Hearings and Business Meetings
September 20, 2005
SD-366 Energy Committee Hearing Room 10:00 AM
Dr. Anne Smith
Prepared Statement of
Anne E. Smith, Ph.D.
Committee on Energy and Natural Resources
United States Senate
September 20, 2005
Mr. Chairman and Members of the Committee:
Thank you for your invitation to participate in today’s hearing. I am Anne Smith, and I am
a Vice President of CRA International. Starting with my Ph.D. thesis in economics at
Stanford University, I have spent the past twenty-five years assessing the most costeffective
ways to design policies for managing environmental risks. For the past fifteen
years I have focused my attention on the design of policies to address climate change risks,
with a particular interest in the implications of different ways of implementing greenhouse
(GHG) gas emissions trading programs. I thank you for the opportunity to share my
findings and climate policy design insights with you. My written and oral testimony today
is a statement of my own research and opinions, and does not represent a position of my
company, CRA International.
I would like to start by summarizing what I think are the most important and overarching
considerations that should be accounted for in devising a sound and effective policy to
mitigate risks of climate change. I will then provide a basis for these points, present more
extensive detail on the trade-offs in policy design alternatives, and summarize results of
analysis my colleagues and I have done of the comparative costs and effectiveness of
proposals now before the Congress.
The key points that I have to offer about designing an effective climate change policy are:
• The linkage between near-term domestic GHG reductions and real reduction of
climate change risk is, for all practical purposes, nonexistent. Near-term domestic
controls cannot have any meaningful impact on global emission levels at any cost
that is currently deemed realistic. Such policies also will not stimulate the kinds of
technological progress necessary to enable meaningful emissions reductions later on
(because one can expect that carbon prices will be driven to the lowest level
necessary to incentivize adoption of important new technology – a level that is too
low to provide innovators with a return of their one-time investment cost).
• The current debate about how to impose ineffectual near-term controls is
encouraging policy makers to neglect much more important, more urgently needed
actions for reducing climate change risks. The top priority for climate change
policy should be a greatly expanded government-funded research and development
(R&D) program, along with concerted efforts to reduce barriers to technology
transfer to key developing countries. Neither of these will be easy to accomplish
effectively, yet they are receiving minimal attention by policy makers.
• Developing new technologies is crucial and it will require long-run, high-risk, highcost
R&D to produce radically new GHG-free energy sources. Even with
moderately expensive GHG limits, the private sector will under-invest in this kind
of R&D, and only government can provide the needed R&D investment. The
existing climate policy proposals, including the McCain/Lieberman (M/L) Bill and
the NCEP or Bingaman proposals, focus on providing subsidies to existing
technologies rather than R&D aimed at developing new technologies. New
government efforts to pick winning technologies and subsidize their deployment
would probably undermine the cost-effectiveness of any emissions control program,
without producing the forward-looking R&D that we really need.
• Although no near-term emissions control program will have much impact on
solving the climate problem, a price on carbon in the near-term can be justified as a
supplement to a meaningful R&D mission once that mission has clearly defined
targets for success. The near-term control program’s role would be to stimulate
emissions reductions that can be achieved now more cheaply than the present value
of future control costs targeted by the R&D program; the maximum near-term
carbon price could therefore be determined by discounting the R&D program’s
defined targets for technology costs and dates of commercial availability.
• The design of such a policy for near-term emissions control matters tremendously.
CRA’s modeling work and the economics literature indicate the relative costeffectiveness
of the various options for the climate change situation.
a. Hard caps are the most costly and least desirable option.
b. The safety valve approach and carbon taxes are alternatives to hard caps that
are much less costly, and that are more consistent with the inherently
subsidiary role of any near-term reductions program. (The contrast between
a safety valve and hard cap approach is especially evident in my comparison
below of results of CRA’s modeling of the McCain/Lieberman Bill and the
cap program of the Bingaman Amendment.)
c. One factor highlighted by CRA’s work but often slighted in other analyses
is the possibility of using allowances to limit the costs of controls. Domestic
GHG controls will cause small but not trivial losses of government revenue.
Auctioning some of the allowances and using proceeds to offset other
expected reductions in Federal revenue would noticeably reduce the
program’s total cost to society.
d. There are no simple analytical methods for determining allocations of
allowances to individual companies or sectors to equitably mitigate the
financial impacts of the policy.
e. There is no allocation design that can make all affected parties better off
under a cap-and-trade or other carbon pricing policy.
f. The inherent complexity of a safety valve approach does not appear to be
justified compared to a simpler carbon tax. A carbon tax would provide
identical emissions reduction incentives at identical costs to those of the
safety valve proposal without the political, institutional, and analytical
complications apparent in today’s safety valve proposals.
To provide a foundation supporting the above statements, I will begin with a review of the
basic elements of climate science and projections of future greenhouse gas emissions that
are relevant to economic questions about the design of climate policies. In section 2, I will
describe the range of potential policy designs, which include carbon-pricing schemes and
technology strategies. Section 3 will focus on just the carbon-pricing approaches in more
detail, and will include a comparative analysis that my colleagues and I have done of the
costs and effectiveness of proposals now before the Congress. I will address costs and
risks to the economy from different policy designs, the ability of economically feasible
mandatory caps on emissions to accomplish long-term climate goals, the role of allocations
in policy design, and alternatives to “mandatory limits on greenhouse gas emissions.” In
Section 4, I will turn to technology strategies. I will explain the reasons for my conclusion
that the most important first step for the Congress to take in developing a cost-effective US
climate policy is to provide incentives for R&D into new energy technologies.
In all of the following, I wish to be clear that I use the term R&D as a distinctly different
concept from providing subsidies for the initial uptake of existing but yet-to-be deployed
technologies. By R&D, I mean investment to create technologies that do not exist today,
and which would require major new scientific breakthroughs before they could become an
option that any private entity might consider proposing in a competition for actual
implementation under a subsidy program. The R&D may entail basic science as well as
work that is identifiably on an energy technology with low or zero carbon emissions.
