Technical Paper Series
Congressional Budget Office
Washington, DC
The Economic Costs of Reducing Emissions
of Greenhouse Gases:
A Survey of Economic Models
Mark Lasky
May 2003
2003-3
Technical
papers in this series are preliminary and are circulated to stimulate
discussion
and critical comment. These papers are not subject to CBO’s formal
review
and editing processes. The analysis and conclusions expressed in them
are
those of the author and should not be interpreted as those of the
Con
gressional Budget Office. References in publications should be cleared with
the
authors. Papers in this series can be obtained by sending an email to
The Economic Costs of Reducing Emissions
of Greenhouse Gases:
A Survey of Economic Models
by
Mark Lasky
Congressional Budget Office
May 2003
I thank Robert Dennis, Douglas Hamilton, Robert Shackleton, and John Sturrock for
many helpful comments and suggestions. I am indebted to Robert Shackleton and
John Sturrock for much of the material in Chapter 1. Several of the modelers
provided helpful information and useful insights about their models. I thank John
Weyant for providing data from the EMF-16 model runs. All errors remain my own.
The analysis and conclusions expressed in this paper are those of the author and
should not be interpreted as those of the Congressional Budget Office.
Mailing Address: Congressional Budget Office, Macroeconomic Analysis Division,
2
nd
and D Streets, S.W., Washington, DC 20515. E-mail: MarkL@cbo.gov.
i
Table of Contents
INTRODUCTION 1
CHAPTER 1. THE KYOTO PROTOCOL 4
CHAPTER 2. ECONOMIC MODELS EXAMINED IN THIS SURVEY 7
CHAPTER 3. PERMIT PRICES UNDER THE KYOTO PROTOCOL 9
Key Determinants of Permit Prices 10
Required Reduction in Emissions 10
Price Sensitivity of Emissions 12
Baseline Price of Carbon-Energy 15
Model Estimates of Permit Prices 16
No International Trade of Permits 16
Unrestricted Trading of Permits Among Annex B Countries 17
Unrestricted Global Trading of Permits 18
Limitations on Permit Trading 19
Estimates of Permit Prices Under Various Scenarios: Model Synthesis 21
Scenarios Consistent with the Kyoto Protocol 23
Scenarios Requiring Amendments to the Kyoto Protocol 28
CHAPTER 4. THE EFFECTS OF THE KYOTO PROTOCOL
ON ENERGY PRICES 30
Model Estimates of Effects on Energy Prices 31
Coal Prices 31
Gasoline Prices 31
Natural Gas Prices 32
Electricity Prices 33
Estimates of Effects on Energy Prices: Model Synthesis 34
CHAPTER 5. MACROECONOMIC AND DISTRIBUTIVE EFFECTS
OF CUTTING GREENHOUSE GAS EMISSIONS 35
Direct Cost 36
Model Results 36
Synthesis of Results 37
ii
Impact on U.S. GDP and Consumption 38
Model Estimates of the Impact on U.S. GDP and Consumption 39
Impact on U.S. GDP and Consumption: Model Synthesis 42
Potential Impacts on Incomes 44
Winners: The Value of Permits Allocated 47
Losers: The Value of Permits Used 48
APPENDIX A. THE PRICE SENSITIVITY OF EMISSIONS
AND THE PRICE OF CARBON-ENERGY 50
Price Sensitivity of Carbon Emissions 50
The Baseline Price of Carbon-Energy 51
APPENDIX B. ORIGINS OF MODEL SYNTHESIS ESTIMATES 56
Emissions Baselines and Caps 56
Price Sensitivity of Carbon Emissions 57
Price Sensitivity of Emissions of Other Greenhouse Gases 62
Range of Uncertainty for Estimates of Permit Prices 63
Impact of the Clean Development Mechanism 64
Domestic Direct Cost 64
Impact on GDP 64
Impact on Consumption 68
Change in Global Emissions 68
Gasoline Price 69
Price of Natural Gas 70
Electricity Prices 70
Bibliography 72
BOXES
3-1. Emissions Leakage 24
5-1. The Economic Impact of Auctioning Permits 45
TABLES
2-1. Studies Analyzing the Impact of Emissions Reductions, by Model
and Institutions of Authors 75
3-1. Percentage Reduction in Emissions of Carbon Dioxide in 2010
iii
Required in Various Regions Under Alternative Permit-Trading
Scenarios 76
3-2. Impact on U.S. Carbon Emissions of a 1 Percent Increase in the
Price of Carbon-Energy, Assuming No International Trading of
Permits, Selected Years 77
3-3. Impact of a 1 Percent Increase in the Price of Carbon-Energy on
Carbon Emissions in 2010 Under Alternative Permit-Trading
Scenarios, by Region 78
3-4. Model Estimates of U.S. Permit Prices in 2010, Without
International Trade of Permits 79
3-5. Model Estimates of Annex B Permit Prices in 2010, with Permit-
Trading Among Annex B Countries 80
3-6. Model Estimates of Global Permit Prices in 2010, with Global
Permit Trading 81
3-7. Model Synthesis Estimates of Permit Prices and Reductions of
Emissions in 2010 Under Various Scenarios 82
4-1. Impact on Energy Prices of a $100 per mtc Permit Price 83
4-2. Model Synthesis Estimates of the Impacts on Energy Prices in
2010 of Various Scenarios to Restrict Greenhouse Gases 84
5-1. Impact of Emissions Reductions on U.S. GDP and Consumption
in 2010, with No International Permit Trading and Non-Auctioned
Permits 85
5-2. Model Estimates of the Cost of Emissions Permits in the United
States in 2010, With and Without International Permit Trading 86
5-3. Impact of Emissions Reductions on U.S. GDP and Consumption
in 2010, with International Permit Trading and Non-Auctioned
Permits 87
5-4. Model Synthesis Estimates of the Effect of Emissions Reductions
on U.S. GDP and Consumption in 2010 Under Various Scenarios,
with Non-Auctioned Permits 88
5-5. Model Estimates of the Value of Emissions Permits Allocated and
Used in the United States in 2010 89
5-6. Model Synthesis Estimates of the Value of Emissions Permits
Allocated and Used in the United States in 2010 90
B-1. Adjustments Made to Model Estimates of Price Sensitivity in
Constructing the Model Synthesis 91
FIGURE
5-1. Model Estimates of Permit Price and Percent Loss in GDP in 2010, 92
with No International Permit Trading
1.
Congressional Budget Office,
The Economics of Climate Change: A Primer
(April 2003).
1
INTRODUCTION
In
response to fears about the damages global warming may cause in the future, most
of
the world’s nations signed the Kyoto Protocol in December 1997, agreeing to cap
human-induced
emissions of carbon dioxide and other human-induced greenhouse
gases
in the United States and 37 other industrial nations beginning in 2008. At that
time, they left many details unresolved. Since then, the Bush Administration has
indicated
it would not submit the protocol to the Senate for ratification, and the other
parties have agreed to rules for implementation that will likely result in a much smaller
reduction
in emissions than originally envisioned by the protocol.
1
Nonetheless, the
considerable
volume of studies of the Kyoto agreement as originally intended may
have
some lessons for any future attempts to limit greenhouse gases. This paper
reviews
those studies in order to assess the economic cost of limiting greenhouse
emissions
through a system of tradable emissions permits and investigates the impact
of alternative rules for trading.
Economi
sts have used a variety of models to estimate the costs of complying with
emissions caps.
These models assume that a permit system would raise the cost of
goods
or services that cause greenhouse gas emissions when they are either used or
produced.
Carbon dioxide from the combustion of coal, oil and natural gas accounts
for
the bulk of such emissions, so prices of energy and energy-intensive goods and
services
would rise the most. The higher the price of such goods rose, the fewer of
them people would use, until emissions were reduced to the level of the cap.
This
paper surveys and synthesizes the predictions of several such economic models
of
the cost of meeting the Kyoto caps. The models produce a wide variety of cost
estimates,
depending on many factors. The most important of those factors are
differen
ces in the amount of permit trading assumed and different model assumptions
about
the sensitivity of energy usage to energy prices and the response of the
econom
y to higher inflation. In some cases, differences in the assumed path of
baseline
emissions, the response of labor supply to the real wage, and the impact of
internat
ional capital flows are also important. All models are simplifications of reality,
and
thus to some extent make unrealistic assumptions— often different ones in
different
models. However, by comparing model assumptions and results, it is
possible
to assess the effect of those assumptions and to adjust for them. The
synthesis
presented in this paper provides an integrated view that minimizes the role
of unrealistic assumptions, and thus gives a clearer view of the likely economic impact
of alternative rules for permit trading.
2.
This
finding assumes that there are no additional benefits to reducing greenhouse emissions beyond
changes in global climate.
2
Depending
on how much permit trading is allowed, a wide range of costs is possible.
Based
on forecasts released in 2000, gasoline prices in the United States could be 12
to
38 cents per gallon (1997 dollars) higher in 2010 than they would without emission
limits, depending on trading rules.
(Throughout this paper, prices are expressed in
1997
dollars. To convert 1997 dollars into 2002 dollars, multiply by 1.085.) The
price
of natural gas to households could be 13 percent to 42 percent higher, while the
price
of electricity to households could be 13 percent to 36 percent higher. Real GDP
in the United States under the
protocol could be 0.5 percent to 1.2 percent lower in
2010
than otherwise, while real consumption could be 0.4 percent to 1.0 percent
lower.
In every case, costs are smallest when no restrictions are imposed on
international
permit trading. (Taking account of uncertainty widens the range of
estimates further.)
Although
different rules for permit trading would have a large impact on the cost of
emissions
restrictions, they would have little impact on the environmental benefits,
th
at is, on the reduction in global emissions of greenhouse gases. In most cases,
restrictions
like those specified in the Kyoto Protocol would reduce global emissions
of carbon dioxide in 2010 by about 6 percent from what it would otherwise be, or
from
40 percent above 1990 levels to 33 percent above 1990 levels. Only if countries
could
not sell their excess permits (permits from a cap in excess of baseline emissions)
would the reduction
in emissions be larger. Unfortunately, costs are also largest in
this
case. Thus, one of the key findings of this paper is that the United States would
be better off the fewer restrictions there were on international permit trading.
2
If
no trading were allowed at all, costs would rise well above those in the worst case
scenario
for the protocol. On the other hand, if developing and newly developed
nations
accepted caps on emissions, essentially creating a global market for permits,
costs
could be reduced to half of what they would be under the best case scenario for
the Kyoto Protocol.
The
distributional impacts of restrictions on greenhouse emissions could be large.
Such
limits, if implemented through permits, would essentially transfer income from
energy
users to permit recipients. (If the permits are auctioned by the government
and
the revenues recycled as tax cuts or transfer payments, then the recipients are
those people who receive the tax cuts
or transfers.) To the extent that the users are
the same as the recipients, there is no net redistribution. However, the amount at
stake
is potentially large: depending on the amount of permit trading allowed, the
value
of permits used in the United States would be between 0.9 percent and 2.0
percent of GDP in 2010.
3
In the interest of simplicity, the models, and thus the model synthesis, leave out
certain aspects of the possible costs of emissions caps. For example, most models
assume that energy usage will decline in the future by the same percentage amount for
a given percentage increase in energy prices as it has in the past. However,
opportunities to reduce energy usage today may be either more or less abundant than
in the past. In addition, the models assume that emissions can be monitored and
permits transacted at no cost, and that countries do not cheat. Also, the models do
not estimate how much costs could be reduced by substituting new forest growth for
the most costly reductions in emissions.
The estimates in this paper are based on projections of emissions and energy prices
released by the U.S. Department of Energy’s Energy Information Administration
(EIA) in 2000. In more recent projections, EIA has raised its forecasts of both energy
prices in the United States and emissions in most signatories of the Kyoto Protocol.
Those upward revisions would increase the adverse impact of the protocol on energy
prices, GDP, and consumption, but also would increase the impact on emissions.
However, the comparative impacts of restrictions on permit trading or of a global
market for permits would be similar.
This paper looks only at the costs of emissions restrictions like those in the Kyoto
Protocol and their effect on global emissions. A full evaluation of the treaty would
also assess the science that underlies the claim that a significant risk of global warming
exists, and the benefits of lower emissions. Such an assessment, however, is beyond
the scope of this paper.
3.
This
information, along with further details on negotiations subsequent to the Kyoto Protocol, can be
found
in Congressional Budget Office,
The Economics of Climate Change: A Primer
(April 2003), pp.
47-48.
4
CHAPTER 1
THE KYOTO PROTOCOL
In
December 1997, most of the world’s nations signed a draft treaty—the Kyoto
Protocol—that would limit human-induced emissions of greenhouse gases in the
United
States and 37 other industrial nations (listed in Annex B). Other signatories
do
not face any limitations under the protocol. Instead, they may undertake joint
abatement
projects with other industrial countries, thereby providing the latter with
credit
against their national limits in exchange for a fee or other compensation. The
p
rotocol enters into force 90 days after 55 signatories have formally ratified it,
provided
those signatories include nations that emitted at least 55 percent of Annex
B’s
carbon dioxide in 1990. Thus, the protocol cannot go into effect unless either the
United
States or Russia ratifies the protocol, since those two countries accounted for
about half of Annex B’s carbon dioxide emissions in 1990.
Recent
developments suggest that a much more limited version of this agreement may
soon
take effect. In early 2001, the Bush Administration indicated that it would not
continue to negotiate the terms of the protocol or submit
the protocol to the Senate
for
ratification. In November 2001, the other parties agreed to allow some credit for
existing
forests, effectively easing the emissions limits. The European Union and
Japan
ratified the protocol in 2002. As of March 2003, ratification by Russia would
bring the treaty into effect within the countries having ratified it.
3
The
Kyoto Protocol covers the six main types of greenhouse gases stemming from
human
activity: carbon dioxide, methane, nitrous oxide, and three kinds of synthetic
gases—hydrofluorocarbons,
perfluorocarbons, and sulphur hexafluoride. The treaty
uses the “global warming potential” of each gas—a measure of its contribution to
global warming—to translate the raw weight of its emissions into an equivalent
weight
of carbon dioxide. By convention, scientists further translate that equivalent
weight
of carbon dioxide into the weight of its carbon alone. For that reason, this
paper reports emissions in metric tons (tonnes) of carbon (mtc).
The
treaty restrictions apply to net emissions, so a country can receive credit for the
carbon
dioxide removed from the air by the net growth of forests that meet certain
qualifications.
To qualify as a Kyoto forest, a woodland must result from human
inter
vention since 1990. However, the signatories have held widely divergent
opinions
as to what should count under this provision. A key point of contention has
been
how much credit should be given for “existing effort,” or forest growth that
4.
A
cap of 92 percent of benchmark year emissions was individually and collectively agreed to by the
European
Union and its members: Austria, Belgium, Denmark, Finland, France, Germany, Greece,
Ireland,
Italy, Luxembourg, Netherlands, Portugal, Spain, Sweden, and United Kingdom. Other countries
agreeing
to a 92 percent cap included Bulgaria, Czech Republic, Estonia, Latvia, Liechtenstein,
Lithuania,
Monaco, Romania, Slovakia, Slovenia, and Switzerland. The following list contains the other
Annex B countries (and the limits they agreed to as percentages of benchmark year levels): Australia
(108),
Canada (94), Croatia (95), Hungary (94), Iceland (110), Japan (94), New Zealand (100), Norway
(101), Poland (94), Russia (100) and Ukraine (100).
5
would
have occurred whether the protocol were ratified or not. In November 2001,
the
remaining parties to the protocol agreed to grant each participant a specified
credit
for existing effort. Granting credit for existing effort reduces both the costs and
the benefits of the protocol.
For
each of the countries listed in Annex B, the protocol specifies a cap on average
net emissions during a commitment period. Caps have been specified only for the first
commitment
period—the five years from 2008 through 2012. Each country’s cap for
that
period is an agreed percentage of its total benchmark year emissions. For most
countries,
the benchmark year is 1990. Some eastern European nations are given the
option
of using a different year, and 1995 is the benchmark year for some gases. The
United
States agreed to a cap of 93 percent of 1990 emissions. As a whole, the
Annex
B cap is roughly 95 percent of 1990 emissions.
4
The protocol does not specify
caps beyond 2012.
Under
the treaty, each Annex B nation receives tradable allowances, or permits, that
certify
the right to release greenhouse emissions up to its respective allowable total
for
the commitment period. The permits may be distributed any way the country
chooses,
and they may be bought and sold, at home or abroad. In addition, a country
can
receive tradable credits for verified greenhouse abatement that it realizes in a
country
outside of Annex B through the Clean Development Mechanism (CDM).
Any
permits not used or sold may be banked for use in a future commitment period.
One of the most important details left unresolved by the protocol is how much
international
permit trading it allows. The treaty merely states that “trading shall be
supplemental
to domestic actions for the purpose of meeting quantified emission
limitation
and reduction.” The parties have never decided how much
“supplementarity”
should be allowed. A restrictive set of rules on trading of permits
would have a major impact on the protocol’s cost.
This
paper assesses the potential economic costs to the United States of reducing
emis
sions under several scenarios consistent with the Kyoto Protocol as originally
negotiated. Lowest costs prevail under “ideal implementation,” in which:
no cartel manipulates permit prices for its own advantage;
the provisions for CDM work as intended;
no restrictions are placed on permit trading; and
6
caps apply to all greenhouse gases.
Costs rise as, one by one, these assumptions are changed. First, a cartel of former
Soviet bloc nations acts to restrict permit supply. Next, the provisions of the CDM
fail to work as intended. Then restrictions on each country’s permit sales are added.
As an alternative, restrictions are placed on each country’s permit purchases. Finally,
only carbon dioxide is capped; other greenhouse gases remain unregulated. This
paper also assesses costs in the illustrative cases of no international permit trading and
of an amended Kyoto Protocol in which all countries accept emissions caps.
5.
Weyant, John
P., ed.,
The Costs of the Kyoto Protocol: A Multi-Model Evaluation, Special Issue of the
Energy Journal
(Cleveland, OH: Energy Economics Educational Foundation, Inc., 1999).
6.
Council
of Economic Advisors,
The Kyoto Protocol and the President’s Policies to Address Climate
Change
(July 1998); Energy Information Administration,
Impacts of the Kyoto Protocol on U.S. Energy
Markets and Economic Activity
(October 1998); Standard & Poor’s DRI,
The Impact of Meeting the
Kyoto Protocol on Energy Markets and the Economy
(Lexington, MA: Standard & Poor’s DRI, 1998);
WEFA,
Inc.,
Global Warming: The High Cost of the Kyoto Protocol, National and State Impacts
(Eddystone, PA: WEFA, Inc., 1998).
7
CHAPTER 2
ECONOMIC MODELS EXAMINED IN THIS SURVEY
The
economic models surveyed in this paper treat the economy as an interactive
system.
Impacts begin in the energy sector, where restrictions on emissions boost the
price
of energy, reducing the amount of energy used in the rest of the economy. This
in
turn affects businesses’ investment and hiring decisions. To the extent that gross
domestic
product (GDP) falls in response, demand for energy declines further.
Economic
models are designed to capture these and other interactions among the
various
sectors of the economy. In that sense, they differ from purely technology-
based models, which focus overwhelmingly on the energy sector.
While
most of the economic models surveyed base projected responses of energy
users
to changes in prices on past responses, several also incorporate choices among
specific
technologies, especially in electricity generation. In such models, electric
utilities
are assumed to choose the mix of technologies that meets empirically-based
demand for electricity at the lowest cost.
Eleven
of the models surveyed in this paper were used during Round 16 of Stanford
University's
Energy Modeling Forum (EMF-16), which examined the potential
economic
impact of the Kyoto Protocol (see Table 2-1). Those studies are published
in
a special issue of the Energy Journal.
5
(That issue contains descriptions of the
models.)
This survey also includes the results of four other models frequently cited
in discussions of the costs of reducing greenhouse gas emissions.
6
I
generally follow the Modeling Forum's convention of referring to the studies by the
name
of the model that each study uses. In the case of studies produced by private
forecasters–DRI, Oxford and WEFA–the model is named for the institution producing
the study. The Energy Information Administration (EIA) uses its own model
(NEMS)
for energy sector results but uses the DRI model for economy-wide results.
I denote this combination as EIA. Studies prepared by the Clinton Administration and
by
Battelle Pacific Northwest National Laboratory both use the SGM model. These
studies are referred to as SGM-Administration and SGM-PNNL, respectively.
8
A key difference among the models is that some use a general equilibrium structure,
while others use a macroeconometric structure. General equilibrium models assume
that the labor market is always in equilibrium, meaning that any worker willing to
work at the going wage can always find a job. In those models, monetary and fiscal
policies can affect output only by changing worker productivity or labor supply.
Macroeconometric models, by contrast, assume that some unemployment is
involuntary. These models leave scope for policy to affect output through changes
in the utilization of labor and capital. The AIM, CETA, EPPA, GTEM, JWS,
MERGE, MS-MRT, RICE, SGM and WorldScan models assume a general
equilibrium structure, while DRI, EIA, Oxford and WEFA use macroeconometric
models. The G-Cubed model uses a hybrid of these two structures.
Not surprisingly, different models have different strengths and weaknesses. Models
that make seemingly unrealistic assumptions in one sector are sometimes better than
other models in representing reality in another sector. In addition, assumptions that
are unrealistic to use in analyzing the impact of emissions restrictions on the
macroeconomy over a ten-year horizon may produce realistic results when applied to
another policy or over a longer time horizon.
The Kyoto Protocol specifies that average annual emissions during the 2008-2012
commitment period should equal specified limits, unless emissions credits are
“banked” for use in later periods. Several models simply assume that the caps are for
2010, and all models assume there is no banking of permits. To ensure comparability
among studies, I focus on results for 2010 and assume no banking of permits.
Banking of permits would increase the impact of the protocol on energy prices during
2008-2012, but would reduce it in later years.
The study using the JWS model does not use the emissions cap specified in the Kyoto
Protocol, but rather a return to 1990 emissions levels in 2010. While the properties
of this model are incorporated into the model synthesis estimates, results from the
study that depend on meeting its looser emissions caps are not presented.
7.
Even
if a business received permits free of charge, it would still set prices as if it had paid for the
permits.
By using the permits, the business foregoes the revenue it could have received from selling the
permits, and it will charge its customers a higher price to cover this foregone opportunity.
8.
To translate permit prices into more familiar terms, 426 gallons of gasoline contain a metric ton of
carbon,
so each $100 rise in permit prices adds $100 per 426 gallons, or 23.8 cents per gallon, to the price
of
gasoline. The actual increase in gasoline prices will be somewhat smaller, because crude oil prices
and refining margins will fall. For further details, see the section on energy prices.
9
CHAPTER 3
PERMIT PRICES UNDER THE KYOTO PROTOCOL
As
originally intended by U.S. negotiators, the Kyoto Protocol would limit emissions
of
greenhouse gases by Annex B countries by creating a system of tradeable permits
that
firms would be required to hold in order to produce energy from coal, natural gas
or
petroleum, or to engage in any other activity that produces greenhouse gases.
Whether
firms purchased permits or received them without charge, the permit price
would
eventually be passed along to users of energy from fossil fuels and to final
purchasers of any other goods or services
produced by emitting greenhouse gases.
7
In
the end, those higher prices would reduce demand for goods and services that
produce greenhouse gases.
Several
studies have used economic models to analyze the impact of a such a system
of
tradeable emissions permits on the economy (see Box 2-1). These models
incorporate
interactions between the energy sector, which is responsible for most
greenhouse
gas emissions, and the rest of the economy. Most of the models assume
that
energy users respond to changes in the price of energy in the same way that they
have in the past.
The studies encompass a wide range of outcomes. Differences in outcomes stem from
differences
in assumptions about the implementation of the permit system, the
operation of the economy,
and the response of consumers and businesses to higher
energy prices. In
general, the fewer the restrictions on permit trading and the more
sensitive
energy users are to energy prices under the protocol, the lower the permit
price and the smaller the impact on the economy.
Synthesizing
results from several studies, permit prices under the Kyoto Protocol
could range from $56 per metric ton of carbon to $178 per metric ton in 2010 (in
1997 dollars),
depending on how the protocol was to be implemented.
8
(Appendix
B
explains how the synthesis results were constructed.) Uncertainty about the size
of consumer and business
responses to higher energy prices raises the range to $41
to
$226 per metric ton of carbon (mtc). Those estimates assume that permits could
be
traded among industrial countries in Annex B and that the protocol was effective
10
in reducing global emissions. If such trade were prohibited, permit prices could be
substantially higher, while permit prices could be lower if the protocol were amended
to include countries outside Annex B. If Annex B countries received full credit for
greenhouse gases absorbed by forest growth and other land use changes, permit prices
would be zero in 2010, but the protocol would have no effect on emissions.
As discussed in the introduction, synthesis estimates of permit prices are based on
projections of emissions and energy prices released by the Energy Information
Administration (EIA) in 2000. More recent projections of emissions and energy
prices are higher, meaning that updated synthesis estimates of permit prices would be
higher in every scenario.
KEY DETERMINANTS OF PERMIT PRICES
Permit prices will depend primarily on three factors: the required reduction in
emissions; the response of households and businesses to an increase in prices of
greenhouse gas-related goods and services; and the level of such prices in the baseline.
A fourth factor, the effect of the Kyoto agreement on gross domestic product (GDP),
is of little importance: even studies that find large GDP losses estimate that lower
GDP would produce less than one-seventh of the required reduction in emissions.
Required Reduction in Emissions
The required reduction in emissions can have a significant impact on permit prices.
The Kyoto agreement would reduce emissions because the permits would increase the
cost of using carbon-energy and goods and services produced using other greenhouse
gases. (Following Nordhaus and Boyer, this paper uses the term “carbon-energy” as
short-hand for energy from coal, natural gas and petroleum.) Not surprisingly, the
larger the required reduction in emissions, the higher the permit price would have to
be.
When permits can be traded across borders, the international market determines the
permit price in each country unless limits are placed on purchases of permits. So,
with international trade of permits, the percentage reduction in emissions in the entire
trading bloc, and not just in the United States, determines the permit price. This is
true for any commodity that is traded internationally. For example, the price of crude
oil in the United States plunged in 1998, even though U.S. consumption rose and
production fell, because the price of oil is determined in a global market, and the
Asian crisis reduced global demand for oil.
9.
Energy
Information Administration,
International Energy Outlook 2000
,
DOE/EIA-0484 (March 2000).
According to the 2002 projection, required reductions in
emissions would be larger for both the United
States and for the rest of Annex B.
10.
Energy Information Administration,
International Energy Outlook 2000
.
11
Although
the studies present a range of estimates, they agree that the percentage
reduction
in emissions for the United States without trading would be much larger
than
the percentage reduction in Annex B or global emissions with trading. If permits
coul
d not be traded and the cap applied only to carbon dioxide, the United States
would
have to reduce domestic emissions by 30 percent to comply with the emissions
caps
specified in the protocol, according to estimates released in 2000 by EIA (but
not
reflected in the model results).
9
However, carbon emissions would have to be cut
by
only 12 percent in the Annex B region if permits were freely traded within that
region.
If permits were traded globally, world emissions would have to be cut by only
6 percent. The models make qualitatively similar assumptions (see Table 3-1).
