Resources for the Future scholars lay out their new calculation of the social cost of carbon—an estimate of the economic damages that result from an incremental ton of carbon dioxide emitted into the atmosphere—which can help assess the benefits and costs of US policy actions. Their new value of the social cost of carbon is substantially higher than the previous values used by the US government.
Setting the Scene
Since 2017, Resources for the Future (RFF) has been working toward updating the scientific basis that underlies the social cost of carbon (SCC). The SCC is an estimate of the economic damages, in dollars, resulting from the addition of an incremental ton of carbon dioxide (CO₂) into Earth’s atmosphere. The value has been used widely to quantify the economic benefits of policies that reduce greenhouse gas emissions, including vehicle fuel economy standards, power plant regulations, and rules that reduce emissions from oil and gas infrastructure.
As part of these efforts, RFF’s Social Cost of Carbon Initiative assembled a large group of multidisciplinary researchers across many institutions to update the science that underlies the SCC in a manner fully responsive to a series of recommendations from a landmark 2017 report published by the National Academies of Sciences, Engineering, and Medicine (NASEM). That work has proven timely, given President Joe Biden’s January 2021 executive order, which instructs his administration to update the official value for the SCC so that it takes into consideration these NASEM recommendations and the recent and ongoing scientific progress that the NASEM guidance has steered.
Accordingly, the federal government has been working to update the SCC over the past year, and our extensive work on this issue at RFF is perfectly positioned to contribute to the necessary science. In an article in the Fall 2021 issue of Resources, we discussed the improvements that we made to three key aspects of SCC estimation: socioeconomic and emissions projections, climate modeling, and economic discounting. Here, we discuss our progress since then, which led to a comprehensive new estimate of the SCC that we published recently in the leading peer-reviewed scientific journal Nature.
Our new paper delivers an updated SCC that is fully responsive to the NASEM near-term recommendations for improving the scientific basis of the SCC. Before we dive into the details below, the key takeaway is that our comprehensive update delivers a central estimate of the SCC of $185 per ton of CO₂, which is 3.6 times larger than the US government’s current interim estimate of $51 per ton. This updated estimate implies that the benefits of climate policies are much larger than what the US government historically has concluded.
Incorporating New Science on the Impacts of Climate Change
Building on the work we discussed in the fall 2021 issue of Resources, our latest research in the new Nature article concludes our comprehensive update of the SCC by completing our work on the last part—the “damage functions,” which translate changes in global temperatures to dollar-value impacts for specific concerns, like human health, agriculture, and sea level rise. This last step completes the scientific updates to these four elements of SCC estimation as recommended by NASEM. The result is a complete, integrated climate-economic model for SCC estimation, which we call the “Greenhouse Gas Impact Value Estimator” (GIVE).
We have discussed the first three parts of the GIVE model at length in our previous articles; here, we dive into more detail on the updates to the damage functions. The damage functions are key inputs for understanding the economic consequences of greenhouse gas emissions, but many previously used damage functions are dated and do not adequately reflect recent research that estimates climate impacts.
To bring things up to date, we reviewed the scientific literature on climate damages that have been published recently in peer-reviewed journals across multiple disciplines. Prior studies emphasize the following four sectors, around which we built the damage components of the GIVE model: human mortality related to heat, agricultural productivity, energy expenditures for heating and cooling buildings, and the coastal impacts of rising sea levels.
Rising temperatures affect human mortality in many ways; for instance, by increasing the prevalence of and worsening the prognoses for cardiovascular and respiratory ailments, along with increasing the spread of some pathogenic diseases. These effects differ in severity among different regions of the globe. In a new meta-analysis published this year by Kevin R. Cromar and colleagues, a panel of 26 health experts combine results from numerous studies to characterize the effect of increased temperature on mortality from all causes for each of 10 regions around the world. We incorporate these results in our model by estimating the excess heat-driven deaths for each country for each future year, using the regionally disaggregated estimates produced in the study. We also measure uncertainty in the estimates by using the corresponding standard errors from the meta-analysis. All the estimates are monetized to calculate undiscounted damages for each year and country using the value of a statistical life, adjusted to reflect a given country’s projected GDP per capita.
