A recent analysis of the social cost of carbon—the cost to society of an additional ton of carbon dioxide emissions—ultimately supports a higher estimate of this metric.
On Inauguration Day, the Trump administration issued its Unleashing American Energy executive order that, among other things, directed the Environmental Protection Agency (EPA) to consider eliminating the use of the social cost of carbon (SCC) by the federal government. Just days later, researchers at the Heritage Foundation released a white paper that the authors claim justifies the elimination of the federal SCC, based on their assessment of the sensitivity of SCC values estimated by Resources for the Future (RFF) and the University of California, Berkeley, through the RFF-Berkeley Greenhouse Gas Impact Value Estimator (GIVE) model.
The Heritage Foundation paper asserts that uncertainties underlying the estimates from the GIVE model justify valuing damages from climate change at precisely zero. Such an approach of ignoring uncertainty, rather than systematically accounting for it, would be arbitrary, lacks scientific basis, and has been rejected previously on those grounds by the courts. Further, the technical analysis of the Heritage Foundation does not support their conclusion; rather, it demonstrates that, for the most reasonable deviations from the state of the science, the SCC is still in the range of $100–$200 per ton of carbon dioxide, which supports increasing the SCC well above the levels used by the Obama and first Trump administrations.
The GIVE model at the heart of the Heritage Foundation’s critique was developed under RFF’s SCC Initiative, an independent collaboration between RFF and UC Berkeley, along with researchers at eight other leading academic institutions. The multidisciplinary research driven by the initiative, which formed the basis for the GIVE model, began in 2017 and yielded multiple peer-reviewed papers (such as Müller, Stock, and Watson 2022; Newell, Pizer, and Prest 2022; Rennert and coauthors 2021; Rennert and coauthors 2022; and Raftery and Ševčíková 2023). In 2022, our work culminated in a major study in the scientific journal Nature and yielded a central SCC estimate of $185 per ton of carbon dioxide in 2020 dollars for the benchmark 2020 emissions year. GIVE later was used as an input to EPA’s comprehensive SCC modeling update, which itself was subject to public comment and comprehensive scientific peer review by an independent panel in 2023.
Central to the SCC Initiative’s work is the recognition—highlighted by the widely accepted recommendations from the National Academies of Sciences, Engineering, and Medicine—that many of the factors relevant to the impacts of climate change are uncertain, and that accounting for such uncertainties merits serious and systematic attention. Our work leverages the state of the art from the scientific literature on uncertainties in future population growth, economic growth, greenhouse gas emissions, the climate system, the impacts of climate change on society and the economy, and discount rates—and how those factors interact. Our papers fully and explicitly quantify myriad uncertainties, in terms of both best estimates and ranges, which is critical for allowing the resulting SCC estimates to properly value risk in economic terms. In other words, our analysis is not based on singular assumptions about important variables, but rather a comprehensive range of factors.
The source code for the GIVE model is freely available to promote transparency and ongoing improvement as the state of the science evolves. The Heritage Foundation’s new white paper taps that capability, replicating the results of our Nature paper (which also are available through our nifty SCC Data Explorer) and assessing the robustness of our conclusions to three changes in modeling approaches, each of which deviate from the state of the science and the recommendations of the National Academies: (1) reverting to outdated approaches to discounting; (2) dropping most of the model years; and (3) replacing an important climate parameter with alternative values. While these “change-a-variable” analyses can be informative diagnostics, they are no substitute for the more systematic approaches to uncertainty that involve Monte Carlo analysis and explicit risk valuation.