Subsidies are aimed at bringing technologies into the market, and by definition, such
technologies must be already reasonably well developed, if not yet cost-effective to use
under current prices without supporting funding. There may be a sometimes unclear line
dividing the two, but it is clear that we do not yet have enough forms of energy
technologies that could, as a group, provide a carbon-free energy economy at any
reasonable cost. Creating that capability should be the mission of an R&D program.
1. Key Points from Climate Science and Global Emissions Scenarios
The key points from climate science and emissions scenarios that are critical to the
economic analysis of policy options are:
• Increases in global average temperatures are related to the concentration of
greenhouse gases in the atmosphere. Once emitted, greenhouse gases remain in the
atmosphere for many decades, so cumulative emissions over a long period of time
determine changes in greenhouse gas concentrations. As a result, climate change
risk is a function of cumulative greenhouse gas emissions, not emissions in any
• Discussions of long-term objectives for climate policy usually focus on stabilizing
greenhouse gas concentrations at some level, so as to limit temperature increases.
The concentration of greenhouse gases in the atmosphere will continue to increase
as long as there are net additions of greenhouse gases. To achieve stabilization of
concentrations and temperature at any level will require that average economy-wide
greenhouse gas emissions be reduced to nearly zero.
• Given the scale of projected increases in global greenhouse gas emissions,
achieving zero net carbon emissions by the middle of the next century will require
producing at least as much energy as is now produced from all sources by means of
processes that have near-zero net carbon emissions. It is not possible to accomplish
this with current technologies at anything close to the current or projected cost of
energy produced from oil, natural gas, and coal.
• Within the next decade or two, developing countries will overtake the industrial
world in total greenhouse gas emissions, so that by 2025 more than half of global
annual emissions of greenhouse gases will be coming from developing countries.
Thus no long-term objective of climate policy can be achieved without effective
actions to reduce emissions from developing countries. Moreover, comparison of
greenhouse gas intensity between developing and industrial countries suggests that
there is a large potential for near-term emission reductions in developing countries
at costs far lower than comparable emission reductions in the United States and
other industrial countries.
These features of the climate problem have some very strong implications for policy
design. Since only cumulative emissions over long time periods matter for climate risk,
mandatory caps that place specific limits on near-term emissions in each year create
significant cost risks without accompanying benefits. Near-term limits on greenhouse gas
emissions require the use of current technology for reducing greenhouse gas emissions, and
as I will discuss in Section 4, they provide no credible incentive for research and
development aimed at wholly new and more affordable technologies.
Nearly-zero greenhouse gas emissions cannot be achieved with current technologies
without massive disruption to standards of living. Once technologies are developed that
can make massive emissions cuts affordable (even if still quite costly) then it will be
possible to “make up for” reductions that we might not undertake today. Therefore the
only reductions in emissions that make sense economically until zero-carbon energy
becomes affordable as the mainstay of our energy system are those that are very cheap
These considerations suggest that the most important long-term feature of any policy
initiative is the impact it will have on investment in R&D and the development of new
technologies to provide essentially carbon-free energy at an affordable cost. For near-term
emission reductions, the most cost-effective emission reductions available today are in
1 Their cost should be less than the present value of the cost of “making up” for them when the zero-carbon
economy becomes viable. For example, if nearly all GHG emissions could be eliminated or offset at
$25/tonne CO2-equivalent starting in 2050 (i.e., $93/tonne carbon-equivalent), then the most that it makes
economic sense to pay for emissions reductions in 2010 is about $3.60/tonne CO2-eq. (or $13/tonne carboneq.),
using a 5% real discount rate.
developing countries, placing a high priority on near-term control policies to bring about
changes in how energy is used in developing countries.
2. Overview of Range of Available Policy Approaches
Proposed approaches for climate change policies that involve a commitment by the
government to bring about changes in future greenhouse gas emissions include:
• Pure cap-and-trade
• Cap-and-trade with a safety valve
• Carbon tax
• An R&D-focused “technology strategy”
• Market transformation and technology transfer in developing countries
These policy approaches form a continuum, all of which can be implemented in a marketbased
manner. At one end of the scale are policy designs that impose specific, rigid limits
on greenhouse gas emissions on specified dates. These are the pure cap-and-trade
programs, which place a cap on emissions and allow trading of allowances between
regulated parties to create an incentive for choice of the most cost-effective mitigation
options. Much attention has been paid to these designs, which have been used successfully
in other environmental areas such as the Acid Rain program (Title IV) of the Clean Air
Act. The McCain/Lieberman amendment to the 2005 Senate Energy Bill (S.1151) falls in
The rigidity of emission limits is progressively loosened in proposals for combining capand-
trade with a “safety valve” or for simply using a carbon tax to penalize the use of all
fossil fuels in proportion to their carbon content. Both safety valve and carbon tax
approaches avoid the imposition of a rigid cap, and instead rely on the economic incentive
of putting a price on carbon emissions to achieve changes in emissions levels.
An R&D-focused technology strategy would commit the Federal government to supporting
the research to create new technologies whose adoption in the future will enable much
larger and more cost-effective emission reductions than are possible today; this also can,
and should be designed in a market-based manner.
Thus, even now, the Congress is looking at a continuum of proposals, with the most rigid
being the M/L Bill with its specific targets and timetables for near-term reductions, and the
least rigid and potentially most cost-effective being a focus on devising and implementing a
major and comprehensive R&D program to produce affordable, zero-emitting technology
that will be possible to adopt on a massive scale, throughout our economy and that of the
The pure cap-and-trade approach places the highest priority on achieving fixed and
predictable emission reductions, and accepts whatever the cost of achieving those emission
reductions may be. The pure carbon tax limits the cost of achieving emission reductions to
be no greater, per unit of carbon removed, than the tax. The emission reduction achievable
from a carbon tax is uncertain, because it depends on how much emission reduction is
possible at a cost equal to or less than the tax. Thus the carbon tax places the highest
priority on cost containment, while tolerating some uncertainty in the level of emission
reductions to be achieved. The safety valve becomes indistinguishable from a carbon tax
once the price limit on emission allowances is reached.
Technology policy and policies toward developing countries address the two features of
climate policy that are not addressed by mandatory limits on near-term emissions. These, I
will suggest, offer far more potential for cost-effective emission reductions in the near and
long term, and are appropriate places for Congress to consider immediate action.
