Chemical Manufacturers Association v. Environmental Protection Agency Carol M. Browner, Administrator, United States Environmental Protection Agency

28 F.3d 1259, 307 U.S. App. D.C. 392, 38 ERC (BNA) 2055, 1994 U.S. App. LEXIS 17734

• Court: Court of Appeals for the D.C. Circuit
• Decided: July 19, 1994
• Docket: 93-1178
• Status: Published

Opinion

Opinion for the Court filed by Circuit Judge GINSBURG.

GINSBURG, Circuit Judge:

Methylene diphenyl diisocyanate (MDI) is a precursor of various polyurethane foams and resins used in the manufacture of upholstery, insulation, wood and automobile coatings, and other products. The Chemical Manufacturers Association petitions for review of an EPA rule designating MDI a “pollutant! ] for which high risks of adverse public health effects may be associated with exposure to small quantities.” The high risk designation limits the extent to which a manufacturer or user of MDI may participate in the Early Reduction Program for hazardous air pollutants under the Clean Air Act.

The CMA claims that the EPA inadequately addressed its comments with respect to the agency’s generic air dispersion model; that both the model and the scientific database containing study data for MDI should have been subjected to notice and comment procedures before the EPA could base its decision upon them; and that the EPA acted arbitrarily and capriciously in designating MDI a high risk air pollutant on the basis of a model that bears no rational relationship to the physical properties of the chemical and on the basis of a minor health effect. We agree with the last point in both its aspects; the EPA’s rule designating MDI a high risk air pollutant is for both reasons substantively arbitrary and capricious. Accordingly, we vacate that decision.

I. BACKGROUND

The Clean Air Act Amendments of 1990 require the EPA to establish a program, based upon maximum achievable control technology (MACT), to reduce emissions of 189 listed hazardous air pollutants, of which MDI is one. 42 U.S.C. § 7412(d). A facility that produces or uses a hazardous air pollutant may apply to participate in the Early Reduction Program — which requires that emissions of such pollutants be reduced in the aggregate by 90 percent — and thereby postpone for six years the deadline for its compliance with the MACT requirements. 42 U.S.C. § 7412(i)(5)(A). A facility participating in the Early Reduction Program may reduce its emission of one pollutant by less than what would otherwise be required if it reduces its emission of another by an equal amount more than would otherwise be required, see 42 U.S.C. § 7412(i)(5)(E) — except with respect to “pollutants for which high risks of adverse public health effects may be associated with exposure to small quantities including, but not limited to, chlorinated dioxins and furans.” Any reduction of a non-high risk pollutant in excess of what is required can be used to offset only one-tenth that amount of the required reduction of a high risk pollutant. See Final Rule: National Emission Standards for Hazardous Air Pollutants for Source Categories; Compliance Extensions for Early Reductions, 57 Fed.Reg. 61,970, 61,983-84 (1992).

In the rule under review, the EPA placed hazardous air pollutants on the high risk list if they met any one of three criteria. First, the EPA included all chlorinated dioxins and furans, as directed by the statute. 57 Fed. Reg. at 61,980. Second, the EPA listed all of the hazardous air pollutants that are known human carcinogens. Id. Finally, the EPA listed each pollutant that it determined may be toxic at or below a certain threshold ambient concentration. Id. at 61,980-83. As a measure of toxicity the EPA used the inhalation reference concentration (RfC) for the particular pollutant; the RfC represents the estimated maximum exposure to a pollutant, as extrapolated from animal studies, that a human can tolerate continuously for 70 years without experiencing any adverse health effect. The EPA considered for inclusion on the high risk list generally only those hazardous air pollutants for which it has verified an RfC in its Integrated Risk Information System (IRIS) database; for MDI the database contains an EPA-verified RfC, based upon a two-year rat study, of 0.05 /xg/m3.

