Chapter

Chapter 4. The Social Cost of Carbon: Valuing Carbon Reductions in Policy Analysis*

Author(s):
Ruud Mooij, Michael Keen, and Ian Parry
Published Date:
September 2012
Share
  • ShareShare
Show Summary Details
Author(s)
Charles Griffiths, Elizabeth Kopits, Alex Marten, Chris Moore, Steve Newbold and Ann Wolverton 

Key Messages for Policymakers

  • Without action to control rising greenhouse gases (GHGs), scientists predict that climate change will continue over time, bringing higher temperatures, sea level rise, and the potential for abrupt changes in earth system processes, with likely negative impacts on agricultural yields, ecosystems, human health, and more.
  • These impacts are expected to vary widely over time and by geographic region, with developing countries likely to experience disproportionate damages due to limited adaptation opportunities and economic dependence on climate-sensitive sectors.
  • The social cost of carbon (SCC) is the discounted monetary value of the future climate change damages due to one additional metric ton of carbon dioxide (CO2) emissions.
  • The U.S. government recently developed a set of SCC estimates for use in benefit-cost analyses of new regulations that influence CO2 emissions—the central case estimate is $21 per tonne for emissions in the year 2010 (in 2007 U.S. dollars).
  • Other countries could use these SCC estimates in benefit-cost analyses or to help set the initial level of a domestic carbon pricing policy, if those countries accept the policy judgments and methodological assumptions underlying the estimates.
  • Other countries should consider developing their own SCC estimates when they wish to reflect fundamentally different assumptions, develop a long-term strategy to evaluate emission reductions, or evaluate the cost-effectiveness of policies designed to meet a given long-term target.
  • The SCC estimates should be updated over time to reflect changes in emissions, atmospheric concentrations, economic conditions, and advancements in scientific knowledge.

Greenhouse gas (GHG) emissions from human activities—mainly from the burning of fossil fuels, deforestation, and agricultural activities—are accumulating in the atmosphere and altering the earth’s climate and other natural systems.1 Between 1850 and 2005, the atmospheric concentration of carbon dioxide (CO2), the predominant anthropogenic GHG, increased from about 280 to 380 parts per million (ppm). Along with increased atmospheric accumulation of other GHGs, this has significantly contributed to an estimated increase in the global average annual surface temperature of approximately 0.8° C above preindustrial levels. Rising temperatures are also causing the level of the oceans to rise—on average by about 20 cm during the twentieth century. In the absence of serious policy action to abate GHG emissions, atmospheric CO2 concentrations are projected to continue rising, with the potential to increase the global average annual surface temperature by an additional 1.1° C to 6.4° C and the average sea level by an additional 20 to 60 cm by the end of this century. Even if atmospheric GHG concentrations were to be stabilized immediately at 2000 levels, scientists estimate that the average surface temperature would likely increase by an additional 0.3° C to 0.9° C by the end of the century due to a delayed response inherent in the climate system (see Chapter 3 for further discussion of emissions and climate trends). Temperature increases and changes in precipitation will not be uniformly distributed across the globe—the specific magnitude, direction, and spatial pattern of these changes are highly uncertain and are an area of ongoing scientific research.

In addition, the scientific literature has paid increasing attention to the likelihood and nature of potential “climate catastrophes”; that is, high-impact, low-probability earth system changes due to rising GHG concentrations. Possibilities include the collapse of the Greenland or West Antarctic ice sheets, a shutdown or change in the Atlantic Ocean circulation, substantially altered periodic weather patterns, large releases of additional GHGs from methane deposits, massive dieback of tropical or boreal forests, and cascading effects in marine food webs from ocean acidification. However, research quantifying the magnitude and timing of the physical—and especially the economic—impacts from many of these potential risks is still in its infancy.

Current and anticipated climate changes are expected to have a wide range of mostly negative impacts on economies and societies across the globe, including (but not limited to) the inundation of coastal areas, reduced agricultural yields, and increased frequency and severity of tropical storms, droughts, and other extreme weather events. Recent studies suggest that a 2.5° C to 3.0° C increase in the average annual surface temperature above preindustrial levels will lead to aggregate annual damages of between 0 and 2.5 percent of global gross domestic product (GDP). These aggregate figures mask considerable disparity in regional effects. One study estimated damages from a 2.5° C increase in global average temperature ranging from a positive 0.7 percent of GDP for the former Soviet Union to a negative 5 percent of GDP for South Asia and India (Nordhaus and Boyer, 2000). Likewise, a recent assessment found that a global sea level rise of 0.5 m to 2 m could displace 72 to 187 million people over the twenty-first century (assuming no adaptation), about 70 percent of which would be concentrated in east, southeast, and south asia (Nicholls and others, 2011). Developing countries will likely experience larger than average damages due to limited adaptation opportunities, as well as greater dependence of their economies on climate-sensitive sectors like agriculture, although nations’ vulnerability to climate change may diminish with economic growth over the coming centuries.