The
Kyoto caps are easier to reach with trade for two reasons. First, the portion of
the
U.S. cap attributable to carbon emissions in 1990 is set 7 percent below those
levels,
while the portion of the cap for the other Annex B countries attributable to
carbon
emissions in 1990 averages only about 3 percent below those levels. Second,
and
more important, carbon emissions are expected to grow more slowly between
1990
and 2010 in other Annex B countries than in the United States. That projection
stems
partly from the expectation that economic activity in the countries of the former
Sovie
t Union will still be below 1990 levels in 2010. In addition, because the
communist
governments in the former Soviet bloc subsidized fossil fuels heavily and
encouraged
excessive use, those countries are expected to cut back their use of fossil
fuels
as they continue their transition toward market economies. Already, carbon
emi
ssions per dollar of GDP in Eastern Europe have fallen at almost a 5 percent
annual
rate from 1990 to 1999. Over the next decade, EIA expects rapid declines in
energy intensity to continue in eastern Europe, and to spread to the former Soviet
Union.
10
A
few provisions of the Kyoto Protocol can cause Annex B emissions of carbon
dioxide
to diverge from the portion of the caps attributable to 1990 emissions of
carbon
dioxide. First, Annex B countries can get credit for certain reductions in non-
Annex B emissions under the Clean Development Mechanism (CDM). Second, since
the
Kyoto caps include all greenhouse gases, reductions in emissions of other
greenhouse gases (like methane and nitrous oxide) below the caps can be counted
toward
a country’s obligation to reduce carbon emissions. Finally, countries can
receive
credits for forest growth. To the extent that countries get credits for forest
growth
that would have occurred even without the protocol, such credits reduce the
environmental benefits of the protocol as well as its economic costs.
12
Because of questions about how those provisions for the CDM and offsets for other
gases and forest growth would be implemented and enforced, many modelers do not
include impacts from them. The Clinton Administration assumed larger benefits from
these provisions than any other researchers, which is one reason why the
Administration’s estimate of the cost of the Kyoto Protocol is more optimistic than
those of most other modelers.
Price Sensitivity of Emissions
How high must permit prices rise to produce the required decline in carbon emissions?
The answer will depend on how much households and businesses respond to an
increase in the price of carbon-energy. The larger their response, the lower the permit
price would have to be to achieve the Kyoto caps.
The sensitivity of households and businesses to the price of carbon-energy varies
significantly among the models, and it is the most important source of differences
among the models' estimates of the permit price. For very small changes in emissions,
the price sensitivity of a model can be measured approximately as the percentage
change in carbon emissions per GDP produced by a one percent increase in the price
of carbon-energy. For the United States, estimates of price sensitivity in 2010 differ
by a factor of four among the models when there is no international permit trading
(see Table 3-2).
Four factors can influence the price sensitivity of carbon emissions at a given point in
time in any country: the long-run ability of businesses to substitute low-carbon fuels
for high-carbon fuels; the long-run sensitivity of household and business energy usage
to higher energy prices; the speed at which these long-run responses occur; and
differences in these reactions between countries. The models make a variety of
different assumptions about each of these factors. (In addition, the price sensitivity
for emissions of other greenhouse gases may differ from that for carbon dioxide. Only
two studies, by the Clinton Administration and the EPPA modelers, look at this
issue.)
Substitution Among Fuels
. One way businesses and households can respond to higher
costs of emitting carbon is by substituting low-carbon or no-carbon fuels for high-
carbon fuels. Electric generators would make most of these substitutions, replacing
coal with other fuels. In fact, fuel substitution by electric generators alone can
account for between 24 percent (in DRI’s 1997 study) and 48 percent (in EIA’s
study) of the total reduction in emissions.
The amount of fuel substitution can differ among models for many reasons. First,
models may differ on the permit price at which a technology that uses less carbon
11.
The
EPPA, G-Cubed and JWS models implicitly assume unrealistic increases in hydroelectric and nuclear
power. The DRI, GTEM and WEFA models assume that
nuclear plants are retired at baseline rates no
matter what the permit price.
AIM, OEF, MERGE, CETA, RICE and WorldScan do not have separate
electricity sectors, so it is impossible to know if their implicit assumptions are realistic or not.
12.
Unrealistic
substitution of capital and labor for crude oil and natural gas only affects permit prices if
emissions are assumed proportional to usage of fossil fuels, as in the AIM and SGM models. Such
unrealistic substitutions do not affect
permit prices in models that assume emissions are proportional to
refined petroleum products and utility natural gas, such as EPPA, G-Cubed, GTEM and JWS.
13.
The
G-Cubed and SGM models assume that emissions are proportional to gross sales of the coal mining
and oil and gas extraction industries, rather than to the net sales of these industries. For example,
emissions
from coal are assumed proportional to gross sales of coal and coal mining services, including
sales
of coal and coal mining services from one company to another. These models also assume that such
intra-industry sales fall by a larger percentage than net sales
of coal to industries that will actually burn
the coal. Thus, estimated emissions from coal fall by a larger percentage than actual coal usage.
14.
The overall capital
intensity of the economy would nonetheless fall. First, businesses would substitute
labor for capital, which is produced using energy, as well as capital and labor for energy. Second, the
amount of capital used in the energy sector would shrink with the size of that sector.
13
becomes
economical and on the availability of that technology. Unfortunately, many
models
implicitly assume unrealistically large increases in nuclear and hydroelectric
power,
given the high cost and political difficulties of new nuclear plants and the lack
of
potential new dam sites. On the other hand, some models assume that utilities
retire
existing nuclear plants at the rate currently anticipated no matter how high
permit
prices go.
11
Assumptions about nuclear and hydro power can cause permit
prices to vary by nearly 30 percent.
Second,
some models assume that capital and labor can somehow be substituted for
fossil
fuels in the production of fossil-fuel energy–an assumption that seems
implausib
le. For example, some models allow petroleum refiners and natural gas
utilities
to substitute labor and capital for petroleum and natural gas. In other words,
those
models assume that refiners would be able to reduce the amount of crude oil
required
to produce a gallon of gasoline by using more labor and capital.
12
This
assumption
can play a significant role in the analysis; in one case, it reduces estimated
permit prices by more than 15 percent.
Third,
emissions from coal, petroleum and natural gas in some models fall by a much
larger
percentage than actual combustion of these fuels.
13
This assumption can cause
permit prices to be understated by 15 percent.
Long-Run
Sensitivity of Overall Energy Demand
.
Households and businesses would
also
reduce overall energy demand as energy prices rose. Businesses would substitute
capital
and labor for energy (factor substitution), while households would substitute
purchases
of other goods for energy.
14
Many of these responses would only occur
slowly
over time. For example, many households and businesses would upgrade to
15.
These
models assume a unit elasticity of demand for energy—that is, the percentage change in energy
demand
is roughly the same (but in the opposite direction) as the percentage change in real energy prices.
In
the G-Cubed model, this assumption affects only final demands. In the RICE and WorldScan models,
a
Cobb-Douglas production function (with a unit elasticity) is used to derive energy demand throughout
the economy.
14
more
fuel-efficient appliances and equipment only when the existing items would
normally be replaced.
Most
of the models base their expectations of how energy users will react to changes
in
energy prices from studies about how businesses and consumers have reacted in the
past.
Because the large price changes of the 1970s and early 1980s produced
disproporti
onately small changes in usage, most models assume that energy users
would respond relatively modestly to changes in energy prices.
Three
models–G-Cubed, RICE, and WorldScan–assume, for analytical convenience,
much
larger responses to changes in energy prices, without, however, claiming an
empirical
basis for this assumption.
15
The use of demand responses out of line with
the
empirical literature has a larger effect than any other unrealistic assumption,
reducing
the permit price by 30 percent in the model requiring the smallest
adjustment, G-Cubed.
Expectations
and Speed of Adjustment
.
The longer it takes for businesses and
households
to reduce their energy usage and make fuel substitutions in response to
higher
carbon-energy prices, the higher energy prices would have to rise in 2010 to
achieve
a given percentage reduction in emissions. The speed of adjustment can be
measured
by comparing the price sensitivity in 2010 with that in 2020; the smaller the
ratio, the more slowly households and businesses cut emissions (see Table 3-2).
Most
models assume businesses would respond gradually, reflecting the fact that an
industry
can achieve many reductions in energy usage per dollar of output only as it
turns
over its capital stock. In general, differences in techniques for modeling this
gradual
response do not appear to have significant effects on the estimates of permit
prices.
Some
models (CETA, JWS and RICE), however, assume immediate adjustment.
These
models assume that businesses can, for example, instantaneously transform coal
mi
nes into energy-saving industrial equipment, or immediately make existing
equipment
more energy efficient. In such models, the price sensitivity is roughly the
same
in 2010 as it is in 2020. This observation also applies to Oxford and
WorldScan,
which assume that all adjustments to changes in energy prices are
complete by 2010.
16.
The
only model for which this is not true, Oxford’s, does not include eastern Europe or the Ukraine, and
thus gives a smaller weight to the countries of Annex B with low baseline prices of carbon-energy.
15
The
date when people begin to anticipate higher energy prices can also affect the
permit
price. Other things equal, the longer it takes for a policy change to be
announced
or ratified, the more slowly people will respond to it. For example, the
EIA
study that assumes people begin to respond in 2000 finds lower permit prices
than the EIA that assumes people begin to respond in 2005.
Speeds
of adjustment are a big factor accounting for differences among the models.
The long-run price sensitivity in the EIA study appears to be only about 10 percent
smaller than that of studies using the SGM model, such as that of the Clinton
Administration.
A more rapid speed of adjustment in the SGM model thus explains
most
of the difference in permit prices between the EIA and Administration studies.
Price
Sensitivity in Other Countries
.
When permits can be traded across borders, the
U.S.
permit price is determined by the price sensitivity of carbon emissions in all the
countries trading permits, not that just in the United States.
Most models assume that other countries would be less sensitive to energy prices than
the
United States (see Table 3-3). Within the Annex B region, other countries
generate
a larger percentage of their electricity from nuclear and hydroelectric sources
than
from fossil fuels. As a result, those countries have fewer opportunities for
substituting
fuels than the United States. And outside of the Annex B region, other
countries
have not fully demonstrated their ability to use a price mechanism to reduce
their
energy usage. As a result, emissions in those countries may not respond as much
to higher prices for carbon-energy as in Annex B countries.
Baseline Price of Carbon-Energy
The
higher the baseline price of carbon-energy, the higher the permit price must be
to
achieve a given percentage increase in energy prices, for a given price sensitivity
(see
Table 3-4). The price of carbon-energy is an aggregate of the price per ton of
carbon
for each type of carbon-energy, such as gasoline, electricity from fossil fuels,
and residential natural gas.
The
models generally project that the baseline price of carbon-energy would be
somewhat
lower for the overall Annex B region than for the United States (see Tables
3-4
and 3-5).
16
That result stems from low baseline prices in the former Soviet bloc.
Although
baseline prices are higher in Japan and western Europe than in the United
States,
these countries would reduce emissions much less than the former Soviet bloc,
and thus receive a smaller weight.
16
The baseline price of carbon-energy is quite a bit lower in non-Annex B countries than
in the Annex B region. The main reason is these countries’ greater dependence on
coal-based energy, which is much cheaper than energy from petroleum and natural
gas.
MODEL ESTIMATES OF PERMIT PRICES
Although the studies produce a wide range of estimates for the price of emissions
permits, the studies unanimously agree on one point: trading permits across borders
will lower their price in the United States. A corollary to this point is that restrictions
on trading boost the permit price.
No International Trade of Permits
Although the Kyoto Protocol envisions international trade of permits, assuming no
trade provides a useful benchmark. Under this scenario, the models find a wide range
of permit prices in 2010, from $91 per metric ton (in 1997 dollars) in the G-Cubed
study to $407 in the Oxford study (see Table 3-4). The average permit price is $246
per metric ton, while the median price is $232.
Although many factors influence permit prices, variations in the models' assumptions
about price sensitivity account for most of the difference in the estimates. Estimates
of price sensitivity vary nearly four-fold across the models. Models with lower-than-
average permit prices generally make assumptions that overstate price sensitivity,
while some of the models with higher-than-average permit prices make assumptions
that understate price sensitivity (see Appendix B). On average, assumptions
overstating price sensitivity outweigh those understating it, thus reducing estimates
of the permit price.
The other determinants of permit prices are less important. As Table 3-1 shows, the
studies are reasonably consistent in their assumptions about how much U.S. emissions
would have to be reduced if permit trading were not allowed. Although assumptions
about the baseline price of carbon-energy also affect the permit price, they vary less
among models than those about price sensitivity and so explain less of the differences
in permit price among the models.
Studies that assume the United States can use forest growth and reductions in other
greenhouse gases to offset carbon emissions find lower permit prices, other things
equal. For example, the EIA study estimates that the permit price would fall from
$355 to $300 per metric ton if the United States could use forest growth or
17
reductions in other greenhouse gases to meet the Kyoto cap. Because businesses and
consumers do not need to reduce energy usage as much in this case, the protocol can
be satisfied with a lower permit price.
Unrestricted Trading of Permits Among Annex B Countries
The Kyoto Protocol envisions permit trading among Annex B countries, and every
study cited in this paper finds that trade lowers permit prices in the United States.
According to the studies, permit prices would range from $24 per metric ton (in 1997
dollars) to $222 per metric ton, compared with a range of $91 to $407 without
trading (see Table 3-5). Permit prices are lower in the United States when permits
can be traded among Annex B countries because the overall Annex B emissions cap
is easier to hit than the U.S. cap alone. So, other Annex B countries would be willing
to sell permits to the United States at prices lower than the U.S. no-trade price.
Much of the variation among studies stems from differences in price sensitivity. For
example, the Oxford study predicts a permit price about nine times as large as the
WorldScan study, even though both studies assume Annex B must cut carbon
emissions by 21 percent. Almost all of the difference stems from different
assumptions about how sensitive energy users are to changes in prices. The
WorldScan study assumes energy users are unusually sensitive; as a result, only a
small permit price is needed to induce them to make large cuts in their energy use.
(In addition, the WorldScan study assumes a somewhat lower baseline price of
carbon-energy.)
However, unlike the no-trading scenario, differences in assumptions about how much
emissions must be reduced to meet the Kyoto Protocol are also important. Models
disagree about both how far baseline emissions will be above Kyoto caps in 2010 and
what role offsets from CDM, forest growth and cuts in other gases will play. The cut
in carbon emissions varies from 8 percent to 21 percent among the studies.
Models that assume higher baseline growth in emissions, and thus larger percentage
reductions in carbon emissions, estimate higher permit prices, given the same price
sensitivity. For example, SGM-PNNL and GTEM assume similar price sensitivity for
Annex B, but SGM-PNNL assumes a smaller percentage reduction in emissions than
GTEM and thus estimates a lower permit price.
Models assuming offsets from CDM, forest growth and cuts in other gases find, on
average, lower permit prices than those that do not. The impact of offsets is largest
in the Clinton Administration study. The Administration assumed lower non-CO
2
greenhouse gases in the former Soviet bloc and very high price sensitivity for these
18
gases at low permit prices. These assumptions make the caps for these gases easier
to reach.
Unrestricted Global Trading of Permits
Amending the Kyoto Protocol to include non-Annex B countries would further
reduce the permit price. According to the models, prices would range from $13 per
metric ton (1997 dollars) to $80 per metric ton with unrestricted global trading of
permits, compared with a range of $24 to $222 with only Annex B participation (see
Table 3-6). (The range of prices under global trading may be artificially narrow
relative to the non-global trading range, because the studies with the highest and
lowest prices under Annex B trading did not examine the effects of global trading.)
With global trading of permits, the permit price would be determined by worldwide
supply and demand for permits. The worldwide percentage reduction in carbon
emissions under global trading would be smaller than the percentage reduction in the
carbon emissions of Annex B under Annex B trading. Global trading would spread
the Annex B reduction over a broader base, which lowers the percentage reduction.
In addition, baseline prices for carbon-energy average lower in non-Annex B countries
than in Annex B countries. As a result, most models find that emissions would fall by
a larger percentage in non-Annex B countries than in Annex B countries, given the
same permit price. The exceptions are those models that assume lower price
sensitivity in non-Annex B countries.
Permit prices vary among models because of differences in percentage reductions in
emissions, baseline prices of carbon-energy and price sensitivity of users. Not
surprisingly, the RICE model, which has the smallest percentage reduction in
emissions, the third-highest price sensitivity among Annex B energy users, and the
lowest baseline price of carbon-energy among non-Annex B countries, has the lowest
permit price. By contrast, the MERGE model has the least responsive energy users,
and the highest permit price.
Most studies assume that the caps for non-Annex B countries would be set equal to
projected baseline emissions in those countries. (The exception is the MS-MRT
model, which assumes that the non-Annex B countries must be given permits in
excess of their baseline level of emissions to induce them to participate. This
assumption reduces the permit price, but also trims the global reduction in emissions
from what it would be if the non-Annex B caps were set at projected baseline levels
of emissions.) If caps for non-Annex B countries were set lower than projected
baseline emissions, both the permit price and the reduction in global emissions would
be higher.
17.
Modelers
have also examined the effects of creating a European “bubble" (in which Europe collectively
meets
its overall target) alongside unrestricted permit trading among the other Annex B countries, but
that possibility is not discussed in this paper because it seems somewhat unrealistic.
18.
The
analysis assumes that the restrictions are binding. If the restrictions were not binding, U.S. permit
prices would be unaffected.
19.
Studies
using EPPA and MERGE assume permit purchases are restricted to one-third of a country’s
obligation,
while the study using MS-MRT assumed purchases are restricted to 30% of a country’s
obligation.
The MERGE study assumes restrictions in the context of global trading of permits, while the
other
two studies assume restrictions under Annex B trading of permits. In practice, the baseline would
not
be observed if the protocol were ratified, so the actual restriction would be defined in terms of
emissions at some fixed date in the past.
19
Limitations on Permit Trading
While
the Kyoto Protocol envisions permit trading among the Annex B countries, it
does
not rule out the possibility that trade could be limited by regulation or strategic
behavior.
Those limits could include restrictions on the purchases of permits by
permit-importing
countries, controls on sales of permits by permit-exporting
countries,
and the exercise of market power by permit-exporting countries.
17
The
models generally agree that limiting free trade would raise the permit price.
Restrictions on Permit Purchases
.
Restricting the number of permits the United States
could
purchase from other countries would boost permit prices in the United States
an
d in other countries facing the same situation.
18
But, permit prices would fall in
other
(unrestricted) countries, due to lower demand for permits in the international
market.
Wi
th binding restrictions on permit purchases, the advantages of global trading of
perm
its would nearly disappear. The United States would be unable to take
advantage
of the lower global price of permits to buy more permits. The domestic
permit
price would be determined by the tightness of the restriction, rather than by the
international
permit price. The only remaining advantage would be that a lower
international permit price would reduce the cost
of purchasing the fixed number of
permits allowed.
Although
several studies mention the possibility of restricting purchases, only three
examine the potential implication of hypothetically restricting each Annex B country’s
permit
purchases to roughly one-third of the difference between its caps and its
baseline
emissions.
19
Studies using the MERGE and MS-MRT models indicate that
such
a restriction would boost U.S. permit prices significantly. The MS-MRT study
finds that it
would boost the U.S. permit price to $166 per metric ton of carbon (in
1997 dollars), from $94 per ton under unrestricted Annex B trading.
Starting from
unrestricted
global trading, the MERGE study finds that restrictions would boost the
U.S. permit price from $80 to $167 per metric ton of carbon.
20.
A.
Denny Ellerman and Annelene Decaux,
Analysis of Post-Kyoto CO
2
Emissions Trading Using
Marginal Abatement Cost Curves
, Report Series No.
40 (Cambridge, Mass.: MIT Joint Program on the
Science and Policy of Global Change, October 1998).
20
By contrast, a study using the EPPA model finds that such a restriction would be
binding
on Japan and western Europe, but not on the United States or other regions.
20
Thus,
because the restriction would reduce international demand for permits, it would
reduce
the price of permits to U.S. and other OECD buyers from $180 to $162 per
me
tric ton (1997 dollars). Compared with other studies (including the study using the
EPPA
model cited in Table 2-1), however, this study assumes higher emissions
baselines
for other Annex B countries. Lower baseline emissions in other Annex B
countries
would reduce the international permit price low enough that the United
States would want to buy more permits than it was limited to.
Restrictions
on Permit Sales
.
Restricting the ability of permit-exporting countries to
sell
emissions permits would also boost permit prices. Two studies examined the
impact
of preventing Russia and other countries of the former Soviet Union from
selling
the excess permits that arise because their projected levels of emissions in 2010
will
be less than their levels in 1990. The MS-MRT model indicates that such a
restriction
on permit sales would increase permit prices from $94 to $136 per metric
ton
of carbon. The PNNL study, which uses the SGM model, estimates similar
effects. In that study, permit prices rise from $82 to $127 per ton of carbon.
Although
restricting the sale of excess permits from Russia and other countries of the
former
Soviet Union would raise permit prices, it would also increase the
environmental
benefits of Kyoto because it would effectively tighten the overall
emissions cap.
Exercise
of Market Power By Permit Exporters
.
With Annex B trading, the countries
of
the former Soviet Union could form a cartel to exploit their position as the primary
exporters
of permits. The impact of such a cartel would depend on its objective. If
its
objective were to maximize permit revenues from other countries, the cartel would
have
little impact. Although the former Soviet Union would be the primary exporter
of
permits, it would still account for only a small share of the permits used in the other
Annex
B countries. (Such a cartel would have considerably less power than OPEC,
which
supplies more than half of the oil used in western Europe and Japan and a
quarter
of the oil used in the United States.) If the former Soviet Union were to
restrict
permit sales to boost prices, it would lose about as much from lower volume
as it gained from higher prices.
However,
if a cartel pursued a broader economic objective than simply maximizing
permit
revenues, permit prices would rise higher. For example, reducing the number
of
permits exported would reduce permit revenues slightly, but would also increase
21.
The
estimates also assume that subsidies to energy usage, such as the tax-free treatment of employer-
provided parking, are not changed.
21
the
supply of permits to the domestic economies of the cartel, reducing energy prices
and
the associated adverse GDP impacts within those countries. If the former Soviet
Union
decided to maximize permit revenues less the direct economic costs of higher
domestic
energy prices, the PNNL study finds that Annex B permit prices would rise
35
percent, from $82 to $111 per metric ton of carbon (1997 dollars). If Eastern
Europe joined the cartel, prices would rise to $118 per tonne. A study using the
EPPA
model, assuming the same cartel objective, finds a much smaller 12 percent
price
increase. The MS-MRT study, using a broader measure of domestic economic
costs,
finds that a former Soviet bloc cartel would boost permit prices by 42 percent,
from $95 per tonne to $135 per tonne.
Amending
the protocol to include all countries would sharply limit the ability of a
former
Soviet Union cartel to boost permit prices. Any effort by these countries to
boost the international price of permits would produce a
sharp drop in permit sales
from
those countries, as other developing countries boosted their own sales. Those
other
countries could try to form a cartel, but its effectiveness would be limited unless
it covered most permit-exporting countries.
ESTIMATES OF PERMIT PRICES UNDER VARIOUS
SCENARIOS: MODEL SYNTHESIS
The
modelers’ results already indicate a general trend: permit prices in the United
States
are likely to be much higher if no international trading is allowed than if trading
is
unrestricted. However, the wide range of model results makes it hard to get a
quantitative
sense of the importance of various sorts of limits on permit trading. This
section
presents a synthesis of the model results, derived from a reduced-form model
created
using averages of the properties of the models, adjusted where necessary (see
Appendix B for the methodology.)
The
synthesis estimates include several outcomes consistent with the Kyoto Protocol,
as
well as two sets of outcomes that would require amendments to the Protocol: no
international
trade of permits, and global trading of permits (see Table 3-7). The
estimates are derived from model runs that assume a credible commitment to reducing
emissions
was made or expected by 2001. Given that the United States has not
ratified
the protocol, energy users will have less time to react, likely pushing permit
prices higher.
21
22.
Actually,
the range may not be symmetric around the point estimate because price sensitivity has a
nonlinear
impact on the permit price. A given reduction in price sensitivity boosts the permit price by
a
larger percentage than a same-sized increase in price sensitivity reduces it. In that case, the range of
uncertainty
would be 20 percent below the permit price to 34 percent above it. In the example, this
would produce a range of permit prices of $45 to $76 per metric ton.
22
A
wide range of scenarios is consistent with the Kyoto Protocol, were the United
States
to ratify it. Depending on how the protocol is interpreted and implemented,
permit
prices in the United States could range from $56 to $178 per metric ton of
carbon
in 2010. (All prices in this section are in 1997 dollars. To convert 1997
dollars
to 2002 dollars, multiply by 1.085.) Those estimates assume that forest
growth
has little net impact. If countries received unlimited credits for greenhouse
gases
absorbed by baseline (non Kyoto-related) forest growth and other land-use
changes,
permit prices would fall to zero in 2010, but the protocol would have no
effect
on the amount of greenhouse gases in the atmosphere. This paper does not
examine
the impact of the limited credits for baseline forest growth agreed to in
November 2001, or of Kyoto-related forest growth.
The
wide range of scenarios stems from the fact that several important issues are not
yet resolved:
Will
permit-exporting countries exploit their market power and raise permit prices
above competitive levels?
Will the clean development mechanism that allows Annex B countries to take credit
for emissions reductions in non-Annex B countries actually work?
Will
restrictions will be imposed on a country’s ability to purchase or sell its permits
to other countries?
Will the difficulties in monitoring reductions in non-CO
2
greenhouse gases confound
efforts to include these gases under the caps?
This
section also presents a range of possible error around the model synthesis
estimate
of the U.S. permit price for each scenario, reflecting the uncertainty present
in
each model’s estimates. Much of this uncertainty stems from estimates of price
sensitivity.
Modelers must estimate the factors determining price sensitivity, and their
e
stimates have some uncertainty attached. Based on estimates of uncertainty from
empirical
studies of the price sensitivity of energy demand, the actual permit price for
a
given scenario would fall within a range running from 27 percent below to 27
percent
above the estimated permit price for that scenario. For example, an estimate
of $56 per metric ton implies a range of $41 to $71 per metric ton.
22
Possible
errors in projected emissions baselines also add to the uncertainty of the
permit
price. However, such uncertainty has little impact on the relative sizes of the
costs
and benefits of reducing emissions. That is because trying to hit the same cap
from
a lower baseline would mean both a lower permit price and a smaller reduction
in
emissions. Using an extreme case to illustrate the point, if baseline emissions in
23.
Annex B offsets from CDM and other gases total 133 mmtc.
23
2010
equaled the Kyoto cap, the treaty would have no cost–the permit price would
be
zero–but also no benefit–emissions would be unaffected. This study provides no
estimates
of the effect of this type of uncertainty, but it would likely be of a similar
magnitude as that coming from uncertainty about price sensitivity.
Scenarios Consistent with the Kyoto Protocol
The
Kyoto Protocol envisions a system of emissions permits that would be traded
among
Annex B countries. However, the protocol is consistent with a variety of
al
ternative outcomes. The range of outcomes would be widened further if some
provisions of the protocol proved unworkable.
Ideal Implementation
.
Ideally, permit-exporting countries make no attempt to exploit
their
market power, the clean development mechanism works as promised, no
restrictions
are placed on the United States’ ability to purchase permits from abroad,
and
reductions in other greenhouse gases can be used to offset carbon dioxide
emissions.
