Calculating the Cost of Environmental Regulation

Calculating the Cost of Environmental Regulation

William A. Pizer and Raymond Kopp


Resources for the Future


Calculating the Costs of Environmental
William A. Pizer and Raymond Kopp
March 2003 • Discussion Paper 03–06
Resources for the Future
1616 P Street, NW
Washington, D.C. 20036
Telephone: 202–328–5000
Fax: 202–939–3460
© 2003 Resources for the Future. All rights reserved. No
portion of this paper may be reproduced without permission of
the authors.
Discussion papers are research materials circulated by their
authors for purposes of information and discussion. They have
not necessarily undergone formal peer review or editorial
Calculating the Cost of Environmental Regulation
William A. Pizer and Raymond Kopp
Decisions concerning environmental protection hinge on estimates of economic burden. Over the
past 30 years, economists have developed and applied various tools to measure this burden. In this paper,
developed as a chapter for the Handbook of Environmental Economics, we present a taxonomy of costs
along with methods for measuring those costs. At the broadest level, we distinguish between partial and
general equilibrium costs. Partial equilibrium costs represent the burden directly borne by the regulated
entity (firms, households, government), including both pecuniary and nonpecuniary expenses, when
prices are held constant. General equilibrium costs reflect the net burden once all good and factor markets
have equilibrated. In addition to partial equlibrium costs, these general equilibrium costs include welfare
losses or gains in markets with preexisting distortions, welfare losses or gains from rebalancing the
government’s budget constraint, and welfare gains from the added flexibilty of meeting pollution
constraints through reductions in the use of higher-priced, pollution-intensive products. In addition to
both partial and general equilibrium costs, we also consider the distribution of costs across households,
countries, sectors, subnational regions, and generations. Despite improvements in our understanding of
cost measurement, we find considerable opportunity for further work and, especially, better application of
existing methods.
Key Words: social cost, cost-benefit, cost-effectiveness, environmental regulation
JEL Classification Numbers: Q20, Q28, H41, L50, D58
1. Introduction………………………………………………………………………………………………………. 1
2. Environmental Protection Costs and Consequences: Partial Equilibrium…………….. 3
2.1. Direct Compliance Costs ……………………………………………………………………………. 4
2.2. Indirect Costs and Revealed Cost Measures………………………………………………….. 7
2.3. Negative Costs? …………………………………………………………………………………………9
2.4. Government Expenditures on Environmental Protection ………………………………. 10
2.5. Household Regulation………………………………………………………………………………. 10
2.6. Uncertainty……………………………………………………………………………………………… 11
2.7. Discounting…………………………………………………………………………………………….. 12
3. General Equilibrium Effects……………………………………………………………………………… 13
3.1. An Example ……………………………………………………………………………………………. 15
3.1.1. When Partial Equilibrium is Right………………………………………………………15
3.1.2. When Partial Equilibrium is Wrong ……………………………………………………17
3.2. Extended Market Analysis………………………………………………………………………… 21
3.3. Approximating Losses in Other Markets…………………………………………………….. 21
3.4. General Equilibrium Analysis……………………………………………………………………. 23
3.5. Numerical Analysis………………………………………………………………………………….. 26
3.6. Environmental Policy versus Public Good Provision……………………………………. 27
3.7. The Double Dividend……………………………………………………………………………….. 28
3.8. Dynamic General Equilibrium Analysis……………………………………………………… 29
3.9. Other “Costs” in General Equilibrium Models…………………………………………….. 31
4. Distribution of Costs…………………………………………………………………………………………. 32
4.1. Impacts by Household………………………………………………………………………………. 33
4.2. Households, General Equilibrium, and Social Welfare …………………………………. 34
4.3. Multi-country Analysis …………………………………………………………………………….. 35
4.4. Impacts by Sector…………………………………………………………………………………….. 36
4.5. Impacts by Region …………………………………………………………………………………… 37
4.6. Intergenerational Issues ……………………………………………………………………………. 38
5. Conclusions……………………………………………………………………………………………………… 39
References…………………………………………………………………………………………………………….. 46
Calculating the Cost of Environmental Regulation
William A. Pizer and Raymond Kopp∗
1. Introduction
The real and perceived economic costs associated with environmental protection are
easily the greatest obstacles to cleaner air and water, improved preservation of ecosystems and
biodiversity, and slower depletion of natural resources. Over the past 30 years, considerable
effort has been directed at quantifying these costs and improving measurement methods.
Aggregate estimates for the United States suggest that roughly 2% of gross domestic product
(GDP) is spent on environmental protection.1 Data for other countries are less comprehensive but
suggest similar levels of expense.2 More important than these aggregate cost estimates—which
imply a decision whether to protect the environment or not—are increasingly frequent and
detailed studies of the cost of specific initiatives.3
Despite such efforts, the accurate measurement of costs remains challenging. This is true
conceptually—in terms of defining what we include as costs—and especially in practice, where
studies use very different methodological approaches to estimate these costs.
Our goal is to create a taxonomy of the different methods and theories of cost
measurement, to show how they relate to one another, and to explain how various empirical
exercises fit in. Imagine, for a moment, that a policymaker, analyst, or citizen asks what we must
give up in order to protect an endangered species, remedy a polluted area, or prevent future
∗ The authors are Fellow and Senior Fellow, respectively, at Resources for the Future. Michael Batz provided
excellent research assistance. This paper was prepared as Chapter 20 of the Handbook of Environmental Economics,
edited by Karl-Göran Mäler and Jeffrey Vincent. We are grateful to Peter Berck, Lars Bergman, and Larry Goulder
for extremely valuable comments on an earlier draft. Naturally, any remaining errors or omissions should be
attributed to the authors.
1 See U.S. EPA (1990a) and Rutledge and Vogan (1994)
2 See Table 13.1 in OECD (1999) and more recent estimates available on the OECD web-site
(“Selected Environmental Data”). The OECD data suggest that expenditures range from 0.6% to 2.0% of GDP
among OECD countries.
3 For example, see analysis by both EPA and the Energy Information Administration presented during hearings
before the U.S. Senate (Consideration of S. 556, Clean Power Act 2001). Alternatively, see a recent survey of cost
analyses by Harrington et al. (2000).
Resources for the Future Pizer and Kopp
climate change. The answer first depends on the consequences we ascribe to an environmental
action—direct compliance costs, forgone opportunities, lost flexibility, etc. The answer also
depends on whether the policy will meaningfully influence the price of goods and services, in
which case it will be necessary to consider the possible welfare consequences due to preexisting
distortions in other (primarily factor) markets, government behavior, and changes in the terms of
trade. Finally, the answer depends on whether concern exists about the “we” taken as a whole, or
about all the pieces “we” comprise. Our chapter focuses on each of these questions in turn:
partial equilibrium costs, general equilibrium effects and broad economic costs, and the
distribution of burden.
The object of our analysis will be one or more relationships of the form
costs ( , ) i = C a z, (1)
where a is a parameter or vector of parameters describing an environmental policy, z is a vector
of parameters summarizing the current economic equilibrium, and ( , ) i C a z is the associated cost
borne by agent i. The vector z should be viewed as the economic features agent i is likely to
assume are fixed in doing a cost calculation—features such as input prices and output level for a
business, or prices and income for a consumer.4 Once computed, this cost ( , ) i C a z might then be
compared with the cost of an alternative policy, ( , ) i C a′ z , compared with the cost of no policy,
(0, ) i C z , compared with the cost ( , ) j C a z borne by another agent j, compared with an estimate
of benefits, or used to compute a marginal cost ( , ) i ∂C a z ∂a and then compared with marginal
In each of these cases, we can distinguish between partial and general equilibrium costs,
and among costs borne by various agents, by considering both the endogeneity of the vector z
and the enumeration of agents i. When z is held fixed and i is limited to the directly regulated
agent(s), we ignore price and activity changes in other markets and measure only the direct
compliance costs in a partial equilibrium. By instead considering the relationship z(a), we can
evaluate costs when policy effects are transmitted to other markets and the economy equilibrates.
By summing costs across agents that compose final demand, we arrive at a measure of the total
4 We are necessarily vague about z at this point because the environmental policy necessarily disrupts the economic
equilibrium. Holding z fixed, as described below, is meant to capture the notion of a partial equilibrium evaluation.
Resources for the Future Pizer and Kopp
cost to the economy.5 This is the notion of cost most often described by economists (e.g.,
Harberger 1971; Diamond and Mirrlees 1971).
Defining and estimating ( , ) i C a z is far from trivial. It may be difficult for agents to
appreciate fully the costs associated with a, such as the opportunity cost of borrowing employees
from other tasks to focus on pollution control, or of building space for house abatement
equipment. It also requires sophisticated and complete models of the economy to elucidate the
relation z(a) as well as transmission of costs to agents not directly affected by environmental
policy (owners of capital, product consumers, factor suppliers).6 Addressing these questions, we
have divided our discussion into three parts. In the next section we consider the various
consequences of environmental protection for those who are directly regulated. This analysis
typically holds the behavior in the remainder of the economy fixed and is therefore considered
partial equilibrium analysis. In Section 3, we consider broad economic costs, taking into account
the general equilibrium effects outside the market where environmental protection occurs and
focusing on final demand. In Section 4, we consider in more detail the question of who bears the
2. Environmental Protection Costs and Consequences: Partial Equilibrium
Popular debate over environmental protection often centers on the out-of-pocket
expenditures or other negative consequences7 directly associated with pollution reduction paid
by firms, governments, and households. In this section we examine various ways these costs are
defined and estimated in the economic literature, organizing the discussion around the affected
agent and type of cost, and later returning to the common issues of uncertainty and discounting.
For the moment, we ignore how good and factor prices are likely to adjust, both creating new
welfare effects in other markets and shifting the burden to different agents—that is, we are
5 Here we emphasize that total costs involve only components of final demand because costs to businesses are
passed on to consumers in the form of higher prices or reduced factor (capital-labor) income. In an open economy,
this will involve costs potentially borne by the foreign sector (McKibbin et al. 1999).
6 For example, U.S. EPA analysis of recently proposed power plant regulation estimated costs of $3.1 billion to $6.9
billion without completely describing who bears these costs: power plant owners, factor suppliers (e.g., coal
companies), or particular end-users (2001).
7 Such consequences, including effects on jobs (Hahn and Steger 1990; Rosewicz 1990) or international
competitiveness (Jaffe et al. 1995), can be converted to pecuniary effects. Other, more indirect costs are discussed
Resources for the Future Pizer and Kopp
implicitly following a partial equilibrium approach and ignoring actual incidence. We return to
these important issues in Sections 3 and 4, respectively.
The overwhelming focal point of the literature to date has been measurement of the direct
compliance cost to firms. Given sufficient knowledge of production technologies, obtained
through engineering studies or revealed market behavior, the cost of environmental protection
described by (1) is a relatively straightforward calculation for firms. Studies of government
compliance costs, while rarer in the literature, are similarly straightforward thanks to detailed
budget documentation. Analyses of regulatory impacts on households are rarer still—and fraught
with difficulty. It is hard to identify, let alone measure, the “technology” for pollution control
activities and the immediate cost to households of environmental regulations.
2.1. Direct Compliance Costs
Environmental policies applied to firms either directly force changes in production
methods (command-and-control policies) or provide incentives to do so by changing prices
(market-based policies). The direct compliance cost is the change in production costs entailed by
the policy. This cost will depend on the particular technological alternatives available to the firm.
Techniques for modeling and estimating production technology are well described in the
economic literature (Caves and Christensen 1980; Färe 1988; Jorgenson 1986). Engineering
models of pollution and abatement also exist (e.g., U.S. EPA 1990, 1985). Finally, surveys of
business and governments can be used to provide direct estimates of pollution control costs
(McGraw-Hill various years; Bureau of the Census 1973-1997; OECD 1999). All of these
methods have been heavily used—and criticized—in the analysis of environmental protection
A convenient way to understand the roles of these various costing techniques is to
consider the channels through which pollution can be reduced and the data necessary to quantify
relevant abatement costs. In some cases, pollution is associated with a particular input, and
substitution away from that input will reduce pollution. In other cases, pollution reduction arises
from changing the production process or installing “end-of-pipe” equipment to capture pollutants
before they escape.8 When input substitution is the primary mechanism for reducing emissions,
8 End-of-pipe treatment often converts one type of pollution (air) into another (solid), raising the need for integrated
pollution control policies. See Davies (2001)
Resources for the Future Pizer and Kopp
or when process changes or end-of-pipe treatment has already occurred and the study is
retrospective, historical data can be used to estimate relationships between observed or estimated
emissions and production costs. However, absent relevant historical experience with pollution
control or input substitution, econometrically estimable production models are unlikely to
provide useful information about abatement costs.
In the case of prospective cost analysis where input substitution is not the primary means
of pollution control, often the only way to estimate costs is to pose the question to engineers
familiar with abatement technology. Early studies of specific environmental regulation followed
this approach. These efforts, including Atkinson and Lewis (1974), Seskin, et al. (1983), Perl and
Dunbar (1982), and Krupnick (1986), estimate the cost of alternative policies based on linear
programming models of firm response using specific technology options enumerated by
engineering experts. This approach is also used by the U.S. Environmental Protection Agency
(EPA) in its regulatory impact analyses of proposed regulations (e.g., U.S. EPA 1985, 1992).
Through 1979, the president’s Council on Environmental Quality used engineering estimates to
forecast aggregate pollution control costs in the annual publication Environmental Quality.
Discussions of these types of analyses can be found in Tietenberg (1992) and more recently in
Morgenstern (1997).
Although an engineering approach is most useful when tailored to the specific
characteristics of an individual plant, it can be problematic when applied on a broad scale. This
often involves estimates based on a “typical” plant that are then extrapolated to the entire
industry. Technologies differ across plants, as do factor costs and even local (state and
municipal) environmental regulations (CBO 1985). For this reason there has been some concern
about the accuracy of this approach when applied to broad regulatory initiatives.
More recently, there has been a large volume of work on the cost of reducing carbon
dioxide emissions to mitigate the threat of climate change. Like the early analyses of abatement
in the 1970s and 1980s, this work is prospective. Unlike these analyses, however, abatement is
linked directly to reduced use of specific polluting inputs: fossil fuels, including coal, oil, and
natural gas. This has allowed researchers to make use of historical information about fuel
substitution to estimate the cost of emissions reductions, rather than rely on engineering
Resources for the Future Pizer and Kopp
estimates. Surveys of these efforts include Nordhaus (1991), Gaskins and Weyant (1993), and
Weyant (1999).9
Alternatives to prospective cost studies have arisen with the collection of increasing
amounts of data on observed or reported expenditures on environmental protection. Most
notably, in 1973, the Bureau of the Census (1973-1997) began collecting and publishing data on
Pollution Abatement Costs and Expenditures (PACE) based on surveys of individual
manufacturing plants. The data provide aggregate industry-level estimates of the out-of-pocket
expenses associated with environmental protection in the manufacturing sector. Although the
survey approach is widely used, Portney (1981) points out the potential problems with it: sample
size, response rate, and the difficulty among respondents with accurately distinguishing pollution
control expenditures. Streitwieser (1995) similarly reports that question misinterpretation has
been identified as the most serious deficiency of the PACE survey.
As these surveys have been refined and become more widely accepted, they have been
combined with similar estimates from other sectors of the economy to estimate economy-wide
costs, as shown in Table 1 (see also U.S. EPA 1984, 1990, 1999; Vogan 1996). These estimates
represent the out-of-pocket expenses attributed to different agents in the economy, shown as
consumers, business, and government. As noted by Schmalensee (1994), simply tallying these
estimates to estimate total costs ignores many indirect costs and may double-count expenses that
are not part of final demand. We return to this issue in Section 3.
When both prospective engineering estimates and retrospective reported expenditures are
available for a particular regulation, a comparison between the two is possible. Harrington et al.
(2000) do this and observe an interesting pattern of results. Among 18 regulations for which they
were able to establish both prospective and retrospective cost measures, the prospective
estimates were higher in two of three cases. They point out, however, that this frequently occurs
because the amount of pollution control was lower than initially estimated, thereby reducing
Returning to our focus on the functional relationship between environmental protection
and cost, ( , ) i C a z , both engineering and survey approaches are problematic in their inability to
9 These surveys describe aggregate modeling exercises that incorporate elements of the next section on social costs.
However, the exercises are based on production and consumer models that reflect private carbon-abatement costs
based on fuel-input substitution.
Resources for the Future Pizer and Kopp
capture uncounted costs or coincidental benefits—uncounted by either the engineer or the survey
respondent.10 Economists have proposed many reasons why these uncounted costs might be
significant, as well as methods for revealing them. We now turn to these issues.
2.2. Indirect Costs and Revealed Cost Measures
Soon after the appearance of survey-based estimates of compliance costs in the literature,
economists began postulating the existence of additional, uncounted burdens associated with
environmental protection. For example, regulations that required large capital expenditures could
arguably crowd other productive investments (Rose 1983). Or regulations that imposed tighter
limits on new emissions sources could discourage investment in otherwise newer and more
productive equipment (Gruenspecht 1982; Nelson et al. 1993). Finally, there is the general
concern that environmental regulation reduces operating flexibility, slowing productivity
improvements in general (Joshi et al. 1997; Boyd et al. 1998).
For example, many of these concerns about indirect or “hidden” costs have been applied
to the New Source Review (NSR) program in the United States (U.S. EPA 2002). Under this
program, new or substantially modified facilities must meet stringent emissions standards. By
exempting old facilities, older plants become relatively more profitable, and firms tend to operate
them longer rather than investing in new plants. Because of the murky definition of “substantial
modification,” firms may also underinvest in maintaining older plants for fear of triggering NSR.
It is exactly these kinds of undesirable incentives and potentially large indirect costs that have
encouraged greater reliance on market mechanisms in the United States and abroad (Gruenspecht
and Stavins 2002).11
Distinct from these costs associated with unintended or distorted behavior, Schmalensee
(1994) raises several additional concerns about measurement problems that plague survey
estimates: No attempt is made to measure legal fees or paperwork costs. Nor can we easily
capture the cost of operating restrictions, such as the increased logging costs associated with
10 This is less of an issue in approaches based on measures of input substitution, such as carbon abatement costs
from reduced fossil fuel use.
11 For example, the acid rain trading program in the United States and the Kyoto Protocol developed under the
United Nations Framework Convention on Climate Change.
Resources for the Future Pizer and Kopp
spotted owl protection.12 Finally, the potential to redesign entire production lines to prevent
pollution while also improving efficiency makes it difficult to identify specific expenditures
related to environmental protection. This is particular true as efforts turn increasingly to pollution
prevention (Tietenberg 1998). It is unclear whether plant managers can realistically discern the
pollution prevention component of expenditures on process changes and cleaner technologies
embodied in new capital (Morgenstern et al. 1999). Instructions for the reinstituted 1999 PACE
survey ask respondents to include “expenditures where environmental protection was the
primary purpose,” later adding that this should measure the incremental costs by comparing
actual expenditures with that for “the alternative technology that would have been adopted were
environmental protection not a consideration.”13
In light of that difficulty, many empirical papers have sought revealed rather than stated
measures of cost at the plant level. In an early example of this approach, Gollup and Roberts
(1983) estimate the cost of lower sulfur dioxide emissions on electric power utilities. They use a
cross-section of plant-level utility data to relate sulfur emissions to total generating costs. More
recently, Coggins and Swinton (1996) examine the marginal cost of sulfur control among a
smaller set of power plants in Wisconsin. Similarly, Pittman (1981) calculates the cost of
controlling water pollution using a sample of pulp and paper mills in Wisconsin and Michigan.
Using the same data, Färe et al. (1993) estimate plant-specific costs from the data. McClelland
and Horowitz (1999) also examine water pollution control costs at pulp and paper mills, but they
include information on the permitted versus actual emissions levels. Finally, Hartman et al.
(1997) conduct a broad study of abatement costs in the manufacturing sector based on early
PACE data that also includes abatement information.
All of those studies make use of environmental indicators to establish cost units. More
