Picking your own cherries

People have noted that the DoE report misrepresents their research on climate change. This is true for my work too. Did the authors faithfully represent their own research?

Ross McKitrick is a co-author of the report, which cites 16 papers by McKitrick, increasing his citations by 0.4%. Let’s look at three of the cited papers, co-authored with Kevin Dayaratna.

The DoE reports writes: “Uncertainties in ECS [Equilibrium Climate Sensitivity] are highly consequential for policy making. As will be discussed in Chapter 11, economic models use ECS values to project the costs of CO2 emissions. The traditional value (3.0 °C) has typically yielded modest global social costs of CO2 emissions, sufficient to justify some policy actions, but mostly deferred to later in this century. If ECS is very high (above 4.5°C) immediate aggressive emission controls become more imperative, whereas no CO2 emission controls are economically justifiable for ECS below 2.0°C (Dayaratna et al. 2017, 2020).”

In contrast to the suggestion in the DoE report, Dayaratna et al. (2017, 2020) do not show results for point estimates of the ECS.

Dayaratna et al. (2017, 2020) find a mix of social costs and social benefits of carbon, depending on the discount rate used, the assumed distribution of the climate sensitivity, and the assumptions on the strength of CO2 fertilization. The DoE summary omits the social costs, falsely claims that Dayaratna only found social benefits.

The selective omission of some of Dayaratna’s results is inconsistent with other policies of the Trump administration. Carbon dioxide emissions are a social benefit if the discount rate is high and CO2 fertilization large. The President has consistently called for a lower interest rate — implying a lower discount rate.

The benefits of CO2 fertilization fall primarily outside the USA. The first Trump administration lowered the social cost of carbon by excluding the impacts of climate change on other parts of the world.

The DoE report continues: “Obtaining a precise estimate [of the climate sensitivity] is impossible, so policy making needs to account for the uncertainty.”

The Von Neumann-Morgenstern Axioms imply that uncertainty is accounted for by considering the mean of the variable of interest.

Later, the DoE report writes: “Dayaratna et al. (2023) used the updated empirical ECS distribution estimate of Lewis (2022), which assimilates modern instrumental and paleoclimatic temperature records, and allowed for a 30 percent gain in the CO2 fertilization benefit in the FUND model, and found that, even at a low discount rate of 2.5 percent, the median SCC as of 2050 is only $18.67, with a 24 percent probability of the true value being negative. At a five percent discount rate the median SCC value is negative until the mid-2040s and at 2050 was only $0.37 with a 49 percent probability of being negative. Thus, under reasonable assumptions a mainstream IAM using updated scientific inputs yields evidence consistent with the SCC not being significantly greater than zero.”

Again, the DoE report cherry-picks the results by Dayaratna. Just as important, the lightly-reviewed 2023 rejoinder to the 2020 paper switches, without explanation, from the mean to the median. This is just wrong.