Office of Management and Budget
February 26, 2014
Page 39
Attachment 2
NERA Damage Function Report
NERA
ECONOMIC CONSULTING
A Review of the Damage Functions Used in Estimating
the Social Cost of Carbon
Prepared for:
American Petroleum Institute
February 20, 2014
Authors *
Anne E. Smith, Ph.D.
David Harrison, Ph.D.
Meredith McPhail
* The authors acknowledge Dr. Noah Kaufman and Dr. Sugandha Tuladhar for helpful and
insightful comments. The opinions expressed herein do not necessarily represent the views of
NERA Economic Consulting, its clients or any other NERA consultants.
NERA Economic Consulting
1255 23rd Street NW
Washington, DC 20037
Tel: +I 202 466 3510
Fax: +1 202 466 3605
www.nera.com
NERA Economic Consulting
Report Qualifications/Assumptions and Limiting Conditions
Information furnished by others, upon which all or portions of this report are based, is believed
to be reliable, but has not been independently verified, unless otherwise expressly indicated.
Public information and industry and statistical data are from sources we deem to be reliable;
however, we make no representation as to the accuracy or completeness of such information.
The findings contained in this report may contain predictions based on current data and historical
trends. Any such predictions are subject to inherent risks and uncertainties. NERA Economic
Consulting accepts no responsibility for actual results or future events.
The opinions expressed in this report are valid only for the purpose stated herein and as of the
date of this report. No obligation is assumed to revise this report to reflect changes, events or
conditions, which occur subsequent to the date hereof.
All decisions in connection with the implementation or use of advice or recommendations
contained in this report are the sole responsibility of the client. This report does not represent
investment advice nor does it provide an opinion regarding the fairness of any transaction to any
and all parties.
© NERA Economic Consulting
NERA Economic Consulting
Contents
Executive Summary
1
I.
5
II.
Introduction
Overview of Interagency Working Group Calculations of the Social Cost of
Carbon
A. Background on Calculations in Integrated Assessment Models and Social Cost of
Carbon Values Developed by the Interagency Working Group
B. Range of Social Cost of Carbon Values from Integrated Assessment Models and
Implications about Uncertainty in the Damage Function
III. Contrast Between The "Damage Function Method" Used in Benefit-Cost
Analyses and the "Damage Functions" Used For Social Cost of Carbon
Estimates
A. Damage Function Approach in Traditional Regulatory Impact Analysis
B. Contrast with Damage Function in Integrated Assessment Models
IV.
V.
VI.
Formulation of the "Damage Function" in Integrated Assessment Models
A. Summary of the Core Damage Function in Most IAMB
B. Adjustments to Core Damage Function to Keep Projected Damages from Exceeding
100% of Baseline GDP
C. Other Adjustments to the Damage Function
D. Summary on Damage Function Formulations
Theoretical Considerations in Assessing the Functional Form of the "Damage
Function" in Integrated Assessment Models
A. Theoretical Considerations in the Choice of the Damage Function in Integrated
Assessment Models
B. Weitzman's Theoretical Assessments of Alternative Damage Functions
C. Summary on Conceptual Background for the Damage Function
Empirical Bases for Parameters Used for Damage Functions in Integrated
Assessment Models
A. The Calibration Process Used to Determine Damage Function Parameters
B. Limitations of Studies Used in Calibration Process
C. Summary of Empirical Bases for Damage Function Parameters
7
7
8
10
10
12
15
15
16
20
20
22
22
24
25
26
26
28
30
VII. Quantitative Examination of Damage Function Sensitivities
A. Sensitivity of Damages to Alternative Damage Function Parameters
B. Implications of Sensitivity Results
C. Differences Due to Use of the Exponential Function
D. Summary of Sensitivity Analyses
31
31
32
34
35
VIII. Conclusions
36
BIBLIOGRAPHY
38
APPENDIX A. Evolution of Damage Function in DICE
42
NERA Economic Consulting
List of Figures
Figure 1. Climate Change IAM Structure Following Traditional "Damage Function Method". 11
Figure 2. Structure of Climate Change IAMs Used for Estimating SCCs
13
Figure 3. Examples of How Variations in Parameters Affect Damage Estimates from a Core
Function of the Form D = a2 x ATa3
16
Figure 4: Altered Damage Functions to Prevent Damages from Exceeding 100% of Baseline
GDP (based on damage function of equation [2] using a3 = 3 and benchmarked to 5%
loss at 4°C)
18
Figure 5. DICE-2013R Damage Function (before and after adder for non-market impacts)
28
Figure 6. Damage Functions with Varied Powers and Calibration Points
32
Figure 7. Ratio of High to Low GDP Loss at Corresponding Temperature Changes
33
Figure 8. Damage Estimates from Exponential Compared to Polynomial Damage Function with
Same Parameters (using highest damage function in Figure 6)
34
NERA Economic Consulting
ii
EXECUTIVE SUMMARY
Introduction
The "social cost of carbon" (SCC) is a monetary value of the change in societal well-being per
metric ton of change in emissions of carbon dioxide (CO2). The Obama Administration has
developed a set of SCC estimates to be used in the Regulatory Impact Analyses (RIAs) that are
required to identify the costs and benefits of major federal regulations. The SCC values were
developed by an Interagency Working Group (IWG) and updated most recently in a November
2013 report (IWG, 2013b). The IWG developed the SCC estimates using results from three
Integrated Assessment Models (IAMB), which are complex mathematical models that combine
natural science and economic assessments. A key element determining an IAM's SCC results is
the "damage function." This report reviews the literature on IAM damage functions and
provides context for the damage functions used in the IWG's analysis.
Findings and Conclusions
In contrast to the traditional effect-by-effect approach known as the "damage function method"
used in RIAs, the damage functions in IAMs tend to be aggregate relationships between global
temperature change and GDP losses that do not describe the specific impacts that lead to
damages due to temperature change. Thus, it is not possible to evaluate uncertainties in the
damage functions that underlie the SCC values by evaluating uncertainties in specific
relationships between detailed physical effects and global damages.
As MIT Economics Professor Robert Pindyck notes, the lack of clear theoretical or empirical
bases for IAM damage functions means that the parameter values and functional forms for the
damage functions used by the IWG are largely ad hoc and arbitrary (Pindyck 2013). Although
the mathematical form of the damage function is relatively simple, plausible parameters for this
mathematical formulation lead to very different estimates of global damages. We find, for
example, that possible damage estimates at a given point in time can differ by a factor of 20 or
more within the range of parameters and range of temperature changes found in the IAM
literature.
Since the damage estimate is a central input to the SCC estimates, the large uncertainty in the
damage function translates into uncertainty in the SCC estimates that could be correspondingly
large. However, a comprehensive representation of damage function uncertainties — analyzed in
combination with the other IAM input uncertainties — is needed to characterize how much more
uncertain the IWG's SCC estimates would be as a result of that damage function uncertainty.
Neither the IWG nor we have conducted such an analysis.
NERA Economic Consulting
1
Range of Social Cost of Carbon Values and Implications about Uncertainty in the Damage
Function
The IWG values for the SCC have been the subject of considerable commentary among
economists and others. As noted, Pindyck (2013) has written that IAM-based analyses such as
those developed by the IWG create an "illusory and misleading" appearance of knowledge and
precision about the benefits of reducing CO2 emissions. With regard to the damage function
specifically, Pindyck (2013) notes that
[WJe know almost nothing, so developers of IAMs can do little more than make
up functional forms and corresponding parameter values. And that is pretty much
what they have done.'
Pindyck provides a relatively brief summary of his concerns with the IAM damage functions.
This report explores the issues in some detail to provide readers who are not familiar with IAMs
a greater understanding of the role of this component in the SCC calculation and the range of
uncertainty in its implementation.
One preliminary indication of the importance of uncertainty in the damage function is to consider
the differences in the IWG's SCC values across the three IAMs. To develop its SCC estimates,
the IWG standardized several key inputs to those IAMs, but not including their damage function
assumptions. Thus some of the differences in the IWG's SCC estimates across those three
models can be attributed to differences in the damage functions of those models. Table ES-1
shows information on the differences across the three IAMs for the three discount rates used by
the IWG. Holding constant the other variables that IWG standardized across the three models,
the average SCC estimates from the three models differ by a factor of three to eight, depending
on the discount rate.
Table ES-1. Average SCC Estimates by Individual IAMs in IWG's Analysis (*) ($/ton of CO 2
emissions in 2020, stated in 2007$)
Discount
Rate
5.0%
3.0%
2.5%
Lowest Average
SCC Estimate
(from FUND)
$3
$19
$33
Highest Average
SCC Estimate
(from PAGE)
$22
$71
$101
Ratio of
Highest to Lowest
Average SCC
8.3
3.7
3.1
The average dollar values were calculated by taking each model's average SCC value across the IWG probability distribution
of climate sensitivity values for each of the five IWG socioeconomic scenarios, and taking a simple average of those five values.
They have been rounded to the nearest dollar. The ratios are based on the unrounded averages. The underlying data to compute
these averages are in Appendix A of IWG (2013b), Tables A2-A4. In each case, the DICE estimate is the middle value, hence
not affecting the range; DICE's average values are $12, $38 and $57 for the 5%, 3% and 2.5% discount rates, respectively.
()
Pindyck, 2013, p. 867.
NERA Economic Consulting
2
These differences do not provide a clear indication of uncertainties regarding the damage
function, since there are other differences remaining among the three models, but it is reasonable
to expect that some of the variation in SCC values is due to differences in their respective
damage functions. However, these differences across the three IAMs used by the IWG likely
understate the full uncertainty in the damage function. Our assessment of plausible damage
function specifications indicates that the full range of uncertainty in the damage function has not
been addressed in the IWG analysis and thus also is not reflected in the SCC values that the IWG
presents.
Conceptual and Empirical Foundations for Damage Function Used in Integrated
Assessment Models
Pindyck (2013) notes the lack of a theoretical foundation for the functional form of the
aggregated damage function typically used in IAMs. The lack of a theoretical foundation for
such an aggregated concept is not surprising in light of the complicated linkages between
temperature change (or other summary measures of physical impacts) and the dollar value of
damages. Harvard Economics Professor Martin Weitzman has considered the theoretical basis
for the functional form and evaluated an alternative to the functional form used in most IAMs;
his evaluations lead him to prefer an alternative functional form although he notes that his
preferred form is not required as a theoretical principle.
We also find there is very limited empirical basis for setting the parameters of the aggregated
damage function, a condition acknowledged by the modelers. Evidence is limited to historically
observed temperature changes (i.e., less than 3 °C). Even for that range of change, there are only
a small number (less than 20) available empirical studies, and these often are for single regions
of the world (such as the US), for single sectors of the economy, and many are now quite dated
(e.g., from the 1990s). This paucity of evidence to set the damage function parameters means
that a wide variety of damage functions can be plausible and, indeed, the existing functions differ
considerably among IAMs.
Sensitivity of Results to Alternative Damage Function Parameters
The estimates of global damages due to a given temperature change can differ substantially
depending upon the assumed parameters of the damage function. The quantitative importance of
the choice of damage function parameters is illustrated by considering the estimate of global
damages when just two damage function parameters are varied from the lowest to highest values
for each that are discussed in the IAM literature. Figure ES-1 graphs the lowest and highest
resulting damage estimates at temperature changes up to 15°C. Sensitivity results are shown
over this wide range of temperature change because temperature changes up to 13°C appear to
have been projected in some of the IWG's IAM runs by the end of their modeling period, the
year 2300.
NERA Economic Consulting
3
Figure ES-1. Range of Damage Estimates with Variations in Two Damage Function Input
Assumptions
100%
90%
A.
Highest
/damage
IG estimate
•••
80%
• 70%
1
7:1
co
60%
tor
/
A
0
50%
Ld
ar
/
40%
/
'4
13 30%
R
/ /
Range of
Sensitivity
/
E
ci
/
O 20%
/
Lowest
/ damage
1G estimate
10%
0%
0
2.5
7.5
10
5
Temperature Change (Degrees Celsius)
12.5
15
This sensitivity analysis shows a large proportional sensitivity at all temperature changes. The
absolute magnitude of the difference is substantially greater at higher temperature changes.
Although the larger temperature changes are not important in the early and middle years of the
IAM projections, these temperature changes can be relevant in the later years of the projections.
This sensitivity analysis only varied two of the parameters that determine a typical IAM damage
function. A sensitivity analysis that included additional parameter uncertainties would widen the
range even more. A more detailed discussion of this sensitivity analysis is provided in the report.
In our study we also explored sensitivity to the alternative functional form preferred by
Weitzman, an exponential form that is explained in the report. Weitzman indicates a preference
for the exponential form as better capturing the effects of extreme events; but given the way
IAMs currently capture temperature change uncertainty, use of the alternative functional form
does not lead to substantial sensitivity in damage estimates for most levels of temperature change
the IAMs consider.2
2
However, if the alternative functional form were to be included in the sensitivity analysis summarized in Figure
ES-1, the range in that figure would be even wider at temperature changes above about 5°C.
NERA Economic Consulting
4
I. INTRODUCTION
The Obama Administration has developed specific monetary values per metric ton of carbon
dioxide (CO2) that are to be used in Regulatory Impact Analyses (RIAs) required under
Executive Order 12866 to estimate the costs and benefits of major federal regulations. These
dollar values, referred to as the "social cost of carbon" (SCC), were developed by an Interagency
Working Group (IWG) and reported in 2009, 2010 and most recently in 2013. The IWG
identified a range of SCC values, derived using a type of mathematical model called an
integrated assessment model (IAM) of climate change. The 2009 values were based on a
literature review of existing SCC estimates conducted by the IWG and were treated as
preliminary values. The Administration's 2010 and 2013 SCC values were derived from new
IAM model runs conducted under the direction of the IWG using three IAMs and sets of
assumptions specified by the IWG. The IAMs used and the IWG assumptions are identified in
Technical Support Documents (IWG, 2010, 2013a, 2013b).
The IAMs that underlie SCC values are complex mathematical models that are fundamentally
interdisciplinary, combining scientific and economic assessments. This paper focuses on the
point in the flow of computations within an IAM where scientific effects are translated into
dollar values, which is often called the IAM "damage function." This point in the modeling is
commonly considered to be within the domain of the economics discipline.
In a recent paper, Pindyck (2013) explains why he believes that IAM-based analyses such as
those developed by the IWG create an "illusory and misleading" appearance of knowledge and
precision about the benefits of reducing CO2 emissions. Pindyck identifies concerns with
components of an IAM that range from scientific to ethical. Regarding the damage function
specifically, he states:
When assessing climate sensitivity, we at least have scientific results to rely on,
and can argue coherently about the probability distribution that is most consistent
with those results. When it comes to the damage function, however, we know
almost nothing, so developers of IAMs can do little more than make up functional
forms and corresponding parameter values. And that is pretty much what they
have done. 3
Pindyck provides a relatively brief summary of his concerns with the IAM damage functions.
This report explores the IAM damage function in some detail to provide readers who are not
familiar with IAMs a greater understanding of the role of this component of the SCC calculation.
We then review the literature regarding the conceptual and empirical bases for specifying the
damage functions used in IAMs, to help readers better understand the statements by Pindyck.
Finally, we provide numerical examples that illustrate the potential range of uncertainty of
damage estimates based upon different IAM damage function specifications. Since the damage
3
Pindyck, 2013, p. 867.
NERA Economic Consulting
5
estimates are a direct input to the calculation of SCC values, the uncertainties explored in this
paper ultimately contribute to uncertainty on the SCC. The IWG has not attempted to
incorporate damage function uncertainty into its range of SCC estimates in any explicit way.
This paper indicates that the range of potential SCC estimates would be even larger than reported
by the IWG if a more complete account were taken of the uncertainty associated with the damage
function component of IAMs.
This paper provides a general discussion about IAMs that applies more broadly than just to the
three models used to produce the IWG's SCC estimates. However, because much of the interest
in this topic has come from the analyses of the IWG to determine SCC values to be used in RIAs
for federal regulations, our examples from existing IAMs focus on the three well-established
IAMs that the IWG has used for its SCC estimates. These are the DICE model of William
Nordhaus of Yale University, 4 the PAGE model of Chris Hope of University of Cambridge 5 and
the FUND model of Richard Tol of University of Sussex, UK, and Vrije Universiteit,
Amsterdam, Netherlands. 6
4
5
6
Many papers describe the DICE model at different points in time. Examples include Nordhaus (1992b, 2008,
2011, 2013), Nordhaus and Boyer (2000) as well as other documentation at:
http://www.econ.yale.edu/—nordhaus/homepage/.
Many papers describe the PAGE model and how it has evolved over time. Examples include Hope (2006, 2011).
Many papers describe the FUND model and how it has evolved over time. Examples include Tol (2006), Anthoff
and Tol (2008, 2010, 2012, 2013) as well as other documentation on its website at: http://www.fund-model.org .
NERA Economic Consulting
6
II. OVERVIEW OF INTERAGENCY WORKING GROUP
CALCULATIONS OF THE SOCIAL COST OF CARBON
This section provides an overview of the methodology used by the IWG to calculate alternative
SCC estimates and summarizes the results of the most recent evaluation. We note that the values
the IWG developed do not provide direct information on the implications of different damage
functions.
A. Background on Calculations in Integrated Assessment Models and Social
Cost of Carbon Values Developed by the Interagency Working Group
In general terms, an IAM contains a series of mathematical linkages between greenhouse gas
emissions and a dollar value of the societal well-being (called "welfare") that can be achieved
given those emissions. The basic chains in the linkage go from (a) greenhouse gas emissions, to
(b) how those emissions set in motion a sequence of atmospheric, terrestrial and biological
responses that result in various forms of "climate change impacts," and ultimately (c) the
monetary value to humans of those various outcomes. To get an SCC estimate from an IAM, the
IAM is run two times, with the runs differing only in the assumed level of CO2 emissions. The
difference in the two resulting present values of welfare is divided by the difference in tons of
CO2 emitted in the two runs, which becomes the estimate of the SCC ($/ton).
The damage function is the link in the chain of IAM calculations that translates projected climate
change outcomes into economic damages (usually stated as Gross Domestic Product, or GDP).
The damage function step typically provides monetary outcomes by year and by region of the
world. IAMs then condense those various monetized outcomes into a present value of global
welfare by applying a set of assumptions about how much weight to give to economic outcomes
in different time periods and regions, and to account for risk aversion? This paper does not
address issues regarding these weightings because they are not part of the damage function per
se, although they are also very important determinants of a SCC estimate.
The SCC values developed by the IWG are based upon the three IAMB noted in the prior section.
The IWG ran each IAM under five different sets of input assumptions regarding baseline future
GDP and GHG emissions, in various combinations with a range of input assumptions relating
atmospheric CO2 concentrations to global temperature change ("climate sensitivity"), and a
7
Risk aversion is the preference for less uncertainty about possible future outcomes. Risk averse individuals are
willing to spend money (for example, insurance premiums) in order to reduce the variance ("uncertainty') in the
value of losses they may have to absorb in the future. In the context of climate change benefit-cost analysis,
society is presumed to be willing to spend money if it will reduce the value of expected damages by at least as
much. A risk averse society will also assign value to limiting the worst of the possible damage outcomes, so it is
willing to spend yet more than just the expected value of damages if doing so will reduce degree of uncertainty
surrounding the expected damages. Thus, increasing the risk aversion assumed in an IAM run will increase its
SCC estimate.
NERA Economic Consulting
7
range of discount rates. The IWG's SCC value for a given discount rate is a simple average
across the three models' average results for the various other combinations of input assumptions.
B. Range of Social Cost of Carbon Values from Integrated Assessment Models
and Implications about Uncertainty in the Damage Function
In contrast to the baseline input conditions, temperature sensitivity, and discount rates, the IWG
did not directly specify ranges reflecting uncertainty about the damage function, noting that "all
other model features were left unchanged, relying on the model developers' best estimates and
judgments."8 As the 2010 IWG report states, "the sensitivity of the results to ... [the] damage
function...has not been incorporated into these estimates." 9
One preliminary indication of the importance of uncertainty in the damage function is to consider
how much difference remains in the SCC estimates of the individual IAMs after their inputs on
baseline GDP, emissions, and climate sensitivity have all been standardized by the IWG. This
can be inferred from the detailed set of SCC values reported in the appendix of the IWG's
Technical Support Documents. Table 1 shows information on the differences across the three
IAMs for each of the discount rates used by the IWG, based on the IWG's November 2013
release of SCC values. To obtain these, we have averaged across SCC estimates for the five
baseline input assumptions and the probability distribution for climate sensitivity for each of the
three discount rates used by the IWG. Holding constant the other variables that IWG
standardized across the three models, the average SCC estimates from the three models differ by
a factor of three to eight, depending on the discount rate.
Table 1. Average SCC Estimates by Individual IAMs in IWG's Analysise 9 ($/ton of CO 2 emissions
in 2020, stated in 2007$)
Discount
Rate
5.0%
3.0%
2.5%
Lowest Average
SCC Estimate
(from FUND)
$3
$19
$33
Highest Average
SCC Estimate
(from PAGE)
$22
$71
$101
Ratio of
Highest to Lowest
Average SCC
8.3
3.7
3.1
(*)
The average dollar values were calculated by taking each model's average SCC value across the IWG probability distribution
of climate sensitivity values for each of the five 1WG socioeconomic scenarios, and taking a simple average of those five values.
They have been rounded to the nearest dollar. The ratios are based on the unrounded averages. The underlying data to compute
these averages are in Appendix A of IWG (2013b), Tables A2-A4. In each case, the DICE estimate is the middle value, hence
not affecting the range; DICE's average values are $12, $38 and $57 for the 5%, 3% and 2.5% discount rates, respectively.
Determining how much of the model-to-model differences in SCC estimates evident in Table 1 is
due to differences in their damage functions and how much is due to other model assumptions
8
IWG, 2010, p. 6.
9
IWG, 2010, p. 6.
NERA Economic Consulting
8
that the IWG left unstandardized m was not possible within the scope of this review of the
literature. Ackerman and Munitz (2012) offer the only directly relevant information of which we
are aware. They find that almost all of the difference in the IWG's 2010 SCC estimates between
the DICE model and the FUND model can be attributed to differences in their damage function
assumptions." A more complete analysis across all three models has not been done to our
knowledge. Nevertheless, an IAM's damage estimate is a direct input to its final SCC estimate,
so differences in IAM damage function assumptions can readily translate into differences in SCC
estimates. I2 Thus, it is reasonable to expect that a substantial portion of the factor of 3 to 8 intermodel variation in the IWG's SCC estimates is attributable to differences in their damage
function specifications, since the damage functions assumed in the DICE, FUND, and PAGE
models used by IWG do differ in substantive ways. (Later sections of this report describe those
differences.)
Differences across the three IAMs used by the IWG, however, likely understate the uncertainty
in the SCC due to uncertainty in the damage function. As discussed below, our assessment of
plausible alternative specifications of the damage function indicates that the full range of
uncertainty in the damage function has not been addressed in the IWG analysis and thus also is
not reflected in the SCC values that the IWG presents.
10
One example is how each model represents global carbon cycling.
11
Ackerman and Munitz, 2012, p. 220.
Differences that affect estimates of damages at lower temperature increases, and/or in the nearer-term will tend to
cause larger sensitivity in the final SCC estimate because these will affect more of the years, and/or be discounted
less.
12
NERA Economic Consulting
9
III. CONTRAST BETWEEN THE "DAMAGE FUNCTION METHOD"
USED IN BENEFIT-COST ANALYSES AND THE "DAMAGE
FUNCTIONS" USED FOR SOCIAL COST OF CARBON ESTIMATES
This section provides background on the "damage function" used in the IAMs and its contrast
with the well-established "damage function method" in a typical benefit-cost analysis of an
environmental regulation.
A. Damage Function Approach in Traditional Regulatory Impact Analysis
The term "damage function method" has long been used by the US Environmental Protection
Agency (EPA) to describe a bottom-up approach for estimating benefits from pollutant controls.
For example, EPA's 1983 Guidelines for Regulatory Impact Analysis (EPA, 1983) characterizes
the process of estimating benefits for RIAs (Regulatory Impact Analyses) as an effect-by-effect
logical chain, with economic valuation as the final step, to be applied after specific forms of
adverse physical effects have been quantified:
The benefits of decreased pollution are the resulting improvements in health and
aesthetics and reductions in damages to plants, animals, and materials. To
measure benefits, one must ordinarily follow a chain of events from (1) the release
of pollutants by industry, households, agriculture, and municipal sources to (2)
the impact of these releases on ambient quality to (3) exposures of people, plants,
animals, and materials through various media (air, water, etc.) to (4) the adverse
effects to (5), when feasible, what people would pay to avoid these effects. 13
An appendix to EPA's Guidelines (EPA, 1988) labels the above process the "damage function
method" and more explicitly identifies a separation in the damage function method between
scientific research and economic research:
This method is based on a dose response function. It relates changes in a
pollutant to physical changes in receptor organisms or materials. The value of the
physical changes is then estimated by an appropriate method.' 4
The damage function method explicitly distinguishes the scientific (dose-response)
and economic valuation components of benefit estimation. 15
There appear to be parallels between EPA's effect-by-effect "damage function method" and the
"damage function" in climate change IAMs. Figure 1 illustrates a flow of analysis that one
might expect to find in a climate change IAM, with labeling to match the five steps of the
"damage function method" listed in the quote above.
13
EPA, 1983, p. M5.
14
EPA, 1988, p. A-5.
15
EPA, 1988, p. A-9.
NERA Economic Consulting
10
Figure 1. Climate Change IAM Structure Following Traditional "Damage Function Method"
1
Future
human
activities
--,
(1)
(2)
Emissions —Atmos.
Conc.
(3)
—
AT
-->
(4)
AHeat
AWater
ASpecies
APollution
AAmenity
AHealth
etc.
Physical sciences
(5a)
(5b)
Resources
AProductivity 'AGDP
.AVVTP
(6)
.
...
Aggregate
Welfare
.}
Economics
Economics
& Ethics
•
In Step 1, projections of human activities into the future that are exogenously specified
(e.g., population levels, economic activity/GDP levels, technologies available to meet
energy demands from people and economic activity) are combined to project greenhouse
gas emissions into the future.
•
In Step 2, physical sciences provide input assumptions and formulas to project
atmospheric concentrations of various greenhouse gases over time.
•
Step 3 relates the ambient (atmospheric) concentrations to "exposures." The most
common metric to proxy for exposure in climate science is temperature change (AT).
Physical sciences provide input assumptions and formulas to project the change in global
temperature associated with the path of atmospheric concentrations. 16 Other metrics of
exposure for projecting "adverse effects" may include, for example, the rate of change in
temperature and ambient CO2 concentrations.
•
In Step 4, climate change effects would be projected on an effect-by-effect basis,
focusing on effects that have meaning to humans." Effects must be defined with some
degree of specificity in order to be valued using economic techniques. However, the
methods for projecting each effect are still predominantly in the domain of the natural
sciences.
•
In Step 5, inputs from the economics discipline are used to monetize the value of the
projected effects. Market-based values (such are changes in supplies or quality of factors
of production) may be represented by change in GDP (AGDP). Nonmarket values (such
Temperature change projections may be regional, and at any point in time will reflect assumptions about the
dynamics by which changing concentrations evolve towards the equilibrium change. All of these require inputs
from physical sciences, and are simulated through complex mathematical formulas in an JAM.