With the aid of integrated assessment models (IAMs) that combine simplified representations of the climate system, the global economy, and their interactions, analysts can evaluate the economic implications of GHG mitigation policies. In this chapter, we describe recent efforts by the U.S. government to estimate the social cost of carbon (SCC) to value the damages from small changes in CO2 emissions for use in benefit-cost analysis of policies that directly or indirectly reduce GHG emissions. We then discuss the potential applicability of these estimates to policy analysis in other countries and regions. We close by highlighting areas of future research and the need for frequent reassessment of the SCC in light of new information.

The Social Cost of Carbon

Defining the SCC

Comparing the benefits of policies that result in carbon reductions to their economic costs requires a monetized measure for the value of future climate change damages. The SCC is one such measure. It is the present value of future damages associated with an incremental increase (by convention, 1 metric tonne) in CO2 emissions in a particular year expressed in consumption equivalent terms. In theory, it is intended to be a comprehensive measure including, for example, damages from changes in agricultural productivity, human health risks, property damages from increased flood frequencies, and the loss of ecosystem services.

The SCC can be calculated from a global perspective, which incorporates damages to all countries caused by CO2 emissions, or from a domestic perspective, which incorporates only the damages experienced by a country’s own residents. The U.S. government interagency working group chose a global perspective to evaluate the SCC, mainly because climate change is a global externality (CO2 emissions rapidly become well mixed in the atmosphere and therefore contribute to damages around the world no matter their source). The use of a global SCC is consistent with the goal of achieving a globally economically efficient solution. If nations were instead to design their policies independently using domestic SCC estimates (thereby excluding benefits that accrue to nonresidents), abatement would be meaningfully less than the globally economically efficient level.

Integrated Assessment Models

Estimation of the economic impacts of CO2 emissions can be broken into four main steps: (1) projections of future GHG emissions, (2) the effects of past and future emissions on the climate system, (3) the impact of changes in climate on the physical and biological environment, and (4) the translation of these environmental impacts into economic damages, discounted back to the present. IAMs couple physical science and economic models to capture important interactions between these components. The IAMs most often used to estimate the SCC represent highly aggregated, reduced-form approaches. While other more comprehensive models, such as computable general equilibrium models, may better represent the complex interactions among sectors of the economy and trade flows among countries, they typically lack the links between physical impacts due to climate change and economic damages necessary for estimating the SCC.

Most of the published SCC estimates are derived from one of three IAMs: William Nordhaus’ Dynamic Integrated Climate Economy (DICE) model, Richard Tol’s Climate Framework for Uncertainty, Negotiation, and Distribution (FUND) model, and Chris Hope’s Policy Analysis for the Greenhouse Effect (PAGE) model. Each model takes a somewhat different approach to translate changes in climate variables, such as temperature and sea level, into economic damages. Some of these differences arise from the model developers’ choices about the level of aggregation across regions, the climate variables, and which damage categories are explicitly included in each model (see Box 4.1).

Box 4.1.Damages in the PAGE, DICE, and FUND Integrated Assessment Models

As summarized in Table 4.1, the three integrated assessment models (IAMs) used for estimating the social cost of carbon (SCC) vary in their treatment of damages. For instance, the Dynamic Integrated Climate Economy (DICE) model aggregates damages across sectors and regions into a single global damage function, while the Policy Analysis for the Greenhouse Effect (PAGE) model divides damages into economic and noneconomic categories. In contrast, the Climate Framework for Uncertainty, Negotiation, and Distribution (FUND) model separately calculates damages for 11 different market and nonmarket categories. DICE and PAGE also include categories that account for the possibility of large “catastrophic” damages at higher temperatures, while FUND does not. In DICE, parameters are handled deterministically and represented by fixed point estimates; in PAGE and FUND, most parameters are represented by probability distributions. Also, unlike PAGE, DICE and FUND treat GDP as endogenous so damages in early years reduce economic output in later years.

Table 4.1.Summary of Climate Change Impacts by Category
Global Damages at 2.5° C Above Preindustrial Levels (Percent of global output)
DICE 2007PAGE 2002FUND 3.5
Agriculture0.13Economic impacts0.36Agriculture and−0.90
Coastal0.32Noneconomic0.65forestry
Other market sectors0.05impactsCoastal0.02
Health0.10Hurricanes and0.01
Nonmarket amenities−0.29other storms
Human settlements and0.17Health0.07
ecosystemsWater resources0.16
Biodiversity Loss0.13
Cooling0.90
Heating−0.51
Subtotal:0.48Subtotal:1.01Subtotal:0.13
Catastrophes1.02Catastrophes0.43
Total1.50Total1.44Total0.13
Sources: Based on Hope (2006), Nordhaus and Boyer (2000), and Tol (2009).

