Corrosion Proof Fittings v. The Environmental Protection Agency and William K. Reilly, Administrator

An amicus may assist the court by filling gaps or adding detail to an issue already presented by a party with standing. But an amicus cannot use its brief to inject an entirely new claim into the case, such as the EPA's alleged duty to consider the economic development of third-world countries.

The court therefore declined to consider freestanding arguments asserted only by Canada and Quebec. It did, however, draw on amici's arguments concerning asbestos fiber types, fiber sizes, and manufacturing processes because those arguments elaborated on objections that the domestic petitioners had properly raised.

TSCA permits review by an interested person, but that language does not eliminate prudential standing limits. Under the zone-of-interests inquiry, a plaintiff must have interests that Congress could reasonably be understood to have intended the statute to protect or regulate.

The statute required the EPA to consider effects on the national economy and domestic health and environmental risks; it did not require the agency to assess effects on foreign workers, foreign economies, or foreign producers. The court treated that omission as meaningful and found the EPA's decision to exclude foreign effects to be a reasonable interpretation of TSCA.

Cassiar, a Canadian mining company, could not acquire standing merely because it sold asbestos to an American customer. Its domestic vendee was capable of asserting its own rights, and allowing every foreign seller affected by an American regulation to sue would make TSCA standing effectively limitless.

Treaty and trade-agreement arguments did not alter the result. The court noted that agreements such as GATT supply their own dispute-resolution mechanisms and did not give these foreign parties standing to challenge the EPA's rule in this proceeding.

TSCA rulemaking does not require every procedural safeguard associated with an adjudicatory hearing. The agency had discretion to use a hearing officer rather than an administrative law judge, to decline to cross-examine petitioners' witnesses, and to avoid creating a duplicative new expert panel where substantial scientific material already existed.

TSCA nevertheless preserves cross-examination where it is needed to disclose disputed material facts essential to a fair determination. The court said the EPA should have allowed full cross-examination of its major witnesses rather than withholding it from some of them.

A procedural defect alone requires invalidation only when it undermines the judicial-review process. The court did not regard the general limits on cross-examination, standing alone, as enough to vacate the rule. But the absence of cross-examination concerning newly adopted and consequential methodology helped deprive the final rule of substantial evidentiary support.

The EPA adopted analogous-exposure estimates only after the hearings had closed. The estimates extrapolated from exposures in similar activities to estimate asbestos exposure during activities such as repair and disposal, even though the agency had not previously announced that it would use this methodology to support the rule.

The estimates were not a minor addition to the record. They increased the rule's estimated benefits by more than one-third overall, and for certain products—such as some gaskets—they accounted for most of the claimed benefits. Constructive notice from raw data that might have supported the analysis was insufficient when the EPA's ultimate methodology had not been disclosed.

Although agencies need not solicit comments on every item of information they consider, an agency cannot reserve a crucial analysis until public participation has ended and then use it as a central justification for regulation. On remand, the EPA had to reopen the methodology and data for public comment and appropriate examination.

TSCA expressly directs a reviewing court to set aside a section 6 rule that is not supported by substantial evidence in the rulemaking record taken as a whole. Congress specifically displaced the APA's arbitrary-and-capricious standard in this context.

Substantial-evidence review requires the court to examine evidence supporting the agency as well as evidence detracting from its decision, and to ask whether a reasonable mind could accept the record as adequate. The court still owed deference to agency expertise on technical questions, but it could not treat agency expertise as a substitute for record support.

The EPA also bore the burden of explaining why each regulated product or use presented an unreasonable risk and why its chosen remedy was justified. A presumption of regulatory validity did not relieve the agency of that statutory obligation.

TSCA is not a zero-risk statute. It directs the EPA to protect adequately against unreasonable risk through the least burdensome requirements, while considering the environmental, economic, and social effects of its action.

A broad ban is the most burdensome option available under TSCA. Before adopting it, the EPA had to show not merely that a ban would reduce asbestos risks, but also that less restrictive measures—such as labeling, product-specific controls, workplace standards, limits on particular uses, or coordination with other agencies—would not adequately achieve the chosen level of protection.

The EPA compared a world with no new TSCA regulation to a world with a ban, but did not calculate the costs, benefits, and risk reductions associated with meaningful intermediate alternatives. That comparison artificially favored the ban and did not establish that no less burdensome option would work.

The EPA could not justify bypassing regulation by asserting that no exposure level is entirely risk-free. Congress required the agency to address unreasonable risk, not to eliminate every possible risk regardless of cost. Nor could the agency assume that workplace regulation would fail because the government might enforce it poorly; the appropriate response to deficient enforcement is improved enforcement.

