Last Friday, the White House blog announced a “return to science.” How so? Federal agencies will henceforth be guided by “evidence-based estimates of the benefits of reducing climate pollution.” Meaning what, exactly? Agencies’ benefit-cost calculations must now incorporate the Obama administration’s social cost of carbon (SCC) estimates, adjusted for inflation. Some would see this development as a return to flimflam. In a more professorial vein, University of Colorado polymath Roger Pielke, Jr. reports that “The Biden Administration Just Failed Its First Scientific Integrity Test.”
The social cost of carbon is an estimate in dollars of the cumulative long-term damage caused by one ton of carbon dioxide (CO2) emitted in a specific year. That number also represents an estimate of the benefit of avoiding or reducing one ton of CO2 emissions.
The Obama administration in 2009 established an Interagency Working Group (IWG) on the Social Cost of Carbon to estimate SCC values from 2010 to 2050. It produced SCC reports in 2010, 2013, and 2016 plus an addendum on the social cost of methane and nitrous oxide—greenhouse gases (GHGs) that are more potent than CO2 but emitted in lesser quantities.
The IWG estimated the 2020 SCC at $26 per ton in its 2010 report and at $42 in its 2016 report. The Trump administration, arguing that benefit-cost assessments should compare the domestic costs of CO2 reductions to the domestic benefits, not to the much larger global benefits, lowered the SCC to $7 per ton. The Trump administration’s $7 domestic SCC figure is based on (a) the U.S. economy’s above-average resilience to climate variability, especially compared to developing country economies, and (b) the Obama IWG’s estimate that the U.S. domestic benefit of reducing one ton of CO2 emissions is 7 to 10 percent of the global average value (p. 11). That, of course, did not stop climate campaigners from decrying the Trump administration’s apples-to-apples comparison as anti-science.
President Trump disbanded the IWG in March 2017. President Biden reestablished it on January 21. The IWG’s new interim report, published on February 26, estimates the 2020 SCC at $51 per ton (the central estimate, discounted at 3 percent in the figure below). The IWG says it will soon request public comment on the interim estimates and publish final updated SCC estimates in January 2022. The final numbers will undoubtedly be bigger, since the IWG believes the interim values “likely underestimate societal damages from GHG emissions.” The final report is also expected to use discount rates lower than 2.5 percent, which will further increase SCC values.
The IWG uses three computer programs, abbreviated DICE, FUND, and PAGE, to estimate the social cost of carbon. The programs are called integrated assessment models (IAMs) because they combine a climate model, which estimates the physical impacts of CO2 emissions, with an economic model, which estimates the dollar value of climate change effects on agricultural productivity, property values, and other economic variables. In federal agency analyses, the cumulative damage of an incremental ton of CO2 emissions is estimated from the year of the emission’s release until 2300.
SCC analysis looks very scientific to the uninitiated, but it is a highly speculative and subjective exercise. In a recent op-ed, Heritage Foundation statistician Kevin Dayaratna argues that SCC estimates are too dependent on alternative assumptions and too easily manipulated for partisan ends to guide policy making.
SCC estimates are profoundly influenced by the discount rates chosen to calculate the present value of future emissions and reductions, variable assumptions about climate sensitivity (how reactive the climate system is to a doubling of CO2 concentration), and diverse estimates of how much damage results from a given increase in global average temperature.
A critical factor influencing the “damage function” in IAMs is whether the model projects strong, weak, or no agricultural benefits from rising carbon dioxide concentrations (a phenomenon known as atmospheric CO2 fertilization).
Other things being equal, lower discount rates produce higher SCC values, higher climate sensitivity produces higher SCC values, and stronger CO2 fertilization produces lower SCC values.
Although literally hundreds of laboratory and field studies document the substantial increases in crop photosynthesis, water-use efficiency, and yields from rising CO2 concentration, DICE and PAGE—two of the three models used by the IWG—include no CO2 fertilization benefits. Those models are structurally biased and their use as policy making tools flouts scientific standards of objectivity.
In a recent peer-reviewed study, Dayaratna, Guelph University economist Ross McKitrick, and CEI climate scientist Patrick Michaels show that when the FUND model—the only IAM with a carbon fertilization parameter—is run with updated empirical information on climate sensitivity and CO2 agricultural benefits, the social cost of carbon drops to very small numbers, with a 40 percent probability of being negative, through the mid-21st century. A negative cost is another way of saying a net benefit.
Importantly, the researchers found that when the climate sensitivity and carbon fertilization parameters are updated, the choice of discount rate becomes less relevant. For instance, when the FUND model is run with Nicolas Lewis and Judith Curry’s climate sensitivity distribution and a 30 percent boost in agricultural productivity (beyond what is already in FUND), the social cost of carbon remains negative “even at a 2.5 percent discount rate”—currently the lowest rate used by the IWG.
Not only does the IWG interim report not include the Dayaratna et al. study in its list of 115 references, it does not even mention carbon dioxide fertilization. That does not augur well for the final IWG report. It will probably continue to average the results of three models, of which two ignore the abundantly demonstrated agricultural benefits of CO2 fertilization.
