Whitepaper on Valuing Methane Emissions Changes in
Regulatory Benefit-Cost Analysis, Peer Review Charge
Questions, and Responses

-------
Table of Contents
Whitepaper: Valuing Methane Emissions Changes in Regulatory Benefit-Cost Analysis	3
Peer Review Charge Questions	13
Review: Karen Fisher-Vanden	15
Review: John Reilly	19
Review: Steven Rose	23
2

-------
Valuing Methane Emission Changes in Regulatory Benefit-Cost Analysis
1. Introduction
While C02 is the primary source of anthropogenic greenhouse gas (GHG) emissions contributing to climate
change, other GHGs such as methane (CH4) are also important contributors.1 The Fifth Assessment Report
(AR5) of the Intergovernmental Panel on Climate Change (IPCC) concluded that in 2011 the increase in
atmospheric C02 concentration since 1750 contributed 1.82 W m"2 to global mean radiative forcing, or
64% of the total radiative forcing from well mixed GHGs, and the direct effect of increased atmospheric
CH4 accounted for 0.48 W m"2, or 17% of total radiative forcing from well mixed GHGs (Myhre et al. 2013).
In addition, CH4 emissions have indirect impacts on the climate due to their role as a precursor for
tropospheric ozone and stratospheric water vapor, both of which are potent GHGs. Accounting for these
indirect effects and the role that emissions of other substances like NOx and VOCs have on CH4
atmospheric concentrations, AR5 estimated that historical anthropogenic emissions of CH4 have
contributed a total of 0.97 W m"2 to global mean radiative forcing, or almost a third of the radiative forcing
resulting from emissions of well mixed GHGs. Therefore, CH4 emissions are having, and will continue to
have, a significant role on human well-being through their effect on the climate.
The United States Environmental Protection Agency (EPA) has promulgated several regulations that affect
CH4 emissions from a variety of sources. For example, the 2012 New Source Performance Standards and
Amendments to the National Emissions Standards for Hazardous Air Pollutants for the Oil and Natural Gas
Industry are expected to reduce CH4 emissions by 900,000 metric tons annually.2 Additionally, the 2017-
2025 Light-duty Vehicle Greenhouse Gas Emission Standards and Corporate Average Fuel Economy
Standards, promulgated jointly with the National Highway Traffic Safety Administration, is expected to
reduce CH4 emissions by over 100,000 metric tons in 2025 increasing to nearly 500,000 metric tons in
2050.3 It is likely that future EPA rulemakings will also impact CH4 emissions.
Consistent with Executive Order 12866, EPA conducts benefit-cost analysis to inform policy makers and
the public about the economic efficiency of regulatory actions. The value of benefit-cost analysis will, in
part, be determined by the ability to quantify and monetize the relevant outcomes of the regulatory action
under investigation in a scientifically and economically defensible manner. EPA has promulgated
regulations that result in changes in CH4 emissions but has not yet quantified such impacts in its main
benefit-cost analyses. In sensitivity analyses EPA has considered the benefits of CH4 emissions reductions
by using the global warming potential (GWP) metric to convert CH4 emissions into carbon dioxide (C02)
equivalents which are then valued using the U.S. Government's (USG) social cost of carbon (SC-CO2)
1	See EPA Endangerment Finding: Endangerment and Cause or Contribute Findings for Greenhouse Gases Under
Section 202(a) of the Clean Air Act, 74 Fed. Reg. 66,496 (Dec. 15, 2009).
2	http://www.gpo.gov/fdsys/pkg/FR-2012-08-16/pdf/2012-16806.pdf
3	http://www.gpo.gov/fdsys/pkg/FR-2012-10-15/pdf/2012-21972.pdf
3

-------
estimates.4 To date, EPA has not included these indirect CH4 benefit approximations in the main benefit-
cost analyses due to the well-documented caveats associated with the approach (discussed in Section 2).
While some direct estimates of CH4 emissions mitigation benefits have been presented in the scientific
literature, EPA has not used them in benefit-cost analyses because they are inconsistent with USG
estimates of the SC-C02 (discussed in Section 3).5
While it is anticipated that the USG will continue to improve the models and data it uses to estimate the
SC-CO2 in accordance with evolving scientific and economic understanding, this paper illustrates how EPA
could apply the social cost of CH4 (SC-CH4) estimates developed in Marten et al. (2014) to improve upon
the current treatment of methane impacts in regulatory analysis so that they need not be implicitly
assigned a value of zero in USG policy assessment. Marten et al. provide the first set of published SC-CH4
estimates that are consistent with the modeling assumptions underlying the USG SC-C02. This paper
begins by describing the GWP-based approach to valuing CH4 mitigation benefits on the margin and its
limitations. This discussion is followed by a description of the direct approach to estimating the benefits
of marginal CH4 emissions reductions, including a summary of Marten et al. and a comparison of direct
estimates to the GWP-based approach.
2. Global Warming Potential Approximation Approach
The global warming potential (GWP) for CH4 is a measure of the additional energy retained by the Earth's
atmosphere as the result of a pulse of CH4 emissions as compared to a pulse of C02 emissions. Specifically,
the GWP is the time-integrated global mean radiative forcing from one kg of CH4 emissions compared to
one kg of C02 emissions over a given time horizon, such that
pulse of gas i at time zero. The time horizon is typically set to 100 years, but others have used alternative
time horizons, such as 20 or 500 years, and the additional radiative forcing is estimated based on a
constant background concentration for each gas.
The GWP was developed by the IPCC for their First Assessment Report (IPCC 1990) as a simple and purely
physical metric to provide information about the potential impacts of each non-C02 GHG relative to C02
emissions. Under the Kyoto Protocol the GWP was designated for use in translating emissions of non-C02
GHGs into comparable C02 equivalents when estimating GHG sources and sinks.6 Similarly the United
4	The USG SC-CO2 estimates are commonly referred to as the social cost of carbon but are consistent with a one
metric ton change in CO2 emissions.
5	Page 49536 http://www.gpo.gov/fdsys/pkg/FR-2012-08-16/pdf/2012-16806.pdf
6	Decision 2/CP.3 http://unfccc.int/resource/docs/cop3/07a01.pdf#page=31.
GWP
(1)
where T is the time horizon and Q, (t) is the contribution to global mean radiative forcing at time t of a
4

-------
Nations Framework Convention on Climate Change (UNFCCC) requires the use of GWPs with a time
horizon of 100 years when calculating national GHG inventories.7
The GWPs for CH4 as estimated by the IPCC for both the Fourth Assessment Report (AR4) and AR5 are
presented in Table 1. The estimates presented in AR4 include both direct and indirect effects taking into
account the feedback of CH4 emissions on its own lifetime and an estimated 40% increase in the radiative
efficacy of CH4 emissions due to their role as a precursor for tropospheric ozone and stratospheric water
vapor. The CH4 GWP estimates presented in AR5 also include the indirect effects of CH4 emissions, but the
increase in radiative efficacy due to these effects was increased to 55% due to new findings on the role of
CH4 in stratospheric ozone formation. A reassessment of the effect of CH4 on its own lifetime led to an
increase in the effective perturbation lifetime from 12 years in AR4 to 12.4 years in AR5.
Table 1: Global Warming Potential for CH4