often than not, however, data on actual emissions, abatement, or other environmental indicators
are unavailable. This has led to a simpler effort to compare reported environmental expenditures
with total production expenditures. Anything more than a one-to-one relationship would indicate
indirect costs that remain uncounted in the reported environmental expenditures. Gray (1987)
12 Spotted owl protection restricts the harvest of old-growth timber from federally owned lands in the Pacific
Northwest. The magnitude and incidence of policy effects on such a regional timber supply restriction depend on
market features distinct from production technology (Murray and Wear 1998). Schmalensee’s (1994) point is that a
lumber firm is unlikely to go to the trouble to estimate such costs in response to a Census Bureau questionnaire.
13 From the general definition of pollution abatement activities and the additional definition for pollution prevention
capital expenditures, 1999 Survey of Pollution Abatement Costs and Expenditures (Bureau of the Census 2000).
Resources for the Future Pizer and Kopp
and Barbera and McConnell (1990) first studied this question using aggregate data, while
subsequent work by Gray and Shadbegian (1994); Joshi et al. (1997); and Morgenstern et al.
(1999) used detailed plant-level data. Although Gray and Shadbegian (1994) and Joshi et al.
found evidence of uncounted costs, Gray (1987) and Morgenstern et al. did not. Barbera and
McConnell are ambiguous.
The analyses relating measured environmental performance to overall production costs at
the plant level provide the most convincing data concerning the expenditures associated with
specific environmental benefits. They provide a revealed rather than reported measure of
environmental amenity costs. In this way, they avoid the potential pitfalls of both uncounted
categories and misallocated expenses associated with both engineering and survey estimates used
to quantify the relation in (1). In the absence of environmental performance measures, the second
group of papers offers insight into whether survey-based cost estimates should be trusted.
Though not unanimous, many of those studies support the accuracy of survey measures. Survey
accuracy is particularly relevant for estimates of national environmental expenditures, which
abstract from environmental performance and lean almost exclusively on survey measures.
2.3. Negative Costs?
Although considerable work has explored the potential for indirect costs at the firm level,
there has been limited work on the potential for indirect benefits to firms—that is, negative costs.
Porter and van der Linde (1995a, 1995b argue that firms are not always operating efficiently and
that environmental regulation can lead firms to recognize and correct these inefficiencies. This
can lead to a significant indirect firm benefit in terms of increased productivity. DeCanio (1993)
and Lovins (1996) present similar arguments. Based on engineering and econometric studies,
they argue that significant inefficiencies—particularly in energy usage—exist throughout the
Despite the obvious appeal of costless or even profitable environmental improvements,
most economists remain skeptical. Palmer et al. (1995) respond that despite the presence of
some cost-saving offsets, environmental regulation generally must increase costs and lower
profit. They point to both surveys of plant managers and conversations with company officials
indicating that the realized cost savings are small compared with the cost of environmental
protection itself. While conceding that substantial savings might occur in a few cases, they
conclude that the bulk of the empirical evidence supports nonzero costs.
Resources for the Future Pizer and Kopp
The aforementioned work by Gray and Shadbegian (1994), Joshi et al. (1997), and
Morgenstern et al. (1999) provides further evidence against this hypothesis. Although these
authors were searching for uncounted costs, their methodology had the capacity to identify
uncounted cost savings, or negative costs, as well. Morgenstern found evidence of significant
cost saving in one of four industries (pulp and paper) relative to reported expenditures, but it was
not sufficient to result in negative compliance costs overall.
2.4. Government Expenditures on Environmental Protection
Costs borne by private agents, such as businesses or consumers, must be determined by
observation or questionnaires. In contrast, costs borne by governments (federal, state, and local)
are part of the public record. For that reason, much less effort has focused on trying to measure
these costs—they are simply compiled from appropriate government reports (e.g., U.S. Bureau
of the Census 1997). National cost estimates conducted by various government agencies have
differed in their treatment of specific programmatic areas of government expense, such as
Superfund, solid waste disposal, drinking water, and other state and local mandates, as well as
their allocation of capital expense across time (Jaffe et al. 1995; Schmalensee 1994). Despite
these differences, the underlying data are generally considered an accurate measure of
programmatic expense.
In addition to direct expenditures on pollution control and environmental protection, the
government also spends resources on enforcement and monitoring. These costs, which are also
measurable based on government budget information, are typically small (roughly 2%; see Table
1) compared with national expenditures on environmental protection (Vogan 1996).
2.5. Household Regulation
Regulations that effectively take income away from consumers result in direct costs to
households.14 Although the majority of out-of-pocket costs summarized in Table 1 accrue to
businesses and government, some are borne directly by households. Those noted in the table
refer to expenditures on emissions control devices for motor vehicles. Interestingly, this estimate
associated with emissions controls ignores the annual cost of queuing for auto emissions
14 Here, we ignore the potential increase in consumer prices associated with environmental regulation of firms—this
is a general equilibrium and/or incidence issue that we return to in the next sections.
Resources for the Future Pizer and Kopp
inspections. More generally, the cost associated with environmental programs (e.g., recycling) in
terms of household time is ignored.15 Embed these types of programs in models of household
production (Morris and Holthausen 1994) and one sees how the efficiency of household
production is diminished, resulting in loss of household utility.
Regulation can affect households in other ways. In the case of pesticide regulation,
products that pose significant environmental and health threats can be banned. In such cases
households may switch to higher-cost or lower-quality substitutes. A change in product “quality”
reduces household utility and represents a real cost, but it is very difficult to quantify.
A third category of cost results from regulation that restricts household behavior.
Although such regulation represents a small category at present, behavior-restricting regulations,
particularly recreational restrictions to advance ecosystem goals, may grow in the future. For
example, regulations that ban particular types of sport fishing, exclude motorized vehicles from
sensitive habitats, or restrict all human intrusion into special areas cause the value of those
recreational experiences to decline for those who otherwise would engage in them.
The last category of household costs we mention has to do with land use. Development
restrictions on a piece of real estate cause the property value to decline and cause the property
owner to suffer a capital loss. The restrictions might be imposed to protect an ecosystem,
preserve environmental amenities (beaches, for example), or limit suburban sprawl. Increased
concern over each of these issues suggests that land use restrictions will be an important
regulatory tool for some time to come.
2.6. Uncertainty
Few analyses of environmental protection costs consider uncertainty in their analyses.
Some perform a sensitivity analysis over key assumptions (e.g., Chapter 7 of U.S. EPA 1999).
However, a sensitivity analysis merely shows how estimates change with alternative
assumptions; it does not indicate what value to use for a benefit comparison or how decisions
should be made.
Considerable work in statistical decision theory dating back to at least Wald (1950) and
more recently Berger (1985) describes a straightforward, though computationally demanding,
15 Of course, this is not inconsistent with the absence of leisure value in the National Income and Product Accounts.
Resources for the Future Pizer and Kopp
approach to decisionmaking under uncertainty. By assigning probabilities to different outcomes,
it is possible to compute expected costs. More generally, a policymaker willing to specify
preferences over both uncertain costs and benefits—including risk aversion—can identify
policies preferred in expectation.16 Kolstad (1982) and Pizer (1999) provide environmental
examples of this approach. A related issue, when both benefits and costs are considered jointly,
is whether price policies are preferable to quantity policies. Price policies tend to lower both
expected costs and expected benefits relative to quantity policies. Weitzman (1974) derives
conditions where one instrument is preferred.
Dixit and Pindyck (1994) and Pindyck (1995) point out another issue partly but not
wholly related to uncertainty: the irreversibility associated with some environmental policies.
The potential exists for environmental protection activities to involve irreversible costs as well as
for environmental consequences to involve irreversible damage. In the former case, the cost of a
policy is raised by the forgone option value associated with waiting to see whether the
environmental protection is necessary. In the latter case, the cost of a policy is lowered by the
value of the option to prevent possibly irreversible damage. This option value could depend
wholly on the passage of time but most often involves learning about uncertain outcomes.
Computing the option value is further complicated when the choice of policy influences the
learning process (Kelly 1999; Nordhaus 1997).
2.7. Discounting
Most cost analyses focus on current or annualized expenditures, as shown in Table 1.
This facilitates a straightforward comparison with annualized benefits resulting from improved
environmental quality. The emergence of global climate change as a major environmental policy
issue forces us to rethink that approach. In particular, it forces us to consider comparisons of
costs and benefits across very long periods of time, since climate change mitigation expenditures
today yield benefits far in the future.
To compare costs (and benefits) over time, costs in the future are discounted to the
present at a particular rate. The appropriate rate of discount has been subject to much debate
(Lind 1982; Arrow et al. 1996; Bazerlon and Smetters 1999), with different rates leading to
16 The notion of risk aversion captures the idea that we are not indifferent between a risky $10 gain and a certain
$10 gain. In statistical decision theory, this amounts to defining a loss function; in utility theory, it reflects a choice
of utility or welfare function.
Resources for the Future Pizer and Kopp
dramatically different estimates of benefits and costs (Cline 1992). In the case of climate change,
rates near zero support aggressive policy responses, with costs appearing small relative to
benefits; rates exceeding 5% suggest modest policy responses, with costs appearing large relative
to benefits. These observations have led to recent work on the consequences of uncertainty about
the discount rate itself (Weitzman 2001; Newell and Pizer 2003). Even ignoring the question of
the appropriate rate, some qualitative results remain. When the focus of climate change
mitigation policy is a long-term concentration target, rather than a particular emissions profile,
Wigley et al. (1996) demonstrate that discounting as well as adjustment costs favor some delay
in mitigation relative to an immediate reversal in emissions growth.
In dynamic equilibrium models, much of the concern about discounting is embedded in
the model specification via consumer preferences over time. As we will see in the next section,
such models allow one to summarize the total cost associated with environmental controls across
many periods, just as Table 1 summarizes the costs associated with environmental controls
across different agents.
3. General Equilibrium Effects
The concepts described so far have focused on the costs associated with regulatory
compliance by various economic actors—firms, governments, or households. Such analysis
ignores the indirect impact that environmental protection activities in one market can have on
activities in other markets, as well as feedback in the original market, as the economy
equilibrates to these additional burdens. We are most often interested in the total cost including
these indirect effects on welfare—effects that can sometimes be quite large relative to the more
obvious consequences measured by the regulated entities. Stated another way, the question to ask
about a proposed regulation or policy is the overall burden to the national economy, not just the
sum of the direct costs measured in isolation.
Welfare effects in other markets arise because of preexisting distortions in those markets
from current taxes or regulation. Such distortions—a difference between the point where the
market would like to equilibrate, matching supply and demand at one price, and where the
market does equilibrate in the presence of taxes or regulation—create deadweight loss. This
deadweight loss is a cost that reduces the production possibilities of the entire economy. For
example, taxes on labor income discourage people from working as much as they would if they
received the full value of their time. When an environmental regulation is introduced, it can
Resources for the Future Pizer and Kopp
affect the existing deadweight loss in these already-distorted markets by shifting supply or
demand and changing the magnitude of the deadweight loss in that market.
Consideration of equilibrium price changes can affect the original cost estimates for the
directly regulated entities as well as welfare in other markets. For example, the cost of an
electricity policy that encourages fuel substitution from coal to natural gas will rise if increased
natural gas demand raises the price of natural gas versus a case where additional natural gas is
available at the same price. On the other hand, the cost of the policy will fall if consumers
demand less electricity as the price of electricity rises in response to the policy, versus a case
where electricity demand remains constant. To the extent that distortions do not exist elsewhere
in the economy, however, total costs can still be measured in the regulated sector even as the
economy equilibrates.17
As we try to measure these total costs, this overall burden, we increasingly focus on the
real decline in final demand—consumption, investment, and government spending—to avoid
double counting (Schäfer and Stahmer 1989; Schmalensee 1994). Environmental regulations that
raise the price of energy, for example, raise both the cost of energy and the cost of manufactured
goods that are energy-intensive. If we count both the increase in manufacturing costs and the
increase in consumer expenditures on higher-priced manufactured goods, we double-count the
cost of the regulation. This has led to a focus on the consequences for real GDP in many
analyses. Such a focus, however, fails to account for the consumption of leisure (and other
nonmarket transactions) and ignores the distinction between investment in pollution control
equipment and investment in productive capital. This means that typical analyses focused on
“GDP effects” can be misleading, as highlighted below.18
Once the economy equilibrates, the change in final demand will include both the original
costs discussed in the previous section (and passed on to households as either higher prices or
17 That is, the marginal cost of incremental regulation in the regulated sector equals the marginal welfare cost of the
incremental regulation measured across the entire economy. This is a consequence of the First Fundamental Welfare
Theorem—a competitive market equilibrium is welfare maximizing and therefore the market value of the
incremental regulation can be viewed as a shadow price. See, for example, Chapter 16 of Mas-Colell et al. (1995).
18 Weyant and Hill (1999), for example, focus on GDP losses in analyzing global climate change policies. A later
article in the same special issue (McKibbin et al. 1999) shows an example (Table 4) where GDP falls and
consumption rises in 2020. Although GDP and consumption will generally move in the same direction, the
quantitative differences raise interesting questions about how quantitative results should be presented—in terms of
the more familiar (and popular) GDP measure or the more economically relevant consumption measure.
Resources for the Future Pizer and Kopp
lower factor income), as well as indirect effects occurring in other markets. Consumer
preferences provide a natural metric for valuing these changes in final demand. Consumers, in
the guise of both purchasers of goods and owners of productive inputs (as well as the
beneficiaries of investment returns and government provision of public goods), are the implicit if
not explicit focus of most cost analyses.19 A change in their well-being can be monetized by
computing the equivalent variation in consumer income. Equivalent variation measures the
change in income that would lead to the same change in a consumer’s well-being as the policy
under scrutiny.20
3.1. An Example
At this point, it is useful to consider an example to illustrate the different ways in which a
partial equilibrium analysis—as well as GDP measures—can fail to correctly capture aggregate
costs. We imagine a simple economy where a single consumer works for the economy’s sole
employer, a power plant that produces energy (x). The plant uses a linear technology to produce
x0 units of energy with 0 0 p ⋅ x hours of labor. We treat labor as the numeraire good, with a price
of one, so costs, income, and prices are denominated in hours of labor. The central question is,
how do we calculate the cost of an environmental regulation that changes the linear production
technology so that it now costs p1 = p0 + c > p0 to produce each unit of energy?
3.1.1.When Partial Equilibrium Is Right
The simplest approach would be to conclude that given current production x0, the cost of
the regulation is 0 c⋅x . Such an analysis might be the result of an impact analysis for a new
regulation where little was known about the likely response of consumer demand for energy to
higher prices. The calculation would be exactly right if in fact consumer demand were fixed.
Why? With fixed demand for a good, any change in the price is equivalent to a lump-sum
19 That is, in the end what most analyses care about is the welfare of people (households). Given the indirect
benefits to households of investment (in static models) and government spending, ad hoc assumptions are often
made in general equilibrium analyses—for example, requiring investment and government spending to remain
20 Two alternatives to equivalent variation (EV) are discussed in the literature: compensating variation and
Marshallian surplus. Freeman (1985) recommends EV because it represents the desired objective—a money
equivalent of welfare change; see also Jorgenson and Slesnick (1985). The idea originated with Hicks (1942);
Chipman and Moore (1980) show that EV represents an indirect utility function and is therefore appropriate for
welfare comparisons.
Resources for the Future Pizer and Kopp
transfer—the consumer always pays for the good and then uses leftover income to buy other
goods on the margin. The added cost of energy 0 c⋅x in this case is exactly like a 0 c⋅x loss of
income available for other goods. Graphically, we can see this in the left panel of Figure 1, with
vertical demand and horizontal supply curves. Here the area above the price line—the consumer
surplus—falls by 0 c⋅x with the regulation, as indicated by the shaded area.21
Even in this case, where the simple partial equilibrium analysis works, measures like
GDP can be misleading. GDP can be viewed as a measure of either factor income or marketed
final demand—in this case, wage income and purchases of energy. Wage income rises in this
example, as the consumer works more to support her fixed demand—that is, GDP rises in
response to a costly regulation. A more careful analysis might note that real energy use remained
constant, but in no case would a study of GDP reveal a decline because the quantity of marketed
goods remains the same in this example, and GDP does not capture changes in nonmarketed
goods, such as leisure.
Returning to the partial equilibrium analysis, input prices may not be perfectly elastic and
demand may not be perfectly inelastic. For example, suppose the right panel in Figure 1
describes demand for energy. The original analysis based on fixed energy demand would have
overstated costs by ignoring the flexibility of the consumer to substitute away from an
increasingly expensive good. A better analysis might correctly predict—or, after the regulation is
in place, observe—x1 as the new demand for energy and conclude that costs equaled 1 c⋅x . This
would reflect the current expenditures recorded by the power plant as being environmentally
Such a calculation, an analysis of environmental expenditures after regulation is in place,
is precisely the basis of the national cost estimates discussed in the first section, where firms are
surveyed about their abatement expenditures.22 Yet such analyses routinely ignore the loss of
consumer welfare associated with reduced consumption of more expensive, regulated goods—in
our example, reduced energy demand from x0 to x1. In this way, they entirely miss the regulatory
costs associated with goods that are banned or whose regulated costs become so high that no
21 Introductory economics teaches that changes in the area between the demand curve and the price line, referred to
as consumer surplus, can be used to measure changes in consumer welfare as prices change. Intuitively, when the
consumer is forced to pay more for the purchased good, it is as though that much income is being taken away, and as
prices rise and demand falls, there will not be as much of an income loss for the next incremental price increase.
22 For example, U.S, EPA (1990a), Rutledge and Vogan (1994), OECD (1999).
Resources for the Future Pizer and Kopp
production occurs and no control costs are reported. We can fix the calculation in our example
relatively easily: the true welfare cost is the area under the demand curve between p = p0 and p =
p1, or 1( )
2 1 0 c⋅ x+x ; that is, the change in consumer surplus.23 Note that again, the estimated
change in GDP will likely be positive unless the decline from x0 to x1 is large. That is, the change
in GDP equals ( ) 1 0 1 0 c⋅x − x −x p .
3.1.2.When Partial Equilibrium Is Wrong
So far, the partial equilibrium cost analysis of environmental regulation in our simple
model has done well when the consumer’s response is properly captured. However, we have yet
to consider what happens when distortions, especially tax distortions, already exist in other
markets. In this case, we have to consider not only costs in the regulated market, but also
changes in existing costs measured in those other markets.
Figure 2 shows the consequences of a new environmental regulation in the energy (left
panel) and labor markets (right panel) in the presence of a preexisting tax on labor income. When
the regulation is implemented, employment decreases from l0 to l1 while energy demand
decreases from x0 to x1. The shaded region shows deadweight loss in both markets.24 The
important thing to recognize is that the deadweight loss in the labor market changes as
employment shifts from l0 to l1. That is, the environmental regulation not only creates a cost in
the energy market where it is imposed, it also influences the cost of existing policies (a labor tax
in this case) in other markets.
One can make the example more precise algebraically. Consider the following utility
function that underlies Figure 1 and Figure 2:
( ) ( ) ( )
2 *
x x
u x l L l