The court accepted the EPA's general use of discounting, but concluded that the agency should discount both costs and benefits to maintain a comparable analysis. It also criticized the EPA for discounting benefits from the date of exposure rather than the later date when asbestos-related injury would manifest, an especially serious error for diseases with long latency periods.

The EPA quantified benefits only through the year 2000 and treated later benefits as unquantified. Unquantified benefits can matter in close cases, but the court held that the agency could not deliberately use an unreasonably short period of analysis and then invoke the resulting large category of unknown benefits as a decisive justification for high costs.

The agency also appeared to count population exposure twice: once in calculating expected harm through risk multiplied by exposure, and again as an additional independent reason to regulate. That redundant reasoning distorted the analysis.

TSCA required consideration of the benefits of asbestos and the availability of substitutes. For products without adequate existing substitutes, the EPA bore a particularly heavy burden before imposing a ban. A waiver process did not cure that problem, because it shifted the burden to regulated parties to prove why an exemption should be granted.

Most importantly, once parties introduced credible evidence that likely substitutes could be dangerous—including carcinogenic materials or less effective safety products—the EPA had to assess the comparative risk. The agency did not have to test every conceivable substitute, but it could not plausibly claim that its rule improved safety while refusing to evaluate known and credible substitute hazards.

Friction products accounted for nearly three-fourths of the EPA's anticipated mortality benefits, and the court acknowledged that this portion of the record was the agency's strongest. The evidence supported the proposition that workers could face substantial asbestos exposure during manufacture, use, repair, and disposal of brakes and related products.

For new vehicles, the record indicated that manufacturers were already developing or using non-asbestos brake systems. A properly supported ban in that original-equipment market might have been sustainable if the EPA had conducted the required least-burdensome-alternative analysis.

The replacement-brake market posed a different problem. Existing vehicles had brake systems designed around asbestos materials, and EPA-commissioned evidence suggested that non-asbestos replacements might perform less safely or might not yet be available. The agency could not rely on possible future regulation by the National Highway Traffic Safety Administration instead of determining whether the substitutes were technically safe.

The EPA also failed to assess credible evidence that non-asbestos fibrous substitutes carried inhalation-related cancer risks. Without considering both highway-safety risks and substitute toxicity, the agency did not establish that its ban of replacement friction products was reasonable or supported by substantial evidence.

For asbestos-cement pipe, the EPA predicted a cost of roughly $128 million to $227 million to save only three to four discounted lives over thirteen years. Much of the supporting exposure analysis relied on the undisclosed analogous-exposure methodology, and the agency did not adequately respond to challenges to the underlying exposure assumptions.

The EPA acknowledged that likely pipe substitutes, including PVC and ductile iron, could present cancer risks comparable to those associated with asbestos-cement pipe production. It nonetheless declined to conduct a genuine comparative-risk assessment and instead rejected adverse evidence through unsupported assumptions. Speculation and agency assertion could not substitute for substantial evidence.

The court also criticized the EPA for retaining challenged figures while claiming that other, unrelated estimates might be too low. Under substantial-evidence review, an agency must support the method producing its conclusion; it may not defend a flawed input by offsetting it with a separate possible error.

For gaskets, the analogous-exposure estimates made up nearly eighty percent of the claimed benefits, making the failure to give notice and permit scrutiny especially consequential. For roofing materials, coatings, shingles, and paper products, some of which were projected to save less than one statistical life, the EPA had a correspondingly acute obligation to examine less burdensome alternatives before selecting a complete ban.

TSCA authorizes regulation of substances that "present or will present" an unreasonable risk. A product's absence from current domestic production does not show that it is harmless; it may reenter commerce and again expose people to a known risk.

The court rejected the claim that the EPA could never regulate unknown future uses merely because it could not calculate the exact costs and benefits of products not yet invented. The statutory reference to future risk, coupled with the practical uncertainty surrounding such uses, allowed a properly supported cleanup ban.

On the EPA's motion for clarification, the court limited this holding to products that were not manufactured, imported, or processed on July 12, 1989, when the final rule was issued. Products that ceased production only because the phase-one ban had already taken effect remained subject to the court's general vacatur and remand.

The court distinguished between an agency's failure to consider factors TSCA expressly required and a disagreement over scientific judgments within the agency's expertise. Petitioners' objections to the treatment of different asbestos fibers and industries primarily asked the court to reweigh competing evidence.

Judicial review did not make the court a substitute regulator. The EPA remained free on remand to retain or reconsider its judgments concerning fiber potency, fiber length, and varying exposure settings, so long as its ultimate regulation satisfied TSCA's procedural and substantial-evidence requirements.