AR58



No CC
With CC
Time Horizon
AR49
Feedback
Feedback
20
72
84
86
100
25
28
34
500
7.6
-
-
Starting in AR4 the IPCC included climate-carbon (CC) feedbacks in the estimate of the radiative forcing
projection from the C02 emissions pulse. This feedback accounted for the weakening of carbon sinks from
increases in the temperature. However, these feedbacks were not accounted for when estimating the
additional mean global radiative forcing due to a non-C02 emissions pulse, which would also have an
effect on temperature, and in turn carbon sinks. Therefore, in AR5 the IPCC presented estimates both with
and without the additional radiative forcing from the CC feedback associated with non-C02 emissions.
Inclusion of CC feedbacks in calculating the GWP for CH4 increases the estimate from 28 to 34 for a 100
year time horizon.
2.1 Application of GWP-Based Approach to Benefit-Cost Analysis
The SC-C02 is an estimate of the monetized damages associated with an incremental increase in C02
emissions in a given year. It is intended to include (but is not limited to) changes in net agricultural
productivity, human health, property damages from increased flood risk, and the value of ecosystem
services. As such, the SC-CO2 is an estimate of the benefits of reducing C02 emissions at the margin. The
USG first published SC-CO2 estimates in 2010 following an interagency process that included EPA and
other executive branch entities. The USG used three integrated assessment models (1AM) to develop SC-
C02 estimates and selected four global values for use in regulatory analyses. The USG recently updated
these estimates using new versions of each 1AM and published them in 2013. The 2013 update did not
7	http://unfccc.int/ghg_data/online_help/definitions/items/3817.php
8	Source: Table 8.7 in Myhre et al. (2013).
9	Source: Table 2.14 in Forster et al. (2007).
5

-------
revisit the 2010 modeling decisions (e.g., discount rates, reference case socioeconomic and emission
scenarios, or equilibrium climate sensitivity). Rather, updates were confined to those 1AM modifications
that were implemented by the model developers themselves and subsequently used in the peer-reviewed
literature. The February 2010 Technical Support Document (TSD)10 provides a complete discussion of the
methods used to develop the USG SC-C02 estimates and the November 2013 TSD11 presents and discusses
the updated estimates.
The USG has not developed an estimate of the social cost of CH4 emissions for use in regulatory analysis.
As a result benefit-cost analyses informing U.S. federal rulemakings have not fully considered the benefits
associated with CH4 emissions mitigation. To understand the potential implication of these omissions, EPA
has conducted sensitivity analysis in some of its regulatory analyses using the 100-year GWP to convert
CH4 emission reductions to C02-equivalents, which are then valued using the SC-C02. This approach
approximates the social cost of methane (SC-CH4) using the SC-C02 and the GWP, such that
SC-CH4 ~ GWP100 xSC-C02 .	(2)
2.2 Limitations of the Global Warming Potential Approach for Valuing CH4 Emissions Changes
The GWP is a simple, transparent, and well-established metric for assessing the relative impacts of non-
C02 emissions compared to C02 on a purely physical basis. However, the GWP-based approximation of
the SC-CH4 in (2) has several well-documented limitations (e.g., Reilly and Richards 1993; Schmalensee
1993; Fankhauser 1994; Marten and Newbold 2012). Gas comparison metrics, such as the GWP, are
designed to measure the impact of non-C02 GHG emissions relative to C02 at a specific point along the
pathway from emissions to monetized damages (depicted in Figure 1), and this point may differ across
measures. The GWP measures the cumulative radiative forcing from a perturbation of a non-C02 GHG
relative to a perturbation of C02 over a fixed time horizon. The GWP and other gas comparison metrics
are not ideally suited for use in benefit-cost analyses to approximate the social cost of non-C02 GHGs
because they ignore important nonlinear relationships beyond radiative forcing in the chain between
emissions and damages. These can become relevant because gases have different lifetimes. For example,
the SC-C02 takes into account the fact that marginal damages from an increase in temperature are a
function of existing temperature levels. Another limitation of gas comparison metrics for this purpose is
that some environmental and socioeconomic impacts are not linked to all of the gases under
consideration and will therefore be incorrectly allocated. For example, the economic impacts associated
with increased agricultural productivity due to higher atmospheric C02 concentrations included in the SC-
C02 would be incorrectly allocated to CH4 emissions with the GWP-based valuation approach.











Emissions
->
At"Tｻosphe7|-
CLM(L-:ntr,-',)Wt

anve
Rrunfc
H*
Climate
impacts

ar.d Socio-
Eioronrc
Ir1.pc4.ti

More-.ized
i*.
10	http://www.whitehouse.gov/sites/default/files/omb/inforeg/for-agencies/Social-Cost-of-Carbon-for-RIA.pdf
11	http://www.whitehouse.gov/sites/default/files/omb/assets/inforeg/technical-update-social-cost-of-carbon-for-
regulator-impact-analysis.pdf
6

-------
Figure 1: Path from GHG Emissions to Monetized Damages (Source: Marten et al. (2014))
Furthermore, the assumptions made in estimating the GWP are not consistent with the assumptions
underlying SC-C02 estimates in general, including the USG SC-C02 estimates. For example the 100 year
time horizon usually used in estimating the GWP is less than the 300 year horizon used in developing the
USG SC-CO2 estimates. The GWP-approach also treats all impacts within the time horizon equally,
independent of the time at which they occur. This is inconsistent with the role of discounting in economic
analysis, which accounts for a basic preference for earlier over later gains in utility, the small but positive
probability of a large global catastrophe (e.g., large asteroid collision, super volcanic eruption, nuclear
war), and expectations regarding future levels of economic growth. In the case of CH4, which has a
relatively short lifetime compared to C02, the temporal independence of the GWP could lead the
approximation in (2) to underestimate the SC-CH4 with a larger downward bias under higher discount
rates (Marten and Newbold 2012).12
3. Direct Estimation
The SC-CH4 can be directly estimated using an integrated assessment model (1AM) similar to the way in
which the SC-CO2 is estimated. lAMs couple simplified models of atmospheric gas cycles and climate
systems with highly aggregated models of the global economy and human behavior to represent the
impacts of GHG emissions on the climate and human welfare. Within lAMs, the equations that represent
the influence of emissions on the climate are based on scientific assessments, while the equations that
map climate impacts to human welfare are based on economic research that has studied the effects of
climate on various market and non-market sectors. Estimating the social cost of emissions for a given GHG
at the margin involves perturbing the emissions of that gas in a given year and forecasting the increase in
monetized climate damages relative to the baseline. These incremental damages are then discounted
back to the perturbation year to represent the marginal social cost of emissions of the specific GHG in that
year.
Several researchers have directly estimated the social cost of CH4 emissions using lAMs, though the
number of such estimates is small compared to the large number of SC-CO2 estimates available in the
literature. Among these published direct estimates there is considerable variation in the models and input
assumptions. These studies differ in the emission perturbation year, employ a wide range of constant and
variable discount rate specifications, and consider a range of baseline socioeconomic and emissions
scenarios that have been developed over the last 20 years. However, as discussed by Marten et al. (2014),
none of the other published estimates of the SC-CH4 are consistent with the USG SC-CO2 estimates, and
most are likely underestimates due to changes in the underlying science since their publication. Therefore,
Marten et al. provide the first set of direct estimates of the SC-CH4 that are consistent with the USG SC-
C02 estimates.
12 We note that the truncation of the time period in the GWP calculation could lead to an overestimate of SC-CH4
for near term perturbation years in cases where the SC-CO2 is based on a sufficiently low or steeply declining
discount rate.
7