Note: For general illustrative purposes only based on default assumptions in DICE 2007 (output-weighted damages) and deterministic runs of FUND 3.5 and PAGE 2002 using the mode values for all parameters. Damage categories are defined by model authors. Refer to documentation for each of the three models for more detailed information on how these are defined.

Sources: Based on Hope (2006), Nordhaus and Boyer (2000), and Tol (2009).

Note: For general illustrative purposes only based on default assumptions in DICE 2007 (output-weighted damages) and deterministic runs of FUND 3.5 and PAGE 2002 using the mode values for all parameters. Damage categories are defined by model authors. Refer to documentation for each of the three models for more detailed information on how these are defined.

The economic damage estimates vary considerably across models at both lower and higher global average temperature changes. This reflects differences in assumptions about the rate of technological change, the ability of human and natural systems to adapt to the effects of climate change, and the projected vulnerability of developing countries, among others. For instance, FUND projects that climate change is potentially beneficial for a 2.5° C increase due to effects on agriculture and forestry and decreased heating costs. PAGE assumes that impacts occur only above some “tolerable” temperature increase, defined as 2° C for developed countries and 0°C in developing countries. Beyond the tolerable level, developed countries are able to eliminate almost all economic impacts through various adaptation measures (e.g., altering crop varieties or planting dates, building sea walls), while developing countries can eventually eliminate 50 percent of economic impacts through adaptation. Adaptation to noneconomic impacts (e.g., biodiversity loss) is much more difficult for both regions. DICE does not include explicit representation of adaptation, though some forms of adaptation—especially in the agricultural sector—are included implicitly through the choice of studies used to calibrate the aggregate damage function.

Source: Interagency Working Group on Social Cost of Carbon (2010).

While IAMs offer useful guidance about the effects of climate change on human well-being, modeling the complex systems involved often requires assumptions that cannot easily be verified based on historical evidence. For this reason, IAM results should not be interpreted as precise predictions of far future outcomes. This is well understood and is frequently emphasized by the IAM developers themselves, and IAMs are regularly updated as modelers revisit key aspects of their framework (e.g., damage functions, assumptions about adaptation, the representation of natural systems to reflect the evolving scientific and economic research).

SCC Estimates Used in U.S. Regulatory Impact Analysis

In 2009–10, the U.S. government formed an interagency working group to develop a set of SCC estimates to be used by all executive branch agencies to value reductions in CO2 emissions in regulatory analyses. The purpose of this interagency working group was to improve the accuracy and consistency with which agencies value reductions in CO2 emissions. Prior to this effort, SCC estimates were used in some but not all regulatory analyses, and the values employed varied substantially among agencies.

The interagency working group used the DICE, PAGE, and FUND models to estimate SCC values. The climate system submodels and the functions that map climate change to economic damages were left unchanged, but a common set of assumptions for three key inputs was used across all the models—socioeconomic and emissions projections, equilibrium climate sensitivity, and discount rates.

The interagency working group selected five scenarios of GDP, population, and GHG emissions projections from the 2009 Stanford Energy Modeling Forum exercise (EMF-22).2 These scenarios spanned a range of emissions projections (including at least one case where the rest of the world takes significant action to reduce emissions) and plausible outcomes for future population and GDP. The motivation for this approach was to ensure that the GDP, population, and emission trajectories were internally consistent for each scenario considered. Across the five scenarios, atmospheric CO2 concentrations in 2100 ranged from about 450 to 890 ppm (or 550 to 1130 ppm in CO2 equivalent when other GHGs such as methane were included), the average percentage change in global per capita GDP ranged from 1.5 to 2.0 percent per year, and the percentage change in global population was 0.4 to 0.5 percent per year. Future projections of global GDP were based on combining regional GDPs using market exchange rates.3

In reduced-form IAMs, the speed and magnitude of temperature change for a given emissions projection are strongly influenced by the equilibrium climate sensitivity (ECS) parameter, which represents the long-term climate response to atmospheric conditions that are similar to those that would be associated with an atmospheric CO2 concentration of 550 ppm, ignoring all other relevant gases.4 To represent the uncertainty of the responsiveness of the climate system to changing atmospheric conditions, the interagency working group used a probability distribution to represent ECS in all three models. The probability distribution was calibrated to the Intergovernmental Panel on Climate Change consensus statement about this parameter by applying three constraints—a median equal to 3°C, two-thirds probability that the ECS lies between 2 and 4.5°C, and zero probability that it is less than 0°C or greater than 10°C.5

Discount Rates Selected

Since the damages from a tonne of CO2 emissions occur over many decades, the discount rate—which reflects the trade-off between present and future consumption—plays a critical role in estimating the SCC. For policies with both intragenerational and intergenerational effects, U.S. federal agencies traditionally employ constant real discount rates of 3 and 7 percent per year. However, discounting over very long time horizons raises exceedingly difficult questions of science, economics, philosophy, and law. Approaches for determining the discount rate for climate change analysis have been categorized as either “descriptive” or “prescriptive.”