Under that scenario, a synthesis of model results suggests that emissions
permits
would cost $56 per metric ton in 2010, the lowest estimate of permit prices
consistent with the protocol.
That
estimate is lower than most model estimates of the impact of the protocol, for
three
reasons. First, the model synthesis incorporates the positive effect of the clean
d
evelopment mechanism, which puts downward pressure on permit prices. By
c
ontrast, only one model (MERGE) incorporates such an effect. Second, most
modelers
do not account for the impact of offsets from other greenhouse gases.
(Without
the effects of the clean development mechanism or offsets from other gases,
the
permit price would rise to $81 per metric ton.
23
)
Third, the level of future baseline
emissions
assumed in the synthesis estimates is significantly lower than that assumed
by
most modelers. Synthesis assumptions are based on projections prepared by the
U.S. Energy Information Agency in 2000, rather than the
EIA projections prepared
in
1998 that many modelers used. Between 1998 and 2000, the EIA lowered its
projected
level of emissions, particularly in Europe and the former Soviet bloc. In
those
countries, emissions projected for 2010 were about 8 percent lower in 2000
than
what they had been in 1998. Partly offsetting those three factors, the price
sensitivity
of carbon emissions assumed in the synthesis estimates is somewhat lower
than the model average.
Ideal
implementation would reduce global greenhouse emissions in 2010 by about 6
percent
of baseline carbon dioxide emissions. (As a percent of baseline greenhouse
emissions,
the reduction would be smaller, but no one projects a global baseline for
24
other greenhouse gases.) Global emissions of carbon dioxide would be 33 percent
above 1990 levels in 2010, compared with a 40 percent rise in the baseline. The
reduction in global emissions would be larger if not for the fact that the protocol
would boost emissions in non-Annex B countries, a phenomenon known as leakage
(see Box 2-1).
BOX 2-1
EMISSIONS LEAKAGE
Leakage occurs when policies that reduce greenhouse gas emissions in Annex B countries cause
emissions to increase in other countries. Leakage can occur in two ways. First, reductions in Annex
B usage of crude oil would reduce its price in world markets. In response, non-Annex B countries
would increase their purchases of crude oil, although by a smaller amount than the reduction in
Annex B purchases.
Second, energy-intensive industries in Annex B countries would relocate some production to
countries where energy is cheaper to use, exporting this production back to Annex B countries.
Although exchange rates would adjust to offset the impact of such relocations on the overall trade
balance of Annex B countries, the composition of Annex B imports would shift toward energy-
intensive goods. This would increase emissions in the countries to which the industries relocated.
Provisions in the Kyoto Protocol allowing Annex B countries to trade permits reduce leakage, for
two reasons. First, permit trading reduces the price of permits, and thus energy prices, in most
Annex B countries. (Energy prices would be higher than in the no-trading case only in the former
Soviet bloc.) This reduces the incentive for energy-intensive industries to relocate to non-Annex
B countries. On average, the modelers find that increases in non-Annex B emissions would offset
10 percent of Annex B cuts with international trading of permits, but 17 percent without.
Second, trading of permits eliminates leakage to the former Soviet bloc. Any increase in emissions
by the former Soviet bloc reduces the number of permits the former Soviet bloc can sell to other
Annex B countries. Leakage to the former Soviet bloc would occur within the Kyoto Protocol only
if such limits were placed on the number of permits other countries could purchase that the domestic
price of permits in the former Soviet bloc fell to zero. On average, the modelers find that, without
permit trading, increases in emissions of the former Soviet bloc would offset 7 percent of the cuts
in the rest of Annex B.
For similar reasons, amending the Kyoto Protocol to allow unrestricted global permit trading would
eliminate leakage. Any increase in carbon emissions by Annex B countries would reduce the
number of permits they could sell.
Exercise of Market Power by Permit-Exporting Countries. Under ideal
implementation of the Kyoto Protocol, the developed countries would buy permits
from eastern Europe and the Annex B portion of the former Soviet Union. Those
latter countries could try to use their position as the only net sellers of permits to
boost the international price of permits above competitive levels. The ultimate
impact on prices would depend on the objective of the permit-exporting countries in
using their market power, and on how the Annex B countries reacted.
24.
In
the ideal implementation scenario, the countries of the former Soviet Union account for 79 percent of
permit exports, but just 12 percent of permits used by permit-importing countries.
25.
At that price, the former Soviet Union would sell only “hot air”—permits in excess of baseline
emissions—and reduce its domestic permit price to zero.
26.
This
scenario still implies a significant reduction in emissions in the former Soviet bloc, because of low
baseline
prices of carbon-energy in those countries. In fact, those countries would be reducing their
emissions by a larger percentage from baseline than any of the other Annex B countries.
25
If
the countries of the former Soviet Union formed a cartel in order to maximize their
revenues
from the export of permits, the impact on permit prices would be small.
Although
those countries would supply a dominant share of permit exports, they
would supply only a small share of the total permits used in the other Annex B
countries.
24
As those countries tried to boost prices, demand for their exports of
permits
would fall off. A synthesis of model results suggests that those countries
would
maximize their revenues if the permit price was $67 per metric ton of carbon,
which
is only modestly higher than permit price of $56 per tonne under the ideal
implementation scenario described above.
Permit
prices would rise higher if the former Soviet Union instead attempted to
maximize
its GDP plus revenues from permit exports. Withholding permits from the
world
market would not only boost the world permit price, but it would also make
more
permits available for consumption in the countries of the former Soviet Union.
That
development would mute the rise in energy prices in those countries and thus the
loss in their GDP.
World permit prices would rise to $85 per metric ton under this
scenario.
25
This
scenario would not occur if other Annex B countries could credibly threaten to
retaliate,
perhaps by imposing quotas on the import of permits. In that case, a more
likely
outcome would be for the former Soviet Union to restrict exports of permits
below
the levels in the ideal implementation scenario, but not so far as to trigger
retaliat
ion from other countries. For example, if countries of the former Soviet
Union
and eastern Europe restricted permit exports by enough to hold their domestic
permit
prices at half the level of international permit prices, the international permit
price would be $70 per ton.
26
No
Clean Development Mechanism.
If the Clean Development Mechanism proved
unworkable,
the impact on U.S. energy prices could be small, because, according to
the
assumption made by the EMF in examining this scenario, the impact of the Clean
Development
Mechanism itself would be small. Assuming a moderate exercise of
market power by
the former Soviet bloc, the U.S. permit price could rise from $70
per metric ton to $76 per metric ton of carbon under this scenario.
27.
This estimate assumes that the
countries of the former Soviet bloc would exercise a moderate degree of
market
power and restrict permit sales so that their domestic price of permits is one-half the international
price of permits.
26
Restrictions
on U.S. Purchases of Emissions Permits.
The Kyoto Protocol states that
“trading
shall be supplemental to domestic actions for the purpose of meeting
quantified
emission limitation and reduction.” Rules governing trading are to be
developed at a
future “Conference of the Parties.” In the past, the European Union
has
suggested that the term “supplemental” means that limits should be imposed on
each
country’s ability to satisfy its Kyoto obligation by purchasing permits from other
countries.
Such
limits could have a significant effect on permit prices. If the United States was
requ
ired to achieve at least 65 percent of its obligation by reducing domestic
emissions,
the U.S. permit price would jump to $122 per metric ton. (This restriction
is
roughly equivalent to that proposed by the European Union in 2000.) By contrast,
permit
prices in the former Soviet Union and eastern Europe would fall very close to
zero
because those countries could supply the limited demand from the rest of Annex
B countries entirely by selling their “hot air”—permits in excess of baseline emissions.
A
tighter limit on permit purchases would result in a higher permit price, while a
looser
restriction would result in a lower price. Note that with restrictions on permit
purchases,
the CDM would have no impact on permit prices. Any credit obtained
from
a non-Annex B country would reduce the number of permits the United States
could purchase from other Annex B countries.
Restrictions
on Sales of Emissions Permits.
The Conference of Parties could instead
impose
limits on the amount of permits a country can sell. For example, the countries
of
the former Soviet bloc could be prevented from selling their hot air. This
restriction would be equivalent to changing their permit allocation to match their
baseline
level of emissions. (This restriction is not as tight as that proposed by the
European
Union in 2000.) Under this scenario the permit price in Annex B countries
would
rise to $137 per metric ton of carbon.
27
The value of permit sales by the
former
Soviet bloc would only decline by 14 percent, from $38 billion to $33 billion
(in
1997 dollars), because the higher price would nearly offset the decline in the
number of permits sold.
That
scenario would reduce global emissions substantially more than other
interpretations
of the Kyoto Protocol. Global emissions of greenhouse gases in 2010
would
fall by the equivalent of nearly 10 percent of baseline carbon dioxide emissions,
about
one and one-half times as much as in the ideal implementation scenario.
Emissions
would be lower than under a policy of limited purchases because permit
prices
would be higher in every Annex B country, including those of the former Soviet
28.
Department
of State,
United States Submission on Land-Use, Land-Use Change and Forestry
,
August
1, 2000 (available at www.state.gov/www/global/global_issues/climate/000801_unfccc1_subm.pdf).
29.
United
Nations Framework Convention on Climate Change,
Greenhouse Gas Inventory Database
,
2000
(available at www.unfccc.de).
27
bloc.
Nonetheless, global emissions of carbon dioxide would still be 29 percent above
1990 levels in 2010.
In
later years, the impact on both the permit price and emissions of a restriction on
sales
of hot air would fade as hot air disappeared with higher baseline emissions in the
former
Soviet Union bloc. However, if the restriction on sales were tied to the
number
of permits sold, rather than to the amount of hot air sold, it could have a
permanent effect on permit prices and global emissions.
No
Offsets from Other Greenhouse Gases.
If other greenhouse gases proved too
difficult
to monitor, or if other problems arose in imposing permits on sources of these
emissions,
the U.S. permit price would rise still further. With restrictions on sales of
hot
air and caps on carbon dioxide equal to the carbon dioxide portion of the overall
Kyoto
cap, each permit would cost $178 per metric ton of carbon in 2010. Taking
account
of uncertainty in model estimates of price sensitivity, the permit price would
fall
in a range of $130 to $226 per metric ton. This permit price would apply only to
carbon
dioxide, and so would have only a slightly larger impact on the overall level
of U.S. prices than a $137 per ton charge on all greenhouse gases.
Permit prices would be higher under this scenario for two reasons. First, other
greenhouse
gases are cheaper to reduce than carbon dioxide. Second, the percentage
difference
between baseline emissions and the Kyoto caps is much larger for carbon
dioxide than it is for the other greenhouse gases.
Credit
for Baseline Forest Growth.
In August 2000, the U.S. Department of State
argued
that countries should receive credits for the change in carbon stocks on
managed
lands during the commitment period, including changes that would have
occurred
whether or not the protocol was ratified.
28
(Managed lands include
cropland,
grazing land and forests, except those not available or appropriate for wood
production.
) According to estimates submitted by the countries to the United
Nations,
such credits would slightly exceed the difference between projected baseline
emissions
and the Kyoto caps in Annex B in 2010.
29
The supply of emissions permits
and
credits would exceed baseline emissions, driving the permit price to zero in 2010.
In
later years, if the cap were frozen at its 2010 level, rising baseline emissions would
eventually result in positive permit prices.
Al
though this interpretation of the Kyoto Protocol’s provisions on forest growth
would
eliminate the economic costs of the protocol in 2010, it would also eliminate
30.
Department
of State,
Proposal by United States, Canada, Japan: Phase-in for Forest Management in the
First Commitment Period
,
November 21, 2000 (available at www.state.gov/www/global/global_issues/
climate/cop6/001121_phase-in.html).
31.
Bruce
A. McCarl, “Carbon Sequestration via Tree Planting on Agricultural Lands: An Economic Study
of
Costs and Policy Design Alternatives,” Internet draft, November 1998 (available at
http://ageco.tamu.edu/faculty/mccarl/papers/676.pdf).
28
the
protocol’s environmental benefits in that year. Since the protocol would not
impose
a cost to emitting greenhouse gases, there would be no reduction in such
emissions.
These
results assume no limitations on countries’ ability to buy and sell permits. With
such
limits, permit prices would be positive, but far smaller than if credits for baseline
forest
growth were not allowed. For example, with the limit on purchases described
above,
the U.S. permit price would fall to $50 per mtc in 2010, from $122 per mtc
without
credits for forest growth. However, the reduction in global emissions of
greenhouse
gases in 2010 would fall to 3 percent of baseline carbon dioxide
emissions, much lower than in any other Kyoto-consistent case.
In
November 2000, the United States, Canada and Japan proposed a somewhat
stricter
treatment for baseline forest growth, or “existing effort.”
30
Under this
proposal,
baseline forest growth could be used to satisfy about 60 percent of Annex
B’s
required reduction in emissions. In the ideal implementation case, the permit price
would
drop from $56 per metric ton to $24 per metric ton, but the cut in global
emissions would fall from 6 percent to 3 percent of baseline carbon dioxide.
If
credits were only given for forest growth that would have occurred in the absence
of
the protocol, permit prices would be lower than if no credit were given for any
forest
growth. Countries could substitute forest growth for the most expensive
cutbacks
in emissions. This would trim the overall reduction in global emissions, but
increase
absorption of greenhouse gases by the same amount. Unfortunately, it is
impossible
to use the economic models to determine the effect this would have on
permit
prices, because none of them include such an effect. One paper finds that such
effects could be large.
31
Scenarios Requiring Amendments to the Kyoto Protocol
Amending
the Kyoto Protocol to prevent the international trade of permits or to
include countries outside of the Annex B region could produce impacts on permit
prices that are larger or smaller than those shown above.
29
No Permit Trading Between Countries. If the United States could not purchase
permits from other countries, but could use reductions in emissions of other
greenhouse gases to offset reductions in carbon dioxide, emissions permits would cost
$216 per metric ton in 2010. If offsets from other greenhouse gases could not be
used, the permit price would rise to $264 per metric ton in 2010.
The only advantage of eliminating international trade of permits would be that
countries could not use hot air from the former Soviet bloc to reduce their own need
to cut emissions. Consequently, global emissions of greenhouse gases would be
reduced from baseline 2010 levels by the equivalent of 8 percent of baseline carbon
dioxide emissions if no offsets from other gases were allowed and by 9 percent if
offsets were allowed. This advantage would disappear over time as higher baseline
emissions in the former Soviet Union eliminated hot air.
Global Trading of Emissions Permits.
Amending the Kyoto Protocol to include non-
Annex B countries could greatly reduce the impact of the protocol on energy prices.
If emissions caps for these countries were set equal to projected baseline levels for
these countries and there were no limitations on trading, the price of an emissions
permit would drop to $28 per metric ton of carbon. Permit prices are lower because
global trading allows Annex B countries to substitute low-cost reductions in non-
Annex B emissions for more expensive reductions in their own emissions.
The behavior of the former Soviet bloc would have a smaller impact on U.S. permit
prices with global trading because non-Annex B countries would provide a large
alternative source of permits. If countries of the former Soviet Union and eastern
Europe restricted permit exports by enough to hold their domestic permit prices at
half the level of international permit prices, the international permit price would only
rise to $31 per metric ton. Preventing the sale of hot air would only boost the permit
price to $49 per metric ton.
Putting restrictions on permit purchases, however, would boost the permit price to
$122 per metric ton, roughly the same as under Annex B trading of permits. The
United States would be unable to take advantage of the additional permits non-Annex
B countries would be willing to supply. If, in addition, no offsets were available from
reductions in other greenhouse gases below target levels (i.e., if the non-carbon
dioxide portions of the caps were eliminated), the U.S. permit price would rise still
further, to $147 per metric ton.
Expanding the Kyoto Protocol to include non-Annex B countries would reduce global
emissions by eliminating leakage to those countries. The global reduction in emissions
of greenhouse gases would be 8 to 10 percent larger with global trading of permits
than under the Kyoto Protocol.
32.
The
change in gasoline prices is not proportional to the permit price because the drop in crude oil prices,
due to lower demand, depends
on global demand for crude oil rather than the permit price. The change
in
electricity prices is not proportional to the permit price because the mix of fuels used to generate
electricity changes as the permit price rises.
30
CHAPTER 4
THE EFFECTS OF THE KYOTO PROTOCOL ON ENERGY PRICES
A
system of tradeable permits would reduce emissions by boosting prices of energy
made
from coal, petroleum and natural gas. Although consumers would not pay for
emissions permits directly, permit prices would have a large influence on energy
prices,
just as prices of crude oil influence gasoline prices even though motorists do
not pay for crude oil directly.
According to a synthesis
of model results, each $100 per metric ton increase in the
permit
price would add 17 to 22 cents per gallon to gasoline prices, roughly $1.57 per
thousand
cubic feet to the price of natural gas, roughly $55 per short ton to the user
price
of steam coal, and 1.4 to 1.7 cents per kilowatt-hour to the price of electricity.
32
On
a percentage basis, gasoline prices would increase the least and coal prices the
most.
Several
factors determine the impact of permit prices on energy prices. Of these, the
direct
impact of the permit price is the most important. In addition, lower demand for
coal
and crude oil depresses their producer prices, offsetting part of the impact of
permit
prices. Depending on whether demand for natural gas rises or falls, the
wellhead
price of natural gas will rise or fall, adding to or subtracting from the impact
of
permit prices. Changing margins for refiners, distributors and electricity generators
may
also affect energy prices. Finally, changes in the mix of fuels used will help
determine how much electricity prices rise.
The
estimates of the potential impact of the Kyoto Protocol on energy prices reflect
the
changes in permit prices found in the previous chapter. Just as different
interpretations
of the protocol could lead to different permit prices, a wide range of
outcomes
for energy prices is consistent with the protocol. U.S. gasoline prices could
rise
anywhere from 12 to 38 cents per gallon higher than they would be otherwise.
(All
prices in this chapter are in 1997 dollars.) The price of natural gas to households
would
increase between 13 and 42 percent above baseline levels. Electricity prices
to
households would be 13 to 36 percent higher than they would be without emissions
restrictions.
Accounting for the uncertainty present in model estimates of price
sensitivity would widen these ranges somewhat.
As
discussed in the introduction, synthesis estimates of the change in energy prices
are
based on projections of emissions and energy prices released by the Energy
31
Information Administration in 2000. More recent projections of emissions and energy
prices are higher. If synthesis estimates were prepared using the new projections, they
would show larger absolute increases in energy prices and larger percentage changes
in electricity prices in every scenario. The percentage change in natural gas prices
would likely also be higher.
MODEL ESTIMATES OF EFFECTS ON ENERGY PRICES
Although every study provides estimates of permit prices, few translate these prices
into changes in energy prices. This section draws heavily on results from the DRI,
EIA and WEFA studies, which provide the most detail on energy prices. Differences
with other studies are also noted.
Coal Prices
Each $100 per mtc in permit prices directly boosts the price of coal delivered to
electric utilities by roughly $55 per short ton, or between 213 and 245 percent (see
Table 4-1). Higher grades of coal, which have higher heat content, would face larger
absolute price increases, although higher baseline prices for these grades would mean
smaller percentage increases.
The modelers expect minemouth prices for each grade of coal to fall due to lower
demand. WEFA projects that a permit price of $265 per metric ton would reduce
average minemouth prices by $3.60 per short ton. EIA expects average minemouth
prices to rise, but only because of shifts in the mix of coal being mined, not because
of increases in prices for any grade of coal. For example, a ton of coal from the
Powder River basin in Wyoming is cheaper than a ton of Appalachian coal at the
minemouth, but contains more carbon and less sulfur per unit of heat. So, as prices
of greenhouse permits rise and prices of sulfur dioxide permits fall, Powder River
basin coal loses its cost advantage, and the mix of coal used shifts toward more-
expensive Appalachian coal, boosting the average minemouth price, even though the
price of each grade falls.
Gasoline Prices
The impact of the Kyoto Protocol on the price of gasoline depends on three factors:
the direct impact of the permit price; the indirect impact of reduced demand for crude
oil on crude oil prices; and the impact of lower volumes on refiner and distributor
33.
Energy Information Administration,
Annual Energy Outlook 2000
, DOE/EIA-0383 (December 1999).
32
margins.
The first effect boosts the gasoline price, while the other two effects partly
offset this increase.
Direct
Effects.
The carbon content of gasoline alone determines the direct impact of
permit
prices. Each gallon of gasoline contains about 5.2 pounds of carbon, so that
each
$100 per mtc (metric ton of carbon) increase in the permit price boosts the price
of
gasoline by 23.8 cents per gallon, all else equal.
33
The carbon content of other
petroleum
products is somewhat higher than that of gasoline, so a $100 permit price
would
directly add almost 26 cents per gallon to jet fuel prices and about 27 cents per
gallon
to distillate (diesel) prices. The direct impact of permit prices on gasoline
prices is smaller
in models that assume that capital and labor can be substituted for
crude oil in the production of gasoline.
Effects on Crude Oil Prices.
Higher prices for
petroleum-based energy would lead
consumers
and businesses to reduce their purchases of it, reducing world-wide
demand
for crude oil. As the impact of lower Asian demand on world oil markets in
1998
demonstrated, reduced demand would push crude oil prices down. The size of
the
drop in prices would not depend on the U.S. permit price, but rather on the global
reduction in crude oil demand.
On
average, the models find that the percentage drop in crude oil prices would be
nearly twice as large
as the percentage drop in global petroleum consumption. So,
a
3 percent drop in worldwide petroleum usage would produce a nearly 6 percent
drop
in crude oil prices. This works out to 2 to 3 cents per gallon for each $100 per
mtc in permit prices.
Margins.
Most studies, including those of DRI and WEFA, assume that changes in
demand
for petroleum-based energy do not impact refiners’ profit margins. The EIA
study,
however, predicts that low capacity utilization would cause refiners to reduce
margins
in order to compete for business. This would reduce gasoline prices by about
3 cents per gallon for each $100 per mtc increase in permit prices.
Natural Gas Prices
Each
$100 per mtc in permit prices directly adds almost $1.50 to the price of a
thousand cubic feet of natural
gas. In addition, demand for natural gas, unlike that
for
coal and petroleum, may rise due to higher demand from electric generators,
pushing
prices up further. Among the modelers publishing natural gas usage, DRI,
EIA,
SGM-PNNL and WEFA expect an increase in demand from electric utilities to
outweigh
lower demand for natural gas by households, businesses and governments,
33
while the G-Cubed, JWS and MS-MRT models predict overall natural gas usage to
fall. Estimates of the total impact of a $100 per metric ton permit price on wellhead
prices (excluding permit costs) thus range from a drop of about 30 cents per thousand
cubic feet to an increase of almost 25 cents.
EIA projects that higher demand for natural gas would boost distributors’ margins.
This would push the price each natural gas customer faces higher, although because
demand would shift from high-margin residential and commercial customers to low-
margin utilities, the average price to all users would fall. The DRI, EIA and WEFA
studies find that each $100 increase in permit prices would add $1.55 to $1.83 per
thousand cubic feet to the residential price of natural gas.
Electricity Prices
Unlike refiners and natural gas distributors, electric utilities can substitute biomass,
solar or wind power for fossil fuels and natural gas for coal, substantially lowering the
carbon content of a kilowatt-hour of electricity without reducing the amount of
electricity generated. In fact, the modelers find that changes in the fuel mix account
for reductions in emissions that are at least as large as those from cuts in electricity
demand.
Although fuel substitutions reduce the direct impact of the permit program, they
introduce a cost as well: fuel and generating costs are higher. (Otherwise, utilities
would already be using the new mix of fuels.) As a result, the net impact of the Kyoto
Protocol on electricity prices would be smaller than the permit cost of the original mix
of fuels, but larger than the permit cost of the final mix of fuels.
Using the baseline mix of fuels, generating technologies, and fossil fuel prices, each
$100 per mtc increase in permit prices would boost electricity prices by 1.7 cents per
kilowatt-hour, according to the DRI, EIA and WEFA studies. Although shifting
from coal lowers the direct impact of the permits on electricity prices considerably,
it boosts other costs, eliminating most of the savings. In the EIA study, for example,
fuel shifting reduces the costs of the permits to less than 1 cent per kilowatt-hour per
$100 permit price, but higher wellhead prices for natural gas add 0.1 cent, and higher
fuel and generating costs add another 0.5 cent. In the end, electricity prices rise
nearly 1.6 cents per kilowatt-hour per $100 permit price anyway, and they are only
9 percent less than they would be if utilities used the baseline mix of fuels. Savings
from switching fuels are smaller in the DRI study and are actually negative in the
WEFA study.
The increase in the electricity price is larger than permit costs and a simple estimate
of higher generating and fuel costs (discussed in Appendix B) would produce. Part
34.
Estimated
prices for natural gas and electricity do not include the cost of permits for methane emitted by
coal mines and natural gas system, and thus are probably too low.
34
of
the reason generating costs are surprisingly high in the DRI and WEFA studies is
that
these models assume that electric generators’ decisions about plant type take into
account
likely permit costs over the entire service lifetime of the plant. Both studies
anticipate
that permit prices will rise after 2010, so, while a new natural gas-fired
plant
may increase costs in 2010, it may nonetheless cost less to operate over the
whole
life of the plant. Thus, costs and prices may rise by more than the value of
permits saved in 2010.
In
addition, the studies assume that, at least in some regions, electricity prices are set
according
to the cost of the last kilowatt-hour generated, rather than according to the
average
cost of all electricity generated. As coal-fired plants are moved from
providing
baseload generation to providing marginal generation, marginal costs rise
more
than average costs, so prices rise more than the average cost of fuel-switching
would predict. By 2020, both of these factors
are less important, as coal plants are
retired.
ESTIMATES OF EFFECTS ON ENERGY PRICES: MODEL SYNTHESIS
Permit
prices play a key role in determining the effect of the Kyoto Protocol on
energy
prices. Scenarios that produce high permit prices, such as limits on
international
trading of permits, also produce large changes in energy prices (see
Table
4-2). Depending on the scenario, U.S. gasoline prices would rise 12 to 38 cents
per
gallon (in 1997 dollars) above baseline levels in 2010. Natural gas prices to
households
would rise between 13 and 42 percent above baseline levels, while
electricity
prices to households would increase between 13 and 36 percent.
34
Scenarios
requiring amendments to the protocol, such as global trading of permits or
no
international trade of permits, could produce smaller or larger impacts. The same
uncertainty
present in estimates of the permit price is also present in estimates of
changes in energy prices.
35.
Emission-producing goods and
services can produce emissions either when they are used or when they
are produced.
36.
I interpret the economic impacts
in 2010 as representative of impacts over a longer period of time. The
actual
impacts in 2010 may be larger, while the impacts in some other years during 2005 to 2015 may
be smaller. See Appendix B.
35
CHAPTER 5
MACROECONOMIC AND DISTRIBUTIVE EFFECTS OF CUTTING
GREENHOUSE GAS EMISSIONS
The various economic studies surveyed find that restrictions on emissions of
greenhouse gases by a system of tradeable permits would reduce both output and
consumption
and transfer a significant amount of income from producers and
consumers of emission-producing goods and services to recipients of permits.
35
However, auctioning permits and using the receipts to cut tax rates could reduce
those losses in output and consumption.