-------
The estimation approach of Marten et al. (2014) used the same set of three lAMs, five socioeconomic-
emissions scenarios, equilibrium climate sensitivity distribution, and three constant discount rates used
to develop the USG SC-C02 estimates. Marten et al. also used the same aggregation method as the USG
SC-CO2 to distill the 45 distribution of the SC-CH4 produced for each emissions year into four estimates:
the mean across all models and scenarios using a 2.5%, 3%, and 5% discount rate, and the 95th percentile
of the pooled estimates from all models and scenarios using a 3% discount rate.
The primary modeling challenge addressed by Marten et al. (2014) is that two of the three lAMs as
implemented by the USG are not "turn-key" ready to estimate the SC-CH4 due to their lack of an
atmospheric stock-flow model of CH4 emissions and their influence on global mean radiative forcing.
Instead, two of the three model implementations use exogenous projection of aggregate non-C02
radiative forcing, which prevents the direct perturbation of CH4 emissions within the models. Therefore,
to estimate the SC-CH4 Marten et al. applied a simple model to estimate the path of additional radiative
forcing from a CH4 perturbation, which is then added to the exogenous non-C02 radiative forcing
projection to estimate the incremental damages compared to the baseline.
The simple model applied by Marten et al. (2014) used an exponential decay function to project
atmospheric CH4 concentrations from the CH4 emissions projections in the five socioeconomic-emissions
scenarios. They set the average lifetime of CH4 to 12 years following the findings of the IPCC in AR4. The
direct radiative forcing associated with the atmospheric CH4 concentration was estimated using the
functional relationships presented in the IPCC's Third Assessment Report and used in AR4. To account for
the indirect effects of CH4 as a precursor for tropospheric ozone and stratospheric water vapor, Marten
et al. followed the approach of the IPCC in AR4 of increasing the direct radiative forcing by 40%.
The USG SC-CO2 modeling exercise assumed that overall radiative forcing from non-C02 sources remains
constant past 2100 without specifying the projections for individual GHGs that were implicit in that
assumption. This broad assumption was sufficient for the purposes of the USG in estimating the SC-C02;
however, estimating the SC-CH4 requires explicit projections of baseline CH4 emissions to determine the
atmospheric concentration and radiative forcing off of which to compare the perturbation. Marten et al.
(2014) chose to interpret the USG SC-CO2 assumption for non-C02 radiative forcing past 2100 as applying
to each gas individually, such that the emissions of each gas fall to their respective rate of atmospheric
decay. This has the effect of holding global mean radiative forcing due to atmospheric CH4 constant past
2100. Marten et al. showed that, due to the relatively short lifetime of CH4, alternative methods for
extrapolating CH4 emissions past 2100 have only a negligible effect (less than 0.5%) on the SC-CH4.
The SC-CH4 estimates developed by Marten et al. (2014) are presented in Table 2 along with the USG SC-
C02 estimates. For more detailed results and a comparison to other published estimates we refer the
reader to the discussion in Marten et al.
8

-------
Table 2: SC-C02 and SC-CH4 Estimates [2007$ per metric ton] (Source: Marten et al. (2014))
SC-CO2	SC-CH4

5.0%
3.0%
2.5%
3%
5.0%
3.0%
2.5%
3%
Year
Mean
Mean
Mean
95th
Mean
Mean
Mean
95th
2010
11
32
51
89
370
870
1,200
2,500
2015
11
37
57
109
460
1,100
1,400
2,900
2020
12
43
64
128
550
1,200
1,600
3,200
2025
14
47
69
143
660
1,400
1,800
3,800
2030
16
52
75
159
780
1,600
2,100
4,300
2035
19
56
80
175
920
1,900
2,300
5,000
2040
21
61
86
191
1,100
2,100
2,600
5,600
2045
24
66
92
206
1,200
2,300
2,900
6,300
2050
26
71
97
220
1,400
2,500
3,100
6,900
3.1	Application of Direct Estimates to Benefit-Cost Analysis
The application of direct estimates from Marten et al. (2014) to benefit-cost analysis of a regulatory action
is analogous to the use of the SC-C02 estimates. Specifically, the direct estimates would be used to value
decreases in CH4 emissions anticipated from the rulemaking. Forecast reductions in CH4 emissions in a
given year resulting from the regulatory action are multiplied by the SC-CH4 estimate based on a
perturbation in that year. To obtain a present value estimate, the monetized stream of future CH4 benefits
are discounted back to the analysis year using the same discount rate used to estimate the SC-CH4.
Specifically, the present value of benefits from a regulatory action leading to reductions in CH4 emissions
A｣t , t = 0,. ..,H, is
^A｣txSC-CH4itx(l + r)_t,	(3)
t=0
where r is the discount rate used to estimate the SC-CH4. The SC-CH4 estimates would be applied in the
same way to calculate CH4 dis-benefits of a rulemaking that leads to an increase in CH4 emissions.
3.2	Comparison with the Global Warming Approach
The Marten et al. (2014) estimates are based on the conclusions presented in AR4, which was the latest
assessment available when they conducted their modeling and analysis, and therefore GWP estimates
based on the same assumptions would provide the most consistent comparison. As noted in Table 1, the
AR4 100-year GWP for CH4 is 25. However, based on the direct estimates in Table 2 the social cost of CH4
emissions in 2020 are 25-46 times higher than for C02 depending on the discount rate. For emissions in
2050 the SC-CH4 is 31-54 times higher than the SC-C02. Therefore the GWP-based approach to estimating
9

-------
the value of CH4 emissions based on the SC-C02 will likely provide an underestimate particularly for higher
discount rates and future emissions years in this application.
To illustrate the difference between the direct SC-CH4 and GWP-based estimates, Table 3 recalculates the
methane co-benefits of the EPA 2017-2025 Light-duty Vehicle Greenhouse Gas Emission Standards and
Corporate Average Fuel Economy Standards13 using both the Marten et al. (2014) SC-CH4 estimates and
the GWP-based approach. The GWP-based approach underestimates the climate co-benefits of the
expected methane emission reductions by 14% to 50% depending on the discount rate assumption.
Table 3. Methane Co-benefits of 2017-2025 Light Duty Vehicle GHG Standards Using Alternative
Valuation Methodologies [Billion 2007$]14
USG Discount rate
assumption
Using Marten et al.
(2014)
sc-ch4
GWP-Based
Approach
(AR4 100 Year)
% Difference
5% Mean
2.5
1.2
-50%
3% Mean
8.1
6.1
- 26%
2.5% Mean
11.6
9.7
-16%
3% 95th percentile
21.9
18.8
-14%
It should be noted that since the Marten et al. (2014) estimates are based on the IPCC AR4 conclusions,
in some cases the GWP-based approach using AR5 100-year GWP estimates would yield higher benefits
than the current Marten et al. estimates. This occurs for low discount rates and emissions years in the
near term and is due to the higher CH4 indirect effects and climate-carbon feedbacks included in the AR5
GWP estimates. As such, the estimates of Marten et al., by being based on AR4, may be considered
conservative in these regards. The inclusion of new AR5 findings in the approach of Marten et al. is
expected to increase the SC-CH4 estimates, and the relative difference between those updated estimates
and the GWP-based approach using the AR5 100-year GWP are expected to be similar to those discussed
above.
4. Valuing Other N011-CO2 GHG Emissions
While this white paper focuses on methane, the valuation of other non-C02 GHG emissions is relevant to
EPA regulatory analyses and remains an ongoing area of research. At least one promulgated rulemaking
to date, the 2017-2025 Light-duty Vehicle Greenhouse Gas Emission Standards, has been expected to
reduce other non-C02 GHG emissions, specifically N20 and HFC-134a. Marten et al. (2014) provides an
analysis of the SC-N20 parallel to their SC-CH4 analysis. They use the same methodology described for SC-
CH4, replacing the simple CH4 atmospheric gas cycle model with a simple N20 atmospheric cycle, and
found the directly modeled SC-N20 estimates generally exceed those from the GWP-based approach.
13	http://www.gpo.gov/fdsys/pkg/FR-2012-10-15/pdf/2012-21972.pdf
14	NPV of climate benefits resulting from 2017-2050 Cm emission reductions, discounted back to 2012. See RIA,
Table 7.1-4, http://www.epa.gov/otaq/climate/documents/420rl2016.pdf.
10