To determine the likely level of human exposure to a hazardous air pollutant, the agency developed a generic air dispersion model; using this model it predicted the ambient concentration, at a certain radius from the source, of a hypothetical (i.e., generic) air pollutant emitted from a typical industrial facility under average meteorologic conditions. The EPA then established a threshold “exposure [level] to small quantities” of the generic hazardous air pollutant, using ten tons of emissions per year per facility as the measure of a small quantity. 57 Fed.Reg. at 61,981. (Because the model does not take into account the peculiar characteristics of each pollutant, the ambient concentration derived from the model — 5.02 |xg/m3 — is the same for all of the 189 identified hazardous air pollutants.) Finally, the EPA designated as high risk every hazardous air pollutant, including MDI, for which the threshold concentration exceeded by at least one order of magnitude (i.e., was at least 10 times) the RfC for that pollutant. Id; Memorandum from Martha H. Keating to Early Emission Reduction Program Docket A-90-47 at 11, Table 32 (June 16, 1992).

Initially the EPA had also proposed, as what it called a “reality cheek,” to consider for high risk designation only those pollutants for which actual emissions exceed ten tons per year from at least one reporting facility; in this way it could avoid designating as a high risk any pollutant actually emitted in a quantity too insignificant to pose a high risk to human health. National Emission Standards for Hazardous Air Pollutants for Source Categories: Proposed Regulations Governing Compliance Extensions for Early Reductions of Hazardous Air Pollutants, 56 Fed.Reg. 27,338, 27,362-63. In the final rule, however, the EPA eliminated this criterion on the ground that its actual emissions data are incomplete because only large facilities report emissions; reliance upon their reports could result in overexcluding from the high risk list a pollutant emitted in significant quantity by a nonreporting facility. 57 Fed. Reg. at 61,981. As a result, the EPA rejected as irrelevant the data that the CMA submitted to demonstrate that no facility emits more than ten tons of MDI per year. National Emission Standards for Hazardous Air Pollutants; Compliance Extensions for Early Reductions — Background Information for Promulgated Standards 2-77 (1992) [hereinafter “BID”}.

II. Analysis

The CMA mounts two procedural and two substantive challenges to the final rule as applied to MDI. As to procedure, the CMA contends that the EPA should have subjected both its generic air dispersion model and its IRIS database to notice and comment rule-making, and that the EPA failed to respond adequately to the CMA’s comments challenging the agency’s application of the model to MDI. As to substance, the CMA challenges both the use of the generic air dispersion model as a rational approximation of the behavior of MDI in a typical facility environment, and the validity of the RfC for MDI as a reasonable approximation of the threshold exposure level for avoiding significant public health effects.

A. Procedural Issues

In evaluating a procedural challenge to an emission limitation, the Clean Air Act requires that we determine whether (1) the agency failed to observe the procedures required by law; (2) any such failure was arbitrary and capricious; (3) the petitioner objected to the agency’s dereliction with reasonable specificity during the time for public comment; and (4) “the errors were so serious and related to matters of such central relevance to the rule that there is a substantial likelihood that the rule would have been significantly changed if such errors had not been made.” 42 U.S.C. § 7607(d)(8) & (9)(D); cf. 42 U.S.C. § 7607(d)(1)(C) (applying the § 7607 standard of review to emission standards or limitations promulgated under 42 U.S.C. § 7412(d)); 42 U.S.C. § 7412(d)(1) (requiring Administrator to promulgate emission standards for hazardous air pollutants listed therein); 42 U.S.C. § 7412(i)(5) (describing the Early Reduction Program as an “alternative emission limitation”). In this case, however, we need not venture beyond the first step, at which we conclude that the EPA did not depart from the required procedures.

The CMA’s notice and comment argument is that the generic air dispersion model and the IRIS database each took on the force of a rule when the EPA dropped its initial proposal to include the “reality cheek” of actual emissions in determining whether a pollutant represents a high risk to human health. In most cases, that is, a hazardous air pollutant will be deemed a high risk solely because the hypothetical ambient concentration predicted by the generic air dispersion model exceeds the RfC benchmark (i.e., 10 times the RfC) in the IRIS; that the actual exposure level may be less than the RfC benchmark will not be relevant.

