenues or pro?tability.
Overall, therefore, it appears that, similar to broader measures of corporate environmental disclosure, projections of
future spending on environmental control lack value.
Ó 2004 Elsevier Ltd. All rights reserved.
Introduction
Since at least the late 1970s (see, e.g., Dierkes &
Preston, 1977; Churchill, 1978; Nader, 1978)
?nancial report environmental disclosures have
been criticized as being nothing more than ‘‘self-
serving, inaccurate accounts of corporate envi-
ronmental performance’’ (Wiseman, 1982, p. 54).
Building on this criticism, proponents of legiti-
macy theory (e.g., Hogner, 1982; Patten, 1991,
1992; Deegan & Rankin, 1996; Lindblom, 1994;
Walden & Schwartz, 1997; O’Donovan, 1999)
more speci?cally argue that such disclosures are
used as manipulative and misleading tools for
*
Tel.: +1-309-438-7857; fax: +1-309-438-8431.
E-mail address: [email protected] (D.M. Patten).
0361-3682/$ - see front matter Ó 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.aos.2004.06.001
Accounting, Organizations and Society 30 (2005) 457–468
www.elsevier.com/locate/aos
addressing corporate exposure to public policy
pressure. A key premise underlying the argument
developed by the legitimacy theorists is the failure
of previous studies (e.g., Ingram & Frazier, 1980;
Wiseman, 1982; Freedman & Wasley, 1990: Fek-
rat, Inclan, & Petroni, 1996; Patten, 2002b) to ?nd
a signi?cant positive association between the ex-
tent of ?nancial report environmental disclosure
and measures of environmental performance.
This study suggests that one possible reason for
the lack of signi?cance in the studies of the relation
between environmental disclosure and environ-
mental performance is their reliance on broad
measures of disclosure.
1
Further, while ?nancial
report environmental disclosure may not corre-
spond well with measures of past environmental
performance (which are used in each of the prior
studies), it is possible that this disclosure, or at least
elements of it, may have value as a signal of future
corporate environmental action. This study ad-
dresses these issues by examining whether one
speci?c item of corporate environmental disclo-
sure––projections for future spending for pollution
abatement andcontrol equipment corresponds with
subsequent environmental action (actual spending).
Disclosure as a tool of legitimacy
This study is concerned with the extent to which
corporations use ?nancial report environmental
disclosure rather than actual environmental per-
formance to seek or maintain social legitimacy.
The concept of social legitimacy is based on the
belief that organizations must continually dem-
onstrate they are operating within the norms and
values of society. Where the actual or perceived
actions of a ?rm depart from these values, the
social legitimacy of the corporation is threatened
(Dowling & Pfe?er, 1975; Milne & Patten, 2002),
which, according to Patten (1991, 1992) can lead
to increased likelihood of public policy actions (see
Walden & Schwartz, 1997) against the company.
Because such actions are potentially costly, cor-
porations have an incentive to maintain their so-
cial legitimacy. Dowling and Pfe?er (1975, p. 127)
argue that while an organization can ‘‘adapt its
output, goals, and methods of operation to con-
form to prevailing de?nitions of legitimacy’’, it
might instead use communication in an attempt to
alter the de?nition of, or to project an image of,
legitimacy. Thus, corporations may be using
?nancial report environmental disclosure as a tool
for gaining or maintaining social legitimacy.
Several recent studies provide evidence that
companies tend to increase the disclosure of posi-
tive environmental information in their ?nancial
reports in reaction to increased exposures due to
negative environmental performance. To illustrate,
Deegan and Rankin (1996) report that while only
two of 20 Australian companies successfully
prosecuted for environmental violations by the
Environmental Protection Authority disclosed
the existence of the o?ence, all companies in the
sample provided signi?cantly greater amounts of
positive environmental disclosure in their annual
reports than the amount of negative disclosure.
2
These results are consistent with the ?ndings of
Patten (2000). Patten’s study of US companies
documents that increases in the provision of po-
sitive environmental information in 10K reports
from the mid-1980s to the mid-1990s were signi?-
cantly correlated with mandated ?rm-speci?c in-
creases in disclosure identifying Superfund
exposures for the ?rms. Finally, Patten (2002a)
documents a signi?cant correlation between public
disclosure of company pollution performance as
measured by toxic releases into the environment
and increases in the extent of ?nancial report
environmental disclosure for those ?rms. Worse
performing companies (companies with higher
levels of toxic releases) had signi?cantly more
extensive o?-setting and mitigating environmental
disclosures.
Deegan and Rankin (1996) and Patten (2000);
Patten (2002a) all suggest companies are using
1
As noted below, these studies used content schemes
including from eight to 13 di?erent items of environmental
disclosure.
2
Deegan and Rankin (1996) de?ne positive disclosures as
information about corporate activities having a bene?cial
impact on the environment.
458 D.M. Patten / Accounting, Organizations and Society 30 (2005) 457–468
environmental disclosure as a manipulative device
rather than as a tool for full disclosure of actual
performance. This is consistent with the argument
presented by Neu, Warsame, and Pedwell (1998),
who, rather than looking at changes in environ-
mental disclosure, document that di?erences in the
extent of information provided for a sample of 33
Canadian ?rms can be seen as attempts at man-
aging corporate relations with di?ering relevant
publics. Thus, rather than representing meaningful
attempts at communicating real environmental
information, Neu et al. (1998, p. 268) suggest
‘‘these discourses o?er organizations a way to
sustain organizational legitimacy without neces-
sarily (their emphasis) changing organizational
modes of operation.’’