=− + − (2)
23 Here the demand schedule is linear. The welfare analysis is not correct unless we are careful to use compensated
(holding utility constant) rather than uncompensated (holding income constant) demand schedules. Note that as the
price goes up by an initial increment dp, the cost of maintaining the current level of well-being rises—which we
correctly measure as x0 dp. Now, we want to continue to measure the cost of maintaining that same level of wellbeing
as the price rises by the next increment, not the cost of maintaining the new, lower level of well-being
associated with a higher price and fixed income. Therefore, the demand schedule must hold welfare, not income,
constant. When preferences are quasi-linear (e.g., the marginal utility of one good equals unity), income and utility
do not affect demand for any good other than the quasi-linear good, and both compensated and uncompensated
demand schedules are the same.
24 In both cases, we show compensated labor supplies, as remarked in the preceding footnote.
Resources for the Future Pizer and Kopp
where x is demand for energy, l is supply of labor (so L – l is demand for leisure), and x* is a
parameter describing a level of satiation with energy. The consumer faces a budget constraint
that now allows for distinction between the wage paid by the employer (unity) and the wage
received by the consumer (w):
p⋅x+w⋅(L−l) =w⋅L+T (3)
where expenditures on energy, p ⋅ x , plus implied expenditures on leisure, w⋅(L−l), must equal
the value of the labor endowment, w⋅L, plus any income transfer T from the government . With
the wage paid by the power plant equal to one, we have w = 1 – t where t is the labor tax. The
first-order conditions imply that
x x* p
= − and l p x*p L e px*p T
w w w w w w
=  − + − =  − −
   
providing a solution for energy demand and labor supply. This, in turn, allows us to write
expenditure and indirect utility functions using (2) and (4)
( )
( )
, , * 1
, , 1 *
e p wu p x p w u p
w w
v p w e p e p x p
w w w w
=  − +  +   
= −   + −  − 
   
where u is utility, v is the indirect utility function, and e is total expenditures, including leisure,
and equal to w⋅L+T in (3).
Previously, we asserted that the cost to consumers without a labor tax (when w = 1)
equals 1( )
2 1 0 c⋅ x+x in Figure 1. Therefore, we should be able to see the following relation:
( ) 1 ( ) ( )
0 no labor tax, no regulation 2 1 0 0 no labor tax, regulation e p ,1,u −c⋅ x +x =e p ,1,u (6)
That is, if we took away 1( )
2 1 0 c⋅ x+x in income from the untaxed, unregulated expenditure
level—keeping the prices the same—utility would fall to the same level created by the
regulation. How can we see this in our model? Without taxes, ( ) 0 no labor tax, no regulation e p ,1,u equals
( ) 0 no labor tax, regulation e p +c,1,u equals L. That is, the equilibrium expenditure level always equals the
income level, and without taxes this always equals L.25 Therefore,
25 Note that the equilibrium expenditure in the face of particular policy equals the expenditure function evaluated at
the price and utility resulting from the particular policy; it necessarily equals w⋅L+T .
Resources for the Future Pizer and Kopp
( ) ( )
( ) ( )
0 no labor tax, no regulation 0 no labor tax, regulation
0 no labor tax,regulation 0 no labor tax, regulation
,1, ,1,
,1, ,1,
e p u e p u
e p c u e p u
− =
+ −
But this equals 1( )
2 1 0 c⋅ x+x through manipulation of the expenditure function in (5), and so (6)
is true.26
With a preexisting labor tax, the calculation is slightly more complex. Again, we want to
compute the change in income equivalent to the utility loss associated with the regulation—but
now with a labor tax present,
( ) ( ) 0 labor tax, no regulation 0 labor tax, regulation e p ,w,u −?=e p ,w,u
analogous to (6). But unlike the untaxed case in (6), where equilibrium expenditures always
equaled the same level of income L, equilibrium expenditures now equal w⋅L+T , where T
changes as we move from the unregulated to the regulated equilibrium. In equilibrium, we know
that T=t⋅l because the government transfer must equal the tax revenue. Using this, we have
( ) ( ) ( ) 0 labor tax, no regulation 1 0 0 labor tax, regulation e p ,w,u + t ⋅ l −l = e p +c,w,u
That is, the taxed equilibrium expenditure without regulation, plus the change in tax revenue
associated with the regulation, equals the taxed equilibrium expenditure with regulation. As
before, ( ) ( ) 0 labor tax,regulation 0 labor tax, regulation e p +c,1,u −e p ,1,u equals 1 ( )
2 1 0 c⋅ x+x , so we have
( ) ( ) 1 ( ) ( )
0 labor tax, no regulation 0 labor tax, regulation 2 1 0 1 0 e p ,w,u −e p ,w,u =c⋅ x +x +t⋅ l −l (7)
Intuitively, welfare cost equals the cost measured in the partial equilibrium energy market
analysis, plus the change in employment times the tax wedge (e.g., the difference between the
marginal benefit and marginal cost of an hour worked).
How does this compare with simpler analyses? As before, we could imagine assuming
that the before-regulation output level x = x0 remains fixed, that the after-regulation output level
x = x1 remains fixed, or that output changes based on the compensated demand schedule. With
per unit regulatory costs c, we would estimate partial equilibrium costs of 0 c⋅x , 1 c⋅x , or
1( )
2 1 0 c⋅ x+x , respectively (the latter corresponds to the shaded area in the left panel of Figure 2).
All of these estimates ignore costs in the labor market arising from the existing labor tax.
26 Note that ( )( ( )) ( ) ( ) ( ) 0 0 0 0 0 0 1
1 1 1
p +c x*−2p +c −p x*−2p = 2x*−2p −c =c⋅2 x +x . It is also evident by the
definition of the compensated demand curve in Figure 2, which equals the derivative of the expenditure function
with respect to price.
Resources for the Future Pizer and Kopp
How does this compare with a general equilibrium calculation based on Figure 2? One
could imagine using Figure 2 to compute the cost not only in the energy market but in the labor
market as well—as we suggested at the beginning of this section. This would not be entirely
correct, either. The compensated demand curves used for welfare analysis in these diagrams
measure welfare consequences for price changes in their own markets. They cannot be used to
accurately measure welfare changes in one market arising from price changes in another market
because they are computed holding those other prices fixed. That is, adding the shaded region
between l = l0 and l = l1 in the right panel of Figure 2 to the shaded area in the left panel of
Figure 2 yields too small an estimate.
Table 2 summarizes the variations we just went through and the consequences that would
be measured based on various notions of costs. The first line shows the costs that might show up
on a survey where firms were asked to report their expenditures on environmental protection.
This ignores the change in production level that may be a result of the regulation. An accurate
partial equilibrium analysis, on the second line, shows what a good analysis of the regulated
market would yield. The third line shows how a model trained to measure GDP impacts would
register costs (measurable as either the change in final goods sold or the change in production
factors purchased). Finally, the fourth line shows the true welfare costs, measured as the loss of
income that would lead to a welfare level equivalent to the imposed regulation—the equivalent
variation in income.
The key observation is that no measure of costs correctly measures welfare costs in all
cases except the general equilibrium measure. This illustrates the first main point of Section 3.
Namely, both the measurable costs to the firm in a single market as well as changes in GDP can
be misleading measures of the real cost to consumers. Costs to the firm measured in a partial
equilibrium analysis ignore general equilibrium changes in output and prices and miss potential
welfare changes from altered production levels and distortions in other markets. Changes in GDP
are misleading because they ignore the consumption of leisure and treat regulatory expenses as if
they directly benefited consumers—as if they were expenditures on food or housing.
The remainder of this section discusses both approximate and exact methods of
measuring the costs of environmental regulation based on a general equilibrium model. This
discussion builds up to the second main point of this section: an accurate measure of the total
cost of environmental regulation depends on a complete model of technology, preferences, and
government behavior. Different assumptions about these features of the economy can lead to
different estimates of the impact of environmental regulation.
Resources for the Future Pizer and Kopp
3.2. Extended Market Analysis
A central failure of simple partial equilibrium analysis is its inability to handle price and
output changes in related markets. Therefore, one solution is to extend the analysis to markets
related vertically or horizontally (Just et al. 1982; Whalley 1975; Kokoski and Smith 1987). A
vertically expanded analysis would involve computing the input factor supply schedules, rather
than treating input prices as fixed, and computing consumer and producer surplus loss in those
markets. In the horizontal case, this would involve computing a demand schedule for the output
of the firm undertaking environmental activities, rather than assuming output remains fixed, and
then using that schedule to compute a change in surplus.
Although such extensions would seem to offer improvements, that is not the case
empirically. Whalley (1975) computes the effect of capital tax reform in the United Kingdom
and finds that extending the partial equilibrium to include horizontal effects worsens most of the
welfare estimates relative to the true social costs. In an environmental example, Kokoski and
Smith (1987) examine the cost of unmitigated global warming. In separate horizontal and
vertical extensions to a simple partial equilibrium analysis, they find that both are considerably
worse when only one market is affected by global warming. When multiple markets are affected
by global warming, the extensions do better.
In our earlier example, summarized in Table 2, an extended market analysis helps. By
considering the substitution opportunities in the leisure market, an extended market analysis
would properly capture the welfare effects in the cases without taxes. With taxes, however, this
approach fails to capture welfare loss in the labor market. An increase in the consumption of
leisure (and decrease in labor supply) due to substitution exacerbates the deadweight loss
associated with the labor tax. These effects are ignored by the extended market analysis.
3.3. Approximating Losses in Other Markets
An alternative to these limited extensions is to consider an approximation based on the
effect in all markets, focused on deadweight loss. Both Diamond and Mirrlees (1971) and
Harberger (1971) provide guidance, suggesting expressions of the form:
cost ( ) (0) ( ( ) (0)) j j i i i
i j
C C t X X

