NCEE#
NATIONAL CENTER FOR
ENVIRONMENTAL ECONOMICS
A Change of PACE:
Comparing the 1994 and 1999
Pollution Abatement Costs and Expenditures Surveys
Randy A. Becker and Ronald J. Shadbegian
Working Paper Series
Working Paper # 04-08
June, 2004
>• ST
3 2 National Center for Environmental Economics
LU
U.S. Environmental Protection Agency
1200 Pennsylvania Avenue, NW(MC 1809)
pro^&< Washington, DC 20460
http://www.epa.gov/economics
-------
A Change of PACE:
Comparing the 1994 and 1999
Pollution Abatement Costs and Expenditures Surveys
Randy A. Becker and Ronald J. Shadbegian
Correspondence:
Ronald J. Shadbegian
UMass Dartmouth
Department of Economics
North Dartmouth, MA 02747
email: rshadbegian@umassd.edu
508-999-8337
and
U.S. Environmental Protection Agency
National Center for Environmental Economics
email: shadbegian.ron@epa.gov
NCEE Working Paper Series
Working Paper # 04-08
June, 2004
DISCLAIMER
The views expressed in this paper are those of the author(s) and do not necessarily represent
those of the U.S. Environmental Protection Agency. In addition, although the research described
in this paper may have been funded entirely or in part by the U.S. Environmental Protection
Agency, it has not been subjected to the Agency's required peer and policy review. No official
Agency endorsement should be inferred.
-------
A Change of PACE:
Comparing the 1994 and 1999
Pollution Abatement Costs and Expenditures Surveys
Randy A. Becked and Ronald J. Shadbegianb'°
June 2004
Abstract
Since 1973, the Pollution Abatement Costs and Expenditures (PACE) survey has been
the principle source of information on U.S. industries' capital expenditure and operating
costs associated with pollution abatement efforts. The PACE survey was discontinued
after 1994 and then revived in 1999 for one year - in a substantially different form than
the preceding surveys however, making longitudinal analysis quite difficult. Conceptual
differences include matters as fundamental as the scope and meaning of pollution
abatement as well as the definition of operating costs. A number of other critical
changes also exist, including ones of industrial coverage and sample selection. This
paper is the first comprehensive effort to document the many changes in the PACE
survey across these years and to provide a detailed guide for researchers and
policymakers who wish to compare the 1994 and 1999 data. Overall, we find a 27%
decline in environmental spending by the manufacturing sector between these two years,
though there appears to be significant heterogeneity across industries. We discuss
potential reasons for this dramatic decline, focusing mainly on issues of survey
methodology and design. This paper should help inform current efforts to redevelop the
PACE survey and re-establish it as a regular, annual survey.
Subject Area: Costs of Pollution Control
Keywords: pollution abatement, survey data, environmental costs, manufacturing
a Center for Economic Studies, U.S. Bureau of the Census
b Department of Economics, University of Massachusetts at Dartmouth
0 National Center for Environmental Economics, U.S. Environmental Protection Agency
The authors have benefited from the helpful comments of Jeffrey Fritzsche, Wayne Gray,
and Kelly Maguire, as well as participants at the 2004 Allied Social Science Associations
meetings and Statistics Canada's Economic Conference 2004. The opinions and
conclusions expressed herein are those of the authors and do not necessarily represent the
views of the Bureau of the Census or the Environmental Protection Agency. All errors
are our own.
-------
I. Introduction
Since 1973, the Pollution Abatement Costs and Expenditures (PACE) survey has
been the principle source of information on U.S. industries' capital expenditure and
operating costs associated with pollution abatement efforts. The data published from this
survey (e.g., U.S. Bureau of the Census, 1994) are widely used by government agencies,
academic researchers, and industry to estimate the costs of environmental regulations and
analyze their effects. For budgetary reasons the PACE survey was discontinued after
1994 and then revived for just one year - in 1999 - in a substantially different form than
the preceding surveys.1 These 1999 data were released in November 2002.2
Our study is the first comprehensive effort to document the many conceptual
changes in the PACE survey between 1994 and 1999. It is extremely important to note
that many of these changes make it impossible to directly compare even such "core" data
items as pollution abatement operating costs. As we will make clear, things similarly
named are not necessarily similarly defined - a fact that is not necessarily apparent from
the 1999 PACE publication (U.S. Bureau of the Census, 2002). Here, we provide a
detailed guide - with appropriate cautions and caveats - for researchers and policymakers
who wish to utilize the 1999 data in a longitudinal context. Our paper also discusses a
number of critical measurement issues and challenges, including ones of survey
methodology and design. This study should help inform current efforts to redevelop the
PACE survey and re-establish it as a regular, annual survey.
The paper proceeds as follows. In Section II, we briefly highlight the importance of
the PACE survey and discuss just some of the many ways that its data have been used. In
1 Note that a survey was also not conducted for reference year 1987.
2 See Iovanna et al. (2003) for a history of the 1999 PACE survey.
1
-------
Section III of the paper, we detail many of the important changes in the survey between
1994 and 1999. Here we highlight the changes in form design that directly affect data
comparability. It is true that many expenditure items were collected in a substantially
different manner on the 1999 survey than on previous surveys. For example, some costs
were no longer disaggregated by pollution media (i.e., air, water, solid waste, etc.) and/or
separated into capital expenditures and operating costs. Some new data items were
introduced in 1999; other long-standing ones were eliminated however, such as capital
depreciation costs and cost offsets. In Section III, we also discuss the critical conceptual
differences between the recent survey and its predecessors. Perhaps most important,
pollution abatement was fundamentally redefined to exclude pollution prevention,
recycling, and disposal - leaving just the treatment of pollution. Pollution abatement also
excluded related administrative activities as well as monitoring & testing. Additionally,
operating costs were redefined to exclude capital depreciation and permit-related
expenses. Finally, there were notable differences in sample selection and industrial
coverage that also affect the comparability of the 1999 PACE estimates with those from
prior years. Most significant were the addition of establishments with fewer than 20
employees, the switch in the U.S. industrial classification system, and changes in the way
in which data from the non-manufacturing sector were collected and presented.
Nevertheless, with appropriate aggregations of and adjustments to the published
PACE statistics from the respective years, a comparison of pollution abatement spending
in 1994 and 1999 can be made, albeit cautiously. This is the subject of Section IV.
Overall, we find a 27% decline in environmental spending by the manufacturing sector
between these two years. However, there appears to be some significant heterogeneity
2
-------
across industries. For example, we find declines in expenditures of 16% and 60% in the
plastics and petroleum industries, respectively, but increased expenditure among steel
mills and pulp mills.
Section V discusses some of the potential reasons for the dramatic changes in
environmental spending between these two years, focusing mainly on issues of survey
methodology and design relating to the 1999 version of the PACE. In particular, we
believe that limited data editing, conservative imputation, altered questionnaire design,
less-than-explicit instructions, and some other factors may have resulted in 1999
expenditures that were too low. Finally, we conclude in Section VI with some
recommendations for any future PACE surveys.
II. Importance of the PACE survey
The importance of the PACE survey to policymakers is perhaps best summarized
in a January 15, 2002 letter from the Environmental Economics Advisory Committee
(EEAC) of the EPA's Science Advisory Board to Governor Christie Whitman, then
Administrator of the U.S. Environmental Protection Agency (EPA):
The PACE Survey data provide a truly unique tool for evaluation
of the costs of compliance with environmental regulations. The
collection of these data has provided the United States with an
important source of information to facilitate the evaluation of
environmental programs and, in turn, to improve the design and
performance of these programs. EPA has used the PACE data in
its Cost of Clean reports, the Section 812 Clean Air Retrospective
Cost Analysis, numerous sector-specific studies, Regulatory
Impact Analyses, analyses of recycling activities, and national
studies of environmental protection activities. The relatively low
cost of the PACE Survey, combined with its great benefits to
EPA, means that the annual Survey provides the Agency with a
tremendous return on its investment.
3
-------
Furthermore the EEAC noted:
At once, the PACE Survey provides a means to assess the costs of
environmental regulations in aggregate and individually, and it
provides important data with which to compare the cost-
effectiveness of alternative regulatory approaches. The systematic
collection of information on these costs of regulation is essential
to meet expanding legal requirements for review of the costs of
regulation, and it is important for EPA's efforts to develop sound
and effective regulations.
Over the years, a number of academic studies have used data from the PACE
survey to examine how environmental regulation impacts U.S. manufacturing. Among
these studies are those that have used the published aggregate statistics (i.e., data at the
state-level or by industry) and those that have used the plant-level microdata.3 Among
the former, Gray (1986, 1987) examined the impact of environmental regulation on the
productivity of 450 manufacturing industries and found that industries with high
pollution abatement costs had larger productivity slowdowns in the 1970s. Barbera and
McConnell (1987, 1990) also used industry-level PACE data and found that the costs
associated with environmental regulation reduced productivity in three of the four
industries in their 1987 study, and in all five industries in their 1990 study. More
recently, Levinson (2001) used published PACE data to construct an industry-adjusted
index of state environmental compliance costs and found that states with apparently more
stringent environmental regulations experienced less foreign direct investment in the
chemical manufacturing industry.
In terms of studies that have used the plant-level data from the PACE survey,
Gray and Shadbegian (1998) examined the impact of investment in pollution abatement
3 The establishment-level survey data are confidential, collected and protected under Title 13 of the U.S.
Code. Restricted access to these data can be arranged through the U.S. Census Bureau's Center for
Economic Studies. See http://www.ces.census.gov/ for details.