A key premise underlying the criticisms of
environmental disclosure outlined above is the
substantial body of evidence suggesting that cor-
porate environmental disclosure does not correlate
(or, as in Patten, 2002b, correlates negatively) with
actual environmental performance. For example,
three early studies of the disclosure/performance
relation (Ingram & Frazier, 1980, Wiseman, 1982
and Freedman & Wasley, 1990), examined whe-
ther the extent of ?nancial report environmental
disclosure was related to indices of environmental
performance issued by the Council on Economic
Priorities (CEP) in the early to mid 1970s. Each of
these studies examined ?nancial reports from the
year corresponding to the year of the CEP report,
and each used a content analysis scheme to mea-
sure the extent of disclosure. In essence, content
analysis awards points based on the presence, and
in some cases (Wiseman, 1982; Freedman &
Wasley, 1990) the degree of speci?city, of disclo-
sure of various items of environmental concern.
The scales used in these studies were fairly broad
with 13 items, encompassing topics addressing
economic, litigation, pollution abatement, and
general environmental areas. None of these early
studies found a signi?cant correlation between
disclosure and performance.
A more recent study of the relation between
environmental performance and environmental
disclosure, Fekrat et al. (1996), is based on 1991
annual report disclosures and uses a di?erent
measure of environmental performance than used
in the earlier studies. Fekrat et al.’s measure, al-
though obtained from the CEP, was based not on
CEP-based analyses, but on ?rm-reported toxic
emissions data required through the Emergency
Planning and Community Right-to-Know Act of
1986. These ?rm speci?c data were made available
in 1990 for 1988 calendar year emissions. The CEP
reported the data adjusted for ?rm size, and Fek-
rat et al. then ranked the companies based on these
adjusted measures. Fekrat et al. employed the
content analysis scheme developed by Wiseman
(1982) to quantify the environmental disclosures
for a sample of 26 of the largest US ?rms, and then
used the rankings based on these scores as the
disclosure index. Correlations of the rankings on
performance and disclosure were negative but not
statistically signi?cant.
Finally, Patten (2002b) also used toxic emis-
sions data as his measure of environmental perfor-
mance. Based on a sample of 131 US companies,
Patten documented that, when size and industry
classi?cation are controlled for, both line count
and content analysis measures of ?nancial report
environmental disclosure are signi?cantly, and
negatively, associated with pollution performance.
Although the content scheme utilized by Patten
(2002b) was less broad than those used in previous
studies (e.g., it did not include litigation disclo-
sures), it still covered eight di?erent disclosure
topics.
In summary, ?nancial report environmental
disclosure has been widely criticized as being
misleading because companies appear to increase
the provision of positive disclosure in response to
increased exposures facing the ?rm and because
this disclosure does not appear to be an accurate
measure of ?rm environmental performance.
However, there are at least two issues that need to
be addressed before this disclosure can be com-
pletely condemned as meaningless with respect to
assessments of corporate environmental actions.
First, a potential problem is the earlier studies’
focus on broad measures of environmental dis-
closure. Hidden within those broader measures
may be pieces of meaningful information. Second,
while the disclosure examined previously may not
correspond well with past environmental perfor-
mance, it, or at least parts of it, may still provide
D.M. Patten / Accounting, Organizations and Society 30 (2005) 457–468 459
information for assessing future environmental
action. This study attempts to address these
shortcomings by examining one speci?c, but
potentially useful, category of environmental dis-
closure; projections of future spending for pollu-
tion abatement and control equipment.
As noted by Zuber and Berry (1992, p. 46), the
Securities and Exchange Commission (SEC) ‘‘re-
quires companies to disclose the material e?ects
compliance with federal, state, and local environ-
mental laws may have on capital expenditures’’ and
‘‘to reveal the material estimated capital expendi-
tures for environmental control facilities for the
current ?scal year and such further periods as the
registrant may deem material.’’ And although only
a limited number of companies actually provide
speci?c disclosure in these areas (see, e.g., Blac-
coniere & Patten, 1994; Gamble, Hsu, Kite, &
Radtke, 1995), projections of future environmental
capital expenditures would appear tobe anexcellent
item to examine relative to identifying whether
disclosures are meant to be useful or misleading. As
argued by Ingram and Frazier (1980, p. 614), ‘‘for
the disclosures to be useful, there should be a cor-
respondence between the disclosures and actual
events.’’ In this case, it would be expected that, if the
information is intended to be meaningful disclosure
of future environmental actions, the projections
should, across companies and over time, closely
mirror subsequent expenditures. However, because
the disclosures are only projections and not actual
costs, they could be used to merely project an image
of pending environmental action that is not subse-
quently carried out. If this were indeed the case,
subsequent actual expenditures would be expected
to be signi?cantly lower than the projections.