= a− −Σ a− , (8)
where as before the vector a parameterizes the policy, ( ) (0) j j C a −C is the loss of social surplus
in the market bearing the direct out-of-pocket costs, ti measures preexisting taxes in each of the
Resources for the Future Pizer and Kopp
remaining markets i≠ j, and ( ) (0) i i X a −X indicates the change in output in each of those
markets due to the policy.27
Note that when we apply (8) to our example in Section 3.1, it delivers the correct answer
with a preexisting labor tax. In addition to the partial equilibrium cost in the energy market,
shown in the left panel of Figure 1 and equal to 1 ( )
2 1 0 c⋅ x+x , we have costs in the labor market
equal to the labor tax t times the change in labor supply,l1 – l0. This yields the correct welfare
cost 1( ) ( )
2 1 0 1 0 c⋅ x+x +t⋅ l−l . With a single representative consumer and a convex production
technology, we can develop a more explicit expression for (8). Suppose y is the vector of (net)
goods produced by the firm, and x is the vector of (net) goods consumed by the consumer. We
can define Y as the matrix of first derivatives of y with respect to producer prices p and S as the
matrix of compensated first derivatives of x with respect to consumer prices q.28 Suppose a is
simply a description of the amount of pollution abated—a scalar a—and let abatement of that
pollutant be good j in the above description. Consider a small change in pollution abated, da. If
we assume that the existing distortions t=q−p do not change, we have dx=S⋅dp (assuming
no change in utility) and dy =Y⋅dp with dp=dq. If we also assume that the existing excess
supply (government and/or foreign supply), y − x , remains constant, we have j dy−dx=eda
where ej is unit vector j. Combining these relations, we have (since taxes are fixed)
( ) 1
j d d eda − p= q= Y−S
The actual change in consumption would be given by
( ) 1
j d d eda − x=S⋅ p=S Y−S (9)
Normally, marginal changes in consumption have no effect on welfare since, at the
margin, producer cost equals consumer benefit. In the presence of distortionary taxes, however,
consumer benefit exceeds producer cost by the tax rate. Increased consumption will raise
27 See Equation (92) in Diamond and Mirrlees; Equation ( 5′′′ ) in Harberger. Diamond and Mirrlees present marginal
conditions for optimal provision of a public good, where their marginal cost expression equals the derivative of the
right-hand side of (8) with respect to a, replacing marginal partial equilibrium costs Cj’ with the derivative of an
aggregate production function with respect to the regulated good Gj (they use k as the regulated good). In this way,
we view Equation (8) as a discrete approximation to the cost side of Diamond and Mirrlees, and in turn an
approximate metric for comparison with monetized benefits. The comparison with Harberger is more
straightforward, as he is already focused on a discrete approximation. The only difference is that his expression for
the cost of a tax in market j, 1 *
2 j j T Δx is replaced by our cost of the regulation in market j, ( ) (0) j j C a −C .
28 Note that this approach requires strict convexity in production and preferences.
Resources for the Future Pizer and Kopp
welfare; decreased consumption will lower welfare. To compute the change in welfare, we
multiply change in consumption (9) by the tax rate:
( ) 1
j d eda − t⋅ x=t⋅S Y−S
This is exactly the second term in (8) as derived by Bruce and Harris (1982) in their
analysis of cost-benefit criteria for evaluating small projects.29 Rather than applying a marginal
cost-benefit analysis to the incremental cost of producing a, ( ) j C′ a , this suggests that we should
base our analysis on
( ) ( ) 1
j j C a e − ′ −t⋅S Y−S , (10)
adjusting for surplus changes in other markets. Diewert (1983) extends this approach even
further, considering the effect of import tariffs for a small open economy.
As the preceding discussion suggests, it is not easy to estimate ( ) (0) i i X a −X , the effect
of a particular policy in every market. This requires a local model of general equilibrium effects,
captured by Y and S in (10). However, as Harberger (1971) notes, the only real concern will be
those markets that have large distortions (ti) and are significantly affected by the policy (large
i ΔX ). Interestingly, this bit of intuition has pervaded a large volume of literature on the
importance of general equilibrium analysis over the intervening 30 years.30
3.4. General Equilibrium Analysis
Efforts to include social costs outside the directly regulated market lead naturally to
general equilibrium cost analyses. Where the previous approach in (10) uses linear
approximations of preferences and technology to work out new equilibria, a general equilibrium
model works directly with utility and profit functions, maximizing both subject to market
equilibrium conditions. A general equilibrium model can also be used to consider alternative
assumptions about taxes and spending (where the previous approach assumed fixed tax rates and
fixed government purchases).
29 See Equation (7) in Bruce and Harris (1982).
30 In a recent article on the use of “Harberger triangles,” Hines (1999) notes that despite the criticism that analysis
based on such triangles ignores general equilibrium effect, Harberger himself consistently emphasized the
importance of spillover effects in distorted markets.
Resources for the Future Pizer and Kopp
General equilibrium analyses can involve both numerical simulation and analytic results.
Analytic work on general equilibrium problems has focused on establishing conditions for
second-best optima and the prices appropriate for marginal project evaluation. Just as Harberger
and others revealed the importance of costs associated with tax distortions, this work has
revealed the importance of fiscal responses to environmental and other public policies. It is
always possible—and frequently necessary—to adjust taxes or spending in response to nonfiscal
policies. The necessity arises from the need to maintain the government budget constraint as the
tax base changes. These adjustments can have significant welfare consequences.
Consider a simple model with a single consumer, a single firm, and a government. The
consumer has endowment k and utility u(x), where both k and x are N element vectors spanning
the good and factor space of the economy. The firm possesses a constant-returns-to-scale
technology with unit profit function π(p) that is operated at an arbitrary scale α, provided profit
is nonnegative based on producer prices p. Finally, the government collects taxes on commodity
transactions between the consumer and firm. The government also purchases a fixed bundle of
goods g with any government balance (negative or positive), resulting in a transfer to the
consumer b. Letting e(p + t, u) be the expenditure function dual of u(x) for consumer prices p +
t, we can write the equilibrium conditions as:
Household budget: e(p + t, u) = b + (p + t) · k (11)
Government budget: t · x(p + t, u) = b + p · g
Zero profit: p · y(p) = 0
Market clearing: y(p) α + k = g + x(p + t, u),
where utility u, producer prices p, output scale α, and government transfers b adjust to reach an
equilibrium. By Walras law, one of the budget or market-clearing equations is redundant. To fix
the nominal variables p, t, and b, one price must also be fixed as numeraire. A typical approach
is to fix first good as numeraire (p1 = 1) and to drop the market-clearing constraint on that first
good. This model is a simplified version of the kinds of models used for analytic and numerical
work on public good provision in a second-best general equilibrium setting (Diamond and
Mirrlees 1971; Diewert 1983; Stiglitz and Dasgupta 1971; Bovenberg and Goulder 1996;
Rutherford 1999).
An important observation about these equilibrium conditions is that to close the model, it
is necessary to make an additional assumption about how the government meets its budget
constraint. In all, the model (11) involves N + 1 endogenous variables—p (except p1), u, and α—
Resources for the Future Pizer and Kopp
and N + 2 equations (recall that one is dropped by Walras law). One additional degree of
freedom must be introduced among the otherwise fixed government variables, t, g, or b. The
approximation in (10) assumed that taxes and government purchases were fixed, thereby
endogenizing government transfers b by default. Alternatively, b can be fixed and the tax rates
can be chosen optimally subject to the budget constraint. Or, any one of the tax rates can be
allowed to adjust and b can be fixed.31 When we consider policy evaluations, these different
assumptions often have important consequences for the implied welfare change.
Using this model, the cost associated with providing some incremental amount of a
public good j dg =ede can be computed two ways. In an analytic model, we can maximize u
subject to the above constraints. If γ is the Lagrange multiplier on the government budget
constraint, and ρ
j is the Lagrange multiplier on the jth market-clearing constraint, the social cost
of the incremental public good will be
( ) j j γ p +ρ de (12)
In a numerical simulation model we can simulate the change du as we increase g by dg. Using a
money-metric utility scaling (Samuelson 1974) such that e(q, u) = u at the initial equilibrium, du
can be interpreted as the equivalent variation cost of the policy for a small dg.
Stiglitz and Dasgupta (1971), Ballard and Fullerton (1992), and others use the analytic
approach to derive results with optimal taxes and no transfers. They show that the cost of a
public good equals its price multiplied by one plus the marginal excess burden of the tax system.
From expression (12), optimal tax policy must imply that ρ
j is proportional to pj if this rule is
going to hold for all goods j. This is similar to the results of Bovenberg and de Mooij (1994) and
Bovenberg and Goulder (1996), who consider optimal Pigouvian taxes on pollution in the
presence of distortionary taxes. Work by Bruce and Harris (1982) and Diewert (1983) reveals the
form of ρ
j when taxes are instead exogenous and transfers adjust. In contrast to Stiglitz and
Dasgupta, they find ρ to be a matrix-weighted average of producer and consumer prices.32
Boadway (1975) provides guidance when some taxes are fixed and other taxes adjust.
31 In this case, we can measure the excess burden of a particular tax by varying b.
32 This result can be seen in (10) by taking C’(a) to be a producer price p and replacing t with q – p.
Resources for the Future Pizer and Kopp
3.5. Numerical Analysis
Analytical models are limited by the need for tractability. Numerical models, however,
are limited only by computational power. An early example of numerical general equilibrium
analysis is Ballard et al. (1985), who use a numerical general equilibrium model of the U.S.
economy to estimate marginal excess tax burdens ranging from 17 to 56 cents on the dollar.
Environmental applications of numerical general equilibrium models include Conrad and
Schröder (1991), Bergman (1991), Whalley and Wigle (1991), and Böhringer and Rutherford
(1997). Bergman considers the loss of gross national product (GNP) and marginal control cost of
SO2, NOx, and CO2 in Sweden. Conrad and Schröder use a computable general equilibrium
model to estimate the carbon tax required to generate a specified annual emissions reduction and
then measure the impacts on GNP and SO2 and NOx emissions in Germany. Böhringer and
Rutherford also consider carbon policies in Germany, but unlike Bergman and Conrad and
Schröder, they focus on welfare consequences. Similarly, Whalley and Wigle consider global
consequences of CO2 reductions with an emphasis on welfare as well as changes in patterns of
When the control costs used in general equilibrium analyses are based on surveys of
regulatory expense, there is often little information concerning the removal process or input
substitutions used to reduce emissions. Consequently, most models make simplifying
assumptions about abatement technology: either input usage for abatement is proportional to
overall input usage33 (Jorgenson and Wilcoxen 1990; Hazilla and Kopp 1990), or abatement
involves only capital and labor (Ballard and Medema 1993). Recent work by Nestor and Pasurka
(1995) demonstrates the potential problems with that approach, using detailed information on
abatement technology (Schäfer and Stahmer 1989). They find that although simplifying
assumptions about abatement technology lead to similar conclusions about which sectors are
most adversely affected by regulation (based on declines in output), the magnitudes are
significantly different. The proportional input model tends to underestimate output changes, but
the capital-labor model overestimates output changes. This occurs, the authors argue, because the
primary factor (capital-labor) intensity of the actual expenditures lies between the primary factor
intensity implied by the proportional input and capital-labor models.
33 This is equivalent to assuming that environmental regulation is a Hicks-neutral productivity effect.
Resources for the Future Pizer and Kopp
3.6. Environmental Policy versus Public Good Provision
The question of optimal provision of public goods motivated many of the analytic
exercises we have mentioned (Diamond and Mirrlees 1971; Diewert 1983; Stiglitz 1979). These
studies focused on the problem of procuring market goods for public use and on the appropriate
way to value those goods, leading to Equation (8) and subsequent variations. In contrast,
environmental regulation focuses on nonmarket goods and frequently involves the manipulation
of property rights. For example, emissions standards may restrict the right of firms to emit above
a certain rate without any direct market consequence.
The intuition behind Equation (8) is the same in both cases, however. There is one market
where the public policy imposes primary losses of social surplus. In the first case, increased
government consumption of a particular good raises production costs in that market in order to
increase output. In the second case, environmental regulation raises production costs in the
regulated firms’ output market while the output level remains the same. The first term in (8)
reflects these direct costs. In response to these changes, there will be equilibrating effects in
other markets. Tax-induced discrepancies between the marginal cost and marginal benefit of
goods in those markets lead to welfare changes reflected in the second term in (8).
When market mechanisms are used to reduce pollution, we can focus on the pollution
market directly, rather than the output market for regulated firms. The initial factor supply curve
in this market is flat and passes through the origin—equilibrium emissions set marginal
abatement costs equal to zero. The direct cost of any regulation is the change in area under the
pollution demand curve or the integral ∫ p (a)da, where a measures the quantity abated and p(a)
is the market price at abatement level a (or the tax required to induce abatement level a). Indirect
welfare changes, as before, are computed as the sum of changes in consumption times the tax
rate in each market.
The use of market-based mechanisms—tradable permits and emissions taxes—changes
the otherwise straightforward application of earlier results from public finance. In addition to
encouraging abatement and creating pollution control costs, these mechanisms also associate
costs with uncontrolled emissions. These costs reflect the value of permits or taxes associated
with unabated emissions. In other words, firms must pay for their inframarginal emissions, not
