The “social cost of carbon” (henceforth “SCC”) is one of the major points of controversy on energy policy, amidst the major changes taking place in Washington. A recent Pro Publica article by Andrew Revkin wondered aloud whether the new Trump Administration would respect the Obama Administration’s use and estimated magnitude of the SCC. The controversy is particularly relevant to us here at IER, because our own work has been cited in the debate. In this article, I’ll clarify what IER’s position on the SCC has been all along—we’ll see that the Pro Publica piece was a bit inaccurate. Let me also state clearly that the present post reiterates IER’s position and is not tied to what officials in the Trump Administration may believe.
Revkin Lays Out the Controversy
In his article, Revkin explains what the SCC is, and how it has been used to influence federal policy under the Obama Administration:
[T]here’s probably no more consequential and contentious a target for the incoming administration than an arcane metric called the “social cost of carbon.”
This value is the government’s best estimate of how much society gains over the long haul by cutting each ton of the heat-trapping carbon-dioxide emissions scientists have linked to global warming.
Currently set at $36 per ton of carbon dioxide, the metric is produced using a complex, and contentious, set of models estimating a host of future costs to society related to rising temperatures and seas, then using a longstanding economic tool, a discount rate, to gauge how much it is worth today to limit those harms generations hence…
The contention arises because the social cost of carbon underpins justifications for policies dealing with everything from power plants to car mileage to refrigerator efficiency. The carbon valuation has already helped shape 79 regulations.
Revkin later in the piece claims that an IER economist (and here he was referring to me, see video or text) had testified to the Senate that “under a proper calculation, the social cost of carbon ‘would probably be close to zero, or possibly even negative.’” Revkin then went on to quote other economists to argue that the SCC was much higher than zero, and that if anything, the Obama Administration’s official estimates were too low.
Clarifying IER’s Position on the SCC
Although Revkin correctly quoted me, without the context he (perhaps unintentionally) gave the wrong impression about IER’s position. Strictly speaking, it’s not that IER has been arguing that the “social cost of carbon” is lower than $36/ton.
Rather, we have been arguing that the very concept itself is far too dubious to be used in federal policymaking. As we put it in our formal Comment submitted back in February 2014:
[T]he use of the SCC as an input into federal regulatory actions is totally inappropriate. The Administration is treating the SCC as if it is a scientifically valid, objective fact of the external world, akin to the charge on an electron or the boiling point of water at sea level. However, the SCC is no such thing, at least in our present state of understanding. Rather, the SCC is an arbitrary output from very speculative computer models. It can be adjusted up or down as the analyst wishes, simply by changing a few key parameter choices. Simply by adjusting the parameter and modeling choices in plausible ways, a knowledgeable economist can generate SCC estimates that are very high, very low, or even negative—meaning that carbon dioxide emissions actually shower “positive externalities” on humans beyond the direct benefits to the emitters, and therefore should (according to the Administration’s logic) receive federal subsidies.
When the average person hears that experts are arguing over the “social cost of carbon,” he probably imagines the debate concerns how much the Earth will warm for a specified amount of greenhouse gas emissions, or perhaps the disputes focus on the amount of damage that a certain amount of warming will cause.
While those disagreements and uncertainties are crucial in the debate, they are not the main drivers over the possible range of the official SCC. For even if we agree—for the sake of argument—how much net damage (measured in dollar terms) an additional ton of carbon dioxide emitted today will cause, in each year for the next three centuries, we still can make the official SCC really high, close to zero, or (depending on which of the officially selected models we use) even negative.
The two main parameters to achieve this flexibility are the discount rate we use, and whether we are counting global versus domestic benefits (or harms). The simple fact is that the standard computer simulations that project large damages from human-caused climate change don’t show too large of an impact until decades into the future. Indeed, one of the three computer models chosen by the Obama Administration’s working group actually shows net benefits to humanity from global warming at least through the year 2050. (Some of the benefits include fewer elderly deaths in the winter, and longer growing seasons in some areas.)
So if the bulk of the computer-projected damages occur decades or even centuries in the future, then the annual rate at which we discount those damages—to turn $100 of damage in the year 2300 into its equivalent dollar-figure today—is incredibly influential on our answer. For example, the Obama Administration’s own estimates suggest that the “social cost of carbon” in 2015 was $36/ton if we adopt a 3% discount rate—this was the knob on the dial that Revkin used for his article. However, making the same computer projections about climate change and its impact on humans, by simply increasing the discount rate to 5%, the 2015 SCC estimate drops down to $11, as the table at the link shows (in the top row).
The reason that the choice of discount rate matters so much is that the time horizon considered in the calculation of the social cost of carbon is from today to the year 2300. The difficulty in calculating the economic impacts of carbon dioxide emissions in 2300 is a daunting task. It is a task similar to asking a young Adam Smith to calculate the economic impact of the steam engine, including the economic impact of the increased use of coal as a result of steam engines.
Like the choice of discount rate, we see a similar result when we consider the decision of whether to restrict our cost/benefit analysis to the United States, or to the world. As it turns out, most of the computer-simulated damages of climate change will impact foreign countries, not Americans. Indeed, the Obama Administration’s own analysis admits that if we wanted to estimate the SCC as it pertains to Americans, we would reduce the official estimate (for a given discount rate) by anywhere from 77% to 93%. (!) For example, if we stipulate Revkin’s choice of the discount rate of 3%, and say that the SCC is $36/ton, we are talking about a global figure. The domestic SCC at a 3% discount rate is—according to the Obama Administration’s own suggested procedure—somewhere between $3 and $8 per ton.
Now here’s the kicker: When the Office of Management and Budget laid out guidelines for regulatory analysis (as codified for example in Circular A-4) it instructed the reader that, “Your analysis should focus on benefits and costs that accrue to citizens and residents of the United States,” and it also was quite explicit that any costs and benefits needed to be reported for both a 3% and a 7% discount rate.
Yet the Obama Administration’s working group on the social cost of carbon flatly ignored both requirements. That is, they reported the SCC for at most a 5% discount rate, and they used global estimates. They gave reasons for their decisions (which we critique here), but the simple fact is that they ignored the government’s own guidelines for regulatory cost/benefit analysis.
We are now in a position to circle back to Revkin’s quote from my Senate testimony. I wasn’t saying that I thought the best estimate of the SCC was “close to zero or possibly even negative,” but rather I was saying if the Obama Administration had followed the OMB’s guidelines then that’s what the number would have been—even stipulating the computer simulations that they chose for the estimation.
Conclusion
To reiterate, the “social cost of carbon” is not a fact about the universe, akin to the charge on an electron or the mass of the moon. Rather, it is an amorphous concept, which can be given an enormous range depending on very controversial (and arbitrary) modeling decisions by the analyst. This is why we at IER argued that the SCC is inappropriate for use in federal policymaking.
For an analogy, the EPA didn’t propose using the “social cost of methane” until 2015. Does that mean the personnel at EPA thought the social cost of methane was exactly $0 before then? Of course not. Rather, even they knew that the social cost of methane was a poorly studied concept that was far too dubious to be used to influence policy.
We are making the same point about the “social cost of carbon.” We are not arguing, “It’s probably $0,” rather we are saying that it is too malleable and arbitrary—and prone to manipulation by the analyst who knows what answer he or she wants—to be used in federal policymaking.
Our analysis on this narrow point doesn’t preclude or justify any particular measure taken to mitigate climate change. Regardless of what one thinks of the necessity for federal regulation in this area, using an estimated $36/ton for the “social cost of carbon” gives the public an unwarranted confidence in the precision and scientific standing of that number. As we have demonstrated, that number deserves no such respect.