This Congress has considered proposals in four out of these five categories. The proposed
McCain/Lieberman amendment to the Senate Energy Bill of 2005 fell in the first category,
of mandatory caps. Proposals by the National Commission on Energy Policy (NCEP) and
Senator Bingaman (S.A. 868) fall into the second, safety valve, category. An approach to
reducing emissions from developing countries was passed into law as Title XVI of the
Energy Policy Act of 2005, along with some of the elements of a technology strategy.
Only carbon taxes per se are not talked about in Washington, but the choice between
carbon caps and carbon taxes is a very important one in the literature on climate policy.
I do not believe it is appropriate to narrow consideration at this time to only “mandatory”
programs in the sense of binding caps on specific schedules, which if taken literally would
include only the McCain/Lieberman proposal. Although used and justifiable in other
environmental areas, this is not the most suitable policy design for climate change. Costs
of large near-term reductions are high, mandatory caps create large risks and uncertainty
about cost, and even mandatory caps cannot provide a credible incentive for R&D to
develop needed technologies. Safety valve proposals, which become indistinguishable
from carbon taxes once the safety valve becomes effective, offer additional flexibility and
need not imply greater climate risks.
Therefore, I would encourage you to include in your thinking about “mandatory” programs
all policies that force households and businesses to take into account a cost of greenhouse
gas emissions. This would recognize that carbon taxes, as well as the NCEP and Senator
Bingaman’s approaches, are all “mandatory” approaches to emissions reductions.
However, none of the mandatory programs aimed at putting limits on future carbon
emissions will provide a credible incentive for R&D or actions by developing countries.
Such mandatory programs are not the only actions that can be taken today. I have
concluded that commitments to support technology development and bring about change in
the rate of growth of emissions from developing countries are a more effective and
appropriate focus for current action on climate policy. To my mind, it makes the most
economic sense to start where resources committed to mitigation of climate change can
achieve the greatest gains, considering both near-term and long-term outcomes as a whole.
Therefore, we should start with a clearly articulated and carefully implemented R&D
program for developing affordable zero-carbon emitting technologies. Neither the
technologies nor the necessary R&D program to create them presently exists.
The first three approaches (pure cap-and-trade, cap-and-trade with a safety valve, and
carbon taxes) all function by placing a direct price on GHG emissions. In the next section, I
will discuss each of the individually, highlighting their respective strengths and
weaknesses. I will then provide comparisons of the outcomes under a pure cap-and-trade
proposal (i.e., the McCain/Lieberman proposal) to those of a proposal that directly limits
costs rather than emissions (i.e., the GHG cap program of the Bingaman/NCEP proposal)
to highlight how they tend to differ in their impacts. I complete the next section by
addressing a number of issues related to allocation of allowances that I feel are greatly
misunderstood, yet extremely important if a cap-and-trade approach is selected instead of a
carbon tax approach for imposing a price on carbon emissions. The following, and last
section, will then turn to an important limitation of all approaches that directly price
emissions in the unique situation of climate change policy, and the reasons that an R&Dfocused
technology strategy needs to be the first and foremost consideration in any policy
to address climate change risks. It is my view that none of the proposed policies to date
properly address this R&D need. In general, they have confused subsidies with the need
for R&D on new technologies, and for the most part the subsidy programs that have been
proposed are also unnecessary for motivating a least-cost response under a carbon-pricing
3. Approaches that Place a Direct Price on Emissions
A. Pure cap-and-trade with rigid emission limits
Emission caps are enforced, under cap-and-trade proposals, by distributing a set of
emission allowances, limited to the quantitative cap. These emission allowances can be
traded, so that emission reductions will occur where they are most cost-effective given
current technology. The cost of a cap-and-trade program depends on how tightly the caps
are set initially and how they are tightened over time. How emission allowances are
distributed also affects the overall economic impact of this policy approach.
Near-term caps, such as those proposed by Senators McCain and Lieberman, can only be
met through use of costly measures based on today’s technology. This raises their costs
substantially compared to a policy sequence in which new and more affordable
technologies are developed first, so that much larger emission reductions can be achieved
at much lower cost.
Emission caps, even if never tightened, will become more expensive over time, because
energy needs are always growing as population increases and the economy expands.
Holding greenhouse gas emissions constant in the face of ever-increasing energy demand
requires going to ever more costly control options. The depth of the cuts required can be
seen by comparing business-as-usual, or current-policy emissions to emissions under the
cap. Based on the current EIA Annual Energy Outlook forecast for emissions under
current policies, the limits proposed by Senators McCain and Lieberman would require
total CO2 emissions from covered sectors to be reduced to 15% below current policy levels
in 2010 and 26% below current policy levels in 2020. Continuing the McCain/Lieberman
cap to 2050 would require a reduction of emissions to 48% below current policy levels in
that year. Tightening the cap to a level consistent with current proposals for programs that
could stabilize greenhouse gas concentrations in the atmosphere would require emissions to
be reduced to more than 80% below what CRA International projects for current policy
emissions in 2050.
The imposition of rigid limits creates unnecessary cost risks, even in the near-term, because
rigid limits can become very expensive if economic growth exceeds expectations or if costs
of measures required to reduce emissions turn out to be higher than assumed. Since
climate risks are not affected by variations in emissions from one year to the next, but only
by cumulative emissions over long time periods, these cost risks associated with rigid caps
are completely unnecessary to achieving long-term climate goals.
The perception that fixed caps create excessive cost risks is, I believe, widely shared. The
McCain/Lieberman amendment would have created specific fixed and mandatory caps.
Other policy approaches before Congress are based on a recognition that setting this kind
of mandatory cap is not the only way to take effective action to address climate change.
All the other approaches before the Congress involve market based incentives, but do not
place a rigid cap on emissions. These approaches are more suitable to the nature of the
B. Cap-and-trade with a safety valve
Combining cap-and-trade with a safety valve has the purpose of reducing the cost risk
associated with the pure cap-and-trade approach. Senator Bingaman and NCEP’s
proposals also reflect a concept called an “intensity-based” cap, but this only serves to
reduce the expected costs of the policy. The real reason that these proposals have reduced
risk of unexpectedly (and unacceptably) high costs lies wholly in the safety valve
The original concept of an “intensity-based” target is that caps would only be tightened in
relation to economic activity levels. If economic activity is high, an intensity approach
would allow a somewhat looser cap to accommodate the extra need for energy, rather than
to choke it off by having a rigid cap no matter what the level of economic activity.
