In the wake of the Supreme Court’s decision in Loper Bright Enters. v. Raimondo, 244 S.Ct. 2244 (2024), the general breadth and scope of agency decision-making has been called into question. In its recent decision, Huntsman Petrochemical LLC v. EPA, No. 23-1045, 2024 WL 3763355 (D.C. Cir. Aug. 13, 2024), the D.C. Circuit has made it clear that where statutory interpretation is not implicated, the Court will continue to afford EPA’s conclusions involving technical expertise a significant degree of deference. While neither the parties nor the Court attempted to address or reference Loper Bright, the Court articulated a clear standard applicable to agency actions involving statistical and modeling analyses: the Court will examine each step of an agency’s analysis to satisfy themselves that the agency has not “departed from a rational course.” Only where a statistical model “bears no rational relationship to the characteristics of data to which it was applied” will agency action be deemed arbitrary and capricious. Accordingly, regulated entities should be aware that the concept of deference lives on when challenging agency decision-making, even in the wake of the fall of Chevron Deference.

In 2020, the U.S. Environmental Protection Agency issued a regulation applicable to chemical manufacturing facilities that emit ethylene oxide. See National Emission Standards for Hazardous Air Pollutants: Miscellaneous Organic Chemical Manufacturing Residual Risk and Technology Review (“Rule”), 85 Fed. Reg. 49084 (Aug. 12, 2020). EPA derived its authority from Section 7412 of the Clean Air Act, 42 U.S.C. § 7412, finding that ethylene oxide emissions posed an unacceptable risk to human health. The risk was calculated by multiplying the “estimated lifetime exposure to ethylene oxide” by EPA’s cancer-risk assessment (also known as an “Integrated Risk Information System (IRIS) Value”).

Petitioners, including Huntsman Petrochemical LLC, the American Chemistry Council and the Louisiana Chemical Association, challenged the Rule on three grounds: first, they asserted that EPA’s IRIS Value and underlying modeling of the cancer risk of ethylene oxide was arbitrary and capricious. Second, that there were procedural deficiencies in EPA’s promulgation of the Rule. And third, that CAA Section 7412(f) constitutes an unconstitutional delegation of congressional authority to EPA.

With respect to the first claim, the Court stated that “EPA’s evaluation of scientific data within its area of expertise” is accorded an “extreme degree of deference,” which is “particularly true for statistical and modeling analysis.” The Court’s review, therefore, will turn on an evaluation of “each step of EPA’s analysis” to ensure that the Agency “has not departed for a rational course.”  In this respect, statistical models must bear “rational relationships to the characteristics of data to which [they] were applied.” Applying this standard, the Court found that EPA did not stray from its rational course, primarily because EPA’s IRIS Value for ethylene oxide was the culmination of an 18-year process, beginning in 1998 where EPA chose underlying data about cancer rates in populations exposed to ethylene oxide to use, including data from a study conducted by the National Institute for Occupational Safety and Health (“NIOSH”). EPA then developed multiple statistical dose-response models from the data, and picked which models best fit the data using fit metrics and assessment of visual fit calculations. The Court stated that this method allowed EPA to “make sense of a large data set of 17,000 individual data points,” and stated that the agency picked a model “seen for many occupational carcinogens.”  

Ultimately, the Court was “cognizant of [its] limited role in assessing EPA’s evaluation of scientific data within its area of expertise,” and found that EPA explained its rationale relating to the calculation and use of the IRIS Value in its response to Petitioners’ comments during the rulemaking process. The Science Advisory Board, which peer-reviewed EPA’s modeling process throughout the 18 years—agreed that the NIOSH data was most appropriate. Moreover, to the extent Petitioners claimed that EPA’s approach in developing models was “oversimplified,” the Court was persuaded that the tools employed in the process were commonplace in the field of epidemiology.  The Court also found Petitioners arguments in favor of an alternative model produced by the Texas Commission on Environmental Quality unpersuasive, stating that “[t]he fact that some modelers may have chosen petitioners’ approach to this calculation does not automatically render EPA’s approach unreasonable.”

The Court also held that EPA’s IRIS Value did not, on its own, need to be subject to notice-and-comment because it did not amount to a final agency action. In fact, the Court found that the value was not shielded from notice-and-comment at all but was subject to public comment when used in the rulemaking at issue. Additionally, because Petitioners failed to raise its argument that Section 7412 constitute and unconstitutional delegation of authority during the period for public comment, the Court held that they forfeited such.

Thus, this case makes clear that where EPA offers substantive explanations of its technical decisions—especially those which are the culmination of many years of effort—courts will find “a rational connection between the facts and the choices EPA made[.]” Accordingly, in these situations, parties should expect courts to be tentative to intervene and assess whether they would have reached a different conclusion looking at the same data.

Share | Email| Print| Comments|  Topics: Air, Chevron Deference, Clean Air Act, D.C. Circuit, Emissions, EPA, HAPs, Hazardous Air Pollutants, Rulemaking