Rising temperatures also will affect the suitability of different agricultural products in different regions of the world. A 2017 study from Frances C. Moore and colleagues carries out a meta-analysis that quantifies the effects of temperature change on the crop yields of four of the world’s most common crops: maize (corn), rice, wheat, and soybeans. The authors find that, while a few regions may benefit, the vast majority of regions are likely to suffer reduced crop yields and associated economic impacts due to the effects of climate change. This result is based on models that show how changes in crop yields affect regional production, trade flows, and consumption for a wide range of goods, while also accounting for adaptations that might be implemented in agriculture as a result of climate change, such as crop switching. Ultimately, the study presents three damage functions that represent low, central, and high estimates, each disaggregated across 16 globally comprehensive regions, which we incorporate into GIVE with three sets of coefficients that represent uncertainty in our modeling.
Heating and Cooling Costs
Due to both climate change and population growth, an increasing number of people are projected to live in hot places, where economic development will continue to increase access to air conditioning. As a result, a hotter climate is expected to increase the cost of cooling, while at the same time reducing the cost of heating buildings. A 2018 study by Leon Clarke and coauthors uses the well-established Global Change Analysis Model (GCAM), an integrated assessment model that estimates the net effects of increased cooling costs and decreased heating costs amid future rising global temperatures and changing socioeconomic patterns. The authors find that, on net, continued future warming is likely to lead to increased building energy expenditures, but that these effects are relatively modest—on the order of 0.1 percent of global GDP for a 2oC rise in global temperatures. They also find substantial variation in this effect across countries, however. For example, continued warming would lead to large increases in overall energy costs in regions that already are hot, like the Middle East, Africa, and Southeast Asia—but also energy savings in frigid countries like Russia. We incorporate their results into our model by using their estimated monetary effects per degree of warming, accounting for variation in the impact of warming across the 12 regions in their model.
Sea Level Rise
Rising sea levels associated with climate change are expected to harm coastal properties over time. Yet, communities can adapt to rising sea levels with responses such as relocating structures inland (“retreat”) or constructing physical infrastructure like seawalls and dikes to prevent flooding (“protect”). Nonetheless, these adaptation measures are themselves imperfect and often costly. A 2016 study by Delavane B. Diaz presents a Coastal Impact and Adaptation Model that estimates, for more than 12,000 individual coastal segments around the world, the costs of rising sea levels in the presence of these adaptation options. We incorporate this model directly into GIVE, coupled with a model of sea level rise built by Tony E. Wong and colleagues, allowing for an internally consistent estimate of the damages from sea level rise and accounting for adaptation and uncertainty about the magnitude of future sea level rise.
The Product: A New Social Cost of Carbon
The results of this multi-year effort are shown in Figure 1, which depicts uncertainty distributions of our SCC estimates, in terms of dollars per ton of CO₂, under two alternative approaches to discount rates. Discount rates translate the value of future impacts to equivalent values experienced today, reflecting that people generally tend to value benefits more when those benefits are received sooner rather than later. For example, a person who would demand a guaranteed $1.03 next year in exchange for $1.00 today is applying a discount rate of 3 percent. The discount rate can have major implications for how we evaluate impacts that occur far in the future.
Figure 1. Social Cost of Carbon Values, with Uncertainty Distribution
The figure illustrates two key points. First, while the SCC often is referred to as a single number, it is important to note that a distribution of SCC values in fact reflects uncertainty about the inputs into its estimation. These inputs include the range of potential future economic growth, baseline emissions, and climate dynamics, as well as the economic impacts of rising temperatures embedded in the damage functions. As is standard in scientific analysis, we account for these uncertainties in key inputs by using Monte Carlo simulations, in which the SCC is estimated many thousands of times using different inputs within ranges that align with evidence about the estimated probabilities of those inputs. These simulations result in many thousands of SCC estimates, and the smoothed curves in the figure reflect their frequency distributions. These distributions are typically summarized by their average, or “expected,” values, which are indicated at the top of the vertical dashed lines.