However, as implemented in these two current proposals, the “intensity-based” cap would,
in fact, still be an absolute cap, computed up to ten years in advance and rigid thereafter.
Its primary novelty is that by computing a cap that is tied to economic growth rather than
historical outcomes, it would more gradually phase in the cap’s apparent stringency. This
certainly makes such a cap less costly than a tighter cap that prevents any further emissions
growth at all. However, as long as the cap is binding at all (which is the intention), there is
still uncertainty on how costly it will actually be to attain, especially given its rigidity over
ten-year periods. (A cap that is truly flexible from year to year in response to economic
activity outcomes might somewhat mitigate this cost uncertainty, but would require
continual, year-to-year updating of allocations. This updating would probably be more
detrimental than helpful in producing compliance planning certainty, while still not
assuring that costs of control would remain below some planned level.)
Nevertheless, the Bingaman and NCEP proposals do have much less cost risk than previous
cap-and-trade proposals, entirely due to the safety valve provision. The safety valve places
a ceiling on the price of carbon allowances under the cap provision. This would be
accomplished by allowing companies to achieve compliance by paying the safety valve
price to the government in lieu of turning in actual allowances that have been issued.
Alternatively, the government could issue more allowances at the safety valve price, which
would then be turned in along with originally-issued allowances. Either way, the effect of
the safety valve is to make the cap itself flexible rather than rigid. However, its flexibility
is linked to the cost of control rather than to economic activity per se.
In summary, the safety valve is a very important way of minimizing cost risk under a
carbon emissions control policy, and it does so by converting the carbon cap into a carbon
tax if the cost of control to meet the cap is higher than the pre-agreed safety valve price.
By design (and also just like a tax), this can alter the amount of emission reduction that is
achieved, thus making emissions reductions uncertain instead.
C. Carbon taxes
Once the safety valve becomes effective, the environmental outcomes and control costs
under a program based on safety valves become indistinguishable from a carbon tax.
However, a carbon tax policy would avoid creating the costs and bureaucracy associated
with allocating allowances and administering an emission trading and enforcement system.
All of these approaches – rigid caps, caps with safety valve, and carbon taxes – share a
common feature of mandatory but market-based emission limitation. They require an
emitter to pay for its legal emissions, either by purchasing an allowance, foregoing
revenues from the sale of an allowance it was allocated, or paying a tax. Each creates
revenues, and the choice must be made in designing the policy of who will collect these
revenues and how they will be used. This is the choice between auction and allocation of
allowances under a cap-and-trade system.
The safety valve involves the sale of emission allowances for a fixed price. It is equivalent
to charging a carbon tax on the use of fossil fuels, with the tax rate set for each fuel based
on its carbon content. Use of a carbon tax would also leave control of how to use the
revenues under the normal budget process. In contrast, revenues from auctioning
allowances or from selling allowances under a safety valve can be placed outside the
budget process, as they are in both the McCain-Lieberman and Bingaman proposals. Using
a free allocation of carbon allowances to compensate some of those harmed by the
imposition of limits on greenhouse gas emissions is also a use of potential revenues that
could accrue to the government, and removes decisions about that use of revenues from the
normal budget and authorization and appropriation process. This has a very important
influence on overall economic costs.
The proposed cap-and-trade and safety valve programs are likely to impose higher costs
than a carbon tax. In part, this is true because they are likely to have greater administrative
costs than an explicit tax. But more importantly, by taking revenues outside the normal
budget process, these policy designs eliminate the possibility of using some or all of the
revenues to replace taxes that would otherwise have to be raised through other federal tax
programs. As I discuss in subsection E below, not allowing revenues from allowance
auctions to be used to offset impacts of emission limitations on total government revenues
substantially increases the cost of the Bingaman and McCain/Lieberman approaches.
Otherwise, the effects of cap-and-trade with a safety valve and a carbon tax are
indistinguishable. Consumers of energy will experience increases in the cost of energy, in
one case by the price that energy producers must pay for carbon allowances and the other
by the carbon tax they must pay. The response of businesses and households to these
altered prices will be identical. Differences will arise only from how potential revenues
from the safety valve or carbon tax are utilized.
D. Analysis of the costs of mandatory caps and safety valves
In order to quantify the costs and emission impacts of the McCain/Lieberman and
Bingaman amendments, my colleagues and I have used CRA’s Multi-Region National
Model of the U.S. economy.2 This model has been used in a variety of studies over the past
10 years, and was used by the National Commission on Energy Policy in its own analysis
of the economic impacts of its proposals.
We have analyzed a range of estimates for the impacts of the McCain/Lieberman proposal
(M/L) and of the carbon cap program in the Bingaman Amendment (BA). For M/L our
range was based on assumptions about the cost at which a carbon free “backstop”
technology will become available and how that cost will drop over time; the availability
and cost of “offsets” to CO2 emissions in covered sectors; and the choices that will be made
about long-term emission limits after 2020. For the BA, at the low end of the range, we
assume that regulated non-CO2 GHGs are able to be costlessly reduced up to the point
where the marginal cost of reducing those other GHGs exceeds the safety valve price,
based on marginal abatement cost curves prepared by MIT. At the high end of the range,
we assume that non-CO2 GHGs are reduced costlessly only up to the point where they
achieve their own share of the intensity targets. Our baseline or “current policy” emissions
trajectory was based on the AEO 2005 reference case forecast of CO2 emissions, and was
not varied, though this would be another source of cost uncertainty, especially for the M/L
The form of Senator Bingaman’s carbon cap proposal that we analyzed sets a cap on
greenhouse gas (GHG) emissions from 2010 onward. The cap is to be calculated in 2006
so that it will cause greenhouse gas intensity (GHG emissions divided by GDP) to fall by
2.4% per year from 2010 to 2020, and then to fall by 2.8% per year thereafter. The
required improvements in GHG intensity are converted to fixed caps for the next decade by
multiplying the required GHG intensity times the level of GDP in each year that is
projected as of 2006.
2 For documentation of the MRN model, see http://www.crai.com/pubs/pub_3694.pdf.
We applied a “safety valve” which allows regulated entities to purchase carbon allowances
for a price of $7 per ton of CO2 in 2010, escalating at 5% per year (nominal). Both the
safety valve escalation and the annual improvement in GHG intensity can be revised by
joint resolution. The bill requires the President to report to Congress on what other
countries are doing to reduce GHG emissions as a basis for recommending such revisions.
The proposal includes some, but not all, emissions of non-CO2 GHGs in the calculation of
GHG intensity and allows banking of allowances for use in future years.
We assumed that under Senator Bingaman’s proposal a large fraction of carbon allowances
will be “allocated” to businesses that face disproportionately large negative impacts, and
that 5% in 2010, rising to 10% by 2020, of the allowances will be auctioned to provide
funding for subsidies for the development and deployment of selected energy technologies.
Sources of economic impacts. Economic impacts arise from four major sources. Direct
costs of complying with emission limitations or of adjusting energy supply and use in
response to a safety valve/carbon tax are incurred by energy producers and consumers.
These costs arise from the necessity of diverting resources from other productive uses to
reducing greenhouse gas emissions. The activities involved include substituting more
costly but lower carbon forms of energy for fossil fuels, making investments and incurring
higher costs to improve energy efficiency, and losing the benefits of foregone energy
A second set of costs arises from an increased excess burden of existing taxes. Both the
Bingaman and M/L proposals provide for allocations of allowances and specify how
revenues from allowance auctions will be utilized. They do not allow proceeds to be used
to reduce other taxes. It is widely accepted among economists who study the Federal tax
system that the current set of income, payroll and corporate taxes impose a deadweight loss
on the U.S. economy. It has been found in a number of studies that a system of emission
limits or carbon taxes that raises energy costs effectively increases the burden of existing
taxes on the economy.3 Using the revenues from sale of carbon allowances or from a
carbon tax to substitute for revenues that would otherwise be raised through conventional
taxes can reduce or eliminate this distortion. Allocating emission allowances at no cost
removes that ability to reduce the distortions of the tax system and contributes to higher
costs, as does reserving revenues for new spending programs that are created by the policy.
CRA’s analyses have revealed a need for governments to use allowance auctions under a
GHG cap to generate a certain amount of new government revenue to offset likely
reductions in existing tax revenues due to a decline in economic activity from the cost of
the policy. If such offsetting revenues are not tapped from the value of the allowances,
then governments will either have to cut services or else raise existing tax rates. The latter
action would actually exacerbate the costs of the policy, and thereby create an inefficiency
due to tax distortions even while the carbon allowance market may function in a perfectly
efficient manner in achieving cost-effective emissions reductions to meet the cap. Neither
3 On the issue of how existing tax distortions are magnified by emission limits, see Larry Goulder, Ian Parry
and Dallas Burtraw, “Revenue-Raising vs. Other Approaches to Environmental Protection: The Critical
Significance of Pre-Existing Tax Distortions,” RAND Journal of Economics, Winter 1997; and Larry Goulder
and Lans Bovenberg, “Optimal Environmental Taxation in the Presence of Other Taxes: General Equilibrium
Analyses,” American Economic Review, September, 1996.
of the proposals analyzed provides for any revenues to be used to offset tax base erosion.4
Although there are some revenues from auctions and safety valve sales, these revenue
sources are earmarked for new spending programs rather than to supplement other, falling
sources of government revenues.
A third cost element arises from the transition costs of job search which are triggered by
the changes in real wages and shift in industry structure causes by emission limits or safety
valve/carbon tax policies. This cost element shows up directly in the results as an increase
in transitional unemployment, and contributes to reduced GDP and to lower household
consumption and welfare.
Finally, since MRN is a fully dynamic computable general equilibrium model with
forward-looking expectations, the prospect of rising carbon allowance prices and future
economic impacts leads households to change their current saving and investment
behavior. Households reduce their current consumption, in order to save and provide for
higher future income to cover the increasing costs of tighter emission limits and rising
safety valve/carbon taxes. This anticipatory behavior makes future costs show up in the
present. The banking option included in both M/L and BA also encourages businesses to
undertake emission reductions in early years in excess of those required by carbon limits, in
order to avoid even higher future mitigation costs due to tightening emission limits or
higher safety valve/carbon taxes. This also contributes to costs in early years.
I provide details of our comparison of the impacts of M/L and BA in Exhibits 1 – 4 at the
end of this testimony. Generally speaking, our results suggest that M/L impacts are about 3
to 4 times larger than the impacts for the BA cap program. Key economic indicators all
follow this pattern:
• For 2010, the GDP loss under BA would be about $21 billion to $34 billion, or a
0.1% to 0.2% reduction (compared to 0.3% to 1.0% for M/L). The GDP loss
increases over time, both because the percent impact of BA increases with time, and
because GDP increases with time.
• GDP loss under BA in 2020 is $70 billion to $96 billion. (This compares to $214 to
$517 billion for M/L) It reflects a 0.3% to 0.4% reduction in 2020 GDP (compared
to 0.8% to 1.9% for M/L).
• Per household consumption losses under BA are $135 to $147 in 2010 and $147 to
$164 in 2020. (Comparable M/L losses are in the $450 to $800 range.)
• Job losses under BA in 2020 are 281,000 to 326,000 (compared to 793,000 to
1,306,000 under M/L).
• Reduction in coal output in 2020 is 8% to 11% (compared to 23% to 42% for M/L).
• Reduction in refined oil output in 2020 is 2% (compared to 6% - 13% for M/L).
4 In very approximate-terms, the share of the allowances that the government would need to offset tax base
erosion and thus avoid exacerbating policy costs appears to be between 30% and 60%. This is based on
multiple scenarios analyzed by CRA International using its MRN model, and apparently has been
corroborated by analyses by Prof. Goulder of Stanford University (personal communication).
• Carbon prices are $13 to 18/tonne C in 2010 and $21 to 29/tonne C in 2020. (M/L
carbon prices are $47-$130 in 2010and $75-$209 in 2020.)
• Under BA, carbon allowance prices hit the safety valve price in 2020 in the high
case and 2035 in the low case.
The greater cost certainty associated with the safety valve is apparent in the fact that our
M/L cost ranges are much wider than those we estimated for the BA cap. However, the
lower overall costs of the BA cap simply reflect the fact that it imposes a much less
stringent demand for near-term emissions reductions. Over the period from 2020 to 2050
BA provides emission reductions that total between one-third and two-fifths (32% to 40%)
of those provided by M/L. Since the costs of BA range from 25% to 33% of M/L, the
comparison also illustrates the law of diminishing returns, in that it costs proportionately
more to achieve the larger emission reductions required by M/L.
E. Issues in allocating allowances
Allocations versus auctions. I understand the committee is very interested in the issue of
allocation of allowances under Senator Bingaman’s proposal. This is a feature of policy
design for which there are several alternatives. Senator Bingaman distinguishes between
auction and allocation. Some allowances would be allocated to parties that suffer
disproportionate harm from emission limits, and some would be auctioned and provide
The first question that the Committee might want to consider is who should control the use
of the revenues from any auctioned or safety valve allowances. If revenues are placed in
the general fund, then Congress will retain the ability to make the decisions about how the
revenues will be utilized. This will allow Congress to consider all societal needs together,
and to balance competing needs as they evolve. To place the proceeds into a Trust Fund
that earmarks them for spending only related to climate policy is tantamount to deciding
now that climate-related spending needs to be separated from all other government
spending decisions, and given a separate, more elevated priority than all other societal
needs, including future needs that may not be anticipated at present. Public finance
practitioners generally frown on the idea of earmarking funds from particular revenue
sources to particular purposes, because the amount of money that will be collected from a
particular source is only connected loosely, if at all, to the amount that it is wise to spend
on even a related purpose. Thus earmarking is likely to produce either too much or too
little funding, and it removes the decision about how much should be spent from the
normal budget, authorization, and appropriation process.
In this regard, I also note that free allocation of allowances is not the only way to provide
for compensation of affected parties. Any compensation that can be achieved by a free
allocation formula could, in principle, be replicated under a 100% auction – it would only
require that the auction revenues be returned to companies by the same formula that would
have been used for allocations. Funds could be appropriated to provide compensation for
those disproportionately harmed, or specific tax credits could be enacted. Determining
how to make this compensation using normal budget processes would be no harder than
determining how to allocate allowances under the procedures outlined in Senator
While general principles of public finance suggest that separation of revenues from such a
policy into a Trust Fund is probably unwise, my personal research has found that such an
approach also could exacerbate the total costs of any carbon-pricing policy, and thus would
be inconsistent with principles of minimizing policy costs. Paradoxically, allocating all of
the allowances at no cost to affected parties, and/or using all of the proceeds from sale of
allowances to fund new spending programs, can lead to far larger costs to the economy
than necessary. This policy cost inflation can be averted by using some allowance or
carbon tax revenues to replace other taxes that would have to be raised to meet budget
targets. By allowing carbon policy revenues to flow to the general fund, Congress retains
its ability to determine how much of the proceeds from allowance sales or carbon taxes
should be used for replacement of other tax revenues that can be expected to decline under
the carbon policy.
Free allocations cannot compensate all businesses and households. Impacts on households
and industry are not determined by where regulations are put in place. An upstream system
like that in Senator Bingaman’s proposal still imposes costs on households and industries.
Not all the costs are borne by fuel suppliers, even if they are the point of regulation. All
users of energy have higher production costs. Some will be able to pass some of these
costs to their consumers, while others will have little ability to pass costs through, and the
brunt of the financial impact will be borne by their shareholders. In the end, households
cannot pass the costs on to anybody, and they ultimately bear the entire cost, as consumers
of higher cost of goods and services, and as shareholders in companies that cannot pass the
Conceptually, allocations could be used to help compensate the companies that bear an
exceptional and unfair burden. We have, in other contexts, estimated the average loss in
capital value to owners of assets in aggregated economic sectors such as the oil, gas, coal
and electricity generation sectors. However, there is no simple formula to identify exactly
which companies these are, or what amount of allocation would actually provide for an
equitable burden sharing arrangement. Companies within the same economic sector may
face diverse impacts, so that an estimate of the “average” loss of profitability for each
sector may bear no correlation to the sum of losses across the negatively affected
companies within each sector. Even if one could identify reasonable allocations to each
sector of the economy, comparable allocations to each company within a sector would have
little chance of equalizing burdens within the sector. Attempts to analytically identify
company-specific burdens within a sector would be even more challenging than attempts to
identify needs by sector, as the relevant data are not even publicly available. Thus, the
idealized concept of mitigating the impact of the rule on individual companies cannot be
estimated quantitatively at the level of detail needed to define company-level allocations,
let alone be condensed to a relatively simple formula.
It is also important to realize that the energy sectors (including non-regulated entities in the
energy sectors) are not the only sectors that will bear losses of capital value as a result of a
carbon pricing policy. All sectors of the economy will be affected to some degree, as all
are consumers of energy to varying degrees. As more and more of the needs to be
compensated are recognized, the identification of a “fair” allocations rule will become
More importantly, once it is recognized that needs for compensation include all individual
energy consumers, and not just companies, policy makers will have to realize that it is not
possible to offset losses for everyone through allocations of allowances. The total cost of a
cap-and-trade system will always exceed the total value of the allowances in that system:
• This is because companies must pay (1) to reduce emissions down to the level of
the cap and also (2) for every ton of emissions that remains after meeting the cap.
The value of the allowances equals only the second component of total costs. At
most, the government can give that entire value back to the companies by free
allocation of 100% of the allowances, but that leaves companies still incurring the
first cost component, and without any way to compensate them for that cost – which
is the real net cost to society.