just the emissions they reduce. Yet these costs on inframarginal emissions are not really costs at
all—they are transfers to whoever owns the emission property rights. This could be firm owners,
consumers, or the government. In either case, this transfer is an ancillary feature of market-based
environmental regulation that is not present in the standard case of public good provision.
Resources for the Future Pizer and Kopp
Goulder et al. (1998) discuss the different outcomes created by market and nonmarket
Goulder et al. (1997) echoes our earlier analysis and notes that environmental regulations
can lead to large deadweight losses in the labor market—distinct from the out-of-pocket costs of
environmental regulation—if labor taxes are high and labor supply declines (even a small
amount) in response to the regulation. This is an important point to bear in mind when
considering the ability to raise government revenue (via permit auctions or tax payments).
Revenue raised by market-based policies can reduce the social cost of environmental policies if
the new revenue is used to offset tax distortions elsewhere in the economy.
Recognition that market-based regulatory mechanisms can be used to raise government
revenue, and that this revenue can be used to reduce existing taxes and lower the economy’s
overall tax and regulatory burden, has lead to an explosive amount of research into so-called
revenue recycling. This research in turn has lent additional credibility to arguments for the
auctioning of emissions permits rather than allocating them gratis.
3.7. The Double Dividend
The potential for revenue-raising environmental regulation to both reduce pollution and
lower the cost of the tax system has been referred to as a double dividend. Specifically, the
second dividend is the reduction in tax distortions caused by the replacement of undesirable taxes
on labor and capital with environmentally beneficial taxes on pollution (Pearce 1991). However,
this remains an empirical question: setting aside any environmental benefit, can the substitution
of taxes on pollution for existing taxes in fact lower the distortionary cost of the tax system as a
Goulder (1995b) points out that there are several versions of the double dividend. The
weak version argues that the tax swap (green taxes replace conventional taxes) raises welfare
relative to an identical environmental tax policy that returns the revenue lump-sum. In contrast,
the strong version argues that swapping environmental taxes for a typical or representative
distortionary tax (e.g., labor) is either welfare-neutral or welfare-improving relative to the nopolicy
Resources for the Future Pizer and Kopp
The weak form of the double dividend hypothesis is relatively uncontroversial. As
Goulder explains, as long as existing taxes are indeed distortionary, using new revenue to reduce
such distortions—versus returning it lump-sum—must reduce the distortion and therefore raise
welfare relative to the case where the revenue is returned lump-sum.34 This is the main point of
Goulder et al. (1997). The stronger forms, however, have provoked considerable debate. An
important insight comes from Bovenberg and de Mooij (1994), who demonstrate that the
hypothesis hinges on the sign of the uncompensated elasticity of labor supply. A negative
elasticity supports the hypothesis; a positive elasticity rejects it. With most empirical studies
favoring positive values (Hausman 1985), this has been viewed as substantial evidence against
the strong version of the double dividend hypothesis.35
Nonetheless, the Bovenberg and de Mooij result is based on a relatively simple model
that ignores many features of the economy and the tax code. Bovenberg and van der Ploeg
(1996) consider a model with involuntary unemployment and find more support for the strong
double dividend. Simulation results from more complex economic models summarized in
Goulder (1995b) lean against the strong version of the double dividend hypothesis, but the
evidence is mixed. Most of the results indicate that substitution of environmental taxes for other
taxes lowers welfare, though several indicate marginal losses or welfare improvements.
Therefore, despite a general sentiment against the strong double dividend hypothesis, the
question remains open.
3.8. Dynamic General Equilibrium Analysis
Static models are limited by the fact that they provide a snapshot of policy consequences
at a single point in time. Since the cost of an environmental policy may vary over time, a single
snapshot can be misleading. Static models are also limited by their inability to model savings and
investment correctly. Savings and investment decisions reflect a trade-off among consumption in
different periods of time. Since investment is a significant portion of final demand, static models
may not even present an accurate snapshot of the period they attempt to describe, if there is a
large impact on investment—a point made via the example at the beginning of the section.
34 Although the efficiency properties of lump-sum versus distortionary taxes are uncontroversial, the equity
properties are not. Kaplow (1996) provides an example of such concerns.
35 A recent paper by Jaeger (2000) argues against the Bovenberg and de Mooij result but has not been widely
Resources for the Future Pizer and Kopp
One approach to dynamic analysis is solving a system of linked static equilibrium
models, as demonstrated by Ballard et al. (1985) and Conrad and Schröder (1991). These models
involve a myopic, intertemporal trade-off where consumers assume the current return to capital
investment is fixed. True dynamic general equilibrium models, however, solve these problems
by simultaneously satisfying market equilibrium and budget constraints in every period.
In our model (11), three modifications are required. First, we index quantities and prices
over both goods and time. Second, we have to consider utility over both goods and time,
resulting in an intertemporal expenditure function. Third, our endowments k are fixed in the
zeroeth period but otherwise evolve based on investment i, an additional element of final
demand. The modified model becomes
Household budget: e(p + t, u) = b + (p0 + t0) · k0 (13)
Government budget: t · x(p + t, u) = b + p · g
Zero profit: pt · y(pt) = 0
Market clearing: y(pt) α
t + kt = gt + xt(p + t, u) + it
Capital accumulation: kt+ 1 = (I – Δ )kt + f(it ),
where I is the identity matrix and Δ is a diagonal matrix of depreciation rates. Note that the
household and government budget constraints are now intertemporal constraints, based on the
intertemporal prices p. The vector-valued function f() describes the tranformation of investment
goods into new capital. This could be a simple linear relationship describing how outputs in one
period become inputs the next period. Or it could be a convex relationship representing the
adjustment costs associated with rapid increases in the capital stock. Finally, we note that this
general specification allows for multiple capital stocks that may be malleable between uses.
Models of the form (13) are not generally solvable using analytic means and are instead
solved by numerical algorithms (Codsi et al. 1992; Lipton and et al. 1982; Rutherford 1999).
Hazilla and Kopp (1990) and Jorgenson and Wilcoxen (1990) provide the best examples of
dynamic general equilibrium modeling applied to environmental regulation. The main finding of
those studies is that static productivity losses due to environmental regulation are amplified by
the long-term effect on capital accumulation. Intuitively, environmental regulation lowers the
marginal product of capital. In the long term, this leads to a lower capital stock, decreased
output, and reduced welfare. The additional cost of this accumulation effect on welfare can be as
much as 40% above the static cost that ignores changes in the capital stock.
Resources for the Future Pizer and Kopp
When we consider the transition between unregulated and regulated regimes,
assumptions about capital stock malleability become important. Models are frequently
categorized as either “putty-putty” when they assume a single capital stock that can be moved
costlessly between uses (see, for example, Hazilla and Kopp 1990 and Jorgenson and Wilcoxen
1990) or “putty-clay” when they assume specialized capital stocks that must be retired and
replaced (Atkeson and Kehoe 1999) or incorporate convex adjustment costs (Goulder 1995a).
This is especially relevant for recent applications concerning the prospective costs of climate
change mitigation efforts (Jorgenson and Wilcoxen 1993; Goulder 1995a; Weyant and Hill 1999;
McKibbin et al. 1999). Although Jorgenson and Wilcoxen require taxes of $25 per ton of carbon
emissions in 2020 to stabilize emissions at 1990 levels, Goulder requires a tax of $50 per ton.
This partially reflects Goulder’s adjustment cost specification vis-à-vis the putty-putty
specification in Jorgenson and Wilcoxen. The possibility of significant adjustment costs due to
less malleable capital stocks emphasizes the importance of dynamic efficiency in long-term
stabilization efforts, referred to as “when flexibility” by Wigley (1996).
3.9. Other “Costs” in General Equilibrium Models
The attention given to specific areas of public concern—productivity growth,
employment, and trade—has led to an emphasis on these issues as alternative measures of “cost.”
Jorgenson and Wilcoxen (1990), for example, focus on the growth consequences associated with
environmental regulation rather than welfare measures—even though their general equilibrium
model can measure welfare effects. Other studies employ nongeneral equilibrium models to
compute these effects. Christainsen and Havemen (1981) survey a collection of studies on the
productivity slowdown of the 1970s and conclude that perhaps 8–12% of the slowdown can be
attributed to environmental regulation. A more recent survey of the literature by U.S. Office of
Technology Assessment (1994) concluded that as much as 44% might be attributable to
environmental protection in individual industries. Looking at specific industries, Barbera and
McConnell (1990) attribute 10–30% of the slowdown to environmental regulations. More recent
concerns have included trade impacts (McKibbin et al. 1999) and labor consequences
(Morgenstern et al. 1998).
Resources for the Future Pizer and Kopp
4. Distribution of Costs
Cost and cost-benefit analyses of environmental regulation tend to focus on aggregate
consequences. Yet these consequences are not evenly distributed among all members of society.
For many reasons, the distribution of consequences and, specifically, costs is important.
First and foremost, most societies have some desire for equality and fairness. This is
revealed by the existence of, for example, civil rights, progressive income taxes, and welfare
programs. Equality can be defined in many ways and is subject to intense debate. Regardless of
the definition, however, the costs associated with a particular regulation or program can be held
to the definition and evaluated. The same evaluation can also be used to design compensatory
schemes to redress identified inequalities.
In cases of international environmental policy, the distribution of costs across countries is
a point of negotiation. Notions of equality and fairness play an important role in the design of
international agreements, with the additional complication that countries participate voluntarily.
This makes a careful analysis even more important, as no supranational authority exists to
resolve differences.
Even without a desire for fairness, there are frequently practical considerations that
motivate a focus on distribution. When costs are narrowly focused on a few stakeholders—such
as restrictions on Northwest logging companies to protect wildlife habitats, or restrictions on
chemical companies manufacturing chlorofluorocarbons to protect stratospheric ozone—those
stakeholders are likely to be more vocal than when costs are spread over a diffuse group. It often
makes sense for policymakers to work with these stakeholders to design policies that at least
minimize their distress.36
Practical concerns often become political concerns, especially when costs are
concentrated geographically. Concentration of costs may arise because of a concentration of
polluting raw materials (coal mines in West Virginia), a concentration of energy-intensive or
pollution-intensive productive activities (steel mills in Pennsylvania), or a concentration of
consumers of energy-intensive or pollution-intensive final goods (heavy use of home heating oil
in New England). All of these effects are ameliorated to the extent that environmental benefits
are primarily local, raising questions of environmental federalism (Oates and Schwab 1996).
36 These observations are attributed to Olson (1965).
Resources for the Future Pizer and Kopp
Finally, environmental policies frequently transcend generations, with one generation
bearing the costs and another the benefits (e.g., current mitigation efforts to prevent future global
warming). Unlike the previous distributional concerns, there are currently no advocates for future
generations to draw attention to equity, practical, or political issues. Nonetheless, concern over
this generational equity continues to influence the policy process.
In the remainder of this section, we summarize different approaches for measuring the
distribution of regulatory costs broken down by household, sector, region, and generation.
4.1. Impacts by Household
Given the pervasiveness and magnitude of environmental regulation, one would think
that comprehensive studies of the cost and benefit distribution of these policies would be
bountiful. Ironically, the contrary is true. What does exist is a small set of fairly straightforward
analyses of price changes and the effect of those changes on household budgets, coupled with an
even smaller number of sophisticated general equilibrium studies.
A recent study by Metcalf (1999) provides one kind of analysis. Metcalf uses inputoutput
tables to construct estimates of the pollution content of different commodities purchased
by consumers. Under a proposed green tax reform that would tax consumption in proportion to
its pollution content, he can then estimate effective rates on each commodity. Finally, he
constructs consumption bundles differentiated by income, marital status, and age based on the
Consumer Expenditure Survey.37 Combining these calculations, he is able to compute the tax
burden for different demographic groups. In a report by the Congressional Budget Office (2000),
similar calculations are carried out concerning the cost to different income groups of a $100-perton
carbon tax. Bull et al. (1994) consider the impact of both a carbon tax and a Btu tax on
households differentiated by income and Census region.38
These studies indicate that pollution taxes in isolation tend to be regressive, with poorer
households spending a larger fraction of their income on environmental taxes. As Metcalf points
out, compensatory reductions in income and payroll taxes targeted at poorer households can
37 Sabelhaus (1996) discusses use of the Consumer Expenditure Survey; Lawson (1997) discusses input-output
38 There are four Census regions for individuals in urban areas (West, Northeast, South, and Midwest), plus rural.
Resources for the Future Pizer and Kopp
offset the regressive effect. Of course, alternative tax reforms could further heighten the
Dowlatabadi et al. (1995) consider a different question. Instead of broad consumption
