A variety of rules and regulations considered by the U.S. federal government—such as energy efficiency standards, fuel economy standards, and power plant regulations—affect the emissions of carbon dioxide (CO2) and other greenhouse gases into the atmosphere.1 For more than three decades, presidential Executive Orders (EOs) have required that federal agencies consider the monetized impact of effects when conducting regulatory impact analyses: see Box 1-1. This report takes a pragmatic approach in offering conclusions and recommendations that are consistent with this approach to regulatory analysis.
In 2008, a ruling by the Ninth Circuit Court of Appeals remanded a fuel economy rule to the Department of Transportation, concluding that it was “arbitrary and capricious” to not monetize the benefits of the CO2 emission reductions in the rule’s regulatory impact analysis.2 In 2009 an interagency working group was formed and developed an approach for estimating the “social cost of carbon” that has been used in dozens of benefit-cost analyses since 2010. The social cost of carbon (SC-CO2)3 is
1 A recent Congressional Research Service report, Federal Citations to the Social Cost of Greenhouse Gases, includes a table that lists federal actions that used the SC-CO2 estimates; the earliest action is April 2008. The report is available at https://fas.org/sgp/crs/misc/R44657.pdf [December 2016].
2 Center for Biological Diversity v. National Highway Traffic Safety Administration, U.S. Court of Appeals, Ninth Circuit, 538 F.3d 1172 (9th Cir. 2008).
3 The acronym for the social cost of carbon in the committee’s interim report was “SCC,” following the then-standard acronym. In late August 2016, the newly renamed Interagency
defined for a given year as the present discounted value of the future damage4 caused by a 1 metric ton increase in CO2 emissions to the atmo-
Working Group on the Social Cost of Greenhouse Gases (previously, the Interagency Working Group on the Social Cost of Carbon) introduced the acronym “SC-CO2.” This report uses the new acronym, except when referring to text from previously published documents.
4 Throughout this report, “damage” represents the net effects of both negative and positive economic impacts of climate change. When incorporated in a benefit-cost analysis, such as a regulatory impact analysis, these net damages are reflected as a benefit of emissions reduction. In benefit-cost analysis, the benefit of a commodity is measured by what people are willing to pay for it. It is important to note that willingness to pay is constrained by ability to pay. The notion that the value attached to a commodity should be constrained by resources is fundamental to economics. When one values the output of a commodity sold in markets one uses a demand function, which reflects the willingness to pay of consumers for purchasing additional units of the good. This is conditional on the distribution of income in society. It is when costs and benefits are added together to determine the net benefits of a decision that principles of benefit-cost analysis enter in. In measuring the economic net benefits of an action, one compares the benefits, as defined above, with the costs of the action. This is an appropriate decision-making criterion, but it does not directly take distributional issues
sphere in that year, or, equivalently, the benefits of reducing CO2 emissions by the same amount in that year.
The Interagency Working Group on the Social Cost of Greenhouse Gases (IWG),5 composed of experts from multiple federal agencies, develops and maintains the SC-CO2 estimates. The current estimation approach
into account. Some individuals may face net costs from the action, and others may face net gains. To provide information on distributional impacts for policy making, the U.S. Office of Management and Budget (OMB) guidance suggests that the distribution of costs and benefits be measured in regulatory impact analysis. Distributional effects can also be reflected in benefit-cost analysis using welfare weights, although this is rarely done in practice and is not permitted in regulatory impact analysis.
5 The IWG is cochaired by the Council of Economic Advisors and the Office of Management and Budget; the other members are the Council on Environmental Quality, the Domestic Policy Council, the Department of Agriculture, the Department of Commerce, the Department of Energy, the Department of the Interior, the Department of Transportation, the Department of the Treasury, the Environmental Protection Agency, the National Economic Council, and the Office of Science and Technology Policy.
was developed in 2009-2010 and released in 2010. The approach has not changed since this initial release, although individual model modifications and other changes were made in 2013 and 2015. The IWG is considering more significant updates to the approach used to estimate the SC-CO2 and asked the National Academies of Sciences, Engineering, and Medicine (hereafter referred to as the Academies) to make recommendations on potential approaches that warrant consideration in future updates of the SC-CO2 estimates. The charge to the Academies also requested recommendation for research that would advance the science in areas that are particularly useful for estimating the SC-CO2. See Box 1-2 for the full statement of task for the committee.