16
EPA's Guidelines uses the term "adverse effects." We use the neutral term "effects" instead because all effects,
whether adverse or beneficial, should be included in a benefit-cost analysis.
17
NERA Economic Consulting
11
as changes in environmental amenities or in health) may be represented by change in
willingness to pay (AWTP). 18
We add a Step 6 to the Figure that is not separately identified in EPA's Guidelines. In this step,
intertemporal and interregional weighting schemes condense projections of economic damages
across many years and regions into a single present-value global measure of welfare. For most
pollutant control policies, this step could be viewed as just another economics component.
However, in the case of climate change, many of the impacts are very far in the future (up to 300
years hence, in the case of the IWG analyses), and also highly variable in terms of the region
affected. Thus, this sixth step raises issues regarding inter-generational and inter-regional equity
that seem largely ethical rather than economic.
B. Contrast with Damage Function in Integrated Assessment Models
The framework described above reflects the traditional "damages function method" in which
economic assessments are narrowly confined to valuing a specific set of projected adverse
effects, with each effect having its own valuation process. Climate change IAMs diverge from
that framework, however. Climate change IAMs tend to fall into one of two categories, neither
of which has all of the steps in Figure 1. The first category contains IAMs that are quite detailed
in their science components—including projecting various adverse effects such as those listed in
Step 4—but do not monetize the effects. In other words, they tend to conduct only Steps 1
through 4. 19
The second category contains IAMs that monetize climate change impacts. As monetization is
essential for estimating a SCC, the models used by the IWG (DICE, PAGE, and FUND) are in
this category. 2° In order to develop dollar values, however, these models tend to simplify the
science-oriented information, with the ultimate simplification represented by combining Steps 4
and 5, as illustrated in Figure 2.
Changes in GDP are not comparable to changes in WTP; they measure different things, but both have implications
for social welfare, which is the endpoint of concern for a benefit-cost analysis.
18
19
IAMs that fall into this effects-oriented type of IAM include IMAGE, ICLIPS, CIAS and AIM (which has
monetized impacts only in its vegetation module).
Other IAMs in this category include RICE, MERGE, WITCH, CETA, CRED, and ICAM. Not all of these models
have been used to generate estimates of SCC, however.
20
NERA Economic Consulting
12
Figure 2. Structure of Climate Change IAMs Used for Estimating SCCs
( 1)
Future
human
activities
(2)
(3)
Emissionstmos.--> AT
Conc.
(5)
Aggregated Damage Function
(6)
AGDP -AAggregate
Welfare
fl
Physical sciences
Economics
Economics
& Ethics
The result of such a simplification is that the component that is called the "damage function" in
an SCC calculation tends to become a highly aggregated formula that predicts economic loss
stated as change in GDP (AGDP) directly as a function of the projected change in temperature
(AT). Note that this type of damage function can no longer be viewed as strictly in the domain
of economics, as it implicitly contains assumptions about physical impacts from a change in
temperature as well as assumptions about how humans value various physical impacts. Indeed,
the damage function in an IAM is more of a replacement for the traditional "damage function
method," and not an application of that method as many people may surmise from the
terminology.
The IWG was aware of this distinction. Figure 2 provides a conceptual illustration of what the
IWG was describing when it wrote the following about the DICE, PAGE, and FUND models it
uses to produce estimates of the SCC:
These models are useful because they combine climate processes, economic
growth, and feedbacks between the climate and the global economy into a single
modeling framework. At the same time, they gain this advantage at the expense of
a more detailed representation of the underlying climatic and economic systems.
DICE, PAGE, and FUND all take stylized, reduced form approaches. Other
IAMs may better reflect the complexity of the science in their modeling
frameworks but do not link physical impacts to economic damages. (emphasis
added). 21
Not all of the IAMs used by the IWG simplify the linkages as much as suggested in Figure 2.
For example, FUND has features that make it a bit of a hybrid of a reduced-form and bottom-up
model. On the one hand, it specifies separate damage functions for multiple effects categories. 22
Tol argues that this disaggregated effects method is superior, stating that "aggregate impacts do
21
IWG, 2010, p. 5, citations omitted.
For example, FUND version 3.7, has 11 impact categories: agriculture, forestry, water resources, energy
consumption (space heating/cooling), sea-level rise, ecosystems, human health (diarrhoea, vector-borne diseases,
cardiovascular/respiratory mortality), extreme weather (tropical storms, extratropical storms).
22
NERA Economic Consulting
13
not follow a smooth, readily simplified curve. Separate modelling of impacts therefore seems
necessary to paint a realistic picture of aggregate impacts." 23 On the other hand, with the
exception of its sea-level rise damage function, each individual damage function in FUND is
aggregate in character. That is, each projects the monetary losses for a damage category directly
from either temperature change, the rate of change in temperature, and/or CO2 concentration (in
the case of agriculture, forestry, and extratropical storms). 24
23
Tol, 2002b, p. 156.
The greater complexity of FUND does not necessarily make its damage estimates more theoretically or
empirically correct. One negative effect is that the added complexity makes the model less accessible or
transparent to outside users or reviewers. This barrier to review creates greater potential for programming errors
that are likely to be detected and corrected only by the original modelers, as has occurred already (see, e.g., IWG,
2013b, p. 9 and Appendix B).
24
NERA Economic Consulting
14
IV. FORMULATION OF THE "DAMAGE FUNCTION" IN
INTEGRATED ASSESSMENT MODELS
This section provides information on the nature of the core damage function typically used in an
IAM. We also summarize various adjustments that are made from the core damage estimate.
A. Summary of the Core Damage Function in Most IAMs
The mathematical calculation used for the typical IAM "damage function" behind an SCC
estimate starts with a core formula that can be generalized as follows:
D=a1 x AT + a2 x Ar 3,
[
1
]
where D is the percentage reduction in baseline GDP, and is a function of change in global
average temperature, AT. 25 The variables labeled al, a2, and a3 are called "parameters" and are
the basic input assumptions that determine the shape of the damage function. Even if two IAMs
use this same core formula, they can produce very different damage estimates for a given
temperature change if they assume different parameter values. An illustrative example of this
point is provided below.
Versions of this core damage function equation are used in the three IAMs used by the IWG as
well as by other IAMs (e.g., CETA and CRED). The first term (i.e., a l x AT) allows for the
possibility of economic benefits from small increases in global temperature. Often, this term is
not used (i.e., al is often set to zero). For example, PAGE did not include a l until the 2009
model version, while DICE included al only in its 1999 version. For the majority of the models
and model versions in which al is set to zero the core damage function is simply the following
formula:
D = a2 x ATa3
[2]
This simpler, two-parameter formulation means that global GDP is always reduced as global
temperature increases. Figure 3 illustrates how different this two-parameter function can be for
just a few hypothetical alternative values of its parameters. Here, we have represented several
damage functions that all pass through the same point, a loss of 5% of GDP at a 4°C temperature
change, but with values of a3 ranging from 1 to 4. (One can force curves having different values
of a3 to go through one same point by appropriately resetting the value for a2 for each alternative
value of a3 . The respective values of a2 for each a3 in this figure are identified in the table
embedded in Figure 3.) Again, if one were to want a damage function that projects net economic
benefits for the first few increments of temperature change above zero, one would also use a l ,
setting it to a negative value. Although it is not illustrated in the figure, this would cause the
Each IAM makes some further adjustments to this core formula. In particular, further adjustments are made to
ensure that D never exceeds 100% of total baseline GDP. These adjustments are described later in this section.
25
NERA Economic Consulting
15
damage curve to dip below the x-axis (implying a negative GDP loss, or a benefit) and then be
steeper thereafter than its equivalent curve without a included.
Figure 3. Examples of How Variations in Parameters Affect Damage Estimates from a Core
Function of the Form D = a2 x Nra3
75%
If u,
equals...
2
3
0.1
4
C.1
t,
Calibrated
tt, Value
0.0125
0.0031
0.0008
0.0002
Quartic (a, = 4)
50%
0
to
Cubic (a, = 3)
a,
a.
CC
CC
r;
25%
CC
Quadratic (a, = 2)
E
Linear (a, = 1)
5%
0%
0
2.5
4°C
5
7.5
Temperature Change (Degrees Celsius)
This figure is intended only to illustrate some of the diversity of damage functions that the
simple formula of equation [2] can produce, and is not intended to represent any of the actual
damage functions in use. We discuss actual choices of parameter values in IAM practice, and
why IAM modelers have chosen those values, in Section VI.
B. Adjustments to Core Damage Function to Keep Projected Damages from
Exceeding 100% of Baseline GDP
An obvious problem with using the core formula of equation [1] or [2] without further
adjustment is that as projected temperature change increases, at some point it will predict
damages that exceed 100% of baseline output — a nonsensical outcome. Temperature change
may have to be very high for this to be a problem in a given IAM run, but the chance of it
occurring cannot be ignored. For example, for the illustrative curves in Figure 3 with a3 equal to
4 and 3, damages would exceed 100% of GDP when projected temperature change rises above
about 8.5 °C and 11 °C, respectively. While such extreme temperature changes may rarely occur
in typical IAM scenarios, we conclude that they have occurred in at least some of the IWG's
NERA Economic Consulting
16
scenarios, 26 and might become more common as IAMs are adapted to try to explore risks of
"runaway" climate changes.
All three of the IAMs used by the IWG make adjustments to their respective core damage
function to prevent its projected losses from ever exceeding 100% of GDP. This is done by
layering onto D some further mathematical adjustments that reduce the rate of increase in
damages as the value of D starts to approach 100%. In this way, the shapes of curves such as
those illustrated in Figure 3 are transformed into an S-shaped curve, or a rough approximation of
one. Because these adjustments can alter the entire shape of the core damage function, we
discuss the various adjustments in some detail here. Because each is essentially ad hoc, we see
no reason to suggest that one approach is technically superior to another.
FUND 3.7 employs the simplest type of adjustment. After calculating damages from each of its
sector- and region-specific damage functions, it applies a constraint that the sum of losses from
economic or market sectors must be less than or equal to baseline income. This constraint is
applied in each region in each year. That is, once economic damages start to exceed income,
they are simply reset equal to income. Figure 4 on the next page provides an illustration of an
unadjusted and the resulting adjusted damage function. The black line is an unadjusted damage
function (using a3 = 3 for illustrative purposes only). The blue line shows what the actual
damage calculation would be when applying the adjustment employed in FUND. 27
With one recent exception, the DICE model has implicitly adjusted the core damage function D
using the following equation:
D' = D+ (1 +D)
[3]
where D' is the percent reduction in baseline GDP that is actually used in the IAM calculations. 28
The result of applying this adjustment is illustrated in Figure 4 as a transformation of the black
To draw this inference, we ran the DICE model over the range of IWG assumptions about emissions growth and
climate sensitivity values to see what values of temperature increase were being projected in those runs (a result
not reported in any of the IWG's technical support documents). Results indicated temperature changes exceeded
11°C in about 10% of the runs performed for 3 of the 5 emissions projections adopted by the IWG for its SCC
analysis.
26
We note that FUND's not-to-exceed constraint is not applied to non-market (WTP-based) damages, so total
damages in FUND can actually exceed total income, but only for that part of the damages that are non-economic.
(However, one can make a reasoned argument that WTP should not exceed income either, as it is bounded by
ability to pay.)
27
Equation [3] states damages as percentage change in baseline (gross) GDP. However, in all versions of DICE
model code and documentation prior to DICE-2013R, Equation [1] is said to define D as damages relative to net
GDP (the remaining GDP after accounting for damages). Given this alternative definition of D, the term 1/(1+D)
is interpreted as the fraction of baseline GDP that remains after accounting for temperature change, and one finds
that in DICE code, the damage calculation applies 1/(1+D) to baseline GDP, not D itself, i.e. GDP(AT)= (1/(1+D))
x baseline GDP. This alternative definition of D is also algebraically equivalent to saying that D/(1+D) is the
percent reduction in baseline (gross) GDP. Since D/(I+D) will never exceed 100% of baseline GDP, the
adjustment in DICE is implicitly achieved when DICE calibrates 1/(1+D) to empirical data rather than D.
28
NERA Economic Consulting
17
▪•▪
curve into to the red curve. As a matter of mathematics, at very low levels of temperature
change, this formula produces adjusted damage estimates that are nearly identical to those from
the unadjusted formula. However, as the temperature change increases, adjusted values diverge
rapidly from the unadjusted values in such a manner that the adjusted values continue to rise, but
ever more slowly, never to exceed 100%. 29
Figure 4: Altered Damage Functions to Prevent Damages from Exceeding 100% of Baseline GDP
(based on damage function of equation [2] using a3 = 3 and benchmarked to 5% loss at 4°C) 3°
200%
— D
— D" = D/(1+D)
175%
— D with Saturation
— D=<Income
cz 150%
a,
eai 125%
4.
O
a, 100%
.
cat
75%
V
tao
R 50%
25%
0%
0.0
2.5
5.0
7.5
10.0
12.5
15.0
17.5
20.0
Temperature Change (Degrees Celsius)
The exception we have found for DICE is in its most recent version, DICE-2013R. In this
version, the core damage function is used as-is. 31 It appears that Nordhaus did not intend for this
core damage function to be used for the higher temperature changes in which damages could
exceed GDP because his documentation states that "this limits the usefulness of this approach for
catastrophic climate change. [It] needs to be examined carefully or re-specified in cases of
higher warming."32 But he has also added a new constraint in the code that caps damages at
29
In technical terms, D' asymptotically approaches 100% as temperature change approaches infinity.
This damage function is not meant to represent a specific damage function in any of the IWG models but is
merely used to illustrate the effects of the various methods employed in IAMs to prevent damages from exceeding
output.
30
The text of the documentation for DICE-2013R (Nordhaus and Sztorc, 2013) says it uses the standard DICE
damage formula: GDP(AT)= (1/(1+D)) x baseline GDP. However, we find that the GAMS code for that model
version uses the unadjusted damage function, D, to calculate GDP in the presence of temperature change, i.e.,
GDP(AT)=(1-D) x baseline GDP. Also, the definition of D provided in the code itself now states that it represents
damages relative to gross output, not net output as in all prior DICE model versions.
31
32
Nordhaus and Sztorc, 2013, p. 11.
N ERA Economic Consulting
18
100% of baseline GDP. This is essentially same method used in FUND (i.e., the DICE-2013R
adjusted damage function also will follow the blue line in Figure 4 for a core damage function
that follows the black line). This new constraint will not likely result in any actual adjustment in
the core damage estimates in a DICE-2013R run because, given the specific core damage
function parameters Nordhaus has assumed in that version, the core damage estimate will only
exceed 100% of baseline GDP for values of temperature change above about 19°C. 33 Thus, this
constraint serves as a precaution against the model producing nonsensical results if someone uses
it to run a scenario that involves an extremely high temperature change, or if a user of this
version might alter the parameters of the damage function so that it exceeds 100% of baseline
GDP at a much lower temperature change. 34
PAGE takes the most complex adjustment approach. To prevent damages from exceeding 100%
of GDP, PAGE09 adjusts core damages (as a fraction of baseline GDP) when the core damage
estimates exceed an assumed "saturation point" expressed as a percentage of consumption
reduction projected by the core damage function. When the computed core damages exceed the
saturation point, damages are scaled downward by a complex formula that ensures the actual
damage estimates used in the model gradually rise toward but never exceed 100%. 35 In
PAGE09, the saturation point is selected probabilistically over a range of 20% to 50% with a
most likely value of 30%. We have found no explanation in the PAGE documentation for the
choice of this range. However, we note that the saturation point value of 30% produces an
adjusted damage function (the green line in the figure) that is fairly similar to that obtained by
the DICE approach (red line in the figure), provided they both start with the same core damage
function. 36 In practice, however, these two models do not use the same parameters for their core
damage functions; it should be kept in mind that the curves in Figure 4 are not the actual adjusted
damage curves of any of the three models, and are used here merely to illustrate the general
effect of each type of adjustment.
In summary, IAM modelers are clearly aware that the core damage functions based on the
standard IAM formula (e.g., equation [1] or [2]) could, in some extreme temperature scenarios,
project losses exceeding 100% of baseline GDP. In an effort to avoid such nonsensical
outcomes, each model applies some form of ad hoc adjustment to their core damage functions.
Two of the three adjustment formulas start to substantively adjust damage estimates at
33
The parameters used in DICE-2013R are 0 3=2 and 02.00267.
One should be aware, however, that this change in DICE's adjustment method will tend to raise this version's
damage estimates (and hence SCC estimates) relative to those from the prior DICE versions. This is because the
prior DICE versions all used the adjustment formula in equation [3], which projects damages along the red line,
which are lower than those along the blue line.
34
This adjustment is applied to each of PAGEO9's impact categories (sea-level rise, economic, and non-economic)
and by region.
35
36 The
saturated damages depicted in Figure 4 were computed by applying PAGEO9's saturation formula using a
saturation point of 30% to the illustrative cubic damage function (black line in the figure). This calculation has not
included mitigation of impacts through adaptation, which are also part of the PAGE09 adjustments to its core
damage formula.
NERA Economic Consulting
19
temperature changes well below the point at which projected GDP losses would actually exceed
100% of baseline GDP. For example, in the illustrative example of Figure 4, unadjusted
damages would not exceed baseline GDP until temperature changes exceed I 1 °C, but the
adjusted damages in DICE and PAGE (red and green lines) would diverge from the unadjusted
damage function (black line) starting at about 5° to 7.5 °C. 37
Note, however, that the illustrative examples in Figure 4 exaggerate the differences that will
typically result due to these adjustments, because it is conducted against a relatively steep
unadjusted damage function in which a3 = 3. This value is at the high end of the range of values
used in IAMs. The differences are less pronounced for lower values of 03 that are more
frequently used. However, the conclusion does hold that it can in some instances be misleading
to compare IAMs' damage functions solely on the basis of the parameter values a l , a2, or a3 that
define their core damage functions.
C. Other Adjustments to the Damage Function
The three IAMs used by the IWG make other adjustments that also affect their ultimate estimate
of damages. These adjustments vary from model to model, thus creating further inter-model
differences. All of them are ad hoc in nature, although the phenomena they attempt to simulate
are valid concerns that, in principle, should be accounted for in a SCC estimate. We mention
two of them here.
Some IAMs have attempted to model the possibility of the occurrence of unexpected
catastrophic events (e.g., the collapse of the thermohaline circulation). PAGE incorporates an ad
hoc adjustment for climatic tipping points by modeling damages from such a discontinuity if it
were to occur (including separate probability distributions for the probability of its occurrence,
the temperature at which it would occur, and the resulting loss). DICE and FUND, in contrast,
do not include the possibility of catastrophic events in their damage functions.
Accounting for adaptation represents another example of an adjustment that differs across the
three models. Adaptation indicates the extent to which society will take actions to avoid adverse
effects of changes in temperature or other effects of CO2 emissions. PAGE and FUND explicitly
incorporate some assumptions regarding how adaptation may mitigate climate change impacts.
DICE does not include explicit assumptions on the role of adaptation in reducing global
damages.
D. Summary on Damage Function Formulations
In summary, the typical IAM damage function starts with a highly simplified and abstract
representation of the detailed set of calculations that in principle link atmospheric GHG
37
At this point, the adjusted damage function also implies that the rate of increase in damages decreases as
temperature increases, which may seem counterintuitive to some.
NERA Economic Consulting
20
concentrations to dollar damages (as illustrated as Steps 4 and 5 of Figure 1 in the previous
section). 38 The inherent limitations of this approach are captured by an early scientific
commenter on the DICE model. When the first versions of DICE were being produced,
Nordhaus (1994) surveyed several experts in the field of climate change to gauge their subjective
perceptions of potential impacts at different levels of temperature change. One expert had the
following response to Nordhaus's request:
I must tell you that I marvel that economists are willing to make quantitative
estimates of economic consequences of climate change where the only measures
available are estimates of global surface average increases in temperature. As
[one] who has spent his career worrying about the vagaries of the dynamics of
the atmosphere, I marvel that they can translate a single global number, an
extremely poor surrogate for a description of the climatic conditions, into
quantitative estimates of impacts of global economic conditions. 39
Other adjustments modify the core damage function and can affect the damage calculations.
Ad hoc adjustments are made to keep the models from inadvertently producing greater than
100% GDP losses at relatively high projected temperature changes, a sensible limitation.
However, such adjustments can alter the shape of the damage function so that one key feature —
an implication that there is a higher rate of increase in damage as temperature change increases —
can be lost as a result of the adjustment. Further adjustments are also seen as desirable to
account for possibilities such as discontinuously rapid growth in damages (raising damage
estimates) and adaptation (lowering damage estimates). These adjustments also are inherently
ad hoc. The net result is that each IAM ends up with a set of mathematical formulas that
determine monetary damages as a function of change in temperature. Setting aside the many
different ad hoc adjustments, each set of damage calculations starts with a specific setting of
functional form and associated parameters. We now turn to the potential theoretical and
empirical bases for those damage function specifications.
Again, we note that FUND's damage function is more complex, with multiple effect-specific loss equations and
the inclusion of independent variables other than just AT. Their choice of functional forms and their
parameterization, however, appear to be of a similar nature to the processes that we describe next.
38
89
Nordhaus, 1994, p. 51.
NERA Economic Consulting
21
V. THEORETICAL CONSIDERATIONS IN ASSESSING THE
FUNCTIONAL FORM OF THE "DAMAGE FUNCTION" IN
INTEGRATED ASSESSMENT MODELS
Pindyck (2013) notes the lack of a theoretical foundation for the functional form of the damage
function used in IAMs (i.e., equation [1] or its simpler version [2]). Our review of the IAM
literature substantiates this conclusion. In seeking evidence of justifications for the functional
forms adopted by the various modelers, we found their papers typically posited the functional
form and then focused on explaining how they developed its parameters.
The lack of a theoretical foundation for a particular damage function is not surprising in light of
the complicated linkages between temperature change (or other summary measures of scientific
impacts) and the dollar value of damages, a condition that means that the choice of a particular
damage function is largely an empirical issue. The only papers we found that consider the
theoretical basis for functional forms for the damage function are ones by Weitzman (2009,
2010), written after the standard practice we have described above had become established. As
discussed below, Weitzman's evaluations lead to a preference for a particular functional form
rather than a conclusion that his preferred form is required as a theoretical principle.
A. Theoretical Considerations in the Choice of the Damage Function in
Integrated Assessment Models
Weitzman has discussed the lack of ability of IAMs to deal with what is sometimes described as
"fat-tailed" probability distributions, by which is meant a small probability of truly catastrophic
impacts. ° As discussed below, Weitzman concludes that the functional forms common in IAMs
are not properly responsive to "fat-tailed" uncertainties. In the course of developing this
argument, Weitzman offers his understanding of the original source of the damage function
formula in IAMs.
Weitzman comments that the functional form, particularly the simple quadratic version in which
GDP loss is predicted to rise with the square of AT (i.e., a3=2) was adopted largely because of its
familiarity to IAM developers with an economics background. He points to the fact that many
dynamic equilibrium economics models use similar quadratic terms to incorporate adjustment
costs to obtain reasonable solution properties. The analytical tractability of this formula could be
expected to carry over into the complex mathematical setting of IAMs. Weitzman notes: "There
was never any more compelling rationale for this particular loss function than the comfort that
economists feel from having worked with it before." 41
We find that the literature does contain evidence of more substantive reasoning regarding the
choice of functional forms in existing IAMs, but such discussions deal more with the value for
40
More technically, the probability distribution for climate risk has infinite variance.
41
Weitzman, 2009, p. 16.
NERA Economic Consulting
22
the formula's parameter a3 than the use of that particular type of formula. Nordhaus (1992), for
example, reports that the case for non-linearity (i.e., a3>1) comes from studies of coastline
impacts. Other discussions point more to anecdotal and subjective judgments than to theoretical
arguments. For example, Nordhaus has always used a power of 2 in DICE (i.e., a3=2). In a
history of 1AM development, Schneider and Lane (2007) describe an informal survey that
Nordhaus undertook (Nordhaus, 1994) to inform the question of how large damages would be at
higher temperature changes. Economist and ecologists had extremely different subjective views
about the level of damages, but they did agree that the relationship should be non-linear:
Despite the difference in magnitude of damage estimates between economists and
ecologists, the shape of the damage estimate curve was similar. All respondents
indicated accelerating costs with more climate changes. 42
Similarly, documentation of the PAGE model cites a study by Cline in 1992 that purportedly
produced an average estimate of the parameter a3 of 1.3 — slightly greater than linear, but less
than 2.43 However, review of the original source (Cline, 1992), shows that Cline largely
assumed that damages for various categories of effect increase with temperature change by
powers that vary from 1.25 to 1.5. Although Cline offers reasons why one would expect some
categories of climate damage to increase more than linearly with temperature change, the
specific numerical powers that he uses in his analysis appear to have been his subjective
judgment, with no explanation of his reasons for choosing each numerical value.
Until its 2009 version, the PAGE model used a probability distribution for the parameter a3 that
was centered on 1.3 (with a range of 1 to 3), based on Cline. 44 Starting in 2009, the PAGE
modelers re-centered that probability distribution on 2, and also raised the minimum value to 1.5.
We have not found an explanation in PAGE09 documentation for making this change in 2009,
which increases estimated damages and hence also the SCC estimate. However, a paper by
Hope (2011) mentions that PAGE09 has a damage function similar to that in a study by
Ackerman et al. (2009). The latter, in turn, states that it uses a central value for a 3 slightly
greater than 2 because 2 is "the value used in many recent models." 45 This would appear to be
an example of the "circularity" mentioned by Pindyck 46 in which choices of damage functions in
IAMs are often justified by reference to each other. Such circularity will tend to narrow the
range of parameters being used across models, causing them as a group to understate the degree
of uncertainty about the true damage relationships. This is one of the reasons why we conclude
that the variation in damage estimates across the three IWG models is unlikely to provide a
sufficient characterization of damage function uncertainty.
42
43
Schneider and Lane, 2007, p. 47.
Warren et al., 2006, p. 31.
44
Technically, the probability distribution was a triangular distribution between 1 and 3 with its mode (apex) at 1.3.
45
Ackerman et al., 2009, p. 8.
46
lc 2013, p. 868.
Pindyck,
NERA Economic Consulting
23
Thus, we find that subjective judgments, informed by some reasoning, have supported the
assumption that damages should increase at an increasing rate as temperature increases, which is
what is achieved by assuming the parameter a3 is greater than 1 (recall the illustrative examples
in Figure 3). We also find that IAM modelers rarely, if ever, assume a value for a3 greater than
3. Here, it seems the reason is pragmatic. For example, using CETA, Peck and Teisberg (1992)
tested a damage function with a3=3 in order to demonstrate that "under certain assumptions a
large reduction of CO2 can be optimal" 47 and concluded that a power of 3 should be sufficiently
high to achieve this. We have found no physical or economic theory or reasoning to limit the
upper end of the range of values for a3 to 3, but we have found no values above 3 in the IAM
1 iterature.
