EPA and Data Transparency

The New York Times recently reported on a proposed policy change at the Environmental Protection Agency that would require the agency to only rely on scientific research with publicly available data when setting pollution exposure standards. Proponents of the rule argue that the practice would allow other researchers to examine and replicate findings, an essential characteristic of the scientific method. Opponents argue the rule would exclude large amounts of research that rely on confidential health information that cannot be public. The Times quotes opponents who view the policy change as an attempt by the Trump administration to attack regulations they don’t agree with by undermining the scientific results on which they are based.

Increased transparency in data used in empirical research and the facilitation of replication of studies are like mom and apple pie. In an ideal world, such practices are the very essence of the scientific method. In practice, academic journals in many disciplines already require that data used in empirical and experimental work be available for replication. The International Committee of Medical Journal Editors, no strangers to the limitations of data that include personal information, recently affirmed their commitment to responsible data sharing.

While opposition to transparency certainly has bad optics, the opponents of this rule change do have a point. The struggle over transparency isn’t really about transparency. Instead, it is simply the latest chapter in the scrum over two studies whose results are the bases of EPA decisions about appropriate clean air exposure standards.

The Harvard Six Cities Study (SCS) and the American Cancer Society Study (ACS), both published in the 1990s, provide the data on which the EPA estimates all mortality risks from air pollution. Both studies looked at the health damage caused by particulate matter (PM), which accounts for 90 percent of the health benefits from emission regulation, and found that higher levels of PM exposure were associated with increased mortality. And these studies both rely on individual health data given under conditions of confidentiality

So the opponents are probably correct that the transparency rule is a clever attempt to undermine the current basis for EPA regulation of PM. And reduced PM exposure is the exclusive basis on which current conventional pollution regulation is justified because the benefits of additional emissions controls on other conventional pollutants are low. In this environmentalists’ nightmare, a transparency mandate ends additional regulation of air quality by the EPA.

But contrary to the Times assertion, this a not simply a tale of good science versus evil polluters. The SCS has been the subject of intense scientific scrutiny and much criticism because of results that are biologically puzzling. The increased mortality was found in men but not women, in those with less than high school education but not more, and those who were moderately active but not sedentary or very active. Among those who migrated away from the six cities, the PM effect disappeared. Cities that lost population in the 1980s were rust belt cities that had higher PM levels and those who migrated away were younger and better educated. Thus, had the migrants stayed in place it is possible that the observed PM effect would have been attenuated.

Furthermore, a survey of 12 experts (including 3 authors of the ACS and SCS) asked whether concentration-response functions between PM and mortality were causal. Four of the 12 experts attached nontrivial probabilities to the relationship between PM concentration and mortality not being causal (65 percent to 10 percent). Three experts said there is a 5 percent probability of noncausality. Five said a 0-2 percent probability of noncausality. Thus 7 out of the 12 experts would not reject the hypothesis that there is no causality between PM levels and mortality. Based on these findings, a 95 percent confidence interval would include zero mortality effect for any reductions in PM concentration below 16 micrograms per cubic meter.

So the scientific fragility of the two studies has been known for some time. Despite that fragility, in December 2012, EPA set a much lower fine-PM standard of 12 micrograms per cubic meter of air to be met by 2020.

If the EPA is forced to use other studies, such as two recent papers by Michael L. Anderson and by Tatyana Deryugina, Garth Heutel, Nolan H. Miller, David Molitor, and Julian Reif, the estimated benefits from PM exposure reduction are reduced. Anderson’s estimate is 60 percent smaller. Deryugina et al conclude that declining PM concentrations from 1999 to 2011 resulted in an additional 150,000 life-years per year, which, if valued at $100,000 per life year, would equal $15 billion in annual benefits. The EPA estimates that the annual compliance costs of the 1990 Clean Air Act standards were $44 billion in 2010. With these lopsided costs and benefits, it is certainly true that eliminating the ACS and SCS studies from consideration and forcing reliance on other studies would result in less stringent regulation except in areas with bad geography, such as Los Angeles, that prevents pollution dispersion.

The Times article mentions the fact that the Congress has had the opportunity to enact language achieving the same goal of transparency but has not done so. Libertarians have long criticized the growth of the discretionary administrative state. Thus because Congress has explicitly considered but failed to enact a research transparency requirement, libertarians should be cautious about using administrative discretion to achieve their preferred outcome.

Written with research assistance from David Kemp.