Second, as is well known, the SCC depends strongly on the discount rate. Historically, the US government has used a central discount rate of 3 percent for the SCC. (As explained in a previous Resources article, we use growth-linked discounting that’s consistent with the indicated discount rates in the near term.) Using a 3 percent discount rate, we find an expected value for the SCC of $80 per ton of CO₂, which is about 60 percent higher than the interim value of $51 per ton that was issued by the US federal government in early 2021, which also was based on a 3 percent discount rate. But that 3 percent discount rate derives from an analysis of market interest rates that was conducted nearly two decades ago, whereas interest rates have fallen significantly since then. Recent research has consistently documented persistent declines in long-run interest rates, suggesting that lower discount rates are appropriate, particularly for long-lived impacts like climate change.
Reasonable discount rates suggested by the economics literature are typically around 2 percent, which we adopt as our preferred value for the SCC calculation. As mentioned previously and as seen in Figure 1, this preferred estimate of the average SCC is $185 per ton of CO₂, which is much higher than the value of $80 per ton under a 3 percent discount rate because the long-lived impacts of CO₂ emissions are discounted less.
Understanding the Change
Our preferred estimate of the SCC of $185 per ton of CO₂ is much higher than the $51-per-ton interim value that was issued by the Biden administration in 2021, and which simply adjusted the previous government modeling results for inflation. As the 2017 NASEM report explains, that modeling does not reflect the recent scientific developments that we have accounted for here. Because we have updated all the inputs that underlie the SCC modeling—the socioeconomic and emissions projections, climate model, discounting approach, and damage functions—the $134-per-ton total change between the $185-per-ton and $51-per-ton estimates reflects the combination of all those changes.
To better illustrate how each change affects the estimate of the SCC, Table 1 decomposes the changes relative to the SCC value resulting from the widely used Dynamic Integrated Climate Economy (DICE) model developed by Nobel Laureate William Nordhaus. As seen in the first row of the table, the 2016 version of the DICE model produces an SCC value of $44 per ton (which is similar to the Biden administration’s interim estimate of $51 per ton, which is based in part on the DICE model). In the second row, we retain the DICE damage function but implement the climate model, socioeconomic and emissions projections, and discounting procedure from GIVE, using a 3 percent near-term discount rate. The combination of these changes increases the average SCC from $44 to $59 per ton—an increase of $15 per ton (about 11 percent of the total change). The third row shows that updating the damage functions discussed above increases the average SCC from $59 to $80, a further $21 per ton increase (or about 15 percent of the total change). Finally, changing the discount rate from 3 percent to 2 percent increases the average SCC from $80 to $185—or $105 per ton (and about 74 percent of the total change).
Table 1. Average Values for the Social Cost of Carbon Vary Among Different Models and Inputs
This decomposition illustrates that the change in the discount rate is driving the majority of the increase in the SCC. This result is expected and is consistent with previous research. Nonetheless, even absent any change to the discount rate, our work shows a substantial increase in the average SCC to $80 per ton, due to the updated socioeconomic projections, improved climate model, and damage functions.
Implications for Policy
The SCC is an important input for many types of analysis that help inform policy decisions—including regulatory actions by the US federal government, such as emissions standards for power plants and vehicles, appliance efficiency standards, procurement decisions, and environmental impact analyses. As such, it is vital to ensure that the SCC values used are based on the latest science and calculated in a fully transparent and reproducible process.
Our research finding—that the value of the SCC is substantially higher than the previous values used by the US government—implies that, based on the latest science and economic conditions, policies to reduce greenhouse gas emissions provide significantly greater benefits than have been accounted for previously. Though most regulatory and policy decisions are not based solely on benefit-cost analysis, higher values of the SCC in general support more stringent policies while retaining positive net benefits.
The scientific process by which our SCC estimates have been generated—which incorporates the latest peer-reviewed research and involves making our source code freely available to download and run—is equally consequential for policymakers and public confidence in the updated estimates. In our work, we adhere to the recommendations of the NASEM report and assemble the GIVE model from scientific components, each of which has undergone independent scientific review. Additionally, the GIVE model itself has been subject to rigorous scientific review. We have designed our research approach to promote transparency and quantification of uncertainty in the final estimates, and these methods align with the principles laid out for the Interagency Working Group’s update of the SCC estimates under Executive Order 13990 in early 2021. Independent, peer-reviewed research is vital for informing public policy on issues that are as critically important as climate change.