B. Weitzman's Theoretical Assessments of Alternative Damage Functions
As noted, Weitzman has discussed theoretical considerations in the choice of a damage function,
focusing on whether the functional form can account for the degree of welfare loss associated
with very small probability climate events of extreme magnitude ("fat-tailed risk"), which he
believes is a fundamental characteristic of climate change risk. Weitzman (2009) concludes that
an exponential form of the damage function, such as
D = 1 - exp(
-
a2 x ATa3)
[4]
is preferable to the simple polynomial used in IAMs (i.e., the term inside the brackets in
equation [4]). 48 His concern is that the simple polynomial form "thins down" the welfare
implications of a fat tail of physical impact risk because the exponential form makes welfare loss
proportional to absolute damages, whereas under the standard polynomial form the welfare loss
is only proportional to proportional increases in physical damages. Weitzman notes, however,
that there is no firm rule in economic theory that welfare loss must be proportional to the
absolute increase in damages (and thus provide greater importance to the "fat tail" of physical
climate change risk). He thus concludes that his preference cannot be justified as a theoretical
principle: "I cannot prove that my favored choice is the more reasonable of the two...but no one
can disprove it either — and this is the point here." 49
Weitzman (2010) explores alternative ways that the damage function, D, could be translated into
a societal welfare loss. He compares the theoretical properties of IAMs' standard multiplicative
47
Peck and Teisberg, 1992, p. 68.
Weitzman, 2009, p. 16. In his paper, Weitzman expresses his preferred damage function as the fraction of
baseline GDP that will remain in the presence of a given temperature change. Equation [4] states his preferred
formula as percentage loss in baseline GDP, for consistency with the other descriptions in this paper of the damage
function. We note that the formula in equation [4] will never exceed 100% of baseline GDP, so it is equivalent to
the adjusted damage functions that IAMs actually apply, rather than to the core damage function 0 7 x AT'''. Also,
while Weitzman uses a quadratic form 0 2 x AT2 in his example, he is clear that the issue applies to damage
functions with any value of a 3 .
48
49
Weitzman, 2009, p. 16.
NERA Economic Consulting
24
use of D in the societal welfare function to an alternative, additive use of D in the welfare
function (which, as with the exponential damage function, preserves fat-tailed physical risk).
However, he finds both options preserve theoretically desirable properties of a societal welfare
function and so again, he concludes "it is essentially impossible to discriminate empirically
between the forms...so that readers must subjectively decide for themselves which...they feel
most comfortable with." 5°
C. Summary on Conceptual Background for the Damage Function
There is no clear theoretical basis for choosing a specific functional form for the damage
function, which is not surprising given the scientific complexities that are implicit in the linkages
between temperature change and the dollar value of global damages. Moreover, although
Weitzman explores some additional conceptual issues and develops a preference for a
formulation that differs from that used by the three IAMs used by the IWG, he concludes that
there is no clear theoretical basis for choosing the functional form for the damage function used
in IAMs (and thus that generate SCC estimates). We will now turn to the evidence that has been
used to develop the specific parameters adopted by IAM developers.
5° Weitzman, 2010, p. 64.
NERA Economic Consulting
25
VI. EMPIRICAL BASES FOR PARAMETERS USED FOR DAMAGE
FUNCTIONS IN INTEGRATED ASSESSMENT MODELS
This section explains the empirical bases for the specific parameters that are used to define the
damage function. As explained in the prior section, values for a3 that have been used in IAMs
have ranged from 1 to 3. In contrast, the values for the other parameters (a2 and, if used, a 1 ) are
determined by a "calibration" to impact estimates that lie within the range of historical
temperature changes, given the modeler's choice of a3. This section summarizes the empirical
bases used to inform this calibration of damage functions in IAMs. We describe the calibration
process and consider limitations in the studies available to develop the calibrated values.
A. The Calibration Process Used to Determine Damage Function Parameters
The determination of the parameters of the damage function is based upon an empirical process
referred to as "calibration" by IAM modelers. Calibration is similar to the process of fitting a
curve to a set of data points. At first blush, it may appear innocuous for a modeler to choose
parameters to create a damage function that is consistent with point estimates of damages at
historically-experienced temperature changes, founded in climate change research. However,
fitting IAM damage functions often involves only one data point, called a "benchmark," which is
an estimate of the percent of GDP lost at one particular temperature change. It is possible to
perform calibration with just one single data point when the damage function is a2 x AT° 3 with a
pre-determined value for a3, because only one parameter (a2) needs to be estimated.
When a single benchmark is used, calibration is just an algebraic calculation rather than a
statistical process of curve-fitting. For example, if a 3 is set to 2, and the benchmark point is a
GDP loss of 3% if temperature change AT is 4 °C, then a2 is derived as the solution to
3 = a2 x 42, which means a2=3/16. This ensures a core damage function that will project
damages of 3% of baseline GDP in any year in which the IAM projects temperature change of
4°C, but it also determines damages at all other temperature changes too. If a different
benchmark is selected, a totally different damage function will result. As we will explain below,
the choice of what benchmark to use is a judgment each modeler makes, not based on a stable set
of observational data. For this reason, damage functions can differ dramatically from model to
model, but each is ad hoc and with no well-defined empirical standards to resolve which might
be more reliable than another.
The PAGE model allows the values to be determined probabilistically over an exogenously
determined range that allows for some representation of the modelers' subjective uncertainty
about the damage function in this model. The damage function parameters of PAGE vary based
on a Monte Carlo simulation, which means that the model is run through many iterations, each
with randomly selected parameter values, and the many different SCC estimates are then
averaged together. In each model iteration, the parameter a3 is drawn from a triangular
distribution and a random value for the calibration benchmark is drawn from another probability
NERA Economic Consulting
26
distribution. 5 ' In each model iteration, the value of a2 is then calculated to fit the selected
damage function through the selected benchmark point. The value for a 3 in PAGE2002 varies
between 1 and 3 with a most likely value of 1.3. In PAGE09 it varies from 1.5 to 3, with a most
likely value of 2. The 2009 update to the PAGE model also introduced al in order to allow for
small initial benefits for small temperature changes. Therefore, fully specifying the damage
function in PAGE is as simple as choosing values of a3 (and al in PAGE09), choosing a
calibration point, and then solving the equation of one unknown through that point to derive a2.
The apparent complexity associated with PAGE is that this process is performed over many
iterations using Monte Carlo simulation and incorporates ad hoc representation of possible
discontinuities due to catastrophic events (i.e., dramatic sudden increases in damages not
captured by the smoothly increasing damage functions that have been described in this paper).
DICE has always set 03 at 2 and then typically calibrated 02 (and a l when used) such that the
global damage function matches outputs from the regional version of the model (RICE), which is
calibrated by using sector-specific impact indices based on existing studies and expert guesses. 52
The most recent version, DICE-2013R, departs from this methodology and instead specifies the
damage function by (1) fitting a quadratic through a set of 13 impact assessments spanning the
timeframe 1994 to 2006 (five of which are his own, and others include FUND and PAGE), then
(2) increasing the damage function value by an ad hoc 25% to allow for non-monetized impacts.
The fitted line and the actual core damage function (i.e., with the non-market impacts adder
included, which is the blue line) can be seen in Figure 5 below along with the 13 data points to
which it was calibrated. 53 The sources for the 13 data points in Figure 5 are provided in Table 2
below.
The specific choices of calibration points and functional forms have evolved in all models over
time. Appendix A provides a brief description of the various changes in the DICE damage
function specification over the years as an example of the process that all models go through.
For the 2002 version of PAGE, the probability distribution for selecting calibration benchmarks is said to be based
on information in the IPCC (2001) report. However, the relationship between the specific numerical values in the
PAGE model and the information on the referenced pages of the IPCC report is not obvious. PAGE09 uses a
different distribution on calibration benchmarks that the model code indicates are based (without detailed
references) on Mastrandrea and FUND.
51
The version of DICE used in IWG's 2013 analysis (the DICE 2010 model) is sparsely documented and rarely
cited in publications. Thus, we have not been able to determine the origins of that version's parameter values.
52
The range shown for the "IPCC estimate" was not included in the calibration, and appears on the figure only to
show how the resulting core damage function (the blue line) compares to the "consensus" statement in IPCC
(2007).
53
NERA Economic Consulting
27
Figure 5. DICE-2013R Damage Function (before and after adder for non-market impacts)
Actual damage function in DICE2013R is increased by 25% above
curve fitting actual data points
6
Tol survey
5
Curve fitted to 13 data points
from Tol (2009) survey
A
—DICE-2013R model
I PCC estimate
-g
2
0
-3
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
4.5
5.0
Global mean temperature increase (°C)
Source: Nordhaus and Sztorc,"DICE-2013R: Introduction and User's Manual," Oct 2013. (Blue curve added to Nordhaus'
figure by NERA to show damage function with the 25% adder assumed by Nordhaus to reflect non-monetized effects.)
B. Limitations of Studies Used in Calibration Process
Modelers typically rely upon studies that estimate damages from changes in temperature to
calibrate the damage function that are invariably incomplete and often relatively old. In their
2006 paper reviewing the four most commonly used IAMs (DICE, PAGE, FUND, and
MERGE), Warren et al. note that all of the models (at that time) were based on literature from
2000 or earlier. Though DICE/RICE at one point conducted sector-specific calibrations to
existing studies, the literature has become increasingly outdated for use in the later model
versions. The studies employed for RICE99 to create sector-specific impacts (described in
Nordhaus and Boyer, 2000) were from the early to mid-1990s — recent enough in history for that
vintage of the model, but one has to question the relevance in 2010 if those studies remained as
part of the basis of determining the damage functions. This seems to be the reason why
Nordhaus modified his calibration approach in the 2013 model as he observes that "those
estimates were increasingly outdated and unreliable." 54 However, it bears repeating that his new
methodology is to fit a quadratic damage function to data points from studies as early as 1994
(see Table 2 below for the sources of the different data points). Thus, little new information is
conveyed in the updated damage function.
54
Nordhaus and Sztorc, 2013, p. 11.
NERA Economic Consulting
28
Table 2. Studies Used by Nordhaus in DICE-2013R (from Tol 2009)
Study
Nordhaus (1994a)
Nordhaus (1994b)
Fankhauser (1995)
Tol (1995)
Nordhaus and Yang
(1996)
Plambeck and Hope
(1996)
Mendelsohn,
Schlesinger, and
Williams (2000)
Nordhaus and Boyer
(2000)
Tol (2002)
Maddison (2003)
Rehdanz and Maddison
(2005)
Hope (2006)
Warming
(°C)
3
3
2.5
2.5
Impact
(% of
GDP)
-1.3
-4.8
-1.4
-1.9
2.5
1.7
2.5
Notes
-2.5
0. 0
2.5
0.1
(**)
"Experimental" model; only includes market
i mpacts.
"Cross-sectional" model; only includes market
impacts.
2.5
-1.5
1
2.5
2.3
-0.1
Only considers market impacts on households.
1
-0.4
Only considers market impacts on households
2.5
0.9
The numbers used by Hope (2006) are averages of
previous estimates by Fankhauser and Tol.
Nordhaus (2006)
2.5
-0.9
(*) The table in Tol (2009) shows this as a +2.5 but the original source and Nordhaus's use indicate it is
-2.5, as reported in this table too.
(**) It appears that Nordhaus used only the "experimental" model result of 0.0 to represent Mendelsohn
et al. (2000), but the value for the "cross-sectional" model appears in Tol (2009)'s original table.
In spite of having a more bottom-up approach, FUND still faces the same limited data
availability issues. For example, the water resources impact category in FUND is calibrated to
results by Downing from the mid-1990s which, in a 2002 paper, Tol refers to as being
incomplete and thus "the model used... is therefore ad hoc."55 Indeed, one of the difficulties of a
more detailed model is that many sources of data must be kept up-to-date.
Authors note the need to rely upon incomplete information to calibrate their models. As Tol
states, "This literature is to a large extent outside the economic discipline. It is therefore hard to
access and appreciate the difficulties." 56 Similarly, Nordhaus writes that "although the literature
in this area is extensive, there are many gaps in coverage of sectors and countries, and many of
55
Tol, 2002b, p. 142.
56
Tol, 2002a, p. 49.
NERA Economic Consulting
29
the most important impacts have not been satisfactorily quantified and monetized." 57 Thus
modelers must use what studies are available.
In some cases, the studies used as "data" by the modelers are not independent of the modeler.
Pindyck (2013) provides the following example of a potential lack of independence of models
and "input" data. "Nordhaus (2008) points out (page 51) that the 2007 IPCC report states that
`global mean losses could be 1-5% GDP for 4 ° C of warming.' But where did the IPCC get those
numbers? From its own survey of several IAMB. Yes, it's a bit circular." 58
Another complication is the use of region-specific data that must be extrapolated to the model
regions used in PAGE, FUND, and DICE/RICE. The aforementioned most-likely a3 value of1.3
in PAGE2002 comes from Cline (1992) and is US-specific. 59 However, the same value is used
to calculate the damages for PAGE's EU region (from which a regional multiplier is used to
arrive at damages for other regions). Although the latter assumption may be plausible, it
highlights the need to make judgments about a correspondence between a study's regions and the
IAM model's regions — even when an underlying study is global in scope.
C. Summary of Empirical Bases for Damage Function Parameters
The calibration process used to develop the damage function parameters is based upon limited
information on the linkage between temperature change and global or regional damages. There
are simply few (and inherently limited) studies that can provide the data for the calibration
process.
The modelers clearly recognize and readily concede the limitations in the empirical evidence that
lies behind their models, including the damage function. Tol admits that this "does not result in
a climate change impact model that is adequate. The accompanying static impact assessment is
far from perfect, with many pieces missing and a lot of questionable assumptions." 6° Nordhaus
similarly states that "the damage functions continue to be a major source of modeling
uncertainty."61 However, as will be discussed in the next section, the range of possible
parameters leads to very large differences in estimated damage values.
57
Nordhaus and Boyer, 2000, p. 69.
58
Pindyck, 2013, p. 868.
59
Warren et al., 2006, p. 31.
60
61
Tol, 2002b, p. 136, citations omitted.
Nordhaus, 2008, p. 51.
NERA Economic Consulting
30
VII. QUANTITATIVE EXAMINATION OF DAMAGE FUNCTION
SENSITIVITIES
This section illustrates the variation in the dollar value of damages under alternative plausible
values of the parameters of the damage function. These results show that the projected GDP
changes due to a given temperature change can vary widely for relatively small differences in the
parameter values.
A. Sensitivity of Damages to Alternative Damage Function Parameters
We illustrate the importance of specific damage function parameters by considering the estimate
of global damages as a function of temperature change under four possible damage functions that
reflect the extreme values of damage function parameters we found being used in IAMs.
Figure 6 graphs the values that these four different damage functions would project at
temperature changes up to 15°C. Sensitivity results are shown over this wide range of
temperature change because temperature changes up to 13°C appear to have been projected in
some of the IWG's IAM runs by the end of their modeling period, the year 2300. 62 The blue
brackets below the x-axis on the figure indicate the range of temperature changes in 2300 at the
10th percentile value of the IWG's ECS distribution, and the rightmost brackets represent the
temperature changes at the 90 th percentile value of the IWG's ECS distribution. They indicate
that the IWG's SCC estimates for 2300 include temperature changes above roughly 3 to 5°C in
about 90% of the IAM runs and that exceed 8 to 13°C in about 10% of the IAM runs.
The y-axis on Figure 6 indicates the damage function estimate for each temperature change level.
The metric for D in this figure is what the loss of GDP will be in the scenario being run (for each
amount of temperature change on the x-axis) stated as a percentage of what it would be absent
any temperature change due to CO2 emissions. The four lines graphed in the figure represent
four alternative possible damage functions, which are created by taking the two extreme values
of a3 (1 and 3) and calibrating them to two points based on the extremes of IPCC's "consensus"
range (i.e., IPCC's 1% and 5% loss at a 4°C temperature change). These functions have also
been adjusted to prevent damages from exceeding 100% of baseline GDP (i.e., the functions take
on an S-shape) using the adjustment method of equation [3].
The IWG runs assumed a probability distribution on the parameter for the equilibrium climate sensitivity (ECS)
parameter, which determines temperature change given a path of emissions of greenhouse gases. The 80%
confidence interval for ECS values used by the IWG is from about 2 to 5.86. We ran the four IWG socioeconomic
scenarios that did not assume atmospheric stabilization through DICE2007 with the ECS fixed at 2 and 5.86. The
blue brackets at the bottom of the figure indicate the range of temperature changes that these runs projected in the
year 2300 for the two ECS values. (Stabilization scenarios inherently have lower temperature changes that are not
included in the ranges shown on the figure.)
62
NERA Economic Consulting
31
Figure 6. Damage Functions with Varied Powers and Calibration Points
100%
Damages as a Percent of Baseline GDP
90%
80%
70%
60%
alib. Pt. = 5% at 4°
(a3 varied from 1 to 3)
50%
40%
30%
alib. Pt. = 1% at 4°
(a3 varied from 1 to 3)
20%
10%
0%
0
2.5
7.5
AT23oo
at ECS=2
10
12.5'
15
Temperature Change
(Degrees Celsius)
AT23oo
at ECS=5.86
Note: Roe & Baker distribution on equilibrium climate sensitivity (ECS) has 10th percentile at —2 and
90th percentile at 5.86
For this sensitivity analysis, all four functions set al to 0. Incorporating a non-zero ai is not
possible with only one calibration point except by exogenously specifying its value. If we were
to include any non-zero value of ai, however, the range of uncertainty would be larger than in
this particular figure.
B. Implications of Sensitivity Results
As can be seen in Figure 6, the projected damages are sensitive to both the calibration point and
the choice of exponent. In particular, the predicted dollar value of damages is substantially
greater with a larger exponent (i.e., an exponent of 3 rather than 1) and when the calibration
point is assumed to be a 5% reduction in global GDP rather than a 1% reduction in global GDP
at a temperature change of 4 °C. By comparing the solid lines to the dotted lines of the same
color, one can see the effect of using an a3 of 3 versus 1. By comparing the blue lines to their
respective red lines, one can see the effect of using a higher or lower calibration point from those
widely cited in the literature.
The results also depend substantially on the temperature change, with the absolute differences
due to alternative parameter values substantially greater at higher temperatures. It is worth
recalling here that the majority, if not all, of IAMB are calibrated to empirical evidence limited to
less than 3°C temperature increase. Nordhaus warns against extrapolating damage functions
NERA Economic Consulting
32
outside of the range in the underlying studies: "In reality, estimates of damage functions are
virtually non-existent for temperature increases above 3°C." 63 However, as noted above, the
climate scenarios used by the IWG include temperature changes well outside of this range.
At temperature changes that are both relatively high and relatively low compared to the
temperature change used for the benchmark (4 °C) in the functions in the figure above, the
damage functions produce very different damage estimates. The damage functions shown in
Figure 6 project GDP losses that differ by at least a factor of 5, which is the difference that
occurs at the benchmark of 4 °C. However, they differ by much more at other temperature
changes. Figure 7 shows the ratio of damage estimates between the highest and lowest damage
function in Figure 6. Differences in the damage estimates across the four alternative functions
exceed a factor of 10 for AT less than about 3°C and greater that about 6°C. Given that
sensitivity in the damage estimate will translate into sensitivity in the SCC estimate, this large
uncertainty in damages likely implies substantial uncertainty in SCC estimates. 64
Figure 7. Ratio of High to Low GDP Loss at Corresponding Temperature Changes
25
„,20
tt'L
15
O
10
x
O
2
'2
5
0
5
Temperature Change (Degrees
63
10
15
Celsius)
Nordhaus and Sztorc, 2013, p. 11.
The SCC sensitivity resulting from damage function sensitivity cannot be determined without a comprehensive
sensitivity analysis using multiple runs of models with multiple different damage function settings, which was
outside the scope of this study. However, absolute differences in damage estimates that occur nearer to the present
time, and at temperature changes that are most frequently projected in IAM scenarios will have the most impact on
sensitivity of the ultimate SCC estimate. While the variation in GDP loss at low and high temperature changes
may be about the same in relative terms, the absolute impact on JAM damage estimates (and hence on the absolute
SCC estimates) is much smaller at the lower temperature changes.
64
NERA Economic Consulting
33
C. Differences Due to Use of the Exponential Function
The sensitivity analysis in Figures 6 and 7 only consider variations in the parameters of a
polynomial damage function. As noted in the prior section, Weitzman indicates a theoretical
preference for an exponential damage function.
Our numerical analyses indicate that the increase in damage estimates from using the exponential
form in place of its equivalent polynomial form is greater as a3 increases from 1 to 3 (the range
shown in Figure 6). Nevertheless, this change of functional form does not appear to result in
exceptionally large change in damage estimates within the ranges of temperature change
projected in the IWG's SCC estimates. For example, Figure 8 shows the exponential damage
function compared to the highest of the four polynomial damage functions in Figure 6. Although
the exponential form does result in greater damage estimates relative to the comparable
polynomial form, the numerical differences are not as large as those due to various settings of
parameters of the polynomial mentioned in the IAM literature. However, if the exponential form
were to be included as another uncertainty in the sensitivity analysis that produced Figure 6, the
range in that figure would be wider at temperature changes above about 5 °C. (More specifically,
the upper line in Figure 6 would rise from the level of the red line in Figure 8 to the level of the
blue line in Figure 8.)
Figure 8. Damage Estimates from Exponential Compared to Polynomial Damage Function with
Same Parameters (using highest damage function in Figure 6)
100%
Exponential of Cubic:
1-exp(-a, * AT*3)
41
80%
a
E
A 20%
0%
0
2.5
5
7.5
10
12.5
15
Temperature Change (Degrees Celsius)
Note that Weitzman's concern is that the ranges in projected temperature changes themselves are
understated, and it is in the higher ranges that an exponential form of the damage function will
produce noticeably higher losses than the standard polynomial form. In the end, Weitzman
NERA Economic Consulting
34
concludes IAMs are insufficient for dealing with the kind of uncertainty that climate change
actually poses. Nevertheless, it appears that greater risk of extreme temperature or other climatic
outcomes also needs to be incorporated into IAMs' scientific assumptions before his concerns
with the damage functional form can become apparent in IAM model results.
D. Summary of Sensitivity Analyses
The estimates of global damages due to a given temperature change can differ substantially
depending upon the parameters of the presumed damage function. Our sensitivity analyses show
large differences in global damages over plausible ranges of two elements of the damage
function: (1) the exponent of the damage function and (2) the point at which it is calibrated.
Moreover, the magnitude of the difference depends upon the level of temperature change, with
the sensitivity greater at higher temperature changes. Although the large temperature changes
are not as important in the early and middle years of the projections, these temperature changes
can be relevant in the later years of the projections. This sensitivity analysis understates the
range of uncertainty because it has assumed for simplicity that the parameter a i is always zero.
Incorporation of plausible non-zero values for a l would widen the estimated range on both sides
of the assumed calibration point.
In contrast to the sensitivity across different parameters of the damage function, the empirical
results are not as sensitive to a shift to an exponential functional form. Weitzman indicates a
preference for the exponential form as better capturing the effects of extreme events; but given
the way IAMs currently capture temperature change uncertainty, use of his proposed exponential
form does not lead to substantial sensitivity in damage estimates for most potential levels of
temperature change the IAMs consider. Nevertheless, if our sensitivity analysis on parameter
uncertainty were to be expanded to include consideration of Weitzman's preferred (exponential)
functional form in addition to the standard (polynomial) form, the estimated sensitivity range for
damage estimates would further widen at higher temperature changes.
NERA Economic Consulting
35
VIII. CONCLUSIONS
In contrast to the traditional effect-by-effect approach known as the "damage function method"
used in RIAs, the damage functions in IAMs tend to be aggregate relationships between global
temperature change and GDP losses that do not describe the specific impacts that lead to
damages due to temperature change. Thus, it is not possible to evaluate uncertainties in the
damage functions that underlie the SCC values by evaluating uncertainties in specific
relationships between detailed physical effects and global damages.
It also is not possible to evaluate effects of uncertainties in the damage function by considering
the specific SCC values reported by the IWG, which nevertheless differ by a factor of three to
eight (i.e., 200% to 700%), as was demonstrated by Table 1 in Section II. These differences do
not provide a clear indication of uncertainties regarding the damage function because there are
other aspects of the IAMs that also were not held constant in the IWG's analyses. However, it is
reasonable to expect that some of that variation in IWG's SCC values is due to differences in the
respective IAMs' damage functions. Our assessment of plausible damage function specifications
indicates that the full range of uncertainty in the damage function has not been addressed in the
IWG analysis and thus also is not reflected in the SCC values that the IWG presents.
The lack of clear theoretical or empirical bases for the specific damage functions used by the
IWG means that we cannot use empirical or conceptual considerations to reduce the range of
possible damage value results. Lacking clear guidance, modelers tend to rely on a familiar
polynomial function and limited empirical information. This specification of the damage
function is further complicated by the theoretical need to modify the damage function to prevent
it from allowing damages to exceed global output even though the likelihood of such a result is
relatively small.
The implication of these difficulties is that the parameter values and calibration procedure for the
damage functions used in the IAMs are largely ad hoc and arbitrary. Using a pre-determined
value for a3 and setting al to 0 means that the damage function is largely dictated by the
subjective choice of a benchmark estimate — often a single data point. Although the
mathematical form of the damage function is relatively simple, plausible parameters for this
mathematical formulation lead to very different estimates of global damages. We find, for
example, that possible damage estimates at a given point in time can differ by up to a factor of 20
within the range of parameters and range of temperature changes found in the IAM literature.
Even this analysis understates the range of uncertainty as it has assumed for simplicity that al is
0, and that there is no uncertainty about the appropriate functional form. A sensitivity analysis
that would also include alternative plausible non-zero values for al, and alternative functional
forms (e.g., considering exponential as well as polynomial formulations, as suggested by
Weitzman) would further increase the range of possible damage values.
NERA Economic Consulting
36
Modelers seem clearly to acknowledge the uncertainties in the damage functions in the IAMs.
Tol reminds the reader that "a lot of questionable assumptions" 65 had to be made in developing
the damage estimates. Nordhaus states that "providing reliable estimates of the damages from
climate change over the long run has proven extremely difficult." 66
Since the damage estimate is a central input to the SCC estimates, the large uncertainty in the
damage function translates into uncertainty in the SCC estimates that could be correspondingly
large. However, a comprehensive representation of damage function uncertainties — analyzed in
combination with the other IAM input uncertainties — is needed to characterize how much more
uncertain the IWG's SCC estimates would be as a result of that damage function uncertainty.
Neither the IWG nor we have conducted such an analysis.
65
Tol, 2002b, p. 136.
Nordhaus and Sztorc, 2013, p. 10.
NERA Economic Consulting
37
BIBLIOGRAPHY
Ackerman, F., and Munitz C. 2012. "Climate Damages in the FUND Model: A Disaggregated
Analysis," Ecological Economics, 77:219-224.
Ackerman, F., Stanton, E., and Bueno, R. 2010. "CRED: A New Model of Climate and
Development," Stockholm Environment Institute Working Paper WP-US-1003.
Ackerman, F., Stanton, E., Hope, C. and Alberth, S. 2009. "Did the Stern Review Underestimate
US and Global Climate Damages?" Energy Policy, 37(7):2717-2721.
Anthoff, D., and Tol, R. S. 2008. "The Climate Framework for Uncertainty, Negotiation and
Distribution (FUND), Technical Description, Version 3.3," retrieved November 2013,
from model website: http://www.fund-model.org/versions.
Anthoff, D., and Tol, R. S. 2010. "The Climate Framework for Uncertainty, Negotiation and
Distribution (FUND), Technical Description, Version 3.5," retrieved November 2013,
from model website: http://wwwfund-model.org/versions.
