American Petroleum Institute v. Environmental Protection Agency

United States Court of Appeals

FOR THE DISTRICT OF COLUMBIA CIRCUIT

Argued November 3, 2016 Decided July 7, 2017

No. 09-1038

AMERICAN PETROLEUM INSTITUTE,

PETITIONER

v.

ENVIRONMENTAL PROTECTION AGENCY,

RESPONDENT

AMERICAN CHEMISTRY COUNCIL, ET AL.,

INTERVENORS

Consolidated with 15-1083, 15-1085, 15-1088, 15-1089,

15-1094

On Petitions for Review of a Final Regulation Promulgated

by the United States Environmental Protection Agency

Jeremy C. Marwell and Thomas Sayre Llewellyn argued

the causes for Industry Petitioners. With them on the briefs

were Stacy R. Linden, Matthew A. Haynie, Aaron J. Wallisch,

Laura E. Boorman, Kevin A. Gaynor, John P. Elwood, Roger

R. Martella, Jr., Joel Visser, Linda E. Kelly, Quentin Riegel,

and Leslie A. Hulse. Wayne D=Angelo, Harry M. Ng, and

Michael R. See entered appearances.

2

Kenneth M. Kastner was on the brief for amici curiae

Eastman Chemical Company and Solvay USA Inc. in support

of Industry Petitioners.

James S. Pew argued the cause for Environmental

Petitioners. With him on the briefs was Khushi K. Desai.

Daniel R. Dertke, Attorney, U.S. Department of Justice,

argued the cause for respondents. With him on the brief were

John C. Cruden, Assistant Attorney General, Douglas M.

Bushey, Attorney, U.S. Department of Justice, and Alan

Carpien, Attorney, U.S. Environmental Protection Agency.

Donald J. Patterson, Jr. argued the cause for Industry

Intervenor-Respondents. With him on the brief were Eric L.

Klein, Aaron J. Wallisch, Laura E. Boorman, John L.

Wittenborn, Wayne D=Angelo, Kevin A. Gaynor, John P.

Elwood, Jeremy C. Marwell, Leslie A. Hulse, Linda E. Kelly,

Quentin Riegel, Roger R. Martella, Jr., Joel Visser, James W.

Conrad, Jr., Thomas Sayre Llewellyn, Stacy R. Linden, and

Matthew A. Haynie. Douglas H. Green entered an appearance.

David R. Case, James S. Pew, Khushi K. Desai, and

Vincent Atriano were on the joint brief for respondent-

intervenors and movant-intervenor Gulf Chemical and

Metallurgical Corp.

Before: TATEL and KAVANAUGH, Circuit Judges, and

WILLIAMS, Senior Circuit Judge.

Opinion for the Court filed PER CURIAM.

Opinion dissenting in part filed by Circuit Judge TATEL.

3

PER CURIAM: This case arises from the Environmental

Protection Agency’s latest effort to define the term “solid

waste” under the Resource Conservation and Recovery Act. In

2015, EPA promulgated a final rule governing when certain

hazardous materials qualify as “discarded” and hence are

subject to the agency’s regulatory authority. Environmental

and Industry Petitioners have each petitioned for review of that

rule, arguing that numerous aspects of it are unlawful and

arbitrary and capricious. For the reasons explained, we grant

the Industry petition for review with respect to Factor 4 of the

legitimacy test and to the Verified Recycler Exclusion and we

dismiss the Environmental petition for review.

I. Introduction

The Resource Conservation and Recovery Act (“RCRA”),

42 U.S.C. §§ 6901-6992k, empowers EPA to manage solid and

hazardous waste. The statute defines solid waste as “garbage,

refuse, sludge . . . and other discarded material.” 42 U.S.C.

§ 6903(27). Hazardous waste is a subset of solid waste that

may pose a substantial threat to human health or the

environment when improperly managed. § 6903(5)(B). If a

material qualifies as hazardous waste, it is subject to regulation

under RCRA Subtitle C, §§ 6921-6939g, which imposes

comprehensive reporting and operating requirements. Material

that is not solid waste, and therefore not hazardous waste, is

exempt from Subtitle C.