The committee interpreted the charge as focusing specifically on the SC-CO2 for its use in federal regulatory impact analysis. As discussed above, the committee therefore developed its conclusions and recommendations to be consistent with an overall analytical approach based on the computation of expected net present value, taking into account that
the SC-CO2 is one of a large number of variables that enter into a typical regulatory impact analysis. In doing so, the committee notes that the particular regulations of interest are typically of only incremental impact in the context of total U.S. or global CO2 emissions. The resulting SC-CO2 estimates are therefore not necessarily applicable for use as the basis of very large-scale policy issues, such as a comprehensive national carbon price. At a minimum, care needs to be taken in such applications of the SC-CO2.
The IWG’s formulation of the SC-CO2 also differs from much academic work on the issue, which often focuses on optimal global CO2 control: in this work, an optimal emissions control level is set so that its marginal cost is equal to marginal damage, and an SC-CO2 estimate in this case is computed using the optimal emissions pathway. The committee also notes that the analytical framework used in developing the IWG SC-CO2 estimates is based on probability-weighted present value. Although this is appropriate for its application in regulatory impact analysis, it is not
the only framework relevant to decision making under uncertainty in the context of national and international climate policy. Approaches to the treatment of uncertainty are discussed in Chapter 2.
Academic research into the estimation of the social costs of greenhouse gas emissions began with work by economist William Nordhaus in the early 1980s (Nordhaus, 1982) and was continued by numerous researchers in the early 1990s (e.g., Ayres and Walter, 1991; Nordhaus, 1991; Haraden, 1992; Peck and Teisberg, 1992; Reilly and Richards, 1993; Fankhauser, 1994).6 Researchers continued to explore the SC-CO2 over the subsequent two decades. This research base informed the initial estimates and the current approach adopted by the IWG.7
Prior to 2008, changes in CO2 emissions associated with proposed policies were generally not valued in federal regulatory impact analyses (RIAs). As noted earlier, following a 2008 court ruling, federal agencies began to account for the impact of CO2 emissions in their analyses. Agencies estimated dollar values for the SC-CO2 using a variety of methodologies.
In 2009, the Obama Administration formed the IWG and charged it with developing a consistent set of SC-CO2 estimates to be used in regulatory impact analyses. The IWG comprised relevant subject-matter experts from federal agencies; all federal agencies were welcome to participate. For developing the SC-CO2 estimates and making decisions on updates, the IWG used consensus-based decision making, relied on existing academic literature and models, and took steps to disclose limitations and incorporate new information (U.S. Government Accountability Office, 2014).8
The IWG initially established interim SC-CO2 values using estimates obtained from the existing literature. These interim values were first used by the U.S. Department of Energy in an RIA for an energy efficiency standard for beverage vending machines in August 2009 (74 Federal Register 44914). The IWG continued working on a more in-depth process to estimate the SC-CO2. In February 2010 it published a set of SC-CO2 estimates
7 The Intergovernmental Panel on Climate Change Fifth Assessment Report provides a database summarizing academic studies on the estimates of the welfare impact of climate change from 1982 to 2012 (Arent et al., 2014).
for the years 2010 through 2050 and described the technical methodology for estimation in a Technical Support Document.9 The methodology used the three most widely cited integrated assessment models (IAMs) that are used in benefit-cost analysis of climate policy to produce estimates of the SC-CO2.10 This report refers to those models as SC-IAMs.
Four updates to the Technical Support Documents related to the SC-CO2 estimates have occurred since the 2010 release: two in 2013 and one each in 2015 and 2016. None of the updates changed the fundamental methodology used to construct the 2010 SC-CO2 estimates.
The technical methodology for constructing the official U.S. SC-CO2 estimates is discussed in detail in the IWG Technical Support Documents (Interagency Working Group on the Social Cost of Carbon, 2010, 2013a, 2013b, 2015a; Interagency Working Group on the Social Cost of Greenhouse Gases, 2016b). Three SC-IAMs were used: DICE (Dynamic Integrated Climate-Economy model), FUND (Framework for Uncertainty, Negotiation and Distribution model), and PAGE (Policy Analysis of the Greenhouse Effect model). Each models the relationship between CO2 emissions and their monetized climate impact. An SC-CO2 estimate is derived following the same causal chain for each of the SC-IAMs: a CO2 emissions pulse is introduced in a particular year, creating a trajectory of CO2 concentrations, temperature change, sea level rise, and climate damages.11 The difference between this damage trajectory and the refer-
9 The Technical Support Document was released as an appendix to rulemaking by the U.S. Department of Energy on small electric motors (Energy Conservation Program: Energy Conservation Standards for Small Electric Motors, 75 Fed. Reg. 10,874 [March 9, 2010]).
10 There are many IAMs in use in the climate change research community for multiple purposes. Generally, IAMs vary significantly in structure, geographic resolution, computational algorithm, and application. In comparison with most other IAMs, the three used by the IWG are specialized in their focus on modeling aggregate global climate damages using highly aggregated economic and climate system representations, referred to as “reduced form IAMs”: (for details, see Box 2-1 in Chapter 2). Although the three SC-IAMs were not developed solely with the purpose of estimating the SC-CO2, they were among the very few models that produced estimates of global net economic damages from CO2 emissions when the IWG was developing its methodology.
11 Damages from global climate change include, but are not limited to, changes in net agricultural productivity, changes in energy use, human health effects, ocean acidification, changes in extreme weather events, and property damages from increased flood risk. Due to the long-lived nature of warming from CO2 emissions, many of the damages from CO2 emissions today may affect economic outcomes for the next several centuries.
ence projection in each year is discounted to the year of the CO2 pulse using an annual discount rate.
The IWG retained most of the SC-IAMs developers’ default assumptions for the parameters and functional forms used in the models. Two key exceptions are that the IWG used a single probability distribution for the equilibrium climate sensitivity (ECS)12 parameter in all three models, as well as a common set of five future socioeconomic and emissions scenarios.13 In addition, three constant discount rates were used to compute the present value of damages from each SC-IAM.
The IWG methodology resulted in 45 sets of estimates (three IAMs, five socioeconomic-emissions scenarios, one ECS distribution, and three discount rates) for the SC-CO2 for a given year, with each set consisting of 10,000 estimates based on draws from the standardized ECS distribution,14 as well as distributions of parameters treated as uncertain in two of the models. For each discount rate, the IWG combined the sets across models and socioeconomic emissions scenarios and then selected four values to be presented in regulatory impact analyses: an average value for each of three discount rates, plus a fourth value, selected as the 95th percentile of estimates based on a 3 percent discount rate. The IWG interpreted the 95th percentile as representing higher-than-expected impacts from temperature changes in the tail of the SC-CO2 estimates: see Figure 1-1.15
The set of four estimates from the most recent results is shown in Table 1-1 for CO2 impulses every 10 years from 2010 to 2050, with interim years interpolated. Percentiles and summary statistics of these estimates are presented in the IWG Technical Support Documents.16
12 ECS measures the long-term response of global mean temperature to a fixed forcing, conventionally taken as an instantaneous doubling of CO2 concentrations from their preindustrial levels (for details, see Box 4-1 in Chapter 4).
13 The committee notes, however, that these scenarios were not fully standardized in implementation due to differences in the SC-IAMs. See Rose et al. (2014b) for details on how the IWG implemented the individual models for the estimates.
14 The IWG selected the Roe and Baker (2007) distribution for the ECS “based on a theoretical understanding of the response of the climate system to increased greenhouse gas concentrations” and that it “better reflects the IPCC judgment that ‘values substantially higher than 4.5°C still cannot be excluded’” (Interagency Working Group on the Social Cost of Carbon, 2010, pp. 13-14).
15 The 150,000 estimates for each discount rate (2%, 3%, and 5%) are calculated by running each of the three models 10,000 times with random draws from the ECS probability distribution and other model-specific uncertain parameters, for each of the five socioeconomic emissions scenarios (three models × 10,000 runs × five socioeconomic emissions scenarios = 150,000 estimates).
16 The full set of the most recent estimates can be found at https://obamawhitehouse.archives.gov/sites/default/files/omb/inforeg/august_2016_sc_ch4_sc_n2o_addendum_final_8_26_16.pdf [January 2017].
The IWG has previously indicated its support for regular updates to the SC-CO2 estimates (Working Group on the Social Cost of Carbon, 2010, p. 3): “[T]he interagency process is committed to updating these estimates as the science and economic understanding of climate change and its impacts on society improve over time.” In 2013, the IWG updated the SC-CO2 estimates using revised models with significant independent, model-specific updates that were made by the model developers themselves: see Table 1-2 for a summary of the model modifications.
TABLE 1-1 Social Cost of Carbon, 2010-2050 (in 2007 dollars per metric ton of CO2)
|Year||5% Average||3% Average||2.5% Average||High Impact (95th Pct at 3%)|
NOTE: See text for discussion.
SOURCE: Interagency Working Group on the Social Cost of Greenhouse Gases (2016b, Table ES-1).
Specifically, the IWG produced revised estimates twice in 2013 using the updated models: first in May, incorporating the revised models with IWG-specific implementation modifications,17 and then in November, making two minor corrections to the May calculations. The IWG has continued to use these versions of the models for subsequent estimates (Interagency Working Group on the Social Cost of Carbon, 2015a; Interagency Working Group on the Social Cost of Greenhouse Gases, 2016b).
These changes resulted in an increase in the SC-CO2 estimates reported in the 2010 Technical Support Document for the year 2020, which were previously reported as $7, $26, and $42 (in 2007 dollars), respectively, for the 5 percent, 3 percent, 2.5 percent discount rates and $81 for the 95th percentile at a 3 percent discount rate. The corresponding four updated SC-CO2 estimates from the May 2013 update for 2020 were $12, $43, $65, and $129 (in 2007 dollars).
The November 2013 updates incorporated two technical corrections to the FUND modeling—correcting the potential dry land loss and the ECS distribution specification. The resulting changes to the final SC-CO2 estimates were generally less than $1 from the May 2013 update.
The 2015 update (Interagency Working Group on the Social Cost of Carbon, 2015a, p. 21) reflected two corrections:
17 Specifically, the May 2013 analysis shifted from using PAGE 2002 to PAGE09 (by Chris Hope), from DICE 2007 to DICE 2010 (by William Nordhaus), and from FUND 3.5 to FUND 3.8 (by Richard Tol and David Anthoff). See Interagency Working Group on the Social Cost of Carbon (2013a, 2013b) and Rose et al. (2014b) for descriptions of model updates and IWG modifications.
TABLE 1-2 Summary of Model Modifications Associated with the 2013 Updates of Estimates of the Social Cost of Carbon
|DICE||Carbon cycle parameters—weaker ocean uptake|
|Sea level dynamics and valuation—explicit modeling|
|Sea level rise and land loss|
|Transient temperature response|
|Methane—account for additional radiative forcing effects|
|PAGE||Sea level rise|
|Revised damage function to account for saturation—modified|
|GDP loss function|
|Regional scaling factors|
|Probability of discontinuity|
|Change in land/ocean carbon uptake|
|Regional temperature change|
NOTES: DICE, Dynamic Integrated Climate-Economy model; FUND, Framework for Uncertainty, Negotiation and Distribution model; GDP, gross domestic product; PAGE, Policy Analysis of the Greenhouse Effect model.
SOURCE: Adapted from Rose et al. (2014b).
First, the DICE model had been run up to 2300 rather than through 2300, as was intended, thereby leaving out the [discounted] marginal damages for the last tear of the time horizon. Second, due to an indexing error, the results from the PAGE model were in 2008 U.S. dollars rather than 2007 U.S. dollars, as was intended.
Figure 1-2 illustrates the relative values of the SC-CO2 estimates from 2010 and 2015 for different years of CO2 emission.
There are significant challenges to estimating a dollar value for CO2 emissions that reflects all of the physical and economic impacts of climate change, and the federal government made a commitment to provide regular updates to the estimates as noted above. The IWG requested this Academies study to guide future revisions of the SC-CO2 in two important ways. First, it requested that this study provide government agencies that are part of the IWG with an assessment of the merits and challenges of a specific near-term update to the SC-CO2 and with recommendations for enhancing the qualitative treatment or characterization of uncertainties
associated with the current SC-CO2 estimates in their reports. The result of this request was the committee’s Phase 1 report (National Academies of Science, Engineering, and Medicine, 2016). The conclusions and recommendations from the Phase 1 report are summarized in the next section.
Second, the IWG requested that the committee consider the merits and challenges of a comprehensive update of the SC-CO2 to ensure that the estimates reflect the best available science. Specifically, it requested that the committee review the currently available science to determine its applicability for the choice of IAMs and damage functions and examine issues related to climate science modeling assumptions; socioeconomic and emissions scenarios; the presentation of uncertainty; and discounting. (The full statement of task is in Box 1-2, above.) The second phase of the study allows for broader consideration of the methodology used for estimating the SC-CO2.
However, the statement of task was limited in its scope. Specifically, the committee was not asked to formally review or critique the current approach to estimating the SC-CO2, though it did consider the current approach in making recommendations. Nor was the committee asked to consider alternatives to the use of SC-CO2 estimates as a regulatory mechanism. These topics, as well as others related to the U.S. government’s and other’s use of SC-CO2 estimates, lie outside of the scope of the committee’s work and this report.
The main body of this report addresses and recommends approaches that warrant consideration in future updates of the SC-CO2 estimates, as well as recommendations for research to advance the science in areas that are particularly useful for estimating the SC-CO2.
In the Phase 1 report, the committee recommended against a near-term update to the SCC18 estimates concluding that changing the ECS alone within the current SCC framework would not significantly improve the estimates. The committee also provided several suggestions about how to improve the communication of uncertainty in the IWG’s Technical Support Documents. The conclusions and recommendations, grouped by the tasks they address, are in Box 1-3. The Phase 1 report also suggested that the IWG consider adopting or developing a common climate module and outlined criteria that the module should satisfy (see Chapter 4 in this report).
In August 2016 the IWG released an updated Technical Support Document. The IWG stated that the release responded to the committee’s recommendations listed above for enhancing the presentation and improving the discussion of the uncertainty in the current estimates of SC-CO2 (Interagency Working Group on the Social Cost of Greenhouse Gases, 2016b). The values for the SC-CO2 estimates did not change from the 2015 release. However, the IWG provided additional material on the sources of uncertainty in the SC-IAMs in additional appendices (in response to Recommendation 2) and made the 150,000 SC-CO2 values underlying each official IWG estimate available for download from the OMB website instead of by request. A new section titled “Treatment of Uncertainty” was added, together with a discussion on the types of uncertainty that are and are not included in the estimation approach (Recommendation 3).
The Technical Support Document also included symmetric high and low values of uncertainty in the estimates and clearly separated the values
18 When referring to documents published prior to 2016, the earlier abbreviation for the social cost of carbon, SCC, is used.
by the discount rate, as shown by the bars below the graph in Figure 1-1 (above) (Recommendation 4). The IWG continues to emphasize the non-symmetric uncertainty in the estimates by including the 95th percentile values in the executive summary table (see Table 1-1, above) despite the committee’s Phase 1 recommendation to present symmetric high and low values from the frequency distribution of SCC estimates with equal prominence, conditional on each assumed discount rate (Recommendation 4; Box 1-3, above). Agencies continue to be directed to use these estimates, but are able to conduct sensitivity analysis if an agency determines it appropriate. Agencies are referred to OMB Circular A-4 for best practices in conducting uncertainty analysis in RIAs.
The IWG also released in August 2016 an addendum to the updated Technical Support Document on estimating the social costs of methane (CH4) and nitrous oxide (N2O) (Interagency Working Group on the Social Cost of Greenhouse Gases, 2016a), noting that the framework for the non-CO2 estimates is the same as that used for SC-CO2 estimation. This report does not review or assess these new estimates for CH4 and N2O.
This study was carried out by a committee of experts appointed by the president of the National Academy of Sciences. The committee consisted of 13 members, working with a technical consultant and study director. Committee expertise spans the issues relevant to the study task: integrated assessment modeling, statistical modeling, climate science, climate impacts, environmental economics, energy economics, decision science, public policy, and regulation. In selecting the committee, care was taken to ensure that the membership possesses the necessary balance between research and practice by including academic scientists and other experts. Committee members were chosen to have the relevant disciplinary expertise and to ensure there are no current connections that might constitute a conflict of interest with the Department of Energy, the Environmental Protection Agency, or other regulatory agency members of the IWG. Biographical sketches of the committee members and staff are provided in Appendix A.
To address the Phase 2 task, the committee held three open meetings to receive information from federal agency staff to understand its study charge and to gather information to explore its charge (see Appendix B). Closed sessions were held to refine and finalize the committee’s conclusions and recommendations. The project included two focused studies by outside experts to support the committee’s analyses: a study on global growth projections as applied to the SC-CO2 estimation problem (see
The report is organized in seven chapters, with the committee’s conclusions and recommendations included in the relevant chapters, and several appendices. Chapter 2 provides a high-level response to the statement of task and an overview of the framework the committee used. Chapters 3-6 provide specific details and recommendations on the implementation of both near-term and longer-term updates. Chapter 3 is focused on updates to socioeconomic and emissions projections; Chapter 4 considers updates to modeling of the Earth system, including temperature change, sea level rise, and ocean acidification; Chapter 5 explores updates to climate impacts and damage estimates; and Chapter 6 presents an updated approach to discounting future damages. Chapter 7 highlights research priorities in key areas that are needed to improve future updates to the SC-CO2 estimates by summarizing research conclusions found throughout the report.
The five substantive appendices provide further technical detail on specific subjects: expert elicitation (Appendix C), projections of global economic growth (Appendix D), calculation of ocean acidification (Appendix E), comparison of the climate components of the SC-IAMs with a simple Earth system model (Appendix F), and model-specific suggestions for near-term improvements to current SC-IAMs damage components (Appendix G).