-------
Their estimates of the SC-N20 are also consistent with the USG SC-C02 estimates and therefore, could also
be used to improve the analysis of regulatory actions projected to influence N20 emissions.
EPA currently does not have directly modeled estimates of the social cost of F-gases, which include a wide
variety of gases spanning a broad range of atmospheric lifetimes and climate impacts. It is difficult to
determine how directly modeled estimates of SC-F gases would compare to estimates from the GWP-
based approach, given the limited number of published estimates at the moment and diversity of gases
in this category.
5. Concluding Remarks
As directed by Executive Orders 12866 and 13563, EPA must use the best available scientific, technical,
economic, and other information to quantify the costs and benefits of regulatory actions. Rigorous
evaluation of costs and benefits has been a core tenet of the EPA rulemaking process for decades. Due to
limitations of the GWP-based approach to value GHG emission impacts and the previous lack of peer-
reviewed SC-CH4 estimates consistent with the USG SC-C02 modeling assumptions, EPA has only
monetized the benefits of CH4 emissions mitigation in sensitivity analysis. However, Marten et al. (2014)
now provides a set of published SC-CH4 estimates consistent with the USG SC-C02 modeling exercise. As
such, the Marten et al. estimates offer a method for improving the analyses of regulatory actions that are
projected to influence CH4 emissions without introducing inconsistency with the manner in which other
C02 mitigation benefits are valued. These estimates can and should be updated if and when the modeling
assumptions underlying the USG SC-C02 estimates are updated to reflect the conclusions of IPCC AR5 or
other evolving scientific and economic knowledge.15
15 The Office of Management and Budget (OMB) recently provided an opportunity for public comment on the
updated November 2013 USG SC-CO2 TSD, in addition to the public comment opportunities available through
particular rulemakings. OMB is currently reviewing the comments received. Any revision to the underlying
modeling assumptions will be addressed separately as the SC-CO2 estimates are updated.
11

-------
References
Fankhauser, S., 1994. The social costs of greenhouse gas emissions: an expected value approach. The
Energy Journal 15 (2), 157-184.
Forster, P., V. Ramaswamy, P. Artaxo, T. Berntsen, R. Betts, D.W. Fahey, J. Haywood, J. Lean, D.C. Lowe,
G. Myhre, J. Nganga, R. Prinn, G. Raga, M. Schulz and R. Van Dorland, 2007: Changes in Atmospheric
Constituents and in Radiative Forcing. In: Climate Change 2007: The Physical Science Basis. Contribution
of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change
[Solomon, S., D. Qin, M. Manning, Z. Chen, M. Marquis, K.B. Averyt, M.Tignor and H.L. Miller (eds.)].
Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA.
IPCC, 1990: Climate Change: The Intergovernmental Panel on Climate Change Scientific Assessment
[Houghton, J.T., G.J. Jenkins, and J.J. Ephraums (eds.)]. Cambridge University Press, Cambridge, United
Kingdom and New York, NY, USA, 364 pp.
Marten, A.L., Newbold, S.C., 2012. Estimating the social cost of non-C02 GHG emissions: methane and
nitrous oxide. Energy Policy 51, 957-972
Marten, A.L., Kopits, E.A., Griffiths, C.W., Newbold, S.C., Wolverton, A. 2014. Incremental CH4 and N20
Mitigation Benefits Consistent with the U.S. Government's SC-C02 Estimates. Climate Policy. Published
online: 20 May 2014.
Myhre, G., D. Shindell, F.-M. Breon, W. Collins, J. Fuglestvedt, J. Huang, D. Koch, J.-F. Lamarque, D. Lee,
B. Mendoza, T. Nakajima, A. Robock, G. Stephens, T. Takemura and H. Zhang. 2013: Anthropogenic and
Natural Radiative Forcing. In: Climate Change 2013: The Physical Science Basis. Contribution of Working
Group I to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change [Stocker, T. F.,
D. Qin, G.-K. Plattner, M. Tignor, S. K. Allen, J. Boschung, A. Nauels, Y. Xia, V. Bex and P. M. Midgley
(eds.)]. Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA.
Reilly, J., Richards, K., 1993. Climate change damage and the trace gas index issue. Environmental and
Resource Economics 3 (1), 41-61.
Schmalensee, R., 1993. Comparing Greenhouse gases for policy purposes. The Energy Journal 14 (1),
245-256.
12

-------
Document Title:_Valuing Methane Emission Changes in Regulatory Benefit-Cost Analysis
Approximate Length: 10 pages
Supporting Materials: Marten et al. (2014) (36 pages excluding Appendices)
Abstract and Charge Questions:
Consistent with Executive Order 12866, EPA conducts benefit-cost analysis to inform policy makers and
the public about the potential economic implications of regulatory actions. EPA has promulgated
regulations that result in changes in CH4 emissions but has not yet quantified such impacts in its main
benefit-cost analyses. Direct estimates of the benefits of mitigating CH4 emissions have been presented
in the scientific literature, but EPA has not used these estimates in benefit-cost analyses because they are
inconsistent with U.S. Government (USG) estimates of the social cost of carbon dioxide (SC-C02).16 A
recently published paper (Marten et al. 2014) presents estimates of the social cost of CH4(SC-CH4) that
are consistent with USG estimates of the SC-C02. While it is anticipated that the USG will continue to
improve the models and data it uses to estimate the SC-C02 in accordance with evolving scientific and
economic understanding, the enclosed paper illustrates how EPA could apply the SC-CH4 estimates from
Marten et al. to improve upon the current treatment of methane impacts in regulatory impact analysis
(RIA) so that they need not be implicitly assigned a value of zero in policy assessment. Consistent with
EPA's peer review guidance, the Agency is seeking review of the application of these new benefit
estimates to regulatory analysis before using them in an RIA. Specifically we seek guidance on the
following questions:
1.	Has EPA correctly interpreted the SC-CH4 estimates provided in Marten et al. (2014) as designed
to measure the monetized value of the climate impacts from marginal changes in CH4 emissions
in a way that is appropriate for use in benefit-cost analysis of regulatory actions projected to
change CH4 emissions?
2.	Do you agree that the Marten et al. SC-CH4 estimates are consistent with the USG SC-C02
estimates?
3.	Do you agree with EPA's characterization of the limitations of using the global warming potential
(GWP) to approximate the SC-CH4 (and other non-C02 GHGs)?
4.	Do you agree with EPA's assessment that direct estimates of the SC-CH4, as developed by Marten
et al., are more appropriate for monetizing changes in CH4 emissions than using the GWP to scale
the USG SC-C02?
16 See the February 2010 Technical Support Document (TSD) and November 2013 TSD Update for a complete
discussion of the methods used to develop the USG SC-CO2 estimates:
http://www.whitehouse.gov/sites/default/files/omb/inforeg/for-agencies/Social-Cost-of-Carbon-for-RIA.pdf,
http://www.whitehouse.gov/sites/default/files/omb/assets/inforeg/technical-update-social-cost-of-carbon-for-
regulator-impact-analysis.pdf.
13