The impacts of the Kyoto Protocol on output, consumption and the distribution of
income would depend on how the protocol is interpreted and implemented. In
particular, the fewer the restrictions on permit trading, the smaller the impacts on the
macroeconomy and the distribution of income.
The macroeconomic impact of reducing emissions of greenhouse gases can be
measured in several ways. This paper focuses on two: the change in GDP and the
change in consumption. The change in GDP measures the effect of emissions
reductions on output, while the change in consumption shows the overall impact of
emissions reductions on standards of living. A third measure, direct cost, captures
only those losses in output suffered directly by consumers and producers of energy.
It thus ignores feedback effects on the rest of the economy.
The synthesis of model results suggests that real U.S. GDP would decline between
0.5 percent and 1.2 percent below baseline levels in 2010 under the Kyoto Protocol,
and real consumption would fall between 0.4 percent and 1.0 percent, depending on
the scenario.
36
The direct cost of the Kyoto Protocol would be between 0.2 percent
and 0.4 percent of GDP. The total value of permits used, which indicates the amount
of income that would be transferred from producers and consumers of energy to
recipients of permits, would total between $108 billion and $245 billion (in 1997
dollars) in 2010, or 0.9 percent to 2.0 percent of GDP. (Many households would
both pay and receive funds.) If the government auctioned the permits, its revenues
would rise by a comparable amount.
37.
The
Harberger triangle is the area between the supply and demand curves, both constructed excluding
taxes, and to the right of actual quantity.
36
Those estimates of losses in GDP and consumption reflect the same uncertainty
present in estimates of the permit price. That is, if the price sensitivity of emissions
is actually higher than estimated, permit price and losses in GDP and consumption
will be lower than estimated, and if price sensitivity is lower than estimated, permit
price and losses in GDP and consumption will be higher than estimated. In addition,
there is some uncertainty about the effect a given permit price has on GDP and
consumption. These latter sources of uncertainty are difficult to quantify, so the
ranges of possible error presented in this chapter focus only on uncertainty in the
estimate of price sensitivity.
As
discussed in the introduction, synthesis estimates of changes in GDP and
consumption
are based on projections of emissions and energy prices released by the
Energy
Information Administration (EIA) in 2000. More recent projections of
emissions
and energy prices are higher, meaning that updated synthesis estimates of
the reduction in GDP and consumption would be larger in every scenario.
DIRECT COST
The direct cost of abatement measures the economic cost that a cap or tax imposes
in the directly-affected market alone
in this case, an implicit market for emissions.
It thus excludes feedback effects, for example those stemming from interactions
between energy markets and the rest of the economy.
Direct cost consists of two parts: domestic direct cost (the Harberger triangle
37
) and
permit purchases from other countries. Domestic direct cost is the loss in value that
users and producers would incur because the Kyoto cap would force users to
substitute away from fossil energy and products that produce other greenhouse gases.
Those substitutions would divert resources from producing fossil energy toward
other uses. But the alternative uses of those resources would be less valuable than
the baseline uses: if they were as valuable, it would not take positive permit prices
to get people to adopt the alternative uses. Permit purchases are the dollar value of
permits to emit greenhouse gases purchased from other countries.
Model Results
The models surveyed in this paper find that the Kyoto caps would carry direct costs
between 0.2 percent and 0.9 percent of baseline GDP in 2010 if permits could not
38.
These
calculations assume that there are no pre-existing taxes on energy, such as gasoline taxes. Such
taxes
increase the adverse impact of additional increases in the price of domestic energy, boosting direct
cost.
Without international trade of permits, pre-existing taxes boost estimates of the U.S. direct cost by
almost a third.
39.
With
restrictions on permit purchases, the price the United States pays for foreign permits can be
anywhere
between the permit prices in permit-exporting countries and in the United States. If restrictions
on
permit purchases push this “import” price down far enough, total direct cost can fall while the permit
price and domestic direct cost rise.
37
be traded internationally.
38
In most cases, models with the highest permit prices find
the highest direct costs, while models with the lowest permit prices find the lowest
direct costs.
The studies agree that the direct cost would be lower with unrestricted trading among
Annex B countries. Even though trading would incur the cost of purchasing permits
from abroad, the direct cost would amount to just 0.1 percent to 0.6 percent of GDP
in 2010. International trading of permits reduces the direct cost because trading
allows energy users to substitute purchases of foreign permits for more costly
reductions in their own emissions. Restricting the trade of permits among Annex B
countries would raise the direct cost.
Revising the Kyoto Protocol to permit global trading of permits would reduce total
direct cost even further, to 0.1 percent to 0.3 percent of GDP. Although the United
States would purchase more permits from other countries under this scenario, permit
prices would fall by enough to reduce the cost of permit purchases from other
countries. In addition, global trading of permits would reduce the amount of
emissions that the United States would have to cut domestically, which would reduce
domestic direct costs.
Synthesis of Results
Based on a synthesis of model results, the Kyoto Protocol would impose total direct
costs of 0.2 percent to 0.4 percent of baseline GDP in 2010. The precise amount
would depend on how the protocol was implemented. Direct cost would be lowest
if no restrictions are placed on trade of permits among Annex B countries. In
general, restrictions on permit trading mean higher permit prices and higher direct
costs.
39
These estimates do not take account of pre-existing taxes on energy;
accounting for such taxes increases direct cost by about 30 percent, to a range of 0.3
percent to 0.6 percent of GDP.
Without
international trade of permits, total direct cost would be 0.4 percent of
baseline
GDP in 2010 if the United States had to meet the Kyoto cap for all
greenhouse
gases; and 0.5 percent of GDP if only carbon dioxide was capped. (These
40.
In addition, higher energy prices
could affect the pace of overall technological change. However, none
of the studies cited in this paper examined this possibility.
38
estimates
do not take account of pre-existing taxes.) By contrast, unrestricted global
trading of permits would reduce total direct cost to just 0.1 percent of GDP.
IMPACT ON U.S. GDP AND CONSUMPTION
In
general, restricting emissions of greenhouse gases would reduce U.S. gross
dome
stic product (GDP) and consumption. The precise amount is uncertain, and
depends
on the details of the proposal. Moreover, those losses could be significantly
reduced—or possibly eliminated—if the government auctioned the emissions permits
(instead
of giving them away) and used the revenues from the auction to reduce
marginal
tax rates. This section and the next assume that permits are given away for
free; the following section explores the implication of auctioning permits.
GDP
is the total market value of goods and services produced domestically during a
given
period, and it is the broadest measure of a country’s economic output.
However, GDP is not a measure of the
standard of living. Consumption is a better
measure
of the standard of living, although it is also imperfect because it does not
include
the value of non-market activities, like leisure. Nonetheless, estimates of the
effects of Kyoto on GDP and consumption are useful benchmarks.
Restrictions
on emissions could affect the economy through several channels. First,
they
would lower potential GDP because higher energy prices would raise the cost
of
capital (which would reduce investment in new plant and equipment) and lower the
real
wage (which would discourage work). Second, reduced energy usage would
render existing labor and capital less productive, further reducing potential GDP.
Third,
higher energy prices might hurt the profitability of new investment in some
countries
more than in others, leading to changes in flows of capital among nations.
Fourth,
if permits were traded internationally, paying for foreign permits would divert
resour
ces from domestic investment and consumption. Fifth, the Federal Reserve
might
have to raise interest rates temporarily to curb inflationary pressures that stem
from
higher energy prices. Such interest rate hikes would lead to higher
unemployment
in the short run. Finally, if the government auctioned the permits, the
addi
tional revenue could be used to cut taxes or increase spending, which would
affect the economy in different ways.
40
41.
The EPPA, SGM-Administration, SGM-PNNL and WorldScan studies do not report changes in GDP.
42.
Many
modelers treat private consumption, government consumption and government investment together.
This
is a close approximation of total consumption. In 1999, 81 percent of government consumption and
investment was government consumption.
39
Model Estimates of the Impact on U.S. GDP and Consumption
Estimates
of losses in GDP and consumption vary widely among studies, depending
on
the model used and on the degree of international trading of permits assumed.
Models
that assume inflation and unemployment can vary from baseline levels
generally
find larger losses than models that do not. And all else equal, models that
assume
energy usage is very sensitive to prices have smaller losses in GDP and
consumption than models that assume energy usage is insensitive.
Permit
prices are a key determinant of GDP loss, so the fewer restrictions placed on
international
trading of permits, the lower the permit price, and the smaller the loss
in GDP. International trade of permits also reduces losses in consumption, despite
the
transfer of income to other countries to purchase foreign permits. When permits
are
traded, prices of emission-producing goods and services are lower, which
outweighs
the negative effect on consumption of purchasing foreign emissions
permits.
No
International Trade of Permits.
Every study finds that losses in GDP and
consumption would be largest if carbon emissions were reduced without international
trade of
emissions permits. Among models reporting effects on GDP, the losses in
2010
vary from 0.4 percent to 4.2 percent of baseline GDP (see Table 5-1).
41
Total
(private plus government) consumption falls 0.2 percent to 3.1 percent below baseline
levels
in 2010, except in the G-Cubed model, which shows a rise in consumption.
42
(That
model finds that consumption is permanently above baseline levels, while GDP
is permanently below baseline levels, a finding that seems implausible.)
Losses
in GDP and consumption are generally larger in studies using
macroeconometric
models than in studies using general equilibrium models. The
larger
losses come from one of two mechanisms. In some macroeconometric models,
higher
energy prices lead the Federal Reserve to raise interest rates, dampening
demand,
raising unemployment, and reducing output. The same mechanism also
slows
investment and productive capacity. In the DRI model, another mechanism
operates:
the rise in energy prices directly slows consumption growth by reducing
consumers’ real incomes without any rise in interest rates.
Among
the general equilibrium models, the CETA and JWS models find unusually
large
GDP losses given their estimates of the permit price (Figure 5-1). In the JWS
model,
this result is due to a much larger decline in labor supply in response to lower
43.
These
models and JWS have the largest ratios of GDP loss to direct cost. Any loss not coming from direct
cost
must come from lower investment or lower labor supply. In JWS, these other losses come primarily
from lower labor
supply. In CETA, MS-MRT and RICE, which hold labor supply constant, they come
from lower investment.
40
real
wages than that of other models. However, it is unclear why GDP loss is larger
in
the CETA model than in the MERGE model, since both models use similar
assumptions.
Among studies using macroeconometric models, Oxford has the
smallest
GDP losses relative to permit price because Oxford assumes that the
reduction
in GDP needed to prevent higher inflation is completed before 2010. As
a
result, by 2010 the relationship between permit price and GDP in the Oxford model
is similar to that in the general equilibrium models.
The
percentage reduction in consumption is usually smaller than the percentage
reduction
in GDP. One reason is that the decline in GDP partly mirrors a decline in
productive
capacity. This means that a smaller share of GDP needs to be devoted to
replacing
depreciated capital, and that a larger share can go toward consumption. In
addition,
reduced rates of return cause people to save a smaller share of their income,
further
softening the impact of lower GDP on consumption. So the general
equilibrium
models in which lower investment accounts for the largest share of GDP
loss
(CETA, MS-MRT and RICE) find unusually small losses in consumption given
the
change in GDP.
43
However, the difference between the percentage declines in
consumption
and GDP is less apparent in the macroeconometric models, because the
more
the Federal Reserve would have to raise interest rates in order to subdue
inflation,
the more equity prices would fall, and the larger would be the decline in
consumption.
Also, higher interest rates would boost saving, further reducing
consumption.
International
Trading of Permits.
Allowing international trading of permits would
reduce
the losses in U.S. GDP and consumption associated with reducing greenhouse
gas
emissions. Permit trading allows the United States to achieve some of its
emissions
reductions in other countries, where those reductions are cheaper. Every
s
tudy finds that, under unrestricted permit trading, the reduction in the cost of
d
omestically allocated permits overwhelms the cost of permits purchased from abroad
(see
Table 5-2). Losses in GDP and consumption would be lowest with global permit
trading,
because permit prices would be lowest. Limiting trade to a smaller area, as
under
the Kyoto Protocol, would boost losses, but they would still be far smaller than
with
no trade. Restricting purchases, so that the U.S. permit price did not equal the
international price, would remove many of the benefits of permit trading.
With
unrestricted permit trading among Annex B countries, the models estimate that
the GDP loss in 2010
would range between 0.2 percent and 2.0 percent of baseline
44.
Most
models treat U.S. purchases of foreign permits as a financial transaction, which is excluded from
the
GDP accounts. The model synthesis estimates also follow this practice. However, the MERGE
model treats permit purchases as an import of a service, which thus subtracts from GDP.
45.
One
should treat the MERGE model’s estimates of GDP loss under international permit trading with
caution.
That model treats purchases of permits as an import, and thus a charge against GDP. This
exaggerates
GDP loss under international permit trading, and exaggerates the difference in GDP between
the buyers’ and sellers’ markets.
41
GDP
if permits are not auctioned (see Table 5-3).
44
Those losses are roughly half as
large as they would be if permits could not be traded among Annex B countries.
Allowing
unrestricted global trading of permits would reduce GDP loss further, to
between
0.1 percent and 1.0 percent of baseline GDP. Macroeconometric models
show larger losses than general equilibrium models.
With
unrestricted permit trading among Annex B countries, estimates of the loss in
consumption
in 2010 range from 0.1 percent to 1.7 percent of baseline levels, also
about half as large
as they would be without permit trading. (G-Cubed is again the
exception;
in that model, consumption rises 0.7 percent above baseline levels).
Allowing
unrestricted global trading of permits would reduce this range to 0.1 percent
to
0.9 percent of baseline levels. For most models, the change in consumption is more
closely
correlated with the change in GDP plus payments for foreign permits than with
the
change in GDP alone. Permit purchases reflect a loss of purchasing power not
captured by GDP.
Restri
cting international trade of permits would increase the losses to GDP and
consumption,
because it would push up permit prices and energy prices. The MS-
MRT
study finds that restricting permit purchases from other countries to 30 percent
of
the difference between a country’s baseline emissions and its cap would boost the
loss
in U.S. GDP from 0.8 percent to 1.2 percent of baseline GDP in 2010.
Removing
restrictions on purchases but preventing the former Soviet bloc countries
from
selling “hot air” would boost U.S. GDP loss to 1.1 percent of baseline GDP in
2010.
According to the MERGE model, adding restrictions on U.S. purchases of
em
issions permits could erase the benefits of an expanded permit market. Exact
losses
would depend on the price that U.S. importers of permits paid foreign sellers,
a
price that could fall anywhere between the domestic U.S. price and the domestic
price
in permit-selling countries. If the price of imported permits equaled the
domestic
price in permit-selling countries—a “buyers’ market”—U.S. GDP loss (0.6
percent
of baseline GDP) would be somewhat smaller than if sellers of permits could
charge
U.S. buyers the same price they paid for domestic permits (0.7 percent of
baseline GDP)—a “sellers’ market.”
45
42
Impact on U.S. GDP and Consumption: Model Synthesis
According to a synthesis of model results (see Appendix B), the Kyoto Protocol
would reduce U.S. GDP in 2010 by 0.5 percent to 1.2 percent below its baseline level
and U.S. consumption in 2010 by 0.4 percent to 1.0 percent below baseline,
depending on how the treaty were implemented (see Table 5-4). (Incorporating
uncertainty about the price sensitivity of emissions expands these ranges to a GDP
loss of 0.4 percent to 1.5 percent and a consumption loss of 0.3 percent to 1.3
percent.) Without any international trade of emissions permits, U.S. GDP would
decline by 1.7 percent to 1.8 percent and U.S. consumption by about 1.2 percent in
2010. By contrast, U.S. GDP and consumption would each decline just 0.2 percent
below baseline levels in 2010 with unrestricted global trading of permits.
GDP loss is closely tied to permit prices through their impact on prices of energy and
other emission-producing goods and services: the larger the impact on energy prices,
the greater the GDP loss. Thus GDP loss is smallest with unrestricted global trading
of permits and largest with no international trade of permits. Within the scenarios that
are consistent with the Kyoto Protocol, restrictions on international trade of permits
raise domestic permit prices and magnify the losses to GDP and consumption. The
percentage change in consumption is smaller than the percentage change in real GDP,
for the same reasons discussed above.
In the case of no international permit trading, the model synthesis produces estimates
of GDP loss that are larger than those from most general equilibrium models, but
smaller than those from macroeconometric models. The general equilibrium models
do not consider the potential impacts of higher unemployment and lower capacity
utilization on GDP, which the synthesis includes. However, the synthesis assumes
that the adverse effects of higher unemployment and lower capacity utilization are
smaller and spread out over a longer period of time than most macroeconometric
models do.
With ideal implementation of international permit trading, the model synthesis
produces estimates of GDP loss comparable to those from general equilibrium
models, and smaller than those from macroeconometric models. On average, the
positive impact of a smaller required reduction in Annex B emissions (and thus a
lower permit price) on GDP in the synthesis estimates is offset by the impact of higher
unemployment and lower capacity utilization in those estimates.
In the case of a cartel of permit-exporting countries, the size of the GDP loss would
depend on the strategy followed by those countries. The figures in Table 5-4 assume
that the countries of the former Soviet bloc sell just enough permits to keep their own
domestic permit price at half the level of the international price. If, instead, those
countries attempted to maximize their gross national income—roughly, GDP plus
43
permit revenues—by selling only their unused permits (known as hot air), losses
would be somewhat larger. U.S. GDP would fall by 0.7 percent below baseline levels
in 2010, and U.S. consumption would fall by 0.6 percent.
With restrictions on permit purchases, losses in GDP and consumption would be
nearly the same under global trading as under Annex B trading, because permit prices
in the United States would be the same in both cases. A difference would arise only
because the price that the United States paid for foreign permits would likely be lower
under global trading, reducing the loss in U.S. income. That price could be anywhere
between the domestic permit price in the United States ($122 per metric ton of
carbon) and the domestic price in permit-exporting countries (zero), and would
depend on the relative bargaining power of buyers and sellers of permits. Since global
trading would increase the number of countries selling permits, the bargaining power
of buyers would rise, pushing down the price of imported permits and losses in GDP
and consumption.
If restrictions were placed on permit purchases and the import price of permits
equaled the domestic permit price of $122 per tonne, GDP would fall by 1.0 percent
below baseline levels and consumption would fall by 0.8 percent. On the other hand,
if the domestic permit price remained unchanged but the price of foreign permits was
zero, GDP would fall by 0.9 percent and consumption would fall by 0.6 percent. (The
difference in consumption is larger than the difference in GDP because permit imports
reduce income, which affects consumption directly but affects GDP only through
reduced saving.) The synthesis estimates for restrictions on permit purchases assume
that the price of imported permits is zero with global permit trading and halfway
between zero and $122 per tonne with Annex B trade of permits.
Among the Kyoto-consistent scenarios, losses in GDP and consumption are
somewhat higher under restrictions on permit sales than under restrictions on permit
purchases. This is at least partly due to the specific restrictions on sales and purchases
chosen. If, for example, the restriction on permit purchases were tightened so that the
United States was required to achieve at least 75 percent of its obligation to reduce
emissions domestically, then losses in GDP and consumption would be larger than
with a prohibition on sales of hot air. Similarly, a looser restriction on permit sales
would reduce losses in GDP and consumption in that case below those in the case of
restrictions on permit purchases.
Although losing offsets from emissions of greenhouse gases other than carbon dioxide
would boost permit prices significantly, it would increase losses to GDP and
consumption only slightly. This is because the total number of permits required would
fall, since permits would no longer be required for activities that produced greenhouse
gases other than carbon dioxide. Losses in GDP and consumption due to reductions
46.
If
greenhouse gases other than carbon dioxide are also capped, then the losers also include consumers of
goods and services whose production causes
such gases to be emitted, as well as the producers of those
goods and services.
44
in
carbon dioxide would rise, but losses due to reductions in other greenhouse gases
would be eliminated.
In
every scenario, if the United States received credits for carbon sequestered through
land
use changes that occurred because of the Kyoto Protocol, then losses in GDP
and
consumption would be smaller than if no such credits were given. Such credits
would
push permit prices lower than they otherwise would be, and reduce the
economic
losses from that source. Partly offsetting this gain, GDP would fall by the
value
of the agricultural products that the reforested land would have otherwise
produced, net of the value of farm inputs that would be freed for other uses.
Using
revenues from a permit auction to cut tax rates would reduce losses in output
and
consumption (see Box 5-1). Unfortunately, results from the surveyed models
were
too few and too varied for me to construct a synthesis estimate of the impact of
lower tax rates.
POTENTIAL IMPACTS ON INCOMES
The
Kyoto Protocol could have a larger impact on the incomes of permit recipients
and
of producers and consumers of energy than on the overall level of income, with
some
households benefitting and some losing. The beneficiaries would be the
recipients
of emissions rights, if such rights were allocated free of charge, or the
recipients
of tax cuts or spending increases, if the permits were auctioned and the
revenues
used to cut taxes or increase spending. The losers would be consumers who
paid
more for energy and for goods and services produced using energy, and energy
producers
whose incomes decline from lower demand for their products.
46
Many
households
would find themselves both winners as stockholders or taxpayers and
losers
as consumers or energy producers. (In addition, taxpayers would have to cover
the
higher cost of government purchases and transfers, but would benefit from higher
profits taxes if emissions rights were allocated to companies free of change.)
T
he JWS model is the only one that directly addresses the potential distributional
impact
of emissions reductions. It finds that wealthy households gain more or lose
less than poor households.
47.
Total consumption (public plus private) would rise 0.9 percent above its baseline level, a larger
percentage
than private consumption alone. In the JWS model, tax rates are exogenous, so government
spending
adjusts to hold the deficit at baseline levels. Higher GDP thus boosts government spending,
while lower GDP reduces government spending. The SGM model follows the same practice.
48.
In
addition, it appears that the JWS study may understate the rise in consumer prices, and thus the fall
in
real wages, from a rise in permit costs. The JWS study reports that permits would add $1.20 per
million
Btu (1996 dollars) to prices of refined petroleum in 2010. Starting from EIA’s baseline price for
refined
petroleum of $7.94 per million Btu (1996 dollars) in 2010, a $1.20 per million Btu carbon charge
would
boost the price of refined petroleum by 15.1 percent. (This increase is likely an underestimate of
the
percentage impact of permit costs on refined petroleum prices in the JWS study, both because the EIA
baseline
includes gasoline taxes, which are not included in the output price for the refined petroleum
industry
used by JWS, and because the real price index for refined petroleum falls between 1996 and
2010
in the JWS study but is roughly unchanged in the EIA study.) The JWS study finds that non-permit
costs of petroleum refiners fall 4.0 percent. Adding
the 15.1 percent rise in permit costs would mean a
10.5
percent rise in the price of refined petroleum. However, the JWS study finds that the price of
refined
petroleum rises just 3.8 percent. Consumer prices of items containing refined petroleum may thus
be understated.
49.
Ian W. H. Parry, Roberton C. Williams, III and Lawrence H. Goulder, “When Can Carbon Abatement
Policies
Increase Welfare? The Fundamental Role of Distorted Factor Markets,”
Journal of
Environmental Economics and Management
, vol. 37, no. 1 (1999), pp. 52-84.
45
BOX 5-1
THE ECONOMIC IMPACT OF AUCTIONING PERMITS
Economic
losses could be smaller than estimated in the previous section if permits were auctioned
and
the revenues used to cut tax rates in ways that improved incentives to work and save. However,
the
amount by which economic losses are reduced by permit auctions would depend on exactly how
auction
revenues were used. Studies disagree on the effects different methods of recycling revenues
would
have. In addition, because the policy discussion has focused on permits given away for free,
few
studies have looked at the implications of permit auctions. It is thus difficult to determine what
the effects of a particular use of auction revenues would be.
In
the absence of international permit trading, the JWS model indicates that if auction revenues were
used
to finance a cut in the corporate tax rate, the GDP loss from reducing emissions to 1990 levels
in
2010 would fall from 1.1 percent of baseline GDP with permits given away for free to 0.4 percent.
The
consumption loss would fall from 0.6 percent to 0.1 percent of the baseline. Although reductions
in
emissions would still reduce investment below baseline levels, cutting taxes on corporate income
significantly eases the impact.
The
JWS model also finds that if auction revenues were used to finance a cut in marginal tax rates
for
individuals, labor hours would jump by 1.1 percent, GDP by 0.4 percent, and private consumption
by
0.7 percent above baseline levels in 2010.
47
This result assumes that labor supply
responds
strongly to real wages and that the policy would raise the marginal after-tax wage rate.
48
In
a paper examining the impact of recycling auction revenues through a cut in the personal tax rate,
Parry,
Williams and Goulder find much smaller positive effects from recycling than the JWS model
does.
49
Although recycling still has a positive impact, it is not enough to overcome the negative
impacts
of higher energy prices. Those authors find that revenue recycling reduces the direct cost of
achieving
a 25 percent reduction in emissions (about the same size as the reduction in the JWS study)
to roughly half of what it would be if permits were not auctioned.
50.
Congressional Budget Office,
Labor Supply and Taxes
, CBO Memorandum (January 1996).
51.
Congressional
Budget Office,
Who Gains and Who Pays Under Carbon-Allowance Trading? The
Distributional Effects of Alternative Policy Designs
(June 2000).
46
BOX 5-1
(continued)
The
JWS model and the Parry, Williams and Goulder model assume that workers are more responsive
to
changes in the marginal after-tax wage than empirical work examined by CBO does.
50
On the
other
hand, most of the other models assume there is no response at all. Thus, most models would
find a much smaller beneficial impact from a cut in personal tax rates than these two models.
The
EIA study finds that using auction revenues to reduce Social Security tax rates of both employers
and
employees would cut GDP losses by almost half. Reducing employer-paid Social Security tax
rates would cut labor costs
and thus prices, which in turn would offset about half of the increase in
consumer
prices from higher energy prices. With lower inflation, the Federal Reserve would not have
to tighten as much and unemployment would not rise as much. In
addition, a lower Social Security
tax
for workers would raise their after-tax real wage, causing them to increase their labor supply.
The
net result is that losses in GDP and consumption would be much smaller than if the permits were
given away for free.
Using
the same macroeconomic model as EIA, DRI finds that, in the absence of international permit
trading,
the gain from recycling permit revenues would be smaller if the revenues were used to
increase
the federal surplus than if they were used to cut marginal tax rates. The positive effects from
increasing the surplus stem from the boost to
national saving. However, national saving would not
improve
by much. While federal government savings would rise, business saving (profits) would fall
because
businesses would no longer receive permits free of charge. Without international trade of
permits,
DRI estimates that losses in both GDP and consumption in 2010 would only be 0.2 percent
smaller than if permits were given away for free.
Only two studies
examine the implications of permit auctions with international trading of permits.
The
EIA study finds that using the auction revenues to cut the Social Security tax rate would reduce
losses in GDP and consumption by roughly
half in scenarios that correspond to Annex B trading of
permits
and global trading of permits. Parry, Williams and Goulder find that auctioning permits and
using
the revenues to cut personal tax rates would reduce the direct cost of achieving a 15 percent
reduction
in U.S. emissions by just over 50 percent of what it would be if permits were not auctioned.
End of Box
An
analysis of which households gain and which households lose and how much they
gain
or lose is beyond the scope of this paper. (CBO has analyzed that issue in
another study.
51
) However, it is easy
to estimate how much could be redistributed.