The descriptive approach is based on observations of people’s actual behaviors, such as saving versus consumption decisions over time and allocations of investment among more and less risky assets. Advocates of this approach argue that because expenditures to mitigate GHGs are a form of investment, discount rates used to evaluate benefits from these expenditures should be based on market rates of return.

The prescriptive approach to discounting specifies a social welfare function that formalizes the normative judgments that the decision maker wants to incorporate into the policy evaluation; that is, how interpersonal comparisons of well-being should be made and how the well-being of future generations should be weighed against that of the present generation. Proponents of the prescriptive approach argue that various market imperfections (e.g., the absence of markets for very long-lived loans) make the market interest rate an unreliable measure of the appropriate trade-off between the consumption of present and future generations; instead, the discount rate should be specified partly based on ethical judgments about intergenerational equity. Often the rates recommended by the prescriptive approach are lower than those based on the descriptive approach.

The interagency working group drew on both approaches but relied primarily on the descriptive approach to inform the choice of discount rate. With recognition of its limitations, the interagency working group felt that this approach was the most defensible and transparent given its consistency with the standard principles of benefit-cost analysis. Regardless of the theoretical approach used to derive the appropriate discount rate(s), it is important to note the inherent conceptual and practical difficulties of adequately capturing consumption trade-offs over many decades or even centuries. In light of disagreement in the literature on the appropriate market interest rate to use in this context and uncertainty about how interest rates may change over time, the interagency working group used three constant discount rates—2.5, 3, and 5 percent per year—to span a plausible range.

Calculating the SCC

Four basic steps are required to calculate the SCC in a particular year t. First, each model is used to project paths of temperature change and aggregate consumption associated with the baseline path of emissions, GDP, and population. Second, each model is re-run with an additional unit of CO2 emissions in year t to determine the projection of temperature changes and aggregate consumption in all years beyond t along this perturbed path of emissions. Third, the marginal damages in each year are calculated as the difference between the aggregate consumption computed in steps 1 and 2. Finally, the resulting path of marginal damages is discounted and summed to calculate the present value of the marginal damages in year t.

The steps above were repeated in each model for multiple future years to 2050. Because the climate sensitivity parameter is modeled probabilistically and because PAGE and FUND incorporate uncertainty in other model parameters, the final output from each model run represents a distribution over the SCC in each year. The exercise produced 45 separate distributions of the SCC for a given year, based on the three models, three discount rates, and five socioeconomic scenarios considered. To produce a range of estimates that reflects this uncertainty but still emphasizes the central tendency, the distributions from each of the models and scenarios were equally weighted and combined to produce three separate probability distributions for the SCC in a given year, one for each assumed discount rate.

Four SCC estimates were selected from these three probability distributions to reflect the global damages caused by one tonne of CO2 emissions: $5, $21, $35, and $65 for 2010 emission reductions (in 2007 U.S. dollars). The first three estimates are based on the average SCC across the three models and five socioeconomic and emissions scenarios for the 5, 3, and 2.5 percent discount rates, respectively. The fourth value is the ninety-fifth percentile of the SCC distribution at a 3 percent discount rate and was chosen to represent potential higher-than-expected impacts from anthropogenic GHG emissions. Figure 4.1 illustrates where these values fall within the wider distribution of SCC values generated by the three IAMs at three different discount rates. Notice that the distribution is skewed toward high values of the SCC. The range of the distribution increases with lower values of the discount rate.

Figure 4.1.Distribution of 2010 Social Cost of Carbon Values at Each Discount Rate

Source: Authors’ calculations based on Interagency Working Group on Social Cost of Carbon (2010).

The SCC estimates also grow over time because future emissions are expected to produce larger incremental damages as the economy grows and physical and economic systems become more stressed in response to greater climatic change. These rates are determined endogenously by the models (see Table 4.2) and are dependent upon a number of assumptions including the socioeconomic and emissions scenario, model structure, parameter distributions, and the discount rate.

Table 4.2.Social Cost of Carbon, 2010–50 (In 2007 U.S. dollars per tonne of CO2)
Year of Emission ReductionDiscount Rate
5 Percent Average3 Percent Average2.5 Percent Average3 Percent 95th
20104.721.435.164.9
20155.723.838.472.8
20206.826.341.780.7
20309.732.850.0100.0
204012.739.258.4119.3
205015.744.965.0136.2
Annualized percent change in SCC, 2010-503.1%1.9%1.6%1.9%

Having four estimates of the SCC raises the possibility that economic analyses of a single policy conducted with different values will generate different qualitative results. So, how should a policymaker interpret the results in such a case? In the United States, policymakers are asked to consider all four estimates of the SCC when conducting benefit-cost analysis, although the average values discounted at 3 percent are treated as central estimates. This is useful for purposes of informing decision makers of the robustness of a policy prescription to a range of values. It is also important to note that economic efficiency is only one of possibly many criteria U.S. policymakers consider when evaluating environmental policies. Furthermore, when there is substantial uncertainty surrounding the economic analysis, this may affect how much weight it is given relative to other criteria.

Using the Social Cost of Carbon for Policy Analysis

The appropriate role of the SCC in policy analysis and the applicability of the U.S. government’s SCC estimates to analyses in other countries or regions depend on the nature of the policy in question, including the magnitude of the expected emission impact and timeframe of consideration. Because the SCC is the net present value of all future damages resulting from an additional tonne of CO2 in year t, the estimates are conditional on forecasts of emissions and socioeconomic conditions from year t onward. Actions taken by the United States, or by other countries, to reduce emissions may change the forecasts of future damages. If these changes are large enough, then the future path of the SCC itself also would change. For this reason, the U.S. government’s SCC estimates are most appropriate for analyzing policies that are expected to have a relatively small impact on global emissions and associated future climate conditions. To date, these values have been used to quantify the benefits of reducing CO2 emissions from U.S. federal regulations such as energy efficiency standards for appliances and CO2 tailpipe emission standards for light and medium heavy-duty vehicles. For long-term policies that have substantial impacts on global emissions, the appropriate SCC is one that accounts for the impact of the policy on forecasts of emissions and socioeconomic conditions.

Using the Social Cost of Carbon in Benefit-Cost Analysis

If policymakers in another country adopt the same set of policy judgments and methodological assumptions used by the U.S. government, then they can directly apply the SCC values described here to calculate monetized global benefits of CO2 reductions resulting from domestic policies. The only adjustments needed involve translating the SCC estimates into the currency of the country where the analysis is to be conducted to allow for direct comparison to the domestic costs and other monetized benefits. In a developing country with a substantial portion of its economy made up of nontraded sectors, the market exchange rate may not provide an accurate assessment of what actions are worth taking, particularly when the SCC is being compared to domestic costs.6 However, given that the SCC values estimated by the U.S. interagency working group are based on global GDP projections that use market exchange rates and the uncertainty involved in projecting purchasing power parity many years into the future, using market exchange rates is likely simpler and more transparent.

If policymakers in another country want to adopt other policy judgments or methodological assumptions, then the SCC estimates would need to be recalculated accordingly. Revised estimates could accommodate a variety of policy and ethical considerations, such as the use of a lower discount rate, used of a domestic SCC value, and equity weighting. We will now briefly discuss each consideration.

Discount Rates

While the interagency working group chose to use three different constant discount rates, other rates may also be consistent with what has been used in the literature. For instance, some decision makers may prefer to select a discount rate that reflects the prescriptive approach or a higher aversion to climate risks. Alternatively, they may want to use a nonconstant or declining rate to reflect greater uncertainty further into the future. If a different discount rate is desired to calculate the present value of avoided damages from one tonne of CO2 emissions, then an entirely new set of SCC values must be generated. In general, use of a lower discount rate will result in a higher average SCC value. Incorporating uncertainty in the discount rate is also expected to increase the average SCC value relative to the equivalent constant discount rate (see Box 4.2 for a more detailed discussion of discount rate uncertainty).

Box 4.2.Treating Uncertainty in the Discount Rate over Long Time Horizons

While the U.S. interagency working group used a range of constant discount rates to generate SCC estimates, there is empirical and theoretical support for using a schedule of discount rates that declines over time. A number of studies in this area have found that uncertainty about future discount rates can have a large effect on net present value. A main result from these studies is that if there is a persistent element to the uncertainty in the discount rate, the effective discount rate declines over time. Consequently, lower discount rates tend to dominate the present value calculation over the very long term.

The proper way to model discount rate uncertainty remains an active area of research. One approach is to employ a model of how long-term interest rates change over time to forecast future discount rates. This type of model incorporates some of the basic features of how interest rates move over time, and its parameters are estimated based on historical observations of long-term rates. Subsequent work on this topic uses more general models of interest rate dynamics to allow for better forecasts that account for the present level of interest rates and the persistence of shocks. A simplified alternative to formally modeling uncertainty in the discount rate is to use a schedule of discount rates as has been done in both the United Kingdom and France. In this case, the analyst would apply a higher discount rate over the first 40 to 50 years of the policy and a graduated schedule of lower discount rates further out in time.

Source: Interagency Working Group on Social Cost of Carbon (2010).

Domestic SCC Values

Likewise, if policymakers wish to limit the analysis to a comparison of domestic benefits and costs, then the SCC values would need to be adjusted to exclude all damages experienced by residents of other nations. It is worth noting, however, that if each nation used an SCC estimate that included only its own domestic damages to evaluate regulations, then the result would be a much lower level of abatement in each country. Furthermore, the global level of abatement that would be realized under this scenario could be achieved at a lower cost if all countries used a common SCC value (or if international trading for emission rights was allowed) since marginal abatement costs would then be equalized across regulated sources.

Equity Weighting

Policymakers also may want to conduct a global social welfare analysis (using an explicit social welfare function that weighs both efficiency and distributional concerns) rather than, or in addition to, a benefit-cost analysis (which typically addresses economic efficiency alone). The SCC estimates developed by the U.S. interagency working group reflect an explicit decision to focus only on economic efficiency, which counts the willingness-to-pay of all individuals who will be affected by climate change equally, no matter their country of residence or income. It is possible to incorporate equity weights that adjust the measure of economic damages for differences in incomes among the affected individuals, but the SCC would need to be recalculated. In such an analysis, the monetized costs of the policy also would need to be adjusted using the same set of equity weights. In Box 4.3, we provide a brief perspective on the use of the SCC in other countries.

Box 4.3.An Illustration of Use of the SCC for Decision Making in Other Countries

The United States is not the only country that has developed social cost of carbon values. The United Kingdom first recommended the use of the SCC for informing national policies on GHG emissions in 2002 and commissioned a series of reports to examine the issue. The most widely known of these reports is the Stern Review. The SCC values estimated by Stern differ from the values used by the U.S. government in a number of ways. For example, Stern uses a lower discount rate and includes equity weighting. Relying on input provided by the Stern Review, the UK government officially set a value for the SCC in 2007 for the express purpose of determining the most appropriate limit on CO2 emissions (referred to as a stabilization target when expressed in terms of the carbon concentration in the atmosphere). This value was set at about $50 per tonne of CO2 equivalent in terms of 2007 U.S. dollars (i.e., £25.5 per tonne), rising at a rate of 2 percent annually. Germany followed the United Kingdom’s lead, relying on the same SCC value to evaluate its own domestic carbon policies, but suggesting sensitivity analysis using, in U.S. dollar terms, $15 and $215 per tonne.

In 2009, the United Kingdom moved away from using an SCC approach. It stated two reasons for this: (1) The SCC requires assumptions about what other countries will do to reduce GHGs and (2) there is a large degree of uncertainty around SCC estimates. In its place, the UK government now uses a shadow price of carbon for policy evaluation. It has two different sets of values, one to assess policies that reduce traded carbon emissions under the European Union’s cap-and-trade policy and another to assess the cost-effectiveness of reducing GHGs in nontraded sectors of the economy.

Sources: DECC (2009); Umwelt Bundes Amt (2008).

SCC and Carbon Taxes

A country or group of countries may also be interested in using the SCC to help set the level of a carbon tax. However, it is important to note that the U.S. government’s SCC values were calculated along a business-as-usual emissions path, not a socially optimal one. Therefore, countries should avoid locking in a long-term tax policy using the U.S. government’s schedule of SCC estimates given in Table 4.2. Instead, the carbon tax in the near term should be set equal to the current best estimate of the global SCC and then adjusted over time to match an SCC that is reestimated as sources adopt new measures to reduce their emissions. Economic efficiency is achieved when the marginal abatement costs are equalized across all sources and are in turn equal to the current updated SCC.

In theory, a global carbon tax could be imposed at the outset without the need for adjusting the tax schedule over time, but only if the ultimate globally economically efficient level of emissions could be determined at the beginning. However, it should be noted that in the short run, the marginal benefit curve is relatively flat for CO2 emission reductions because of the time lag inherent in the climate system and the relatively flat relationship between damages and climate change at relatively small temperature changes. This means that even for a nonmarginal tax policy, the SCC will not differ much from the marginal policy path in the near term. Figure 4.2 illustrates this point by showing the time path of the SCC along two separate emissions trajectories using the DICE 2010 model. The first represents a business-as-usual scenario. The second represents a large policy change, specifically a 50 percent reduction in CO2 emissions each year relative to the business-as-usual path.

Figure 4.2.Social Cost of Carbon Values for Business-as-Usual and Nonmarginal Policy Emissions Paths

Source: Authors’ calculations based on the DICE model.