-------
5.	Are there other existing approaches for monetizing the benefits (or dis-benefits) to society from
reductions (increases) in CH4 emissions that should be considered in regulatory analysis?
6.	Although the focus of this review is on the application of estimates of the social cost of CH4 to
benefit-cost analysis for regulations, do your answers for the questions above hold for the
application of the social cost of N20 estimates provided in Marten et al.?
7.	Are there implementation issues not addressed in the paper that EPA should consider before
applying the Marten et al. estimates in regulatory analysis?
14

-------
Document Title: Valuing Methane Emission Changes in Regulatory Benefit-Cost Analysis
Approximate Length: 10 pages
Supporting Materials: Marten et al. (2014) (36 pages excluding Appendices)
Abstract and Charge Questions:
Consistent with Executive Order 12866, EPA conducts benefit-cost analysis to inform policy
makers and the public about the potential economic implications of regulatory actions. EPA has
promulgated regulations that result in changes in CH4 emissions but has not yet quantified such
impacts in its main benefit-cost analyses. Direct estimates of the benefits of mitigating CH4
emissions have been presented in the scientific literature, but EPA has not used these estimates
in benefit-cost analyses because they are inconsistent with U.S. Government (USG) estimates of
the social cost of carbon dioxide (SC-CO2).1 A recently published paper (Marten et al. 2014)
presents estimates of the social cost of CH4 (SC-CH4) that are consistent with USG estimates of
the SC-C02. While it is anticipated that the USG will continue to improve the models and data it
uses to estimate the SC-C02 in accordance with evolving scientific and economic understanding,
the enclosed paper illustrates how EPA could apply the SC-CH4 estimates from Marten et al. to
improve upon the current treatment of methane impacts in regulatory impact analysis (RIA) so
that they need not be implicitly assigned a value of zero in policy assessment. Consistent with
EPA's peer review guidance, the Agency is seeking review of the application of these new
benefit estimates to regulatory analysis before using them in an RIA. Specifically we seek
guidance on the following questions:
1. Has EPA correctly interpreted the SC-CH4 estimates provided in Marten et al. (2014) as
designed to measure the monetized value of the climate impacts from marginal changes
in CH4 emissions in a way that is appropriate for use in benefit-cost analysis of
regulatory actions projected to change CH4 emissions?
I have read both Marten et a I. (2014) and the review document and feel that the review
document provides an accurate summary of the issues and methodologies discussed in
Marten et al. (2014). I feel that Table 3 of the review document provides a nice example
of how the SC-CH4 estimates from Marten et al. (2014) could be used in BCAs of
proposed regulations and underscores the bias that arises if a GWP-based approach is
used rather than the direct approach proposed by Marten et al. (2014).
There, of course, is a whole host of issues that arise when applying any social cost
measure to regulatory analyses, which have been extensively discussed in the literature
1 See the February 2010 Technical Support Document (TSD) and November 2013 TSD Update for a complete
discussion of the methods used to develop the USG SC-C02 estimates:
http://www.whitehouse.gov/sites/default/files/omb/inforeg/for-agencies/Social-Cost-of-Carbon-for-RIA.pdf,
http://www.whitehouse.gov/sites/default/files/omb/assets/inforeg/technical-update-social-cost-of-carbon-for-
regulator-impact-analysis.pdf.
1

-------
and which I expand on in my responses below. A key issue that I would like to raise here
is that these measures are only appropriate for marginal changes in CH4. These
measures are not designed to be used to evaluate non-marginal changes in methane
emissions (or any other gas, for that matter). Therefore, caution must be used when
applying social cost measures like this.
2.	Do you agree that the Marten et al. SC-CH4 estimates are consistent with the USG SC-
C02 estimates?
"Consistent" can have many interpretations. I will say that the Marten et al. SC-CH4
estimates are computed in a similar way as the SC-C02 estimates, so in this regard, the
two estimates are "consistent." However, C02 is more explicitly modeled in the three
models than CH4 so in this regard they are not "consistent." However, this inconsistency
is due to limitations of the models and I feel that Marten et al. have taken appropriate
steps to address these limitations the best way possible. However, gaps still remain and
should be recognized.
3.	Do you agree with EPA's characterization of the limitations of using the global warming
potential (GWP) to approximate the SC-CH4 (and other non-C02 GHGs)?
The review document (and Marten et al) discusses a number of problems that arise when
GWP is used to approximate SC-CH4: (1) in the introduction and in section 2, the authors
point out that the indirect effects of CH4, as a precursor to tropospheric ozone and
stratospheric water vapor, can amplify radiative forcing significantly (which would not
be captured in the GWP); (2) GWP ignores important nonlinear relationships beyond
radiative forcing in the chain between emissions and damages容.g., increased
agricultural productivity due to C02 fertilization would be incorrectly attributed to CH4 if
the GWP was used; (3) GWP does not account for differences in time horizons between
gases容.g., since CH4 has a shorter lifetime than CO2, the GWP approach would
underestimate the SC-CH4.
Although all three are technically correct, I feel that (1) and (2) could be addressed to a
certain extent (although not perfectly) by adjusting the GWP to account for these biases.
However, the temporal issue raised in (3) seems more difficult to address through simple
adjustments to the GWP.
In sum, I agree with the authors that problems exist and that the direct approach in
theory is the best way to avoid these issues.
4.	Do you agree with EPA's assessment that direct estimates of the SC-CH4, as developed
by Marten et al., are more appropriate for monetizing changes in CH4 emissions than
using the GWP to scale the USG SC-C02?
2

-------
As discussed in my response to question 3 above, I agree that the direct approach is
likely a superior approach to the indirect GWP approach. However, it should be noted
that the direct approach has issues as well. Namely, as discussed in section 3, most
models do not include an atmospheric stock-flow model of CH4; thus, the authors were
forced to develop a separate model to project the path of radiative forcing from a CH4
perturbation, and then incorporate this path into the 1AM exogenously. As a result,
indirect or feedback effects are missed. For instance, climate change impacts on
agriculture will affect methane emissions.
In sum, no approach is perfect but in my opinion, the "direct' approach used by Marten
et al is preferred to the indirect GWP approach for the reasons outlined in the review
document. However, the EPA should continue to seek improvements to the direct
approach put forth by Marten et al.
5.	Are there other existing approaches for monetizing the benefits (or dis-benefits) to
society from reductions (increases) in CH4 emissions that should be considered in
regulatory analysis?
My complaint with pastSC measures is the use of highly aggregated and stylized models
to monetize the benefits of reductions. By using models that represent the global
economy as one aggregate sector, we are missing important subsector interactions and
distributional effects that can only be captured with a more disaggregated model, such
as a computable general equilibrium model. My sense would be that these SC would be
much higher if a more disaggregated model was used. Modeling the economy as one
monolithic sector implies, for instance, perfect substitutability across subsectors which
will underestimate the cost of damages. It also assumes perfect trade which can also
underestimate the cost of damages. (See Chapter 6 of the IPCC WGIII Fifth Assessment
Report which highlights some of these biases that arise with alternative model
characteristics).
The use of these simplified models for SC estimates, I believe, is a large source of the
criticisms we've seen with respect to the SCC reports. The use of more sophisticated
economic models (like those used in the IPCC) is needed, in my opinion.
6.	Although the focus of this review is on the application of estimates of the social cost of
CH4 to benefit-cost analysis for regulations, do your answers for the questions above
hold for the application of the social cost of N20 estimates provided in Marten et al.?
Yes.
7.	Are there implementation issues not addressed in the paper that EPA should consider
before applying the Marten et al. estimates in regulatory analysis?
3