In response, the EPA first claims that it did in fact subject both the air dispersion model and the IRIS database to notice and comment by requesting, in the notice of proposed rulemaking, comments “on all aspects of the proposed rule,” including “the method used to select the high-risk pollutants.” 56 Fed.Reg. at 27,340, 27,355. See also id. at 27,363 (“The EPA requests comment on the appropriateness of the modeling assumptions that were used [in the air dispersion model]”). It is clear enough that the EPA did subject the generic air dispersion model to notice and comment rulemaking. The EPA set out the basis for the model in the notice of proposed rulemaking, id. at 27,362-63, stated its rationale for making various assumptions, id. at 27,362, requested comments on the assumptions it made, id. at 27,363, addressed significant comments in the background paper referred to in its final rule document, see BID at 2-62 through 2-65, and revised some modeling parameters based upon the comments it received. See 57 Fed. Reg. at 61,981.

On the other hand, it does not seem that the EPA subjected the IRIS database to public comment. Insofar as the EPA is relying upon the IRIS Background Paper as support for its claim to have done so, it is being disingenuous with the court; that document was not available to commenters until two months after publication of the final rule and one and a half years after publication of the proposed rule.

In the alternative, the EPA argues that it was not required to solicit comment on the database itself because any time the agency uses the data in it as part of a rulemaking, the scientific basis for, and the application of, those data are then subject to comment. IRIS Background Paper at 5 (1993). Accordingly, the EPA goes on to argue that in this rulemaking it did in fact subject to comment both the RfC for MDI in the IRIS database and the use of that RfC to identify high risk pollutants, pointing once again to the proposed rule document where it requested “comments on all aspects of the proposed rule,” 56. Fed.Reg. at 27,340, 27,355, and to the IRIS Background Paper.

We agree that the EPA was not required to subject the IRIS database to notice and comment rulemaking. The database by itself has no preclusive effect; the data in the database constrain no one until so applied in a particular rule. In this instance, the application of the data — the comparison of the RfC for MDI to the threshold ambient concentration predicted by the air dispersion model — was clearly part of the proposed rule and as such was the subject of notice and comment. In any case, the CMA has not demonstrated that the EPA was, in fact, deaf to the Association’s comments with respect to the RfC for MDI.

The CMA’s second procedural argument is that the EPA failed to respond meaningfully to the Association’s comments objecting to the proposal to list MDI as a high risk pollutant. This claim, however, is subsumed within the CMA’s substantive challenge to the adequacy of the EPA’s rationale for its decision to designate MDI a high risk pollutant, which is based upon the same objections, and to which we now turn.

B. The Substantive Claims

We review the substance of a rule promulgated under the Clean Air Act using the same indulgent standards that we apply when reviewing a rule pursuant to the Administrative Procedure Act. We are to determine only whether the rule is arbitrary and capricious, an abuse of discretion, or otherwise not in accordance with the law. 42 U.S.C. § 7607(d)(9)(A). Not many rules fail the arbitrary and capricious standard, but for the reasons set out below we hold that the rule now under review fails twice, i.e., on both of the grounds advanced by the CMA.

1. The Generic Air Dispersion Model

The CMA’s challenge to the EPA’s use of the generic air dispersion model in order to predict the ambient concentration of MDI in the environment surrounding an MDI-emitting facility is straightforward: the assumptions used in the model bear no rational relationship to the known properties of MDI. In particular, the model rests upon the assumptions that the pollutant will be a gas emitted at 20°C from a single 10-meter stack and that it will disperse as a gas through the atmosphere under certain specified meteorologic conditions. 57 Fed.Reg. at 61,981. The CMA presented evidence—uncontested by the EPA—that MDI is a solid at 20°C, that it has a low vapor point (meaning that it tends not to evaporate), and that it typically disperses as an aerosol (i.e., as fine particles suspended in the air). The CMA also asserts that MDI enters the atmosphere not from a single source but primarily as a fugitive emission from many sources spread throughout a facility. Memorandum from Martha H. Keating to Early Reductions Program Docket No. A-90-47 at 4 (January 31, 1992). If MDI is to be denominated a high risk based solely upon application of the model to the health effects of the chemical without any validation of the result, the CMA argues, then the model must bear some rational relationship to the known properties of MDI emissions; because it does not, MDI was wrongly classified as a pollutant that poses a significant risk to public health.