Note: This example uses a constant consumption discount rate of 3 percent per year, but all remaining parameters are based on the default values in DICE 2010.

Notice that the SCC values for the two scenarios are very close to each other in the early years of the forecast: The SCC on the policy path remains within 5 percent of the SCC on the business-as-usual path for at least the first 45 years, even though the emissions on the policy path are 50 percent lower than those on the business-as-usual path starting immediately after the first time period. This is an extreme example, but it effectively demonstrates that it would be reasonable to use the SCC estimated along a business-as-usual forecast as a guide for setting a domestic carbon tax in the near term, such as in the next 10 to 20 years.7

Social Cost of Carbon and Cost-Effectiveness

The SCC is not the appropriate measure for evaluating projects when the overall policy objective is to meet a predetermined emissions (or concentration or temperature) target at the lowest possible cost. If the environmental target is specified ex ante, then a measure of the benefits of abatement is not needed. Instead, a measure of the marginal cost of abatement—sometimes called the “shadow price of carbon”—can be used to evaluate the cost-effectiveness of the policy (see Box 4.3). The two measures will be equal only when the emissions target is set at the economically efficient level.

Caveats and Reassessing the Social Cost of Carbon in the Future

The SCC estimates developed by the U.S. federal government for use in a benefit-cost analysis are subject to several important caveats. They are based on a number of assumptions and modeling simplifications that are not readily verifiable in the real world. For instance, there are many differences across the IAMs in how damages are modeled, as well as the treatment of technological change, adaptation, and catastrophic damages. Gaps in the literature make modifying these aspects of the models challenging, which highlights the need for additional research. Other key areas for future research include improvements in how predicted physical impacts translate into economic damages for a wide range of market and nonmarket damage categories, better incorporation of sectoral and regional interactions, the treatment of the discount rate in regulatory analyses where costs and benefits are widely separated in time (including how to address uncertainty), and methods for estimating the marginal damages from non-CO2 GHG emissions.

In addition, the SCC estimates are based on a range of socioeconomic scenarios for how emissions will develop over time. As technologies develop, populations and economies grow, and countries formulate policies to reduce emissions, it is a virtual certainty that reality will diverge from what was assumed in the models to estimate the SCC. Thus, to keep pace with new research developments and to reflect diverging long-term trends, the SCC values should be reestimated on a regular basis.

Finally, it is worth emphasizing that attempts to estimate the SCC involve many unquantifiable uncertainties inherent to forecasting complex systems far into the future. This is due to the difficulty in accurately representing the many linkages between natural and economic systems as well as unforeseeable changes in population growth, technological progress, and regional economic growth. Therefore, the results from IAMs such as those discussed in this chapter should not be viewed as highly precise estimates of the SCC—or even precise estimates of the probability distribution over the SCC. Nevertheless, such models still provide invaluable information for policy analysis by quantifying what is currently known about potential climate damages and by providing a rigorous basis from which decision makers can assess the potential implications of the unavoidable modeling simplifications and omissions.

References and Suggested Reading

For background on climate trends and the science of global warming, see the following:

    Intergovernmental Panel on Climate Change, 2007, “Summary for Policymakers,” 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. (Cambridge, UK: Cambridge University Press).

    • Search Google Scholar
    • Export Citation

On sea level rise specifically, see the following:

    Nicholls, Robert J., NatashaMarinova, Jason A.Lowe, SallyBrown, PierVellinga, Diogode Gusmão, JochenHinkel, and Richard S. J.Tol, 2011, “Sea-Level Rise and Its Possible Impacts Given a ‘Beyond 4°C World’ in the Twenty-First Century,” Philosophical Transactions of the Royal SocietyA, Vol. 369, No. 1934, pp. 161–181. DOI: 10.1098/rsta.2010.0291.

    • Search Google Scholar
    • Export Citation

For some discussion on the valuation of climate change damages and the social cost of carbon, see the following:

    ICF International, 2011a, Improving the Assessment and Valuation of Climate Change Impacts for Policy and Regulatory Analysis: U.S. EPA/DOE Workshop Summary Report. Part I: Modeling Climate Change Impacts and Associated Economic Damages, http://yosemite.epa.gov/ee/epa/eerm.nsf/vwRepNumLookup/EE-0564?OpenDocument.

    • Search Google Scholar
    • Export Citation

    ICF International, 2011b, Improving the Assessment and Valuation of Climate Change Impacts for Policy and Regulatory Analysis: U.S. EPA/DOE Workshop Summary Report. Part II: Research on Climate Change Impacts and Associated Economic Damages, http://yosemite.epa.gov/ee/epa/eerm.nsf/vwRepNumLookup/EE-0566?OpenDocument.