-------
I am not sure I would characterize these as "implementation" issues, but I do want to
take this opportunity to stress the importance of being forthcoming with the
shortcomings of these SC estimates. These shortcoming are not specific to any gas.
(1)	As discussed in my response to question 1, these estimates are not appropriate for
evaluating large (non-marginal) changes in emissions of any of these gases.
(2)	As discussed in my response to question 5, the SC values will be underestimated due
to the use of highly aggregated models.
(3)	These estimates do not take into account extreme or threshold events, which could
amplify the estimates significantly.
(4)	These estimates will be biased downward due to the omissions of nonmarket values
and omitted impacts, and will be biased upward due to the lack of adaptation
responses (although FUND does account for some of this).
4

-------
John Reilly, MIT
December 28, 2014
This report is a response to a request from Katherine Kiel (Dec. 10, 2014) to review
a draft EPA (no date) paper, "Valuing Methane Emission Changes in Regulatory
Benefit-Cost Analysis." As part of the request I was provided with the 7 charge
questions repeated below in italics, the EPA paper, and the paper: Alex L. Marten,
Elizabeth A. Kopits, Charles W. Griffiths, Stephen C. Newbold, and Ann Wolverton,
2014 (on line) Incremental C4H and N2O mitigation benefits consistent with the US
Government's SC-CO2 Estimates, Climate Policy, which forms the basis for estimates
provided in the draft EPA paper.
The basic objective of the EPA paper was to outline a process for establishing a
social cost of methane (and possibly N2O) that is consistent with the established
basis for estimating a social cost of carbon previously developed by EPA.
1.	Has EPA correctly interpreted the SC-CH4 estimates provided in Marten et al.
(2014) as designed to measure the monetized value of the climate impacts from
marginal changes in CH4 emissions in a way that is appropriate for use in
benefit-cost analysis of regulatory actions projected to change CH4 emissions?
The Marten et al. (2014) paper follows closely the original social cost of
carbon approach developed by the EPA, using the same 3 IA models,
expanding them to include methane and nitrous oxide. EPA's interpretation
of the paper appears to be correct. The main addition was an explicit
treatment of the lifetime of methane (and nitrous oxide). The formulation
used is obviously a simplification of complex atmospheric chemistry but has
been used in earlier publications and likely approximates a more complex
representation. The ad hoc increase in radiative forcing to account for
indirect effects is another simplification, and obviously has substantial
impacts on the estimates. It is justified by the IPCC indirect estimates. It is
not clear that the method includes the fact that abiogenic methane decays
into CO2 and hence may represent an additional impact of methane release.
(Biogenic methane also decays into CO2 but if that methane is derived from
plant material that regrows it would then not represent an addition of CO2 to
the atmosphere.)
2.	Do you agree that the Marten et al. SC-CH4 estimates are consistent with the
USG SC-CO2 estimates?
While there is considerable controversy about how to estimate a Social Cost
of Carbon from a theoretical standpoint as well as the empirical foundation
for such an estimate, the method put forward by Marten et al. (2014) is
theoretically and empirically consistent with the original Social Cost of
Carbon estimates developed by EPA.

-------
John Reilly, MIT
December 28, 2014
3. Do you agree with EPA's characterization of the limitations of using the global
warming potential (GWP) to approximate the SC-CH4 (and other non-C02
GHGsｦ)?
I agree that using GWP's to scale the social cost of carbon would be
theoretically inconsistent. As the EPA (no date) paper discusses, the Social
Cost of Carbon method appropriately uses a discount rate to weight damages
at different points in time, whereas the GWP approach stops at radiative
forcing and then uses an arbitrary time horizon to truncate the effects,
weighting effects in each year equally. This leads to the controversy about
which GWP time horizon to use. Of course this controversy is not completely
avoided as it resurfaces as a controversy about the appropriate discount rate.
It appears to turn out that given the time path of damages the 100-year GWP
of methane is very similar to the Social Cost of Methane relative to the Social
Cost of Carbon as estimated in the Marten et al (2014) paper. (This was a
conclusion Reilly and Richards (1993) reached.) With a very different path of
damages this result may not hold. For that reason as EPA imagines updating
these estimates, and for theoretical consistency, using the Marten et al.
(2014) method for arriving at a Social Cost of methane (or nitrous oxide)
appears much more defensible.
While it does not affect the basic conclusions, I have some issues with the
paragraph in EPA (no date) repeated below in italics, especially the sentences
highlighted here in bold.
Furthermore, the assumptions made in estimating the GWP are not
consistent with the assumptions underlying SC-CO2 estimates in general,
including the USG SC-CO2 estimates. For example the 100 year time
horizon usually used in estimating the GWP is less than the 300
year horizon used in developing the USG SC-CO2 estimates. The
GWP-approach also treats all impacts within the time horizon equally,
independent of the time at which they occur. This is inconsistent with
the role of discounting in economic analysis, which accounts for a
basic preference for earlier over later gains in utility, the small but
positive probability of a large global catastrophe (e.g., large
asteroid collision, super volcanic eruption, nuclear war), and
expectations regarding future levels of economic growth. In the
case of CH4, which has a relatively short lifetime compared to CO2, the
temporal independence of the GWP could lead the approximation in
(2) to underestimate the SC-CH4 with a larger downward bias
under higher discount rates (Marten and Newbold 2012).1
1 We note that the truncation of the time period in the GWP calculation could lead to an overestimate
of SC-CH4 for near term perturbation years in cases where the SC-CO2 is based on a sufficiently low
or steeply declining discount rate.

-------
John Reilly, MIT
December 28, 2014
Regarding the inconsistency of the 100- and 300-year horizons: Yes, I
suppose this is true but there really is no direct comparison. In the economic
analysis one hopefully has a far enough time horizon so that with discounting
it is irrelevant. In some sense choice of discount rate is a substitute for the
choice of time horizon葉he higher the discount rate the shorter the time
horizon. In a final version of this paper I might rephrase this as something
like. In the USG SC-CO2 estimates a 300-year time horizon was used, long
enough to minimize its effects on estimates given the discount rates used. In
contrast the GWP approach is to truncate estimates at different time horizons (20-
, 100-, 500-years), treating all impacts within the time horizon equally,
independent of the time at which they occur. I think this gets across the point you
want to make without directly suggesting that the 100-year and 300-year horizons
are inconsistent (when in fact that is not even comparable.)
Then the second emboldened sentence raising a huge set of issues and
controversies. I think the sentence would be best deleted. The discount rate
should not theoretically include the risk of catastrophe. Risks should be
separately evaluated with a risk-free discount rate to arrive at an "expected" social
cost of carbon, perhaps with a utility function that more heavily weights bad
outcomes. While an observed rate of return can include a risk premium based on
a specific assessment of the risk (and time profile of the risk) it is inappropriate to
apply a risk premium to a discount rate and then apply that risk-adjusted rate to
many different investment profiles. Here, different characterizations of when
catastrophes may occur. Embedding risk into the discount rate in this manner is
little different than using GWP's with truncated time horizons to implicitly give
different weights (1 or 0) to damages occurring at different times. And while in a
Ramsey model the discount rate is approximately the sum of the pure rate of time
preference plus the growth rate that again is calculation under certainty so using
"expected growth" is inconsistent. Then you have the Weitzman argument that
with uncertainty in the appropriate discount rate, one should use a declining rate.
And bringing up things like asteroid collisions and such just seems distracting
here.
Finally, I guess the last emboldened statement is true but it took me a long time to
figure it out. A higher discount rate, as compared to a lower rate, will lead to a
lower social cost of carbon (or methane). So concluding that it will underestimate
the Social Cost of Methane seemed initially backward. Further a higher discount
rate, while lowering the social cost of both gases, will tend to raise the Social Cost
of Methane relative to that of carbon. But I guess if I fully parse this sentence,
you are saying that taking a specific GWP-horizon (e.g 100 years) and deriving a
SC of methane by applying it to your existing SC of carbon, then if you were to
do this the right way with a high discount rate葉hat SC of methane would be
higher then that derived using the shorthand method. Maybe there is a clearer
way to say this.. .but then I'm not sure why this is important. There seems to be a
concern about underestimating the methane value. You could just as easily say,