bundles, they focus on differences in energy consumption. They analyze energy price increases
arising from carbon taxes and trace the impact of these price increases through three alternative
household energy demand models, which vary by region. Results of this exercise suggest that
costs (in terms of household energy expenditures) between the lowest- and the highest-cost
region differ by as much as 45%.
4.2. Households, General Equilibrium, and Social Welfare
The preceding examples provide cost estimates differentiated by household but ignore the
general equilibrium considerations discussed in Section 3. General equilibrium analysis of
distributional impacts is complicated by both data and computational requirements. Two
approaches have emerged. One considers restricted models of household preferences that (owing
to the restrictions) can be combined easily into a model of aggregate behavior (though not
necessarily via a representative agent; see Jorgenson et al. 1980). The general equilibrium model
can be solved using the simplified model of aggregate behavior, and then individual demand and
welfare can be recovered afterward. Jorgenson et al. (1992) use this approach to study the impact
of carbon taxes on the economic welfare of 1,344 distinct household types. Similarly, Aasness et
al. (1996) consider the effect of environmental taxes in Norway.
A second approach is to model explicitly the behavior of distinct households. In our
earlier general equilibrium model (11), we need to specify distinct endowments for each
household. We then add one equation (a household budget constraint) and one endogenous
variable (utility) for each additional household, leading to a modified model,
Household budgets: ej(p + t, uj) = bj + (p + t) · kj (14)
Government budget: Σj t · xj(p + t, uj) = Σj bj + p · g
Zero profit: p · y(p) = 0
Market clearing: y(p) z + Σj kj = g + Σj xj(p + t, uj),
39 Both Bull et al. (1994) and Metcalf (1999) argue that the regressivity declines when one considers the lifetime
incidence of these taxes.
Resources for the Future Pizer and Kopp
where the definitions are identical to (11) except that household utility (uj), expenditures (ej),
lump-sum transfers (bj), and demand (xj) are now indexed over households j.
The down side to this approach is that computational constraints limit the number of
distinct households. For that reason, it is not usually applied to disaggregate analysis. Such
models are, however, the standard approach to multiregion, international general equilibrium
analyses, discussed in the next section.
In addition to modeling the welfare of distinct households, Jorgenson et al. (1992) also
consider the question of how to quantify in an internally consistent manner society’s aversion to
inequality. They develop a social welfare function with many desirable properties and then use
this function to measure the social cost, inclusive of equity concerns.40 Depending on the
aversion to inequality, they find a welfare loss of $187 billion to $249 billion—suggesting that
equity concerns alone might increase the social cost of policies by a third.
4.3. Multicountry Analysis
Modeling distributional effects across countries or regions is entirely analogous to
measuring distributional effects across households. Regions have their own endowments and
then trade on international markets. Thanks to recent interest in modeling mitigation efforts to
reduce global climate change, a host of such disaggregated global models exist. A recent special
issue of Energy Journal (Weyant and Hill 1999) presents results from more than a dozen models
simulating the consequences of the Kyoto Protocol.
The Second Generation Model (MacCracken et al. 1999), for example, has been used to
study the impact of different implementations of the Kyoto Protocol. In particular, the model is
capable of estimating regional costs associated with various assumptions about international
trading of emissions permits and the supply of additional emissions rights from developing
countries. They demonstrate that substantial cost savings—for all countries—are possible with
international permit trading. Similar analyses are carried out by McKibbin et al. (1999), with a
particular emphasis on trade flows and exchange rate appreciation. They argue that the Kyoto
Protocol leads to significant exchange rate appreciation in developing countries, fed by an
increase in capital inflows and possibly the export of emissions credits. While boosting
40 These properties include unrestricted domain, independence of irrelevant alternatives, positive association,
nonimposition, and cardinal full comparability (see Jorgenson and Slesnick 1983).
Resources for the Future Pizer and Kopp
developing country income, this has a Dutch disease effect, whereby the higher exchange rate
makes their exports less competitive on world markets. They report a decline in developing
countries’ exports of approximately 20%. Other international models with regional detail have
been used to address costs associated with capital malleability (Jacoby and Wing 1999),
incentives for regional participation (Peck and Teisberg 1999), and cost-effectiveness (Nordhaus
and Boyer 1999), among other topics.
4.4. Impacts by Sector
Within the United States (and we expect elsewhere), the economic analysis of a proposed
environmental regulation begins with a quantification of the direct compliance cost. To the
extent that most major regulations have focused on commercial activities, these compliance costs
are borne by production sectors of the economy. As a consequence, production sector studies of
regulatory cost are quite common, considerably more so than household-level studies.
The vast majority of these studies consider only the costs borne by the directly regulated
sector(s). Since 1973, the Pollution Abatement Cost and Expenditure Report has provided
detailed information on expenditures by different sectors in the United States (Bureau of the
Census 1973-1997). Since 1975, the Bureau of Economic Analysis has supplemented that report
with its own analysis (Vogan 1996). Germany has reported even more detailed information,
beginning in 1980 (Schäfer and Stahmer 1989). These studies consistently identify the same
sectors as spending the most per dollar of revenue: steel, petroleum refining, plastics, and paper,
to name a few.
In many cases this is adequate for the purposes of assessing differential impact. However,
in the case of large regulatory programs, especially programs that affect important economic
sectors (e.g., energy), secondary impacts can be significant. In a series of papers beginning with
Hazilla and Kopp (1990) and Jorgenson and Wilcoxen (1990), large-scale, multisector,
computable general equilibrium (CGE) models were used to examine the costs of environmental
regulation and to assess the distribution of those costs across different sectors of economy. Both
Hazilla-Kopp and Jorgenson-Wilcoxen assessed the impact of the U.S. Clean Air and Clean
Water acts. In addition to estimates of macroeconomic changes, these structurally similar models
enabled one to model changes in total cost, output, employment, and capital accumulation
resulting from environmental regulation, and to do so in a dynamic setting.
The results of these models revealed the wide range of sectoral impacts brought about by
major regulatory programs. One example illustrates the point. Hazilla and Kopp report that in
Resources for the Future Pizer and Kopp
1990 the output of the finance, insurance, and real estate sector was almost 5% lower than it
would have been in the absence of the Clean Air and Clean Water regulations and that
employment declined 2.5%. These are costs borne by a sector that had no direct compliance cost
New CGE models take the analysis of regulation and sectoral impacts into the realm of
policy design. Current debates over global climate change and the policy measures needed to
restrict the emissions of greenhouse gases acknowledge that, at least in the near term, these
polices will be focused on the combustion of fossil fuels—most notably coal consumed in
electricity generation. Efforts to restrict carbon emissions severely will impact the coal mining
and processing sector, electricity utilities, and to some extent, other fossil energy industries. A
recent study by Bovenberg and Goulder (2000) uses a CGE model to examine the impact of
carbon taxes. In particular, it determines the level of compensation that would be required to
compensate capital owners in the energy sector for their losses. Climate policies that generate
revenues (e.g., carbon taxes or auctioned permits) could then be combined with compensation
schemes to address directly the sectoral cost distribution issue.
It is important to bear in mind that not all environmental regulations have negative
sectoral impacts. Cap-and-trade permit policies can generate large rents for regulated sectors if
permits are allocated on a gratis basis. The point has recently been made by Bovenberg and
Goulder (2000).
4.5. Impacts by Region
Many studies of distribution by household also consider regional effects. Jorgenson et al.
(1992), Dowlatabadi et al. (1995), and Bull et al. (1994) all discuss impacts at broad regional
levels (between four and nine regions) in the United States. More detailed impacts of broad,
national policy initiatives are difficult because detailed data are lacking. Of course, case studies
of narrow regulatory efforts are often sufficiently detailed to identify actual payees (e.g., Deck
A related line of work has explored whether environmental regulation influences the
choice of plant location. Such choices could have significant local economic impacts that would
appear in few, if any, of the cost measures discussed so far. Studies by Bartik (1988), Low and
Yeats (1992), and Crandall (1993) suggest that firms are sensitive, in general terms, to cost
variations among states when deciding where to locate new facilities. However, there is little
direct evidence of a relationship between stringency of environmental regulation and plant
Resources for the Future Pizer and Kopp
location choices. In an analysis that includes measures of environmental stringency, Bartik found
that neither measures of expenditures nor emissions standards had significant effects on plant
location decisions. These results are similar to those of Levinson (1996) and McConnell and
Schwab (1990), although Levinson did find that the locations of new branch plants of large
multiplant companies in pollution-intensive industries were somewhat sensitive to differences in
regulations. In contrast, a recent study by Gray (1996) finds that states with more stringent
regulation (measured by a variety of state-specific measures) have fewer plant openings.
Rather than directly examining plant location decisions, other work has compared rates of
manufacturing employment growth—not just new plants—in attainment areas versus
nonattainment areas. Papers by Henderson (1996) and Kahn (1997) found relatively lower
growth rates in manufacturing employment in nonattainment counties, compared with those that
attained the air quality standard. Becker and Henderson (1997) found that environmental
regulation reduced births and increased deaths in nonattainment areas, shifting polluting activity
to cleaner areas. With a similar approach, Greenstone (1997) estimates an annual loss of about
8,000 jobs in nonattainment areas over the period 1972–1987. Importantly, his estimates assume
that employment growth at polluting plants in less regulated areas is an appropriate control group
from which to infer the likely change in employment in the absence of regulation.
Pollution control is not the only source of environmental regulation. When one looks to
the future, one sees a growing demand for land use restrictions, fueled in part by environmental
concerns and in part by other motivations. By their very nature, land use restrictions are local or
regional and therefore might be expected to give rise to regional economic disparities—though
without the controversy of federally dictated policy. For example, northern Virginia (the area
nearest Washington, D.C.) is the Silicon Valley of the East. It has experienced rapid employment
and income growth fed by an equally rapid commercialization of once-rural countryside. Had
severe restrictions on land use been in place 15 years ago, one can argue that the current
economic environment might have been negatively affected. The exact effect of current
restrictions on future growth is a topic of intense debate (Webster and Wu 1999a, 1999b;
Alavalapati et al. 1996). In any case, it seems likely that current restrictions will have some
regional economic cost.
4.6. Intergenerational Issues
Many environmental issues, such as hazardous waste disposal, habitat and species
preservation, and global climate change, involve consequences that span generations. Although
Resources for the Future Pizer and Kopp
standard economic theory discounts future consequences quite rapidly, ethical arguments suggest
a more equal treatment.41 Arrow et al. (1996) summarize these opposing views, but the issue
remains unresolved. Recent advances in economic modeling have allowed one to model
discounting within generations without imposing discounting across generations, providing a
formal framework for analysis (see overlapping generations models of Howarth 1998;
Bovenberg and Heijdra 1998). Again, the underlying question of how to compare
intergenerational utility remains unsettled.
The significance of this issue cannot be understated, as small differences in the discount
rate lead to enormous differences in valuation. This feature alone has led to the observation that
ordinary market fluctuations in interest rates could lead to the use of lower rates—even distinct
from the intergenerational concern (Newell and Pizer 2003; Weitzman 1998). Given the
significance of intergenerational welfare in environmental policy, we leave a more complete
discussion of the topic to Chapter XXX of this handbook.
5. Conclusions
Accurate analysis of the social cost of environmental regulation requires a sophisticated
application of welfare economics. This analysis begins with attention to nonpecuniary costs and
unmeasured consequences at the firm and household level, continues with general equilibrium
effects—including tax distortions, revenue recycling, and capital accumulation—and concludes
with particular attention to the distribution of costs and consequences. Each part of this analysis
has important lessons to offer both the practitioner and the researcher.
Without a randomized experiment to understand the consequences of new regulation, it is
impossible to speak confidently about the costs borne by firms. There are reasonable arguments
suggesting that surveys and engineering studies could both under- and overestimate compliance
costs. Case studies, econometric analyses, and retrospective analysis remain inconclusive,
although one could interpret this to mean that such measures are, on average, unbiased.
Economy-wide analyses identify two important effects ignored by even the most
thorough direct cost analysis: welfare losses associated with increased tax distortions and the
potential to offset at least some of these losses with revenue-raising environmental taxes. The
41 Ramsey (1928) called the use of anything greater than a zero percent discount rate between generations
Resources for the Future Pizer and Kopp
former concern, trumpeted by economists for more than 30 years, remains outside the scope of
ordinary policy analysts. The latter concern, recently picked up by environmental advocates