4
-------
capital on the investment in "traditional" capital (used to produce output). They find that,
among pulp and paper mills, expenditure on the former in fact "crowds out" investment
in the latter almost dollar-for-dollar. In another study, Gray and Shadbegian (2002)
found that manufacturing plants in the paper, steel, and oil industries had lower
productivity levels when their environment-related costs were higher. In contrast,
Berman and Bui (2001) found little effect of regulatory costs on the productivity of oil
refineries.4
Becker (2001) has also used establishment-level data from the PACE survey, to
examine the effects of the Clean Air Act on air pollution abatement (APA) capital
expenditure and operating costs. As might be expected, he found that heavy emitters of
the "criteria" air pollutants located in stringently-regulated NAAQS non-attainment
counties generally had higher APA outlays. This study also revealed potential
shortcomings in the PACE survey data. In a more recent study, Becker (2003) found that
certain community characteristics had additional effects on the APA expenditures of
nearby polluters, over and above formal regulatory requirements.
III. Changes in the PACE survey
In this section, we discuss many of the important changes to the PACE survey
between the version in the early-1990s and that in 1999.
Scope of the survey
In terms of sample size, the manufacturing component of the survey is somewhat
4 This result is at least somewhat consistent with the results of Gray and Shadbegian (2002) in that, of the
three industries they studied, the impact of pollution abatement costs on productivity was the smallest for oil.
5
-------
larger in 1999. The 1994 PACE publication reports that 17,800 manufacturing plants
were sampled, while in 1999 well over 19,000 were selected. In both instances, the
sampling frame consisted of establishments in the prior Census of Manufactures (i.e.,
1992 and 1997, respectively) along with plants opening in the intervening years (i.e.,
1993 and 1998). Note that up to and including 1993, the PACE sample had been a strict
subset of the contemporaneous Annual Survey of Manufactures (ASM).
One important difference in sampling between the PACE survey of 1999 and those
since 1976 is that the 1999 sampled establishments with fewer than 20 employees. These
had previously been deemed a relatively insignificant portion of the universe of polluting
plants. In particular, according to PACE publications, "early surveys showed that [these]
establishments contributed only about 2 percent to the pollution estimates while
constituting more than 10 percent of the sample size." Since the 1980 PACE, "no
adjustment is made to account for these establishments." Below, we will examine
whether this group is still a small percentage of total pollution abatement expenditures.
The 1999 sample also differs from the 1994 sample in terms of industrial coverage.
Some of the differences are associated with the replacement, in 1997, of the Standard
Industrial Classification (SIC) system with the North American Industrial Classification
System (NAICS) and therefore are not unique to the PACE survey. In addition to the
significant re-categorization of production activity within manufacturing, which alone
makes cross-year comparisons difficult, some activity previously defined as
manufacturing is now classified as non-manufacturing and vice versa. Table A-l (in
Appendix A) shows industrial activity that has been removedfrom manufacturing under
NAICS, while Table A-2 shows industrial activity that has been moved into
6
-------
manufacturing under NAICS. Details on within-manufacturing changes can be found at
http://www.census.gOv/epcd/ec97brdg/INDXNAI3.HTM#31-33.
Another difference in industrial coverage between previous versions of the PACE
and the 1999 survey is that the 1999 PACE surveyed plants in the apparel industry
(formerly SIC 23). According to PACE publications, the reason for their prior exclusion
was that "these establishments operate primarily in rented quarters where the abatement
of pollution (probably most of which is solid waste) is generally arranged by the landlord.
Capital expenditures for pollution abatement in such establishments are probably
minimal." And indeed, in 1999, NAICS 315 - which accounts for about 70% of the
output of what had been SIC 23 - had a negligible amount of capital expenditure and no
more than $8 million of operating costs.5 Table A-3 shows the NAICS industries that
correspond to the former SIC 23.
The 1999 PACE also canvassed certain non-manufacturing establishments, in
particular those engaged in mining (NAICS 21) and electric power generation (NAICS
22111). This is somewhat of a break from the past. From 1988 to 1994, the Census
Bureau also collected data on pollution abatement expenditures from mining companies
and electric utilities, and statistics for these industries were presented in the PACE
publications. However, the survey instrument used (Form PA-2) was a supplement to the
Census Bureau's Plant and Equipment Survey and was somewhat different from the
PACE survey that manufacturing plants received (Form MA-200). Most importantly, it
did not inquire about pollution abatement operating costs — it only asked about capital
expenditures. Another important difference is that the PA-2 was sent to firms primarily
engaged in those activities, not establishments. A firm-level survey may yield
7
-------
substantially different estimates than an establishment-level survey if firms have a large
amount of "secondary" activity — that is, for example, mining firms with non-mining
ventures.6 For these and other reasons, we will not focus on PACE'S non-manufacturing
industries in this paper.
One extremely important caution we would like to convey here, for those interested
in comparing 1999 PACE expenditures to those in prior years, is that "All industries" in
the tables of previous PACE publications denotes just the manufacturing industries (less
SIC 23), while "All industries" in the 1999 publication implies all manufacturing
industries as well as the mining and electric power generation industries.7 The
manufacturing-only data begin farther down the 1999 tables, with NAICS 31-33. Failing
to realize this can be critical. Table 1 summarizes this as well as the other major "scope"
issues faced by those wishing to compare 1999 PACE data to the older PACE data.
Concepts and definitions
In addition to these changes in industrial coverage and sample selection, there were
significant changes in the data that the PACE survey intended to capture. This is true for
5 Note that 16% of what is NAIC S 315 came from somewhere other than SI C 23.
6 This can actually be seen in the 1994 data. The PA-2 was also sent to firms primarily engaged in
petroleum and coal products (SIC 29), which is also in scope to the establishment-level PACE survey (MA-
200). The estimates based on the firm-level survey suggest that SIC 29 had $4.7 billion of capital
expenditures for the abatement of air, water, solid waste, nonmedia and other pollutants. The PACE
survey, on the other hand, suggests that this industry had just $2.6 billion of such expenditures.
7 Of course, neither captures the pollution abatement expenditures of truly all industries since there are
polluting industries that have been out-of-scope to the PACE survey, then and now. For example, among
the excluded sectors are construction, agriculture (e.g., crop production, hog farms), wholesaling (e.g.,
petroleum, chemicals), retail (e.g., gasoline stations), transportation (e.g., trucking), services (e.g., dry
cleaning, auto repair, hospitals, universities), utilities (e.g., landfills, incinerators), and government-owned
enterprises that are out-of-scope to Census Bureau surveys.
8
-------
Table 1
Summary of main scope issues
(1) Estimates for "All industries" in the 1994 tables are for
manufacturing only, while "All industries" in 1999 also
encompasses mining and electric power generation. Estimates
for just manufacturing (NAICS 31-33) begin farther down the
1999 tables.
(2) Since 1973, the apparel industry (SIC 23) had been excluded
from the survey. This industry was included again however in
the 1999 survey. (See Table A-3.)
(3) With the replacement of the SIC system with NAICS, some
industrial activities are no longer classified as manufacturing
(see Table A-l), while others are newly classified as
manufacturing (see Table A-2).
(4) Since 1976, establishments with fewer than 20 employees had
been excluded from the PACE survey, and no adjustments
were made for their missing expenditures after 1980. In 1999,
these established were once again included in the PACE
sample.
even the most fundamental data items, such as pollution abatement operating costs and
pollution abatement capital expenditures. We now attempt to explain the various changes
in these two core concepts.
Operating costs
In its previous incarnations, the PACE survey opened with a stand-alone question
on pollution abatement operating costs (see Item 3 in Appendix B), where abatement was
broadly defined to include techniques in pollution prevention, recycling, treatment,
collection, and disposal, as well as related administrative activities and monitoring &
testing. Operating costs were to be disaggregated into 3 pollution media - air, water, and
solid/contained waste - and further disaggregated into 5 categories of costs (within each
media): depreciation, salaries & wages, fuel & electricity, contract work/services, and
9
-------
materials/leasing/miscellaneous.8 Item 4 of the survey then asked respondents to estimate
the percent of the air, water, and solid waste totals in Item 3 directed toward the
abatement of hazardous air, water, and solid waste, respectively. Item 5 inquired about
payments to governments (federal, state, county, or local) for pollutant removal, namely
for: (a) sewage services9 and (b) solid/contained waste collection & disposal.10 Operating
costs associated with the prevention, replacement, removal, and monitoring of (inferior or
leaking) underground storage tanks (USTs) were to be reported exclusively in Item 9.
Likewise, operating costs specifically associated with site cleanup were to be reported in
Item 10. Finally, operating costs for the abatement of "other" pollutants (namely
radiation, multimedia pollutants, and noise "that would otherwise disturb the surrounding
community") were captured in Items 11 and 12.
In the 1999 PACE, operating costs were collected in a much different manner (see
Item 2B in Appendix C). As in prior PACE surveys, data were collected separately for
air, water, and solid waste. However, multi-media also appears here, as its own category,
which had not been the case in prior surveys. In addition, the 1999 PACE recognized the
disposal of solid waste and the recycling of solid waste (see Item 2C-1 and Item 2C-2) as
distinct from the abatement of solid waste (Item 2B-3). It is very important to note that
these three had been just a single expense category prior to this survey.11
8 The grouping of expenses in these last three categories has changed over the years. Early on, equipment
leasing was its own category, with "materials, supplies, services, and other costs" as another. Later,
"materials, supplies, fuel, and electricity" was a category along with "services, equipment leasing, and
other costs." Meanwhile, there were always separate categories for depreciation and labor.
9 Except "sanitary" sewage. However, if expenditure on such sewage could not be separated from
expenditure on "industrial" sewage, respondents were to report the entire amount.
10 Except the collection & disposal of office and cafeteria trash. Again, if this could not be separated from
expenditure on "industrial" solid waste, then the entire amount was to be reported.
11 There appears, however, to be some difference in the definition of recycling. In particular, the 1999
survey explicitly states that the "burning of waste materials for fuel are not included in this category."
Furthermore, the 1999 PACE recognized a distinction between post-process and in-process recycling, the
10
-------
Another obvious difference is that the 1999 PACE collected operating costs in the
same item as capital expenditures, and as the percentages attributable to hazardous
pollutants, which had not been the case before. More importantly, respondents were no
longer asked to disaggregate operating costs by category (i.e., depreciation, salaries &
wages, etc.). This is obviously a loss for researchers interested in these types of
expenditures. A more fundamental question is whether this change in any way impacts
the quality (and comparability) of the total operating cost numbers. After all, the same
question, asked two different ways, may yield different answers.
There are also some very significant changes in the definition of pollution
abatement operating costs between 1994 and 1999 that users of these data absolutely
need to be aware of. In particular, the 1999 survey employed a much narrower definition
of pollution abatement operating costs, along a number of dimensions. For example,
costs associated with the testing & monitoring of emissions, conducting environmental
audits & studies, developing pollution abatement operating procedures, completing
environmental reporting requirements, training, preparing documents related to
environmental protection, and other such administrative costs were no longer included in
the definition. Instead, in 1999, these were recognized as distinct environmental
activities and their expenditures were collected near the end of the survey form, in Items
4C (environmental monitoring & testing) and 4D (administration of environmental
programs) in particular. Note that respondents were not asked to disaggregate these
expenditures by pollution media (i.e., air, water, and solid waste). Furthermore, these
particular operating costs were lumped together with their respective capital
former being an end-of-line technique (captured in Item 2C-2) while the latter is a prevention technique
(captured in Item 3B).
11
-------
expenditures. (This fact alone complicates longitudinal comparisons with prior years, as
we shall see in Section IV.) In 1999, the manufacturing sector had $1.7 billion of
expenditure on these two "new" items, which previously would have been reported
among "traditional" pollution abatement operating costs or capital expenditures.
In addition, operating costs associated with pollution prevention techniques were
also no longer reportable among the "traditional" pollution abatement operating costs, as
they once had been.12 Instead, such outlays were to be reported in Item 3B of the 1999
survey, which (again) lumped together both capital expenditures and operating costs, as
well as lumping together the prevention of pollution of all media (see Appendix C).
Furthermore, the instructions on how to report this item were perhaps less explicit than
they could have been. To take one example, in the 1994 survey, respondents were
instructed to include the "incremental costs for consumption of environmentally
preferable materials and fuels" — that is, the cost of "cleaner" inputs minus the cost of
"dirtier" alternatives. No such guidance was offered in the 1999 survey, which leads one
to wonder whether costs such as these were captured or not.
Another very important conceptual change in operating costs is that the 1999
survey no longer deemed the depreciation of pollution abatement capital stock
(equipment and structures) an operating cost. And unlike the abovementioned costs,
depreciation was not collected elsewhere on the 1999 PACE survey. In 1994, the
depreciation of pollution abatement capital in the manufacturing sector totaled almost $3
12 It should be noted that these issues with pollution prevention, administration, monitoring & testing,
disposal, and recycling have more to do with a revised definition of pollution abatement than a changed
definition of operating costs. In particular, for the 1999 survey, pollution abatement was redefined to
include only the treatment of pollutants after their generation (not including recycling and disposal), while
previously, prevention techniques leading to the reduction and elimination of pollution generated had also
been deemed abatement, as had the treatment, recycling, and disposal of pollutants, as well as related
administrative activities and monitoring & testing.
12
-------
billion, representing over 15% of total pollution abatement operating costs.13 Adjusting
for such a major definitional difference (in addition to the others above) is obviously very
important when making cross-year comparisons, as we do in Section IV.
Meanwhile, payments to governments for the removal and disposal of industrial
wastes, previously collected separately, were to be combined with other disposal
operating costs in Item 2C of the 1999 survey.14 Costs associated with leaking or inferior
underground storage tanks (USTs) were again a separate line item in the 1999 survey
(Item 4A-2), however it very likely did not include anything beyond their removal and
replacement. In contrast, Item 9 on the 1994 PACE survey also included the monitoring
of USTs as well as prevention activities (i.e., the "installation of safeguards for existing
underground tanks"). In 1999, expenditures on the former were presumably included in
the new item specifically concerned with environmental monitoring (Item 4C) and the
latter were included in pollution prevention (Item 3B). In any event, in 1999, operating
costs related to USTs were combined together with UST capital expenditures, where
previously they were reported separately. Likewise, operating costs and capital
expenditures associated with site cleanup were also combined together, where previously
they had been distinct. However, in 1999, the operation and maintenance of Superfund
sites was separated from other types of site cleanup (Item 4A-1 and Item 4A-3,
respectively), where previously they were combined. Again, cleanup-related testing &
monitoring, compliance auditing, environmental studies, and other administrative costs -
which previously were included in these operating costs - were imbedded in other items
in the 1999 survey.
13 This percentage had been even higher in the past, peaking at over 24% in 1979.
14 Note that the "payments to governments" item on the 1999 PACE survey (i.e., Item 7A) has absolutely
13
-------
Finally, in the 1994 survey, operating costs explicitly included "permits and costs
incurred obtaining permits."15 In the 1999 survey, costs incurred in obtaining permits
were likely included among administrative costs (Item 4D) while the permits themselves
were captured in Item 7A-1 (i.e., payments to government through permits, fees, and
charges) or Item 7B-1 (tradable permits bought). The 1999 survey also included
questions on payments to governments through fines and penalties (Item 7A-2) and
"other expenditures as a result of penalties, such as payments for supplemental
environmental projects" (Item 7A-3). Presumably environmental taxes, including
Superfund contributions, were to be included in Item 7A-1, along with environmental
permits and fees. The 1994 survey, however, explicitly excluded "taxes, fines, legal fees,
and Superfund taxes and contributions" from its definition of operating costs. This is yet
another reason why it will not be possible to get fully conformable definitions of costs
across these two years.
Capital expenditures
In terms of pollution abatement capital expenditures, the 1994 version of the PACE
survey contained a stand-alone question (see Item 7 in Appendix B) asking respondents
to report such expenditures by 3 pollution media - air, water, and solid/contained waste -
and by 2 types of capital: the end-of-line variety and that associated with production
process enhancements, replacements, additions, and alterations. End-of-line (EOL)
capital treats pollutants after their generation,16 while production process enhancements
nothing in common with the "payments to governments" questions on the older PACE surveys.
15 Prior to 1993, however, permit costs were excluded from the PACE'S definition of operating costs.
16 Examples include: dust collectors, scrubbers, and precipitators for air pollution abatement; trickling
filters, settling ponds, clarifiers, and spill containment dikes for water pollution abatement; cardboard
balers, compactors, bins, improved waste containments, and incinerators for solid/contained waste. (Taken
from page 11 of the 1994 PACE instruction booklet.)
14
-------
(PPE) capital is capital that is embedded in production process changes meant to reduce
or avoid the generation of pollutants.17 As with operating costs, respondents were asked
to estimate the percent of expenditures attributable to the abatement of hazardous air,
water, and solid waste, respectively. Item 8 further inquired as to the proportion of non-
hazardous air pollution abatement capital expenditure devoted toward the six "criteria"
air pollutants covered under the Clean Air Act versus other non-hazardous air pollutants.
Finally, as was the case with operating costs, capital expenditures on USTs, site cleanup,
and the abatement of "other" pollutants (i.e., multi-media, noise, and radiation) were
reported separately in Items 9 through 11, respectively.
In the 1999 PACE survey, capital expenditures were collected in a substantially
different fashion, and much of the preceding discussion on the collection of operating
costs applies here as well. (As such, we will recap the prior discussion only briefly.)
Again, recalling Item 2B, data were collected separately for air, water, and solid waste, as
well as a new independent multi-media category. And, again, the disposal and recycling
of solid waste were recognized as distinct from its abatement/treatment - or from
prevention for that matter. Spending on USTs - with the exception of monitoring as well
as preventative measures - was collected in a separate item, but with no distinction
between operating costs and capital expenditures, as had been the case in prior surveys.
The same is true for site cleanup expenses.
Meanwhile, pollution monitoring & testing equipment - be it for air, water, solid
waste, underground storage tanks, multimedia, etc. - was reported in Item 4C
17 Examples include equipment and structures necessary for the conversion to substitute fuels that generate
fewer air pollutants, conversion to a closed or partially closed loop system for water pollution abatement, or
conversion enabling the recycling of scrap materials. Prior to 1992, this type of expenditure was called
change-in-productionprocess (CIPP) capital.
15
-------
(environmental monitoring & testing), where previously it had been included in the PPE
capital expenditure for the particular media. Furthermore, in 1999, such expenditure was
combined with monitoring & testing operating costs, severely complicating any
longitudinal data analysis. Similarly, the capital expenditures associated with the
administration of environmental programs (e.g., audits, studies, reporting, training, etc.)
were also now reported independent of media (Item 4D), though this change is much
more likely to have impacted the reporting of operating costs than capital expenditures.
In another obvious break from prior PACE surveys, the reporting of capital
expenditures for the treatment of pollutants was decoupled from expenditures on the
prevention of pollution, with Item 2 and Item 3, respectively. It is important to point out
here that one of the more nuanced differences between the 1999 PACE survey and what
came before is that the recent survey strictly defined pollution abatement as the treatment
of pollutants after their generation, excluding recycling and disposal, while process
changes leading to the reduction or elimination of pollution generation was labeled
pollution prevention. In contrast, in 1994, all types of capital expenditures - prevention,
treatment, recycling, disposal, and monitoring & testing - were defined as pollution
abatement. Treatment and disposal were deemed EOL techniques, while prevention,
(post- and in-process) recycling, and monitoring & testing were deemed PPE. Therefore,
the "pollution abatement capital expenditure" in Item 2B of the 1999 survey, together
with "disposal capital expenditures" of Item 2C-1, was meant to be similar (if not
identical) to the EOL capital expenditure in the older surveys. Meanwhile, in 1999, PPE
capital was captured in Item 2C-2 (post-process recycling capital), the Item 3B question
on pollution prevention, and the Item 4C question on monitoring & testing. Of course, in
16
-------
contrast to past surveys, the last two items lumped together capital expenditures and
operating costs and made no distinction between pollution media.
Perhaps more important however than how PPE/prevention capital expenditure was
reported across these years is what was to be reported. Such capital expenditures pose
unique reporting challenges, since the relevant costs are often commingled with capital
investment for non-environmental purposes.18 For example, the conversion to a
production process that uses more environmentally-friendly fuels or materials may
involve the installation of a new piece of production machinery with certain
environmental features imbedded in it. Since the earliest days of the PACE survey,
respondents were instructed to estimate the pollution abatement portion of this
investment as the extra cost of pollution abatement features: "i.e., your actual spending
less what you would have spent without the pollution abatement features built-in."
Presumably this became increasingly difficult to do as production equipment without said
pollution abatement features became rarer. Beginning with the 1992 PACE survey, the
following guidance was added to the instructions, to aid respondents facing such
difficulties:
Special instructions: Estimating the pollution abatement portion
(distinct from production efficiency, energy conservation,
employee safety, etc.) of a process enhancement may not be
feasible in all cases. For these cases, report in this manner: Do
not include any of the project cost unless the primary purpose is
environmental protection. If the primary purpose of the project is
environmental protection, report the whole production process
enhancement project expenditure. Indicate in the remarks section
that this is the case. Caution: A project with the primary purpose
of improving production efficiency may include pollution
abatement features added to meet legal requirements. Since the
primary purpose of such a project is still not environmental
18 In contrast, an end-of-line device - such as a scrubber - has only one purpose, and reporting such capital
expenditures are generally thought to be rather straightforward.
17
-------
protection, do not report any of the production process
enhancement.
Whether or not one agrees with this particular conceptualization, at least it was relatively
explicit: If possible, separate out the pollution abatement portion of capital expenditures;
otherwise, report all [none] of the costs if the primary purpose for undertaking the
investment was [was not] environmental protection.
The 1999 survey, on the other hand, is not explicit on the matter of what to report.
Little guidance was actually offered in the Item 3 instructions. Earlier in the instruction
booklet, however, the following appears under the definition of pollution prevention:
Pollution prevention can be an inherent part of the production
process, in which case production and pollution prevention may
be said to occur jointly in an "integrated process." However, for
the purposes of this survey, in order for any expenditure to be
attributable to pollution prevention the integrated process must
have been selected primarily for environmental protection, i.e., an
alternative production process would have been chosen absent this
consideration.
While this is not necessarily inconsistent with the notion put forth in previous PACE
surveys, it also does not rule out other interpretations. Our point is that such ambiguity
may have implications for the comparability of the capital expenditure estimates from the
two years.
New items
In addition to reformulating some old concepts, the 1999 PACE survey introduced
some new data items as well. For example, expenditures on habitat protection (Item 4B),
which includes wetlands mitigation banking and riparian buffer strips, had never been
collected before by the PACE survey. Likewise, much of the expenditures in Item 7
(other payments) are new to the PACE, including payments to governments through fees,
18
-------
charges, fines, penalties, and other means, as well as the revenues from tradable SO2 and
NOx permits.19 The 1999 PACE also asked a series of 21 yes/no questions concerning
the plant's engagement in various types of pollution abatement (Item 2A) and pollution
prevention (Item 3A) activities, participation in voluntary environmental programs (Item
5), and receipt of tax credits or subsidies for their environmental activities (Item 6). The
results of these dichotomous questions, however, do not appear in the PACE publication.
Perhaps the most important addition to the 1999 survey - with potentially the
greatest impact - was the checkboxes that appear next to the expenditures items
indicating "Information not available or not collected to provide an estimate" (see
Appendix C). Establishments received the following instructions on how to report their
data:
Answer all questions. If you can not answer a question from your
plant records, please estimate the answer carefully.... If you are
unable to provide estimates because your facility does not keep
the necessary records or the records are unavailable (i.e. in storage
or archived), please mark the [stated] box.
The implications of this unprecedented choice are not fully understood. Presumably item
non-response was higher than it would have been without this sanctioned option.
Furthermore, when these boxes were checked, it is not necessarily clear what respondents
had in mind. To some, a reported zero expenditure might be consistent with "information
not available or not collected." To others, checking this box might indicate a lack of
confidence in the submitted response. Still others may have checked the box because
they had expenditures but did not estimate them. A post-survey "response analysis
survey" of respondents suggested that the majority (but certainly not all) of those that
checked these boxes had the final motivation. In any event, it is not necessarily clear
19 The costs of permits, however, had been included in the 1993 and 1994 PACE surveys.
19
-------
whether the checkbox is meant to apply to the capital expenditure item, operating costs,
or both.20
Eliminated data items
In addition to abandoning some of the expenditure detail previously collected (e.g.,
operating costs by category; pollution prevention capital expenditure by media; etc.),
there were other items that the 1999 PACE survey simply did not collect. Most notable is
capital depreciation costs, which was already discussed above. Cost offsets (i.e.,
"operating expenses recovered as a result or an off-shoot of pollution abatement
techniques") were also not measured in 1999. In 1994, these had totaled some $1.7
billion. Also excluded in 1999 were expenditures on the abatement of noise, radiation,
and "other" pollutants, which in 1994 amounted to $39 million in capital expenditure and
$178 million in operating costs. And very notable exclusions from the 1999 publication
are the tables on expenditures by state and 2-digit SIC (3-digit NAICS) industry. While
these data were collected, at least in principle, estimates were not produced chiefly
because of quality concerns.
IV. Comparing PACE expenditures across the years
With the discussion above in mind, we will now attempt to compare 1994 and 1999
PACE expenditures. Comparison of the published data obviously takes some effort.
Concept redefinitions necessitate the addition and subtraction of certain items in order to
achieve comparable numbers. And because the 1999 survey often did not separate out
20 Complicating matters even more, responses of zero were not entered from the survey form. Therefore, it
is impossible to distinguish zero responses (which can be a legitimate value, especially for the type of
20
-------
expenses by pollution media or into capital expenditures and operating costs, significant
aggregation sometimes becomes necessary. We'll note here that, in general, adjusting for
differences in sampling and industrial coverage cannot be done with the published
statistics alone. Below, we make these adjustments using the underlying plant-level
microdata, which we have access to.
End-of-line capital expenditures
In terms of capital expenditures, the simplest comparison to make is that of the end-
of-line (EOL) variety. This was collected in the top line of Item 7 on the 1994 survey
(see Appendix B), and estimates appear by 2-digit SIC industry in Tables 4a-4c of the
1994 publication and by state in Tables 6a-6c. Note that EOL capital expenditures
accounted for 52%, 70%, and 62% of total pollution abatement capital expenditures for
air, water, and solid waste, respectively.
As noted above, in 1999, EOL capital expenditures were disposal capital
expenditures plus pollution abatement capital expenditures. (Therefore this is not the
same as the identically-named pollution abatement capital expenditures that appear in
Tables 1, 2, 3, etc. of the 1994 publication, which also include PPE capital.) These 1999
EOL expenditures were collected in the first column of Item 2B and 2C-1, respectively,
and estimates appear in Tables 1, 3, 4a, and 7a of the 1999 publication.
Beyond this, we must adjust for the four differences in scope summarized in Table
1. Table 2 tallies all the necessary adjustments to the data. Note that the industries that
were reclassified as non-manufacturing under NAICS did not account for much of 1994
EOL capital expenditure - a mere 0.08%. A much more important adjustment is the
expenditures under consideration here) from non-responses (i.e., true missing values).
21
-------
subtraction of the (weighted) expenditures of plants with fewer than 20 employees in
1999. Here they account for 2.1% of expenditures. After this, the final two adjustments
to the 1999 data account for a relatively small amount of the remaining expenditures:
NAICS industries that were once in SIC 23 account for just 0.03% of outlays, and
NAICS industries that are new to manufacturing account for only 0.04%.