• It is true that companies may be able to pass some of these two cost components on
to their customers, and so directly-regulated companies could be given more
compensation than the cost that their shareholders bear if all of the allowances were
allocated to them alone. However, this only means that a part of the net cost has
been spread to other, non-regulated parties, including consumers. They, in turn,
would require their share of the allowance allocation to be compensated for the part
of the cost that was passed to them. There is not enough value in the allowances to
cover all costs to regulated companies if they cannot pass those costs on, and
neither can that value cover all the incurred costs after they are divided up and
spread throughout the entire economy.
Thus, a carbon pricing policy will always impose a real net cost on the economy that
cannot be eliminated through any allocation formula that may be devised. All that an
allocation scheme can do is alter the companies and individual consumers that end up
bearing the burden of that cost.
These challenges in identifying fair allocations are not a result of proposing an upstream
point of compliance. They would be equally difficult under any downstream or hybrid
form of implementation. They do, however, present more prominent issues when using a
cap-and-trade approach than under a carbon tax, because the former system does require
that a specific decision be made for how to distribute the allowances. (At the same time,
needs for compensation and burden sharing would also exist under a carbon tax, and there
would also be equivalent degrees of ability to achieve such compensation under a carbon
Administrative costs and bureaucracy for small and distant emission reductions. I have
estimated that under Senator Bingaman’s proposal, the price of carbon allowances would
rise above the safety valve level between 2020 and 2035. EIA puts this somewhat earlier.
This effectively turns the Bingaman proposal into a carbon tax program, but with the much
higher costs of an administrative apparatus for issuing, enforcing, and trading carbon
allowances that doesn’t actually do anything other than impose a pre-determined price on
This leads to the question of whether it is desirable to create the bureaucracy and
administrative burden of a comprehensive national emission trading program for the small
reductions that are possible with a safety valve. The main differences between safety valve
proposals and simply establishing a federal fuels tax based on carbon content are (1) that
the safety valve has a greater administrative burden, and (2) the safety valve approach
allows revenues that would otherwise go into the normal budget process to be handed out
by an executive agency or quasi-government corporation.
Thus the government cedes the ability to set overall social priorities for the use of the
funds. Further, it sets the stage for automatically spending whatever is collected on
climate-related technologies, without regard to the need for spending at such a level.
Because it is not tied to an R&D program with clearly specified goals and a plan for
meeting those goals, much of the spending is likely to result in subsidies on investments
that would occur anyway (because they are cost-effective under the carbon price) or on
investments that are not desirable (because they are only feasible at a cost that is higher
than the safety valve price, which by definition reflects the maximum that is deemed
reasonable to spend on near-term emissions reductions).5 The use of an outside entity does
not solve the problem of creating a good R&D program; but it does mean that Congress
loses the opportunity to make those R&D spending decisions directly and transparently.
4. The Need for R&D Strategy to Be the Leading Edge of Climate Policy
A. New technology is not encouraged by mandatory limits
Although M/L has much more substantial (and uncertain) costs than the BA cap proposal,
both proposals have substantial costs. But despite these costs to the economy, neither fixed
caps nor safety valve/carbon tax policy designs can provide an adequate incentive for the
critical piece of the solution – which is creation of radically new technologies. In my
opinion, it would be better now to put resources into developing new technologies than in
forcing the use of existing technology to achieve relatively small and costly emission
reductions. Creating an effective R&D program will not be cheap, but it ultimately has to
happen if climate risks are to be reduced. The difficult decisions are how much to spend
now, and how to design programs to stimulate R&D that avoid mistakes of the past.
5 Nuclear power presents a different situation. Although the technology is nearly zero-emitting (there are
some emissions associated with its fuel cycle), available now, and cost-effective under even a modest carbon
pricing scheme, its deployment is hampered by existing policy. Removal of institutional and political barriers
to new nuclear generation might be the most important way of enabling existing nuclear generation
technology to provide cost-effective emissions reductions within the next two decades.
The subsidies to current technology embodied in BA and M/L are not likely to bring about
that change in the fundamental direction of R&D, because they are directed at the
demonstration and use of current technology. These subsidies should be carefully
distinguished from funding for R&D. Most subsidies would be unnecessary under a
carbon-pricing program, as the market price of carbon due to the cap provides the
appropriate financial incentives for the optimal use of the control methods that would then
also benefit from the subsidies. A well-designed policy to address needs for R&D in
entirely new technologies is needed, not subsidies to get existing technologies deployed in
the market place. A very different commitment is needed to create programs that will
change the direction of basic research toward creation of climate friendly, zero carbon
technologies. Subsidies for demonstration and use of currently available technologies do
not create incentives for creation of entirely new technologies.
In BA, the carbon intensity basis for mandatory caps ensures that they rise gradually, so
that there is little change in emissions for the next decade. The safety valve, by design,
takes over from the mandatory cap when its costs begin to rise. By design, the safety valve
will not stimulate the desired level of R&D. By attempting to limit cost to a level deemed
tolerable, it eliminates adequate incentives for R&D on new technology.
Nor will an adequate incentive be provided if the safety valve were eliminated, now or in
the future. This would provide a trajectory of rising allowance prices and tightening limits.
But those future policy results cannot be a credible incentive for current R&D, as I explain
B. Carbon pricing programs cannot provide credible incentives for technology
Whether cap-and-trade or a carbon tax is the policy approach taken, these mandatory
programs cannot achieve the most important need in a climate program, which is to
stimulate development of the kinds of technologies that alone can make significant
mitigation of climate risk possible in the long run.
Emission caps are not only premature and risky for the economy. They are not capable of
stimulating the kind of technology development that is an absolute necessity to achieve any
of the objectives of climate policy. Putting a stop to the continued growth of greenhouse
gas concentrations in the atmosphere requires meeting all of today’s energy needs in a way
that produces zero net carbon emissions, and does so at acceptable cost. That is not
possible with the set of technologies that exist today.
Hoffert et al. argue that “the most effective way to reduce CO2 emissions with economic
growth and equity is to develop revolutionary changes in the technology of energy
production, distribution, storage and conversion.”6 They go on to identify an entire
portfolio of technologies, suggesting that the solution will lie in achieving advances in
more than one of the following categories of research:
6M. I. Hoffert et al., “Advanced Technology Paths to Global Climate Stability: Energy for a Greenhouse
Planet” Science, Vol. 298, Nov.1, 2002, p. 981.