(Repetto and Austin 1997), remains poorly understood outside—and sometimes among—
environmental economists.
Finally, most cost analyses ignore the differential cost of environmental regulation based
on demographics, especially income. Costs are not borne evenly, and many regulations appear to
be regressive. When revenue from environmental taxes is used to cut incomes taxes, especially
on capital, the effect is acutely regressive. The same is true when marketable permits are
provided gratis to owners of capital. An interesting question is whether, given these concerns, the
recent trend toward market-based regulation over command-and-control approaches improves
the well-being of all consumers.
Despite the difficulties, one often hears the refrain that costs are easier to estimate than
the benefits of environmental regulation, or that regulatory cost estimates are simple data
collection and accounting exercises (as performed by the Department of Commerce and
published in the PACE reports). This disparity of view is due, we believe, to the prevailing
opinion among many in the policy community, and some trained economists as well, that the
cost of environmental regulation is synonymous with the private compliance cost contained in
self-reported surveys and engineering studies. The theory and evidence we present in this paper
are meant to dispel that myth and to argue that, indeed, the private compliance costs may be only
a fraction of the true social cost of regulation. Furthermore, unless one takes a very simplistic
view of the policymaking process, the inability to measure benefits accurately does not diminish
the importance of accurate cost measurements.
As we look toward the future of regulatory cost analysis, several topics stand out as both
practical concerns and important research topics. Although private costs seem to be measured
adequately by surveys and, to a lesser extent, engineering studies, social costs consistently
deviate from these costs because of interactions with distortionary taxes in other markets as well
as the effects of various recycling schemes (in the case of revenue-raising policies). Further,
most environmental policies have important equity consequences—regionally, sectorally,
demographically, and intertemporally—that should not be ignored. The most challenging area
Resources for the Future Pizer and Kopp
for future work, however, will likely be improved understanding and modeling of the process of
technological change.42
As pollution control moves away from end-of-pipe abatement and toward pollution
prevention and process changes, it becomes increasingly difficult to identify the operating costs
associated with environmental protection. Even worse, as we turn to research and new
technology to provide cleaner alternatives to polluting activities, the cost of these research and
development activities is virtually unknown. In retrospective studies, it is difficult to know what
improvements may have occurred elsewhere in the absence of environmentally focused R&D
activities. In prospective studies, it is difficult to know how much improved technologies will
Consider the example of global climate policy and, particularly, efforts to reduce carbon
dioxide emissions. Economists would agree that in general, effective policies to limit carbon
dioxide emissions should raise the private of cost of carbon emitted into the atmosphere in the
short run (through the use of tax or permit systems). The price rise will stimulate conservation of
carbon-containing fuels and provide incentives for the development of noncarbon energy
technologies in the future. The price increases will also have a dynamic effect on the distribution
of resources devoted to research and development, innovation, and commercialization, with
relatively more resources going to carbon-saving research and less elsewhere. What we do not
know, and have only begun to conceptualize, is the effect of this altered resource distribution on
social welfare. Will resources be diverted from medical research, nanotechnology, and
telecommunications? And if they are, what social gains have we lost so that we can have carbonfree
Another way to view this challenge is to consider the difficulty of evaluating different
outcomes characterized by large changes in relative prices (e.g., carbon and other pollutants)
over long periods of time. Existing approaches have tended to emphasize marginal changes in
consumption and productions, with local preferences and technology estimable from recent data.
As we consider nonmarginal changes over long periods of time, new tools for cost analysis will
undoubtedly be needed.
42 This topic is the subject of Chapter XXX of this handbook.
Resources for the Future Pizer and Kopp
1992 1993 1994
Pollution Abatement and Control 104.6 110.0 121.8
Pollution Abatement 100.5 105.8 117.6
Consumers 7.9 8.4 9.8
Business 65.9 69.0 76.6
Government 26.6 28.4 31.2
Regulation and Monitoring;
Research and Development
4.2 4.2 4.2
Portion of Expenditures from PACE Survey 21%
Portion of Expenditures from Government
Finance Survey/Census 22%
Source: Vogan (1996), Tables 2 and 10.
Table 1: Current Spending ($Billions) on Pollution Abatement and Control
Resources for the Future Pizer and Kopp
Elastic demand,
no labor tax
Elastic demand, with
labor tax
“Cost” to firm 0 c⋅x 1 c⋅x 1 c⋅x
Partial equilibrium 0 c⋅x 1 ( )
2 1 0 c⋅ x+x 1 ( )
2 1 0 c⋅ x+x
GDP lossa 0 −c⋅x
= l0 – l1
( ) 1 0 1 0 −c⋅x + x −x p
= l0 – l1
( ) 1 0 1 0 −c⋅x + x −x p
= l0 – l1
True welfare loss
(general equilibrium) 0 c⋅x 1 ( )
2 1 0 c⋅ x+x 1 ( ) ( )
2 1 0 1 0 c⋅ x+x +t⋅ l−l
aNegative numbers indicate gains.
Table 2: Different Cost Measures and Different Modeling Assumptions
Resources for the Future Pizer and Kopp
00 2 4 6
x1 x0
00 2 4 6
Fixed demand Downward-sloping demand
Figure 1: Effect of Environmental Controls on Consumer Surplus, No Labor Tax
(shaded area shows deadweight loss)
Resources for the Future Pizer and Kopp
00 2 4 6
x1 x0 l1 l0
0 0 2 4 6 8 10
Energy market Labor market
Figure 2: Effect of Environmental Regulation with Preexisting Labor Tax
(shaded area shows deadweight loss)
Resources for the Future Pizer and Kopp
Aasness, Jorgen, Torstein Bye, and Hans Terje Mysen. 1996. Welfare Effects of Emission Taxes
in Norway. Energy Economics 18 (4):335-46.
Alavalapati, Janaki, Bill White, Pam Jagger, and Adam Wellstead. 1996. Effect of Land Use
Restrictions on the Economy of Alberta: A Computable General Equilibrium Analysis.
Canadian Journal of Regional Science 19 (3):349-65.
Arrow, K.J., W.R. Cline, K.-G. Maler, M. Munasinghe, R. Squitieri, and J.E. Stiglitz. 1996.
Intertemporal Equity, Discounting and Economic Efficiency. In Climate Change 1995 –
Economic and Social Dimensions of Climate Change, edited by J. P. Bruce, H. Lee and
E. F. Haites. Cambridge, UK: Cambridge University Press.
Atkeson, Andrew, and Patrick J. Kehoe. 1999. Models of Energy Use: Putty-Putty versus Putty-
Clay. American Economic Review 89 (4):1028-43.
Atkinson, Scott E., and Donald H. Lewis. 1974. A Cost-Effectiveness Analysis of Alternative
Air Quality Control Strategies. Journal of Environmental Economics and Management 1
Ballard, Charles L., and Don Fullerton. 1992. Distortionary Taxes and the Provision of Public
Goods. Journal of Economic Perspectives 6 (3):117-31.
Ballard, Charles L., and Steven G. Medema. 1993. The marginal efficiency effects of taxes and
subsidies in the presence of externalities: A computational general equilibrium approach.
Journal of Public Economics 52:199-216.
Ballard, Charles L., John B. Shoven, and John Whalley. 1985. General Equilibrium
Computations of the Marginal Welfare Costs of Taxes in the United States. The American
Economic Review 75 (1 (March)):128-138.
Barbera, Anthony J., and Virginia D. McConnell. 1990. The Impact of Environmental
Regulations on Industry Productivity: Direct and Indirect Effects. Journal of
Environmental Economics and Management 18:50-65.
Bartik, Timothy J. 1988. The Effects of Environmental Regulation on Business Location in the
United States. Growth Change 19 (3):22-44.
Resources for the Future Pizer and Kopp
Bazerlon, Coleman, and Kent Smetters. 1999. Discounting Inside the Beltway. Journal of
Economic Perspectives 13 (4):213-228.
Becker, Randy, and Vernon Henderson. 1997. Effects of Air Quality Regulation on Decision of
Firms in Polluting Industries. Cambridge, MA: NBER.
Berger, James O. 1985. Statistical Decision Theory and Bayesian Analysis. 2nd ed. New York:
Bergman, Lars. 1991. General Equilibrium Effects of Environmental Policy: A CGE-Modeling
Approach. Environmental and Resource Economics 1 (1):43-61.
Boadway, R.W. 1975. Cost-benefit Rules in General Equilibrium. Review of Economic Studies
Böhringer, Christoph, and Thomas F. Rutherford. 1997. Carbon Taxes with Exemptions in an
Open Economy: A General Equilibrium Analysis of the German Tax Initiative. Journal
of Environmental Economics and Management 32 (2):189-203.
Bovenberg, A. Lans , and Lawrence H. Goulder. 2000. Neutralizing the Adverse Industry
Impacts of CO2 Abatement Policies: What Does It Cost? In Behavioral and
Distributional Effects of Environmental Policies: Evidence and Controversies
(forthcoming), edited by C. Carraro and G. Metcalf. Chicago: University of Chicago
Bovenberg, A. Lans, and Ruud A. de Mooij. 1994. Environmental Levies and Distortionary
Taxation. The American Economic Review 84 (4):1085-1089.
Bovenberg, A. Lans, and Lawrence H. Goulder. 1996. Optimal Environmental Taxation in the
Presence of Other Taxes: General Equilibrium Analyses. The American Economic Review
86 (4):985-1000.
Bovenberg, A. Lans, and Ben J. Heijdra. 1998. Environmental Tax Policy and Intergenerational
Distribution. Journal of Public Economics 67 (1):1-24.
Bovenberg, A. Lans, and Frederick van der Ploeg. 1996. Optimal Taxation, Public Goods and
Environmental Policy with Involuntary Unemployment. Journal of Public Economics 62
Boyd, James, Alan J. Krupnick, and Janice Mazurek. 1998. Intel’s XL Permit: A Framework for
Evaluation. Washington: Resources for the Future.
Resources for the Future Pizer and Kopp
Bruce, Neil, and Richard G. Harris. 1982. Cost-Benefit Criteria and the Compensation Principle
in Evaluating Small Projects. Journal of Political Economy 90 (4):755-76.
Bull, Nicholas, Kevin A. Hassett, and Gilbert E. Metcalf. 1994. Who Pays Broad-Based Energy
Taxes? Computing Lifetime and Regional Incidence. Energy Journal 15 (3):145-64.
Bureau of the Census, U.S. Department of Commerce. 1973-1997. Pollution Abatement Costs
and Expenditures. Washington, D.C.: U.S. Department of Commerce.
———. 2000. Pollution Abatement Costs and Expenditure Survey 1999 (draft). Washington,
D.C.: U.S. Department of Commerce.
Caves, Douglas W., and Laurits R. Christensen. 1980. Global Properties of Flexible Functional
Forms. American Economic Review 70 (3):422-432.
Chipman, John S., and James C. Moore. 1980. Compensating Variation, Consumer’s Surplus,
and Welfare. American Economic Review 70 (5):933-49.
Christainsen, Gregory B., and Robert H. Haveman. 1981. The Contribution of Environmental
Regulations to the Slowdown in Productivity Growth. Journal of Environmental
Economics and Management 8 (4):381-390.
Cline, William. 1992. The Economics of Global Warming. Washington: Institute of International
Codsi, George, K. R. Pearson, and Peter J. Wilcoxen. 1992. General-Purpose Software for
Intertemporal Economic Models. Computer Science in Economics and Management 5
Coggins, Jay S., and John R. Swinton. 1996. The Price of Pollution: A Dual Approach to
Valuing SO2 Allowances. Journal of Environmental Economics and Management 30
Congressional Budget Office. 1985. Environmental Regulation and Economic Efficiency.
Washington: Congress of the U.S., Congressional Budget Office.
Congressional Budget Office, Congress of the United States. 2000. Who Gains and Who Pays
under Carbon-allowance Trading? The Distributional Effects of Alternative Policy
Designs. Washington: CBO.
Conrad, Klaus, and Michael Schröder. 1991. The Control of CO2 Emissions and Its Economic
Impact: An AGE Model for a German State. Environmental and Resource Economics 1
Resources for the Future Pizer and Kopp
Council on Environmental Quality, Executive Office of the President. various years.
Environmental Quality. Washington: GPO.
Crandall, Robert W. 1993. Manufacturing on the Move. Washington: Brookings Institution.
Davies, Terry. 2001. Reforming Permitting. Washington: Resources for the Future.
DeCanio, Stephen J. 1993. Barriers within Firms to Energy-Efficient Investments. Energy Policy
Deck, Leland. 1997. Visibility at the Grand Canyon and the Navajo Generating Station. In
Economic Analyses at EPA, edited by R. D. Morgenstern. Washington: Resources for the
Diamond, Peter A., and James A. Mirrlees. 1971. Optimal Taxation and Public Production II:
Tax Rules. The American Economic Review 61 (3):261-278.
Diewert, W. Erwin. 1983. Cost-Benefit Analysis and Project Evaluation: A Comparison of
Alternative Approaches. Journal of Public Economics 22:265-302.
Dixit, Avinash K., and Robert S. Pindyck. 1994. Investment Under Uncertainty. Princeton:
Princeton University Press.
Dowlatabadi, Hadi, Raymond J. Kopp, and F. Ted Tschang. 1995. Distributional and
Environmental Consequences of Taxes on Energy: A Partial Equilibrium Model of U.S.
Household Energy Demand. In Advances in the Economics of Energy and Resources:
Energy Use and Sustainable Economic Growth, edited by J. Moroney. Greenwich: JAI
Färe, Rolf. 1988. Fundamentals of Production Theory. Berlin: Springer-Verlag.
Färe, Rolf, Shawna Grosskopf, C. A. Knox Lovell, and Suthathip Yaisawarng. 1993. Derivation
of Shadow Prices for Undesirable Outputs: A Distance Function Approach. The Review
of Economics and Statistics 75 (2):374-385.
Freeman, A. Myrick, III. 1985. Methods for Assessing the Benefits of Environmental Programs.
In Handbook of Natural Resource and Energy Economics, edited by A. V. Kneese and J.
L. Sweeney. New York: North-Holland.
Gaskins, Darius W. Jr., and John P. Weyant. 1993. Model Comparisons of the Costs of Reducing
CO2 Emissions. The American Economic Review: AEA Papers and Proceedings 83
Resources for the Future Pizer and Kopp
Gollup, Frank M., and Mark J. Roberts. 1983. Environmental Regulations and Productivity
Growth: The Case of Fossil-Fueled Electric Power Generation. Journal of Political
Economy 91 (4):654-674.
Goulder, Lawrence H. 1995a. Effects of Carbon Taxes in an Economy with Prior Tax
Distortions: An Intertemporal General Equilibrium Analysis. Journal of Environmental
Economics and Management 29 (3 (Part 1)):271-297.
———. 1995b. Environmental Taxation and the Double Dividend: A Reader’s Guide.
International Tax and Public Finance 2:157-183.
Goulder, Lawrence H., Ian W. H. Parry, and Dallas Burtraw. 1997. Revenue-Raising versus
Other Approaches to Environmental Protection: The Critical Significance of Preexisting
Tax Distortions. RAND Journal of Economics 28 (4):708-31.
Goulder, Lawrence H., Ian W. H. Parry, Roberton C. III Williams, and Dallas Burtraw. 1998.
The Cost-Effectiveness of Alternative Instruments for Environmental Protection in a
Second-Best Setting: Resources for the Future.
Gray, Wayne. 1996. Does State Environmental Regulation Affect Plant Location. Worcester,
MA: Clark University.
Gray, Wayne B. 1987. The Cost of Regulation: OSHA, EPA, and the Productivity Slowdown.
The American Economic Review 77 (5 (December)):998-1006.
Gray, Wayne B., and Ronald J. Shadbegian. 1994. Pollution Abatement Costs, Regulation and
Plant-level Productivity. Washington: Center for Economic Studies, U.S. Department of
Greenstone, Michael. 1997. The Marginal Effects of Environmental Regulations on the