Anthoff, D., and Tol, R. S. 2012. "The Climate Framework for Uncertainty, Negotiation and
Distribution (FUND), Technical Description, Version 3.6," retrieved November 2013,
from model website: http://www.fund-model.org/versions.
Anthoff, D., and Tol, R. S. 2013. "The Climate Framework for Uncertainty, Negotiation and
Distribution (FUND), Technical Description, Version 3.7," retrieved November 2013,
from model website: http://www.fund-model.org/versions.
Cline, W. R. 1992. The Economics of Global Warming. Washington DC: Institute for
International Economics.
Dietz, S., Hope, C., Stern, N., and Zenghelis, D. 2007. "Reflections on the Stern Review (1): A
Robust Case for Strong Action to Reduce the Risks of Climate Change," World
Economics, 8(1):121-168.
EPA. 1988. Analysis of Benefits, Appendix A to Guidelines for Performing Regulatory Impact
Analysis, Office of Policy Planning and Evaluation. March (Reprinted March 1991).
Available: http ://yo sem ite1 .epa. gov/ee/epa/eerm.nsf/vwAN/EE-0228A-1 .pdf/$file/EE0228A-1 .pdf.
EPA. 1983. Guidelines for Performing Regulatory Impact Analysis. Office of Policy Planning
and Evaluation. December (Reprinted March 1991). Available:
http://yosemitel.epa.gov/ee/epa/eerm.nsf/vwAN/EE-0228A-1.pdf/$file/EE-0228A-1.pdf.
Fankhauser, S. 1994. "The Economic Costs of Global Warming Damage: A Survey," Global
Environmental Change, 4(4):301-309.
NERA Economic Consulting
38
Greenstone, M., Kopits, E., and Wolverton, A. 2013. "Developing a Social Cost of Carbon for a
US Regulatory Analysis: A Methodology and Interpretation," Review of Environmental
Economics and Policy, 7(1):23-46.
Hanemann, W. M. 2008. "What is the Economic Cost of Climate Change?" CUDARE Working
Papers. Available: http://escholarship.org/uc/item/9g11z5cc.
Hope, C. 2006. "The Marginal Impact of CO2 from PAGE2002: An Integrated Assessment
Model Incorporating the IPCC's Five Reasons for Concern," The Integrated Assessment
Journal, 6(1):19-56.
Hope, C. 2011. "The Social Cost of CO2 from the PAGE09 Model," Economics Discussion
Papers, No. 2011-39, Kiel Institute for the World Economy. Available:
http://www.economics-ejournal.org/economics/discussionpapers/2011-39.
Hope, C. 2011. "The PAGE09 Integrated Assessment Model: A Technical Description,"
Cambridge Judge Business School Working Paper, Cambridge. Available:
http://www.jbs.cam.ac.uk/fileadmin/user upload/research/workingpapers/wp1104.pdf.
Hope, C., Anderson, J., and Wenman, P. 1993. "Policy Analysis of the Greenhouse Effect: An
Application of the PAGE Model," Energy Policy, 21(3):327-338.
IPCC. 2001. Impacts, Adaptation, and Vulnerability, Contribution of Working Group II to the
Third Assessment Report, Cambridge University Press.
IPCC. 2007. "Summary for Policymakers," in Climate Change 2007: Impacts, Adaptation, and
Vulnerability, Contribution of Working Group II to the Fourth Assessment Report of the
Intergovernmental Panel on Climate Change, Cambridge: Cambridge University Press.
IWG, U.S. Government. 2010. Technical Support Document:-Social Cost of Carbon for
Regulatory Impact Analysis-Under Executive Order 12866, February. Available:
http://www.epa.gov/oms/climate/regulations/scc-tsd.pdf.
IWG, US Government. 2013a. Technical Support Document:-Technical Update of the Social
Cost of Carbon for Regulatory Impact Analysis-Under Executive Order 12866, May.
Available:
http://www.whitehouse.gov/sites/default/files/omb/inforeg/social cost of carbon for ria
2013 update.pdf.
IWG, US Government. 2013b. Technical Support Document-Technical Update of the Social
Cost of Carbon for Regulatory Impact Analysis- Under Executive Order 12866,
November. Available:
http://www.whitehouse.gov/sites/default/files/omb/assets/inforeg/technical-update-socialcost-of-carbon-for-regulator-impact-analysis.pdf.
Manne, A., Mendelsohn, R., and R. Richels. 1995. "MERGE: A Model for Evaluating Regional
and Global Effects of GHG Reduction Policies," Energy Policy, 23(1):17-34.
NERA Economic Consulting
39
Nordhaus, W. 1991. "Economic Approaches to Greenhouse Warming," In R. D. Dornbush, & J.
M. Poterba, Global Warming: Economic Policy Approaches (pp. 33-68). Cambridge:
MIT Press.
Nordhaus, W. 1992a. "Rolling the 'DICE': An Optimal Transition Path for Controlling
Greenhouse Gases," Cowles Foundation Discussion Paper No. 1019. Available:
http://cowles.econ.yale.edu/P/cd/d10a/d1019.pdf.
Nordhaus, W. 1992b. "The 'DICE' Model: Background and Structure of a Dynamic Integrated
Climate-Economy Model of the Economics of Global Warming," Cowles Foundation
Discussion Paper No. 1009. Available: http://cowles.econ.yale.edu/P/cd/d10a/d1009.pdf.
Nordhaus, W. 1994. "Expert Opinion on Climatic Change," American Scientist, 82:45-51.
Nordhaus, W. 2008. A Question of Balance, New Haven: Yale University Press.
Nordhaus, W. 2010. "Economic Aspects of Global Warming in a Post-Copenhagen
Envrionment," PNAS, 107(26):11721-11726.
Nordhaus, W. 2011. "Estimates of the Social Cost of Carbon: Background and Results from the
RICE-2011 Model," Cowles Foundation Discussion Paper No. 1826. Available:
http://www.econ.yale.edu/-nordhaus/homepage/documents/CFDP1826.pdf.
Nordhaus, W., and Boyer, J. 2000. Warming the World: Economic Models of Global Warming,
Cambridge: MIT Press.
Nordhaus, W., and Sztorc, P. 2013. "DICE 2013R: Introduction and User's Manual," retrieved
November 2013, from
http://www.econ.yale.edui-nordhaus/homepage/documents/DICE Manual 100413r1.pdf.
Peck, S., and Teisberg, T. 1992. "CETA: A Model for Carbon Emissions Trajectory Assessment,"
Energy Journal, 13(1):55-77.
Plambeck, E. L., and Hope, C. 1996. "PAGE95: An Updated Valuation of the Impacts of Global
Warming," Energy Policy, 24(9):783-793.
Pindyck, R. 2013. "Climate Change Policy: What Do the Models Tell Us?" Journal of
Economic Literature, 51(3):860-872.
Schneider, S., and Lane, J. 2007. "Integrated Assessment Modeling of Global Climate Change:
Much Has Been Learned—Still a Long and Bumpy Road Ahead," The Integrated
Assessment Journal, 5 (1):41-75.
Stanton, E., Ackerman, F., and Kartha, S. 2008. "Inside the Integrated Assessment Models: Four
Issues in Climate Economics," Stockholm Environment Institute Working Paper WP-US0801.
NERA Economic Consulting
40
Tol, R. S. 2002a. "Estimates of Damage Costs of Climate Change — Part 1: Benchmark
Estimates," Environmental and Resource Economics, 21:47-73.
Tol, R. S. 2002b. "Estimates of Damage Costs of Climate Change — Part 2: Dynamic Estimates,"
Environmental and Resource Economics, 21:135-160.
Tol, R. S. 2006. "The Climate Framework for Uncertainty, Negotiation and Distribution (FUND),
Technical Description, Version 2.8," retrieved November 2013, from model website:
http://wwwfund-model.orgiversions .
Tol, R. S. 2009. "The Economic Effects of Climate Change," Journal of Economic Perspectives,
23(2):29-51.
Tol, R. S., and Fankhauser, S. 1997. "On the Representation of Impact in Integrated Assessment
Models of Climate Change," Environmental Modeling and Assessment, 3:63-74.
Warren, R., Hope, C., Mastrandrea, M., Tol, R., Adger, N., and Lorenzoni, I. 2006. "Spotlighting
Impacts Functions in Integrated Assessment: Research Report Prepared for the Stern
Review on the Economics of Climate Change," Tyndall Centre for Climate Change
Research, Working Paper 91.
Weitzman, M. 2009. "On Modeling and Interpreting the Economics of Catastrophic Climate
Change," The Review of Economics and Statistics, 91(1):1-19.
Weitzman, M. 2010. "What is the 'Damages Function' for Global Warming — and What
Difference Might it Make?" Climate Change Economics, 1(1):57-69.
NERA Economic Consulting
41
APPENDIX A. EVOLUTION OF DAMAGE FUNCTION IN DICE
In this Appendix, we summarize how DICE's damage function has evolved over the years. This
summary includes the more prominently documented versions of DICE as a simple exposition of
how particular modelling decisions have changed over the years. Most notably, there is a
consistency in the use of a quadratic function of temperature change (though sea-level rise was
briefly included) while determinants of the coefficient, or benchmark estimate, changed in 1999
and again in 2013.
DICE 1992-1994
In Nordhaus's 1992 paper, the author states that net economic damages in the US from a 3°C
temperature increase would be roughly .25% of GNP. He claims that the global number actually
should be 1.33% to include non-market impacts and to reflect greater vulnerability to climate
damages in other parts of the world. 67
A quadratic function is assumed, supported by the "evidence about the nonlinear nature of
impact of temperature rise ... from studies of coastal impacts." 68
In summary, the early version of DICE used the following damage function is:
D(t) = .0133 * [T(t)/3]^2 * Q(t)
where t is time, D(t) is loss of global output, T(t) is temperature, and Q(t) is output.
RICE/DICE 1999
In the 1999 update to the model, the coefficients of DICE's damage function were chosen in
order to match results from RICE-99 (again, the quadratic form was assumed). 69 The resulting
damage function took the following form, which allows for benefits to output at low levels of
temperature change:
D(t) = -.0045 * T(t) + .0035 * T(t)^2
with D(t) now defined as climate damage as a fraction of net output and T(t) is temperature
increase from 1900.
Much more detail is given concerning calibration of the RICE-99 damage function. Nordhaus
employs 'impact indices' specific to each RICE region and sector (agriculture, 'other vulnerable
market', coastal, health, nonmarket time use, and settlements). These indices are constructed by
67
For a more detailed discussion, see Nordhaus, 1992.
68
Nordhaus, 1992, footnote on pg. 45.
69
For a more detailed discussion, see Nordhaus and Boyer, 2000.
NERA Economic Consulting
42
assuming a sector-specific functional form of the impact of temperature based on existing studies
and literature. For each region, the sector impacts are calculated for temperature changes of
2.5°C and 6°C then summed to come up with one damage value per region and temperature
change. With these two points, determining the values of the coefficients of the damage function
is then as simple as solving a system of two equations.
DICE 2007
Nordhaus follows the same approach as in 1999 in his 2007 update but with refined and revised
estimates at the region/sector leve1. 70 However, one result of the recalibration is that there are no
longer benefits at low levels of temperature change:
D(t) = .0028388 * T(t)A2
RICE/DICE 2010
As stated in the IWG 2013, Nordhaus made significant changes to the damage functions for the
2010 model-year version of DICE — he revised the relationship to explicitly include sea-level rise
in addition to temperature change. However, in RICE 2011 Nordhaus reverts back to damages
solely as a function of temperature change claiming that "damages can be reasonably well
approximated by a quadratic in temperature over the near term." 71 Nevertheless, the DICE 2010
damage function takes the following form (where SLR = Sea-level rise):
D(t) = .00205 * T(t)A2 + .0052 * SLR + .0031 * SLRA2
It should be noted that a technical description or user's manual does not exist for DICE 2010 so
less is known about the derivation of the parameters of model.
DICE-2013R
For the most recent version of DICE, Nordhaus appears to have abandoned his previous
calibration exercise of building up the damage function from calibrations to RICE (which, for its
part, aggregated across sector-specific damages). In his October 2013 model, it appears that
Nordhaus first fitted a quadratic damage function through estimates of output impacts in Tol's
2009 paper, then increased the impact by 25% to reflect non-monetized impacts. 72 It is worth
noting that 5 out of the 13 studies used to calibrate the 2013 damage function were previous
incarnations of Nordhaus's own impact estimations. The resulting damage function is:
D(t) = .00267 * T(t)A2
70
For a more detailed discussion, see Nordhaus, 2008.
71
Nordhaus, 2011, p. 5.
72
For a more detailed discussion, see Nordhaus, 2013.
NERA Economic Consulting
43
NERA
ECONOMIC CONSULTING
NERA Economic Consulting
1255 23rd Street NW
Washington, DC 20037
Tel: +1 202 466 3510
Fax: +1 202 466 3605
www.nera.com
Office of Management and Budget
February 26, 2014
Page 40
Attachment 3
DICE 2013. Introduction and User's Manual
William Nordhaus with Paul Sztorc. May, 2013
Pages 10-11
(5)
(t) = wi TAT (t) +i1 [[TAT (t) j 2
Equation (5) involves the economic impacts of climate change, which is the thorniest issue in
climate-change economics. These estimates are indispensable for making sensible decisions
about the appropriate balance between costly emissions reductions and climate damages.
However, providing reliable estimates of the damages from climate change over the long run has
proven extremely difficult.
The present study relies on estimates from earlier syntheses of the damages, with updates in light
of more-recent information. The basic assumption is that the damages from gradual and small
climate changes are modest, but that the damages rise non-linearly with the extent of climate
change. These estimates also assume that the damages are likely to be relatively larger for poor,
small, and tropical countries than for rich, large and mid-latitude countries.
The damage function in (5) has been dramatically simplified from earlier DICE/RICE versions.
Earlier versions relied on Nordhaus and Boyer (2000). However, further work indicated that
those estimates were increasingly dated and unreliable. The new model instead uses a highly
simplified damage function that relies on current estimates of the damage function. More
precisely, DICE-2013 uses estimates of monetized damages from the Tol (2009) survey as the
starting point. However, current studies generally omit several important factors (biodiversity,
ocean acidification, and political reactions), extreme events (sea-level rise, changes in ocean
circulation, and accelerated climate change), impacts that are inherently difficult to model
(catastrophic events and very long term warming), and uncertainty (of virtually all components
from economic growth to damages). I have added an adjustment of 25 percent of the
monetized damages to reflect these non-monetized impacts. While this is consistent with the
estimates from other studies (see Hope 2011, Anthoff and Tol 2010, FUND 2013), it is
recognized that this is largely a judgmental adjustment. The current version assumes that
damages are a quadratic function of temperature change and does not include sharp thresholds or
tipping points, but this is consistent with the survey by Lenton et al. (2008).
Office of Management and Budget
February 26, 2014
Page 41
Attachment 4
[Pindyck (2013).]
Journal of Economic Literature 2013, 51(3), 860-872
http://dx. dot org/10.1257/jel.51.3. 860
Climate Change Policy:
What Do the Models Tell Us?'
ROBERT S. PINDYCK *
Very little. A plethora of integrated assessment models (IAMs) have been constructed
and used to estimate the social cost of carbon (SCC) and evaluate alternative
abatement policies. These models have crucial flaws that make them close to useless
as tools for policy analysis: certain inputs (e.g., the discount rate) are arbitrary, but
have huge effects on the SCC estimates the models produce; the models' descriptions
of the impact of climate change are completely ad hoc, with no theoretical or empirical
foundation; and the models can tell us nothing about the most important driver of the
SCC, the possibility of a catastrophic climate outcome. IAM-based analyses of climate
policy create a perception of knowledge and precision, but that perception is illusory
and misleading. (JEL C51, Q54, Q58)
1.
limiting carbon emissions, and is the focus of
policy-oriented research on climate change.
So how large is the SCC? Here there is
plenty of disagreement. Some argue that climate change will be moderate, will occur in
the distant future, and will have only a small
impact on the economies of most countries.
This would imply that the SCC is small, perhaps only around $10 per ton of CO 2. Others
argue that without an immediate and stringent
GHG abatement policy, there is a reasonable
chance of substantial temperature increases
that might have a catastrophic economic
impact. If so, it would suggest that the SCC is
large, perhaps as high as 5200 per ton of CO 2. 1
Introduction
here is almost no disagreement among
economists that the full cost to society
T
of burning a ton of carbon is greater than its
private cost. Burning carbon has an external cost because it produces CO 2 and other
greenhouse gases (GHGs) that accumulate
in the atmosphere, and will eventually result
in unwanted climate change—higher global
temperatures, greater climate variability, and
possibly increases in sea levels. This external
cost is referred to as the social cost of carbon
(SCC). It is the basis for taxing or otherwise
* Massachusetts Institute of Technology. My thanks to
Millie Huang for her excellent research assistance, and
to Janet Currie, Christian Gollier, Chris Knittel, Charles
Kolstad, Bob Litterman, and Richard Schmalensee for
helpful comments and suggestions.
Go to http://dx.doi.org/10.1257/je1.51.3.860 to visit the
article page and view author disclosure statement(s).
I The SCC is sometimes expressed in terms of dollars
per ton of carbon. A ton of CO 2 contains 0.2727 tons of
carbon, so an SCC of $10 per ton of CO 2 is equivalent to
$36.67 per ton of carbon. The SCC numbers I present in
this paper are always in terms of dollars per ton of CO2.
860
Pindyck: Climate Change Policy: What Do the Models Tell Us?
Might we narrow this range of disagreement over the size of the SCC by carefully
quantifying the relationships between GHG
emissions and atmospheric GHG concentrations, between changes in GHG concentrations and changes in temperature (and other
measures of climate change), and between
higher temperatures and measures of welfare such as output and per capita consumption? In other words, might we obtain better
estimates of the SCC by building and simulating integrated assessment models (IAMB),
i.e., models that "integrate" a description of
GHG emissions and their impact on temperature (a climate science model) with projections of abatement costs and a description of
how changes in climate affect output, consumption, and other economic variables (an
economic model).
Building such models is exactly what some
economists interested in climate change policy have done. One of the first such models
was developed by William Nordhaus over
twenty years ago. 2 That model was an early
attempt to integrate the climate science and
economic aspects of the impact of GHG
emissions, and it helped economists understand the basic mechanisms involved. Even
if one felt that parts of the model were overly
simple and lacked empirical support, the
work achieved a common goal of economic
modeling: elucidating the dynamic relationships among key variables, and the implications of those relationships, in a coherent
and convincing way. Since then, the development and use of IAMB has become a growth
industry. (It even has its own journal, The
Integrated Assessment Journal.) The models
have become larger and more complex, but
unfortunately have not done much to better
elucidate the pathways by which GHG emissions eventually lead to higher temperatures,
which in turn cause (quantifiable) economic
2
See, for example, Nordhaus (1991, 1993a, 1993b).
861
damage. Instead, the raison d'etre of these
models has been their use as a policy tool.
The idea is that by simulating the models, we
can obtain reliable estimates of the SCC and
evaluate alternative climate policies.
Indeed, a U.S. Government Interagency
Working Group has tried to do just that. It
ran simulations of three different IAMB, with
a range of parameter values, discount rates,
and assumptions regarding GHG emissions,
to estimate the SCC. 3 Of course, different
input assumptions resulted in different SCC
estimates, but the Working Group settled
on a base case or "average" estimate of $21
per ton, which was recently updated to $33
per ton. 4 Other IAMB have been developed
and likewise used to estimate the SCC. As
with the Working Group, those estimates
vary considerably depending on the input
assumptions for any one IAM, and also vary
across IAMB.
Given all of the effort that has gone into
developing and using IAMB, have they helped
us resolve the wide disagreement over the
size of the SCC? Is the U.S. government
estimate of $21 per ton (or the updated estimate of $33 per ton) a reliable or otherwise
useful number? What have these IAMB (and
related models) told us? I will argue that the
answer is very little. As I discuss below, the
models are so deeply flawed as to be close to
useless as tools for policy analysis. Worse yet,
3 The three IAMS were DICE (Dynamic Integrated
Climate and Economy), PAGE (Policy Analysis of the
Greenhouse Effect), and FUND (Climate Framework for
Uncertainty, Distribution, and Negotiation). For descriptions of the models, see Nordhaus (2008), Hope (2006),
and Tol (2002a, 2002b).
4 See Interagency Working Group on Social Cost of
Carbon (2010). For an illuminating and very readable discussion of the Working Group's methodology, the models
it used, and the assumptions regarding parameters, GHG
emissions, and other inputs, see Greenstone, Kopits, and
Wolverton (2011). The updated study used new versions
of the DICE, PAGE, and FUND models, and arrived at
a new "average" estimate of $33 per ton for the SCC. See
Interagency Working Group on Social Cost of Carbon
(2013).
Journal of Economic Literature, Vol. LI (September 2013)
862
their use suggests a level of knowledge and
precision that is simply illusory, and can be
highly misleading.
The next section provides a brief overview
of the IAM approach, with a focus on the
arbitrary nature of the choice of social welfare function and the values of its parameters.
Using the three models that the Interagency
Working Group chose for its assessment of
the SCC as examples, I then discuss two
important parts of JAMS where the uncertainties are greatest and our knowledge is
weakest—the response of temperature to an
increase in atmospheric CO 2, and the economic impact of higher temperatures. I then
explain why an evaluation of the SCC must
include the possibility of a catastrophic outcome, why IAMs can tell us nothing about
such outcomes, and how an alternative and
simpler approach is likely to be more illuminating. As mentioned above, the number of
IAMs in existence is large and growing. My
objective is not to survey the range of IAMs
or the JAM-related literature, but rather to
explain why climate change policy can be
better analyzed without the use of IAMs.
2.
Integrated Assessment Models
Most economic analyses of climate change
policy have six elements, each of which can
be global in nature or disaggregated on a
regional basis. In an IAM-based analysis,
each of these elements is either part of the
model (determined endogenously), or else is
an exogenous input to the model. These six
elements can be summarized as follows:
I. Projections of future emissions of a CO 2
equivalent (CO2e) composite (or individual GHGs) under "business as usual"
(BAU) and one or more abatement
scenarios. Projections of emissions in
turn require projections of both GDP
growth and "carbon intensity," i.e., the
amount of CO2e released per dollar of
GDP, again under BAU and alternative
abatement scenarios, and on an aggregate or regionally disaggregated basis.
2. Projections of future atmospheric CO 2e
concentrations resulting from past, current, and future CO 2e emissions. (This
is part of the climate science side of an
IAM.)
3. Projections of average global (or
regional) temperature changes—and
possibly other measures of climate
change such as temperature and rainfall variability, hurricane frequency, and
sea level increases—likely to result over
time from higher CO 2e concentrations.
(This is also part of the climate science
side of an IAM.)
4. Projections of the economic impact,
usually expressed in terms of lost GDP
and consumption, resulting from higher
temperatures. (This is the most speculative element of the analysis, in part
because of uncertainty over adaptation
to climate change.) "Economic impact"
includes both direct economic impacts
as well as any other adverse effects of
climate change, such as social, political, and medical impacts, which under
various assumptions are monetized and
included as part of lost GDP.
5. Estimates of the cost of abating GHG
emissions by various amounts, both now
and throughout the future. This in turn
requires projections of technological
change that might reduce future abatement costs.
6. Assumptions about social utility and the
rate of time preference, so that lost consumption from expenditures on abatement can be valued and weighed against
future gains in consumption from the
Pindyck: Climate Change Policy: What Do the Models Tell Us?
reductions in warming that abatement
would bring about.
These elements are incorporated in the
work of Nordhaus (2008), Stern (2007),
and others who evaluate abatement policies though the use of IAMB that project
emissions, CO2e concentrations, temperature change, economic impact, and costs
of abatement. Interestingly, however,
Nordhaus (2008), Stern (2007), and others come to strikingly different conclusions
regarding optimal abatement policy and the
implied SCC. Nordhaus (2008) finds that
optimal abatement should initially be very
limited, consistent with an SCC around $20
or less, while Stern (2007) concludes that
an immediate and drastic cut in emissions
is called for, consistent with an SCC above
200. 5 Why the huge difference? Because
the inputs that go into the models are so
different. Had Stern used the Nordhaus
assumptions regarding discount rates,
abatement costs, parameters affecting temperature change, and the function determining economic impact, he would have
also found the SCC to be low. Likewise, if
Nordhaus had used the Stern assumptions,
he would have obtained a much higher
SCC. 6
2.1 What Goes In and What Comes Out
And here we see a major problem with
IAM-based climate policy analysis: the
modeler has a great deal of freedom in
choosing functional forms, parameter values, and other inputs, and different choices
5 In an updated study, Nordhaus (2011) estimates the
SCC to be $12 per ton of CO 2.
6 Nordhaus (2007), Weitzman (2007), Mendelsohn
(2008), and others argue (and I would agree) that the Stern
study (which used a version of the PAGE model) makes
assumptions about temperature change, economic impact,
abatement costs, and discount rates that are generally
outside the consensus range. But see Stern (2008) for a
detailed (and very readable) explanation and defense of
these assumptions.
863
can give wildly different estimates of the
SCC and the optimal amount of abatement.
You might think that some input choices
are more reasonable or defensible than others, but no, "reasonable" is very much in
the eye of the modeler. Thus these models
can be used to obtain almost any result one
desires.'
There are two types of inputs that lend
themselves to arbitrary choices. The first
is the social welfare (utility) function and
related parameters needed to value and
compare current and future gains and losses
from abatement. The second is the set of
functional forms and related parameters that
determine the response of temperature to
changing CO,e concentrations and (especially) the economic impact of rising temperatures. I discuss the social welfare function
here, and leave the functional forms and
related parameters to later when I discuss
the "guts" of these models.
2.2 The Social Welfare Function
Most models use a simple framework for
valuing lost consumption at different points
in time: time-additive, constant relative risk
aversion (CRRA) utility, so that social welfare is
(1)
W=
1 - 77
E0 f ()° C"e -st dt,
where 77 is the index of relative risk aversion (IRRA) and S is the rate of time preference. Future consumption is unknown,
so I included the expectation operator E,
although most IAMB are deterministic in
nature. Uncertainty, if incorporated at all,
is usually analyzed by finning Monte Carlo
simulations in which probability distributions
7 A colleague of mine once said "I can make a model tie
my shoe laces."
864
Journal of Economic Literature, Vol. LI (September 2013)
are attached to one or more parameters. 8
Equation (1) might be applied to the United
States (as in the Interagency Working Group
study), to the entire world, or to different
regions of the world.
I will put aside the question of how meaningful equation (1) is as a welfare measure,
and focus instead on the two critical parameters, 6 and 77. We can begin by asking what is
the "correct" value for the rate of time preference, 6? This parameter is crucial because
the effects of climate change occur over very
long time horizons (50 to 200 years), so a
value of 6 above 2 percent would make it hard
to justify even a very moderate abatement
policy. Financial data reflecting investor
behavior and macroeconomic data reflecting
consumer and firm behavior suggest that 6 is
in the range of 2 to 5 percent. While a rate
in this range might reflect the preferences
of investors and consumers, should it also
reflect intergenerational preferences and
thus apply to time horizons greater than fifty
years? Some economists (e.g., Stern 2008
and Heal 2009) have argued that on ethical
grounds 6 should be zero for such horizons,
i.e., that it is unethical to discount the welfare of future generations relative to our own
welfare. But why is it unethical? Putting aside
their personal views, economists have little
to say about that question. 9 I would argue
that the rate of time preference is a policy
parameter, i.e., it reflects the choices of policy makers, who might or might not believe
8 A recent exception is Cai, Judd, and Lontzek (2013),
who developed a stochastic dynamic programming version
of the Nordhaus DICE model. Also, Kelly and Kolstad
(1999) show how Bayesian learning can affect policy in a
model with uncertainty.
9 Suppose John and Jane both have the same incomes.
John saves 10 percent of his income every year in order to
help finance the college educations of his (yet-to-be-born)
grandchildren, while Jane prefers to spend all of her disposable income on sports cars, boats, and expensive wines.
Does John's concern for his grandchildren make him more
ethical than Jane? Many people might say yes, but that
answer would be based on their personal values rather than
economic principles.
(or care) that their policy decisions reflect
the values of voters. As a policy parameter,
the rate of time preference might be positive, zero, or even negative.'" The problem
is that if we can't pin down 6, an IAM can't
tell us much ; any given IAM will give a wide
range of values for the SCC, depending on
the chosen value of 6.
What about 77, the IRRA? The SCC that
comes out of almost any IAM is also very
sensitive to this parameter. Generally, a
higher value of 71 will imply a lower value
of the SCC." So what value for m should be
used for climate policy? Here, too, economists disagree. The macroeconomics and
finance literatures suggest that a reasonable
range is from about 1.5 to at least 4. As a
policy parameter, however, we might consider the fact that 77 also reflects aversion to
consumption inequality (in this case across
generations), suggesting a reasonable range
of about 1 to 3. 12 Either way, we are left with
a wide range of reasonable values, so that any
given TAM can give a wide range of values
for the SCC.
Disagreement over 6 and n boils down to
disagreement over the discount rate used to
1 () Why negative? One could argue, perhaps based on
altruism or a belief that human character is improving over
time, that the welfare of our great-grandchildren should be
valued more highly than our own.
11 The larger is n the faster the marginal utility of consumption declines as consumption grows. Since consumption is expected to grow, the value of additional future
consumption is smaller the larger is 7/. But ri also measures
risk aversion; if future consumption is uncertain, a larger
n makes future welfare smaller, raising the value of additional future consumption. Most models show that unless
risk aversion is extreme (e.g., 97 is above 4), the first effect
dominates, so an increase in n (say from 1 to 4) will reduce
the benefits from an abatement policy. See Pindyck (2012)
for examples.
12 If a future generation is expected to have twice the
consumption as the current generation, the marginal utility of consumption for the future generation is 1/2 1/ as
large as for the current generation, and would be weighted
accordingly in any welfare calculation. Values of above
3 or 4 imply a relatively very small weight for the future
generation, so one could argue that a smaller value is more
appropriate.
,
Pindyck: Climate Change Policy: What Do the Models Tell Us?
put gains and losses of future consumption
(as opposed to utility) in present value terms.
In the simplest (deterministic) Ramsey
framework, that discount rate is R = 6 + 77g,
where g is the real per capita growth rate of
consumption, which historically has been
around 1.5 to 2 percent per annum, at least
for the United States. Stern (2007), citing
ethical arguments, sets 6 0 and 7] = 1 , so
that R is small and the estimated SCC is very
large. By comparison, Nordhaus (2008) tries
to match market data, and sets 6 = 1.5 percent and 77 = 2, so that R 5.5 percent and
the estimated SCC is far smaller.'`' Should
the discount rate be based on "ethical"
arguments or market data? And what ethical arguments and what market data? The
members of the Interagency Working Group
got out of this morass by focusing on a middle-of-the-road discount rate of 3 percent,
without taking a stand on whether this is the
correct" rate.
"
3.
The Guts of the Models
Let's assume for the moment that economists could agree on the "correct" value for
the discount rate R. Let's also assume that
they (along with climate scientists) could also
agree on the rate of emissions under BAU
and one or more abatement scenarios, as
well as the resulting time path for the atmospheric CO2e concentration. Could we then
use one or more IAMs to produce a reliable
estimate of the SCC? The answer is no, but
to see why, we must look at the insides of the
models. For some of the larger models, the
"guts" contain many equations and can seem
intimidating. But in fact, there are only two
key organs that we need to dissect. The first
13 Uncertainty over consumption growth or over the
discount rate itself can reduce R, and depending on the
type of uncertainty, lead to a time-varying R. See Gollier
(2013) for an excellent treatment of the effects of uncertainty on the discount rate. Weitzman (2013) shows how
the discount rate could decline over time.
865
translates increases in the CO 2e concentration to increases in temperature, a mechanism that is referred to as climate sensitivity.
The second translates higher temperatures
to reductions in GDP and consumption, i.e.,
the damage function.
3.1 Climate Sensitivity
Climate sensitivity is defined as the temperature increase that would eventually result
from an anthropomorphic doubling of the
atmospheric CO 2e concentration. The word
"eventually" means after the world's climate
system reaches a new equilibrium following
the doubling of the CO 2e concentration, a
period of time in the vicinity of fifty years.
For some of the simpler IAMs, climate sensitivity takes the form of a single parameter; for
larger and more complicated models, it might
involve several equations that describe the
dynamic response of temperature to changes
in the CO2e concentration. Either way, it can
be boiled down to a number that says how
much the temperature will eventually rise if
the CO2e concentration were to double. And
either way, we can ask how much we know
or don't know about that number. This is an
important question because climate sensitivity is an exogenous input into each of the
three IAMs used by the Interagency Working
Group to estimate the SCC.
Here is the problem: the physical mechanisms that determine climate sensitivity involve crucial feedback loops, and the
parameter values that determine the strength
(and even the sign) of those feedback loops
are largely unknown, and for the foreseeable future may even be unknowable. This is
not a shortcoming of climate science; on the
contrary, climate scientists have made enormous progress in understanding the physical
mechanisms involved in climate change. But
part of that progress is a clearer realization
that there are limits (at least currently) to our
ability to pin down the strength of the key
feedback loops.
866
Journal of Economic, Literature, Vol. LI (September 2013)
The Intergovernmental Panel on Climate
Change (2007) (IPCC) surveyed twenty-two
peer-reviewed published studies of climate
sensitivity and estimated that they implied
an expected value of 2.5° C to 3.0° C for climate sensitivity." Each of the individual
studies included a probability distribution
for climate sensitivity, and by putting the distributions in a standardized form, the IPCC
created a graph that showed all of the distributions in a summary form. A number of
studies—including the Interagency Working
Group study—used the IPCC's results to
infer and calibrate a single distribution for
climate sensitivity, which in turn could be
used to run alternative simulations of one or
more IAMs. 15
Averaging across the standardized distributions summarized by the IPCC suggests
that the 95th percentile is about 7°C, i.e.,
there is roughly a 5 percent probability that
the true climate sensitivity is above 7°C. But
this implies more knowledge than we probably have. This is easiest to see in the relatively simple climate model developed by
Roe and Baker (2007). Using their notation,
let Ao be climate sensitivity in the absence of
any feedback effects, i.e., absent feedback
effects, a doubling of the atmospheric CO 2e
concentration would lead to an increase in
radiative forcing that would in turn cause a
temperature increase of AT o = VC. But as
Roe and Baker explain, the initial temperature increase AT0 "induces changes in the
underlying processes . . which modify the
effective forcing, which, in turn, modifies
ST." Thus the actual climate sensitivity is
given by
A=
(2)
where f (0 < f < 1.) is the total feedback factor (which in a more complete and complex
model would incorporate several feedback
effects).
Unfortunately, we don't know the value
of f Roe and Baker point out that if we
knew the mean and standard deviation off,
denoted byf and Uf respectively, and if af is
small, then the standard deviation of A would
be proportional to af/(1 —f) 2 . Thus uncertainty over A is greatly magnified by uncertainty over f, and becomes very large if f is
close to I. Likewise, if the true value off is
close to I., climate sensitivity would be huge.
As an illustrative exercise, Roe and Baker
assume that f is normally distributed (with
mean f and standard deviation ad, and
derive the resulting distribution for A, climate sensitivity. Given their choice of f and
af , the resulting median and 95th percentile
are close to the corresponding numbers that
come from averaging across the standardized
distributions summarized by the IPCC.'
The Interagency Working Group calibrated the Roe—Baker distribution to fit
the composite IPCC numbers more closely,
and then applied that distribution to each
of the three IAMB as a way of analyzing the
16 The
Roe-Baker distribution is given by:
g (A if, 0 f, 0)
14 The
IPCC also provides a detailed and readable
overview of the physical mechanisms involved in climate
change, and the state of our knowledge regarding those
mechanisms.
15 Newbold and Daigneault (2009) and Pindyck (2012)
(who fit a gamma distribution to the IPCC's summary
graph) used the distribution to infer the implications of
uncertainty over climate sensitivity for abatement policy.
But as discussed below, they probably underestimated the
extent of the uncertainty.
Ao
—
-
1
0-.f V271- z2
exp
(
2
k 1 _f
- VI
1/z
where z = A + O. The parameter values are f = 0.797,
at- = 0.0441, 0 = 2.13. This distribution is fat-tailed, i.e.,
declines to zero more slowly than exponentially. Weitzman
(2009) has shown that parameter uncertainty can lead to a
fat-tailed distribution for climate sensitivity, and that this
implies a relatively high probability of a catastrophic outcome, which in turn suggests that the value of abatement
is high. Pindyck (2011a) shows that a fat-tailed distribution
by itself need not imply a high value of abatement.
Pindyck: Climate Change Policy: What Do the Models Tell Us?
sensitivity of their SCC estimates to uncertainty over climate sensitivity.
Given the limited available information,
the Interagency Working Group did the best
it could. But it is likely that they—like others who have used IAMs to analyze climate
change policy—have understated our uncertainty over climate sensitivity. We don't know
whether the feedback factor f is in fact normally distributed (nor do we know its mean
and standard deviation). Roe and Baker simply
assumed a normal distribution. In fact, in an
accompanying article in the journal Science,
Allen and Frame (2007) argued that climate
sensitivity is in the realm of the "unknowable."
3.2 The Damage Function
When assessing climate sensitivity, we at
least have scientific results to rely on, and can
argue coherently about the probability distribution that is most consistent with those
results. When it comes to the damage function, however, we know almost nothing, so
developers of IAMs can do little more than
make up functional forms and corresponding
parameter values. And that is pretty much
what they have done.
Most IAMs (including the three that were
used by the Interagency Working Group to
estimate the SCC) relate the temperature
increase T to GDP through a "loss function" L(T), with L(0) = 1 and /AT) < 0.
Thus GDP at time t is GDPt = Lad GDP;,
where GDP; is what GDP would be if
there were no warming. For example, the
Nordhaus (2008) DICE model uses the following inverse-quadratic loss function:
(3)
L = 1/[1
7r1
T + 7r2(T)2].
Weitzman (2009) suggested the exponentialquadratic loss function:
(4)
L(T) = exp [-13 (T) 2] ,
which allows for greater losses when T is
large. But remember that neither of these
867
loss functions is based on any economic (or
other) theory. Nor are the loss functions that
appear in other IAMs. They are just arbitrary functions, made up to describe how
GDP goes down when T goes up.
The loss functions in PAGE and FUND,
the other two models used by the Interagency
Working Group, are more complex but
equally arbitrary. In those models, losses
are calculated for individual regions and (in
the case of FUND) individual sectors, such
as agriculture and forestry. But there is no
pretense that the equations are based on any
theory. When describing the sectoral impacts
in FUND, Tol (2002b) introduces equations
with the words "The assumed model is:"
(e.g., pages 137-39, emphasis mine), and at
times acknowledges that "The model used
here is therefore ad hoc" (142).
The problem is not that IAM developers
were negligent and ignored economic theory; there is no economic theory that can tell
us what L(T) should look like. If anything,
we would expect T to affect the growth rate
of GDP, and not the level. Why? First, some
effects of warming will be permanent; e.g.,
destruction of ecosystems and deaths from
weather extremes. A growth rate effect
allows warming to have a permanent impact.
Second, the resources needed to counter the
impact of warming will reduce those available for R&D and capital investment, reducing growth." Third, there is some empirical
support for a growth rate effect. Using data
IlAdaptation to rising temperatures is equivalent to
the cost of increasingly strict emission standards, which,
as Stokey (1998) has shown with an endogenous growth
model, reduces the rate of return on capital and lowers the growth rate. To see this, suppose total capital
K = Kp + Ka (T) , with KL(T) > 0, where is directly productive capital and Ka(T) is capital needed for adaptation
to the temperature increase T (e.g., stronger retaining
walls and pumps to counter flooding, more air conditioning and insulation, etc.). If all capital depreciates at
rate 6K, K,, = K — K„ = I — 6KK — K„(T)T, so the rate of
growth of is reduced. See Brock and Taylor (2010) and
Fankhauser and Tol (2005) for related analyses.
868
Journal of Economic Literature, Vol. LI (September 2013)
on temperatures and precipitation over fifty
years for a panel of 136 countries, Dell,
Jones, and Olken (2012) have shown that
higher temperatures reduce GDP growth
rates but not levels. Likewise, using data for
147 countries during 1950 to 2007, Bansal
and Ochoa (2011, 2012) show that increases
in temperature have a negative impact on
economic growth.' 8
Let's put the issue of growth rate versus
level aside and assume that the loss function
of eqn. (3) is a credible description of the
economic impact of higher temperatures.
Then the question is how to determine the
values of the parameters 7r i and 7r2. Theory
can't help us, nor is data available that could
be used to estimate or even roughly calibrate the parameters. As a result, the choice
of values for these parameters is essentially
guesswork. The usual approach is to select
values such that L(T) for T in the range of
2°C to 4°C is consistent with common wisdom regarding the damages that are likely to
occur for small to moderate increases in temperature. Most modelers choose parameters
so that L(1) is close to 1 (i.e., no loss), L(2)
is around 0.99 or 0.98, and L(3) or L(4) is
around 0.98 to 0.96. Sometimes these numbers are justified by referring to the IPCC
or related summary studies. For example,
Nordhaus (2008) points out that the 2007
IPCC report states that "global mean losses
could be 1-5 percent GDP for 4°C of warming" (51). But where did the IPCC get those
numbers? From its own survey of several
IAMs. Yes, it's a bit circular.
The bottom line here is that the damage
functions used in most IAMs are completely
made up, with no theoretical or empirical
foundation. That might not matter much if
18 See Pindyck (2011b, 2012) for further discussion and
an analysis of the policy implications of a growth rate versus level effect. Note that a climate-induced catastrophe,
on the other hand, could reduce both the growth rate and
level of GDP.
we are looking at temperature increases of 2
or 3°C, because there is a rough consensus
(perhaps completely wrong) that damages
will be small at those levels of warming. The
problem is that these damage functions tell
us nothing about what to expect if temperature increases are larger, e.g., 5°C or more.'
Putting T = 5 or T = 7 into equation (3) or
(4) is a completely meaningless exercise. And
yet that is exactly what is being done when
IAMs are used to analyze climate policy.
I do not want to give the impression that
economists know nothing about the impact
of climate change. On the contrary, considerable work has been done on specific aspects
of that impact, especially with respect to
agriculture. One of the earliest studies of
agricultural impacts, including adaptation, is Mendelsohn, Nordhaus, and Shaw
(1994); more recent ones include Deschenes
and Greenstone (2007) and Schlenker and
Roberts (2009). A recent study that focuses
on the impact of climate change on mortality, and our ability to adapt, is Deschenes and
Greenstone (2011). And recent studies that
use or discuss the use of detailed weather
data include Dell, Jones, and Olken (2012)
and Auffhammer et al. (2013). These are just
a few examples; the literature is large and
growing.
Statistical studies of this sort will surely
improve our knowledge of how climate
change might affect the economy, or at least
some sectors of the economy. But the data
used in these studies are limited to relatively
short time periods and small fluctuations in
temperature and other weather variables—
the data do not, for example, describe what
19 Some modelers are aware of this problem. Nordhaus
(2008) states: "The damage functions continue to be a
major source of modeling uncertainty in the DICE model"
(51). To get a sense of the wide range of damage numbers
that come from different models, even for T = 2 or 3°C,
see table 1 of Tol (2012). Stern (2013) argues that IAM
damage functions ignore a variety of potential climate
impacts, including possibly catastrophic ones.
Pindyck: Climate Change Policy: What Do the Models Tell Us?
has happened over twenty or fifty years following a 5°C increase in mean temperature. Thus these studies cannot enable us
to specify and calibrate damage functions of
the sort used in IAMs. In fact, those damage
functions have little or nothing to do with the
detailed econometric studies related to agricultural and other specific impacts.
4.
Catastrophic Outcomes
Another major problem with using IAMs
to assess climate change policy is that the
models ignore the possibility of a catastrophic climate outcome. The kind of outcome I am referring to is not simply a very
large increase in temperature, but rather
a very large economic effect, in terms of a
decline in human welfare, from whatever climate change occurs. That such outcomes are
ignored is not surprising; IAMs have nothing
to tell us about them. As I explained, IAM
damage functions, which anyway are ad hoc,
are calibrated to give small damages for small
temperature increases, and can say nothing
meaningful about the kinds of damages we
should expect for temperature increases of
5°C or more.
4.1 Analysis of Catastrophic Outcomes
For climate scientists, a "catastrophe" usually takes the form of a high temperature outcome, e.g., a 7°C or 8°C increase by 2100.
Putting aside the difficulty of estimating the
probability of that outcome, what matters in
the end is not the temperature increase itself,
but rather its impact. Would that impact be
"catastrophic," and might a smaller (and
more likely) temperature increase be sufficient to have a catastrophic impact?
Why do we need to worry about large temperature increases and their impact? Because
even if a large temperature outcome has
low probability, if the economic impact of
that change is very large, it can push up the
SCC considerably. As discussed in Pindyck
869
(2013a), the problem is that the possibility of
a catastrophic outcome is an essential driver
of the SCC. Thus we are left in the dark;
IAMs cannot tell us anything about catastrophic outcomes, and thus cannot provide
meaningful estimates of the SCC.
It is difficult to see how our knowledge of
the economic impact of rising temperatures
is likely to improve in the coming years.
More than temperature change itself, economic impact may be in the realm of the
"unknowable." If so, it would make little
sense to try to use an IAM-based analysis to
evaluate a stringent abatement policy. The
case for stringent abatement would have to
be based on the (small) likelihood of a catastrophic outcome in which climate change is
sufficiently extreme to cause a very substantial drop in welfare.
4.2 What to Do?
So how can we bring economic analysis
to bear on the policy implications of possible catastrophic outcomes? Given how little
we know, a detailed and complex modeling
exercise is unlikely to be helpful. (Even if we
believed the model accurately represented
the relevant physical and economic relationships, we would have to come to agreement
on the discount rate and other key parameters.) Probably something simpler is needed.
Perhaps the best we can do is come up with
rough, subjective estimates of the probability
of a climate change sufficiently large to have
a catastrophic impact, and then some distribution for the size of that impact (in terms,
say, of a reduction in GDP or the effective
capital stock).
The problem is analogous to assessing the
world's greatest catastrophic risk during the
Cold War—the possibility of a U.S.–Soviet
thermonuclear exchange. How likely was
such an event? There were no data or models
that could yield reliable estimates, so analyses had to be based on the plausible, i.e., on
events that could reasonably be expected to
Journal of Economic Literature, Vol. LI (September 2013)
870
play out, even with low probability. Assessing
the range of potential impacts of a thermonuclear exchange had to be done in much
the same way. Such analyses were useful
because they helped evaluate the potential
benefits of arms control agreements.
The same approach might be used to assess
climate change catastrophes. First, consider
a plausible range of catastrophic outcomes
(under, for example, BAU), as measured by
percentage declines in the stock of productive capital (thereby reducing future GDP).
Next, what are plausible probabilities?
Here, "plausible" would mean acceptable
to a range of economists and climate scientists. Given these plausible outcomes and
probabilities, one can calculate the present
value of the benefits from averting those outcomes, or reducing the probabilities of their
occurrence. The benefits will depend on
preference parameters, but if they are sufficiently large and robust to reasonable ranges
for those parameters, it would support a
stringent abatement policy. Of course this
approach does not carry the perceived precision that comes from an IAM-based analysis,
but that perceived precision is illusory. To
the extent that we are dealing with unknowable quantities, it may be that the best we
can do is rely on the "plausible."
5.
Conclusions
I have argued that IAMs are of little or
no value for evaluating alternative climate
change policies and estimating the SCC.
On the contrary, an IAM-based analysis suggests a level of knowledge and precision that
is nonexistent, and allows the modeler to
obtain almost any desired result because key
inputs can be chosen arbitrarily.
As I have explained, the physical mechanisms that determine climate sensitivity involve crucial feedback loops, and
the parameter values that determine the
strength of those feedback loops are largely
unknown. When it comes to the impact of
climate change, we know even less. JAM
damage functions are completely made up,
with no theoretical or empirical foundation.
They simply reflect common beliefs (which
might be wrong) regarding the impact of 2°C
or 3°C of warming, and can tell us nothing
about what might happen if the temperature increases by 5°C or more. And yet those
damage functions are taken seriously when
IAMs are used to analyze climate policy.
Finally, IAMs tell us nothing about the likelihood and nature of catastrophic outcomes,
but it is just such outcomes that matter most
for climate change policy. Probably the best
we can do at this point is come up with plausible estimates for probabilities and possible
impacts of catastrophic outcomes. Doing
otherwise is to delude ourselves.
My criticism of IAMs should not be taken
to imply that, because we know so little, nothing should be done about climate change
right now, and instead we should wait until
we learn more. Quite the contrary. One can
think of a GHG abatement policy as a form
of insurance: society would be paying for a
guarantee that a low-probability catastrophe
will not occur (or is less likely). As I have
argued elsewhere, even though we don't
have a good estimate of the SCC, it would
make sense to take the Interagency Working
Group's $21 (or updated S33) number as a
rough and politically acceptable starting
point and impose a carbon tax (or equivalent
policy) of that amount.'" This would help to
establish that there is a social cost of carbon,
and that social cost must be internalized in
the prices that consumers and firms pay.
(Yes, most economists already understand
this, but politicians and the public are a different matter.) Later, as we learn more about
the true size of the SCC, the carbon tax could
be increased or decreased accordingly.
2() See Pindyck (2013b). Litterman (2013) and National
Research Council (2011) come to a similar conclusion.
Pindyck: Climate Change Policy: What Do the Models Tell Us?
REFERENCES
Allen, Myles R., and David J. Frame. 2007. "Call Off
the Quest." Science 318 (5850): 582-83.
Auffhammer, Maximilian, Solomon M. Hsiang, Wolfram Schlenker, and Adam Sobel. 2013. "Using
Weather Data and Climate Model Output in Economic Analyses of Climate Change." National Bureau
of Economic Research Working Paper 19087.
Bansal, Ravi, and Marcelo Ochoa. 2011. "Welfare Costs
of Long-Term Temperature Shifts." National Bureau
of Economic Research Working Paper 17574.
Bansal, Ravi, and Marcelo Ochoa. 2012. "Temperature,
Aggregate Risk, and Expected Returns." National
Bureau of Economic Research Working Paper 17575.
Brock, William A., and M. Scott Taylor. 2010. "The
Green Solow Model." Journal of Economic Growth
15 (2): 127-53.
Cai, Yongyang, Kenneth L. Judd, and Thomas S.
Lontzek. 2013. "The Social Cost of Stochastic and
Irreversible Climate Change." National Bureau of
Economic Research Working Paper 18704.
Dell, Melissa, Benjamin F. Jones, and Benjamin A.
Olken. 2012. "Temperature Shocks and Economic
Growth: Evidence from the Last Half Century."
American Economic Journal: Macroeconomics 4 (3):
66-95.
Deschenes, Olivier, and Michael Greenstone. 2007.
"The Economic Impacts of Climate Change: Evidence from Agricultural Output and Random Fluctuations in Weather." American Economic Review 97
(1): 354-85.
Deschenes, Olivier, and Michael Greenstone. 2011.
"Climate Change, Mortality, and Adaptation: Evidence from Annual Fluctuations in Weather in the
US." American Economic Journal: Applied Economics 3 (4): 152-85.
Fankhauser, Samuel, and Richard S. J. Tol. 2005. "On
Climate Change and Economic Growth." Resource
and Energy Economics 27 (1): 1-17.
Gollier, Christian. 2013. Pricing the Planet's Future:
The Economics of Discounting in an Uncertain
World. Princeton and Oxford: Princeton University
Press.
Greenstone, Michael, Elizabeth Kopits, and Ann Wolverton. 2011. "Estimating the Social Cost of Carbon
for Use in U.S. Federal Rulemakings: A Summary
and Interpretation." National Bureau of Economic
Research Working Paper 16913.
Heal, Geoffrey. 2009. "Climate Economics: A Metareview and Some Suggestions for Future Research."
Review of Environmental Economics and Policy 3
(1): 4-21.
Hope, Chris. 2006. "The Marginal Impact of CO 2 from
PAGE2002: An Integrated Assessment Model Incorporating the IPCC's Five Reasons for Concern."
Integrated Assessment 6 (1): 19-56.
Intergovernmental Panel on Climate Change. 2007.
Climate Change 2007: Synthesis Report. Geneva:
Intergovernmental Panel on Climate Change.
Kelly, David L., and Charles Kolstad. 1999. "Bayesian
871
Learning, Growth, and Pollution." Journal of Economic Dynamics and Control 23 (4): 491-518.
Litterman, Bob. 2013. "What Is the Right Price for
Carbon Emissions?" Regulation 36 (2): 38-43.
Mendelsohn, Robert. 2008. "Is the Stern Review an
Economic Analysis?" Review of Environmental Economics and Policy 2 (1): 45-60.
Mendelsohn, Robert, William D. Nordhaus, and Daigee Shaw 1994. "The Impact of Global Warming on
Agriculture: A Ricardian Analysis." American Economic Review 84 (4): 753-71.
National Research Council. 2011. America's Climate
Choices. Washington, D.C.: National Academies
Press.
Newbold, Stephen C., and Adam Daigneault. 2009.
"Climate Response Uncertainty and the Benefits of
Greenhouse Gas Emissions Reductions." Environmental and Resource Economics 44 (3): 351-77.
Nordhaus, William D. 1991. "To Slow or Not to Slow:
The Economics of the Greenhouse Effect." Economic Journal 101 (407): 920-37.
Nordhaus, William D. 1993a. "Optimal GreenhouseGas Reductions and Tax Policy in the 'DICE' Model."
American Economic Review 83 (2): 313-17.
Nordhaus, William D. 1993b. "Rolling the 'DICE': An
Optimal Transition Path for Controlling Greenhouse
Gases." Resource and Energy Economics 15 (1):
27-50.
Nordhaus, William D. 2007. "A Review of the Stern
Review on the Economics of Climate Change." Journal of Economic Literature 45 (3): 686-702.
Nordhaus, William D. 2008. A Question of Balance:
Weighing the Options on Global Warming Policies.
New Haven and London: Yale University Press.
Nordhaus, William D. 2011. "Estimates of the Social
Cost of Carbon: Background and Results from the
RICE-2011 Model." National Bureau of Economic
Research Working Paper 17540.
Pindyck, Robert S. 2011a. "Fat Tails, Thin Tails, and
Climate Change Policy." Review of Environmental
Economics and. Policy 5 (2): 258-74.
Pindyck, Robert S. 2011b. "Modeling the Impact of
Warming in Climate Change Economics." In The
Economics of Climate Change: Adaptations Past and
Present, edited by Gary D. Libeca-D and Richard H.
Steckel, 47-72. Chicago and London: University of
Chicago Press.
Pindyck, Robert S. 2012. "Uncertain Outcomes and
Climate Change Policy." Journal of Environmental
Economics and Management 63 (3): 289-303.
Pindyck, Robert S. 2013a. "The Climate Policy
Dilemma." Review of Environmental Economics and
Policy 7 (2): 219-37.
Pindyck, Robert S. 2013b. "Pricing Carbon When We
Don't Know the Right Price." Regulation 36 (2):
43-46.
Roe, Gerard H., and Marcia B. Baker. 2007. "Why Is
Climate Sensitivity So Unpredictable?" Science 318
(5850): 629-32.
Schlenker, Wolfram, and Michael J. Roberts. 2009.
"Nonlinear Temperature Effects Indicate Severe
872
Journal of Economic Literature, Vol. LI (September 2013)
Damages to U.S. Crop Yields under Climate
Change." Proceedings of the National Academy of
Sciences 106 (37): 15594-98.
Stern, Nicholas. 2007. The Economics of Climate
Change: The Stern Review. Cambridge and New
York: Cambridge University Press.
Stern, Nicholas. 2008. "The Economics of Climate
Change." American Economic Review 98 (2): 1-37.
Stern, Nicholas. 2013. "The Structure of Economic
Modeling of the Potential Impacts of Climate
Change: Grafting Gross Underestimation of Risk
onto Already Narrow Science Models." Journal of
Economic Literature 51 (3): 838-59.
Stokey, Nancy L. 1998. "Are There Limits to Growth?"
International Economic Review 39 (1): 1-31.
Tol, Richard S. J. 2002a. "Estimates of the Damage
Costs of Climate Change, Part I: Benchmark Estimates." Environmental and Resource Economics 21
(1): 47-73.
Tol, Richard S. J. 2002b. "Estimates of the Damage
Costs of Climate Change, Part II: Dynamic Estimates." Environmental and Resource Economics 21
(2): 135-60.
Tol, Richard S. J. 2012. "Targets for Global Climate
Policy: An Overview" Unpublished.
U.S. Interagency Working Group on Social Cost of
Carbon. 2010. "Technical Support Document: Social
Cost of Carbon for Regulatory Impact Analysis."
Washington, D.C.: U.S. Government
U.S. Interagency Working Group on Social Costs
of Carbon. 2013. "Technical Support Document:
Technical Update of the Social Cost of Carbon for
Regulator)/ Impact Analysis." Washington, D.C.: U.S.
Government.
Weitzman, Martin L. 2007. "A Review of the Stern
Review on the Economics of Climate Change." Journal of Economic Literature 45 (3): 703-24.
Weitzman, Martin L. 2009. "On Modeling and Interpreting the Economics of Catastrophic Climate
Change." Review of Economics and Statistics 91 (1):
1-19.
Weitzman, Martin L. 2013. "Tail-Hedge Discounting
and the Social Cost of Carbon." Journal of Economic
Literature 51 (3): 873-82.
This article has been cited by:
1. Nicholas Stern. 2013. The Structure of Economic Modeling of the Potential Impacts of Climate
Change: Grafting Gross Underestimation of Risk onto Already Narrow Science Models. Journal of
Economic Literature 51:3, 838-859. [Abstract] [View PDF article] [PDF with links]
Office of Management and Budget
February 26, 2014
Page 42
Attachment 5:
5 sets of Associations' Comments on DOE Energy Conservation Standards:
- Metal Halide Lamp Fixtures
- Walk-In Coolers and Freezers
- Commercial Refrigeration Equipment
- Residential Furnace Fans
- Microwave Ovens (Reconsideration)
yia
energ
A
NATIONAL ASSOCIATION 0
1 Manufacturers
1 0 FENt
4,
ti
Portland Cement Association
314 ER
o vc,
U.S. CHAMBER OF COMMERCE
AFPNI
American
Fuel & Petrochemical
Manufacturers
THE AMERICAN RESOURCE
American
Forest & Paper
. Association
doe
October 21, 2013
VIA ELECTRONIC FILING AND ELECTRONIC MAIL
U.S. Department of Energy
Office of Energy Efficiency and Renewable Energy
Building Technologies Program, EE-2J
1000 Independence Ave. SW
Washington, D.C. 20585-0121
Attn: Lucy deButts (metal halide lamp [email protected] )
U.S. Department of Energy
October 21, 2013
Page 2 of 8
Re: Docket No. EERE-2009-BT-STD-0018: Energy Conservation Program:
Energy Conservation Standards for Metal Halide Lamp Fixtures; Proposed
Rule; Federal Register Vol. 78, Number 161 (Tuesday, August 20, 2013)
Dear Sir/Madam:
The U.S. Chamber of Commerce, American Forest & Paper Association, American Fuel
& Petrochemical Manufacturers, American Petroleum Institute, Council of Industrial Boiler
Owners, National Association of Manufacturers, National Mining Association, and Portland
Cement Association (collectively, the "Associations") offer these comments on the Department
of Energy ("DOE")'s proposed rule for Energy Conservation Standards for Metal Halide Lamp
Fixtures, 78 F.R. 51463 (August 20, 2013) ("Metal Halide Lamp Proposed Rule" or "Proposed
Rule"). For the reasons discussed below, the Associations believe that the "social cost of
carbon" ("SCC") should be withdrawn as a basis for the Metal Halide Lamp Proposed Rule, and
that the SCC calculation should not be used in any rulemaking and/or policymaking until it
undergoes a more rigorous notice, review and comment process as outlined below.
The U.S. Chamber of Commerce ("Chamber") is the world's largest business
federation, representing the interests of more than three million businesses and organizations of
all sizes, sectors, and regions, as well as state and local chambers and industry associations, and
dedicated to promoting, protecting, and defending America's free enterprise system.
The American Forest & Paper Association ("AF&PA") serves to advance a sustainable
U.S. pulp, paper, packaging, and wood products manufacturing industry through fact-based
public policy and marketplace advocacy. AF&PA member companies make products essential
for everyday life from renewable and recyclable resources and are committed to continuous
improvement through the industry's sustainability initiative - Better Practices, Better Planet
2020. The forest products industry accounts for approximately 4.5 percent of the total U.S.
manufacturing GDP, manufactures approximately $200 billion in products annually, and
employs nearly 900,000 men and women. The industry meets a payroll of approximately $50
billion annually and is among the top 10 manufacturing sector employers in 47 states. AF&PA's
sustainability initiative - Better Practices, Better Planet 2020 - is the latest example of our
members' proactive commitment to the long-term success of our industry, our communities and
our environment. We have long been responsible stewards of our planet's resources. Our
member companies have collectively made significant progress in each of the following goals,
which comprise one of the most extensive quantifiable sets of sustainability goals for a U.S.
manufacturing industry: increasing paper recovery for recycling; improving energy efficiency;
reducing greenhouse gas emissions; promoting sustainable forestry practices; improving
workplace safety; and reducing water use.
The American Fuel & Petrochemical Manufacturers ("AFPM") is a trade association
representing high-tech American manufacturers of virtually the entire U.S. supply of gasoline,
diesel, jet fuel, other fuels and home heating oil, as well as the petrochemicals used as building
blocks for thousands of vital products in daily life. AFPM members make modern life possible
U.S. Department of Energy
October 21, 2013
Page 3 of 8
and keep America moving and growing as they meet the needs of our nation and local
communities, strengthen economic and national security, and support 2 million American jobs.
The American Petroleum Institute ("API") is a national trade association representing
over 500 member companies involved in all aspects of the oil and natural gas industry. API's
members include producers, refiners, suppliers, pipeline operators, and marine transporters, as
well as service and supply companies that support all segments of the industry. API and its
members are dedicated to meeting environmental requirements, while economically developing
and supplying energy resources for consumers.
The Council of Industrial Boiler Owners ("CIBO") is a broad-based association of
industrial boiler owners, architect-engineers, related equipment manufacturers, and University
affiliates with members representing 20 major industrial sectors. CIBO members have facilities
in every region of the country and a representative distribution of almost every type of industrial,
commercial and institutional (ICI) boiler and fuel combination currently in operation. CIBO was
formed in 1978 to promote the exchange of information within the industry and between industry
and government relating to energy and environmental equipment, technology, operations,
policies, law and regulations affecting industrial boilers. Since its formation, CIBO has been
active in the development of technically sound, reasonable, cost-effective energy and
environmental regulations for industrial boilers. CIBO supports regulatory programs that provide
industry with enough flexibility to modernize -- effectively and without penalty - the nation's
aging energy infrastructure, as modernization is the key to cost-effective environmental
protection.
The National Association of Manufacturers ("NAM") is the largest industrial trade
association in the U.S., representing over 12,000 small, medium and large manufacturers in all
50 states. NAM is the leading voice in Washington, D.C., for the manufacturing economy, which
provides millions of high wage jobs in the U.S. and generates more than $1.6 trillion in GDP. In
addition, two-thirds of NAM members are small businesses, which serve as the engine for job
growth. NAM's mission is to enhance the competitiveness of manufacturers and improve
American living standards by shaping a legislative and regulatory environment conducive to U.S.
economic growth.
The National Mining Association ("NMA") is a national trade association whose
members produce most of America's coal, metals, and industrial and agricultural minerals. Its
membership also includes manufacturers of mining and mineral processing machinery and
supplies, transporters, financial and engineering firms, and other businesses involved in the
nation's mining industries. NMA works with Congress and federal and state regulatory officials
to provide information and analyses on public policies of concern to its membership, and to
promote policies and practices that foster the efficient and environmentally sound development
and use of the country's mineral resources.
The Portland Cement Association ("PCA") represents 26 U.S. cement companies
operating 79 manufacturing plants in 34 states. Accounting for 78% of domestic cement-making
U.S. Department of Energy
October 21, 2013
Page 4 of 8
capacity, PCA members operate distribution centers in all 50 states and nearly every
congressional district.
These Associations' members have a strong interest in this Proposed Rule because they
may be impacted by the SCC precedent set in the rulemaking given that many of them
manufacture products that, when combusted, result in greenhouse gas ("GHG") emissions
(including carbon dioxide ("CO2")), and because, in the course of their business, they emit CO2.
Should this Administration, or any subsequent one, promulgate further regulation of these
products or emissions, such proposals and rules could potentially be based, in large part, on
either the 2010 or 2013 estimates of the SCC ("SCC Estimates") created by the federal
Interagency Working Group ("IWG"). Therefore, the Associations' members have a direct and
meaningful interest in ensuring that any estimates and applications of the SCC are based on
transparent processes, accurate information, rational assumptions, and are within the reach of the
current scientific understanding and impact models.'
I. THE SCC SHOULD UNDERGO A NOTICE AND COMMENT PROCESS
BEFORE IT IS USED IN THE METAL HALIDE LAMP PROPOSED RULE OR
ANY OTHER RULEMAKINGS.
The IWG has defined the SCC as "an estimate of the monetized damages associated with
an incremental increase in carbon emissions in a given year." In the Metal Halide Lamp
Proposed Rule, the DOE uses benefits derived from the SCC to justify the proposed energy
efficiency regulation. The DOE estimates that the Proposed Rule will have cumulative benefits
of $5.371 billion over a 30-year period (2016-2045) at a 3% discount rate. Of that $5.371 billion
in benefits, $3.748 billion is from lower consumer electricity costs, $1.532 billion is from the
SCC, and $91 million is from reduced NO„ emissions. The DOE also estimates that the
Proposed Rule will have cumulative costs of $1.294 billion. Notably, under Title III of the
Energy Policy and Conservation Act (42 U.S.C. 6295), the DOE's findings with regard to the
benefits of the Metal Halide Lamp Proposed Rule are legally sufficient for justifying the rule
without the inclusion of any benefits based upon the SCC.
While the DOE may rely heavily upon the SCC in claiming that the Metal Halide Lamp
Proposed Rule has benefits of $5.371 billion, that does not change the fact that the SCC has not
been adequately noticed and reviewed before being used in this Proposed Rule or any other
rulemaking. As described in the attached Petition for Correction pursuant to the Information
Quality Act, the Associations believe that the 2010 and 2013 Technical Support Documents and
SCC Estimates should be withdrawn and not used in any rulemaking and policymaking,
including the Metal Halide Lamp Proposed Rule, for the following reasons:
To be clear, we question the application of the 2010 and 2013 Interagency Working Group ("IWG") estimates of
the SCC, which are based on complex economic impacts hundreds of years in the future, which in turn are based on
present day understanding of current and future carbon emissions. We are not herein discussing the existence or
potential causes of climate change.
U.S. Department of Energy
October 21, 2013
Page 5 of 8
1. The SCC Estimates fail in terms of process and transparency. The SCC Estimates fail to
comply with OMB guidance for developing influential policy-relevant information under
the Information Quality Act. 2 The SCC Estimates are the product of an opaque process
and any pretensions to their supposed accuracy (and therefore usefulness in policymaking) are unsupportable.
2. The models with inputs (hereafter referred to as "the modeling systems") used for the
SCC Estimates and the subsequent analyses were not subject to peer review as
appropriate.
3. Moreover, even if the SCC Estimate development process was transparent, rigorous, and
peer-reviewed, the modeling conducted in this effort does not offer a reasonably
acceptable range of accuracy for use in policymaking.
4. The IWG has failed to disclose and quantify key uncertainties to inform decision makers
and the public about the effects and uncertainties of alternative regulatory actions as
required by OMB.
5. By presenting only global SCC estimates and downplaying domestic SCC estimates in
2013, the IWG has severely limited the utility of the SCC for use in benefit-cost analysis
and policymaking.
Given all of the concerns summarized above and detailed in the attached petition, neither the
2010 nor 2013 IWG estimates of SCC should be used in the Metal Halide Lamp Proposed Rule,
as well as any other rulemaking and policymaking until the SCC undergoes a more rigorous
notice, review and comment process.
II. THE METAL HALIDE LAMP PROPOSED RULE IS FLAWED IN OTHER
SIGNIFICANT WAYS.
The problems with the Metal Halide Lamp Proposed Rule go beyond the transparency
and process problems associated with the SCC, which the DOE relied upon in its cost-benefits
analysis. As outlined below, there are other errors and omissions in the DOE's cost-benefit
analysis that must be addressed.
A. The DOE Improperly Balances Costs and Benefits
An important principle of cost-benefit analysis is that costs and benefits must be
compared over the same time frame and within the same scope. The DOE's cost-benefit analysis
for the Proposed Rule violates this principle in both ways. With respect to the time frame, the
DOE calculates the present value of the costs of the Proposed Rule to consumers and
manufacturers over a 30-year period. In contrast, the annual SCC values used in the DOE's
2 The SCC Estimates also fail to comply with the OMB Bulletin for Agency Good Guidance Practices, which
requires pre-adoption public notice and comment for economically significant guidance documents. See OMB
Bulletin, 72 FR at 3440 (Sec. IV).
U.S. Department of Energy
October 21, 2013
Page 6 of 8
cost-benefit analysis reflect cumulative benefits that will accrue to individuals up to 300 years in
the future. According to the DOE analysis, the metal halide lamp energy efficiency standards
would lead to the removal of greenhouse gases (GHGs), which in turn would avoid global
warming damages and result in benefits over the next 300 years. The DOE's comparison of 30
years of cost to 300 hundred years of putative, future benefits is inconsistent and improper. 3
The DOE cost-benefit analysis also fails to balance the scope of costs and benefits.
Regulations result in both direct and indirect costs and benefits. Direct costs are the costs of
compliance immediately imposed by the regulation on the regulated entities, i.e. higher
manufacturing and consumer costs for energy-efficient metal halide lamps. Direct benefits
include any economic or health costs that would accrue immediately to the public from the
greater efficiency of metal halide lamps, i.e. lower operating costs from improved energy
efficiency of the metal halide lamps. Indirect costs would include the secondary and tertiary
effects that would be induced by the regulation. Here, the DOE expands its benefits calculations
to include indirect benefits without commensurately expanding its cost analysis to include
potential indirect costs. For example, the DOE analysis includes benefits based upon the SCC,
which would be classified as indirect benefits; however, it does not take into account indirect
costs, such as income loss and job search costs imposed on workers who might be displaced
because of higher metal halide lamp prices and reduced product demand. The DOE analysis for
the Proposed Rule should be balanced in the scope and time frame of any costs and benefits.
B. The DOE Ignores Global Benefits Reduction Rate
In justifying the benefits of the Metal Halide Lamp Proposed Rule, the DOE explicitly
ignores a recommendation from the IWG to reduce global benefits from the SCC in the context
of wholly domestic regulations. In relying upon the SCC in its cost-benefit analysis, the DOE
recognizes the benefits of reduced global CO 2 emissions resulting from the Proposed Rule, but it
fails to consider the correlating global costs of the Proposed Rule. According to the IWG, the
U.S. domestic SCC benefits for the reduction of one ton of CO2 amount to only 7% - 23% of the
global SCC benefits for the reduction of the same ton of CO2. In its Preliminary Technical
Support Document for the Proposed Rule, the DOE acknowledges and then disregards the IWG's
recommendation in the Proposed Rule, stating that "...the interagency group determined that a
range of values from 7 percent to 23 percent should be used to adjust the global SCC to calculate
domestic effects, although DOE will give preference to consideration of the global benefits of
reducing CO 2 emissions."4
3 Although the DOE presents carbon reduction values for each year of the 30-year period over which costs and
benefits are compared, the presentation is not complete. For example, the amount of $43 per ton that is listed in
Table V.9 of the Proposed Rule (see page 51514) is not the actual direct benefit in terms of the economic damages
avoided by eliminating a ton of carbon in the year 2020. Instead, it is the present value in 2020 (calculated at a 3%
discount rate) of the projected economic damage avoided over the future 300 years by all person in all places
worldwide if the subject ton of carbon eliminated in 2020 remains out of the emissions stream for all time forward.
4 See http://wwwl .eere.energy.gov/buildings/appliance standards/commercial/pdfs/mhlfpreanalysis chapter2.pdf,
Preliminary Technical Support Document, p. 64.
U.S. Department of Energy
October 21, 2013
Page 7 of 8
In ignoring the IWG's recommendation and failing to apply a reduction rate for the
global SCC benefits, the DOE's estimate of carbon-reduction benefits for the Proposed Rule is 4
to 14 times greater than it would have been if the DOE had followed the IWG's
recommendation. In other words, the domestic SCC benefits for the Proposed Rule would be
$107 million to $352 million under the IWG's recommendation, instead of the $1.532 billion
claimed by the DOE based on the global benefits perspective. The DOE's decision to depart
from the more appropriate IWG approach on this issue, with no justification or reasoning,
renders it arbitrary.
C. The DOE Fails to Consider EPA's Planned GHG Regulations on Power Plants
The DOE also fails to take into account in the Proposed Rule the Environmental
Protection Agency ("EPA")'s planned GHG regulations for new and existing power plants. In
the President's highly publicized Climate Action Plan, which was released on June 25, 2013, the
White House directed EPA to propose and issue regulations reducing GHG emissions from new
and existing power plants. The Climate Action Plan and accompanying Presidential
memorandum outlined detailed rulemaking schedules for both new and existing power plants.
Specifically, EPA would propose regulations for GHG emissions from new power plants by
September 20, 2013, and similar regulations for existing power plants by June 1, 2014. EPA
already fulfilled the first directive by releasing a proposed regulation for GHG emissions from
new power plants on September 20, 2013. Given the Agency's simultaneous announcement that
it would undertake a two-month outreach to stakeholders on the existing power plant rule, all
indications are that EPA will fulfill the second directive, too.
Despite all of the publicity and media coverage surrounding the Climate Action Plan and
EPA's strong suggestions that it would adhere to the President's directives, the DOE states in its
Preliminary Technical Support Document for the August 20, 2013 Proposed Rule that, "[i]n the
absence of any Federal emissions control regulation of power plant emissions of CO2, a DOE
standard is likely to result in reductions of these emissions." 5 While there may not have been
fully implemented GHG regulations in place for power plants on August 20, 2013, it is clear that
EPA is rapidly and purposefully moving toward proposing and finalizing those regulations. The
DOE's failure to even mention, much less consider, EPA's GHG regulations for power plants is
a significant flaw in its analysis.
III. CONCLUSION
For the reasons stated above, including the incorporation of the arguments posited in the
attached Petition, the DOE should withdraw the SCC calculation as a basis for the Metal Halide
Lamp Proposed Rule, and refrain from using the SCC in any other rulemaking or policymaking
until the SCC undergoes a more rigorous notice, review and comment process. Additionally, the
DOE should address and correct the errors outlined in Section II of these comments.
5 See http://www.regulations.govMdocumentDetail ;D=EERE-2009-BT-STD-0018-0021, Chapter 2, "Analytical
Framework," pages 2-60.
U.S. Department of Energy
October 21, 2013
Page 8 of 8
Thank you for the opportunity to participate in this proceeding. If you have any follow up
questions regarding these comments, please feel free to reach out to William L. Kovacs, Senior
Vice President of Environment, Technology & Regulatory Affairs at the U.S. Chamber of
Commerce at (202) 463-5457 or by e-mail: [email protected].
American Forest & Paper Association
American Fuel & Petrochemical Manufacturers
American Petroleum Institute
Council of Industrial Boiler Owners
National Association of Manufacturers
National Mining Association
Portland Cement Association
U.S. Chamber of Commerce
U.S. CHAMBER OF COMMERCE
A
Portland[ementissocialion
NATIONAL ASSOCIATION OF
I Manufacturers
AFPNI
American
Fuel & Petrochemical
Manufacturers
THE AMERICAN RESOURCE
American
Forest & Paper
Association
dine
November 12, 2013
VIA ELECTRONIC FILING AND ELECTRONIC MAIL
U.S. Department of Energy
Office of Energy Efficiency and Renewable Energy
Building Technologies Program, EE-2J
1000 Independence Ave. SW
Washington, D.C. 20585-0121
Attn: Brenda Edwards ([email protected] )
Re: Docket No. EERE-2008-BT-STD-0015: Energy Conservation Program:
Energy Conservation Standards for Walk-In Coolers and Freezers; Proposed
Rule; Federal Register Vol. 78, Number 176 (Wednesday, September 11,
2013); RIN 1904-AB86
U.S. Department of Energy
November 12, 2013
Page 2 of 10
Dear Sir/Madam:
The U.S. Chamber of Commerce, American Forest & Paper Association, American Fuel
& Petrochemical Manufacturers, American Petroleum Institute, Council of Industrial Boiler
Owners, National Association of Manufacturers, National Mining Association, and Portland
Cement Association (collectively, the "Associations") offer these comments on the Department
of Energy ("DOE")'s proposed rule for Energy Conservation Standards for Walk-In Coolers and
Freezers, 78 F.R. 55782 (September 11, 2013) ("Walk-In Coolers and Freezers Proposed Rule"
or "Proposed Rule"). For the reasons discussed below, the Associations believe that the "social
cost of carbon" ("SCC") should be withdrawn as a basis for the Walk-In Coolers and Freezers
Proposed Rule, and that the SCC calculation should not be used in any rulemaking and/or
policymaking until it undergoes a more rigorous notice, review and comment process as outlined
below.
The U.S. Chamber of Commerce ("Chamber") is the world's largest business
federation, representing the interests of more than three million businesses and organizations of
all sizes, sectors, and regions, as well as state and local chambers and industry associations, and
dedicated to promoting, protecting, and defending America's free enterprise system.
The American Forest & Paper Association ("AF&PA") serves to advance a sustainable
U.S. pulp, paper, packaging, and wood products manufacturing industry through fact-based
public policy and marketplace advocacy. AF&PA member companies make products essential
for everyday life from renewable and recyclable resources and are committed to continuous
improvement through the industry's sustainability initiative - Better Practices, Better Planet
2020. The forest products industry accounts for approximately 4.5 percent of the total U.S.
manufacturing GDP, manufactures approximately $200 billion in products annually, and
employs nearly 900,000 men and women. The industry meets a payroll of approximately $50
billion annually and is among the top 10 manufacturing sector employers in 47 states. AF&PA's
sustainability initiative - Better Practices, Better Planet 2020 - is the latest example of our
members' proactive commitment to the long-term success of our industry, our communities and
our environment. We have long been responsible stewards of our planet's resources. Our
member companies have collectively made significant progress in each of the following goals,
which comprise one of the most extensive quantifiable sets of sustainability goals for a U.S.
manufacturing industry: increasing paper recovery for recycling; improving energy efficiency;
reducing greenhouse gas emissions; promoting sustainable forestry practices; improving
workplace safety; and reducing water use.
The American Fuel & Petrochemical Manufacturers ("AFPM") is a trade association
representing high-tech American manufacturers of virtually the entire U.S. supply of gasoline,
diesel, jet fuel, other fuels and home heating oil, as well as the petrochemicals used as building
blocks for thousands of vital products in daily life. AFPM members make modern life possible
and keep America moving and growing as they meet the needs of our nation and local
communities, strengthen economic and national security, and support 2 million American jobs.
U.S. Department of Energy
November 12, 2013
Page 3 of 10
The American Petroleum Institute ("API") is a national trade association representing
over 500 member companies involved in all aspects of the oil and natural gas industry. API's
members include producers, refiners, suppliers, pipeline operators, and marine transporters, as
well as service and supply companies that support all segments of the industry. API and its
members are dedicated to meeting environmental requirements, while economically developing
and supplying energy resources for consumers.
The Council of Industrial Boiler Owners ("CIBO") is a broad-based association of
industrial boiler owners, architect-engineers, related equipment manufacturers, and University
affiliates with members representing 20 major industrial sectors. CIBO members have facilities
in every region of the country and a representative distribution of almost every type of industrial,
commercial and institutional (ICI) boiler and fuel combination currently in operation. CIBO was
formed in 1978 to promote the exchange of information within the industry and between industry
and government relating to energy and environmental equipment, technology, operations,
policies, law and regulations affecting industrial boilers. Since its formation, CIBO has been
active in the development of technically sound, reasonable, cost-effective energy and
environmental regulations for industrial boilers. CIBO supports regulatory programs that provide
industry with enough flexibility to modernize -- effectively and without penalty - the nation's
aging energy infrastructure, as modernization is the key to cost-effective environmental
protection.
The National Association of Manufacturers ("NAM") is the largest industrial trade
association in the U.S., representing over 12,000 small, medium and large manufacturers in all
50 states. NAM is the leading voice in Washington, D.C., for the manufacturing economy, which
provides millions of high wage jobs in the U.S. and generates more than $1.6 trillion in GDP. In
addition, two-thirds of NAM members are small businesses, which serve as the engine for job
growth. NAM's mission is to enhance the competitiveness of manufacturers and improve
American living standards by shaping a legislative and regulatory environment conducive to U.S.
economic growth.
The National Mining Association ("NMA") is a national trade association whose
members produce most of America's coal, metals, and industrial and agricultural minerals. Its
membership also includes manufacturers of mining and mineral processing machinery and
supplies, transporters, financial and engineering firms, and other businesses involved in the
nation's mining industries. NMA works with Congress and federal and state regulatory officials
to provide information and analyses on public policies of concern to its membership, and to
promote policies and practices that foster the efficient and environmentally sound development
and use of the country's mineral resources.
The Portland Cement Association ("PCA") represents 26 U.S. cement companies
operating 79 manufacturing plants in 34 states. Accounting for 78% of domestic cement-making
capacity, PCA members operate distribution centers in all 50 states and nearly every
congressional district.
U.S. Department of Energy
November 12, 2013
Page 4 of 10
These Associations' members have a strong interest in this Proposed Rule because they
may be impacted by the SCC precedent set in the rulemaking given that many of them
manufacture products that, when combusted, result in greenhouse gas ("GHG") emissions
(including carbon dioxide ("CO2")), and because, in the course of their business, they emit CO2.
Should this Administration, or any subsequent one, promulgate further regulation of these
products or emissions, such proposals and rules could potentially be based, in large part, on
either the 2010 or 2013 estimates of the SCC ("SCC Estimates") created by the federal
Interagency Working Group ("IWG"). Therefore, the Associations' members have a direct and
meaningful interest in ensuring that any estimates and applications of the SCC are based on
transparent processes, accurate information, rational assumptions, and are within the reach of the
current scientific understanding and impact models.'
I. THE SCC SHOULD UNDERGO A NOTICE AND COMMENT PROCESS
BEFORE IT IS USED IN THE WALK-IN COOLERS AND FREEZERS
PROPOSED RULE OR ANY OTHER RULEMAKINGS.
The IWG has defined the SCC as "an estimate of the monetized damages associated with
an incremental increase in carbon emissions in a given year." In the Walk-In Coolers and
Freezers Proposed Rule, the DOE uses benefits derived from the SCC to justify the proposed
energy efficiency regulation. The DOE estimates that the Proposed Rule will have cumulative
benefits of $41.1 billion over a 30-year period (2017-2046) at a 3% discount rate. Of that $41.1
billion in benefits, $31.6 billion is from lower consumer electricity costs, $9.0 billion is from the
SCC, and $550 million is from reduced NO„ emissions. The DOE also estimates that the
Proposed Rule will have cumulative costs of $7.2 billion. Notably, under Title III of the Energy
Policy and Conservation Act (42 U.S.C. 6295), the DOE's findings with regard to the benefits of
the Walk-In Coolers and Freezers Proposed Rule are legally sufficient for justifying the rule
without the inclusion of any benefits based upon the SCC.
While the DOE may include SCC benefits in claiming that the Walk-In Coolers and
Freezers Proposed Rule has benefits of $41.1 billion, that does not change the fact that the
IWG's SCC analysis has not been adequately noticed and reviewed before being used in this
Proposed Rule or any other rulemaking. As described in the attached Petition for Correction
pursuant to the Information Quality Act, the Associations believe that the 2010 and 2013
Technical Support Documents and SCC Estimates should be withdrawn and not used in any
rulemaking and policymaking, including the Walk-In Coolers and Freezers Proposed Rule, for
the following reasons:
1. The SCC Estimates fail in terms of process and transparency. The SCC Estimates fail to
comply with OMB guidance for developing influential policy-relevant information under
To be clear, we question the application of the 2010 and 2013 Interagency Working Group ("IWG") estimates of
the SCC, which are based on complex economic impacts hundreds of years in the future, which in turn are based on
present day understanding of current and future carbon emissions. We are not herein discussing the existence or
potential causes of climate change.
U.S. Department of Energy
November 12, 2013
Page 5 of 10
the Information Quality Act. 2 The SCC Estimates are the product of an opaque process
and any pretensions to their supposed accuracy (and therefore usefulness in policymaking) are unsupportable.
2. The models with inputs (hereafter referred to as "the modeling systems") used for the
SCC Estimates and the subsequent analyses were not subject to peer review as
appropriate.
3. Moreover, even if the SCC Estimate development process was transparent, rigorous, and
peer-reviewed, the modeling conducted in this effort does not offer a reasonably
acceptable range of accuracy for use in policymaking.
4. The IWG has failed to disclose and quantify key uncertainties to inform decision makers
and the public about the effects and uncertainties of alternative regulatory actions as
required by OMB.
5. By presenting only global SCC estimates and downplaying domestic SCC estimates in
2013, the IWG has severely limited the utility of the SCC for use in benefit-cost analysis
and policymaking.
Given all of the concerns summarized above and detailed in the attached petition, neither the
2010 nor 2013 IWG estimates of SCC should be used in the Walk-In Coolers and Freezers
Proposed Rule, as well as any other rulemaking and policymaking until the SCC undergoes a
more rigorous notice, review and comment process. 3
II. THE WALK-IN COOLERS AND FREEZERS PROPOSED RULE IS FLAWED IN
OTHER SIGNIFICANT WAYS
The problems with the Walk-In Coolers and Freezers Proposed Rule go beyond the
transparency and process problems associated with the SCC, which the DOE relied upon in its
cost-benefits analysis. 4 As outlined below, there are other errors and omissions in the DOE's
cost-benefit analysis that must be addressed.
2
The SCC Estimates also fail to comply with the OMB Bulletin for Agency Good Guidance Practices, which
requires pre-adoption public notice and comment for economically significant guidance documents. See OMB
Bulletin, 72 F.R. at 3440 (Sec. IV).
3 Notably, on November 1, 2013, the Administrator of the Office of Management and Budget announced that OMB
was "issuing updated values for the Social Cost of Carbon" and that OMB's OIRA would "provide a new
opportunity for public comment on the [SCC] estimates." See Howard Shelanski, Adm'r, Office of Information and
Regulatory Affairs, Office of Management and Budget, "Refining Estimates of the Social Cost of Carbon," Nov. 1,
2013, available at http://www.whitehouse.gov/blog/20 1 3/1 1 /01 /refining-esti mates-social-cost-carbon.
In using the SCC Estimates, the DOE also fails to adhere to its own guidelines for ensuring and maximizing the
quality, objectivity, utility, and integrity of information disseminated by the DOE. For example, at the direction of
OMB and pursuant to Section 515 of the Treasury and General Government Appropriations Act for Fiscal Year
2001, the DOE implemented guidelines on October 1, 2002 aimed at ensuring the quality of information that it
disseminates. The OMB guidelines sought to "provide policy and procedural guidance to Federal Agencies for
ensuring and maximizing the quality, objectivity, utility, and integrity of information (including statistical
information) disseminated by Federal Agencies." See also DOE Final Report Implementing OMB Information
'
U.S. Department of Energy
November 12, 2013
Page 6 of 10
A. The DOE Improperly Balances Costs and Benefits
An important principle of cost-benefit analysis is that costs and benefits must be
compared over the same time frame and within the same scope. The DOE's cost-benefit analysis
for the Proposed Rule violates this principle in both ways. With respect to the time frame, the
DOE calculates the present value of the costs of the Proposed Rule to consumers and
manufacturers over a 30 year period. In contrast, the annual SCC estimates used in the DOE's
cost-benefit analysis reflect cumulative benefits that will accrue to individuals up to 300 years in
the future. According to the DOE analysis, the proposed walk-in coolers and freezers energy
efficiency standards would lead to the removal of 298 million metric tons of greenhouse gases
(GHGs) over 30 years (2017-2046), 5 which in turn would avoid global warming damages and
result in benefits over the next 300 years. The DOE's comparison of 30 years of cost to 300
hundred years of putative, future benefits is inconsistent and improper. 6
-
The DOE cost-benefit analysis also fails to balance the scope of costs and benefits.
Regulations result in both direct and indirect costs and benefits. Direct costs are the costs of
compliance immediately imposed by the regulation on the regulated entities, i.e. higher
manufacturing and consumer costs for walk-in coolers and freezers. Direct benefits include any
economic or health costs that would accrue immediately to the public from the greater efficiency
of walk-in coolers and freezers, i.e. lower operating costs from improved energy efficiency of the
walk-in coolers and freezers. Indirect costs would include the secondary and tertiary effects that
would be induced by the regulation. Here, the DOE expands its benefits calculations to include
indirect benefits without commensurately expanding its cost analysis to include potential indirect
costs. For example, the DOE analysis includes benefits based upon the SCC, which would be
classified as indirect benefits; however, it does not take into account indirect costs, such as
income loss and job search costs imposed on workers who might be displaced because of higher
prices for new walk-in coolers and freezers and their components, and reduced product demand.
The DOE analysis for the Proposed Rule should be balanced in the scope and time frame of any
costs and benefits.
B. The DOE Ignores Global Benefits Reduction Rate
In justifying the benefits of the Walk-In Coolers and Freezers Proposed Rule, the DOE
ignores a recommendation from the IWG to reduce global benefits from the SCC in the context
of wholly domestic regulations. In relying upon the SCC in its cost-benefit analysis, the DOE
recognizes the benefits of reduced global CO2 emissions resulting from the Proposed Rule, but it
Dissemination Quality Guidelines, at 26-27 (2002) ("[w]here feasible, data should have full, accurate, transparent
documentation, and possible sources of error affecting data quality should be identified and disclosed to users....").
5 See 78 F.R. No. 176, p. 55786.
6 Although the DOE presents carbon reduction values for each year of the 30-year period over which costs and
benefits are compared, the presentation is not complete. For example, the amount of $43 per ton that is listed in
Table V.9 of the Proposed Rule (see page 55975) is not the actual direct benefit in terms of the economic damages
avoided by eliminating a ton of carbon in the year 2020. Instead, it is the present value in 2020 (calculated at a 3%
discount rate) of the projected economic damage avoided over the future 300 years by all person in all places
worldwide if the subject ton of carbon eliminated in 2020 remains out of the emissions stream for all time forward.
U.S. Department of Energy
November 12, 2013
Page 7 of 10
fails to consider the correlating global costs of the Proposed Rule. According to the IWG, the
U.S. domestic SCC benefits for the reduction of one ton of CO2 amount to only 7% - 23% of the
global SCC benefits for the reduction of the same ton of CO2. In its proposed Energy
Conservation Standards for Metal Halide Lamp Fixtures, the DOE acknowledged the IWG's
recommendation, by stating that "...the interagency group determined that a range of values
from 7 percent to 23 percent should be used to adjust the global SCC to calculate domestic
effects, although DOE will give preference to consideration of the global benefits of reducing
CO2 emissions."7
By ignoring the IWG's recommendation and failing to apply a reduction rate for the
global SCC benefits, the DOE's estimate of carbon-reduction benefits for the Proposed Rule is 4
to 14 times greater than it would have been if the DOE had followed the IWG's
recommendation. In other words, the domestic SCC benefits for the Proposed Rule would be
only $630 million to $2.1 billion under the IWG's recommendation, instead of the $9.0 billion
claimed by the DOE based on the global benefits perspective. The DOE's decision to depart
from the IWG approach on this issue, with no justification or reasoning, renders it arbitrary.
C. The DOE Fails to Consider EPA's Planned GHG Regulations on Power Plants
The DOE also fails to take into account in the Proposed Rule the Environmental
Protection Agency ("EPA")'s planned GHG regulations for new and existing power plants. In
the President's highly publicized Climate Action Plan, which was released on June 25, 2013, the
White House directed EPA to propose and issue regulations reducing GHG emissions from new
and existing power plants. The Climate Action Plan and accompanying Presidential
memorandum outlined detailed rulemaking schedules for both new and existing power plants.
Specifically, EPA would propose regulations for GHG emissions from new power plants by
September 20, 2013, and similar regulations for existing power plants by June 1, 2014. EPA
already fulfilled the first directive by releasing a proposed regulation for GHG emissions from
new power plants on September 20, 2013. Given the Agency's simultaneous announcement that
it would undertake a two-month outreach to stakeholders on the existing power plant rule, all
indications are that EPA will fulfill the second directive, too.
Despite all of the publicity and media coverage surrounding the Climate Action Plan and
EPA's strong suggestions that it would adhere to the President's directives, the DOE fails to
consider the impact of EPA's planned GHG power plant regulations on the Walk-In Coolers and
Freezers Proposed Rule. This is significant because EPA's planned GHG regulations for new
and existing power plants likely will materially affect the projections of CO 2 emissions
reductions on which the DOE's SCC-derived benefit calculations are based. The DOE's
projections of baseline CO2 emissions over the 2017-2046 timeframe assume the continuation of
existing patterns of electricity generation by fuel types. It is well known, however, that EPA's
planned GHG regulations on power plants, as well as other existing and proposed regulations,
such as the Mercury and Air Toxics Standards, can be reasonably expected to change the
See htt•://wwwl .eere.ener
0v/buildings/appliance standards/commercial/pdfs/mhlf preanalysis chapter2.pdf,
Preliminary Technical Support Document, p. 64 for Docket No. EERE-2009-BT-STD-0018, Energy Conservation
Standards for Metal Halide Lamp Fixtures, 78 F.R. 161 (Aug. 20, 2013).
7
U.S. Department of Energy
November 12, 2013
Page 8 of 10
baseline pattern of energy generation, including the types of fuels used for electricity generation
and the extent to which they are used. Consequently, in failing to consider EPA's planned GHG
regulations on power plants, the DOE's projections of CO 2 emissions reductions from the WalkIn Coolers and Freezers Proposed Rule are likely invalid.
This example highlights a significant problem with the application of SCC-derived
benefit calculations by regulatory agencies. When different agencies are simultaneously
pursuing regulatory agendas that address similar sources of CO2 emissions, the likelihood of
double-counting of the same putative SCC benefits is high. The result may be to promote
excessive and economically unjustified regulations because the actual benefits have been
overestimated by duplicative emissions reduction claims. Both the DOE and EPA should not
take credit for removing the same ton of CO2, and neither agency should claim benefits from the
removal of that same ton in more than one of its own regulations. Indeed, the potential effects of
EPA's planned GHG regulations on power plants very well may overwhelm any emissions
reduction claims that the DOE may project for its energy efficiency standards. Consequently, the
DOE's failure to consider EPA's planned GHG regulations on power plants, as well as other
related emissions-reducing regulations, 8 is a significant flaw in its analysis for the Proposed
Rule.
D. The DOE Incorrectly Used the IWG's Analysis Showing Increasing SCC
Estimate Values for Future Years
In its cost-benefit analysis for the Walk-In Coolers and Freezers Proposed Rule, the DOE
incorrectly used information from the 2013 IWG Technical Support Document, which lead to an
overestimation of the claimed benefits from the Proposed Rule. Essentially, the DOE used the
IWG's increasing SCC estimates over time to calculate benefits from the Proposed Rule, but
failed to account for the fact that those increasing values do not anticipate any policy changes or
changing emissions trends over time.
In its analysis, the DOE calculated benefits for the reduction of CO2 emissions for each
year from 2017 through 2046. Although the Proposed Rule's projected emissions reduction is
the same for each year — about 9.9 million tons — the benefit amounts rise significantly over the
same period because the DOE multiplies each year's emissions reduction amount by a higher
SCC estimate. Those SCC estimates are derived from the IWG's 2013 table of SCC estimates
8
On June 7, 2013, EPA published a proposed regulation establishing new and more stringent effluent limitation
guidelines and standards for the steam electric generating point source category. 78 FR No. 110, p. 34432. EPA
claimed that the new guidelines would create SCC-derived benefits of $127.6 million per year (at a 3% discount
rate) because the guidelines would lead to the closure of coal-fueled steam plants and the substitution of electric
generating sources that produce less CO 2 emissions. EPA claimed that the new guidelines would result in $3.8
billion in cumulative CO2 emissions reduction benefits over a 30-year period. 78 FR No. 110, p. 34516-17. Given
that there very well may be overlapping emissions reduction benefits between the EPA effluent limitation guidelines
and the DOE's Walk-In Coolers and Freezers Proposed Rule, DOE should be considering the EPA guidelines in its
analysis of the Proposed Rule.
U.S. Department of Energy
November 12, 2013
Page 9 of 10
from 2010-2050. 9 For example, the SCC estimate for a ton of CO2 added to or removed from the
global atmospheric inventory is $33 in 2010, $43 in 2020, and $66 in 2045. 10
The DOE's use of the increasing SCC estimates from the IWG's 2013 table is
problematic because those estimates only describe the SCC estimate for a one-time change in the
specified year, without accounting for any prior changes from the baseline emissions trend in
earlier years. For example, the 2045 value of $66 estimates the value of a ton of CO2 if there
have been no policy changes or new regulations impacting carbon emissions until 2045. 11 The
value of $66, therefore, is not the correct value to apply for 2045 if the emissions reduction for
that year has been preceded by emissions reductions in each prior year since 2017, as is the case
with the policy intervention that is proposed by the DOE Walk-In Coolers and Freezers Proposed
Rule.
The result of continuing the standard in effect in 2045, after having already been in effect
since 2017, would translate into a smaller benefit estimate per ton than $66. The benefit would
be smaller because the alleged future damage of a ton of CO2 in 2045 would have been reduced
by the effect of the tons removed in prior years. Consequently, the DOE's misuse of the SCC
estimates from the IWG's 2013 table in its cost-benefit analysis resulted in the Department
overestimating the claimed benefits from the Walk-In Coolers and Freezers Proposed Rule. 12
E.
The DOE Changed Its Rulemaking Framework for the Proposed Rule
Without Allowing for Notice and Comment
Prior to publishing a proposed energy efficiency standard, the DOE's long-standing
practice has been to prepare and publish for public review and comment a "Rulemaking
Framework" document that describes the salient features of the DOE's planned approach to
analyzing the costs, benefits and other impacts of the proposed standard. With respect to the
Proposed Rule, the DOE published "Rulemaking Framework for Walk-In Coolers and Walk-In
Freezers (RIN: 1904-AB86) on December 22, 2008. 13 In that "Rulemaking Framework," DOE
stated as follows:
Nothing in EPCA, nor in the National Environmental Policy Act, requires
that the economic value of emissions reductions be incorporated in the net
present value analysis of the value of energy savings. Unlike energy
See "Revised Social Cost of CO 2, 2010 — 2050" Table from the Interagency Working Group's "Technical Support
Document: Technical Update of the Social Cost of Carbon for Regulatory Impact Analysis Under Executive Order
12866," May 2013, p. 3.
10 Id. These estimates are based upon a 3% discount rate.
11 See IWG, "Technical Support Document: Social Cost of Carbon for Regulatory Impact Analysis Under Executive
Order 12866," February 2010, pp. 24-25, for a detailed description of the calculation procedure.
12 To address this problem, the IWG could re-run the integrated assessment models which underlie the SCC
calculations to reflect policies that begin in various years to remove a ton of CO 2 from the baseline and continue to
do so annually thereafter. This corrected procedure likely would yield a table that could be used to analyze the
benefits of policies that begin at certain years and continue steadily thereafter.
" The document is included in the current docket at http://www.regulations.gov/MdocumentDetail;D=EERE-2008BT-STD-0015-0008.
9
U.S. Department of Energy
November 12, 2013
Page 10 of 10
savings, the economic value of emissions reductions is not priced in the
marketplace. For those emissions for which real national emission
reductions are anticipated (CO2, Hg, and NOX), only a range of estimated
economic values based on environmental damage studies of varying
quality and applicability is available. Consequently, DOE is reporting and
weighing these values separately and is not including them in the NPV
analysis. 1 4
The DOE now has changed its rulemaking approach for the Proposed Rule by
incorporating SCC estimates for emissions reductions in its benefits analysis. The Department
took those actions without prior notice or opportunity for public comment. As it has done in the
past, the DOE should have published a revised "Rulemaking Framework" for public comment
before altering its analytical approach in the Proposed Rule, i.e. incorporating SCC estimates.
Instead, the DOE published a Proposed Rule lacking in transparency and public input.
III. CONCLUSION
For the reasons stated above, including the incorporation of the arguments posited in the
attached Petition, the DOE should withdraw the SCC calculation as a basis for the Walk-In
Coolers and Freezers Proposed Rule, and refrain from using the SCC in any other rulemaking or
policymaking until the SCC undergoes a more rigorous notice, review and comment process.
Additionally, the DOE should address and correct the errors outlined in Section II of these
comments.
Thank you for the opportunity to participate in this proceeding. If you have any follow up
questions regarding these comments, please feel free to reach out to William L. Kovacs, Senior
Vice President of Environment, Technology & Regulatory Affairs at the U.S. Chamber of
Commerce at (202) 463-5457 or by e-mail: wkovacsguschamber.com .
American Forest & Paper Association
American Fuel & Petrochemical Manufacturers
American Petroleum Institute
Council of Industrial Boiler Owners
National Association of Manufacturers
National Mining Association
Portland Cement Association
U.S. Chamber of Commerce
H /d
pp. 43-44.
energy ila
ARM
Portland Llment Association
U.S. CHAMBER OF COMMERCE
Nt
UNAL ASSOCIATION OF
Manufacturers
AFPIVI
American
Fuel & Petrochemical
Manufacturers
THE AMERICAN RESOURCE
.
American
Forest & Paper
Association
d
November 12, 2013
VIA ELECTRONIC FILING AND ELECTRONIC MAIL
U.S. Department of Energy
Office of Energy Efficiency and Renewable Energy
Building Technologies Program, EE-2J
1000 Independence Ave. SW
Washington, D.C. 20585-0121
Attn: Brenda Edwards ([email protected] )
Re: Docket No. EERE-2010-BT-STD-0003: Energy Conservation Program:
Energy Conservation Standards for Commercial Refrigeration Equipment;
U.S. Department of Energy
November 12, 2013
Page 2 of 10
Proposed Rule; Federal Register Vol. 78, Number 176 (Wednesday,
September 11, 2013); RIN 1904-AC19
Dear Sir/Madam:
The U.S. Chamber of Commerce, American Forest & Paper Association, American Fuel
& Petrochemical Manufacturers, American Petroleum Institute, Council of Industrial Boiler
Owners, National Association of Manufacturers, National Mining Association, and Portland
Cement Association (collectively, the "Associations") offer these comments on the Department
of Energy ("DOE")'s proposed rule for Energy Conservation Standards for Commercial
Refrigeration Equipment, 78 F.R. 55890 (September 11, 2013) ("Commercial Refrigeration
Equipment Proposed Rule" or "Proposed Rule"). For the reasons discussed below, the
Associations believe that the "social cost of carbon" ("SCC") should be withdrawn as a basis for
the Commercial Refrigeration Equipment Proposed Rule, and that the SCC calculation should
not be used in any rulemaking and/or policymaking until it undergoes a more rigorous notice,
review and comment process as outlined below.
The U.S. Chamber of Commerce ("Chamber") is the world's largest business
federation, representing the interests of more than three million businesses and organizations of
all sizes, sectors, and regions, as well as state and local chambers and industry associations, and
dedicated to promoting, protecting, and defending America's free enterprise system.
The American Forest & Paper Association ("AF&PA") serves to advance a sustainable
U.S. pulp, paper, packaging, and wood products manufacturing industry through fact-based
public policy and marketplace advocacy. AF&PA member companies make products essential
for everyday life from renewable and recyclable resources and are committed to continuous
improvement through the industry's sustainability initiative - Better Practices, Better Planet
2020. The forest products industry accounts for approximately 4.5 percent of the total U.S.
manufacturing GDP, manufactures approximately $200 billion in products annually, and
employs nearly 900,000 men and women. The industry meets a payroll of approximately $50
billion annually and is among the top 10 manufacturing sector employers in 47 states. AF&PA's
sustainability initiative - Better Practices, Better Planet 2020 - is the latest example of our
members' proactive commitment to the long-term success of our industry, our communities and
our environment. We have long been responsible stewards of our planet's resources. Our
member companies have collectively made significant progress in each of the following goals,
which comprise one of the most extensive quantifiable sets of sustainability goals for a U.S.
manufacturing industry: increasing paper recovery for recycling; improving energy efficiency;
reducing greenhouse gas emissions; promoting sustainable forestry practices; improving
workplace safety; and reducing water use.
The American Fuel & Petrochemical Manufacturers ("AFPM") is a trade association
representing high-tech American manufacturers of virtually the entire U.S. supply of gasoline,
diesel, jet fuel, other fuels and home heating oil, as well as the petrochemicals used as building
blocks for thousands of vital products in daily life. AFPM members make modern life possible
U.S. Department of Energy
November 12, 2013
Page 3 of 10
and keep America moving and growing as they meet the needs of our nation and local
communities, strengthen economic and national security, and support 2 million American jobs.
The American Petroleum Institute ("API") is a national trade association representing
over 500 member companies involved in all aspects of the oil and natural gas industry. API's
members include producers, refiners, suppliers, pipeline operators, and marine transporters, as
well as service and supply companies that support all segments of the industry. API and its
members are dedicated to meeting environmental requirements, while economically developing
and supplying energy resources for consumers.
The Council of Industrial Boiler Owners ("CIBO") is a broad-based association of
industrial boiler owners, architect-engineers, related equipment manufacturers, and University
affiliates with members representing 20 major industrial sectors. CIBO members have facilities
in every region of the country and a representative distribution of almost every type of industrial,
commercial and institutional (ICI) boiler and fuel combination currently in operation. CIBO was
formed in 1978 to promote the exchange of information within the industry and between industry
and government relating to energy and environmental equipment, technology, operations,
policies, law and regulations affecting industrial boilers. Since its formation, CIBO has been
active in the development of technically sound, reasonable, cost-effective energy and
environmental regulations for industrial boilers. CIBO supports regulatory programs that provide
industry with enough flexibility to modernize -- effectively and without penalty - the nation's
aging energy infrastructure, as modernization is the key to cost-effective environmental
protection.
The National Association of Manufacturers ("NAM") is the largest industrial trade
association in the U.S., representing over 12,000 small, medium and large manufacturers in all
50 states. NAM is the leading voice in Washington, D.C., for the manufacturing economy, which
provides millions of high wage jobs in the U.S. and generates more than $1.6 trillion in GDP. In
addition, two-thirds of NAM members are small businesses, which serve as the engine for job
growth. NAM's mission is to enhance the competitiveness of manufacturers and improve
American living standards by shaping a legislative and regulatory environment conducive to U.S.
economic growth.
The National Mining Association ("NMA") is a national trade association whose
members produce most of America's coal, metals, and industrial and agricultural minerals. Its
membership also includes manufacturers of mining and mineral processing machinery and
supplies, transporters, financial and engineering firms, and other businesses involved in the
nation's mining industries. NMA works with Congress and federal and state regulatory officials
to provide information and analyses on public policies of concern to its membership, and to
promote policies and practices that foster the efficient and environmentally sound development
and use of the country's mineral resources.
The Portland Cement Association ("PCA") represents 26 U.S. cement companies
operating 79 manufacturing plants in 34 states. Accounting for 78% of domestic cement-making
U.S. Department of Energy
November 12, 2013
Page 4 of 10
capacity, PCA members operate distribution centers in all 50 states and nearly every
congressional district.
These Associations' members have a strong interest in this Proposed Rule because they
may be impacted by the SCC precedent set in the rulemaking given that many of them
manufacture products that, when combusted, result in greenhouse gas ("GHG") emissions
(including carbon dioxide ("CO2")), and because, in the course of their business, they emit CO2.
Should this Administration, or any subsequent one, promulgate further regulation of these
products or emissions, such proposals and rules could potentially be based, in large part, on
either the 2010 or 2013 estimates of the SCC ("SCC Estimates") created by the federal
Interagency Working Group ("IWG"). Therefore, the Associations' members have a direct and
meaningful interest in ensuring that any estimates and applications of the SCC are based on
transparent processes, accurate information, rational assumptions, and are within the reach of the
current scientific understanding and impact models.'