• wind, solar and biomass
• nuclear fission
• nuclear fusion
• hydrogen fuel cells
• energy efficiency
• carbon sequestration
Currently available technologies cannot provide sufficient or low cost reductions to meet
the GHG challenge. Developing that supply will require basic science and fundamental
breakthroughs in a number of disciplines. The magnitude of possible reductions in the next
decade or two achievable with today’s technology is dwarfed by the magnitude of
reductions that successful innovation would supply through these routes.7
Emission caps cannot provide adequate incentives. Even combined with an allowance
trading system that puts a price on emissions, fixed caps cannot provide the incentives for
the necessary technological change to occur. Thus, efforts to address climate change by
imposing costly caps or taxes in the near-term will fail to provide long-term reductions.
Additionally, if the R&D externality is being effectively addressed, implementation today
of a cap or tax that will not become stringent until a later date will provide little or no
further benefit in the form of an “announcement effect.” The only role for near-term GHG
caps or taxes would be to achieve emissions reductions that are justifiable immediately
because their cost per ton removed is less than the present value of the cost of avoided
future emission reductions that would come from the future technologies, once they
become available. Any other degree of stringency is unwarranted before R&D is
successful, and unnecessary to supplement policies that will address the fundamental
market failures associated with R&D.
Announcements of high future carbon prices to stimulate R&D are not credible, because
those carbon prices would not be necessary once technologies are developed.8 When new
technology and new capacity investments are the issue, the only policy strategies that
matter immediately are those that will increase incentives to invest in R&D, and direct the
R&D toward technologies that will create a much larger supply of carbon-free energy
alternatives at acceptable costs. Therefore, the only attribute of a cap-and-trade program
that will matter will be the future course of the cap and its implications for future allowance
7 For example, if all of the existing US natural gas-fired combined cycle generating capacity were to suddenly
be fully utilized, we estimate based on our models of the US power sector that current annual US CO2
emissions would be reduced by about 80 MMTC – about a 4% reduction in total US GHG emissions – and it
would come at a cost of about $80/tonne C, even if gas prices would not be inflated by the sudden surge in
natural gas demand.
8 These points are developed in a more rigorous fashion in W. D. Montgomery and Anne E. Smith “Price,
Quantity and Technology Strategies for Climate Change Policy.” To appear in Human-Induced Climate
Change: An Interdisciplinary Assessment, Cambridge University Press, forthcoming 2005.
None of the “mandatory” programs under consideration could stimulate the kind of R&D in
new energy technologies that is required. The “safety valve” in the NCEP program and
Senator Bingaman’s amendment is designed to provide assurance that the price of emission
allowances will not reach economically unsustainable levels. But that policy design causes
the prices to be set at a level far too low to provide an adequate incentive for private
investors to develop radically new technologies.
To motivate the large R&D investments required, it would be necessary for governments to
announce policies that will lead to high enough implicit taxes on carbon emissions to
provide an adequate expected return on R&D investment. This tax will necessarily exceed
the tax needed to induce adoption of the technology once it is developed. Once affordable
technologies are produced, a relatively low carbon tax price will be enough to motivate
companies to adopt the new technologies. That lower carbon price will not be enough to
compensate the investors who paid for the R&D, but it will be enough to get it utilized.
Even if laws passed today served to announce a future emissions tax high enough to create
such an incentive, no future Congress or Administration would keep that commitment once
the technology was developed. As in the case of patents, there is a tradeoff between
efficiency in resource allocation and providing an incentive for R&D. A carbon price
above the level necessary to induce adoption of the new technology will cause avoidable
deadweight losses as all energy supply and use decisions are distorted. Reducing implicit
carbon taxes to the lowest possible level to get the new technology deployed will always be
beneficial to the economy. Therefore, future governments will face irresistible pressure to
let the implicit tax on carbon emissions fall back to a level just sufficient to get the R&D
utilized, taking away all the rewards to innovation.
This leads to a fundamental dynamic inconsistency that makes any effort to set emission
caps or announce future carbon prices sufficient to stimulate R&D not credible. Since
private investors can understand this is the optimal strategy for government – and indeed
would likely be skeptical of the political ability of any government to proceed with what
will look like “corporate welfare” – they will not be motivated to invest in R&D by any
announcement of future climate policy.
C. Design of technology policy
What this argument demonstrates is that it is not possible to rely on caps on future
emissions, or on announcements of a safety valve or carbon tax, to motivate R&D to
develop the new technologies needed for long-term reduction of climate risk. This means
that there is an extraordinarily high priority to designing effective programs to stimulate
that R&D through incentives provided today. I would urge Congress to turn its interest in
climate policy toward a subject it knows well – how to craft a program that will lead to
effective use of private and government funds to carry out the R&D needed to provide the
radically new technologies required to stabilize concentrations of greenhouse gases and
ultimately, global climate.
D. Large opportunities for near-term emission reductions exist in developing
For near-term emission reductions, developing countries offer far larger and more costeffective
opportunity for emission reduction that mandatory emission limits on U.S.
businesses and consumers. There are a number of ways in which the U.S. Congress could
act to increase technology transfer and encourage foreign investment in developing
countries, and these actions could lead to near-term reductions in emission larger than any
of the mandatory limits on U.S. emissions under considerations.
The provisions of the McCain/Lieberman and Bingaman Amendment proposals dealing
with developing countries create no mechanism for bringing about changes in those
countries. A great deal of the difference in greenhouse gas intensity between developing
countries and industrial countries can be explained by fundamental failures of markets and
institutions in developing countries. Much more cost-effective emission reductions are
possible in the near-term through programs directed at developing countries by focusing on
fundamental institutional and market reforms to create the property rights and investment
climate required for private foreign direct investment and technology transfer.9 These
needs are already a focus of the Climate Change Title (Title XVI) of the Energy Policy Act
of 2005, which passed into law after the Bingaman Amendment was released. I believe
that approach of Title XVI should be followed, and further enhanced if necessary. The
more general and less focused provisions expressed in the Bingaman Amendment proposal
are unnecessary additions, and could distract from implementing the more focused
provisions that already exist as law.