Methods and results
This study focuses on 10K report environmen-
tal disclosures by US ?rms from the chemical,
petroleum, metals processing, and paper indus-
tries. These industries have been cited often as
being subject to greater environmental scrutiny
than other industry segments (see, e.g., Cowen,
Ferreri, & Parker, 1987; Patten, 1991, 2002;
Hackston & Milne, 1996). Firms in these industries
also consistently have been shown to exhibit
higher levels of ?nancial report environmental
disclosure than ?rms from less environmentally
sensitive industries (Cowen et al., 1987; Patten,
1991, 2002; Hackston & Milne, 1996). All 10K
reports available on the SEC EDGAR database
for ?rms in the four noted industries for reporting
years 1993 through 2002
3
were reviewed. The
examination centered on identifying disclosures on
capital expenditures, both projected and actual,
for pollution abatement and control. Each obser-
vation set included in the analysis required two
pieces of environmental disclosure. First, a speci?c
dollar amount projection for environmental capi-
tal expenditures had to be made in one year (des-
ignated year i), and second, a speci?c dollar
amount of actual environmental capital expendi-
tures had to be disclosed the following year (des-
ignated year i þ 1). Overall, the 10Ks for 119
di?erent companies were examined.
4
Of these, 61
?rms had at least one set of projected (year i) and
actual (year i þ 1) environmental capital expendi-
ture data disclosed, and 355 observation sets, in
total, were collected. Table 1 summarizes the
source of the data, by industry.
The accuracy of the environmental capital
expenditure projections was calculated by com-
paring the actual expenditure in year i þ 1 with the
projection from the previous year. However, be-
cause the size of the companies making environ-
mental capital expenditure disclosures varied
considerably (as did the size of the expenditures), a
projection error statistic adjusted for size was
computed. That is, projection errors were calcu-
lated as:
ðActual
iþ1
þ Projection
i
Þ=Projection
i
where Actual
iþ1
is the actual environmental capital
expenditure amount as disclosed in year i þ 1, and
3
The SEC ?rst began accepting electronic ?ling of 10K
reports (which are then made available through the EDGAR
database) in 1994. Thus, the ?scal year 1993 reports are the
oldest available through the EDGAR source.
4
Not all 119 companies had 10K reports available for all ten
years (1993–2002). Among other reasons, some companies
began electronic ?ling in later periods, and some ?rms merged
or were taken over by other companies.
460 D.M. Patten / Accounting, Organizations and Society 30 (2005) 457–468
Projection
i
is the projection for year i environ-
mental capital expenditures as disclosed in year i.
Table 2 presents results on the general distri-
bution (negative, exact, or positive) of the projec-
tion errors (Panel A), the mean projection error
amounts by industry (Panel B), and the mean
projection error amounts by year of actual
expenditure (Panel C). Fig. 1 provides a more de-
tailed presentation of the distribution of projection
errors. As noted in Panel A of Table 2, more than
three-fourths of the projection errors (76.1%) were
negative. This means that, for the vast majority of
observation sets, the actual environmental capi-
tal expenditure in year i þ 1 was less than had
been projected in year i. If it is assumed that there
is an equal chance of meeting (exact or positive
Table 1
Summary of the source of sample comparison sets of observa-
tions (environmental capital expenditures year
iþ1
and environ-
mental capital expenditure projection year
i
) by industry
Industry Total
number of
companies
examined
Companies
with at least
one compar-
ison set of
observations
Total
number of
observation
sets
Chemical 49 21 123
Metals 18 11 69
Petroleum 33 20 107
Paper 19 9 56
Total 119 61 355
All 10K reports available on the SEC EDGAR database for
years 1993 through 2002 for each of the 119 companies were
examined for environmental capital expenditure disclosures.
Table 2
Summary of 1-year projection errors [(actual environmental capital expenditures year i þ 1 minus projected environmental capital
expenditures year i) divided by projected environmental capital expenditures year i]
Panel A
Distribution of errors
Negative projection errors
(actual projection) 79 22.2%
Panel B
Mean projection errors by industry
Industry Observation sets Mean error (%)
Chemical 123 )16.6
Metals 69 )22.5
Petroleum 107 )11.7
Paper 56 )17.7
Total 355 )16.4
Panel C
Mean projection errors by year
(of actual expenditure)
Year
1994 35 )14.3
1995 46 )24.6
1996 49 )21.9
1997 50 ) 9.6
1998 44 )13.7
1999 39 )19.9
2000 36 )12.0
2001 28 )14.0
2002 28 )16.1
Total 355 )16.4
a
Binomial probability of this percentage of negative errors -75% -60 to-
75
-45 to -
60
-30 to -
45
-15 to -
30
0 to -15 exact 0 to +15 +15 to
+30
+30 to
+45
+45 to
+60
+60 to
+75
> +75%
Projection Error Amount
1 Year Projection Errors
Fig. 1. Distribution of 1-year projection errors.
462 D.M. Patten / Accounting, Organizations and Society 30 (2005) 457–468
for estimated environmental capital expenditures
for each of the next 2 years (designated year i þ 1
and year i þ 2), and (2) disclosure of actual
spending in the two subsequent years (year i þ 1
and year i þ 2). Far fewer companies made 2-year
projections. In total, only 110 observation sets
were available. Table 3 presents the results of the
2-year comparisons while Fig. 2 provides a more
detailed presentation of the distribution of 2-year
projection errors. Similar to the results for the one-
year analysis, the vast majority of the projection
errors (78.2%) are negative. The binomial proba-
bility of such a distribution of negative errors is
again less than 0.001, two-tailed. The mean 2-year
projection error is a slightly smaller )14.6%. As
highlighted in Fig. 2, the distribution of projection
errors is very similar to the one-year errors, with a
large number of observations between )15% and
)75%. This additional analysis, therefore, appears
to support the ?nding that environmental capital
spending projections lack accuracy.
