For some time now I’ve been trying to convey just how ludicrous it is that the US government’s anti-carbon policies are based on computer simulations of the global economic/climate system through the year 2300. A recent piece by David Kreutzer and Kevin Dayaratna (an economist and computer programmer, respectively, at the Heritage Foundation) is the best I’ve yet seen:
When you switch on the kitchen light tonight, how will it affect Captain James T. Kirk and the intrepid crew of the Starship Enterprise?
Captain Kirk is a fictional character, of course, but the question — thanks to the EPA — is all too real. The agency calls it the “Social Cost of Carbon.” In the SCC they claim to have an estimate, measured in dollars per ton of CO2, for all the damage that your free-and-easy light-switching today will impose on the world from now until the year 2300, at which point Captain Kirk would be 77. To save Kirk and the rest of future Earth from a panoply of speculative, incremental horrors, the EPA has plans to nudge your carbony little fingers away from that switch and many others.
The two authors then go on to describe their results when they tweak 2 of the 3 computer models that the Obama Administration’s Working Group used to calculate the “social cost of carbon”:
Like the EPA, we ran the DICE model (created by William Nordhaus) and the FUND model (by Richard Tol). But we used more recent peer-reviewed estimates of CO2’s impact on world temperature (here and here), and we ran the models using a 7 percent discount rate (that is, the rate at which we value short-term effects over long-term effects), as stipulated by the OMB and curiously ignored by the EPA, which used 2.5 percent, 3 percent, and 5 percent.
The simple substitution of the discount rate drops the SCC estimate by more than 80 percent in the DICE model and pushes it to zero or even negative for the FUND model. (Here it might be appropriate to observe a moment of sympathetic silence for Richard Tol, a lead and convening IPCC author, who was recently excoriated by his former brethren for publicly stating the painfully obvious: Some impacts of warming are good.)
Updating the estimates of CO2’s impact on temperatures drops the SCC estimates by 40 to 80 percent. When the 7 percent discount rate is used in conjunction with the updated CO2 impacts, the FUND model indicates a better-than-even chance the SCC is negative.
That is, using more up-to-date numbers renders a range of estimates so broad that it is not clear whether CO2 emissions should be restricted or subsidized. Are these changes debatable? Absolutely. Is the debate on these topics over? Absolutely not. We cannot calculate how today’s trip to the grocery will harm humanity in three centuries. Nevertheless, the EPA is forging ahead with regulations employing a SCC of about $40 per ton for the near term. [Bold added.]
Incidentally, I haven’t personally vetted their computer runs, but their results are consistent with my own simulations run on the DICE model when I tweaked it for a paper in The Independent Review. And as I explained in my Senate testimony last summer, the FUND model does indeed show “positive externalities” of global warming for modest warming, which means that the use of a high enough discount rate would render a prima facie case for subsidizing carbon dioxide emissions.
Of course, we aren’t seriously proposing government subsidies on the grounds that CO2 emissions confer “social benefits” that aren’t being fully internalized by the emitters. The point is that even taking the Administration’s case on its own terms, it’s not even clear that CO2 emissions are a bad thing. And to repeat, we’re not inventing “the Heritage Foundation computer model.” Rather, we are running 2 of the 3 computer models that the Obama Administration team used to come up with these figures for federal agencies.
Last point: You might wonder why they (and I) didn’t try peeking under the hood of the third model, the so-called PAGE model. The answer is that it’s proprietary, and the creator (Chris Hope) doesn’t make the code publicly available. (If you email him, he gives conditions under which you can use it, which involves paying him for a pretty expensive training session.) That’s of course his prerogative–and I mean that sincerely, I don’t expect him to volunteer his time or post his proprietary code for the world to download–but it makes it impossible for outside researchers to vet the government’s reported figures when one of the computer models involved is inaccessible. One would think that when federal officials selected the representative computer models to use when estimating the “social cost of carbon,” they would pick models that are publicly available.