I. THE SCC SHOULD UNDERGO A NOTICE AND COMMENT PROCESS
BEFORE IT IS USED IN THE COMMERCIAL REFRIGERATION
EQUIPMENT PROPOSED RULE OR ANY OTHER RULEMAKINGS.
The IWG has defined the SCC as "an estimate of the monetized damages associated with
an incremental increase in carbon emissions in a given year." In the Commercial Refrigeration
Equipment Proposed Rule, the DOE uses benefits derived from the SCC to justify the proposed
energy efficiency regulation. The DOE estimates that the Proposed Rule will have cumulative
benefits of $7.6 billion over a 30-year period (2017-2046) at a 3% discount rate. Of that $7.6
billion in benefits, $6.0 billion is from lower consumer electricity costs, $1.5 billion is from the
SCC, and $50 million is from reduced NO emissions. The DOE also estimates that the
Proposed Rule will have cumulative costs of $1.9 billion. Notably, under Title 111 of the Energy
Policy and Conservation Act (42 U.S.C. 6295), the DOE's findings with regard to the benefits of
the Commercial Refrigeration Equipment Proposed Rule are legally sufficient for justifying the
rule without the inclusion of any benefits based upon the SCC.
While the DOE may include SCC benefits in claiming that the Commercial Refrigeration
Equipment Proposed Rule has benefits of $7.6 billion, that does not change the fact that the
IWG's SCC analysis has not been adequately noticed and reviewed before being used in this
Proposed Rule or any other rulemaking. As described in the attached Petition for Correction
pursuant to the Information Quality Act, the Associations believe that the 2010 and 2013
Technical Support Documents and SCC Estimates should be withdrawn and not used in any
rulemaking and policymaking, including the Commercial Refrigeration Equipment Proposed
Rule, for the following reasons:
To be clear, we question the application of the 2010 and 2013 Interagency Working Group ("IWG") estimates of
the SCC, which are based on complex economic impacts hundreds of years in the future, which in turn are based on
present day understanding of current and future carbon emissions. We are not herein discussing the existence or
potential causes of climate change.
U.S. Department of Energy
November 12, 2013
Page 5 of 10
1. The SCC Estimates fail in terms of process and transparency. The SCC Estimates fail to
comply with OMB guidance for developing influential policy-relevant information under
the Information Quality Act. 2 The SCC Estimates are the product of an opaque process
and any pretensions to their supposed accuracy (and therefore usefulness in policymaking) are unsupportable.
2. The models with inputs (hereafter referred to as "the modeling systems") used for the
SCC Estimates and the subsequent analyses were not subject to peer review as
appropriate.
3. Moreover, even if the SCC Estimate development process was transparent, rigorous, and
peer-reviewed, the modeling conducted in this effort does not offer a reasonably
acceptable range of accuracy for use in policymaking.
4. The IWG has failed to disclose and quantify key uncertainties to inform decision makers
and the public about the effects and uncertainties of alternative regulatory actions as
required by OMB.
5. By presenting only global SCC estimates and downplaying domestic SCC estimates in
2013, the IWG has severely limited the utility of the SCC for use in benefit-cost analysis
and policymaking.
Given all of the concerns summarized above and detailed in the attached petition, neither the
2010 nor 2013 IWG estimates of SCC should be used in the Commercial Refrigeration
Equipment Proposed Rule, as well as any other rulemaking and policymaking until the SCC
undergoes a more rigorous notice, review and comment process. 3
II. THE COMMERCIAL REFRIGERATION EQUIPMENT PROPOSED RULE IS
FLAWED IN OTHER SIGNIFICANT WAYS
The problems with the Commercial Refrigeration Equipment Proposed Rule go beyond
the transparency and process problems associated with the SCC, which the DOE relied upon in
its cost-benefits analysis. 4 As outlined below, there are other errors and omissions in the DOE's
cost-benefit analysis that must be addressed.
The SCC Estimates also fail to comply with the OMB Bulletin for Agency Good Guidance Practices, which
requires pre-adoption public notice and comment for economically significant guidance documents. See OMB
Bulletin, 72 F.R. at 3440 (Sec. IV).
3 Notably, on November 1, 2013, the Administrator of the Office of Management and Budget announced that OMB
was "issuing updated values for the Social Cost of Carbon" and that OMB's OIRA would "provide a new
opportunity for public comment on the [SCC] estimates." See Howard Shelanski, Adm'r, Office of Information and
Regulatory Affairs, Office of Management and Budget, "Refining Estimates of the Social Cost of Carbon," Nov. 1,
2013, available at littp://www.whitehouse.goviblog/20 1 3/1 1 /01 /refining-estimates-social-cost-carbon.
4 In using the SCC Estimates, the DOE also fails to adhere to its own guidelines for ensuring and maximizing the
quality, objectivity, utility, and integrity of information disseminated by the DOE. For example, at the direction of
OMB and pursuant to Section 515 of the Treasury and General Government Appropriations Act for Fiscal Year
2001, the DOE implemented guidelines on October 1, 2002 aimed at ensuring the quality of information that it
2
U.S. Department of Energy
November 12, 2013
Page 6 of 10
A. The DOE Improperly Balances Costs and Benefits
An important principle of cost-benefit analysis is that costs and benefits must be
compared over the same time frame and within the same scope. The DOE's cost-benefit analysis
for the Proposed Rule violates this principle in both ways. With respect to the time frame, the
DOE calculates the present value of the costs of the Proposed Rule to consumers and
manufacturers over a 30-year period. In contrast, the annual SCC estimates used in the DOE's
cost-benefit analysis reflect cumulative benefits that will accrue to individuals up to 300 years in
the future. According to the DOE analysis, the Commercial Refrigeration Equipment energy
efficiency standards would lead to the removal of 54.88 million metric tons of greenhouse gases
(GHGs) over 30 years (2017-2046), 5 which in turn would avoid global warming damages and
result in benefits over the next 300 years. The DOE's comparison of 30 years of cost to 300
hundred years of putative, future benefits is inconsistent and improper. 6
The DOE cost-benefit analysis also fails to balance the scope of costs and benefits.
Regulations result in both direct and indirect costs and benefits. Direct costs are the costs of
compliance immediately imposed by the regulation on the regulated entities, i.e. higher
manufacturing and consumer costs for commercial refrigeration equipment. Direct benefits
include any economic or health costs that would accrue immediately to the public from the
greater efficiency of commercial refrigeration equipment, i.e. lower operating costs from
improved energy efficiency of the commercial refrigeration equipment. Indirect costs would
include the secondary and tertiary effects that would be induced by the regulation. Here, the
DOE expands its benefits calculations to include indirect benefits without commensurately
expanding its cost analysis to include potential indirect costs. For example, the DOE analysis
includes benefits based upon the SCC, which would be classified as indirect benefits; however, it
does not take into account indirect costs, such as income loss and job search costs imposed on
workers who might be displaced because of higher prices for new commercial refrigeration
equipment and their components, and reduced product demand. The DOE analysis for the
Proposed Rule should be balanced in the scope and time frame of any costs and benefits.
disseminates. The OMB guidelines sought to "provide policy and procedural guidance to Federal Agencies for
ensuring and maximizing the quality, objectivity, utility, and integrity of information (including statistical
information) disseminated by Federal Agencies." See also DOE Final Report Implementing OMB Information
Dissemination Quality Guidelines, at 26-27 (2002) ("[w]here feasible, data should have full, accurate, transparent
documentation, and possible sources of error affecting data quality should be identified and disclosed to users....").
5 See 78 F.R. No. 176, p. 55893.
6 Although the DOE presents carbon reduction values for each year of the 30-year period over which costs and
benefits are compared, the presentation is not complete. For example, the amount of $43 per ton that is listed in
Table IV-18 of the Proposed Rule (see page 55845) is not the actual direct benefit in terms of the economic damages
avoided by eliminating a ton of carbon in the year 2020. Instead, it is the present value in 2020 (calculated at a 3%
discount rate) of the projected economic damage avoided over the future 300 years by all person in all places
worldwide if the subject ton of carbon eliminated in 2020 remains out of the emissions stream for all time forward.
U.S. Department of Energy
November 12, 2013
Page 7 of 10
B. The DOE Ignores Global Benefits Reduction Rate
In justifying the benefits of the Commercial Refrigeration Equipment Proposed Rule, the
DOE ignores a recommendation from the IWG to reduce global benefits from the SCC in the
context of wholly domestic regulations. In relying upon the SCC in its cost-benefit analysis, the
DOE recognizes the benefits of reduced global CO2 emissions resulting from the Proposed Rule,
but it fails to consider the correlating global costs of the Proposed Rule. According to the IWG,
the U.S. domestic SCC benefits for the reduction of one ton of CO2 amount to only 7% - 23% of
the global SCC benefits for the reduction of the same ton of CO2. In its proposed Energy
Conservation Standards for Metal Halide Lamp Fixtures, the DOE acknowledged the IWG's
recommendation, by stating that "...the interagency group determined that a range of values
from 7 percent to 23 percent should be used to adjust the global SCC to calculate domestic
effects, although DOE will give preference to consideration of the global benefits of reducing
CO2 emissions."'
By ignoring the IWG's recommendation and failing to apply a reduction rate for the
global SCC benefits, the DOE's estimate of carbon-reduction benefits for the Proposed Rule is 4
to 14 times greater than it would have been if the DOE had followed the IWG's
recommendation. In other words, the domestic SCC benefits for the Proposed Rule would be
only $105 million to $246 million under the IWG's recommendation, instead of the $1.5 billion
claimed by the DOE based on the global benefits perspective. The DOE's decision to depart
from the IWG approach on this issue, with no justification or reasoning, renders it arbitrary.
C. The DOE Fails to Consider EPA's Planned GHG Regulations on Power Plants
The DOE also fails to take into account in the Proposed Rule the Environmental
Protection Agency ("EPA")'s planned GHG regulations for new and existing power plants. In
the President's highly publicized Climate Action Plan, which was released on June 25, 2013, the
White House directed EPA to propose and issue regulations reducing GHG emissions from new
and existing power plants. The Climate Action Plan and accompanying Presidential
memorandum outlined detailed rulemaking schedules for both new and existing power plants.
Specifically, EPA would propose regulations for GHG emissions from new power plants by
September 20, 2013, and similar regulations for existing power plants by June 1, 2014. EPA
already fulfilled the first directive by releasing a proposed regulation for GHG emissions from
new power plants on September 20, 2013. Given the Agency's simultaneous announcement that
it would undertake a two-month outreach to stakeholders on the existing power plant rule, all
indications are that EPA will fulfill the second directive, too.
Despite all of the publicity and media coverage surrounding the Climate Action Plan and
EPA's strong suggestions that it would adhere to the President's directives, the DOE fails to
consider the impact of EPA's planned GHG power plant regulations on the Commercial
Refrigeration Equipment Proposed Rule. This is significant because EPA's planned GHG
7 See http://wwwl xere.energy.goy/buildings/appliance standards/commercial/pdfs/mhlf preanalysis chapter2.pdf,
Preliminary Technical Support Document, p. 64 for Docket No. EERE-2009-BT-STD-0018, Energy Conservation
Standards for Metal Halide Lamp Fixtures, 78 F.R. 161 (Aug. 20, 2013).
U.S. Department of Energy
November 12, 2013
Page 8 of 10
regulations for new and existing power plants likely will materially affect the projections of CO 2
emissions reductions on which the DOE's SCC-derived benefit calculations are based. The
DOE's projections of baseline CO 2 emissions over the 2017-2046 timeframe assume the
continuation of existing patterns of electricity generation by fuel types. It is well known,
however, that EPA's planned GHG regulations on power plants, as well as other existing and
proposed regulations, such as the Mercury and Air Toxics Standards, can be reasonably expected
to change the baseline pattern of energy generation, including the types of fuels used for
electricity generation and the extent to which they are used. Consequently, in failing to consider
EPA's planned GHG regulations on power plants, the DOE's projections of CO2 emissions
reductions from the Commercial Refrigeration Equipment Proposed Rule are likely invalid.
This example highlights a significant problem with the application of SCC-derived
benefit calculations by regulatory agencies. When different agencies are simultaneously
pursuing regulatory agendas that address similar sources of CO2 emissions, the likelihood of
double-counting of the same putative SCC benefits is high. The result may be to promote
excessive and economically unjustified regulations because the actual benefits have been
overestimated by duplicative emissions reduction claims. Both the DOE and EPA should not
take credit for removing the same ton of CO2, and neither agency should claim benefits from the
removal of that same ton in more than one of its own regulations. Indeed, the potential effects of
EPA's planned GHG regulations on power plants very well may overwhelm any emissions
reduction claims that DOE may project for its energy efficiency standards. Consequently, the
DOE's failure to consider EPA's planned GHG regulations on power plants, as well as other
related emissions-reducing regulations, 8 is a significant flaw in its analysis for the Proposed
Rule.
D. The DOE Incorrectly Used the IWG's Analysis Showing Increasing SCC
Estimate Values for Future Years
In its cost-benefit analysis for the Commercial Refrigeration Equipment Proposed Rule,
the DOE incorrectly used information from the 2013 IWG Technical Support Document, which
lead to an overestimation of the claimed benefits from the Proposed Rule. Essentially, DOE
used the IWG's increasing SCC estimates over time to calculate benefits from the Proposed
Rule, but failed to account for the fact that those increasing values do not anticipate any policy
changes or changing emissions trends over time.
8
On June 7, 2013, EPA published a proposed regulation establishing new and more stringent effluent limitation
guidelines and standards for the steam electric generating point source category. 78 FR No. 110, p. 34432. EPA
claimed that the new guidelines would create SCC-derived benefits of $127.6 million per year (at a 3% discount
rate) because the guidelines would lead to the closure of coal-fueled steam plants and the substitution of electric
generating sources that produce less CO 2 emissions. EPA claimed that the new guidelines would result in $3.8
billion in cumulative CO2 emissions reduction benefits over a 30-year period. 78 FR No. 110, p. 34516-17. Given
that there very well may be overlapping emissions reduction benefits between the EPA effluent limitation guidelines
and the DOE's Commercial Refrigeration Equipment Proposed Rule, DOE should be considering the EPA
guidelines in its analysis of the Proposed Rule.
U.S. Department of Energy
November 12, 2013
Page 9 of 10
In its analysis, the DOE calculated benefits for the reduction of CO 2 emissions for each
year from 2017 through 2046. Although the Proposed Rule's projected emissions reduction is
the same for each year — about 9.9 million tons — the benefit amounts rise significantly over the
same period because the DOE multiplies each year's emissions reduction amount by a higher
SCC estimate. Those SCC estimates are derived from the IWG's 2013 table of SCC estimates
from 2010-2050. 9 For example, the SCC estimate for a ton of CO2 added to or removed from the
global atmospheric inventory is $33 in 2010, $43 in 2020, and $66 in 2045. 10
The DOE's use of the increasing SCC estimates from the IWG's 2013 table is
problematic because those values only describe the SCC estimate for a one-time change in the
specified year, without accounting for any prior changes from the baseline emissions trend in
earlier years. For example, the 2045 value of $66 estimates the value of a ton of CO2 if there
have been no policy changes or new regulations impacting carbon emissions until 2045. 11 The
value of $66, therefore, is not the correct value to apply for 2045 if the emissions reduction for
that year has been preceded by emissions reductions in each prior year since 2017, as is the case
with the policy intervention that is proposed by the DOE Commercial Refrigeration Equipment
Proposed Rule.
The result of continuing the standard in effect in 2045, after having already been in effect
since 2017, would translate into a smaller benefit estimate per ton than $66. The benefit would
be smaller because the alleged future damage of a ton of CO 2 in 2045 would have been reduced
by the effect of the tons removed in prior years. Consequently, the DOE's misuse of the SCC
estimates from the IWG's 2013 table in its cost-benefit analysis resulted in the Department
overestimating the claimed benefits from the Commercial Refrigeration Equipment Proposed
Rule. 12
III. CONCLUSION
For the reasons stated above, including the incorporation of the arguments posited in the
attached Petition, the DOE should withdraw the SCC calculation as a basis for the Commercial
Refrigeration Equipment Proposed Rule, and refrain from using the SCC in any other rulemaking
or policymaking until the SCC undergoes a more rigorous notice, review and comment process.
Additionally, the DOE should address and correct the errors outlined in Section II of these
comments.
See "Revised Social Cost of CO 2 , 2010 — 2050" Table from the Interagency Working Group's "Technical Support
Document: Technical Update of the Social Cost of Carbon for Regulatory Impact Analysis Under Executive Order
12866," May 2013, p. 3.
10 Id. These estimates are based upon a 3% discount rate.
11 See IWG, "Technical Support Document: Social Cost of Carbon for Regulatory Impact Analysis Under Executive
Order 12866," February 2010, pp. 24-25, for a detailed description of the calculation procedure.
12 To address this problem, the IWG could re-run the integrated assessment models which underlie the SCC
calculations to reflect policies that begin in various years to remove a ton of CO 2 from the baseline and continue to
do so annually thereafter. This corrected procedure likely would yield a table that could be used to analyze the
benefits of policies that begin at certain years and continue steadily thereafter.
9
U.S. Department of Energy
November 12, 2013
Page 10 of 10
Thank you for the opportunity to participate in this proceeding. If you have any follow up
questions regarding these comments, please feel free to reach out to William L. Kovacs, Senior
Vice President of Environment, Technology & Regulatory Affairs at the U.S. Chamber of
Commerce at (202) 463-5457 or by e-mail: [email protected].
American Forest & Paper Association
American Fuel & Petrochemical Manufacturers
American Petroleum Institute
Council of Industrial Boiler Owners
National Association of Manufacturers
National Mining Association
Portland Cement Association
U.S. Chamber of Commerce
AFPM
American
Fuel & Petrochemical
Manufacturers
U.S. CHAMBER OF COMMERCE
energyiralipl
American
Forest & Paper
Association
cibe
January 23, 2014
Ms. Brenda Edwards,
U.S. Department of Energy, Building
Technologies Office, Mailstop EE-5B,
1000 Independence Avenue S.W.,
Washington, DC 20585-0121
Filed Electronically
RE: Docket ID No. EERE-2010—BT—STD-0011
Dear Ms Edwards:
The American Fuel & Petrochemical Manufacturers' ("AFPM"), the U.S. Chamber of Commerce
("the Chamber")2, the Council of Industrial Boiler Owners 3, the American Forest & Paper Association'',
1 The American Fuel & Petrochemical Manufacturers is a national trade association of more than 400 companies,
including virtually all U.S. refiners and petrochemical manufacturers. AFPM members operate 122 U.S. refineries
comprising approximately 98% of U.S. refining capacity. AFPM petrochemical members make the chemical
building blocks that go into products ranging from medical devices, cosmetics, furniture, appliances, TVs and
radios, computers, parts used in every mode of transportation, solar power panels and wind turbines. As an energy
intensive industry, AFPM members are directly affected by the government's use of the Social Cost of Carbon in
cost-benefit analyses underlying federal regulations.
The U.S. Chamber of Commerce is the world's largest business federation representing the interests of more than 3
million businesses of all sizes, sectors, and regions, as well as state and local chambers and industry associations.
The Chamber is dedicated to promoting, protecting, and defending America's free enterprise system.
2
3
The Council of Industrial Boiler Owners ("CIBO") is a broad-based association of industrial boiler owners,
architect-engineers, related equipment manufacturers, and University affiliates with members representing 20 major
industrial sectors. CIBO members have facilities in every region of the country and a representative distribution of
almost every type of industrial, commercial and institutional (ICI) boiler and fuel combination currently in
and the American Petroleum Institute s (collectively, "the Associations") submit these comments
responding to the Department of Energy's ("DOE") Notice of Proposed Rulemaking for Energy
Conservation Standards for Residential Furnace Fans. 6 The Associations object to DOE's continued use
of the Social Cost of Carbon ("SCC") in the cost benefit analysis performed in connection with this
Proposed Rule.
The Associations believe the SCC should be withdrawn as a basis for the Proposed Rule. The
SCC calculation should not be used in any rulemaking and/or policymaking until it undergoes a more
rigorous notice, review and comment process. These arguments were more fully developed in comments
filed by AFPM, the Chamber, and several other trade associations on DOE's Energy Conservation
Standards for Commercial Refrigeration Equipment' and are incorporated by reference herein. 8
Thank you for your consideration of this important matter. If you have any further
questions, please feel free to reach out to David Friedman, Vice President of Regulatory Affairs
at American Fuel & Petrochemical Manufacturers at (202) 552-8461 or by e-mail at
[email protected] .
Respectfully submitted,
American Forest & Paper Association
American Fuel & Petrochemical Manufacturers
American Petroleum Institute
Council of Industrial Boiler Owners
U.S. Chamber of Commerce
operation. CIBO was formed in 1978 to promote the exchange of information within the industry and between
industry and government relating to energy and environmental equipment, technology, operations, policies, law and
regulations affecting industrial boilers. Since its formation, CIBO has been active in the development of technically
sound, reasonable, cost-effective energy and environmental regulations for industrial boilers. CIBO supports
regulatory programs that provide industry with enough flexibility to modernize -- effectively and without penalty the nation's aging energy infrastructure, as modernization is the key to cost-effective environmental protection.
4
American Forest & Paper Association (AF&PA) is the national trade association of the forest products industry,
representing pulp, paper, packaging and wood products manufacturers, and forest landowners. AF&PA serves to
advance a sustainable U.S. pulp, paper, packaging, and wood products manufacturing industry through fact-based
public policy and marketplace advocacy.
The American Petroleum Institute ("API") is a national trade association representing over 500 member companies
involved in all aspects of the oil and natural gas industry. API's members include producers, refiners, suppliers,
pipeline operators, and marine transporters, as well as service and supply companies that support all segments of the
industry. API and its members are dedicated to meeting environmental requirements, while economically
developing and supplying energy resources for consumers.
5
6
7
8
78 Federal Register 64067 (October 25, 2013) (the "Proposed Rule").
78 Federal Register 55890 (September 11, 2013).
See Comments of the U.S. Chamber of Commerce, American Forest & Paper Association, American Fuel &
Petrochemical Manufacturers, American Petroleum Institute, Council of Industrial Boiler Owners, National
Association of Manufacturers, National Mining Association, and Portland Cement Association; Docket No. EERE2010-BT-STD-0003-0079; http://www.regulations.gov/MdocumentDetail;D=EERE-2010-BT-STD-0003-0079
A merican'
Chemistry
Council
API®
American
Forest & Paper
Association
cibe
00-•"%tak4
NAHR
National Association of Home Builders
-04
NMA.
THE AMERICAN RESOURCE
PCA-tk
Portland Cement Association
September 16, 2013
VIA ELECTRONIC FILING
U.S. Department of Energy
1000 Independent Ave. SW
Washington, D.C. 20585-0121
Re: Docket No. EERE-2013-BT-PET-0043: Petition for Reconsideration, Energy
Conservation Program for Consumer Products: Landmark Legal Foundation;
Federal Register Volume 78, Number 159 (Friday, August 16, 2013)
Dear Sir/Madam:
The Department of Energy seeks comment on whether to undertake the reconsideration
suggested in the petition filed by the Landmark Legal Foundation (LLF). LLF has petitioned
DOE to reconsider the final rule of Energy Conservation Standards for Standby Mode and Off
Mode for Microwave Ovens, 78 FR 36316 (June 17, 2013) ("Microwave Final Rule"). LLF cites
the use of the 2013 Interagency Work Group ("IWG") Social Cost of Carbon ("SCC") in the
final rule, but not the proposal or supporting analyses, as grounds for the reconsideration and
asserts that this change in the values used in estimating the economic benefits of the Rule should
have been subject to a prior opportunity for public comment. Further, LLF asserts that without
reconsideration of the Microwave Final Rule, DOE might now rely on its prior use of the 2013
2
I
1:
ASSOC
SCC values in the Microwave Final Rule when it endeavors to enact new energy conservation
standards in the future.
The undersigned associations support the reconsideration of the Microwave Final Rule.
We do not seek to have the final standards issued in that rule changed; rather our concerns are
how that rule referenced the SCC. Our members may be impacted by the precedent set in the
Microwave Final Rule because many of them manufacture products that, when combusted, result
in greenhouse gas ("GHG") emissions (including carbon dioxide ("CO2") and because, in the
course of their business, they emit CO 2 . Should this Administration, or any subsequent one,
promulgate further regulation of these products or emissions, such proposals and rules could
potentially be based, in large part, on either the 2010 or 2013 estimates of the SCC. Therefore
our members have a direct and meaningful interest in ensuring that any estimates and
applications of the SCC are based on transparent processes, accurate information, rational
assumptions, and are within the reach of the current scientific understanding and impact models.
To be clear, we question the application of the 2010 and 2013 IWG estimates of the SCC,
which are based on complex economic impacts hundreds of years in the future, which in turn are
based on present day understanding of current and future carbon emissions. We are not herein
discussing the existence or potential causes of climate change.
As described in the attached Petition for Correction pursuant to the Information Quality
Act, we believe that both the 2010 and 2013 Technical Support Documents and SCC Estimates
should be withdrawn and not used in rule-making and policy-making for the following reasons:
1. The SCC Estimates fail in terms of process and transparency. The SCC Estimates fail to
comply with OMB guidance for developing influential policy-relevant information under
the Information Quality Act. The SCC Estimates are the product of an opaque process
and any pretensions to their supposed accuracy (and therefore usefulness in policymaking) are unsupportable.
2. The models with inputs (hereafter referred to as "the modeling systems") used for the
SCC Estimates and the subsequent analyses were not subject to peer review as
appropriate.
3. Moreover, even if the SCC Estimate development process was transparent, rigorous, and
peer-reviewed, the modeling conducted in this effort does not offer a reasonably
acceptable range of accuracy for use in policy-making.
4. The IWG has failed to disclose and quantify key uncertainties to inform decision makers
and the public about the effects and uncertainties of alternative regulatory actions as
required by OMB.
5. By presenting only global SCC estimates and downplaying domestic SCC estimates in
2013, the IWG has severely limited the utility of the SCC for use in domestic benefit-cost
analysis and policy-making.
3 I'
Given all of the concerns summarized above and detailed in the attached petition, we do not
believe that either the 2010 or 2013 IWG estimates of SCC should be used in rule-making and
policy-making at this time.
In this particular rule-making, the DOE's change of the SCC without notice and comment is
particularly egregious. Specifically:
•
The DOE's February 14, 2012, supplementary notice of proposed rule-making referenced
a $21 per ton social cost of carbon value based on the IWG 2010 document.
•
The DOE's June 17, 2013, final rule notice referenced a $33 social cost of carbon value
based on the IWG 2013 document. This value is 57% higher than the value used in the
proposed rule-making.
In conclusion, given 1) the significant issues regarding both the IWG's 2010 and 2013 SCC
summarized above and described in the attached petition and 2) the DOE's failure to provide
notice and comment on a significant change in its proposal, the undersigned associations support
the LLF petition for reconsideration of the Microwave Final Rule. Again, our support for the
petition is due to the rule's reference to the social cost of carbon, and does not reflect a position
on the final standards themselves.
The American Chemistry Council
The American Forest & Paper Association
The American Petroleum Institute
The Council of Industrial Boiler Owners
The National Association of Home Builders
The National Mining Association
The Portland Cement Association