The
total amount gained is just the value of permits issued to U.S. households,
directly
and indirectly, which is the volume of permits issued (the emissions cap)
multiplied
by their price. (With restrictions on permit purchases from abroad, the
value
of import quotas would also be part of the gains.) The amount lost is the value
of
permits issued plus the value of permits purchased from abroad, less losses by
47
foreign energy producers. In every study, the value of permits issued and used is
larger than the change in GDP. Consumers, rather than energy producers, bear the
lion’s share of the losses. However, consumer losses are more evenly distributed
across the population than producer losses are.
Winners: The Value of Permits Allocated
The studies indicate that the value of permits issued in the United States would be
large (see Table 5-5). With unrestricted trading of permits among Annex B countries,
the models find that the value of permits issued would be between $32 billion and
$281 billion (in 1997 dollars) in 2010. Eliminating international trade of permits
would boost permit prices, raising the value of permits issued to $114 billion to $524
billion. Global trading of permits would reduce permit prices, reducing the value of
permits issued in the United States to $17 billion to $105 billion. Those amounts
would go to recipients of the permits if permits are distributed free of charge, or to
the recipients of tax cuts or spending increases if the permits are auctioned and the
proceeds used to cut taxes or increase spending.
Using the synthesis model, I estimate that the value of permits allocated to the United
States under the Kyoto Protocol would range from $86 billion to $223 billion (in
1997 dollars) in 2010, or 0.7 to 1.8 percent of GDP, depending on how the protocol
was interpreted and implemented (see Table 5-6). The lower estimate is almost 7
percent as large as CBO’s January 2001 projection of revenues from individual
income taxes in 2010, and 36 percent as large as CBO’s projection of revenues from
corporate income taxes. The upper estimate is 17 percent as large as CBO’s
projection revenues from individual income taxes in 2010, and 94 percent as large as
CBO’s projection of revenues from corporate income taxes. Thus, with restrictions
on permit sales by the former Soviet bloc and no offsets from reductions in other
greenhouse gases, receipts from a permit auction would be large enough to finance
a 17 percent reduction in tax rates on individual income.
If a limit were placed on permit purchases, a quota system would be needed to hold
permit purchases below the limit. In that case, the total value of permits issued would
include the value of emissions permits and the value of import permits (quotas). The
value of each quota would equal the difference between the U.S. permit price and the
price charged by the foreign seller. The latter price could be anywhere between zero
and the U.S. price of emissions permits, putting the value of quotas between zero and
$25 billion (in 1997 dollars), according to a synthesis of model results. (That estimate
assumes that the United States would be constrained to achieve at least 65 percent of
its Kyoto obligation with cuts in domestic emissions.)
52.
This
includes only lower margins of energy producers, e.g., lower wage rates for coal miners who retain
their jobs and lower profit
margins for oil producers. It does not include reductions in incomes of labor
and
capital no longer employed in the energy industry, such as laid-off coal miners or abandoned coal
mines. These are not redistributed to anyone, but are simply lost.
48
Without
international trading of permits, the value of permits would rise to $331
billion (1997 dollars)
in 2010, or 2.7 permit of GDP, 25 percent as large as CBO’s
January 2001 projection of revenues from individual
income taxes in 2010, and 39
percent
larger than CBO’s projection of revenues from corporate income taxes. If the
Kyoto
Protocol were amended to include all countries, however, the value of permits
would
be much smaller. The value of permits would be smaller than under the
existing
Kyoto Protocol, as long as significant limits were not put on U.S. purchases
of foreign permits.
Losers: The Value of Permits Used
These
gains would come at the expense of energy producers and households
consuming
goods and services whose production or use causes greenhouse gases to
be
emitted. Consumers would suffer most of these losses. Under the Kyoto Protocol,
higher
prices for U.S. consumers would account for between 94 percent and 96
percent
of the value of permits used, and income losses by energy producers would
account for the remaining amounts.
52
Moreover, if restrictions on permit trading were
imposed,
permit prices would rise proportionately more than crude oil prices fall, so
that
the consumer share of the total loss would also rise. Thus, eliminating
international
trade of permits would boost consumers’ share of the loss to 96 percent
of the value of permits used, while unrestricted global trading of permits would
reduce it to 90 percent.
Foreign producers would absorb more than half
of producers’ share of losses. The
biggest
source of loss for producers would be lower oil prices. (The United States
is
projected to import more than two thirds of the crude oil it uses in 2010.) Both
domestic and foreign refiners would have excess capacity and would be forced to trim
their
margins. The price of coal would also fall, resulting in lower incomes for miners
and
mine owners. On the other side, if usage of natural gas rose, natural gas prices
would
rise, raising incomes of natural gas producers. Most natural gas used in the
United
States is also produced here, so this factor would primarily help raise U.S.
incomes.
With
international trade of permits, wealth would be transferred from the United
States
to other countries, on net. That is, the value of permits purchased from abroad
would
greatly exceed the drop in foreign income from lower prices for fossil fuels
consumed
in the United States. Without international trade of permits, payments for
foreign
permits would disappear, but there would be a small net inflow from foreign
53.
The
issue of net transfers is complicated by foreign ownership of stock in U.S. companies that receive
permits and domestic ownership of stock in foreign companies that receive permits.
49
crude
oil producers to U.S. consumers of petroleum products. As noted above,
however,
the transfer from domestic energy consumers to permit recipients would
grow significantly.
53
50
APPENDIX A
THE PRICE SENSITIVITY OF EMISSIONS AND THE PRICE OF
CARBON-ENERGY
The price sensitivity of carbon emissions measures how carbon emissions change in
response to changes in the price of carbon-energy due to emissions charges. The
price of carbon-energy is the price that users of energy generated from coal, oil and
natural gas pay per metric ton of carbon embodied in these fuels and emitted when
they are burned. This appendix defines these concepts in more detail, shows how they
are used in this study and how they were calculated from the models’ results.
Price Sensitivity of Carbon Emissions
The price sensitivity of carbon emissions equals the logarithm change in carbon
intensity (carbon emissions divided by real GDP), divided by the logarithm of one plus
the ratio of the permit price to the baseline price of carbon-energy. This is equivalent
to dividing the logarithm change in carbon emissions minus the logarithm change in
GDP by the logarithm of one plus the ratio of the permit price to the baseline price
of carbon-energy. Mathematically,
s
E
GDP
E
GDP
T
P
E
E
GDP
GDP
T
P
final
final
baseline
baseline
final
baseline
final
baseline
=
−
+
=
−
+
ln
ln
ln
ln
ln
ln
,
1
1
where s is price sensitivity, E is carbon emissions, T is the permit price, and P is the
baseline price of carbon-energy. Price sensitivity is a negative number.
Logarithms are preferred to percentage changes because the percentage increase in
the price of carbon-energy needed to reduce emissions by a given percentage amount
rises as the level of emissions falls. That is, doubling the permit price leads to less
than a doubling of the amount of carbon emissions mitigated.
If one knows price sensitivity, one can rearrange the above formula to determine the
level of emissions produced at a given permit price:
.
E
E
GDP
GDP
T
P
final baseline
final
baseline
s
=×
×+
1
Alternatively, the same formula can be rearranged to show what permit price is
required to achieve a given reduction in emissions:
51
TP
E
E
GDP
GDP
final
baseline
final
baseline
s
=×
−
1
1
.
The definition of price sensitivity assumes that, all else equal at a given time,
emissions are proportional to real GDP. That is, an extra one percent of GDP will
push emissions up one percent. Most economic models make assumptions about
energy usage that guarantee a similar result. For the models that do not report
GDP—EPPA, SGM and WorldScan—I used direct cost as an estimate of the loss in
real GDP in calculating price sensitivity. Direct cost almost certainly understates the
actual loss in GDP, and thus leads to an overstatement of price sensitivity in these
models. That is, the less of the reduction in emissions explained by lower GDP, the
more explained by higher energy prices.
The price sensitivity of carbon emissions is similar in many ways to a demand
elasticity for carbon-energy, but there are two important differences. First, the price
sensitivity of emissions captures both changes in demand for carbon-energy and
substitutions between fuels with different carbon contents. Second, price sensitivity
measures the response of emissions to the permit price, rather than to changes in the
price of carbon-energy, and thus includes supply effects. Differences in supply prices
(the price of energy excluding permit costs) cause differences between the permit
price and the change in the price of carbon-energy. For example, if lower demand
causes the price of crude oil to fall, the change in the price of carbon-energy will be
smaller than the permit price. The first factor will make the price sensitivity of carbon
emissions larger than the elasticity of demand for carbon-energy, while the second
factor will partially offset this effect.
The Baseline Price of Carbon-Energy
The price of carbon-energy is calculated from the prices paid by the end users of
energy. Except for the case of fuel-switching by electric utilities, the amount of
emissions produced is determined by these end users. For example, motorists base
their decisions on the type of car to buy and how much to drive it on the price they
pay at the pump. The price of crude oil will affect their consumption of gasoline, and
the emissions from it, only to the extent that it affects the retail price of gasoline.
Similarly, consumers of coal-based electricity will base their usage of electricity on the
price of that electricity, not on the price of the coal used to make it, except to the
extent that the coal price affects the electricity price.
The baseline price of carbon-energy must be aggregated from the prices of many
individual energy products, such as gasoline, diesel fuel, jet fuel, heating oil, delivered
natural gas, and electricity, each with their own price per metric ton of emissions.
52
Since emissions data are only available by fossil fuel type and by broad industrial
category from any of the models, the first step in the aggregation is to create a price
of carbon-energy for each fuel—coal, natural gas and petroleum. This is done by
dividing total dollars spent on final energy from each fuel by emissions from that fuel.
The second step aggregates prices of energy from each fuel into a single price of
carbon-energy that can be used in calculating the price sensitivity of emissions. For
this purpose, I assume that the price sensitivity of each fuel, excluding substitution
between fuels, is equal, and that the effect of substitution between fuels on emissions
can be captured mathematically by an expression that relates the permit price to the
overall price of carbon-energy. That is,
E
GDP
T
P
E
GDP
T
P
E
GDP
T
P
E
GDP
T
P
z
E
GDP
T
P
s
c
c
r
n
n
r
p
p
r
x
1
1
1
1
1
1
+
=+
++
++
++
−
,
where r is the price sensitivity of emissions of each fuel, excluding the effects of
substitution between fuels, the subscripts c, n and p denote coal, natural gas and
petroleum, respectively, and E
i
and P
i
denote baseline emissions from fuel i and the
baseline price of energy from fuel i. The parameter z is a rough approximation of the
total share of emissions that can be eliminated by fuel substitution, and the parameter
x governs how this substitution relates to the permit price. Each side of the equation
is an expression for economy-wide emissions intensity (the ratio of emissions to
GDP).
Because the above equation is nonlinear, there is no single P that, given the prices for
each fuel, will satisfy this equation exactly for all permit prices T. For the sake of
simplicity, this study defines the equation to hold around a permit price of zero. In
other words, starting from a permit price of zero, the baseline price of carbon-energy
is defined so that the left- and right-hand sides of the equation both produce the same
ratio of emissions to GDP for small increases in the permit price. Taking derivatives
of both sides of the above equation with respect to T, and then setting T to zero yields
the following equation:
()
sxz
E
GDP P
r
E
GDP P
r
E
GDP P
r
E
GDP P
c
c
n
n
p
p
−=++
1
1
1
1
.
The reduction in overall emissions intensity (from s) less the reduction in emissions
intensity from substitution effects (from xz) equals the reduction in emissions intensity
from higher costs of using each type of fuel (the right-hand side of the equation).
An expression for P can be obtained by solving this equation for P and setting s equal
to r plus xz, i.e., s-xz=r. Thus,
P
E
E
P
E
E
P
E
E
P
c
c
n
n
p
p
=++
−− −
−
11 1
1
,
53
a CES (constant elasticity of substitution) weighting of energy prices, with the
substitution parameter equal to 1. The baseline price for carbon-energy is the
reciprocal of a weighted average of the reciprocals of the prices of energy from each
fuel.
The same formula is used to aggregate baseline energy prices across countries. In this
case, the subscripts refer to countries instead of fuels. The baseline price is the
reciprocal of a weighted average of the reciprocals of the prices of energy from each
country. This aggregation assumes that the price sensitivity in each country being
aggregated is the same.
In theory, there are two reasons the results could be distorted, though any distortion
seems likely to be small. First, as the permit price rises and emissions fall, the relative
weight of the higher-cost fuels rises, because reductions in the lower-cost fuels are
disproportionately large. Second, as the permit price rises, opportunities for further
reductions in emissions from fuel-switching disappear faster than opportunities for
further reductions in final use of carbon-energy. Both of these should tend to cause
the estimated price sensitivity of emissions to decline as permit prices rise, and
consequently the price sensitivity should be lowest when no permit trading is allowed.
However, the models give mixed results: roughly one-third find this expected
relationship, another third find that trading lowers price sensitivity, and another third
find that trading has a negligible impact on price sensitivity (less than a 5 percent
change).
Constructing the Baseline Price of Carbon-Energy for the United States.
The price
of carbon-energy is the price that users of energy generated from coal, oil and natural
gas pay per metric ton of carbon emissions. For each fuel, I calculate this price by
dividing the market value of the energy produced by that fuel by the carbon emissions
generated by that fuel. For the non-electricity portion of the energy from each fuel,
the market value is the amount of energy delivered to end-users times the average
sales price of that energy. (For petroleum, I exclude the value of non-energy products
such as asphalt, most plastics and motor oil, since these do not produce carbon
emissions.) For the electricity portion, the market value for a given fuel is that fuel’s
share of the total inputs to electricity generation times the total market value of
electricity. This implicitly assumes that electricity from every source is sold at the
same price.
Several studies provide all or most of the data needed for these calculations. The EIA
study and editions of the EIA’s Annual Energy Outlook used by SGM-Admin and
SGM-PNNL contain all the requisite data. The DRI and WEFA studies contain most
of the data needed. Missing data on shares of each type of fuel going to various end
uses were filled in using data from the EIA study.
54.
One
study projects lower consumption of each fossil fuel in 2010 than the EIA study, but at the same time
projects higher carbon emissions.
55.
In
the EPPA, price data are in units of efficiency labor. I assume that the price of efficiency labor rises
at the same rate as the GDP price index.
56.
International
Energy Agency,
Energy Prices & Taxes: Quarterly Statistics
,
no. 1 (1999). Countries
covered
include the United States, OECD Europe (including the Czech Republic, Hungary and Poland),
Australia,
Canada, China, India, Indonesia, Japan, Kazakhstan, South Korea, Mexico, New Zealand,
Romania, Russia, Slovakia, South Africa, Taiwan, Thailand, Turkey and Venezuela.
54
Several
other studies—those using the AIM, EPPA, G-Cubed, GTEM, JWS, MS-
MRT and Oxford models—provide price indexes for electricity, refined petroleum (or
gasoline)
and natural gas, but incomplete or inconsistent data for quantities of fuels.
54
These
prices are converted to prices for final users by assuming that real distribution
costs
and indirect taxes (most important in the case of refined petroleum) are constant
in
real terms.
55
Emissions and quantities of energy used are taken from EIA’s Annual
Energy
Outlook 1998 (AEO98), which contains data very similar to that used in the
EIA
study. The estimated price of carbon-energy is not very sensitive to the choice
of data source for emissions and energy use.
One
model, RICE, contains its own measure of the price of carbon-energy. This
equals
the wholesale price of carbon-energy, which is assumed to be constant across
all
regions, plus a markup over the wholesale price, which varies by region. Historical
values
of the price of carbon-energy for the United States are somewhat larger than
I calculate, while historical values for some developing countries are smaller.
Data
for energy prices were not available for the studies using the CETA, MERGE
and WorldScan models. For these general equilibrium models, I use the average price
of
carbon-energy from the eight general equilibrium models for which a price of
carbon-energy
can be calculated. (RICE is not included in this average, since its
historical values are inconsistent with those assumed for the other models.)
Con
structing Baseline Prices of Carbon-Energy for Other Countries.
None of the
studies
contain energy forecasts for other countries with the same detail as that used
to
construct the baseline price of carbon-energy for the United States. To construct
forecasts
of the baseline price of carbon-energy for other countries, I essentially
calculate
historical differences in prices of carbon-energy between the United States
and other countries, and then add these to the forecast for U.S. prices.
T
he International Energy Agency provides price data for electricity to households and
industry,
regular and premium gasoline, light fuel oil for industry, commercial and
non-commercial
diesel fuel, high sulphur fuel oil to industry, and natural gas to
industry
and households for several foreign countries.
56
Data for 1996, denominated
in
foreign currencies, are converted to U.S. dollar values using 1996 exchange rates.
57.
The
value added tax (VAT) would be applied to the permit component of final energy prices, just as it
is to the
non-permit (baseline) components of energy prices. To make the baseline price comparable to
the permit price, one therefore must remove VAT from both.
58.
The
Czech Republic, Hungary, Poland, Romania and Slovakia are used for unavailable eastern European
countries,
Kazakhstan and Russia are used for unavailable countries of the former Soviet Union,
Indonesia,
Thailand and Taiwan are used for missing Asian countries, excluding the Middle East, Brazil
is
used for missing Central and South American countries, South Africa is used for the rest of Africa,
excluding
the Middle East, and the oil exporters Indonesia, Mexico and Venezuela are used for the
Middle East, excluding Turkey.
55
Value
added tax is removed from prices subject to it.
57
For each country, prices per
Btu
of fossil fuel for various end uses of that fuel are weighted together using EIA
data
on end use consumption by country, or using U.S. weights where country data
are
missing. For countries for which price data is unavailable, price data from similar
countries
or countries in the same region are used.
58
These steps produce estimates
of the price of energy per Btu of fossil fuel by fuel and by country for 1996.
To
obtain the price of carbon-energy by fuel and country, I then multiply these
estimates
by the U.S. ratio of the price of energy per metric ton of carbon to the price
of
energy per Btu, for each fuel. These estimates of the price of carbon-energy for
each fuel by country are then
aggregated to estimates of the price of carbon-energy
for
each fuel for broader regions using the same formula used to aggregate the price
of carbon-energy across fuels.
To
obtain model-specific forecasts of the baseline price of carbon-energy in 2010, it
is
assumed that the difference between foreign and U.S. prices of carbon-energy, by
fuel,
are the same in 2010 as in 1996. In other words, the differences obtained above
for
1996 are added to each model’s 2010 estimate for the price of carbon-energy, by
fuel.
(In cases where differences lead to unrealistically low prices in China and the
former
Soviet Union, a multiplicative adjustment is made.) These prices are then
aggregated
across fuels using the CES aggregation formula developed above, to
obtain
estimates of the baseline price of carbon-energy for Annex B and the non-
Annex B countries. Finally, these estimates are converted to 1997 dollars.
59.
Energy
Information Administration,
International Energy Outlook 2000
,
DOE/EIA-0484 (March 2000).
60.
The EPPA estimates are from John Reilly,
Ronald G. Prinn, Jochen Harnisch, Jean Fitzmaurice, Henry
D.
Jacoby, David Klicklighter, Peter H. Stone, Andrei P. Sokolov and Chien Wang,
Multi-Gas
Assessment of the Kyoto Protocol
,
Report Series No. 45 (Cambridge, MA: MIT, Joint Program on the
Science and Policy of Global Change, January 1999).
56
APPENDIX B
ORIGINS OF MODEL SYNTHESIS ESTIMATES
This
appendix shows how the various estimates of the effects of emissions reductions
on
the U.S. economy were synthesized into a single estimate using a simple reduced-
form
model. The guiding principle of this model is to base its properties on the
properties
of the models wherever possible. In some cases, that principle meant
making
adjustments to models that made unrealistic assumptions. In other cases, it
meant
using only the subset of models that examined a particular issue, such as
emissions
leakage. In cases where modelers use other sources, it meant using updated
projections from those sources.
Emissions Baselines and Caps
Few
modelers claim special expertise in forecasting baseline emissions of carbon
dioxide.
Instead, many of them base their estimates on work done by the EIA. This
paper
uses the projections of carbon dioxide emissions from EIA’s March 2000
report.
59
For consistency, I used estimates of carbon dioxide caps for the Kyoto
protocol
from the same publication. Because EIA’s projections of emissions for many
countries
were lower in 2000 than they had been when modelers prepared their
analysis
in 1998, smaller percentage reductions in emissions are required in the model
synthesis
estimates than in most of the studies. Since 2000, however, EIA has revised
its emissions projections up, pushing required reductions in emissions higher again.
Unfortunately,
EIA does not project emissions of the five other greenhouse gases:
methane,
nitrous oxide, perfluorocarbons, hydrofluorocarbons, and sulfur
hexaflouride.
However,
studies from three modelers—EPPA, SGM-Administration
and SGM-PNNL—do.
60
For Annex B countries, I used the average emissions
baselines
and Kyoto caps for other gases from those three studies. None of the
studies
provided estimates of other greenhouse gases for non-Annex B countries, but
instead
assumed that any extension of the Kyoto Protocol to include non-Annex B
countries
would exempt emissions of these gases in those countries. I made the same
assumption.
61.
United
Nations Framework Convention on Climate Change,
Greenhouse Gas Inventory Database
,
2000
(available at www.unfccc.de).
57
Forest
Growth.
Baseline estimates of greenhouse gases sequestered by cropland,
grazing
land and managed forests are based on country submissions to the United
Nations
Framework Convention on Climate Change.
61
The estimate of credits for
carbon
sinks in 2010 is the amount of carbon sequestered in the most recent year for
which
an estimate is available in that document. For most Annex B countries, this is
1998.
GDP
and Emissions.
The model synthesis uses CBO’s
January 2000 projection of
real
GDP for the United States, and EIA projections for other countries. Although
CBO’s projection
for U.S. GDP in 2010 is nearly 9 percent higher than EIA’s, it is
difficult
to determine whether CBO’s higher GDP would imply higher emissions than
EIA’s.
For example, if the extra output in CBO’s projection stems from higher
investment in computers and semiconductors, emissions would probably be similar to
EIA’s.
Consequently, I simply used EIA’s emissions projection without adjustments.
A
higher emissions projection would mean higher permit prices and greater losses in
GDP
and consumption, but a greater environmental benefit. The baseline price of
carbon-energy was derived as described in Appendix A.
Price Sensitivity of Carbon Emissions
Every
study surveyed in this paper estimates what permit prices would be in the
United
States without international trade of permits. Thus, this case provides a useful
benchmark.
I use estimates of price sensitivity from that benchmark to develop
estimates of price sensitivity for the other trading scenarios and for other countries.
U.S.
Price Sensitivity, with no International Trade of Permits.
To develop an estimate
of
the price sensitivity of U.S. carbon emissions, I adjusted the price sensitivities of
each
model for known problems (if possible), and then took a geometric average of
the
resulting price sensitivities (see Table B-1). My adjustments attempt to deal with
four types of problems found in some of the models:
•
theoreti
cal assumptions about responses to energy prices contradicted by the
empirical evidence;
•
reductions
in carbon dioxide emissions that exceed the reduction in the consumption
of carbon-energy;
•
unrealistically
large or small implied increases in nuclear and hydroelectric power;
and
•
responses
to the expected rise in energy prices that either begin too late or are
completed too quickly.
62.
DRI
uses a Cobb-Douglas production function to determine the impact of labor, capital and energy on
potential
GDP, but not to determine the demand for energy. While this is internally inconsistent, it
allows energy demand to be modeled consistently with the empirical evidence.
58
Another
possible factor that artificially inflates price sensitivity in at least one model,
but
is not explored here, is the treatment of lower coal exports as a reduction in U.S.
emissions. Coal exports do not count as emissions under the Kyoto Protocol.
The
G-Cubed, RICE and WorldScan models assume that the elasticity of energy
demand
is 1.0; that is, each one percent rise in energy prices produces a one percent
fall
in demand. This estimate of the elasticity of energy demand is much higher than
that
in the empirical literature. In G-Cubed, this assumption affects only final demand,
through
a Cobb-Douglas utility function. The RICE and WorldScan models use
Cobb-Douglas
production functions to determine energy demand throughout the
economy.
62
Since price sensitivities in those two models depend entirely on this
assu
mption, they are excluded from the synthesis calculations. G-Cubed uses
estimat
ed elasticities in industry production functions, and so provides some
empirically-based
information, but the adjustment required for the demand elasticities
is large.
Severa
l models assume that the percentage reduction in emissions can exceed the
percentage
reduction in carbon-energy. This assumption can take different forms.
For example, some models assume that refiners can produce the same amount of
gasoline
by using more labor and capital and less crude oil. (If a model assumes that
emissions
are proportional to sales of refined products rather than crude oil, then
unrealistic
assumptions about crude oil usage do not affect estimated emissions or
price
sensitivity.) In the AIM and SGM models, such substitution can take place in
both
the petroleum refining and natural gas utility industries. In EPPA, the amount
of
substitution is greater, but it takes place only in the natural gas utility industry,
which is lumped together with the electricity industry.
E
missions reductions can also exceed reductions in consumption of fossil fuels in
models
in which emissions from fossil fuels are assumed proportional to total sales of
refined
products, including intra-industry sales. The problem in these models is that
intra-industry
sales frequently do not generate emissions. For example, when gas
util
ities sell less gas to each other or coal mines buy fewer services from mining
service
companies, the total amount of emissions does not change. In the G-Cubed
and
JWS models, however, reductions in such sales are assumed to reduce emissions.
Those
models assume that firms cut back on intra-industry sales disproportionately
as
the permit price rises, which artificially reduces emissions. For example, in the G-
Cubed
model, total sales of coal fall by 45 percent in 2010 without international trade
of
permits, while non intra-industry sales of coal fall by just 40 percent. Thus, coal
59
emissions are estimated to fall by 45 percent, even though net coal usage declines only
40 percent.
Any model that does not separate electricity by fuel source may inadvertently assume
unrealistically large increases in electricity from non-fossil sources. In making my
adjustments, I make the judgment that any increases in electricity from nuclear and
renewable sources (e.g., wind and biomass) that are 50 percent larger than those
found by EIA at a similar permit price in 2020 are unrealistically large. Instead, I
assume that the additional electricity would be generated by natural gas instead.
(Assuming generation by coal would increase the adjustment.) Such adjustments are
required for EPPA, G-Cubed, and JWS. Although the MS-MRT model does not
break out electricity by fuel source, it does not appear to imply unrealistically large
increases in electricity from non-fossil sources.
Other models assume that nuclear energy remains at baseline levels when the costs of
using fossil fuels rise, an assumption that seems somewhat unrealistic given that the
lifetime of nuclear plants can be extended. To address this issue, I made small
adjustments to estimates of price sensitivity from the DRI, GTEM, Oxford and WEFA
models, by assuming the same increase in nuclear generation over baseline levels as
in the EIA model.
The CETA, JWS, RICE and WorldScan models all assume that capital and labor can
be adjusted immediately at no cost. In other words, businesses can change the energy
efficiency of existing equipment at no cost, and can transform coal mines and mining
equipment into nuclear power plants at no cost. Although such adjustments are nearly
costless in the long run (when existing equipment has depreciated and decisions about
new investment have to be made), they are not costless in the short run. I adjusted
the estimates of price sensitivity in these models by the average ratio of price
sensitivity in 2010 to that in 2020 for the other models (see Table 3-2). (This
adjustment is conservative, since many models assume adjustments continue after
2020.) Such ratios range between 0.48 in the EPPA model, in which adjustment takes
place most slowly, to 1.00 in the Oxford model, in which adjustments are not
immediate but are nonetheless completed by 2010.