Credit: Chip Somodevilla / Getty Images
In contrast to the concluding recommendations of the Heritage Foundation, the technical analysis of their report in fact reinforces the robustness of the SCC estimates from the GIVE model. When appropriate discount rates are incorporated that are in line with the consensus of the economic literature and appropriately account for risk, the SCC is well above the Obama-era estimate of $50 per ton of carbon dioxide and typically falls in the $100–$200 range (sometimes higher). To give a few examples of the Heritage Foundation’s SCC estimates, using a higher 2.5 percent discount rate yields an SCC of $119 per ton, dropping the entire second half of the years modeled yields $109, and ignoring high-impact/low-probability scenarios by considering only the median of the distribution yields $155. None of those changes reflect best practices in economic analysis (for example, we have long known from decision theory that the rational approach to decisionmaking in the presence of uncertainty is through expected values—that is, through means, not medians); nonetheless, the analysis by the Heritage Foundation still produces relatively large SCC estimates.
Only when using inappropriately high discount rates of 5 percent or more (levels that are out of line with the economic literature and market interest rates) do estimates from the Heritage Foundation fall below the Obama-era $50 value. The GIVE model’s central estimate of $185 instead is based on a central 2 percent discount rate, which reflects the consensus of the economic literature and longstanding downward trends in market interest rates. Recent rate hikes by the Federal Reserve have reinforced the case for the central 2 percent discount rate, as real Treasury yields have held steady over the past several years at their long-run average of about 2 percent.
The Heritage Foundation also considers a highly technical aspect of discounting—whether to use a “deterministic” discount rate or a stochastic “Ramsey” rate. While the deterministic rate ignores risk, the Ramsey rate properly prices it. We previously demonstrated the importance of the Ramsey approach, a result that is underscored and replicated by the results produced in the Heritage Foundation analysis: if our GIVE analysis had used the inappropriate deterministic rates, the SCC estimates would be much higher and overstated, potentially by a large amount.
Finally, the Heritage Foundation considers the effects of two alternative estimates of the Earth’s “climate sensitivity”—that is, the degree to which carbon dioxide emissions drive global temperature rise. In GIVE, we use the range of estimates of this climate sensitivity that is embraced by the Intergovernmental Panel on Climate Change, which reflects the broad state of the science across many studies. Heritage instead tests two alternatives from individual papers that were published in 2017 and 2022, which yield SCC estimates of $68 and $164, respectively, again both above the $50 Obama-era benchmark. The Heritage Foundation also considers the probability of a “negative” SCC stemming from the small probability of climate change yielding net benefits to society; for example, from increased crop yields in historically frigid regions of the world, such as Russia. GIVE natively includes this possibility but nonetheless finds a more than 99 percent probability that the costs of climate change outweigh any benefits. In the Heritage Foundation’s two alternative approaches to modeling the effect of carbon dioxide on the climate, these probabilities of costs outweighing benefits become 95–98 percent and 98–99 percent. In all these cases, the degree of confidence that climate change is costly to society exceeds the standard 95 percent confidence test routinely used in scientific research.
It should be encouraging that the analysis from the Heritage Foundation confirms that the upward revision to the SCC is appropriate and robust, even against changes outside the current state of the science. A surprising note, then, is what follows the long technical portion of the Heritage Foundation’s white paper: a section on policy recommendations that diverges so starkly with the analysis that precedes it. Citing uncertainty in their estimate, they call for the Trump administration to eliminate the SCC and endorse a new law to prohibit its use in the future. This proposed omission of the SCC would replace an uncertain range of the SCC with an overly precise estimate of exactly $0. Not only would their suggested approach understate the level of—and uncertainty in—the SCC, but it also is contrary to federal case law. In Center for Biological Diversity v. National Highway Traffic Safety Administration (2008), federal courts ruled that, although SCC estimates may be uncertain (as is the case for all estimations), “the value of carbon emissions reduction is certainly not zero.”
Beyond the legal issues, such an approach to banning the SCC would be unscientific. Even if the analysis from the Heritage Foundation had found higher uncertainty than it did, the appropriate response to uncertainty is not to throw up one’s hands, but rather to use appropriate tools to systematically assess and quantify the uncertainty, as the GIVE model is designed explicitly to do. Just as homeowners value insurance that protects them against risk exposure to highly uncertain events like fires, the rational response to uncertainty is to take it seriously and value it accordingly.