-------
John Reilly, MIT
December 28, 2014
that methane would be overvalued for low discount rates. I'd think you just want
the CS of carbon and methane to be consistent and unbiased in either direction.
4. Do you agree with EPA's assessment that direct estimates of the SC-CH4, as
developed by Marten et al., are more appropriate for monetizing changes in
CH4 emissions than using the GWP to scale the USG SC-CO2?
Yes, see above.
5.	Are there other existing approaches for monetizing the benefits (or dis-
benefits) to society from reductions (increases) in CH4 emissions that should be
considered in regulatory analysis?
Not of which I am aware. As the Reilly and Richards (1993) paper referred to
in the Marten et al. (2014) paper the multiple impacts of these different
gases, beyond climate change, could in principle be incorporated into the
analysis but that raises further complications. E.g. CO2 has some benefit to
crop growth (disputed) but ozone (of which methane is a precursor) has not
only climate implications but also damages to crops and health. However,
with all the recognized limitations to the empirical foundation for the SC
estimates, the chosen approach is theoretically sound.
6.	Although the focus of this review is on the application of estimates of the social
cost ofCH4 to benefit-cost analysis for regulations, do your answers for the
questions above hold for the application of the social cost ofN20 estimates
provided in Marten et al.?
Yes, the method is equally applicable to N20.
7.	Are there implementation issues not addressed in the paper that EPA should
consider before applying the Marten et al. estimates in regulatory analysis?
As the paper itself points out, the current approach of using a social cost of 0
is clearly not right and so whatever the limitations of existing methods its
seems better to use something rather than nothing. Of course one could use
a value that is so high that zero would be preferable, but I don't see that error
here. More to the point: Accepting the Social Cost of Carbon estimates, this
approach consistently applies the concept to methane (and potentially other
GHGs).
John Reilly
MIT

-------
Peer Review of EPA Proposed Methodology "Valuing Methane Emission Changes in Regulatory
Benefit-Cost Analysis"
Steven Rose, Ph.D., Energy and Environmental Analysis Research Group, EPRI
January 15, 2015
This is an extremely challenging research area due to the breadth of physical and social sciences
represented and the vast uncertainty inherent in modeling global biophysical and economic systems for
centuries to come. EPA should be commended for seeking peer review feedback. Peer review is
important for producing scientifically defensible results and instilling public confidence. However, given
the regulatory importance of the social cost of greenhouse gas estimates, a different peer review
process should be pursued going forward that will increase public confidence in the ultimate outcome
(see suggestion at the end of my comments).
For this review activity, EPA requested peer review feedback on seven charge questions. Below I have
responded to each. Overall, I am concerned about moving forward with direct non-C02 social cost
estimates based on the USG SC-C02 methodology before having a peer reviewed SC-C02 methodology.
Responses to EPA Charge Questions
1.	Has EPA correctly interpreted the SC-CH4 estimates provided in Marten et al. (2014) as designed
to measure the monetized value of the climate impacts from marginal changes in CH4 emissions
in a way that is appropriate for use in benefit-cost analysis of regulatory actions projected to
change CH4 emissions?
Yes, the Marten et al. SC-CH4 estimates are derived with marginal global methane pulses and as
such would be conceptually appropriate for valuing incremental changes in global methane
emissions such as those likely to result from U.S. regulatory actions. However, as discussed
below, there are computational issues with the specific Marten et al. SC-CH4 estimates that
need to be considered; and, implementation issues associated with using the SC-CH4 estimates
to value regulatory action methane changes.
2.	Do you agree that the Marten et al. SC-CH4 estimates are consistent with the USG SC-C02
estimates?
The Marten et al. estimates are mostly consistent with the USG SC-C02 estimates. They are
consistent in a variety of ways with the USG SC-C02 estimates due to the common experimental
design (e.g., same three integrated assessment (IA) models, uncertainty specification, scenario
runs, and results aggregation procedure). However, they do not appear to be entirely
consistent, and the implications are not clear to me. For instance, the simple model used to
compute CH4 concentrations and radiative forcing is different from the modeling used to
construct non-C02 forcing for the PAGE and DICE model's USG SC-C02 reference and pulse
scenarios. It is not clear to me whether this is a big deal. One would want to compare reference
and incremental perturbation responses for all the EMF-22 emissions scenarios. Marten et al.
makes some comparison to MAGICC 5.3, but doesn't comment on reference scenario
differences (that could impact SC-C02 estimates) or on the differences for higher and lower

-------
emissions scenarios. Another inconsistency is the implementation of the CH4 and C02
perturbations. The C02 perturbations in the USG SC-C02 calculations vary by model, differing in
temporal implementation and magnitude (see Section 5 in Understanding the Social Cost of
Carbon: A Technical Assessment, http://epri.co/3002004657). In DICE, a 1 GtC shock was added
over the decade which straddles year t, in FUND, a 1 million metric ton carbon (1 MtC) shock
was added to every year within a decade from year t forward, and in PAGE, 100 billion metric
tons of C02 (100 GtC02, 27 GtC) was distributed evenly over the decades preceding and
subsequent to year t.1 In Marten et al. however, the CH4 perturbation was a 1 MtCH4 pulse in a
single year t.
More importantly, while consistency is the driving motivation for the Marten et al. paper, it
creates a serious problem. The Marten et al. SC-CH4 estimates inherit all the issues associated
with the USG SC-C02 estimates. Overall, consistency with the USG SC-C02 methodology is a
scientifically pragmatic and laudable objective for SC-CH4 calculations. However, it requires that
the SC-C02 methodology be scientifically sound. However, the USG SC-C02 methodology, and
the IA models themselves, have not undergone peer review; and, a number of fundamental
issues have been identified that could merit and motivate revisions to the USG SC-C02
methodology (and resulting SC-C02 & SC-CH4 estimates). Scientifically, it would be inappropriate
at this stage to propagate issues with the SC-C02 methodology by moving forward with the
proposed Marten et al. (2014) SC-CH4 (and SC-N20) estimates. Unfortunately, consistency alone
is not adequate justification for using the Marten et al. estimates. A peer reviewed SC-C02
methodology is needed before creating consistent SC-X estimates for non-C02 gases. See reply
below to Charge Question #4 for some of the important issues that need to be considered.
3.	Do you agree with EPA's characterization of the limitations of using the global warming potential
(GWP) to approximate the SC-CH4 (and other non-C02 GHGs)?
Yes, conceptually, I agree with the limitations noted. However, practically, there are
methodological issues that should be re-considered, and potentially revised, before one can
legitimately begin to claim that the SC-CH4 estimates are improvements over GWP application.
See reply to Charge Question 4.
4.	Do you agree with EPA's assessment that direct estimates of the SC-CH4, as developed by Marten
et al., are more appropriate for monetizing changes in CH4 emissions than using the GWP to
scale the USG SC-C02?
Conceptually, direct estimates of SC-CH4 are more appropriate than using GWPs to value CH4
emissions changes associated with US regulatory actions. Practically, however, is another issue.
It depends on the scientific soundness of the SC-CH4 estimation methodology. Unfortunately,
the current USG SC-C02 methodology, and therefore the proposed SC-CH4 methodology, has not
been peer reviewed to establish scientific soundness, and as noted, fundamental issues with the
current methodology have been identified. For instance, we recently completed and published a
very extensive technical assessment of the USG SC-C02 modeling that coded up individual
1 Note that with PAGE, the emissions pulse is initially introduced as a uniform increase in average annual C02
emissions over the given period associated with year t. However, within PAGE'S climate model, emissions for years
t-1 and t are averaged. Thus, the emissions pulse enters PAGE'S carbon cycle as uniform (but half-sized) increases
in average annual emissions in both the decades preceding and following year t.