In its original study, EIA assumes that households and businesses outside the electric
generating sector do not begin to respond to the prospect of higher energy prices until
2005. In a later study, EIA assumes those responses begin in 2000. For EIA, I use
the estimate of price sensitivity derived from the latter study.
I use the geometric mean of the adjusted estimates of price sensitivity from all but the
RICE and WorldScan models to develop an estimate for the model synthesis. That
estimate is -0.536, which is lower than the geometric mean of the unadjusted
estimates because most of the adjustments made to estimates from the studies reduce
63.
Ca
rol Dahl, “A Survey of Energy Demand Elasticities in Support of the Development of the NEMS”
(working
paper, Colorado School of Mines, October 1993). In using the estimates from this survey, I set
negative
cross-price elasticities to zero, and reduced large positive cross-price elasticities so that they
were equal in absolute value to the corresponding own-price elasticities.
60
price
sensitivity. To avoid counting the SGM model twice, results from the studies
by
the Administration and Battelle PNNL are averaged together. Although the
CETA,
MERGE and MS-MRT models share many parameters, the adjusted price
sensitivities
in these models are close enough to the model synthesis estimate that
averaging
these models would have little impact. DRI and EIA use the same model
for GDP results, but they use different models for energy demand, and thus are
treated separately.
U.S.
Price Sensitivity, with International Trade of Permits.
I use the same price
sensitivity
for the United States with permit trading as without permit trading.
Although
theory suggests that price sensitivity might be higher at lower permit prices
(see
Appendix A), the models provide little support for this proposition. On average,
price
sensitivity is 4 percent higher with Annex B trade of permits but 12 percent
lower
with global trade of permits than without international trade of permits. (These
figures
are not comparable, since fewer models examine global trade of permits than
Annex
B trade of permits.) One way to interpret this observation is that fuel
substitution
becomes more economical above a threshold permit price (somewhere
around
$100 per metric ton of carbon), but that opportunities for fuel substitution are
slowly
exhausted at higher prices. Since the variations in price sensitivity are small,
I use a single estimate for all scenarios.
Empirical
Estimates of the Price Sensitivity of Carbon Emissions in the United States.
The
estimate of price sensitivity thus derived from the models agrees closely with the
available
empirical evidence. In 1993, Carol Dahl surveyed estimates of energy
demand
elasticities from more than 400 studies, providing summary estimates of
short-run
and long-run own-price and cross-price elasticities for coal, oil, natural gas
and
electricity in the residential, commercial, industrial and transportation sectors.
63
I
obtained medium-run elasticities by averaging the long-run and short-run elasticities.
These
elasticities were then combined with changes in energy prices from the EIA
scenarios
to determine what emissions would be if EIA’s NEMS model used these
elas
ticities instead of its own. (Fuel substitution in electricity generation was assumed
to occur at the same
rates as in EIA’s early start study.) Using this emissions data,
price
sensitivities could then be calculated for the various scenarios consistent with
Dahl’s summary estimates of price elasticities.
The
estimates of price sensitivity derived from this exercise are surprisingly close to
those
obtained from the models. Price sensitivity averages -0.53 over the EIA
scenarios,
about the same as the adjusted model average. As in the models, price
sensitivity
roughly equals this value with no international trade of permits (a permit
64.
These
figures do not include results from the study using the G-Cubed model. For the former Soviet bloc,
that
study uses results from the SGM model, which are not consistent with those that would be obtained
using the G-Cubed model for this region.
61
price
of about $350 per metric ton of carbon), is somewhat lower (-0.49) with global
trade of permits
(a permit price of $70 per metric ton), and is slightly higher (-0.55
to -0.56) in intermediate cases (permit prices of $130 to $300 per metric ton).
Those
price sensitivity estimates assume that energy users begin to anticipate higher
future
energy prices in 2000 or 2001. If energy users began to anticipate higher
energy
prices at a later date, energy users would have less time to respond, and
demand
elasticities would be closer to the short-run estimates, pushing price
sensitivity
lower. On the other hand, if caps applied to 2020 instead of 2010, demand
elasticities
would be closer to the long-run estimates, pushing price sensitivity higher.
The
Price Sensitivity of Carbon Emissions in Other Countries.
The models disagree
on
whether price sensitivity in other Annex B countries would be lower or higher than
in
the United States. Most models assume price sensitivity would be lower in other
countries,
but some models, most notably GTEM and WorldScan, assume price
sensitivity
would be much higher in other countries than in the United States. Thus,
while
the median ratio of price sensitivity in overall Annex B to that in the United
States
is 0.89, GTEM and WorldScan push the geometric average ratio up to 1.02.
64
The
geometric average of the ratio is 0.95 if one excludes those two models and the
two
models with the lowest ratios of price sensitivity in overall Annex B to that in the
United
States (RICE and SGM). That is the assumption used in the model synthesis.
The
models also disagree about the price sensitivity in countries outside of Annex B.
Some models (AIM, CETA and MERGE) assume price
sensitivity would be much
lower
among non-Annex B countries than within Annex B, while SGM assumes it
would
be much higher. Other models looking at non-Annex B emissions with global
trade
of permits (G-Cubed, MS-MRT and RICE) assume price sensitivity outside
Annex
B would be no more than 11 percent above or below its value inside Annex B.
On
average, the models find price sensitivity somewhat lower outside Annex B than
inside it.
I
n calculating model synthesis estimates, I assume that price sensitivity in both overall
Annex
B and in the non-Annex B countries is 95 percent as large as in the United
States.
In addition to the model evidence, theory suggests two reasons why price
sensitivity
should be lower outside the United States than inside it. First, because of
their
greater dependence on nuclear power, Japan and Europe have fewer
opportunities
for fuel substitution in the electricity industry. Second, energy users in
the
former Soviet bloc and the non-Annex B countries may not respond as readily to
65.
John
Reilly, Ronald G. Prinn, Jochen Harnisch, Jean Fitzmaurice, Henry D. Jacoby, David Kicklighter,
Peter
H. Stone, Andrei P. Sokolov and Chien Wang,
Multi-Gas Assessment of the Kyoto Protocol
Report
No. 45 (Cambridge, MA: MIT, Joint Program on the Science and Policy of Global Change, January
1999).
See also John Reilly, Monika Mayer and Jochen Harnisch,
Multiple Gas Control Under the Kyoto
Agreement
Report No. 58 (Cambridge, MA: MIT, Joint Program on the Science and Policy of Global
Change, March 2000).
66.
Interlaboratory
Working Group on Energy-Efficient and Low-Carbon Technologies,
Scenarios of U.S.
Carbon Reductions: Potential Impacts of Energy-Efficient and Low Carbon Technologies by 2010 and
Beyond
(Oak Ridge National Laboratory, Lawrence Berkeley National Laboratory, Pacific Northwest
National
Laboratory, National Renewable Energy Laboratory, and Argonne National Laboratory,
62
changes
in energy prices as energy users in the United States, who have a long
experience with free markets.
Price Sensitivity of Emissions of Other Greenhouse Gases
While
many economists have implicitly examined how carbon dioxide emissions
respond
to changes in prices of carbon-energy, few if any have studied how emissions
of
other greenhouse gases respond to changes in their price. The economic models
surveyed
in this paper take two approaches: either they assume that emissions of
other
greenhouse gases respond the same as emissions of carbon dioxide (SGM-
PNNL),
or they use technology-based estimates of how much emitters theoretically
could
reduce emissions at various prices (SGM-Administration and a recent study
using
EPPA
65
). Neither of these approaches seems appropriate. The first option
makes
no use of available information from technology-based studies; the second
op
tion treats the information from the technology-based studies as consistent with
information on carbon emissions from economic studies.
This
paper takes a different approach: I derive a synthesis estimate of the price
sensitivity
of emissions of other greenhouse gases by adjusting the price sensitivity of
carbon
emissions obtained from the economic studies by the ratio between estimates
of
price sensitivities of carbon dioxide and methane obtained from two representative
technology-based
studies. (Methane is the most important of the other greenhouse
gases.)
In other words, the percentage reduction in methane in one technology-based
study
is higher than the percentage reduction in carbon dioxide in another technology-
based
study, so I assume that the price sensitivity of other greenhouse gases is larger
than
the price sensitivity of carbon dioxide by a similar amount. This procedure
assumes
that the excess of price sensitivity in technology-based models relative to that
in economic models is the same for all greenhouse gases.
A
technology-based study of carbon emissions by five scientific laboratories finds that
a
permit price of $50 per metric ton of carbon equivalent produces a 23 percent
reduction in emissions.
66
An EPA technology-based study of methane emissions finds
September 1997).
67.
Environmental
Protection Agency, Office of Air and Radiation,
U.S. Methane Emissions 1990-2020:
Inventories, Projections, and Opportunities for Reductions
(September 1999). That study looks at
reductions
in methane that would be profitable with a permit price of $50 per metric ton of carbon
equivalent
on methane, but no price charged on the carbon dioxide emitted when the methane is burned.
Under the Kyoto Protocol, such
a charge would be imposed when methane from natural gas systems or
coal
mining is burned. The resulting carbon dioxide would cost $6.55 for each $50 of methane emissions
prevented. For these sources,
I use the percentage reduction in methane at a permit price of $43.45 per
ton as the EPA estimate of methane reductions at a permit price of $50 per ton.
68.
This estimate of the price sensitivity of other greenhouse gases is consistent with the use of the baseline
price for carbon emissions. If the true baseline price for other greenhouse gases is lower than that for
carbon dioxide, then price sensitivity for those gases is correspondingly lower.
63
that
the same permit price yields roughly a 35 percent reduction in emissions.
67
Using
EIA's
baseline price of carbon emissions for the sake of comparability, these numbers
imply
price sensitivities of -1.66 for carbon dioxide and -2.74 for methane. Thus, I
assume
the price sensitivity for other greenhouse gases is 1.65 (2.74 divided by 1.66)
times as large as the price sensitivity for carbon dioxide, or -0.88 for the United
States.
68
Range of Uncertainty for Estimates of Permit Prices
This study only
examines the uncertainty in permit prices coming from uncertainty
about
the estimate of price sensitivity. (As explained in Chapter 2, errors in the
forecast
of baseline emissions would change both the costs and benefits of the Kyoto
Protocol
in the same direction.) This uncertainty is best measured by looking at
standard
errors from estimates of the elasticity of energy demand. Using the range
of model estimates of price sensitivity would not provide a good measure of
uncertainty,
since differences in these estimates stem more from model assumptions
than
from uncertainty about how responsive energy users are to changes in the price
of energy.
The
range of uncertainty in permit prices presented in Chapter 2 roughly corresponds
to one standard error
above and below the estimate of price sensitivity. According
to
Dahl’s study, the median standard error of estimates of all types of energy demand
in
the medium run is 0.07. Surprisingly, the average standard error between studies,
roughly
0.13, is much larger. I use the average of these standard errors, 0.10, and
then
augment it to account for uncertainty about fuel substitution, which is not
r
eflected in the elasticity of energy demand. Assuming that the error for fuel
substitution is of the same magnitude as, but uncorrelated with, the error for total
energy
demand, the final standard error is somewhat less than 0.13. Reducing the
absolute
magnitude of price sensitivity of carbon emissions by 0.13, from -0.53 to
64
0.40, boosts permit prices by 34 percent, while a price sensitivity of -0.66 reduces
permit prices 20 percent. The average error is thus 27 percent.
Impact of the Clean Development Mechanism
It is difficult to gauge how effective the clean development mechanism (CDM) would
be in reducing emissions in non-Annex B countries. The only study that quantifies the
impact, the study using MERGE, uses the EMF-16 assumption that 15 percent of the
non-Annex B reductions made under global trading would be available as CDM
projects under Annex B trading. I use this assumption in developing model synthesis
estimates. CDM projects would then provide 43 mmtc of emissions reductions in the
ideal implementation scenario.
Domestic Direct Cost
Domestic direct cost is calculated as the area under the marginal abatement cost
curve. This curve plots the permit price against the corresponding reduction in
domestic emissions, excluding reductions due to lower GDP. When accounting for
the impact of pre-existing taxes in energy markets, the marginal abatement cost curve
plots the permit price plus pre-existing taxes against the reduction in emissions.
Eliminating pre-existing taxes reduces the U.S. baseline price of carbon from $298 to
$264 per metric ton of carbon.
Impact on GDP
If restrictions were placed on emissions, GDP losses could flow from six sources:
• domestic direct cost;
• a lower capital stock because of higher prices for plant and equipment;
• impacts of capital flows;
• reduced labor supply;
• a lower capital stock because of the decline in income arising from buying permits
from abroad; and
• impacts of higher interest rates on unemployment and the capital stock.
Unfortunately, the studies provide no estimates of the impact of any individual factor
on GDP. Nonetheless, enough information can be gleaned from available model
evidence to develop synthesis estimates of GDP loss.
Domestic Direct Cost and Higher Prices for Plant and Equipment.
One can use GDP
loss in general equilibrium models in the case of no international trade of permits as
an estimate of the combined impact of domestic direct cost and higher prices for plant
69.
In
a model that uses chain-type aggregation to determine real GDP, real consumption can reflect changes
in consumer surplus. None of the general equilibrium models use this type of aggregation, however.
65
and
equipment on GDP. Removing permit trading eliminates the effect of permit
purchases
from other countries. In addition, most general equilibrium models have
small
or nonexistent capital flows and exogenous labor supply and unemployment, so
they
effectively remove these effects on GDP. One exception is G-Cubed, in which
the
United States has a capital inflow and higher unemployment. However, those
factors
have roughly offsetting effects on GDP, so GDP loss yields a rough estimate
of
the effects of domestic direct cost and higher capital prices in this model as well.
Another
exception is the JWS model, which has large labor supply responses. In that
model,
the GDP loss combines the impacts of domestic direct cost, lower investment
and lower labor supply.
With
no international trade of permits, the geometric mean of the ratio of GDP loss
to
domestic direct cost is 2.5 in the general equilibrium models that report GDP,
except
JWS. These ratios range from 1.5 in AIM to 4.8 in CETA. However, as
permit
prices fall, domestic direct cost declines proportionately faster than the decline
in
investment. Thus, the ratio of GDP losses stemming from domestic direct cost and
higher
investment costs to domestic direct cost rises as permit prices fall.
Consequently,
a method other than a simple ratio to domestic direct cost must be used
to
determine the impact of domestic direct cost and higher investment costs on GDP
when
countries can trade permits. Fortunately, theory provides a way to quantify
those two effects on GDP.
The loss in GDP stemming from domestic direct cost differs from domestic direct cost
for
three reasons. First, a portion of domestic direct cost reflects reduced consumer
surplus,
and thus does not affect GDP.
69
Second, domestic direct cost as calculated
by
the models ignores the impact of pre-existing taxes on energy, and thus understates
the
impact on GDP. These two factors happen to roughly offset each other in the
case
of no international permit trading, leaving a loss in GDP roughly equal to
domestic direct cost. Third, however, this loss in GDP reduces income and thus
saving,
feeding back into investment and causing a further loss in GDP. Overall, the
loss
in GDP from domestic direct cost should be between 1.3 and 1.4 times as large
as domestic direct cost.
The
permit price affects investment through its impact on the cost of producing new
plant
and equipment. In the DRI model, a permit price that raises the overall price
level
by one percent boosts capital prices by a little more than 0.5 percent, and thus
reduces
the desired capital stock by roughly the same percentage. Using information
on
the effect of the capital stock on potential GDP and on the feedback effects of
resulting
changes in saving on investment, one can determine the impact of emissions
restrictions
on GDP through lower investment. For the case of no international trade
70.
Much
of the large response of labor supply in the JWS model results from the assumption of a
representa
tive consumer with an infinite lifetime. This exaggerates the intertemporal tradeoff of labor
and leisure beyond what it would be in a model assuming consumers with finite lifetimes.
71.
Congressional Budget Office,
Labor Supply and Taxes
, CBO Memorandum (January 1996).
66
of
permits, this impact is between 1.1 and 1.2 times as large as domestic direct cost.
Adding
this to the effect from domestic direct cost produces a GDP loss about 2.5
times
as large as domestic direct cost in the case of no international permit trading,
the same as the estimate from the models.
Capital
Flows.
G-Cubed is the only model in which capital flows have a large impact
on
GDP. In the study using this model, the United States has lower permit prices than
the
other developed countries when there is no international trade of permits, and so
draws
capital flows from those countries. These flows boost investment, and thus
GDP.
According
to the model synthesis, however, only Japan would have a higher permit
price
than the United States when there is no international trade of permits allowed.
And
with Annex B trade of permits, only Australia, Canada and the former Soviet
bloc
would see a larger percentage increase in their overall price level, a plausible
measure
of the increase in the cost of doing business, and thus of the reduction in the
return
to capital. Thus, if anything, the Kyoto Protocol would likely lead to capital
outflows
from the United States. Given the difficulty of judging the size or effect of
these
flows, the possible impact of capital flows on GDP has been left out of the
model synthesis.
Reduced Labor Supply.
Most
of the general equilibrium models assume that labor
supply
is exogenous—that is, it does not respond to changes in the marginal after-tax
real
wage. On the other hand, the JWS model assumes that labor supply is highly
responsive
to the marginal after-tax real wage, with an elasticity of about 1.0.
70
The
actual response would be between these two extremes.
In
a memorandum looking at the empirical evidence, CBO concluded that “a 10
percent
increase in after-tax wages would raise total hours of work by between zero
and
3 percent,” indicating a labor supply elasticity of 0 to 0.3. The study went on to
state
that “those estimates may somewhat overstate the responsiveness of the
economy’s
labor supply,” because they did not account for how married men and
women
would respond to changes in a spouse’s after-tax wage rate.
7
1
The labor
supply
elasticity in the DRI model, 0.06, falls in this range, although below its
midpoint.
Consequently, this estimate is used to calculate the change in labor supply
and
the resulting impact on GDP in the model synthesis results. Using the midpoint
elasticity of 0.15 instead would boost estimates of GDP loss by about 15 percent.
67
Permit Purchases from Other Countries. Unless purchases of permits from other
countries are counted as an import of a service, such purchases have no direct impact
on GDP. However, the permits must ultimately be paid for with higher net exports.
Those purchases reduce the share of GDP going to investment, which in turn reduces
potential GDP. This impact is similar to the feedback effects of lower saving on GDP
discussed above, and is likewise small.
Impacts of Higher Interest Rates on Unemployment and Investment. The timing of
GDP loss from higher interest rates depends strongly on when the increase in the
general price level stemming from higher energy prices is assumed to occur, and how
long unemployment remains above baseline levels in response to that increase. This
can be seen most clearly in the two EIA studies, which use the same model. In the
study in which energy prices begin to rise in 2005, unemployment is still well above
baseline levels in 2010, and real GDP falls 4.2 percent below its baseline level. If
energy prices begin to rise in 2000, however, unemployment is back to baseline levels
in 2010, and real GDP falls less than one third as much, 1.2 percent, even though the
permit price is nearly as large as in the other case.
I interpret the economic impacts in 2010 as representative of impacts over a longer
period of time. A simple-minded focus on the economy’s response in 2010 alone
would exaggerate the effect on GDP of fighting higher inflation over this longer
period. According to the Kyoto Protocol, permits would first be imposed in 2008,
so the effect on unemployment would be near its peak in 2010. Instead, the model
synthesis estimates assume that the effect of fighting higher inflation on GDP is spread
evenly over a ten year period, so that the estimated effect in 2010 will be
representative of this longer period of time.
The Federal Reserve would focus on the portion of inflation that it believed would be
permanent if not counteracted by higher interest rates. In the DRI model, a one-time
1.0 percent upward shock to the general price level would trigger a permanent 0.11
percentage point rise in the inflation rate if not offset by higher unemployment and
lower capacity utilization. Thus, the 3.3 percent increase in the general price level
occurring with no international trade of permits would lead to a permanent 0.37
percentage point rise in the inflation rate.
Eliminating this extra inflation would reduce real GDP by an average of nearly 0.5
percent per year over 10 years. A loss of nearly 0.3 percent per year would be
directly associated with the higher unemployment and lower capacity utilization
needed to bring inflation back down. An additional 0.2 percent per year would be lost
because the higher interest rates needed to slow the economy would hurt investment,
reducing potential GDP. The 0.5 percent reduction in GDP from macroeconometric
effects in the no-trade case is smaller than those in the DRI and WEFA studies and
72.
For studies publishing both private and government consumption (DRI, EIA and WEFA), this ratio is
calculated
using the percentage change in total (private plus government) consumption. The model
synthesis
estimates assume that the percentage change in private consumption is the same as the
percentage
change in government consumption. This seems a more realistic long run assumption than
assuming that government consumption and investment do not respond to changes in GDP.
68
the
first EIA study (which assumes responses begin in 2005) but larger than those in
the
Oxford study and the second EIA study (which assumes responses begin in 2000).
GDP
Losses in Other Countries.
The model synthesis estimates assume that GDP
loss
in other countries is determined in the same way that GDP loss in the United
States
is. This is not likely to hold exactly, because inflation may respond differently
to
unemployment in other countries than in the United States, among other things.
However, GDP loss
in other countries only affects the U.S. results insofar as lower
GDP
in those countries reduces their demand for permits and thus the international
permit price. Those effects will be small, so using a reasonable approximation for
foreign
GDP loss should not have much effect on permit price or GDP estimates for
the United States.
Impact on Consumption
Without
international trade of permits or reductions in greenhouse gases other than
carbon
dioxide, the ratio of the percentage drop in consumption to the percentage
drop
in GDP averages 0.70 among the models.
72
(To avoid double-counting the DRI
model,
this average uses only the ratio from the original EIA study. The ratio in the
DRI
study is higher, while that in the EIA early start study is lower). In most models,
as
international trade of permits is added, the ratio of the percentage change in
consumptio
n is more closely tied to the percentage change in GDP less permit
purchases
than to the percentage change in GDP alone. That is, an extra $100 of lost
GDP
has about the same effect on consumption as an extra $100 spent on foreign
permits.
The model synthesis estimates thus assume that the ratio of the percentage
change
in consumption to the percentage change in GDP less permit purchases
declines gradually from 0.70 without
international trade of permits or sinks to 0.60
with unrestricted global trade of permits—the average estimates from the models.
Change in Global Emissions
The
change in global emissions from baseline levels in 2010 under the Kyoto Protocol
would
equal emissions reductions in countries constrained by the protocol less
emissions
increases in countries unconstrained by the protocol. Such increases in
emissions,
resulting from greater oil consumption in response to lower global oil
prices
and a relocation of energy-intensive industries from constrained to
69
unconstrained countries, are known as leakage. Leakage would boost emissions in
any country with a domestic permit price of zero. Such countries would include the
former Soviet bloc if limits were placed on permit imports or international trade of
permits were blocked altogether, and would include non-Annex B countries if there
were no global trade of permits. (Leakage to non-Annex B countries would also
occur under global trade of permits if limits were placed on permit imports.) In the
discussion that follows, the leakage rate is defined as the increase in emissions in
unconstrained countries as a percentage of the decline in emissions in constrained
countries.
The leakage rate depends on the permit price in countries where emissions are
constrained. The higher the permit price in these countries, the more leakage there
will be. Thus, on average, the models find that the leakage rate to non-Annex B
countries drops from 17 percent with no international trade of permits to 10 percent
with unrestricted Annex B trade of permits. (These averages exclude models that
assume no leakage or that do not specify emissions in countries other than the United
States.) Model synthesis estimates of leakage are calculated using the relationship
between leakage rates and permit prices established by these two data points in cases
where the permit price in non-Annex B countries is zero. In models in which the
permit price in the former Soviet bloc is also zero when there is no international trade
of permits, leakage to these countries is 40 percent as large as leakage to the non-
Annex B countries. So leakage to the former Soviet bloc is assumed to be 40 percent
as large as non-Annex B leakage when permit prices in the former Soviet bloc are
zero.
It is difficult to be certain whether these leakage estimates are consistent with the
model synthesis estimates of price sensitivity. The amount of leakage per dollar of
permit price and the price sensitivity of emissions should be positively correlated: the
more easily emissions can move from one country to another, the greater price
sensitivity will be in a given country. However, these concepts show little correlation
across the models.
Gasoline Price
The change in gasoline prices consists of three pieces: the direct impact of the permit
price; the impact of lower demand for gasoline on refiner margins; and the impact of
lower global oil demand on crude oil prices. The direct impact is 23.8 cents a gallon
per each $100 per metric ton increase in the permit price, according to data from EIA.
Refiner margins fall by 0.8 cents a gallon per each $100 per metric ton increase in the
permit price, according to averages from the DRI, EIA and WEFA studies. The
percentage change in the price of crude oil is found by combining model responses of
crude oil prices to global demand for crude oil with estimated changes in global
73.
Acco
rding to Dahl’s demand elasticities, demand for natural gas would fall below baseline levels, so
wellhead
prices for natural gas would drop. However, the Dahl elasticities also imply smaller reductions
in
petroleum than the model synthesis, reducing the drop in crude oil prices. Thus, using the Dahl
elasticities
for individual fossil fuels would produce higher gasoline and lower natural gas prices than the
elasticities implied by the models.
70
demand
for crude oil. According to the model average, the global elasticity of supply
for crude oil is between 0.5 and 0.6 over a ten-year horizon.
Price of Natural Gas
The
change in the price of natural gas also consists of three pieces: the direct impact
of
the permit price; the impact of higher natural gas demand on wellhead prices; and
the impact of lower residential and commercial demand on distribution costs. Permits
directly add $1.48 per thousand cubic feet per each $100 per
metric ton increase in
the
permit price. The increase in wellhead prices is calculated by combining the
responses
of wellhead prices to U.S. demand for natural gas in the DRI, EIA and
WEFA studies with the rise in demand implied by the model synthesis estimates. (The
model synthesis estimates focus on changes in U.S. demand because of the difficulties
of
transporting natural gas overseas.) This rise in natural gas demand is smaller than
in
the EIA study, because the drop in non-electricity usage is larger.
73
Finally,
distribution
costs per cubic foot of gas delivered to residential and commercial
customers
would rise as the same fixed costs were spread over a smaller consumption
base.
Electricity Prices
In
the model synthesis estimates, electricity prices rise for two reasons: the direct
impact
of the permit price; and the higher generating costs associated with fuel
switching.
These both depend on the amount of fuel switching. A shift from coal to
natural
gas or non-fossil sources reduces the direct impact of the permit price but
increases
generating costs. However, generators will only want to switch fuels if the
increase in generating costs is smaller than the savings in permit expenses.
Determining
the change in the price of electricity requires several calculations. First,
I
calculate a change in emissions from electricity generation consistent with model-
based
and empirical estimates of overall price sensitivity and the price sensitivity of
non-electricity
emissions. Multiplying the permit price by the resulting level of
emissions
yields the total value of permits required to produce electricity. Model-
based
and empirical estimates of the price sensitivity of electricity demand are
combined
with the change in emissions from electricity generation to determine the
portion
of the drop in electricity emissions resulting from fuel switching. The total
71
increase in generating costs due to fuel switching is assumed to equal the change in
emissions resulting from fuel switching times the average permit price at which those
switches are made, i.e., one half the permit price. Adding together the direct impact
of the permit price and the increase in generating costs due to fuel switching and
dividing by final electricity consumption yields the rise in electricity prices per
kilowatt-hour. (The change in electricity consumption is calculated from the change
in electricity emissions that does not result from fuel switching.)