Manufacturing Sector: Evidence from the 1970 and 1977 Clean Air Act Amendments.
Princeton, NJ: Economics Department, Princeton University.
Gruenspecht, H. K. 1982. Differentiated Regulation: The Case of Auto Emission Standards.
American Economic Review 72:328-331.
Gruenspecht, Howard, and Robert Stavins. 2002. A level field on pollution at power plants.
Boston Globe, January 26.
Hahn, Robert, and Wilbur Steger. 1990. An Analysis of Jobs at Risk and Job Losses From the
Proposed Clean Air Act Amendments. Pittsburgh, PA: CONSAD Research Corporation.
Resources for the Future Pizer and Kopp
Harberger, Arnold C. 1971. Three Basic Postualtes for Applied Welfare Economics: An
Interpretive Essay. Journal of Economic Literature 9:789-797.
Harrington, Winston, and Richard Morgenstern. 2000. On the Accuracy of Regulatory Cost
Estimates. Journal of Policy Analysis and Management 19 (2):297-322.
Hartman, R.S., D. Wheeler, and M. Singh. 1997. The Cost of Air Pollution Abatement. Applied
Economics 29:759-774.
Hausman, Jerry A. 1985. Taxes and Labor Supply. In Handbook of Public Economics, edited by
A. J. Auerbach and M. S. Feldstein. Amsterdam: North-Holland.
Hazilla, Michael, and Raymond J. Kopp. 1990. Social Cost of Environmental Quality
Regulations: A General Equilibrium Analysis. Journal of Political Economy 98 (4):853-
United States Senate. 2001. Environment and Public Works Committee. Hearing to receive
testimony on how S. 556 would affect the environment, the economy, and any
improvements or amendments that should be made to the legislation. 107th Congress.
November 1.
Henderson, Vernon J. 1996. Effects of Air Quality Regulation. American Economic Review 86
Hicks, J.R. 1942. Consumers’ Surplus and Index-Numbers. Review of Economic Studies 9
Hines, James R., Jr. 1999. Three Sides of Harberger Triangles. Journal of Economic
Perspectives 13 (2):167-188.
Howarth, Richard B. 1998. An Overlapping Generations Model of Climate Economy
Interactions. Scandinavian Journal of Economics 100 (3):575-591.
Jacoby, Henry, and Ian S. Wing. 1999. Adjustment Time, Capital Mobility and Policy Cost.
Energy Journal Special Issue.
Jaeger, William K. 2000. Optimal Environmental Taxes. Eugene, OR: University of Oregon.
Jaffe, Adam B., Steven R. Peterson, Paul R. Portney, and Robert N. Stavins. 1995.
Environmental Regulation and the Competitiveness of U.S. Manufacturing: What Does
the Evidence Tell Us? Journal of Economic 33 (1 (March)):132-163.
Resources for the Future Pizer and Kopp
Jorgenson, Dale W. 1986. Econometric Methods for Modeling Producer Behavior. In Handbook
of Econometrics, edited by Z. Griliches and M. D. Intriligator. Amsterdam: North-
Jorgenson, Dale W., Lawrence J. Lau, and Thomas M. Stoker. 1980. Welfare Comparison under
Exact Aggregation. American Economic Review 70 (2):268-72.
Jorgenson, Dale W., and Daniel T. Slesnick. 1983. Individual and Social Cost-of-Living Indexes.
In Price Level Measurement., edited by W. E. Diewert and C. Montmarquette. Ottawa:
Statistics Canada.
———. 1985. General Equilibrium Analysis of Economic Policy. In New Developments in
Applied General Equilibrium Analysis. New York and Melbourne: Cambridge University
Jorgenson, Dale W., Daniel T. Slesnick, and Peter J. Wilcoxen. 1992. Carbon Taxes and
Economic Welfare. In Brookings Papers on Economic Activity.
Jorgenson, Dale W., and Peter J. Wilcoxen. 1990. Environmental Regulation and U.S. Economic
Growth. Rand Journal of Economics 21 (2):314-340.
———. 1993. Reducing US carbon emissions: An econometric general equilibrium assessment.
Resource and Energy Economics 15 (1 (March)):7-25.
Joshi, S., L. Lave, J.-S. Shih, and F. McMichael. 1997. Impact of Environmental Regulations on
the U.S. Steel Industry. Pittsburgh: Carnegie Mellow University.
Just, Richard E., Darrell L. Hueth, and Andrew Schmitz. 1982. Applied Welfare Economics and
Public Policy. Englewood Cliffs: Prentice-Hall.
Kahn, Mathew E. 1997. Particulate Pollution Trends in the United States. Journal of Regional
Science and Urban Economics 27 (1):87-107.
Kaplow, Louis. 1996. The Optimal Supply of Public Goods and the Distortionary Cost of
Taxation. National Tax Journal 49 (4):513-533.
Kokoski, Mary F., and V. Kerry Smith. 1987. A General Equilibrium Analysis of Partial-
Equilibrium Welfare Measures: The Case of Climate Change. The American Economic
Review 77 (3):331-341.
Kolstad, Charles Durgin. 1982. Economic and Regulatory Efficiency in Air Pollution Control.
PhD, Economics, Stanford University, Stanford, CA.
Resources for the Future Pizer and Kopp
Krupnick, Alan J. 1986. Costs of Alternative Policies for the Control of Nitrogen Dioxide in
Baltimore. Journal of Environmental Economics and Management 13:189-197.
Lawson, Ann M. 1997. Benchmark Input-Output Accounts for the U.S. Economy 1992. Survey
of Current Business:36-82.
Levinson, Arik. 1996. Environmental Regulations and Manufacturers’ Location Choices:
Evidence from the Census of Manufactures. Journal of Public Economics 62 (1-2):5-29.
Lind, Robert. 1982. A Primer on the Major Issues Relating to the Discount Rate for Evaluating
National Energy Options. In Discounting for Time and Risk in Energy Policy, edited by
R. Lind. Washington: Resources for the Future.
Lipton, David, and et al. 1982. Multiple Shooting in Rational Expectations Models [The Solution
of Linear Difference Models under Rational Expectations]. Econometrica 50 (5):1329-33.
Lovins, Amory B. 1996. Negawatts. Energy Policy 24 (4):331-343.
Low, Patrick, and Alexander Yeats. 1992. Do ‘Dirty’ Industries Migrate? In International Trade
and the Environment. Washington: The World Bank.
MacCracken, Chris, J.A. Edmonds, S. Kim, and R. Sands. 1999. Economics of the Kyoto
Protocol. Energy Journal Special Issue.
Mas-Colell, Andreu, Michael D. Whinston, and Jerry R. Green. 1995. Microeconomic Theory.
New York: Oxford University Press.
McClelland, John D., and John K. Horowitz. 1999. The Costs of Water Pollution Regulation in
the Pulp and Paper Industry. Land Economics 75 (2 (May 1999)):220-232.
McConnell, Virginia D., and Robert M. Schwab. 1990. Impact of Environmental Regulation on
Industry Location Decisions: The Motor Vehicle Industry. Land Economics 66 (1):67-
McGraw-Hill. various years. Annual McGraw-Hill Survey of Pollution Control Expenditures.
New York: McGraw Hill Publishing.
McKibbin, Warwick J., M. Ross, R. Shackleton, and P. Wilcoxen. 1999. Emissions Trading,
Capital Flows and the Kyoto Protocol. Energy Journal Special Issue.
Metcalf, Gilbert E. 1999. A Distributional Analysis of Green Tax Reforms. National Tax Journal
52 (4):655-681.
Resources for the Future Pizer and Kopp
Morgenstern, Richard D., ed. 1997. Economic Analyses at EPA. Washington: Resources for the
Morgenstern, Richard D., William A. Pizer, and Jhih-Shyang Shih. 1998. Jobs versus the
Environment: An Industry-level Perspective. Washington, DC: Resources for the Future.
———. 1999. The Cost of Environmental Protection. Washington: Resources for the Future.
Morris, Glenn E., and Duncan M. Holthausen, Jr. 1994. The Economics of Household Solid
Waste Generation and Disposal. Journal of Environmental Economics and Management
26 (3):215-34.
Murray, Brian C., and David N. Wear. 1998. Federal timber restrictions and interregional
arbitrage in U.S. lumber. Land Economics 74 (1):76-91.
Nelson, R.A., T. Tietenberg, and M.R. Donihue. 1993. Differential Environmental Regulation:
Effects on Electric Utility Capital Turnover and Emissions. Review of Economics and
Statistics 77 (2):368-373.
Nestor, Deborah Vaughn, and Carl A. Jr. Pasurka. 1995. Alternative specifications for
environmental control costs in a general equilibrium framework. Economics Letters 48
Newell, Richard, and William A. Pizer. 2003. Discounting the Distant Future: How Much Do
Uncertain Rates Increase Valuations? Journal of Environmental Economics and
Management Forthcoming.
Nordhaus, William, and Joseph Boyer. 1999. Requiem for Kyoto: An Economic Analysis.
Energy Journal Special Issue.
Nordhaus, William D. 1991. The Cost of Slowing Climate Change: a Survey. The Energy
Journal 12 (1):37-65.
Oates, Wallace E., and Robert M. Schwab. 1996. The Theory of Regulatory Federalism: The
Case of Environmental Management. In The Economics of Environmental Regulation.,
edited by W. E. Oates. Cheltenham, U.K.: Elgar.
OECD. 1999. OECD Environmental Data Compendium. Paris: OECD.
Olson, Mancur, Jr. 1965. The Logic of Collective of Action: Public Goods and the Theory of
Groups. Cambridge: Harvard University Press.
Resources for the Future Pizer and Kopp
Palmer, Karen, Wallace E. Oates, and Paul Portney. 1995. Tightening Environmental Stantards:
The Benefit-Cost or the No-Cost Paradigm. Journal of Economic Perspectives 9 (4 (Fall
Pearce, David W. 1991. The Role of Carbon Taxes in Adjusting to Global Warming. Economic
Journal 101:938-948.
Peck, Stephen, and Thomas Teisberg. 1999. CO2 Emission Control Agreements: Incentives for
Regional Participation. Energy Journal Special Issue.
Perl, Lewis J., and Frederick C. Dunbar. 1982. Cost Effectiveness and Cost-Benefit Analysis of
Air Quality Regulations. The American Economic Review 72 (2):208-213.
Pindyck, Robert S. 1995. Sunk Costs and Sunk Benefits in Environmental Policy: I. Basic
Theory. Cambridge, MA: MIT.
Pittman, Russell W. 1981. Issue in Pollution Control: Interplant Cost Differences and Economies
of Scale. Land Economics 57 (1):1-17.
Pizer, William A. 1999. Optimal Choice of Climate Change Policy in the Presence of
Uncertainty. Resource and Energy Economics 21:25-287.
Porter, Michael E., and Claas van der Linde. 1995a. Green and Competitive: Ending the
Stalemate. Harvard Business Review 73 (5 (Sept-Oct 1995)):121-134.
———. 1995b. Toward a New Conception of the Environment-Competitiveness Relationship.
Journal of Economic Perspectives 9 (4):97-118.
Portney, Paul R. 1981. The Macroeconomic Impacts of Federal Environmental Regulation. In
Environmental Regulation and the U.S. Economy, edited by H. M. Peskin, P. R. Portney
and A. V. Kneese. Baltimore: Johns Hopkins University Press for Resources for the
Ramsey, F.P. 1928. A Mathematical Theory of Saving. The Economic Journal 38 (152):543-559.
Repetto, Robert, and Duncan Austin. 1997. The Costs of Climate Protection: A Guide for the
Perplexed. Washington: World Resources Institute.
Rose, A. 1983. Modeling the Macroeconomic Impact of Air Pollution Abatement. Journal of
Regulatory Science 23 (4):441-459.
Rosewicz, Barbara. 1990. Americans Are Willing to Sacrifice to Reduce Pollution, They Say.
Wall Street Journal, April 20, 1990, A1.
Resources for the Future Pizer and Kopp
Rutherford, Thomas F. 1999. Applied General Equilibrium Modeling with MPSGE as a GAMS
Subsystem: An Overview of the Modeling Framework and Syntax. Computational
Economics 14 (1-2):1-46.
Rutledge, {Gary L., and Christine R. Vogan}. 1994. Pollution Abatement and Control
Expenditures 1972-92. Survey of Current Business 74 (5):36-49.
Sabelhaus, John. 1996. Consumer Expenditure Survey: Family-Level Extracts, 1980:1 -1994:1.
Washington: Congressional Budget Office.
Samuelson, Paul A. 1974. Complementarity—An Essay on the 40th Anniversary of the Hicks-
Allen Revolution in Demand Theory. Journal of Economic Literature 12 (4):1255-1289.
Schäfer, Dieter, and Carsten Stahmer. 1989. Input-Output Model for the Analysis of
Environmental Protection Activities. Economic Systems Research 1 (2):203-28.
Schmalensee, Richard. 1994. The Costs of Environmental Protection. In Balancing Economic
Growth and Environmental Goals, edited by M. B. Kotowski. Washington: American
Council for Capital Formaation, Center for Policy Research.
Seskin, Eugene P., Robert J. Anderson, and Robert O. Reid. 1983. An Empirical Analysis of
Economic Strategies for Controlling Air Pollution. Journal of Environmental Economics
and Management 10 (2):112-124.
Stiglitz, J. E. 1979. Equilibrium in Product Markets with Imperfect Information. American
Economic Review 69 (2):339-45.
Stiglitz, Joseph E., and P. Dasgupta. 1971. Differential Taxation, Public Goods and Economic
Efficiency. Review of Economic Studies 38 (114):151-74.
Streitwieser, Mary L. 1995. Evaluation and Use of the Pollution Abatement Costs and
Expenditures Survey Micro Data. Washington: Center for Economic Studies, U.S.
Bureau of the Census.
Tietenberg, Tom. 1992. Environmental and Natural Resource Economics: HarperCollins.
———. 1998. Disclosure Strategies for Pollution Control. Environmental and Resource
Economics 11 (3-4):587-602.
U.S. Bureau of the Census. 1997. Census of Governments, Volume 4, Number 5, Compendium
of Government Finances. Washington: GPO.
Resources for the Future Pizer and Kopp
U.S. Environmental Protection Agency. 1984. Final Report: The Cost of Clean Air and Water:
Report to Congress 1984. Washington, D.C.: U.S. Environmental Protection Agency.
———. 1985a. Compilation of Air Pollutant Emissin Factors. Research Triangle Park, NC.
———. 1985b. Costs and Benefits of Reducing Lead in Gasoline: Final Regulatory Impact
Analysis. Washington: Office of Policy Analysis.
———. 1990a. Environmental Investments: The Cost of a Clean Environment. Washington,
D.C.: U.S. Environmental Protection Agency.
———. 1990b. OAQPS Control Cost Manual. Research Triangle Park, NC.
———. 1992. Regulatory Impact Analysis of Worker Protection Standard for Agricultural
Pesticides. Washington: U.S. EPA. Office of Pesticide Programs.
———. 1999. Final Report to Congress on Benefits and Costs of the Clean Air Act, 1970 to
1990. Washington, D.C.: U.S. Environmental Protection Agency.
———. 2001. Analysis of Multi-Emissions Proposals for the U.S. Electricity Sector, Requested
by Senators Smith, Voinovich, and Brownback. Washington: US EPA.
———. 2002. New Source Review: Report to the President. Washington, D.C.: U.S.
Environmental Protection Agency.
U.S. Office of Technology Assessment. 1994. Industry, Technology and the Environment:
Competitiveness, Challenges and Business Opportunities. Washington.
Vogan, Christine R. 1996. Pollution Abatement and Control Expenditures, 1972-94. Survey of
Current Business, September.
Wald, A. 1950. Statistical Decision Functions. New York: Wiley.
Webster, C. J., and F. Wu. 1999a. Regulation, Land-Use Mix, and Urban Performance. Part 1:
Theory. Environment and Planning A 31 (8):1433-42.
———. 1999b. Regulation, Land-Use Mix, and Urban Performance. Part 2: Simulation.
Environment and Planning A 31 (9):1529-45.
Weitzman, Martin L. 1998. Why the Far-Distant Future Should Be Discounted at Its Lowest
Possible Rate. Journal of Environmental Economics and Management 36 (3):201-208.
———. 2001. Gamma Discounting. American Economic Review 91 (1):260-271.
Resources for the Future Pizer and Kopp
Weyant, John P., and Jennifer Hill. 1999. The Costs of the Kyoto Protocol: A Multi-Model
Evaluation, Introduction and Overview. Energy Journal Special Issue.
Whalley, John. 1975. How Reliable is Partial Equilibrium? Review of Economics and Statistics
Whalley, John, and Randall Wigle. 1991. Cutting CO2 Emissions: The Effects of Alternative
Policy Approaches. Energy Journal 12 (1):109-24.
Wigley, T.M.L., R. Richels, and J.A. Edmonds. 1996. Economic and environmental choices in
the stabilization of atmospheric CO2 concentrations. Nature 379:240-243.



Por favor, inicia sesión con uno de estos métodos para publicar tu comentario:

Logo de

Estás comentando usando tu cuenta de Cerrar sesión / Cambiar )

Imagen de Twitter

Estás comentando usando tu cuenta de Twitter. Cerrar sesión / Cambiar )

Foto de Facebook

Estás comentando usando tu cuenta de Facebook. Cerrar sesión / Cambiar )

Google+ photo

Estás comentando usando tu cuenta de Google+. Cerrar sesión / Cambiar )

Conectando a %s

A %d blogueros les gusta esto: