EPA and NHTSA Lowball Estimates of Carbon Costs in Proposed Tailpipe Emissions Standard

Once upon a time, EPA and other agencies labored under the yoke of a cruel regime that was contemptuous of the “reality-based community,” but intimately aware of the needs and desires of the energy industry. Climate policy didn’t really happen in those days. Then the world changed.

In the first year of the new regime, EPA and NHTSA proposed a standard for tailpipe emissions, including an estimate of the “social cost of carbon,” or the value of the incremental damages caused by greenhouse gas emissions. Someone needs to tell the authors of this standard that we are free at last to take climate change seriously; you don’t have to keep censoring yourself, as you may have in the past. And once we start exercising that freedom, we will find that the social cost of carbon is much larger than the EPA/NHTSA estimates, which include values as low as $5 per ton of carbon dioxide.

The analysts who proposed the new standard narrowed their scope at the outset, ducking the tough questions and relying on the most conservative sources – and in some cases, misrepresenting data and references. There are three stages of the analysis, each of them problematic: estimates of climate damages; choice of a discount rate; and near-dismissal of worst-case catastrophic risks. (The following remarks are highlights of comments I submitted to the agencies on the proposed tailpipe emissions standards, which include citations to references and data sources.)

To estimate the value of damages that will result from climate change, the EPA/NHTSA analysis first restricts itself to a literature review by Richard Tol, then emphasizes its exclusive reliance on peer-reviewed journal articles, then narrows its attention even further to three economic models which it identifies as “the leading” models in the field.

In fact, Tol’s literature review includes more than 200 estimates of the social cost of carbon – but more than half of them are from his own work! Tol suggests that the Stern Review, the massively researched study of climate economics sponsored by the British government, is an outlier in the field; Tol and a handful of other economists have complained that the Stern Review did not appear in a peer-reviewed journal. Thus the EPA/NHTSA emphasis on peer review as a criterion serves to exclude consideration of the Stern Review in U.S. policymaking. (Despite the lack of peer-reviewed publication, the American Economics Association invited Nicholas Stern to give a major lecture on climate economics at the group’s annual meeting; the lecture was published in the top-ranked journal in the field, the American Economics Review.)

The three models chosen by EPA and NHTSA, namely DICE, PAGE, and FUND, are not the most comprehensive or detailed; on the contrary, they are among the simplest and most easily modified climate economics models. This has allowed many investigators to experiment with these models – showing that the results are extremely sensitive to small changes in input data. Choosing these models does not determine the cost of climate damages; that depends on the input data as well. In my detailed comments on the proposed rule, I present reasons to think that the default data sets used by the models’ developers are, in all three cases, erring on the side of understating climate damages.

Those who discuss climate economics are condemned to debate about discount rates -- the concept that anything valuable, be it money or human life, will be worth a predictable amount less in the future than it is today. The EPA/NHTSA analysis reads as if it began by deciding that 3% and 5% were the correct discount rates, and then set about marshalling evidence in support of these numbers. The proposed rule suggests that 3% is a reasonable estimate of a discount rate based on the real rate of return on risk-free assets, such as Treasury bonds. This is empirically wrong: Treasury data show that 2% is a much better current estimate of real long-term rates. Economists examining longer-run data have often estimated the risk-free rate of return to be even lower than 2%.

Alternatively, the proposed rule considers a discount rate derived under the so-called “Ramsey equation” of economic theory. EPA and NHTSA assert that 5% is a typical discount rate based on this approach, and footnote five academic sources in support of that claim. But only one of the five sources actually advocates a discount rate as high as 5%! Three of the five sources suggest a discount rate of 2% or lower; two of them express serious doubts about the relevance of the “Ramsey equation” framework for discounting climate costs and benefits.

Moreover, the proposed rule takes no position on the arguments for discount rates that decline over time, describing them as new to the literature. Yet the major innovations in this area were published from 1998 through 2003. The EPA/NHTSA estimates of the social cost of carbon range from $5 per ton of carbon dioxide (at a constant 5% discount rate) to $54 per ton (at a discount rate starting at 3%, then declining), underscoring the importance of the choice of a discount rate. At the 2% rate suggested by the academic literature and data discussed in the proposed rule, the social cost of carbon would be even higher.

Finally, one of the most crucial questions in the whole field of climate economics is all but dismissed – sorry, I meant relegated to a closing “caveat.” How should we respond to worst-case catastrophic risks, such as collapse of the Greenland ice sheet, eventually resulting in 7 meters of sea-level rise? In a path-breaking theoretical analysis, Martin Weitzman has demonstrated that in cases of uncertain probabilities of unbounded risks (such as climate change), the marginal value of a unit of risk reduction – that is, the social cost of carbon – is infinitely large. This provides a rigorous foundation for the precautionary principle; it argues for a large-scale initiative in climate protection, interpreted as social insurance against catastrophe rather than as the result of a cost-benefit calculation.

But that’s not happening this year, in the world of EPA and NHTSA. It will take more than a year, it seems, to realize that we are now free to take climate change seriously.