These calculations assume that three possible additional impacts on electricity prices
are negligible. First, there is no net impact from changes in fossil fuel prices. That
is, the increase in natural gas costs from higher wellhead prices is assumed to offset
the reduction in coal costs from lower minemouth prices. Second, the change in
marginal cost is assumed to equal the change in average cost. The EIA study instead
argues that marginal costs would rise more than average costs, pushing prices higher
than what this paper assumes. Third, all customers are assumed to face the same
absolute increase in electricity prices. The DRI study assumes this, and the WEFA
study assumes something close to it. The EIA study, however, assumes that the
percentage increase in electricity prices is roughly the same for all customers, meaning
that the absolute increase for residential customers is much larger than the increase
for other customers. If this is true, the model synthesis estimates understate the
percentage increase in residential electricity prices.
72
Bibliography
Congressional Budget Office, Labor Supply and Taxes, CBO Memorandum (January
1996).
Congressional Budget Office, The Economics of Climate Change: A Primer (April
2003).
Congressional Budget Office, Who Gains and Who Pays Under Carbon-Allowance
Trading? The Distributional Effects of Alternative Policy Designs (June 2000).
Council of Economic Advisors, The Kyoto Protocol and the President’s Policies to
Address Climate Change (July 1998).
Dahl, Carol, “A Survey of Energy Demand Elasticities in Support of the Development
of the NEMS” (working paper, Colorado School of Mines, October 1993).
Department of State, Proposal by United States, Canada, Japan: Phase-in for Forest
Management in the First Commitment Period, November 21, 2000 (available at
www.state.gov/www/global/global_issues/climate/cop6/001121_phase-in.html).
Department of State, United States Submission on Land-Use, Land-Use Change and
Forestry, August 1, 2000 (available at www.state.gov/www/global/global_issues/
climate/climate_2000_submiss.html).
DRI/McGraw-Hill, The Impact of Carbon Mitigation Strategies on Energy Markets,
the National Economy, Industry, and Regional Economies (study prepared for
UMWA-BCOA, Lexington, MA, July 1997).
Edmonds, J.A., H.M. Pitcher, D. Barns, R. Baron and M.A. Wise, “Modelling future
greenhouse gas emissions: The second generation model description,” in L.R.
Klein and Fu-Chen Lo, eds., Modelling Global Change (Tokyo, Japan: United
Nations University Press, 1995), pp. 295-340.
Ellerman, A. Denny and Annelene Decaux, Analysis of Post-Kyoto CO
2
Emissions
Trading Using Marginal Abatement Cost Curves, Report Series No. 40
(Cambridge, MA: MIT Joint Program on the Science and Policy of Global
Change, October 1998).
Energy Information Administration, Analysis of the Impacts of an Early Start for
Compliance with the Kyoto Protocol (July 1999).
Energy Information Administration, Annual Energy Outlook 1998, DOE/EIA-0383
(December 1997).
Energy Information Administration, Annual Energy Outlook 2000, DOE/EIA-0383
(December 1999).
Energy Information Administration, Annual Energy Outlook 2003, DOE/EIA-0383
(January 2003).
Energy Information Administration, Impacts of the Kyoto Protocol on U.S. Energy
Markets and Economic Activity (October 1998).
Energy Information Administration, International Energy Outlook 2000, DOE/EIA-
0484 (March 2000).
Energy Information Administration, International Energy Outlook 2003, DOE/EIA-
0484 (March 2003).
73
Environmental Protection Agency, Office of Air and Radiation, U.S. Methane
Emissions 1990-2020: Inventories, Projections, and Opportunities for
Reductions (September 1999).
Geurts, Ben, Arjen Gielen, Richard Nahuis, Paul Tang and Hans Timmer, Scanning
WorldScan: Final report on the presentation and evaluation of WorldScan, a
model of the WORLD economy for SCenario ANalysis, Report No. 410200008
(The Hague, Netherlands: CPB Netherlands Bureau for Economic Policy
Analysis, 1997).
Interlaboratory Working Group on Energy-Efficient and Low-Carbon Technologies,
Scenarios of U.S. Carbon Reductions: Potential Impacts of Energy-Efficient
and Low Carbon Technologies by 2010 and Beyond (Oak Ridge National
Laboratory, Lawrence Berkeley National Laboratory, Pacific Northwest National
Laboratory, National Renewable Energy Laboratory, and Argonne National
Laboratory, September 1997).
International Energy Agency, Energy Prices & Taxes: Quarterly Statistics, no. 1
(1999).
Jorgenson, Dale W., Richard J. Goettle, Peter J. Wilcoxen and Daniel T. Slesnick,
Carbon Mitigation, Permit Trading and Revenue Recycling (prepared for the
Environmental Protection Agency, November 1998).
Kainuma, Mikiko, Yuzuru Matsuoka and Tsuneyuki Morita, The AIM/ENDUSE
Model and Case Studies in Japan (Tsukuba, Japan: Center for Global
Environmental Research, undated).
Manne, Alan, Robert Mendelsohn and Richard Richels, “MERGE: A model for
evaluating regional and global effects of GHG reduction policies,” Energy
Policy, vol. 23, no. 1 (1995), pp. 17-34.
Manne, Alan and Richard Richels, “On stabilizing CO
2
concentrations–cost-effective
emission reduction strategies,” Environmental Modeling and Assessment, vol.
2, no. 4 (1997), pp. 251-265.
McCarl, Bruce A., “Carbon Sequestration via Tree Planting on Agricultural Lands:
An Economic Study of Costs and Policy Design Alternatives,” Internet draft,
November 1998 (available at http://ageco.tamu.edu/faculty/mccarl/papers/676.
pdf).
McKibbin, Warwick J. and Peter J. Wilcoxen, The Theoretical and Empirical
Structure of the G-Cubed Model, Discussion Papers in International Economics,
No. 118 (Washington, DC: Brookings Institution, December 1995).
Nordhaus, William D. and Joseph Boyer, Warming the World: Economic Models of
Global Warming (Cambridge, MA: MIT Press, 2000).
Parry, Ian W. H., Roberton C. Williams, III and Lawrence H. Goulder, “When Can
Carbon Abatement Policies Increase Welfare? The Fundamental Role of
Distorted Factor Markets,” Journal of Environmental Economics and
Management, vol. 37, no. 1 (1999), pp. 52-84.
Peck, Stephen C. and Thomas J. Teisberg, “CETA: A Model for Carbon Emissions
Trajectory Assessment,” The Energy Journal, vol. 13, no. 1 (1992), pp. 55-77.
74
Reilly, John, Monika Mayer and Jochen Harnisch, Multiple Gas Control Under the
Kyoto Agreement Report No. 58 (Cambridge, MA: MIT, Joint Program on the
Science and Policy of Global Change, March 2000).
Reilly, John, Ronald G. Prinn, Jochen Harnisch, Jean Fitzmaurice, Henry D. Jacoby,
David Klicklighter, Peter H. Stone, Andrei P. Sokolov and Chien Wang, Multi-
Gas Assessment of the Kyoto Protocol, Report Series No. 45 (Cambridge, MA:
MIT, Joint Program on the Science and Policy of Global Change, January 1999).
Standard & Poor’s DRI, The Impact of Meeting the Kyoto Protocol on Energy
Markets and the Economy (Lexington, MA: Standard & Poor’s DRI, 1998).
United Nations Framework Convention on Climate Change: Subsidiary Body for
Scientific and Technological Change, Mechanisms Pursuant to Articles 6, 12
and 17 of the Kyoto Protocol: Text for further negotiation on principles,
modalities, rules and guidelines (April 2000).
WEFA, Inc., Global Warming: The Economic Cost of Early Action, National
Impacts (Eddystone, PA: WEFA, Inc., 1997).
WEFA, Inc., Global Warming: The High Cost of the Kyoto Protocol, National and
State Impacts (Eddystone, PA: WEFA, Inc., 1998).
Weyant, John P., ed., The Costs of the Kyoto Protocol: A Multi-Model Evaluation,
Special Issue of the Energy Journal (Cleveland, OH: Energy Economics
Educational Foundation, Inc., 1999).
Yang, Z., R.S. Eckaus, A. D. Ellerman and H.D. Jacoby, The MIT Emissions
Prediction and Policy Analysis (EPPA) Model Report No. 6 (Cambridge, MA:
MIT, Joint Program on the Science and Policy of Global Change, May 1996).
75
Table 2-1. Studies Analyzing the Impact of Emissions Reductions, by Model and
Institutions of Authors
Model Institutions of Authors
AIM
a
(A
sian-Pacific Integrated Model)
NIES (National Institute for Environmental
Studies, Japan) and Kyoto University
CETA
a
(Model for Carbon Emissions Trajectory
A
ssessment)
EPRI (Electric Power Research Institute) and
Teisberg Associates
DRI
(DRI Macro Model)
Standard & Poor's Data Resources, Inc.
EIA
Energy sector impacts: NEMS
(N
ational Energy Modeling System);
Macro impacts: DRI
Energy Information Administration
EPPA
a
(Emissions Prediction and Policy Analysis
Model)
MIT (Massachusetts Institute of Technology)
G-Cubed
a
(Global General Equilibrium Growth Model)
Australian National University, Brookings
Institution, Environmental Protection
Agency, and University of Texas
GTEM
a
(Global Trade and Environment Model)
ABARE (Australian Bureau of Agricultural and
Resource Economics)
(JWS
)
(J
orgenson-Wilcoxen-Slesnick Model)
Dale W. Jorgenson Associates
MERGE
a
(Model for Evaluating Regional and Global
E
ffects of Greenhouse Gas Reduction
Policies)
EPRI (Electric Power Research Institute) and
Stanford University
MS-MRT
a
(Multi-Sector Multi-Region Trade Model)
Charles River Associates and University of
Colorado
Oxford
a
(Oxford Global Macroeconomic and Energy
Model)
Oxford Economic Forecasting
RICE
a
(Regional Dynamic Integrated Model of
C
limate and the Economy)
Yale University
SGM-Administration
(see SGM-PNNL)
Clinton Administration
SGM-PNNL
a
(Second Generation Model)
Batelle Pacific Northwest National Laboratory
WEFA
Macro Model
Wharton Econometric Forecasting Associates
WorldScan
a
(Model of the World Economy for Scenario
An
alysis)
RIVM (National Institute of Public Health and
the Environment, Netherlands)
NOTES:
a. Participants in Round 16 of Stanford University’s Energy Modeling Forum.
76
Table 3-1. Percentage Reduction in Emissions of Carbon Dioxide in 2010 Required in
Various Regions Under Alternative Permit-Trading Scenarios
(Percentage Reduction from Baseline)
Model
Emissions Reduction
in U.S. with No
International Permit
Trade
Total Emissions
Reduction in Annex B
Countries with Permit
Trading Among Annex B
Countries
Global Emissions
Reduction with
Global Permit
Trading
Models with No Offsets
AIM 25 13 7
CETA 29 8 5
DRI 29 n.a. n.a.
EIA 30 n.a. n.a.
EPPA 29 18 n.a.
G-Cubed 30 17 9
GTEM 28 20 n.a.
MS-MRT 30 15 7
a
Oxford 31 21
b
n.a.
RICE 25 10 5
WEFA 27 n.a. n.a.
WorldScan 27 21 n.a.
Models with Offsets
EIA 27 n.a. n.a.
MERGE 29 16 10
SGM-Administration 28 11 7
SGM-PNNL 29 14 10
Memorandum
Average of All Models 29 16
c
8
SOURCE: Author’s calculations, using: Council of Economic Advisors,
The Kyoto Protocol and the President’s Policies to
Address Climate Change
(July 1998); Energy Information Administration,
Impacts of the Kyoto Protocol on U.S.
Energy Markets and Economic Activity
(October 1998); William D. Nordhaus and Joseph Boyer,
Warming the
World: Economic Models of Global Warming
(Cambridge, MA: MIT Press, 2000); Standard & Poor’s DRI,
The
Impact of Meeting the Kyoto Protocol on Energy Markets and the Economy
(Lexington, MA: Standard & Poor’s
DRI, 1998); WEFA, Inc.,
Global Warming: The High Cost of the Kyoto Protocol, National and State Impacts
(Eddystone, PA: WEFA, Inc., 1998); John P. Weyant, ed.,
The Costs of the Kyoto Protocol: A Multi-Model
Evaluation, Special Issue of the Energy Journal
(Cleveland, OH: Energy Economics Educational Foundation, Inc.,
1999); and personal communications from Richard Richels and John Weyant.
NOTES: For the United States, the 2010 cap is 93 percent of 1990 emissions. Percentage reduction refers to the percentage
reduction in 2010 baseline emissions required to meet the cap. In some models, the actual percentage reduction differs
slightly from this figure.
Offsets are the amount of carbon dioxide offset by forest growth and reductions in other greenhouse gases beyond their
share of the cap.
The model average assumes no offsets. Targets for non-Annex B countries equal baseline emissions in those countries.
n.a. = not available
a. The MS-MRT scenario for global trade of permits assumes that emissions caps for non-Annex B countries are the emissions
they produce when the Annex B countries meet their caps without international trade of permits.
b. Oxford data for Annex B excludes Ukraine, eastern Europe, Australia and New Zealand.
c. This average excludes Oxford, which does not have data for all of Annex B.
77
Table 3-2. Impact on U.S. Carbon Emissions of a 1 Percent Increase in the Price of Carbon-
Energy, Assuming No International Trading of Permits, Selected Years
Model
Percentage Change in Carbon
Emissions from Baseline in
Impact in 2010 as a
Percentage of the Impact
in 2020
2010 2020
AIM -0.68 -1.01 68
CETA -0.67 -0.75 89
DRI
a
-0.49 -0.59 84
EIA
b
-0.41 -0.60 68
EIA-early start
b
-0.47 -0.67 71
EPPA -0.73 -1.53 48
G-Cubed -1.54 -2.10 74
GTEM -0.40 -0.49 81
JWS
c
-1.28 -1.26 101
MERGE -0.53 -0.72 73
MS-MRT -0.51 -0.62 83
Oxford -0.42 -0.42 100
RICE -0.90 -0.83 108
SGM-Administration -0.68 n.a. n.a.
SGM-PNNL -0.69 -0.74 93
WEFA -0.42 -0.54 78
WorldScan -1.17 -1.13 104
SOURCE: Author’s calculations, using: Council of Economic Advisors,
The Kyoto Protocol and the President’s Policies to
Address Climate Change
(July 1998); DRI/McGraw-Hill,
The Impact of Carbon Mitigation Strategies on
Energy Markets, the National Economy, Industry, and Regional Economies
(study prepared for UMWA-
BCOA, Lexington, MA, July 1997); Energy Information Administration,
Analysis of the Impacts of an Early
Start for Compliance with the Kyoto Protocol
(July 1999); Energy Information Administration,
Impacts of the
Kyoto Protocol on U.S. Energy Markets and Economic Activity
(October 1998); Dale W. Jorgenson, Richard J.
Goettle, Peter J. Wilcoxen and Daniel T. Slesnick,
Carbon Mitigation, Permit Trading and Revenue Recycling
(prepared for the Environmental Protection Agency, November 1998); William D. Nordhaus and Joseph Boyer,
Warming the World: Economic Models of Global Warming
(Cambridge, MA: MIT Press, 2000); WEFA, Inc.,
Global Warming: The High Cost of the Kyoto Protocol, National and State Impacts
(Eddystone, PA: WEFA,
Inc., 1998); John P. Weyant, ed.,
The Costs of the Kyoto Protocol: A Multi-Model Evaluation, Special Issue of
the Energy Journal
(Cleveland, OH: Energy Economics Educational Foundation, Inc., 1999); and personal
communications from Richard Richels and John Weyant.
NOTES: All models assume the 2020 emissions cap is the same as the 2010 cap.
These numbers provide an estimate of how energy users adjust their use of carbon-based energy in response to changes
in its price. That price sensitivity is negative because energy use falls as its price rises. Larger absolute values indicate
a larger response. The numbers are calculated by using logarithms (see Appendix A for details). Because price
sensitivity is nonlinear, these numbers should not be scaled up for larger price changes using simple multiplication.
n.a. = not available
a. Data for DRI are geometric weighted averages of results from two scenarios assuming the Kyoto target is 90 percent and
100 percent, respectively, of 1990 emissions.
b. In these scenarios, EIA assumes emissions of carbon dioxide are reduced 7 percent below 1990 levels.
c. This model assumes emissions return to 1990 levels.
78
Table 3-3. Impact of a 1 Percent Increase in the Price of Carbon-Energy on Carbon Emissions in
2010 Under Alternative Permit-Trading Scenarios, By Region
(Percentage Change from Baseline Emissions)
Annex B Permit Trading Global Permit Trading
United States
All Annex B
Countries United States
All Annex B
Countries
Rest of
World
AIM -0.62 -0.73 -0.57 -0.72 -0.34
CETA -0.52 -0.45 -0.58 -0.51 -0.30
DRI -0.48
a
n.a. n.a. n.a. n.a.
EIA -0.41
b
n.a. -0.29
c
n.a. n.a.
EIA early start -0.47
b
n.a. -0.34
c
n.a. n.a.
EPPA -1.05 -0.88 n.a. n.a. n.a.
G-Cubed -1.43 -1.09 -1.17 -1.03 -1.15
GTEM -0.33 -0.59 n.a. n.a. n.a.
MERGE -0.56 -0.49 -0.54 -0.46 -0.23
MS-MRT -0.50 -0.55 -0.46 -0.46 -0.41
Oxford -0.50 -0.47 n.a. n.a. n.a.
RICE -0.86 -0.68 -0.83 -0.68 -0.69
SGM-Administration -0.73 -0.65 -0.70 -0.66 -1.14
d
SGM-PNNL -0.80 -0.57 -0.82 -0.55 -1.02
WorldScan -2.12 -3.06 n.a. n.a. n.a.
SOURCE: Author’s calculations, using: Council of Economic Advisors,
The Kyoto Protocol and the President’s Policies
to Address Climate Change
(July 1998); Energy Information Administration,
Analysis of the Impacts of an
Early Start for Compliance with the Kyoto Protocol
(July 1999); Energy Information Administration,
Impacts
of the Kyoto Protocol on U.S. Energy Markets and Economic Activity
(October 1998); William D. Nordhaus
and Joseph Boyer,
Warming the World: Economic Models of Global Warming
(Cambridge, MA: MIT Press,
2000); Standard & Poor’s DRI,
The Impact of Meeting the Kyoto Protocol on Energy Markets and the
Economy
(Lexington, MA: Standard & Poor’s DRI, 1998); John P. Weyant, ed.,
The Costs of the Kyoto
Protocol: A Multi-Model Evaluation, Special Issue of the Energy Journal
(Cleveland, OH: Energy Economics
Educational Foundation, Inc., 1999); and personal communications from Richard Richels and John Weyant.
NOTES: These numbers provide an estimate of how energy users adjust their use of carbon-based energy in response to
changes in its price. That price sensitivity is negative because energy use falls as its price rises. Larger absolute
values indicate a larger response. The numbers are calculated by using logarithms (see Appendix A for details).
Because price sensitivity is nonlinear, these numbers should not be scaled up for larger price changes using simple
multiplication.
Only the Annex B trading scenario is consistent with the Kyoto Protocol.
n.a. = not available
a. Case 2 in DRI’s study of Annex B trading of permits.
b. U.S. emissions are reduced to a level 9 percent above 1990 levels.
c. U.S. emissions are reduced to a level 24 percent above 1990 levels.
d. Includes only China, India, Korea and Mexico.
79
Table 3-4. Model Estimates of U.S. Permit Prices in 2010, Without International Trade of
Permits
Key Determinants of Permit Prices
Models
Permit Price (in
1997 dollars
per mtc)
a
Percentage
Change in
Carbon
Emissions
Percentage
Change in GDP
Price Sensitivity
of Carbon
Emissions
Baseline Price of
Carbon-Energy
(in 1997 dollars
per mtc)
Models Using Kyoto Targets without Offsets
AIM 184 -25 -0.5 -0.68 362
CETA 201 -29 -1.9 -0.67 332
b
DRI
c
254 -29
d
-2.3 -0.49 274
EIA 355 -31
d
-4.2 -0.41 296
EIA early start 322 -30
d
-1.2 -0.47 296
EPPA 232 -29 n.a.
e
-0.73 387
G-Cubed 91 -30 -0.4 -1.54 361
GTEM 389 -28 -2.0 -0.40 340
MS-MRT 287 -30 -1.9 -0.51 300
Oxford Econ. 407 -30
d
-1.8 -0.42 334
RICE 184 -25 -1.0 -0.90 496
WEFA 270 -27
d
-3.2 -0.42 291
WorldScan 101 -27 n.a.
e
-1.17 332
b
Models Using Kyoto Targets with Offsets
EIA 300 -27
d
-3.5 -0.41 296
MERGE 286 -29 -1.0 -0.53 332
b
SGM-Admin. 192 -28 n.a.
e
-0.68 314
SGM-PNNL 189 -29 n.a.
e
-0.69 298
Memo: Model Targeting 1990 Emissions without Offsets
JWS 70 -25 -1.1 -1.28 292
SOURCE: Author’s calculations using: Council of Economic Advisors,
The Kyoto Protocol and the President’s Policies to
Address Climate Change
(July 1998); DRI/McGraw-Hill,
The Impact of Carbon Mitigation Strategies on Energy
Markets, the National Economy, Industry, and Regional Economies
(study prepared for UMWA-BCOA,
Lexington, MA, July 1997); Energy Information Administration,
Analysis of the Impacts of an Early Start for
Compliance with the Kyoto Protocol
(July 1999); Energy Information Administration,
Impacts of the Kyoto
Protocol on U.S. Energy Markets and Economic Activity
(October 1998); Dale W. Jorgenson, Richard J. Goettle,
Peter J. Wilcoxen and Daniel T. Slesnick,
Carbon Mitigation, Permit Trading and Revenue Recycling
(prepared
for the Environmental Protection Agency, November 1998); William D. Nordhaus and Joseph Boyer,
Warming the
World: Economic Models of Global Warming
(Cambridge, MA: MIT Press, 2000); WEFA, Inc.,
Global Warming:
The High Cost of the Kyoto Protocol, National and State Impacts
(Eddystone, PA: WEFA, Inc., 1998); John P.
Weyant, ed.,
The Costs of the Kyoto Protocol: A Multi-Model Evaluation, Special Issue of the Energy Journal
(Cleveland, OH: Energy Economics Educational Foundation, Inc., 1999); and personal communications from
Richard Richels and John Weyant.
NOTES: Price sensitivity of carbon emissions is a measure of how energy users adjust their use of carbon-based energy in
response to 1 percent change in its price. I use logarithms in preparing the estimate, so it cannot be scaled up for larger
percent changes by simple multiplication (see Appendix A for details). Price sensitivity is negative because energy use
falls as its price rises. Larger absolute values indicate a larger response.
mtc=metric ton of carbon
n.a. = not available
a. Permit prices were converted from other base years to 1997 dollars using the GDP price deflator.
b. The data needed to calculate the price of carbon-based energy are unavailable from these studies. I used the average
price of carbon-based energy in eight other studies using general equilibrium models.
c. Emissions and GDP data for DRI are weighted averages of results from two scenarios assuming the Kyoto target is 90
percent and 100 percent, respectively, of 1990 emissions.
d. Emissions in 2010 differ from target levels by small amounts.
e. The change in GDP is not available for these models. Price sensitivity is calculated using direct cost.
80
Table 3-5. Model Estimates of Annex B Permit Prices in 2010, with Permit-Trading
Among Annex B Countries
Key Determinants of Permit Prices in Annex B
Models
Permit Price
(in 1997
dollars per
mtc)
a
Percentage
Change
in Carbon
Emissions
Percentage
Change
in GDP
b
Price
Sensitivity of
Carbon
Emissions
Baseline Price of
Carbon-Energy
(in 1997 dollars
per mtc)
Models Using Kyoto Targets without Offsets
AIM 78 -13 0
c
-0.73 364
CETA 55 -8 -0.8 -0.45 305
d
EIA 166 n.a. n.a. n.a. n.a.
EIA early start 152 n.a. n.a. n.a. n.a.
EPPA 91 -18 n.a.
e
-0.88 364
G-Cubed 64 -17 0
c
-1.09 344
GTEM 128 -20 -0.6 -0.59 293
MS-MRT 94 -15 0.2 -0.55 266
Oxford Econ. 222 -21 -0.9 -0.47 351
RICE 41 -10 -0.3 -0.70 274
WorldScan 24 -21 n.a.
e
-3.06 291
d
Models Using Kyoto Targets with Offsets
DRI 115 n.a. n.a. n.a. n.a.
MERGE 116 -16 -0.7 -0.50 299
d
SGM-
Administration 54 -11 n.a.
e
-0.65 270
SGM-PNNL 82 -14 n.a.
e
-0.57 269
SOURCES: Same as Table 3-3.
NOTES: With Annex B trade of permits, the U.S. permit price equals the Annex B permit price.
Price sensitivity of carbon emissions is a measure of how energy users adjust their use of carbon-based energy in
response to 1 percent change in its price. I use logarithms in preparing the estimate, so it cannot be scaled up for larger
percent changes by simple multiplication (see Appendix A for details). Price sensitivity is negative because energy use
falls as its price rises. Larger absolute values indicate a larger response.
mtc=metric ton of carbon
n.a. = not available
a. Permit prices were converted from other base years to 1997 dollars using the GDP price deflator.
b. Regional percentage changes in GDP are weighted together using emissions data.
c. These changes round to zero.
d. Data needed to calculate the price of carbon-based energy are unavailable from these studies. For the United States, I use
the average price of carbon-based energy in eight other studies using general equilibrium models. Prices for Annex B then
incorporate regional differences in emissions between studies.
e. The change in GDP is not available for these models. Price sensitivity is calculated using direct cost.
81
Table 3-6. Model Estimates of Global Permit Prices in 2010, with Global Permit Trading
Key Determinants of Permit Prices
Models
Permit
Price
(in 1997
dollars per
mtc)
a
Percentage
Change in Carbon
Emissions
Percentage Change
in GDP
Price Sensitivity of
Carbon Emissions
Baseline Price of
Carbon-Energy (in
1997 dollars per
mtc)
Annex B Other Annex B Other Annex B Other Annex B Other
Models Using Kyoto Targets without Offsets
AIM 46 -8 -5 0
b
0.2 -0.72 -0.34 364 259
CETA 31 -5 -8 -0.2 0.1 -0.51 -0.30 305
c
202
c
EIA
d
68 n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a.
EIA early start
d
63 n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a.