    • Search Google Scholar
    • Export Citation

    Interagency Working Group on Social Cost of Carbon, 2010, Social Cost of Carbon for Regulatory Impact Analysis under Executive Order 12866, February, http://www.whitehouse.gov/sites/default/files/omb/inforeg/for-agencies/Social-Cost-of-Carbon-for-RIA.pdf.

    • Search Google Scholar
    • Export Citation

    National Research Council, 2009, Hidden Costs of Energy: Unpriced Consequences of Energy Production and Use (Washington: National Academies Press).

    • Search Google Scholar
    • Export Citation

For more detail on the integrated assessment models used in the U.S. interagency working group report on social cost of carbon, see the following:

    Hope, Chris, 2006, “The Marginal Impact of CO2 from PAGE 2002: An Integrated Assessment Model Incorporating the IPCC’s Five Reasons for Concern,” The Integrated Assessment Journal, Vol. 6, No. 1, pp. 19–56.

    • Search Google Scholar
    • Export Citation

    Nordhaus, William, and JosephBoyer, 2000, Warming the World: Economic Models of Global Warming (Cambridge, Massachusetts: MIT Press).

    Tol, Richard, 2009, “An Analysis of Mitigation as a Response to Climate Change” (Copenhagen: Copenhagen Consensus on Climate).

For perspectives on discounting climate damages, see the following:

    Portney, Paul, and JohnWeyant, eds., 1999, Discounting and Intergenerational Equity (Washington: Resources for the Future Press).

For other country perspectives on the SCC, see the following:

    DECC, 2009, Carbon Valuation in U.K. Policy Appraisal: A Revised Approach (London: Department of Energy and Climate Change).

    Umwelt Bundes Amt, 2008, Economic Valuation of Environmental Damage: Methodological Convention for Estimates of Environmental Externalities (Dessau-Rosslau, Germany: Federal Environment Agency).

    • Search Google Scholar
    • Export Citation
*

The views expressed in this chapter are those of the authors and do not necessarily represent those of the U.S. Environmental Protection Agency. We thank Joe Aldy, Terry Dinan, Robert Mendelsohn, Michael Keen, Ian Parry, and Tom Tietenberg for helpful comments and suggestions.

1

Fossil fuel combustion contributes about 26 Gt of CO2 a year, and land-use changes contribute another 6 Gt. Agricultural activities are the primary source of other potent GHGs such as methane and nitrous oxide. Figures cited in this chapter are from the Intergovernmental Panel on Climate Change’s Working Group I, “Summary for Policymakers” (IPCC, 2007).

2

The socioeconomic scenarios are available at Stanford’s Energy Modeling Forum, a group of well-regarded energy modeling teams from Asia, Australia, Europe, and North America. See: http://emf.stanford.edu/research/emf22/.

3

While the EMF-22 models use market exchange rates (MER) to calculate global GDP, it is also possible to use purchasing power parity. Purchasing power parity takes into account that some countries consume very different baskets of goods, including domestically produced nontradable goods. MERs tend to make low-income countries appear poorer than they actually are. Because many models assume convergence in per capita income over time, use of MER-adjusted GDP gives rise to projections of higher economic growth in low-income countries. There is an ongoing debate about how much this will affect estimated climate impacts. Critics of the use of MER argue that it leads to overstated economic growth and hence significant upward bias in projections of GHG emissions and unrealistically high future temperatures. Others argue that convergence of the emissions-intensity gap across countries at least partially offsets the overstated income gap so that differences in exchange rates have less of an effect on emissions. Nordhaus argues that the ideal approach is to use purchasing power parity but recognizes the practical data limitations of doing so over long time periods, although he also notes that exchange rate conversion issues are probably far less important than uncertainties about population or technological change.

4

Specifically, the ECS represents the increase in the annual global-average surface temperature from a sustained doubling of atmospheric CO2 relative to preindustrial levels.

5

The truncation at 10° C is reasonable considering the very long time lags associated with such high climate sensitivity values (i.e., such high temperature outcomes could only occur well beyond the relevant timeframe for policy analysis using the discount rates employed by the interagency working group).

6

This is not the case if a country uses the SCC values to approximate the level of a “global” carbon price. In this case, the ultimate objective of a social planner would be to maximize global benefits minus global costs. Any payments made across countries for emission reductions would occur at the market exchange rate.

7

Also note that in this example, the SCC increases when emissions are reduced. One reason this occurs is that in the DICE model, climate damages are represented as the loss of a fraction of gross economic output in each period, where the fraction of output lost depends only on the temperature anomaly in the period. Economic output net of climate damages is allocated to consumption and investment, so reduced damages in early periods can lead to increased economic output, and therefore increased absolute damages, in later periods.

    Other Resources Citing This Publication