5
A second additional analysis examines com-
pany projections for total capital expenditures.
Rather than being misleading disclosure, the
inaccuracy of environmental capital expenditure
projections could be a function of the di?culty in
projecting spending on capital assets. If so, a
comparison of actual capital expenditures in total
with previous year projections would be expected
to yield a distribution of inaccuracy similar to that
for environmental capital expenditures only. A
review of those 10K reports from which the envi-
ronmental capital expenditure observation sets
were drawn resulted in 193 sets of projection/ac-
tual data for total capital expenditures.
Panel A of Table 4 presents the results of the
comparisons of total capital expenditures with
their previous year projections, while Panel B of
Table 4 presents the corresponding environmental
capital expenditure comparison results for these
company/years only. Fig. 3 presents a comparison
of the distribution of projection errors for the total
capital expenditures and the corresponding envi-
ronmental capital expenditures. As highlighted in
Panel A of Table 4, almost exactly half of the total
capital expenditure projection errors are positive
(50.3%) and almost half (49.1%) are negative. The
binomial probability of observing this level of
negative errors is 0.886, two-tailed. In contrast,
and similar to the distribution for the total sample,
77.7% of the projection errors for environmental
capital expenditures are negative. The binomial
probability for this distribution is less than 0.001,
two-tailed. Further, the mean projection error for
the total capital expenditure amounts is a positive
4.0% in contrast to a mean error of minus 17.5%
for the corresponding environmental capital
expenditures. Fig. 3 highlights that while the dis-
tribution of environmental capital expenditure
projection errors is again heavily weighted toward
the )15% to )75% range, the error terms for the
total capital expenditure projections are much
more closely distributed around zero. These results
suggest that, on average and across time, the
projections for total capital expenditures are rea-
sonably accurate, while projections for environ-
mental capital expenditures are not.
6
Finally, this analysis also identi?es whether
the environmental capital expenditure projection
Table 3
Summary of 2-year projection errors
Negative projection errors
(actual projection) 23 20.9%
Total 110 100%
Mean 2-year projection error )14.6%
a
Binomial probability of this percentage of negative errors
-75% -60 to-
75
-45 to -
60
-30 to -
45
-15 to -
30
0 to -15 exact 0 to +15 +15 to
+30
+30 to
+45
+45 to
+60
+60 to
+75
> +75%
Projection Error Amounts
2 Year Projection Errors
Fig. 2. Distribution of 2-year projection errors.
7
Pearson product-moment correlation coe?cients for the
relation between the environmental capital expenditure projec-
tion errors and the change in revenues and change in return on
assets are 0.063 and 0.075, respectively. Neither measure is
statistically signi?cant at conventional levels (p < 0:10, two-
tailed).
464 D.M. Patten / Accounting, Organizations and Society 30 (2005) 457–468
0
10
20
30
40
50
60
70
> -75% -60 to-
75
-45 to -
60
-30 to -
45
-15 to -
30
0 to -15 exact 0 to
+15
+15 to
+30
+30 to
+45
+45 to
+60
+60 to
+75
>
+75%
Projection Error Amounts
Comparison of Projection Errors for Environmental Capital Expenditures versus Total Capital
Expenditures
Envir. Cap. Exp.
Total Cap. Exp.
Fig. 3. Comparison of distribution of total capital expenditure projection errors with environmental capital expenditure projection
errors.
Table 4
Summary of projection errors for companies with data on total capital expenditures
Panel A
Total capital expenditure projection errors
Negative projection errors
(actual projection) 93 48.2%
Total 193 100%
Mean projection error + 4.0%
Panel B
Environmental capital expenditure projection errors (companies with total capital expenditure data)
Negative projection errors
(actual projection) 40 20.7%
Total 193 100%
Mean projection error )17.5%
a
Binomial probability of this percentage of negative errors ¼0.886, two-tailed.
b
Binomial probability of this percentage of negative errors
Financial report environmental disclosure has been widely criticized because the extent of disclosure both varies
in response to exposures facing the firm and because it is not an accurate measure of firm environmental performance.
This study suggests there are at least two potential problems that need to be addressed before this disclosure
can be completely condemned as meaningless. A first potential problem is the earlier studies’ focus on
broad measures of environmental disclosure. Hidden within those broader measures may be pieces of meaningful
information. Second, while the disclosure examined previously may not correspond well with past environmental
performance, it, or at least parts of it, may still provide meaningful information for assessing future environmental
action. This study attempts to address these shortcomings by examining one specific, but potentially
useful, category of environmental disclosure: projections of future spending for pollution abatement and control
equipment.
The accuracy of ?nancial report projections of
future environmental capital expenditures: a research note
Dennis M. Patten
*
Department of Accounting, Illinois State University, 2200 North & School Streets, Normal, IL 61761-5520, USA
Abstract
Financial report environmental disclosure has been widely criticized because the extent of disclosure both varies
in response to exposures facing the ?rm and because it is not an accurate measure of ?rm environmental perfor-
mance. This study suggests there are at least two potential problems that need to be addressed before this dis-
closure can be completely condemned as meaningless. A ?rst potential problem is the earlier studies’ focus on
broad measures of environmental disclosure. Hidden within those broader measures may be pieces of meaningful
information. Second, while the disclosure examined previously may not correspond well with past environ-
mental performance, it, or at least parts of it, may still provide meaningful information for assessing future envi-
ronmental action. This study attempts to address these shortcomings by examining one speci?c, but potentially
useful, category of environmental disclosure: projections of future spending for pollution abatement and control
equipment.
Based on 355 sets of projected/actual spending drawn from 10K reports ?led with the US’s Security and Exchange
Commission between 1993 and 2002, inclusive, results suggest that the projections may be more misleading than
meaningful. Actual spending was lower than the projected amount for more than 75% of the observations, and the
mean projection error (the di?erence adjusted for the size of the projection) was a negative 16.4%. Analysis of
projection errors for a 2-year, rather than a 1-year window revealed a similar distribution. In contrast to the accuracy
of the environmental capital expenditure projections, the actual/projection di?erence for total capital expenditures
was very small, suggesting that it is not a di?culty in estimating future capital spending that is driving the error
results. The projection errors also did not appear to be a function of changes in company revenues or pro?tability.
Overall, therefore, it appears that, similar to broader measures of corporate environmental disclosure, projections of
future spending on environmental control lack value.
Ó 2004 Elsevier Ltd. All rights reserved.
Introduction
Since at least the late 1970s (see, e.g., Dierkes &
Preston, 1977; Churchill, 1978; Nader, 1978)
?nancial report environmental disclosures have
been criticized as being nothing more than ‘‘self-
serving, inaccurate accounts of corporate envi-
ronmental performance’’ (Wiseman, 1982, p. 54).
Building on this criticism, proponents of legiti-
macy theory (e.g., Hogner, 1982; Patten, 1991,
1992; Deegan & Rankin, 1996; Lindblom, 1994;
Walden & Schwartz, 1997; O’Donovan, 1999)
more speci?cally argue that such disclosures are
used as manipulative and misleading tools for
*
Tel.: +1-309-438-7857; fax: +1-309-438-8431.
E-mail address: [email protected] (D.M. Patten).
0361-3682/$ - see front matter Ó 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.aos.2004.06.001
Accounting, Organizations and Society 30 (2005) 457–468
www.elsevier.com/locate/aos
addressing corporate exposure to public policy
pressure. A key premise underlying the argument
developed by the legitimacy theorists is the failure
of previous studies (e.g., Ingram & Frazier, 1980;
Wiseman, 1982; Freedman & Wasley, 1990: Fek-
rat, Inclan, & Petroni, 1996; Patten, 2002b) to ?nd
a signi?cant positive association between the ex-
tent of ?nancial report environmental disclosure
and measures of environmental performance.
This study suggests that one possible reason for
the lack of signi?cance in the studies of the relation
between environmental disclosure and environ-
mental performance is their reliance on broad
measures of disclosure.
1
Further, while ?nancial
report environmental disclosure may not corre-
spond well with measures of past environmental
performance (which are used in each of the prior
studies), it is possible that this disclosure, or at least
elements of it, may have value as a signal of future
corporate environmental action. This study ad-
dresses these issues by examining whether one
speci?c item of corporate environmental disclo-
sure––projections for future spending for pollution
abatement andcontrol equipment corresponds with
subsequent environmental action (actual spending).
Disclosure as a tool of legitimacy
This study is concerned with the extent to which
corporations use ?nancial report environmental
disclosure rather than actual environmental per-
formance to seek or maintain social legitimacy.
The concept of social legitimacy is based on the
belief that organizations must continually dem-
onstrate they are operating within the norms and
values of society. Where the actual or perceived
actions of a ?rm depart from these values, the
social legitimacy of the corporation is threatened
(Dowling & Pfe?er, 1975; Milne & Patten, 2002),
which, according to Patten (1991, 1992) can lead
to increased likelihood of public policy actions (see
Walden & Schwartz, 1997) against the company.
Because such actions are potentially costly, cor-
porations have an incentive to maintain their so-
cial legitimacy. Dowling and Pfe?er (1975, p. 127)
argue that while an organization can ‘‘adapt its
output, goals, and methods of operation to con-
form to prevailing de?nitions of legitimacy’’, it
might instead use communication in an attempt to
alter the de?nition of, or to project an image of,
legitimacy. Thus, corporations may be using
?nancial report environmental disclosure as a tool
for gaining or maintaining social legitimacy.
Several recent studies provide evidence that
companies tend to increase the disclosure of posi-
tive environmental information in their ?nancial
reports in reaction to increased exposures due to
negative environmental performance. To illustrate,
Deegan and Rankin (1996) report that while only
two of 20 Australian companies successfully
prosecuted for environmental violations by the
Environmental Protection Authority disclosed
the existence of the o?ence, all companies in the
sample provided signi?cantly greater amounts of
positive environmental disclosure in their annual
reports than the amount of negative disclosure.
2
These results are consistent with the ?ndings of
Patten (2000). Patten’s study of US companies
documents that increases in the provision of po-
sitive environmental information in 10K reports
from the mid-1980s to the mid-1990s were signi?-
cantly correlated with mandated ?rm-speci?c in-
creases in disclosure identifying Superfund
exposures for the ?rms. Finally, Patten (2002a)
documents a signi?cant correlation between public
disclosure of company pollution performance as
measured by toxic releases into the environment
and increases in the extent of ?nancial report
environmental disclosure for those ?rms. Worse
performing companies (companies with higher
levels of toxic releases) had signi?cantly more
extensive o?-setting and mitigating environmental
disclosures.
Deegan and Rankin (1996) and Patten (2000);
Patten (2002a) all suggest companies are using
1
As noted below, these studies used content schemes
including from eight to 13 di?erent items of environmental
disclosure.
2
Deegan and Rankin (1996) de?ne positive disclosures as
information about corporate activities having a bene?cial
impact on the environment.
458 D.M. Patten / Accounting, Organizations and Society 30 (2005) 457–468
environmental disclosure as a manipulative device
rather than as a tool for full disclosure of actual
performance. This is consistent with the argument
presented by Neu, Warsame, and Pedwell (1998),
who, rather than looking at changes in environ-
mental disclosure, document that di?erences in the
extent of information provided for a sample of 33
Canadian ?rms can be seen as attempts at man-
aging corporate relations with di?ering relevant
publics. Thus, rather than representing meaningful
attempts at communicating real environmental
information, Neu et al. (1998, p. 268) suggest
‘‘these discourses o?er organizations a way to
sustain organizational legitimacy without neces-
sarily (their emphasis) changing organizational
modes of operation.’’
A key premise underlying the criticisms of
environmental disclosure outlined above is the
substantial body of evidence suggesting that cor-
porate environmental disclosure does not correlate
(or, as in Patten, 2002b, correlates negatively) with
actual environmental performance. For example,
three early studies of the disclosure/performance
relation (Ingram & Frazier, 1980, Wiseman, 1982
and Freedman & Wasley, 1990), examined whe-
ther the extent of ?nancial report environmental
disclosure was related to indices of environmental
performance issued by the Council on Economic
Priorities (CEP) in the early to mid 1970s. Each of
these studies examined ?nancial reports from the
year corresponding to the year of the CEP report,
and each used a content analysis scheme to mea-
sure the extent of disclosure. In essence, content
analysis awards points based on the presence, and
in some cases (Wiseman, 1982; Freedman &
Wasley, 1990) the degree of speci?city, of disclo-
sure of various items of environmental concern.
The scales used in these studies were fairly broad
with 13 items, encompassing topics addressing
economic, litigation, pollution abatement, and
general environmental areas. None of these early
studies found a signi?cant correlation between
disclosure and performance.
A more recent study of the relation between
environmental performance and environmental
disclosure, Fekrat et al. (1996), is based on 1991
annual report disclosures and uses a di?erent
measure of environmental performance than used
in the earlier studies. Fekrat et al.’s measure, al-
though obtained from the CEP, was based not on
CEP-based analyses, but on ?rm-reported toxic
emissions data required through the Emergency
Planning and Community Right-to-Know Act of
1986. These ?rm speci?c data were made available
in 1990 for 1988 calendar year emissions. The CEP
reported the data adjusted for ?rm size, and Fek-
rat et al. then ranked the companies based on these
adjusted measures. Fekrat et al. employed the
content analysis scheme developed by Wiseman
(1982) to quantify the environmental disclosures
for a sample of 26 of the largest US ?rms, and then
used the rankings based on these scores as the
disclosure index. Correlations of the rankings on
performance and disclosure were negative but not
statistically signi?cant.
Finally, Patten (2002b) also used toxic emis-
sions data as his measure of environmental perfor-
mance. Based on a sample of 131 US companies,
Patten documented that, when size and industry
classi?cation are controlled for, both line count
and content analysis measures of ?nancial report
environmental disclosure are signi?cantly, and
negatively, associated with pollution performance.
Although the content scheme utilized by Patten
(2002b) was less broad than those used in previous
studies (e.g., it did not include litigation disclo-
sures), it still covered eight di?erent disclosure
topics.
In summary, ?nancial report environmental
disclosure has been widely criticized as being
misleading because companies appear to increase
the provision of positive disclosure in response to
increased exposures facing the ?rm and because
this disclosure does not appear to be an accurate
measure of ?rm environmental performance.
However, there are at least two issues that need to
be addressed before this disclosure can be com-
pletely condemned as meaningless with respect to
assessments of corporate environmental actions.
First, a potential problem is the earlier studies’
focus on broad measures of environmental dis-
closure. Hidden within those broader measures
may be pieces of meaningful information. Second,
while the disclosure examined previously may not
correspond well with past environmental perfor-
mance, it, or at least parts of it, may still provide
D.M. Patten / Accounting, Organizations and Society 30 (2005) 457–468 459
information for assessing future environmental
action. This study attempts to address these
shortcomings by examining one speci?c, but
potentially useful, category of environmental dis-
closure; projections of future spending for pollu-
tion abatement and control equipment.
As noted by Zuber and Berry (1992, p. 46), the
Securities and Exchange Commission (SEC) ‘‘re-
quires companies to disclose the material e?ects
compliance with federal, state, and local environ-
mental laws may have on capital expenditures’’ and
‘‘to reveal the material estimated capital expendi-
tures for environmental control facilities for the
current ?scal year and such further periods as the
registrant may deem material.’’ And although only
a limited number of companies actually provide
speci?c disclosure in these areas (see, e.g., Blac-
coniere & Patten, 1994; Gamble, Hsu, Kite, &
Radtke, 1995), projections of future environmental
capital expenditures would appear tobe anexcellent
item to examine relative to identifying whether
disclosures are meant to be useful or misleading. As
argued by Ingram and Frazier (1980, p. 614), ‘‘for
the disclosures to be useful, there should be a cor-
respondence between the disclosures and actual
events.’’ In this case, it would be expected that, if the
information is intended to be meaningful disclosure
of future environmental actions, the projections
should, across companies and over time, closely
mirror subsequent expenditures. However, because
the disclosures are only projections and not actual
costs, they could be used to merely project an image
of pending environmental action that is not subse-
quently carried out. If this were indeed the case,
subsequent actual expenditures would be expected
to be signi?cantly lower than the projections.
Methods and results
This study focuses on 10K report environmen-
tal disclosures by US ?rms from the chemical,
petroleum, metals processing, and paper indus-
tries. These industries have been cited often as
being subject to greater environmental scrutiny
than other industry segments (see, e.g., Cowen,
Ferreri, & Parker, 1987; Patten, 1991, 2002;
Hackston & Milne, 1996). Firms in these industries
also consistently have been shown to exhibit
higher levels of ?nancial report environmental
disclosure than ?rms from less environmentally
sensitive industries (Cowen et al., 1987; Patten,
1991, 2002; Hackston & Milne, 1996). All 10K
reports available on the SEC EDGAR database
for ?rms in the four noted industries for reporting
years 1993 through 2002
3
were reviewed. The
examination centered on identifying disclosures on
capital expenditures, both projected and actual,
for pollution abatement and control. Each obser-
vation set included in the analysis required two
pieces of environmental disclosure. First, a speci?c
dollar amount projection for environmental capi-
tal expenditures had to be made in one year (des-
ignated year i), and second, a speci?c dollar
amount of actual environmental capital expendi-
tures had to be disclosed the following year (des-
ignated year i þ 1). Overall, the 10Ks for 119
di?erent companies were examined.
4
Of these, 61
?rms had at least one set of projected (year i) and
actual (year i þ 1) environmental capital expendi-
ture data disclosed, and 355 observation sets, in
total, were collected. Table 1 summarizes the
source of the data, by industry.
The accuracy of the environmental capital
expenditure projections was calculated by com-
paring the actual expenditure in year i þ 1 with the
projection from the previous year. However, be-
cause the size of the companies making environ-
mental capital expenditure disclosures varied
considerably (as did the size of the expenditures), a
projection error statistic adjusted for size was
computed. That is, projection errors were calcu-
lated as:
ðActual
iþ1
þ Projection
i
Þ=Projection
i
where Actual
iþ1
is the actual environmental capital
expenditure amount as disclosed in year i þ 1, and
3
The SEC ?rst began accepting electronic ?ling of 10K
reports (which are then made available through the EDGAR
database) in 1994. Thus, the ?scal year 1993 reports are the
oldest available through the EDGAR source.
4
Not all 119 companies had 10K reports available for all ten
years (1993–2002). Among other reasons, some companies
began electronic ?ling in later periods, and some ?rms merged
or were taken over by other companies.
460 D.M. Patten / Accounting, Organizations and Society 30 (2005) 457–468
Projection
i
is the projection for year i environ-
mental capital expenditures as disclosed in year i.
Table 2 presents results on the general distri-
bution (negative, exact, or positive) of the projec-
tion errors (Panel A), the mean projection error
amounts by industry (Panel B), and the mean
projection error amounts by year of actual
expenditure (Panel C). Fig. 1 provides a more de-
tailed presentation of the distribution of projection
errors. As noted in Panel A of Table 2, more than
three-fourths of the projection errors (76.1%) were
negative. This means that, for the vast majority of
observation sets, the actual environmental capi-
tal expenditure in year i þ 1 was less than had
been projected in year i. If it is assumed that there
is an equal chance of meeting (exact or positive
Table 1
Summary of the source of sample comparison sets of observa-
tions (environmental capital expenditures year
iþ1
and environ-
mental capital expenditure projection year
i
) by industry
Industry Total
number of
companies
examined
Companies
with at least
one compar-
ison set of
observations
Total
number of
observation
sets
Chemical 49 21 123
Metals 18 11 69
Petroleum 33 20 107
Paper 19 9 56
Total 119 61 355
All 10K reports available on the SEC EDGAR database for
years 1993 through 2002 for each of the 119 companies were
examined for environmental capital expenditure disclosures.
Table 2
Summary of 1-year projection errors [(actual environmental capital expenditures year i þ 1 minus projected environmental capital
expenditures year i) divided by projected environmental capital expenditures year i]
Panel A
Distribution of errors
Negative projection errors
(actual projection) 79 22.2%
Panel B
Mean projection errors by industry
Industry Observation sets Mean error (%)
Chemical 123 )16.6
Metals 69 )22.5
Petroleum 107 )11.7
Paper 56 )17.7
Total 355 )16.4
Panel C
Mean projection errors by year
(of actual expenditure)
Year
1994 35 )14.3
1995 46 )24.6
1996 49 )21.9
1997 50 ) 9.6
1998 44 )13.7
1999 39 )19.9
2000 36 )12.0
2001 28 )14.0
2002 28 )16.1
Total 355 )16.4
a
Binomial probability of this percentage of negative errors -75% -60 to-
75
-45 to -
60
-30 to -
45
-15 to -
30
0 to -15 exact 0 to +15 +15 to
+30
+30 to
+45
+45 to
+60
+60 to
+75
> +75%
Projection Error Amount
1 Year Projection Errors
Fig. 1. Distribution of 1-year projection errors.
462 D.M. Patten / Accounting, Organizations and Society 30 (2005) 457–468
for estimated environmental capital expenditures
for each of the next 2 years (designated year i þ 1
and year i þ 2), and (2) disclosure of actual
spending in the two subsequent years (year i þ 1
and year i þ 2). Far fewer companies made 2-year
projections. In total, only 110 observation sets
were available. Table 3 presents the results of the
2-year comparisons while Fig. 2 provides a more
detailed presentation of the distribution of 2-year
projection errors. Similar to the results for the one-
year analysis, the vast majority of the projection
errors (78.2%) are negative. The binomial proba-
bility of such a distribution of negative errors is
again less than 0.001, two-tailed. The mean 2-year
projection error is a slightly smaller )14.6%. As
highlighted in Fig. 2, the distribution of projection
errors is very similar to the one-year errors, with a
large number of observations between )15% and
)75%. This additional analysis, therefore, appears
to support the ?nding that environmental capital
spending projections lack accuracy.
5
A second additional analysis examines com-
pany projections for total capital expenditures.
Rather than being misleading disclosure, the
inaccuracy of environmental capital expenditure
projections could be a function of the di?culty in
projecting spending on capital assets. If so, a
comparison of actual capital expenditures in total
with previous year projections would be expected
to yield a distribution of inaccuracy similar to that
for environmental capital expenditures only. A
review of those 10K reports from which the envi-
ronmental capital expenditure observation sets
were drawn resulted in 193 sets of projection/ac-
tual data for total capital expenditures.
Panel A of Table 4 presents the results of the
comparisons of total capital expenditures with
their previous year projections, while Panel B of
Table 4 presents the corresponding environmental
capital expenditure comparison results for these
company/years only. Fig. 3 presents a comparison
of the distribution of projection errors for the total
capital expenditures and the corresponding envi-
ronmental capital expenditures. As highlighted in
Panel A of Table 4, almost exactly half of the total
capital expenditure projection errors are positive
(50.3%) and almost half (49.1%) are negative. The
binomial probability of observing this level of
negative errors is 0.886, two-tailed. In contrast,
and similar to the distribution for the total sample,
77.7% of the projection errors for environmental
capital expenditures are negative. The binomial
probability for this distribution is less than 0.001,
two-tailed. Further, the mean projection error for
the total capital expenditure amounts is a positive
4.0% in contrast to a mean error of minus 17.5%
for the corresponding environmental capital
expenditures. Fig. 3 highlights that while the dis-
tribution of environmental capital expenditure
projection errors is again heavily weighted toward
the )15% to )75% range, the error terms for the
total capital expenditure projections are much
more closely distributed around zero. These results
suggest that, on average and across time, the
projections for total capital expenditures are rea-
sonably accurate, while projections for environ-
mental capital expenditures are not.
6
Finally, this analysis also identi?es whether
the environmental capital expenditure projection
Table 3
Summary of 2-year projection errors
Negative projection errors
(actual projection) 23 20.9%
Total 110 100%
Mean 2-year projection error )14.6%
a
Binomial probability of this percentage of negative errors
-75% -60 to-
75
-45 to -
60
-30 to -
45
-15 to -
30
0 to -15 exact 0 to +15 +15 to
+30
+30 to
+45
+45 to
+60
+60 to
+75
> +75%
Projection Error Amounts
2 Year Projection Errors
Fig. 2. Distribution of 2-year projection errors.
7
Pearson product-moment correlation coe?cients for the
relation between the environmental capital expenditure projec-
tion errors and the change in revenues and change in return on
assets are 0.063 and 0.075, respectively. Neither measure is
statistically signi?cant at conventional levels (p < 0:10, two-
tailed).
464 D.M. Patten / Accounting, Organizations and Society 30 (2005) 457–468
0
10
20
30
40
50
60
70
> -75% -60 to-
75
-45 to -
60
-30 to -
45
-15 to -
30
0 to -15 exact 0 to
+15
+15 to
+30
+30 to
+45
+45 to
+60
+60 to
+75
>
+75%
Projection Error Amounts
Comparison of Projection Errors for Environmental Capital Expenditures versus Total Capital
Expenditures
Envir. Cap. Exp.
Total Cap. Exp.
Fig. 3. Comparison of distribution of total capital expenditure projection errors with environmental capital expenditure projection
errors.
Table 4
Summary of projection errors for companies with data on total capital expenditures
Panel A
Total capital expenditure projection errors
Negative projection errors
(actual projection) 93 48.2%
Total 193 100%
Mean projection error + 4.0%
Panel B
Environmental capital expenditure projection errors (companies with total capital expenditure data)
Negative projection errors
(actual projection) 40 20.7%
Total 193 100%
Mean projection error )17.5%
a
Binomial probability of this percentage of negative errors ¼0.886, two-tailed.
b
Binomial probability of this percentage of negative errors