The Congress did not specify any particular method by which the EPA must identify high risk pollutants. Therefore, we should uphold the EPA’s choice of a model for determining an exposure level above which a hazardous pollutant may be said to pose a high risk if the Clean Air Act can reasonably be read to authorize the agency's choice. Chevron U.S.A. Inc. v. Natural Resources Defense Council, Inc., 467 U.S. 837, 843, 104 S.Ct. 2778, 2781-82, 81 L.Ed.2d 694 (1984). See also Edison Electric Institute v. EPA, 2 F.3d 438, 445-46 (D.C.Cir.1993).

The agency is charged under the Act with evaluating no less than 189 hazardous air pollutants that “span a range of emission release and exposure situations ... across source categories.” .56 Fed.Reg. at 27,363. That the Congress gave it the discretion to use a model for that purpose is surely a reasonable reading of the Act. Such modeling is an established technique of environmental analysis; it conserves agency resources and facilitates timely decision making. _ Edison Electric Institute, 2 F.3d at 443; see also Eagle-Picher Industries v. EPA, 759 F.2d 905, 921-22 (D.C.Cir.1985); Small Refiner Lead Phase-Down Task Force v. EPA, 705 F.2d 506, 535 (D.C.Cir.1983). The air dispersion model in particular is well-suited, moreover, to the mandate given to the agency, which is to identify pollutants for which exposure to small quantities “may” be associated with significant adverse human health effects; the Congress apparently contemplated that the high risk list would be overinclu-sive, and because it proceeds from conservative assumptions the model tends to produce that result.

It necessarily follows that in designing the model the agency must have broad discretion to make simplifying assumptions—to model the pollutant as a gas, for example—and to establish parameters or set values for variables, such as choosing ten tons of emissions per facility per year as a conservative output from which to determine a likely exposure level. That the model does not fit every application perfectly is no criticism; a model is meant to simplify reality in order to make it tractable. See Milton Friedman, The Methodology of Positive Economics, in Essays in Positive Economics 3, 14-15 (1953) (“A hypothesis is important if it ‘explains’ much by little, that is, if it abstracts the common and crucial elements from the mass of complex and detailed circumstances surrounding the phenomena to be explained and permits valid predictions on the basis of them alone. To be important, therefore, a hypothesis must be descriptively false in its assumptions”); cf. Small Refiner Lead Phase-Down Task Force, 705 F.2d at 535 (court defers to agency’s decision balancing “the cost and complexity of a more elaborate model against the oversimplification of a simpler model”). Hence the normal criterion by which to evaluate a model is not the accuracy of the assumptions from which it proceeds but the utility of the results it produces. While the agency must if challenged “provide a full analytical defense” of its model, see Eagle-Picher Industries, 759 F.2d at 921, it need not justify the model on an ad hoc basis for every chemical to which the model is applied, even when faced with data indicating that it is not a perfect fit. To require as much would be to defeat the purpose of using a model.

If we are to earn our keep, however, judicial deference to the agency’s modeling cannot be utterly boundless; we must reverse the agency’s application of the generic air dispersion model as arbitrary and capricious if there is simply no rational relationship between the model and the known behavior of the hazardous air pollutant to which it is applied. See Motor Vehicle Mfrs. Ass’n v. State Farm Mut. Auto. Ins. Co., 463 U.S. 29, 43, 103 S.Ct. 2856, 2866-67, 77 L.Ed.2d 443 (1983); cf. Small Refiner Lead Phase-Down Task Force, 705 F.2d at 535 (court may strike down model “so oversimplified that the agency’s conclusions from it are unreasonable”). To illustrate, the reasonable assumption that a certain type of fish comes from the sea leads directly to the prediction that a fish of that type will die when put in an aquarium without salt water; but if one should learn that the particular fish comes from a lake, and thus that the prediction is certainly wrong and that the fish will die without fresh water, then it would be wrongheaded in the extreme to persist in the original assumption.

Generally, the court defers the determination of fit between the facts and the model to the EPA, so that the agency rather than the court may balance marginal losses in accuracy against marginal gains in administrative efficiency and timeliness of decision making. The more inflexibly the agency intends to apply the model, however, the more searchingly will the’court review the agency’s response when an affected party presents specific detailed evidence of a poor fit between the agency’s model and that party’s reality. Cf. McLouth Steel Products Corp. v. Thomas, 838 F.2d 1317, 1320-22 (D.C.Cir.1988) (model deemed a legislative rule because inflexible application so constrained EPA discretion).

The CMA claims that the EPA’s model, as applied to MDI, is a very poor fit indeed in several respects. First, the model treats MDI as though it were released into the atmosphere from a point source rather than as a fugitive emission. The agency acknowledged that “modeling fugitive emissions as a point source release will generally result in higher predicted concentrations” of the pollutant. BID at 2-65. Its decision not to vary the model for fugitives was a reasonable one, however, in view of the record before it. Most facilities do emit most hazardous air pollutants from a point source, id., and, most important, the CMA did not produce any specific evidence indicating that MDI is atypical in that respect.

With respect to the CMA’s other objections, however, the EPA has not pointed to any record evidence that shows a rational relationship between the generic air dispersion model and the physical properties of MDI, even in the face of the specific scientific evidence to the contrary adduced by the CMA. See Edison Electric, 2 F.3d at 446 (EPA has burden to “demonstrate ... on the record” existence of rational relationship between generic mismanagement scenario and wastes at issue). When the EPA eliminated the proposed “reality check” in the final rule, it did not substitute any other mechanism by which to validate the predicted level of human exposure to a pollutant by reference to actual emissions; instead it applied the generic air dispersion model inflexibly to all 189 hazardous air pollutants without regard to whether the basic assumptions of the model comported, within an acceptable margin of error, to the physical characteristics of the chemicals modeled.

When the CMA presented specific objections to the EPA’s application of its model to MDI, the EPA responded in a high-handed and conclusory manner.- For example, as noted above, the EPA accepted the CMA’s point that MDI is a solid at the ambient temperature (20°C) at which the model assumes that the generic pollutant is a gas, BID at 2-77, yet it dismissed the CMA’s objection that a solid cannot act as a gas with the facile assertion that “[i]t is likely that MDI is emitted at temperatures higher than ambient and so would disperse much like any other pollutant,” that is, as a gas. Id. (In fact, the record shows that MDI is still a solid at 37°C.) This is exactly the sort of “speculative factual assertion[ ]” that we found insufficient to support the “waste management scenario” or model in Edison Electric. 2 F.3d at 446. And, we may add, it bespeaks a “let them eat cake” attitude that ill-becomes an administrative agency whose obligation to the public it serves is discharged if only it avoids being arbitrary and capricious.

In response to the CMA’s further challenge that MDI, if emitted, would likely be in the form of a rapidly settling aerosol, the EPA replied in an unsupported and concluso-ry fashion that it “still believes that the dispersion model used [is] appropriate for MDI.” BID at 2-77. This response added nothing to the agency’s defense of its thesis except perhaps the implication that it was committed to its position regardless of any facts to the contrary. We can hardly say that the EPA has demonstrated that its generic air dispersion model bears a rational relationship to the physical properties of MDI when the agency itself disdains making any effort to do so. See Tex Tin Corp. v. EPA, 992 F.2d 353, 354-55 (D.C.Cir.1993) (EPA’s reliance upon generic studies in face of conflicting detailed and specific scientific evidence held arbitrary and capricious). For want of a rational relationship between the model and the molecule, we hold that the rule designating MDI a high risk is arbitrary and capricious.

2. Toxicity of MDI

Recall that the Congress delegated to the EPA the task of deciding which of 189 listed hazardous air pollutants are “high risk,” by which it meant those “for which high risks of adverse public health effects may be associated with exposure to small quantities including, but not limited to, chlorinated dioxins and furans.” 42 U.S.C. § 7412(i)(5)(E). An ordinary reading of the Congress’s mandate to the EPA would suggest that the agency must first identify whether exposure to a small amount of a hazardous air pollutant can ever cause an “adverse public health effect” and then determine the likelihood of that effect occurring, the concern being only with “high risks.” Oddly, though, neither the EPA nor the CMA urge that reading of the statute upon us; instead, both confound the concepts of risk and outcome and treat the two as a single question of the “seriousness” of the health effect. Thus, the EPA asserts only that nasal irritation resulting from excessive exposure to MDI is the type of serious health effect associated with a “high risk” hazardous air pollutant, while the CMA claims that it is too minor a health effect to be classified as “high risk” within the meaning that the Congress intended by that term.

We take the dispute as we find it, of course. And because the Congress did not specifically define its terms, and the statute is not otherwise clear on its face, again we must uphold the EPA’s interpretation of the statute if it is merely reasonable. Chevron, 467 U.S. at 843, 104 S.Ct. at 2781-82.

In order to determine which pollutants are sufficiently toxic to be designated “high risk,” the EPA as we have seen interpreted the statute to encompass (1) chlorinated dioxins and furans; (2) known human carcinogens; and (3) those pollutants that are relatively toxic to humans at or above a certain level of exposure. (The first two categories are not at issue here.) In determining the relative toxicity of pollutants in the third category, the EPA interpreted the words “high risks of adverse public health effects” as encompassing certain “serious” “health effects endpoints”:

The health effects endpoints considered were carcinogenicity, reproductive and developmental toxicity, acute lethality, and systemic effects other than acute lethality (e.g., neurologic disorders).

56 Fed.Reg. at 27,361; 57 Fed.Reg. at 61,-980-81.

The EPA would now have us believe that this list of “serious” health effects endpoints was meant not to be exclusive but only to show that the agency considered “impacts other than carcinogenicity.” In effect, the agency is suggesting that it meant all along that any pollutant for which exposure is associated with either a low risk of a serious health effect or a high risk of a non-serious health effect should be included on the high risk list.

Even if the list of health effects endpoints was meant to be illustrative rather than exhaustive, however, the preamble to the final rule clearly demonstrates that the EPA was looking only at serious health effects. The EPA stated there that “the purpose of this analysis was to identify pollutants whose emissions could potentially present a high risk, not a de minimis risk.” 57 Fed.Reg. at 61,982. While east in terms of risk, the EPA was clearly using “high risk” to mean “associated with a serious health effect.” And understandably so; after all, an insignificant health problem is still insignificant, even if it is a virtual certainty, whereas even a low risk of a serious problem represents something of a “public health effect.” Moreover, reading that statement literally in terms of risk implies that the agency could not list known human carcinogens from which the risk of cancer is very slight — a result it never even contemplated. See id. (EPA listed as high risk certain substances solely “based on the fact that these are known human carcinogens”).

Accordingly, we understand the EPA to have interpreted its statutory mandate as requiring it to designate as high risks those pollutants exposure to which at low levels could cause a serious health problem for humans — in effect disregarding the probability of such effect occurring. Although, as we have said, that is surely not the interpretation that flows most naturally from the statute, neither is it an unreasonable interpretation.