Although science can inform SCC estimation, SCC estimates are not science because futurology is not science. No one can foresee how the global economy will evolve over the next 280 years. Consequently, no one can predict the baseline emission trajectory of the global economy out to 2300. But it is only in relation to some assumed baseline that the incremental effects of the next ton of CO2 might be estimated.
The same conclusion results from another obvious fact of life. Human beings use technology to adapt to environmental conditions. Consequently, the damage functions in IAMs—the projected impacts of climate change on consumption, climate-sensitive industries, and human health—depend on assumptions about how technology will develop as the world warms. Nothing is harder to forecast than long-term technological change.
One thing, though, is easy to predict. IAMs built by regulatory advocates will incline to underestimate human adaptive capabilities. For example, in the PAGE model, “Beyond 2°C, no adaptation is assumed to be possible to mitigate the impacts of climate change” (p. 14). That does not compute. As climate economist Richard Tol reminds us, “Humans are a tough and adaptable species. People live on the equator and in the Arctic, in the desert and in the rainforest. We survived ice ages with primitive technologies.”
In his recent book False Alarm, Bjorn Lomborg discusses a study with a worst-case scenario in which by 2100 rising sea levels flood up to 350 million people every year, with costs reaching $100 trillion or 11 percent of global GDP annually. However, those damages are projected only if people do nothing to adapt.
If reasonable adaptations are taken, flood costs increase from $11 billion in 2000 to $38 billion in 2100. Similarly, dike costs increase from $13 billion to $48 billion. However, Lomborg notes, “the total cost to the economy will actually decline, from 0.05 percent of GDP to 0.008 percent.” Moreover, the number of annual flood victims drops from 3.4 million in 2000 to 15,000 in 2100—a 99.6 percent reduction in flood victims! In other words, with reasonable adaptation, people are projected to be much safer, and the global economy much less affected in 2100, despite high-end warming.
“The academic literature has many valid critiques of the methodologies for estimating a SCC as well as critiques of the very idea of a SCC,” Pielke, observes. However, that is not the argument he develops in his recent commentary. Instead, he emphasizes that the baseline emission trajectories from which the IWG calculates SCC values are implausible. Reliance on those unrealistic baselines means that the IWG’s SCC estimates are “much closer to quantification theater than to scientific rigor.”
As indicated above, to estimate the incremental impact of one ton of CO2 emissions, SCC modelers must first estimate how global emissions and concentrations will change over time. Such estimates are only as credible as the socio-economic development scenarios on which they are based. The IWG calculates SCC values with five emission trajectories. Four are no-climate-policy (“baseline”) emission trajectories projected by four socio-economic models participating in a 2009 Stanford Energy Modeling Forum study known as EMF-22. The fifth, a climate policy scenario, is the average trajectory produced by the same four models run with a CO2 stabilization target of 550 parts per million. For more detail, see the Electric Power Research Institute’s 2014 technical assessment of the IWG.
Here is the key point. All the EMF models estimated emissions growth through 2100. The IWG then extended their trajectories out to 2300. According to EPRI, “the extensions lack a coherent, viable, and intuitive storyline (or set of storylines)” that could explain the emission trajectories after 2100. To repeat, nobody can foresee how the global economy will evolve centuries into the future. The IWG did not even try to guess how economies would develop after 2100, yet nonetheless plotted emissions growth over the next 200 years.
But wait, it gets zanier than that. All four reference (no-climate-policy) scenarios (USG1 – USG4) “result in post-2100 cumulative CO2 emissions in excess of estimated fossil reserves.”
For example, in the USG2 scenario, cumulative CO2 emissions reach 22,024 gigatons in 2200 and 33,023 gigatons in 2300—multiples of the estimated reserves (3,674 – 7,113 gigatons). Thus, the IWG’s SCC estimates “envision cumulative carbon dioxide emissions that are far, far in excess of any plausible current expectation about the future,” Pielke observes. “In fact,” he continues, “to even approach these massive amounts of cumulative emissions, the world would have to make it a policy goal to burn as much coal as possible over the coming centuries. That seems unlikely.”
Pielke also notes that it no longer makes sense to treat no-climate-policy scenarios as reference scenarios. Granted, the world is not on course to achieve zero-net emissions by 2050—the professed goal of today’s climate campaigners. However, achieving zero-net emissions by 2100 and certainly by 2200 is far from implausible. If the world achieves net-zero by 2200, cumulative CO2 emissions will reach ~3,150 gigatons. Cumulative emissions in the IWG’s reference scenarios are 3.4 to 10.4 times larger.
Pielke concludes: “If the world economy does not actually emit into the atmosphere tens of thousands of gigatons of carbon dioxide, as envisioned by the IWG, then the majority of the IWG SCC estimates are simply imaginary—setting aside all other methodological issues that might be raised.”
(This post was updated on March 8, 2021.)