-------
components of the models and ran diagnostic scenarios (Understanding the Social Cost of
Carbon: A Technical Assessment, http://epri.co/3002004657). From the analyses, we identified a
number of fundamental issues with the USG SC-C02 approach that should be considered:
a.	Significant structural & response differences across models that need to be evaluated to
determine whether they are providing useful information or are differences to reconcile or
address explicitly as an uncertainty. For instance, the models do not consider the same sets
of emissions and radiative forcing categories葉he drivers for projected temperature. The
models also have stark differences in key pieces of the climate modeling (e.g., the carbon
cycle, climate sensitivity, climate feedbacks, and projected climate change uncertainty), with
some elements excluded entirely from some models. Furthermore, unique model specific
factors dominate results in the damage components and therefore raise questions about
their representation within each model and across models容.g., agricultural C02
fertilization, cooling energy demand, global damages dominated by China, regional scaling
of damages, rapidly growing global non-economic damages, potential discontinuity
damages, and damages that increase quadratically with temperature. Finally, the study also
finds dramatic differences in estimated damages across models for comparable regions and
sectors that are not explored or explained.
b.	Reasonable alternative specifications, additional uncertainties, and some variation that is
artificial due to, for instance, difference in model implementation. Together, these findings
suggest the need to revisit the representation of uncertainty in the experimental design.
c.	Inconsistencies across modeling, as well as inter-model dependency, that raises an issue
about the statistical comparability of results produced by the three different models.
Statistical comparability and independence is required for USG SC-C02 approach which
combines 150,000 results from the three IA models into a distribution in order to derive a
single USG SC-C02 for a given year and discount rate.
d.	The current USG SC-C02 estimates may not be robust (i.e., insensitive to alternative
assumptions) given that (i) the study finds the underlying climate and damage results from
the models (e.g., concentrations, radiative forcing, temperature, and sector and regional
specific damages over time) to be very sensitive to alternative assumptions, and (ii) the
study finds reasonable alternatives to the assumptions and modeling used in the USG SC-
C02 experiment.
e.	Issues with the overall experimental design, in particular the use of multiple models, which
creates the consistency and comparability challenges and issues noted previously.
Based on these findings, the study makes a number of recommendations that could help
increase scientific and public confidence in the SC-C02 results:
a. Internally review the modeling to evaluate differences, improve comparability and
uncertainty representation, and enhance robustness.
b. Revisit the experimental design, especially given the challenges with the multi-model
approach.

-------
c.	Evaluate robustness to reasonable alternative assumptions and modeling to insure that the
results are stable.
d.	Peer review the approach and the models used. The USG SC-C02 approach is novel and peer
review would be valuable and practical. Model review would also be practical given the
regulatory use of the models.
e.	Additional documentation and justification for methodological choices to facilitate
communications & interpretation, and increase public confidence.
f.	Application guidance to insure proper application of USG SC-C02 estimates.
Other researchers, of course, have also identified issues with the current USG SC-C02 approach.
Marten et al. cite a number of studies (Arrow et al., 2013; Kopp & Mignone 2012; Marten, 2011;
O'Neil, 2010; Warren, Mastrandrea, Hope, & Hof, 2010). And recently, the scientific community
has called for a more scientific process with greater scientific community engagement and
formal peer review (Pizer et al., 2015).
5.	Are there other existing approaches for monetizing the benefits (or dis-benefits) to society from
reductions (increases) in CH4 emissions that should be considered in regulatory analysis?
Global temperature potentials could be considered, but they would have similar time period
issues as GWPs.
6.	Although the focus of this review is on the application of estimates of the social cost of CH4 to
benefit-cost analysis for regulations, do your answers for the guestions above hold for the
application of the social cost of N20 estimates provided in Marten et a I. ?
Yes, I have similar concerns about the Marten et al. SC-N20 estimates and would be reluctant to
move forward with the Marten et al. SC-N20 and SC-CH4 estimates.
7.	Are there implementation issues not addressed in the paper that EPA should consider before
applying the Marten et a I. estimates in regulatory analysis?
Yes, in addition to the issues raised in my comments above regarding the Marten et al. SC-CH4
(and SC-N20) estimates, there are a few issues regarding use of SC-CH4 and SC-N20 values. First,
in regulatory applications, it will be essential to estimate net global changes in emissions due to
proposed rules in order to appropriately utilize SC-CH4 and SC-N20 estimates, which reflect the
marginal value of net global changes in methane and nitrous oxide respectively. Second, it will
be important to think about consistency in the underlying assumptions in benefit and cost
calculations, e.g., those used for computing social costs of GHGs, GHG emissions reductions, and
compliance costs. Third, current USG guidance to use all SC-C02 estimates will presumably be
the same for SC-CH4 and SC-N20 estimates. As such, the guidance needs to be expanded to
provide direction on how agencies should use the multiple resulting C02 (CH4, or N20) benefit
estimates in benefit-cost analyses and regulation proposal decisions.
Additional comment

-------
Given the regulatory importance of the social cost of greenhouse gas estimates, a more extensive and
public peer review process should be pursued going forward that will give the public greater confidence
in the ultimate values. Specifically, peer review of the USG SC-C02 methodology (and the subsequent
non-C02 social cost methodologies) should be a public process with a scientific review panel (created
through a public selection process) that produces a single report reflecting the panel's critique and
recommendations. This sort of review process is typical for important regulatory metrics and
methodologies, and a key function of groups like EPA's Science Advisory Board. The peer review I'm
participating in here is a useful means of soliciting scientific feedback, but not a substitute for the review
process needed (and described above).
References
Arrow, K., Cropper, M., Gollier, C., Groom, B., Heal, G., Newell, R., Weitzman, M. (2013). Determining
benefits and costs for future generations. Science, 341, 349-350.
Kopp, R. E., B.K. Mignone (2012). The U.S. government's social cost of carbon estimates after their first
two years: Pathways for improvement. Economics, 6(15), 1-41.
Marten, A. L. (2011). Transient temperature response modeling in lAMs: The effects of over
simplification on the SCC. Economics, 5, 2011-2018.
O'Neill, B. (2010). Multi-century scenario development and socioeconomic uncertainty. In Improving the
assessment and valuation of climate change impacts for policy and regulatory analysis:
Modeling climate change impacts and associated economic damages. US Environmental
Protection Agency/US Department of Energy. Retrieved from
http://yosemite.epa.gov/ee/epa/eerm.nsf/vwAN/EE-0564-115.pdf/$file/EE-0564-115.pdf
Pizer, W., M. Adler, J. Aldy, D. Anthoff, M. Cropper, K. Gillingham, M. Greenstone, B. Murray, R. Newell,
R. Richels, A. Rowell, S. Waldhoff, J. Wiener (2014). Using and improving the social cost of
carbon, Science 5: 1189-1190.
Warren, R., Mastrandrea, M. D., Hope, C., & Hof, A. F. (2010). Variation in the climatic response to SRES
emissions scenarios in integrated assessment models. Climatic Change, 102, 671-785.

-------