G-Cubed
e
24 -7 -12 -0.2 -0.4 -1.03 -1.15 344 210
MS-MRT
f
32 -5 -8 -0.1 0
b
-0.46 -0.41 266 139
RICE 13 -3 -7 -0.2 -0.2 -0.68 -0.69 268 134
Models Using Kyoto Targets with Offsets on Annex B Emissions
MERGE 80 -11 -8 -0.4 -0.9 -0.46 -0.23 299
c
199
c
SGM-
Administration 22 -5 -15 n.a.
g
n.a.
g
-0.66 -1.14 270 144
SGM-PNNL 29 -6 -16 n.a.
g
n.a.
g
-0.55 -1.02 269 159
SOURCES: Author’s calculations, using: Council of Economic Advisors,
The Kyoto Protocol and the President’s Policies to Address Climate
Change
(July 1998); Energy Information Administration,
Analysis of the Impacts of an Early Start for Compliance with the Kyoto
Protocol
(July 1999); Energy Information Administration,
Impacts of the Kyoto Protocol on U.S. Energy Markets and Economic
Activity
(October 1998); William D. Nordhaus and Joseph Boyer,
Warming the World: Economic Models of Global Warming
(Cambridge, MA: MIT Press, 2000); John P. Weyant, ed.,
The Costs of the Kyoto Protocol: A Multi-Model Evaluation, Special
Issue of the Energy Journal
(Cleveland, OH: Energy Economics Educational Foundation, Inc., 1999); and personal communications
from Richard Richels and John Weyant.
NOTES: Price sensitivity of carbon emissions is a measure of how energy users adjust their use of carbon-based energy in response to a 1
percent change in its price. CBO uses logarithms in preparing the estimate, so it cannot be scaled up for larger percent changes by
simple multiplication (see Appendix A for details). Price sensitivity is negative because energy use falls as its price rises. Larger
absolute values indicate a larger response.
mtc = metric ton of carbon
n.a. = not available
a. Permit prices were converted from other base years to 1997 dollars using the GDP price deflator.
b. These changes round to zero.
c. Data needed to calculate the price of carbon-based energy are unavailable from these studies. For the United States, I use the average price
of carbon-based energy in eight other studies using general equilibrium models. Prices for other countries then incorporate regional
differences in emissions between studies.
d. EIA does not specify which scenario corresponds to global permit trading. These data are from EIA scenarios which assume U.S. emissions
of carbon dioxide are 24 percent above baseline levels in 2010.
e. The G-Cubed modelers assume that Mexico and OPEC do not participate in reducing emissions.
f. MS-MRT assumes that non-Annex B countries are given targets equal to their emissions under no international permit trading, rather
than their lower baseline emissions.
g. The change in GDP is not available for these models. Price sensitivity is calculated using direct cost.
82
Table 3-7. Model Synthesis Estimates of Permit Prices and Reductions of Emissions in 2010
Under Various Scenarios
U.S. Permit Price
(1997 dollars per mtc)
Reduction in Global
Emissions
a
(Percent of baseline
emissions of CO
2
)
Kyoto-Consistent Scenarios
Ideal Implementation 56±15 6
Cartel
b
70±19 6
Plus: No CDM 76±21 6
Plus: Restrictions on Permit Sales
c
137±37 10
Plus: No Offsets from Reductions in
Other Greenhouse Gases
d
178±48 9
Restrictions on Permit Purchases
e
122±33 6
Full Credit for Baseline Forest Growth
f
00
No International Trading of Permits
Ideal Implementation 216±58 9
No Offsets from Reductions in Other Greenhouse
Gases
d
264±71 8
Global Trading of Permits
g
Ideal Implementation 28±8 7
Cartel
b
31±8 7
Plus: Restrictions on Permit Sales
c
49±13 11
Restrictions on Permit Purchases
e
122±33 6
Plus: No Offsets from Reductions in Other
Greenhouse Gases
d
147±40 6
SOURCE: Author’s calculations.
NOTES: CDM= Clean Development Mechanism, by which Annex B countries can take credits for projects that reduce emissions
in non-Annex B countries.
mtc = metric ton of carbon
CO
2
= carbon dioxide
To convert 1997 dollars to 2002 dollars, multiply by 1.085.
Estimates of the U.S. permit price include a range of possible error, reflecting uncertainty about exactly how much
businesses, consumers and government would adjust their energy usage in response to higher prices.
a. Reductions are measured from the 2010 baseline. In this baseline, global emissions of carbon dioxide in 2010 are 40
percent above 1990 levels.
b. The countries of the former Soviet Union and eastern Europe are not permitted to sell permits they receive in excess of
their baseline emissions. This is equivalent to reducing their allocation of permits to baseline levels.
c. Countries cannot offset emissions of carbon dioxide by reducing emissions of other greenhouse gases below target
levels.
d. Each country must achieve at least 65 percent of its obligation to reduce emissions domestically.
e. Each country receives credit for the change in carbon stocks on cropland, grazing land, and forests, except those not
available or appropriate for wood production.
f. The global trading cases assume partial exercise of market power by permit-exporting countries within Annex B. Non-
Annex B countries are assumed not to control emissions of other greenhouse gases, and their caps are assumed equal to
their baseline emissions of carbon dioxide.
83
Table 4-1. Impact on Energy Prices of a $100 per mtc Permit Price
(Change from Baseline Prices unless Otherwise Noted)
Price of Coal to
Utilities
(Dollars per
short ton)
Price of
Gasoline
(Cents per
gallon)
Price of Natural Gas
to Households
(Dollars per
thousand cubic feet)
Price of
Electricity
to all Users
(Cents per
kwh)
Direct Impact
a
55 23.8 1.48 0.9 to 1.4
Change in Fossil Fuel Prices -1 to 1 -2 to -4 0.05 to 0.23 0 to 0.1
Other -1
-3 to 0 0 to 0.12 0.2 to 0.5
Total Impact
a
53 to 56 19 to 22 1.55 to 1.83 1.6-1.7
Memorandum
:
Baseline Price in 2010, in 23 to 25 127 to 136 5.83 to 6.30 5.1 to 6.3
1997 Dollars
Total Impact
(Percent of Baseline Price) 213 to 245 14 to 17 25 to 31 26 to 33
SOURCES: DRI/McGraw-Hill,
The Impact of Carbon Mitigation Strategies on Energy Markets, the National Economy,
Industry, and Regional Economies
(study prepared for UMWA-BCOA, Lexington, MA, July 1997); Energy
Information Administration,
Impacts of the Kyoto Protocol on U.S. Energy Markets and Economic Activity
(October
1998); WEFA, Inc.,
Global Warming: The Economic Cost of Early Action, National Impacts
(Eddystone, PA:
WEFA, Inc., 1997).
NOTES: kwh = kilowatt hour
mtc = metric ton of carbon
a. Direct impact figures for coal, gasoline, and natural gas are from the EIA study only. Total impact figures are taken from
the studies. Thus, the three components of total impact may not add up to the total impact.
84
Table 4-2. Model Synthesis Estimates of the Impacts on Energy Prices in 2010 of Various
Scenarios to Restrict Greenhouse Gases
U.S. Permit
Price (in 1997
dollars per mtc)
Gasoline Price
(Change from
Baseline in 1997
cents per gallon)
Price of Natural
Gas to Households
(Percent Change
from Baseline)
Price of
Electricity to
Households
(Percent Change
from Baseline)
Kyoto-Consistent Scenarios
Ideal Implementation 56±15 12±3 13±4 13±4
Cartel
a
70±19 15±4 16±4 15±4
Plus: No CDM 76±21 16±4 18±5 17±5
Plus: Restrictions on
Permit Sales
b
137±37 29±8 32±9 29±8
Plus: No Offsets from
Reductions in Other
Greenhouse Gases
c
178±48 38±10 42±11 36±10
Restrictions on Permit Purchases
d
122±33 26±7 29±8 26±7
Full Credit for Baseline Forest
Growth
e
00 0 0
No International Trading of Permits
Ideal Implementation 216±58 47±13 51±14 43±12
No Offsets from Reductions in
Other Greenhouse Gases
c
264±71 57±15 62±17 50±14
Global Trading of Permits
f
Ideal Implementation 28±8 5±1 7±2 6±2
Cartel
a
31±8 6±2 7±2 7±2
Plus: Restrictions on Permit
Sales
b
49±13
9±2 11±3 11±3
Restrictions on Permit Purchases
d
122±33 26±7 29±8 26±7
Plus: No Offsets from
Reductions in Other
Greenhouse Gases
c
147±40 32±9 34±9 31±8
SOURCES: Author’s calculations.
NOTES: CDM=Clean Development Mechanism, by which Annex B countries can take credits for projects that reduce
emissions in non-Annex B countries.
Estimates include a range of possible error, reflecting uncertainty about exactly how much businesses, consumers
and government would adjust their energy usage in response to higher prices.
To convert 1997 dollars into 2002 dollars, multiply by 1.085.
mtc = metric tons of carbon
a. The countries of the former Soviet Union and eastern Europe are assumed to limit exports of permits such that the
domestic price of emissions permits is one-half the international price.
b. The countries of the former Soviet Union and eastern Europe are not permitted to sell permits they receive in excess of
their baseline emissions. This is equivalent to reducing their allocation of permits to baseline levels.
c. Countries cannot offset emissions of carbon dioxide by reducing emissions of other greenhouse gases below target
levels.
d. Each country must achieve at least 65 percent of its obligation to reduce emissions domestically.
e. Each country receives credit for the change in carbon stocks on cropland, grazing land, and forests, except those not
available or appropriate for wood production.
f. The global trading cases assume partial exercise of market power by permit-exporting countries within Annex B. Non-
Annex B countries are assumed not to control emissions of other greenhouse gases, and their caps are assumed equal to
their baseline emissions of carbon dioxide.
85
Table 5-1. Impact of Emissions Reductions on U.S. GDP and Consumption in 2010,
with No International Permit Trading and Non-Auctioned Permits
(Percentage Change from Baseline)
Percentage Change in Memo:
Permit Price
(1997 dollars per
mtc)
GDP Consumption
General Equilibrium Models
AIM -0.5 -0.4
a
184
CETA -1.9 -0.6 201
G-Cubed -0.4 1.4 91
GTEM -2.0 -2.1 389
MERGE -1.0 -1.1 286
MS-MRT -1.4 -0.4 287
RICE -1.0 -0.2 184
SGM-PNNL n.a. -0.7
a
189
Macroeconometric Models
DRI
b
-2.9 -2.9 254
EIA -4.2 -3.1 355
EIA early start -1.2 -0.5 322
Oxford -1.8
c
-2.5
a
407
WEFA -3.2 -1.8 270
SOURCE: Author’s calculations, using: DRI/McGraw-Hill,
The Impact of Carbon Mitigation Strategies on Energy
Markets, the National Economy, Industry, and Regional Economies
(study prepared for UMWA-BCOA,
Lexington, MA, July 1997); Energy Information Administration,
Analysis of the Impacts of an Early Start
for Compliance with the Kyoto Protocol
(July 1999); Energy Information Administration,
Impacts of the
Kyoto Protocol on U.S. Energy Markets and Economic Activity
(October 1998); William D. Nordhaus and
Joseph Boyer,
Warming the World: Economic Models of Global Warming
(Cambridge, MA: MIT Press,
2000); WEFA, Inc.,
Global Warming: The High Cost of the Kyoto Protocol, National and State Impacts
(Eddystone, PA: WEFA, Inc., 1998); John P. Weyant, ed.,
The Costs of the Kyoto Protocol: A Multi-Model
Evaluation, Special Issue of the Energy Journal
(Cleveland, OH: Energy Economics Educational
Foundation, Inc., 1999); and personal communication from John Weyant.
NOTES: Consumption is private consumption plus government consumption and investment, unless noted otherwise.
Except for G-Cubed, general equilibrium models assume that unemployment and inflation cannot vary from baseline
levels. Macroeconometric models and G-Cubed assume that unemployment and inflation can vary from baseline
levels.
Estimates assume emissions of carbon dioxide in 2010 are cut to 93% of 1990 levels.
Studies using the EPPA, SGM and WorldScan models do not publish changes in GDP, and only the SGM-PNNL
study publishes the change in consumption.
n.a.=not available
mtc = metric tons of carbon
a. Percentage change in private consumption only.
b. Data for DRI are derived from a weighted average of two scenarios assuming the Kyoto target is 90 percent and 100
percent, respectively, of 1990 emissions.
c. The 1.8 percent decline in actual GDP is smaller than the 2.5 percent decline in potential GDP shown in the Oxford study.
86
Table 5-2. Model Estimates of the Cost of Emissions Permits in the United States in 2010, With
and Without International Permit Trading
(Billions of 1997 dollars)
No
International
Permit
Trading
Annex B Trading of Permits
Model Domestically Allocated
Permits
Permit
Purchases
All Permits
Value
Change from
No Trading Value
Change from
No Trading
AIM 218 93 -125 16 109 -109
CETA 253 69 -184 20 89 -164
EIA 441 207
a
-235
a
36
a
243
a
-198
a
EPPA 297 117 -180 15 132 -165
G-Cubed 114 80 -34 10 90 -24
GTEM 524 173 -351 41 214 -310
MERGE 373 152 -221 27 179 -194
MS-MRT 359 118 -242 29 146 -213
Oxford 516 281 -235 25 307 -209
RICE 237 53 -183 13 67 -170
SGM-Admin 291
b
82
b
-209
b
17
b
99
b
-192
b
SGM-PNNL 291
b
127
b
-165
b
20
b
147
b
-144
b
WorldScan 136 32 -104 6 38 -98
SOURCES: Author’s calculations, using: Council of Economic Advisors,
The Kyoto Protocol and the President’s Policies
to Address Climate Change
(July 1998); Energy Information Administration,
Impacts of the Kyoto Protocol
on U.S. Energy Markets and Economic Activity
(October 1998); William D. Nordhaus and Joseph Boyer,
Warming the World: Economic Models of Global Warming
(Cambridge, MA: MIT Press, 2000); John P.
Weyant, ed.,
The Costs of the Kyoto Protocol: A Multi-Model Evaluation, Special Issue of the Energy Journal
(Cleveland, OH: Energy Economics Educational Foundation, Inc., 1999); and personal communication from John
Weyant.
NOTES: With no international trading of permits the value of domestically allocated permits equals the value of all permits.
Unless otherwise indicated, figures are for the value of carbon dioxide permits only.
n.a.= not available
To convert 1997 dollars into 2002 dollars, multiply by 1.085.
a. Under Annex B trading of permits, U.S. emissions are reduced to a level 9 percent above 1990 levels.
b. These figures include the value of permits for all greenhouse gases.
87
Table 5-3. Impact of Emissions Reductions on U.S. GDP and Consumption in 2010, with
International Permit Trading and Non-Auctioned Permits
(Percentage Change from Baseline)
Annex B Trading of Permits Global Trading of Permits
GDP Consumption GDP Consumption
General Equilibrium Models
AIM -0.3 -0.3
a
-0.2 -0.2
a
CETA -0.7 -0.4 -0.4 -0.3
G-Cubed -0.2 1.0 -0.1 0.6
GTEM -0.4 -1.0 n.a. n.a.
MERGE -0.6 -0.5 -0.3 -0.2
a
MS-MRT -0.9 -0.3 -0.4 -0.1
RICE -0.3 -0.1 -0.1 -0.1
SGM-PNNL n.a. -0.4
a
n.a. -0.1
a
Macroeconometric Models
DRI -1.1 -1.6 n.a. n.a.
EIA
b
-2.0 -1.7 -1.0 -0.9
EIA early start
b
-0.7 -0.4 -0.5 -0.4
Oxford -1.0
c
-1.4
a
n.a. n.a.
SOURCE: Author’s calculations, using: Energy Information Administration,
Impacts of the Kyoto Protocol on U.S.
Energy Markets and Economic Activity
(October 1998); William D. Nordhaus and Joseph Boyer,
Warming the World: Economic Models of Global Warming
(Cambridge, MA: MIT Press, 2000);
Standard & Poor’s DRI,
The Impact of Meeting the Kyoto Protocol on Energy Markets and the
Economy
(Lexington, MA: Standard & Poor’s DRI, 1998); WEFA, Inc.,
Global Warming: The High
Cost of the Kyoto Protocol, National and State Impacts
(Eddystone, PA: WEFA, Inc., 1998); John P.
Weyant, ed.,
The Costs of the Kyoto Protocol: A Multi-Model Evaluation, Special Issue of the Energy
Journal
(Cleveland, OH: Energy Economics Educational Foundation, Inc., 1999); and personal
communication from John Weyant.
NOTES: Consumption is private consumption plus government consumption and investment, unless noted
otherwise.
Except for G-Cubed, general equilibrium models assume that unemployment and inflation cannot vary from
baseline levels. Macroeconometric models and G-Cubed assume that unemployment and inflation can vary
from baseline levels.
Studies using the EPPA, SGM and WorldScan models do not publish changes in GDP, and only the
SGM-PNNL study publishes the change in consumption.
n.a.=not available
a. Percentage change in private consumption only.
b. EIA does not specify which of its scenarios correspond to Annex B and global trading of permits. This table uses the
EIA scenario in which U.S. emissions of CO
2
are 9 percent above 1990 levels in 2010 for Annex B trading and the
scenario in which U.S. emissions of CO
2
are 24 percent above 1990 levels in 2010 for global trading.
c. The 1.0 percent decline in actual GDP is smaller than the 1.4 percent decline in potential GDP shown in the Oxford
study.
88
Table 5-4. Model Synthesis Estimates of the Effect of Emissions Reductions on U.S. GDP
and Consumption in 2010 Under Various Scenarios, with Non-Auctioned
Permits
(Percentage Change from Baseline)
Percentage Change in
Scenario GDP Consumption
Kyoto-Consistent Scenarios
Ideal Implementation -0.5±0.1 -0.4±0.1
Cartel
a
-0.6±0.2 -0.5±0.1
Plus: No CDM -0.6±0.2 -0.5±0.1
Plus: Restrictions on Permit Sales
b
-1.1±0.3 -0.9±0.2
Plus: No Offsets from
Reductions in Other
Greenhouse Gases
c
-1.2±0.3 -1.0±0.3
Restrictions on Permit Purchases
d
-1.0±0.3 -0.7±0.2
Full Credit for Baseline Forest Growth
e
00
No International Trading of Permits
Ideal Implementation -1.7±0.5 -1.2±0.3
No Offsets from Reductions in Other
Greenhouse Gases
d
-1.8±0.5 -1.2±0.3
Global Trading of Permits
f
Ideal Implementation -0.2 -0.2
Cartel
a
-0.3±0.1 -0.2
Plus: Restrictions on Permit Sales
b
-0.4±0.1 -0.4±0.1
Restrictions on Permit Purchases
d
-0.9±0.2 -0.6±0.2
Plus: No Offsets from Reductions in
Other Greenhouse Gases
c
-0.9±0.2 -0.6±0.2
SOURCE: Author’s calculations.
NOTES: CDM= Clean Development Mechanism, by which Annex B countries can get credits for projects that reduce
emissions in non-Annex B countries.
Estimates include a range of possible error, reflecting uncertainty about exactly how much businesses,
consumers and government would adjust their energy usage in response to higher prices.
a. Eastern Europe and the former Soviet Union are assumed to limit exports of permits such that their domestic price
of emissions permits is one-half the international price.
b. The countries of the former Soviet Union and eastern Europe are not permitted to sell permits they receive in
excess of their baseline emissions. This is equivalent to reducing their allocation of permits to baseline levels.
c. Countries cannot offset emissions of carbon dioxide by reducing emissions of other greenhouse gases below
target levels.
d. Each country must achieve at least 65 percent of its obligation to reduce emissions domestically.
e. Each country receives credit for the change in carbon stocks on cropland, grazing land, and forests, except those
not available or appropriate for wood production.
f. The global trading cases assume partial exercise of market power by permit-exporting countries within Annex B.
Also, non-Annex B countries are assumed not to control emissions of other greenhouse gases.
89
Table 5-5. Model Estimates of the Value of Emissions Permits Allocated and Used in the
United States in 2010
(Billions of 1997 dollars)
No International
Permit Trading
Annex B Trading of
Permits
Global Permit Trading
Model Allocated and Used Allocated Used Allocated Used
AIM 218 93 109 54 67
CETA 253 69 89 39 52
DRI 314
a
150
b
167
b
n.a. n.a.
EIA 441 207
c
243
c
85
d
114
d
EPPA 297 117 132 n.a. n.a.
G-Cubed 114 80 90 30 40
GTEM 524 173 214 n.a. n.a.
MERGE 373 152 179 105 130
MS-MRT 359 118 146 41 55
Oxford 516 281 307 n.a. n.a.
RICE 237 53 67 17 23
SGM-Admin 291
e
82
e
99
e
34
e
44
e
SGM-PNNL 291
e
127
e
147
e
45
e
59
e
WEFA 337 n.a. n.a. n.a. n.a.
WorldScan 136 32 38 n.a. n.a.
SOURCES: Author’s calculations, using: Council of Economic Advisors,
The Kyoto Protocol and the President’s Policies
to Address Climate Change
(July 1998); DRI/McGraw-Hill,
The Impact of Carbon Mitigation Strategies on
Energy Markets, the National Economy, Industry, and Regional Economies
(study prepared for UMWA-
BCOA, Lexington, MA, July 1997); Energy Information Administration,
Impacts of the Kyoto Protocol on U.S.
Energy Markets and Economic Activity
(October 1998); William D. Nordhaus and Joseph Boyer,
Warming the
World: Economic Models of Global Warming
(Cambridge, MA: MIT Press, 2000); Standard & Poor’s DRI,
The
Impact of Meeting the Kyoto Protocol on Energy Markets and the Economy
(Lexington, MA: Standard &
Poor’s DRI, 1998); WEFA, Inc.,
Global Warming: The High Cost of the Kyoto Protocol, National and State
Impacts
(Eddystone, PA: WEFA, Inc., 1998); John P. Weyant, ed.,
The Costs of the Kyoto Protocol: A Multi-
Model Evaluation, Special Issue of the Energy Journal
(Cleveland, OH: Energy Economics Educational
Foundation, Inc., 1999); and personal communication from John Weyant.
NOTES: With no international trading of permits, the value of permits allocated to U.S. businesses and households equals
the value of permits used by U.S. businesses and households.
Unless otherwise indicated, figures are for the value of carbon dioxide permits only.
n.a.= not available
To convert 1997 dollars into 2002 dollars, multiply by 1.085.
a. Data are derived from a weighted average of two scenarios assuming the Kyoto target is 90 percent and 100 percent,
respectively, of 1990 emissions.
b. Case 2 in DRI’s study of Annex B trading of permits.
c. U.S. emissions are reduced to a level 9 percent above 1990 levels.
d. U.S. emissions are reduced to a level 24 percent above 1990 levels.
e. These figures include the value of permits for all greenhouse gases.
90
Table 5-6. Model Synthesis Estimates of the Value of Emissions Permits Allocated and
Used in the United States in 2010
Allocated Used
Scenario Billions of
1997 Dollars
Percent
of GDP
Billions of
1997 Dollars
Percent
of GDP
Kyoto-Consistent Scenarios
Ideal Implementation 86 0.7 108 0.9
Cartel
a
106 0.9 131 1.1
Plus: No CDM 116 1.0 142 1.2
Plus: Restrictions on
Permit Sales
b
208 1.7 231 1.9
Plus: No Offsets from
Reductions in Other
Greenhouse Gases
c
223 1.8 245 2.0
Restrictions on Permit Purchases
d
186-211
e
1.5-1.7 211 1.7
No International Trading of Permits
Ideal Implementation 329 2.7 329 2.7
No Offsets from Reductions in Other
Greenhouse Gases
c
331 2.7 331 2.7
Global Trading of Permits
f
Ideal Implementation 43 0.3 56 0.5
Cartel
a
48 0.4 63 0.5
Plus: Restrictions on Permit
Sales
b
74 0.6 94 0.8
Restrictions on Permit Purchases
d
186-211
e
1.5-1.7 211 1.5-1.7
Plus: No Offsets from
Reductions in Other
Greenhouse Gases
c
184-210
e
1.5-1.7 210 1.7
SOURCE: Author’s calculations.
NOTES: CDM= Clean Development Mechanism, by which Annex B countries can get credits for projects that
reduce emissions in non-Annex B countries.
To convert 1997 dollars into 2002 dollars, multiply by 1.085.
a. Eastern Europe and the former Soviet Union are assumed to limit exports of permits such that the domestic price of
emissions permits is one-half the international price.
b. The countries of the former Soviet Union and eastern Europe are not permitted to sell permits they receive in
excess of their baseline emissions. This is equivalent to reducing their allocation of permits to baseline levels.
c. Countries cannot offset emissions of carbon dioxide by reducing emissions of other greenhouse gases below target
levels.
d. Each country must achieve at least 65 percent of its obligation to reduce emissions domestically.
e. These estimates include the value of import quotas. The lower number in the range assumes that the United States
buys foreign permits at the domestic U.S. permit price. The higher number assumes that the United States buys
permits in their domestic price in permit-exporting countries.
f. The global trading cases assume partial exercise of market power by permit-exporting countries within Annex B.
Also, non-Annex B countries are assumed not to control emissions of other greenhouse gases.
91
Table B-1.
Adjustments Made to Model Estimates of Price Sensitivity in Constructing the Model Synthesis Estimates
Model
Model
Estimate of
Price
Sensitivity
Multiplicative Adjustment for
Adjusted
Price
Sensitivity
Final
Demand
for
Energy
Too
Sensitive
Reductions
in
Emissions
Exceeding
Reductions
in Carbon-
Energy
Usage
Implicit
Increases in
Nuclear and
Hydroelectric
Power That Are
Too
Responses to
Higher Energy
Prices That
Large Small Begin
Too
Late
Are
Ended
Too
Soon
AIM -0.68 0.91 -0.62
CETA -0.67 0.77 -0.52
DRI -0.49 1.06 -0.52
EIA -0.41 1.16 -0.47
EPPA -0.69 0.84 0.83 -0.51
G-Cubed -1.54 0.73 0.78 0.82 -0.72
GTEM -0.40 1.06 -0.42
JWS -1.28 0.89 0.80 0.77 -0.71
MERGE -0.53 -0.53
MS-MRT -0.51 -0.51
Oxford -0.42 1.06 -0.44
SGM-Ad-
ministration
-0.68 0.93 -0.64
SGM-PNNL -0.69 0.93 -0.65
WEFA -0.42 1.07 -0.45
SOURCES: Table 3-2 and author’s calculations.
NOTES: Price sensitivity provides an estimate of how energy users adjust their use of carbon-based energy in response to changes in its
price (see Appendix A for details). That sensitivity is negative because energy use falls as its price rises. Larger absolute values
indicate a larger response.
Except for JWS, all calculations assume that U.S. emissions in 2010 are reduced 7 percent below 1990 levels. JWS assumes
U.S. emissions in 2010 are reduced to 1990 levels.
RICE and WorldScan models assume a unit elasticity of demand for energy. Because it is difficult to know how to adjust
estimates of price sensitivity from these models for this factor, these models are not included in the table.
Figure 5-1. Model Estimates of Permit Price and Percent Loss in GDP in 2010,
with No International Permit Trading
MS-MRT
CETA
RICE
AIM
G-Cubed
MERGE
GTEM
JWS
DRI
WEFA
EIA
Oxford
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
0
50
100
150
200
250
300
350
400
450
Permit Price (1997 Dollars per Metric Ton of Carbon)
Percent Loss in Real GDP
General Equilibrium Models
Macroeconometric Models
Regression Line Relating Percent Loss in GDP to Permit Price for General Equilibrium Models with Kyoto Consistent Caps
Maximum and Minimum Ratios of GDP Loss to Permit Price
Note: JWS assumes a cap equal to 1990 emissions, instead of a Ky
oto
-
consistent cap.
92