After the appropriate adjustments are made, we see that there was a 3.9% decline in
nominal spending on EOL capital by the manufacturing sector between these two years,
and an 11.9% decline in real terms.21 However, a more relevant metric is expenditure
normalized by the level of economic activity, since manufacturing may have increased or
decreased over this period. We therefore compute and present dollars of EOL capital
expenditure per $1,000 of value added.22 We see that, overall, such expenditure fell
25.9% between these two years, with EOL capital for air, water, and solid waste falling
20.9%, 29.8%, and 32.4%, respectively.
Prevention/PPE capital expenditures
Evaluating capital expenditures related to prevention and recycling activities - formerly
known as production process enhancements (PPE) - is impossible since much of these
expenditures were combined with prevention operating costs in Item 3B of the 1999
survey (and not separated by media in any event). However, even if one were to
21 The GDP implicit price deflator implies price increases of about 9.04% between these two years.
22 Nominal value added in the manufacturing sector in 1994 totaled $1,406,467.9 million, after eliminating
SIC 23 and those industries that left manufacturing under NAICS, as well as establishments with fewer
than 20 employees. Nominal value added in 1999 totaled $1,822,171.3 million, after removing activity that
once was SIC 23 or that was in non-manufacturing under the SIC system, and after removing
establishments with fewer than 20 employees. These numbers imply a 29.6% growth in nominal
manufacturing activity and an 18.8% growth in real terms, using this appropriately modified definition of
manufacturing. These numbers were computed from statistics available in the 1994 and 1999 Annual
Survey of Manufactures (ASM) publications and the adjustments implied by tables found in Appendix A.
22
-------
Table 2
End-of-line capital expenditures
(in millions of dollars)
1994
Source
Air
Water
Solid Waste
Total
EOL capital exp. for "All industries"
Tables 4a, 4b, 4c
2,228.1
1,698.4
521.5
4,448.0
NAICS non-manufacturing industries
microdata
-1.0
-1.2
-1.4
-3.6
Adjusted total
2,227.1
1,697.2
520.1
4,444.4
Expenditures per $1,000 of value added (in dollars)
$1.58
$1.21
$0.37
$3.16
1999
Source
Air
Water
Solid Waste
Total
Abatement capital exp. for NAICS 31-33
Table 1
2,313.4
1,588.9
307.9
4,210.2
Disposal capital exp. for NAICS 31-33
Table 7 a
-
-
+153.2
+153.2
Establishments with <20 employees
microdata
-36.1
-47.9
-7.1
-91.1
SIC 23 industries
microdata
-0.0
-0.2
-1.1
-1.3
SIC non-manufacturing industries
microdata
-0.6
-0.9
-0.1
-1.6
Adjusted total (1999 dollars)
2,276.7
1,539.9
452.8
4,269.4
Adjusted total (1994 dollars)
2,087.9
1,412.2
415.3
3,915.4
Percentage change
-6.3%
-16.8%
-20.1%
-11.9%
Expenditures per $1,000 of value added
$1.25
$0.85
$0.25
$2.34
Percentage change
-20.9%
-29.8%
-32.4%
-25.9%
assume that
(a) none of the $2.1 billion dollars in combined prevention expenditure was
operating costs (i.e., it was all capital expenditure);
(b) it was only for air, water, and solid waste (and not also for multimedia, USTs,
etc.);
(c) all of the $507.6 million in monitoring & testing costs in 1999 was capital
expenditure (this was previously treated as PPE capital) and not operating
costs;
(d) plants with under 20 employees contributed nothing to the 1999 prevention
and recycling totals; and
(e) the adjustments for changed industrial coverage would matter little;
there was still an enormous decline in such prevention/PPE capital expenditures. In
particular, even with these ridiculously generous assumptions, prevention/PPE capital
expenditures still fell 32.7% relative to manufacturing activity. This is quite contrary to
expectations, since prevention and recycling as well as monitoring & testing has surely
Calculations are available from the authors upon request.
23
-------
become more - not less - prevalent in the United States. Perhaps, however, the
prevention expenditure in particular is done more for production efficiency reasons than
for environmental protection, in which case these expenditures should have been
excluded from the 1999 PACE survey.
Operating costs, prevention, and other costs
Ideally, we would be able to examine operating costs independently. However,
because operating costs are often combined with capital expenditures in the 1999 survey,
the lowest common denominator between the two surveys quickly becomes quite large.
As discussed above in Section III, there are a number of additions we must make to
the basic pollution abatement operating costs statistics collected in Item 2B of the 1999
survey and found in Tables 1, 3, and 4b of the publication. First, operating costs
associated with disposal and recycling (collected in Item 2C and found in Table 7a) must
be added to the solid waste total in order to be comparable to the 1994 definition of
abatement. Furthermore, operating costs associated with pollution prevention (Item 3B),
administration of environmental programs (Item 4D), and environmental monitoring &
testing (Item 4C) must be also added to achieve comparability. The first is found in
Table 8a of the 1999 publication; the latter two are in Table 9a. However, all of these
23
items also include related capital expenditures!
Therefore, to maintain comparability we must add the appropriate capital
expenditures to 1994's basic pollution abatement operating costs numbers (collected in
23 One can imagine that there is not much capital associated with administrative activities, beyond some
relatively inexpensive office equipment perhaps. Monitoring & testing, however, surely contains some
significant capital investment, as does pollution prevention.
24
-------
Item 3 and found in Tables 1, 2, 7, 8a-8c, 9, lOa-lOc, and 1 la-1 lb).24 In 1994,
monitoring & testing equipment was to be included in PPE capital expenditure (i.e., the
second line of Item 7), as was pollution prevention capital. The appropriate PPE figures
(for air, water, and solid waste) can be found in Tables 4a, 4b, and 4c of the 1994
publication. However, since 1994 PPE capital expenditures (for solid waste) also include
capital for recycling efforts, we must then add this item in to the 1999 total. Recycling
capital expenditures were collected in Item 2C-2 of the 1999 survey and appear in Table
7a of the publication.
This is not all. Note that the scope of the prevention, administration, and
monitoring & testing statistics in 1999 is wider than just air, water, and solid waste.
Prevention, for example, also presumably encompasses multimedia pollutants and
underground storage tanks (USTs), while monitoring & testing and administration cover
those two concerns as well as site cleanup. As a result, we must add expenditures on
these three categories to the 1994 total. Operating costs associated with USTs (Item 9),
site cleanup (Item 10), and multimedia pollutants (Item 11-12) can be found in Table
13a-13b of the 1994 publication. Capital expenditures on those same items can be found
in Table 12a-12b. Both operating costs and capital expenditures must obviously be
included here since they are inseparable in 1999.
These particular additions, however, include more than just prevention,
administration, and monitoring & testing. They obviously also include the
abatement/treatment of multimedia pollutants, the replacement of leaking or inferior
24 Note that the PAOC figures in the 1994 publication include the payments to governments for sewage
services and the collection and disposal of industrial waste (from Item 5), which is fine since such
expenditures are already imbedded in 1999's numbers for water pollution abatement and solid waste
disposal.
25
-------
USTs, the operation and maintenance of Superfund sites (remediation), as well as costs
associated with other types of site cleanup (e.g., leaks and spills). Therefore, these
expenditures must of course be added to the 1999 total to maintain comparability.
Multimedia capital expenditures and operating costs were asked in Item 2B of the 1999
survey and appear in Tables 1,3, and 4b of the publication, while the remaining three
costs were asked in Item 4A and appear in Table 9a.
Finally, operating costs in the 1994 survey subsume the costs of permits.
Therefore, two additional items must be added to the 1999 total: tradable permits bought
(Item 7B-1) and payments to government through permits, fees, and charges (Item 7A-1),
estimates of which appear in Table 10a. However, since the 1994 definition of operating
costs explicitly excluded taxes, fines, and such, the addition of the latter implies that the
1999 total will be somewhat too high.
Table 3 summarizes and tallies the various additions just described. Certain
subtractions must also occur. In particular, as we noted above, the 1999 survey explicitly
excluded capital depreciation from its definition of operating costs, and these costs were
not collected elsewhere on the survey. We must therefore subtract off depreciation costs
from the 1994 total. Depreciation costs for air, water, and solid waste were asked as part
of Item 3 of the survey and were published in Tables 1 la-1 lb. The portion of operating
costs attributable to depreciation was not collected however for USTs, site cleanup, and
multimedia pollutants. We therefore estimate these costs. In particular, we recognize
that depreciation costs for air, water, and solid waste (combined) were 38.95% of their
combined total capital expenditures. We apply this same ratio to the capital expenditure
in these three other areas. The result is $3.05 billion of depreciation costs that are
26
-------
subtracted in 1994. The only other subtraction necessary in 1994 is the expenditures by
industries (or shares of industries) that are no longer classified as manufacturing under
NAICS. Just $76.2 million - or 0.37% of the total - is lost as a result of this adjustment.
Subtractions also need to occur on the 1999 side. The most important of these is
the (weighted) expenditure of plants with fewer than 20 employees. According to the
microdata, these facilities contributed $646.3 million of expenditure, or about 3.2% of the
total here. After this, the adjustments for SIC 23 and non-manufacturing SICs are
relatively small, at $22.3 million and $15.6 million, respectively.
After all the appropriate adjustments have been made, we see a 13.6% decline in
real expenditures across these two years, and a 27.3% decline when measured against
manufacturing value added. This decline is only slightly larger than the one seen earlier
in EOL capital expenditure.
And though we cannot measure the change in total operating costs separately, we
can still estimate a lower bound of its change. In particular, if we eliminate all capital
expenditures from the 1994 total, we are left with $17,361.2 million in operating costs, or
$12.34 per $1,000 of value added. Then if we (unrealistically and generously) assume
that all the joint expenditures (on monitoring & testing, prevention, UST replacement,
etc.) are strictly operating costs and contain absolutely no capital expenditures, we have
$10.56 of operating costs per $1,000 of value added in 1999 (after eliminating capital
expenditures for recycling and multimedia). This implies a decline in operating costs
between these two years of at least 14%. Given the improbable assumptions we employ,
the actual decline is of course much greater.
27
-------
Table 3
Operating costs, prevention, and other costs
(in millions of dollars)
1994
Source
Air
Water
Solid Waste
Total
Abatement operating costs for "All industries"
Table 1
6,139.1
7,031.5a
5,601.4b
18,772.0
Capital depreciation costs for "All industries"
Table 11a
-1,471.7°
-1,102.3°
-378.0C
-2,952.0
PPE capital exp. for "All industries"
Tables 4a/4b/4c
+2,082.6
+730.5
+317.0
+3,130.1
UST capital exp. for "All industries"
Table 12a
-
-
-
+74.1
UST operating costs for "All industries"
Table 13a
-
-
-
+181.4
UST depreciation costs
estimatedd
-
-
-
-28.9
Site cleanup capital exp. for "All industries"
Table 12a
-
-
-
+183.5
Site cleanup operating costs for "All industries"
Table 13a
-
-
-
+1,460.0
Site cleanup depreciation costs
estimated d
-
-
-
-71.5
Multimedia capital exp. for "All industries"
Table 12a
-
-
-
+5.3
Multimedia operating costs for "All industries"
Table 13a
-
-
-
+78.5
Multimedia depreciation costs
estimatedd
-
-
-
-2.1
NAICS non-manufacturing industries
microdata
-76.2
Adjusted total
$20,754.2°
Expenditures per $1,000 of value added
$14.76
1999
Source
Air
Water
Solid Waste
Total
Abatement operating costs for NAICS 31-33
Table 1
3,977.5
4,277.4
1,818.8
10,073.7
Disposal operating costs for NAICS 31-33
Table 7a
-
-
+3,123.4
+3,123.4
Recycling operating costs for NAICS 31-33
Table 7a
-
-
+1,190.9
+1,190.9
Administrative costs for NAICS 31-33
Table 9a
+1,159.4
Monitoring & testing costs for NAICS 31-33
Table 9a
+507.6
Government permits, etc. for NAICS 31-33
Table 10a
+695.0f
Tradable permits bought by NAICS 31-33
Table 10a
+12.0
Pollution prevention costs & exp. for NAICS 31-33
Table 8a
+2,101.9
Recycling capital exp. for NAICS 31-33
Table 7a
-
-
+120.1
+120.1
UST replacement costs & exp. for NAICS 31-33
Table 9a
-
-
-
+57.8
Site cleanup remediation costs & exp. for NAICS 31-33 Table 9a
-
-
-
+761.9
Other site cleanup costs & exp. for NAICS 31-33
Table 9a
-
-
-
+80.0
Multimedia capital exp. for NAICS 31-33
Table 1
-
-
-
+177.1
Multimedia operating costs for NAICS 31-33
Table 1
-
-
-
+166.6
Establishments with <20 employees
microdata
-646.3
SIC 23 industries
microdata
-22.3
SIC non-manufacturing industries
microdata
-15.6
Adjusted total (1999 dollars)
19,543.2
Adjusted total (1994 dollars)
17,923.0
Percentage change
-13.6°/
Expenditures per $1,000 of value added
$10.73
Percentage change
-27.3°/
s The PAOC for water in Table 1 already includes the $1,315.5 million of payments to governments for sewage services. See Table
10b.
b The PAOC for solid waste in Table 1 already includes the $301.8 million of payments to governments for collection and disposal of
industrial waste. See Table 10c.
c Distribution based on the underlying microdata.
d We assume that depreciation costs are 38.95% of relevant capital expenditures.
e According to Tables 12a and 13a, there was another $179.5 million in expenditures for "other pollutants within the scope of this
survey but not reported elsewhere on this form." It is not exactly clear what these other pollutants are and whether they were also in -
scope to the 1999 survey. We do notinclude these costs here.
f While the cost of permits was included in the definition of operating costs in 1994, the payment of taxes, fees, and such were not.
This number, therefore, includes too much expenditure. No reasonable adjustments can be made however.
28
-------
Total costs and expenditures
Since the expenditures we examine in Table 3 do not overlap with the EOL capital
expenditures we examine in Table 2, we can combine them to form a more
comprehensive picture. Table 4 shows that, overall, there was a 27.1% decline in
expenditures per $1,000 of value added. This is closer to the change for operating costs,
prevention, and other costs (27.3%) than for EOL capital expenditures (25.9%), reflecting
the fact that the former costs account for
Table 4
Total costs & expenditures
(in millions of nominal dollars)
1994
1999
EOL capital expenditures (adjusted) 4,444.4
Operating costs, prevention, and other costs (adjusted) 20,754.2
Total (adjusted) 25,198.6
Expenditures per $1,000 of value added $17.92
Percentage change
4,269.4
19,543.2
23,812.6
$13.07
-27.1%
about 82%) of total costs and expenditures. These declines are all the more dramatic
given evidence of double-reporting of expenditures by some 1999 PACE respondents.25
We will also note at this point that, according to Tables 2 and 3, establishments
with fewer than 20 employees spent some $737.4 million in 1999, which accounts for
3.0%) of the expenditure by the entire manufacturing sector. This reflects an increase in
importance since the last time such plants were surveyed - in the late 1970s - when they
25 In particular, according the post-survey response analysis survey, a fair number of establishments
apparently reported disposal & recycling costs in both Items 2C and 2B-3 (i.e., abatement of solid waste),
reported prevention expenditures in both Items 3B and 2B (i.e., pollution abatement), reported recycling
expenses in both Items 2C-2 and 3B (i.e., pollution prevention), and occasionally reported their "other"
expenditures and payments (Items 4 and 7) in Items 2 or 3 as well. Some effort was made to eliminate
"obvious" cases of double-reporting, but significant double-counting surely remains in the 1999 PACE
data.
29
-------
accounted for 2% of expenditures according to PACE publications. And over this time
period, these establishments' share of manufacturing's value added actually decreased
slightly, from 5.2% in 1977 to 5.1% in 1997, according to Census of Manufactures
statistics. This implied increase in the intensity of environmental expenditures for this
segment of the population is entirely consistent with the notion that, for various reasons,
environmental regulators tend to target smaller establishments later than their larger
counterparts (see, for example, Becker and Henderson 2000).
A look at specific industries
Here we repeat the above exercises (in our Tables 2-4) for four specific industries:
petroleum refineries, steel mills, plastic material & resin manufacturing, and pulp mills.
These industries were chosen because they are quite pollution-intensive and therefore
should have had high levels of pollution abatement expenditure. Indeed, in 1994 at least,
three of these industries were among the top five 4-digit SIC industries in terms of
pollution abatement operating costs. They were also chosen because their SIC-NAICS
mappings are fairly simple.26 In the interest of brevity, we present only final, adjusted
expenditures per $1,000 of value added in Table 5.27
We see here that the reported expenditure in the plastics industry declined 15.8%,
which seems fairly substantially, but is in fact a smaller decline than the 27% seen in total
26 In particular, SIC 2821 (plastic materials & resins) is now simply NAICS 325211. That is, it did not
splinter into any other NAICS categories, and NAICS 325211 is comprised only of plants from the former
SIC 2821. Similarly, SIC 2911 (petroleum refining) is now simply NAICS 324110, with no splintering or
merging, and SIC 2611 (pulp mills) is now simply NAICS 322110. SIC 3312 (steel) is a bit more
complicated, but still rather straightforward. It split into two parts: 99.2% of it is now classified as NAICS
331111 (iron & steel mills), with the rest assigned to NAICS 324199 (all other petroleum & coal products).
On the flip side, 99.0% of NAICS 331111 came from the former SIC 3312, and 25.4% of NAICS 324199
came from the former SIC 3312. In contrast, SIC 2869 (industrial organic chemicals n.e.c.), which was the
industry with second highest operating costs in 1994, is not considered here because of the complexity of
its SIC-NAICS mapping, splintering as it did between five NAICS categories.
30
-------
manufacturing. Petroleum refineries, on the other hand, reported over 60% less PACE
expenditure per value added. Meanwhile, steel mills reported 11% more expenditure in
1999 than in 1994, and calculations for pulp mills also suggest an increase in reported
expenditure.28
V. Discussion
Overall, the evidence suggests a dramatic fall in reported expenditures between
1994 and 1999. Furthermore, all expenditure categories have seen a decrease: EOL
capital expenditure (air, water, and solid waste), PPE/prevention capital expenditure, and
total operating costs. Our look at specific industries, however, suggests some significant
heterogeneity, with some
Table 5
Total costs & expenditures for select industries
Plastic materials Petroleum
and resins mfg. refineries Steel mills
Expenditures per $1,000 of value added: 1994 $55.48 $218.24 $57.53
Expenditures per $1,000 of value added: 1999a $46.70 $86.79 $64.00
Percentage change -15.8% -60.2% +11.2%
* Figures include expenditures by establishments with fewer than 20 employees and therefore may be slightly higher than they
should be for comparability.
industries experiencing a decrease and others an increase.
Why is there such a decline? One possibility is that this reflects a real change in
environmental spending, as manufacturing activity (potentially) shifted toward "cleaner''
27 Calculations are available from the authors.
28 Computations for pulp mills are plagued by the suppression of key data items in the 1999 publication (for
confidentiality reasons). Therefore, we can only reliably produce a range for the estimate: from a decrease
of 2.9% in reported expenditures to an increase of 20.3%. Our best estimate is that reported expenditures
increased 8.5% (from $109.17 per $1,000 of value added to $118.46).
31
-------
industries and/or expenditures necessitated by environmental regulations eventually
eased. However, on the first point, it is important to note that particularly big and "dirty"
industries (e.g., petroleum refineries) also experienced dramatic declines, not just total
manufacturing. And the second notion could potentially be true for something like
capital expenditures, but it is much harder to imagine that operating costs and prevention-
related spending would fall - and fall so much so quickly.
Another possibility is that the spending estimates for years leading up to the
survey's hiatus were too high. This of course is hard to prove or disprove. Evidence
suggests however that, if anything, the opposite is true (e.g., Becker 2001, Gray and
Shadbegian 2002, Shadbegian and Gray 2003). A much more likely scenario however is
that issues with the 1999 survey led to reported expenditures that were too low. In
particular, we believe some of the following factors may have played important roles in
1999:
° A long hiatus. Record-keeping and reporting on environmental costs were presumably
much better - especially among large "certainty" establishments - when the PACE was
a regular, annual survey.
° Usual freshman year issues. Most surveys, unless they are extensively pre-tested, take
a number of years to accomplish all that they set out to do. In particular, instructions
and question-wording may need some fine-tuning, and editing and imputation
procedures generally require further development. Given the substantial changes in the
PACE survey, and the long hiatus since the last one, the 1999 PACE survey might
reasonably be viewed as a first-year survey.
° A delayed mail out. For a variety of reasons, the 1999 PACE survey was mailed out
fairly late, in September 2000. With eight months having elapsed since the end of the
reference year, and 20 months since the beginning of that year, it is not hard to imagine
that establishments would have had difficulties recalling and estimating many of their
environmental expenditures. This, of course, is in addition to the normal difficulties
faced by plants reporting under the best of circumstances.
° The "information not available " checkboxes. The introduction of the "Information not
available or not collected to provide an estimate" checkbox next to each expenditure
32
-------
item may have (perhaps inadvertently) raised item non-response. That is, on the
margin, some establishments may have checked this box in lieu of providing their best
estimate.
° Conservative imputation for item non-response. Some imputation for item non-
response was done on the 1999 survey, however it was probably rather conservative
relative to prior PACE surveys. This conservatism was largely due to the fact that there
are few good "signals" on which items actually require imputing and a real lack of
defensible imputation algorithms. Imputation was also limited to just operating costs
for air, water, solid waste, disposal, and recycling —thereby excluding prevention
expenditures as well as spending on administration, monitoring & testing, multimedia
pollutants, and so forth.2
° Limited and "asymmetric " data editing. Some amount of data editing was done on the
1999 survey — to correct suspected cases of double-counting, to fix expenditures
reported in dollars rather than thousands of dollars, and to treat otherwise implausibly
high values. There was no examination of and editing for implausibly low values
however. Again, defensible mechanisms are very hard to come by, especially with no
prior year data to establish norms for each industry.
° Less-than-explicit instructions. Relative to the 1994 PACE at least, the instructions in
1999 were much less explicit. The 1994 instruction booklet, for example, contains
long, detailed lists of items to be included and excluded from expenditures. Another
example is the elaborate directions on how to report PPE/prevention capital
expenditures (see Section III above). It is not necessarily clear that this would have
lead to lower reporting in 1999, but it certainly might have.
° Overly broad questions. In contrast to the 1999 survey, which asked for just a single
operating cost number (per media), the older PACE surveys asked respondents to
attribute their operating costs (by media) to five separate expense categories:
depreciation, salaries & wages, fuel & electricity, contract work, etc. Perhaps this
additional prompting reduces the potential for omitting expenditures. At the very least,
the same question asked two different ways may yield different results.
° Question placement. In 1999, some key expenditures items - particularly,
administration and monitoring & testing - were found near the end of the survey form,
where item non-response may be more prevalent and/or where the implicit suggestion
may be that these are less important expenditures. The importance of question
placement can perhaps be seen in the case of expenditures on multimedia pollutants,
which was asked on the final page of the 1994 survey but on the second page of the
1999 survey. Interestingly, reported multimedia expenditures (per dollar of value
added) rose 327.3% between the two years.
29 Capital expenditures of all types were also exempted from imputation, which had been the case in the
past as well.
33
-------
VI. Recommendations
In light of the above discussions, we offer some recommendations for any future
PACE surveys:30
° Provide clear, detailed, and explicit instructions, along the lines of the 1994 PACE
survey. Plainly define key concepts, offer instructive examples, and explain how
particularly difficult items (e.g., incremental material and fuel costs, PPE/prevention
capital expenditures) are to be reported. Among other benefits, reducing ambiguity
should also lessen the incidence of double-counting.
° Great care should be taken to name and define concepts in a longitudinally-consistent
manner. In the 1999 survey, a number of fundamental concepts took on radically new
meanings (e.g, pollution abatement, operating costs, end-of-line capital,
PPE/prevention capital, and all industries) making cross-year comparisons extremely
difficult, if not impossible, and at the very least extremely confusing to the casual data
user.
° Do not provide establishments with the option to not respond to an item, as was the
case with the "information not available" checkboxes in the 1999 survey. Surely,
plants are better able to estimate their expenditures than the Census Bureau can impute
them.
° Design the survey instrument with the eventual imputation algorithm(s) in mind. As it
turned out, imputation in the 1999 PACE was rather ad hoc, and there were few good
signals on which items actually needed imputing. One idea is to structure questions in
the following manner:
~ Did your facility undertake any of the following air pollution abatement
techniques?
~ If yes, report your facility's operating costs here.
~ Did your facility incur any capital expenditures related to air pollution
abatement?
~ If yes, report your facility's capital expenditures here.
° Data entry must include reported zeros. This is essential for proper imputation (for
item non-response) and is also extremely important for microdata research.
° Examine and edit data for implausibly low values, particularly in critical industries.
Suspicious cases might be identified using industry-level norms developed from
historical expenditure data. A resolution may require following up with a respondent
(e.g., by phone).
° Include capital depreciation in the definition of operating costs.
30 See also Burtrawe/ al. (2001).
34
-------
° If activities such as environmental administration and monitoring & testing are to be
recognized as distinct from the treatment, prevention, recycling, and disposal of
pollution - and they from each other - they should at least be placed together on the
survey form and given equal footing. One can imagine, for example, a grid with these
six basic activities as columns and the various types of expenditures (e.g., salaries &
wages, materials & supplies, contract work, capital investment, etc.) as the rows. Such
an arrangement may also reduce the potential for double-reporting expenditures across
activities, which occurred with some frequency on the 1999 PACE survey.
° Do not have respondents pool capital expenditures and operating costs, as was the case
on a number of items on the 1999 survey. The economic meaning of these
expenditures is quite different and therefore they must be kept distinct.
° Send a sample survey form to intended respondents before the beginning of the
reference year and collect data immediately at the end of the reference year. This may
lead to better record-keeping during the reference year, reduce recall bias, and result in
higher quality estimates.
° Publish state-by-industry (2-digit SIC / 3-digit NAICS) tables on key expenditure
items. Note that achieving high quality estimates at this level of detail may require
larger sample sizes than is typical, but the result is well worth the cost. Such tables
were produced up until 1999 and proved extremely popular.
° The PACE sample should (again) be a strict subsample of the Annual Survey of
Manufactures. This had been the case up until 1993. This would yield a much larger,
more representative sample for microdata research.
° Over-sample newer establishments and ask them for their historical cost data. It turns
out that new facilities with the potential to pollute are generally required (by their
environmental permits) to install pollution abatement technologies before they begin
operation. To the extent that these capital expenditures occur in the calendar year(s)
before the plant begins operation - and therefore before the establishment is in the
Census Bureau's business register and thus capable of being sampled - these costs are
not captured by a traditional "current year" survey instrument like the PACE. This
omission is potentially enormous and is very likely growing in importance. A survey
that includes retrospective questions would help in measuring capital expenditures by
facilities under construction.
° Continue collecting data from establishments with fewer than 20 employees. This
paper has shown that these facilities now account for a greater fraction of expenditure
than they did when they were last surveyed in the late 1970s (i.e., 3% versus 2%) - a
trend that, we think, is likely to continue.
35
-------
References
Barbera, Anthony J. and Virginia D. McConnell. "Effects of Pollution Control on
Industry Productivity," Journal of Industrial Economics, 35(2), 161-172, January
1987.
Barbera, Anthony J. and Virginia D. McConnell. "The Impact of Environmental
Regulations on Industry Productivity: Direct and Indirect Effects," Journal of
Environmental Economics and Management, 18(1), 50-65, January 1990.
Becker, Randy A. "Air Pollution Abatement Costs under the Clean Air Act: Evidence
from the PACE Survey," Center for Economic Studies Discussion Paper, 01-12,
November 2001. (Revised October 2003)
Becker, Randy A. "Pollution Abatement Expenditure by U.S. Manufacturing Plants: Do
Community Characteristics Matter?" Center for Economic Studies Discussion Paper,
03-18, November 2003.
Becker, Randy A. and Vernon Henderson. "Effects of Air Quality Regulations on
Polluting Industries," Journal of Political Economy, 108(2), 379-421, April 2000.
Berman, Eli and Linda Bui. "Environmental Regulation and Productivity: Evidence from
Oil Refineries," Review of Economics and Statistics, 83(3), 498-510, August 2001.
Burtraw, Dallas, Alan Krupnick, Richard Morgenstern, William Pizer, and Jhih-Shyang
Shih. "Workshop Report: Pollution Abatement Costs and Expenditures (PACE)
Survey Design for 2000 and Beyond," Resources for the Future Discussion Paper
Series, 01-09, March 2001.
Gray, Wayne B. Productivity Versus OSHA and EPA Regulations. Ann Arbor, MI: UMI
Research Press, 1986.
Gray, Wayne B. "The Cost of Regulation: OSHA, EPA and the Productivity Slowdown,"
American Economic Review, 77(5), 998-1006, December 1987.
Gray Wayne B. and Ronald J. Shadbegian. "Environmental Regulation, Investment
Timing, and Technology Choice," Journal of Industrial Economics, 46(2), 235-256,
June 1998.
Gray Wayne B. and Ronald J. Shadbegian. "Pollution Abatement Costs, Regulation, and
Plant-Level Productivity," in Economic Costs and Consequences of Environmental
Regulation, Wayne B. Gray (ed.), Burlington, VT: Ashgate Publishing, 2002.
Levinson, Arik. "An Industry-Adjusted Index of State Environmental Compliance
Costs," in Behavioral and Distributional Effects of Environmental Policy, Carlo
36
-------
Carraro and Gilbert E. Metcalf (eds.), Chicago: National Bureau of Economic
Research and The University of Chicago Press, 2001.
Iovanna, Rich, Kelly Maguire, and A1 McGartland. "The Pace of PACE at the U.S.
Environmental Protection Agency," Association of Environmental and Resource
Economists Newsletter, 23(2), 21-24, November 2003.
Shadbegian, Ronald J. and Wayne B. Gray. "Pollution Abatement Expenditures and
Plant-Level Productivity: A Production Function Approach," Center for Economic
Studies Discussion Paper, 03-16; National Center for Environmental Economics
Working Paper, 2003-05, August 2003.
U.S. Bureau of the Census. Pollution Abatement Costs and Expenditures. Washington,
DC: U.S. Government Printing Office, various years.
37
-------
AfififindiiA31
Table A-l: Industrial activity no longer classified as manufacturing under NAICS
100% of SIC 2411 (logging)
100% of SIC 2711 (newspaper publishers)
100% of SIC 2721 (periodical publishers)
100% of SIC 2731 (book publishers)
100% of SIC 2741 (miscellaneous publishers)
99% of SIC 2771 (greeting card publishers)
13% of SIC 3732 (boat repair)
Table A-2: Industrial activity now classified as manufacturing (NAICS industries)
2.5%
of
311330
1.8%
of
311340
7.2%
of
311612
100%
of
311811
18.8%
of
313311
0%
of
313312
15.1%
of
314121
5.7%
of
315999
100%
of
326212
100%
of
334611
42.0%
of
334612
3.7%
of
335312
7.3%
of
337110
2.4%
of
337121
2.9%
of
337122
100%
of
339116
Table A-3: NAICS industries that correspond to the former SIC 23
100% of 313222 (Schiffli machine embroidering)
85% of 314121 (curtain & drapery mills)
100% of 314129 (other household textile product mills)
100% of 314911 (textile bag mills)
100% of 314912 (canvas & related product mills)
73% of 314999 (all other miscellaneous textile product mills)
100% of 315 (apparel manufacturing) except:
100% of 3151 (apparel knitting mills)
36% of 315992 (glove & mitten manufacturing)
6%> of 315999 (other apparel accessories & manufacturing)
50% of 323113 (commercial screen printing)
43% of 336360 (motor vehicle seating & interior trim manufacturing)
16% of 339994 (broom, brush, & mop manufacturing)
31 Source: "1997 Economic Census: Bridge Between NAICS and SIC"
(http://www.census.gov/epcd/ec97brdg/)
38
-------
AEEgfldi^£: Questions from the 1994 PACE survey
ITEM 3 - OPERATING EXPENSES FOR POLLUTION ABATEMENT EXCEPT PAYMENTS TO GOVERNMENT
FOR POLLUTANT REMOVAL
Expense Category
Media
Air
Water
SolittfContained
Waste
a. Depreciation
361
S , ,000
S11
5 ,000
S3'!
S r ,000
b. Salaries &. Wages
SB
S , ,000
mi
$ , pea
123
$ , ,000
c. Fuel & Electricity
3IB
1 , ,001
:m
$ r ,000
as
$ ,000
cl. Contract Woric-Services
3M
$ . ,000
SH
$ , ,000
m
S , ,000
e. Mat# rials,
Leasing,
Miscellaneous
3t&
1 r ,ooo
215
S r ,000
as,
1 , ,001
f. Total
m
1 , ,000
31S
i r ,ooo
m
1 , ,OOQ
ITEM 7 - NEW CAPITAL EXPENDITURES FOR POLLUTION ABATEMENT BY MEDIA
Type
Media
Air
(a)
vV.-itor
>!¦«
C-'Ikl Ctintdiin/il VVu'..te
Eld >if LiiH'
nil
•S , ,000
ni
S , ,000
721
S , ,000
F'r«niu- ti.>ri F'rr
¦l» <1 In tii >n
>il>.n>"'ii>'iit
'MI'
S , ,000
T»
S , ,000
m
s , ,000
Ti>t^l bv Modi.*
.'U.I
S , ,000
T13
S , ,000
173
$ , ,000
Percent
htaz^rdous
C.MI-' It.'lil .H.
714
.0%
724
.0%
39
-------
Annendix C: Questions from the 1999 PACE survey
pfHf!EW own I1TIWM muTCUCHT reienrteai awn Bcrvwiii«
' *
B. Pollution Abatement Costs and Expenditures by Medium
Report total pollution abatement costs and expenditures. Also report the percentage of the total capital expenditures
and operating costs that are for hazardous emissions, discharges, or wastes, To avoid double counting, please
record expenditures in one media category only, Report the total capital expenditures and NOT the Financed or
depreciated amount
(REPORTDOLLARS IN THOUSANDS.}
What were the TOTAL pollution abatement costs
and expenditures for this facility in 1999 for -
1. Abatement of air emissions?
Pollution abatement
capital expenditures
Pollution abatement
operating costs
Information not
available or not
collected to provide
an estimate
220
1 000
2311
$ 080
2*0
~
What percent of expenditures and costs were for
hazardous air emissions?
221
%
231
%
241
~
2. Abatement of water discharges?
222
$ 000
232
$ 000
242
~
What percent of expenditures and costs were for
hazardous water discharges?
223
%
233
%
243
~
3. Abatement of solid wastes?
214
$ 000
234
$ OOO
244
~
What percent of expenditures and costs were for
hazardous solid wastes (including mixed wastes)?
22B
0/
fO
231
%
246
~
A. Abatement of multi-media pollutants?
22©
$ 000
236
$ 000
246
~
What percent of expenditures and costs were for
hazardous multi-media pollutants (including
mixed wastes)?
22?
%
23"
%
247
~
¦ >«„« pollution ABATEMENT, DISPOSAL AND RECYCLING - Continued
C. Disposal and Recycling Costs and Expenditures
Report on-site and off-site disposal and on-site
(post process) and off-site recycling costs and
expenditures in the appropriate categories below.
(REPORT DOLLARS IN THO USANDS.)
1. On-site and off-site disposal?
Disposal capital
expenditures
Disposal operating
costs
Information not
available or not
collected to provide
an estimate
»i
$ ooo
$ 0-00
2«
~
What percent of expenditures were for disposal of
hazardous pollutants {including mixed wastes)?
251
%
2B1
%
291
~
2. On-site (post process! and off-site recycling?
Racy elin a capita 1
expenditures
Recycling operating
costs
Information not
available or not
collected to provide
an estimate
2TI
$ 000
2B0
$ 000
291
~
40
-------
BlMnrB om " iiTinw pncv/cwTinw
A, Indicate MarktJO)those pollution prevention activities that occurred
at this facility in 1 999.
1, Process or procedure modifications?
Y«
No
Don't know
3111
~
~
~
2. Redesign or reformulation of products?
302
~
~
~
3, Substitution or modification of equipment or technologies?
303
~
~
~
4. In-process recycling, recovery, recirculation, and reuse of materials?
304
n
~
n
5, Substitution of raw materials?
305
~
~
~
6. Improvements or modifications to housekeeping, maintenance, training
or inventory control?
306
~
~
~
7. Improvements or modifications to prevent leaks and spills?
307
~
~
~
8. Other? Specify in the "Rem arks" ssctfon, htm S.
308
~
~
~
R. If you check YES to any of the activities, report the total (e.g., capital,
operating! costs and expenditures for these activities.
(REPORT DOLLARS IN THOUSANDS.}
Amount
Information not
awalable or not
collected to provide
an estimate
310
$ 000
311
~
41
------- |