Suppose a progressive just woke up from a coma and you explained, “There is currently a battle between the Trump Administration and public health and environmental activists. One is favoring government proclamations and edicts issued on the basis of secret information, while the other favors transparency so outside parties can review the methods and data.” The progressive could be forgiven for thinking that the activists would surely be the ones favoring open access. And yet, the opposite situation is currently playing out, as critics blast the EPA’s new proposed rule that requires regulations to be based only on scientific findings that are derived from publicly available information.
I am not a lawyer and cannot comment on the specific language used in the proposed rule. However, as an economist who has worked in the climate change policy debate, I definitely agree that the only hope the public has for accountability and fidelity to the actual scientific literature is that government officials should issue rules on the basis of judgments or analyses that can be reproduced—and thus put to the test—by outside groups. It is amazing to me that we are even having this argument, at a time when there is a genuine “crisis of reproducibility” in the social sciences and even in economics, as I discuss in this recent article.
As this Washington Post story makes clear, much of the dispute centers specifically on the Harvard “Six Cities” study of 1993. (I note with irony that the Washington Post’s banner motto is “Democracy Dies in Darkness.” I leave it to the reader to apply that motto to the current dispute.) One of the concerns is that certain health studies are based on medical records given to researchers with the understanding of confidentiality, and perhaps the proposed rule—as currently worded—could prevent the use of such commonplace research.
Other critics of the EPA’s proposal at the Union of Concerned Scientists object that, for example, to insist on “reproduced results” would be perverse in the case of some health risks. In other words, some critics worry that as written, the rule would in effect be saying, “Before the EPA can issue regulations to reduce known health hazards, outside scientists have to first inject kids with toxins to see if it really does hurt them.”
To repeat, I am not a lawyer and cannot speak to the specific wording of the proposed rule. What I am saying is that there is definitely merit in the call for transparent, reproducible data and methods, notwithstanding legitimate concerns about the implementation of such a proposal.
Forget the current political argument over the EPA, and just ask yourself: In general, is it a good idea for government agencies to be issuing policies on the basis of evidence that outsiders cannot independently examine? How can Congress, for example, evaluate the merits of certain regulations implemented by the Executive branch, if the experts upon whom they call have to report, “We have no way of evaluating these claims”?
Exhibit A: The “Social Cost of Carbon”
These are not merely hypothetical worries. When the Obama Administration’s Interagency Working Group on the Social Cost of Carbon produced estimates of the social cost of carbon (SCC) under various parameter choices, they did not follow OMB protocol by reporting the figures using (a) a domestic (as opposed to global) figure and (b) at a 7% discount rate. (We at IER detailed all of these concerns in our formal comment, and I responded to the GAO’s report on the matter here.)
Besides these serious issues—which involved ignoring OMB guidelines for cost-benefit analysis—there was also the matter of pinning down how much of the computer-simulated damages from future climate change occurred after, say, the year 2150. In other words, if the Obama Administration is telling Congress and the public that “the social cost of carbon using a 3% discount rate in the year 2050 is $69” (as of this writing), it puts that number in perspective to know how much of that ostensible $69 in present-discounted-value-of-future-damages is due to climate change damages that the computer simulations think will happen in the next few decades, versus not for a century or more.
Now because two of the three computer models that the Obama Administration Working Group used in their process are publicly available, outside groups could answer questions like the above. For example, the Heritage Foundation produced this report using the FUND model, to show that with a defensible change in some key parameters on the climate sensitivity and discount rate, the estimated social cost of carbon could be negative. (That would mean society benefits from climate change, above and beyond the direct benefits of fossil fuel use to market participants.)
In a different study using the DICE model, the Heritage Foundation showed that at a discount rate of 3%, the year-2050 social cost of carbon estimates had a bit more than 20% of the damage accruing from events that wouldn’t occur until the year 2150 or later. In other words, for the current estimate of $69 in “social cost” for an extra ton of CO2 emitted in the year 2050, only $55 of it is due to damages that the computer simulates will occur within a century of the emission. The point isn’t that “the distant future doesn’t matter,” but rather that the public and Congress surely appreciate that as computer-simulated events are pushed deeper into the future, they become less reliable as a guide for current policy.
However, even though the Heritage team (which involved a professional computer programmer) was able to do “sensitivity analyses” such as the above on two of the computer models, the third model—the PAGE model—was not publicly available. (One had to undergo expensive training from its developer before being given access.) Beyond that, the original Interagency Working Group did not save (let alone make public) intermediate results in their simulations, which would have allowed for very easy analysis along the above lines. Yet it took months (and costly efforts) for the Heritage team to produce the results I have described, because the estimation procedure and data were not hosted in a manner to facilitate outside reproduction.
In an age when it is becoming routine for peer-reviewed journals to insist that their authors make data available to other researchers before agreeing to publication, it should not be controversial for government agencies to adopt similar standards before using any ostensible scientific “findings” in the promulgation of coercive regulations.
It is true that the specific wording of such a rule would need to take into account legitimate issues of medical privacy. But there are plenty of areas in the climate change policy debate that could benefit from more transparency, and shrouding these issues in secrecy has nothing to do with medical records.
The disconnect between the alarmist rhetoric on climate change, and the actual peer-reviewed research, is astonishing. Policymakers and the public should understand exactly what drives reported “scientific findings.” An insistence on transparency in methods and data is an obvious step in that direction.