Nonetheless, having stated that it would consider for the high risk list only those pollutants associated to any degree with a serious health effect, the EPA failed in fact to screen out non-serious health effects. Thus, in determining that MDI is relatively toxic, the EPA relied solely upon the presence in the IRIS database of an agency-verified RfC for MDI, without regard to the seriousness of the health effect associated with that RfC. As EPA counsel frankly acknowledged at oral argument, at no time did the agency examine the RfC data in the IRIS database in order to determine whether the effect of exposure to a chemical was among (or even like) the health endpoints that the agency had earlier determined were “serious.” Indeed, no human health effect was considered too minor to warrant putting the implicated chemical on the “high risk” list. Therefore, the de facto standard that the EPA used to place a chemical on the high risk list was not whether exposure to it may result in a serious human health effect, but whether exposure to it may cause “any human health effect for which there is an RfC value.” Insofar as the agency’s approach had the effect of treating as a “high risk” a pollutant of which the only known health effect is non-serious, it is inconsistent with the agency’s own stated intentions.

The record supports the CMA’s claim that the EPA placed MDI on the high risk list solely in reliance upon an RfC associated with a non-serious health effect. The single study upon the basis of which the EPA established the RfC for MDI involved nasal lesions in rats subjected to prolonged exposure to MDI in aerosol form. The EPA referred to these lesions as “degenerative” with some instances of “basal cell hyperpla-sia,” but they remain, in lay terms, only nasal irritations with a reduced sense of smell in some cases. BID at 2-76. Nasal irritation is not a health effect included within (or even in the same ballpark as) the EPA’s list of “serious” health effect endpoints: carcinogenicity, reproductive and developmental toxicity, acute lethality, and other systemic disorders.

The EPA defends its use of the RfC for MDI in the IRIS database on the ground that it “judged that the nasal effects [on which the RfC is based] ... are of concern with respect to chronic exposure situations” and that it “considers that the spectrum of effects in the nasal cavity is more extensive than minimal irritation.” BID at 2-76. Those concerns may be warranted, but neither supports the conclusion that MDI poses a high risk of adverse public health effects, especially in the face of record evidence suggesting that even the modest health effect of exposure to MDI (i.e., nasal lesion) is reversible. See Memorandum from Martha H. Keating to Early Reduction Docket A-90-47 at II-B-4 (Attachment 4) (“The return to normal of these [MDI exposure] tests for [a group removed from exposure] was suggested to indicate reversibility of effects seen at low level exposures”).

The EPA suggests that, since irritants generally produce nasal effects sooner than the (presumably more severe) lung effects, using an RfC based upon nasal effects will capture both nasal and lung effects. While that may well be, it still does not follow— considering the nature of the four types of health effect endpoints with which the agency was concerned — that nasal lesions are a serious health endpoint such as to justify characterizing a hazardous pollutant as presenting a particularly high risk; and it was upon the basis of the RfC for nasal lesions alone that the agency acted. Accordingly, we conclude that it was arbitrary and capricious for the EPA to list MDI as a high risk pollutant solely upon the basis of the RfC for MDI, without identifying any serious health effect with which it has ever been associated.

III. Conclusion

The decision of the EPA to list MDI as a high risk air pollutant is arbitrary and capricious for the two reasons elaborated above, and therefore we vacate it. No remand for further administrative proceedings is warranted because the EPA did not suggest in the rulemaking under review that there is any alternative basis in the record for including MDI on the high risk list. If the EPA has another basis upon which to proceed, it may of course begin anew.

Although the CMA asks us also to extend the statutory period for a manufacturer to enroll in the Early Reduction Program and thereby to seek an extension of the deadline for complying with the MACT requirements, it presents no good reason for us to do so. The EPA has not yet promulgated MACT standards for most industries that emit MDI, and therefore the enrollment period for a compliance extension has not closed for those industries. See 42 U.S.C. § 7412(i)(3)(A) and (B). Because the CMA points to no specific disadvantage suffered by any member wishing to participate in the Early Reduction Program, let alone one warranting an equitable modification of a statutory requirement, we deny such relief.

Judgment accordingly.