Pursuant to its RCRA authority, EPA has promulgated a

rule defining solid waste as “discarded material” not otherwise

excluded from the agency’s regulations. 40 C.F.R.

§ 261.2(a)(1). A separate regulation lists materials that fall

outside the definition of solid waste. § 261.4. Central to the

issues before us, EPA considers certain materials that are

destined for recycling to be discarded and hence solid waste

4

subject to RCRA regulation. Definition of Solid Waste, 80 Fed.

Reg. 1,694, 1,738/3 (Jan. 13, 2015) (the “Final Rule”).

For our purposes, the relevant history begins in 2007,

when EPA proposed a rule deregulating many hazardous

secondary materials. See American Petroleum Institute v.

EPA, 683 F.3d 382, 385 (D.C. Cir. 2012) (“API II”).

Secondary materials are substances generated as the remainder

of industrial processes; they include spent materials,

byproducts, and sludges. See 40 C.F.R. § 260.10. EPA’s

proposed rule—which became a final rule in October 2008—

excluded hazardous secondary materials from the definition of

solid waste in two circumstances: first, if the company that

generated the materials controlled the recycling of those

materials; and second, if the generator transferred the materials

to an off-site recycler it had audited to ensure compliance with

proper recycling practices. Revisions to the Definition of Solid

Waste, 73 Fed. Reg. 64,668, 64,669/3-70/1-2 (Oct. 30, 2008)

(the “2008 Rule”). These two exemptions were known,

respectively, as the “Generator-Controlled Exclusion” and the

“Transfer-Based Exclusion.” Id. at 64,670/1, 64,675/2

(capitalization added). To qualify for either, secondary

materials had to be recycled “legitimately,” a term EPA defined

by reference to certain “legitimacy factors.” Id. at 64,675/2-3.

EPA adopted this legitimacy requirement to distinguish “true”

recycling from “sham” recycling in which companies claim to

reuse materials they in fact discard. Id. at 64,700/2.

Several organizations challenged the 2008 Rule. One, the

American Petroleum Institute, argued that the rule unlawfully

regulated materials called spent petroleum refinery catalysts,

which are byproducts of the oil refining process. API II, 683

F.3d at 387. Another group, the Sierra Club, asserted that the

rule “was not sufficiently protective of human health and the

environment,” in violation of RCRA. Id. at 389. A third entity,

5

Gulf Chemical and Metallurgical Corporation (“Gulf”), moved

to intervene to defend the rule’s treatment of spent catalysts.

Before this court heard oral argument, EPA entered a

settlement agreement with the Sierra Club. Id. Pursuant to that

agreement, the Sierra Club withdrew its petition, and EPA

agreed to propose a new solid waste rule. Id. As promised,

EPA published a notice of proposed rulemaking in July 2011.

Definition of Solid Waste, 76 Fed. Reg. 44,094 (July 22, 2011)

(the “Proposed Rule”). A year later, we held that API’s

challenge to the 2008 rule was unripe given the forthcoming

final rule. API II, 683 F.3d at 384. We deferred any action on

Gulf’s motion to intervene, which is dealt with in a separate

order published today.

EPA promulgated the Final Rule on solid waste—the one

before us now—in January 2015. 80 Fed. Reg. at 1,694/1. The

2015 Final Rule differs from the 2008 Rule in several ways,

four of which are relevant here. First, the Final Rule revises

the definition of “legitimate” recycling and expands the scope

of the legitimacy factors to cover all recycling. Id. at 1,719/3-

20/1. Second, it establishes that spent catalysts—which were

ineligible for exclusions under the 2008 Rule—could qualify

for the exemptions in the 2015 regulation. Id. at 1,738/1.

Third, the rule defers a decision on whether to add conditions

to 32 previously promulgated exclusions from the definition of

solid waste, which EPA calls the “pre-2008” exclusions. Id. at

1,741/2. Fourth and finally, the rule replaces the transfer-based

exclusion with the “Verified Recycler Exclusion,” a new

standard governing when transferred materials qualify as solid

waste. Id. at 1,695/2. We provide additional detail on each of

these provisions later in this opinion.

Multiple organizations petitioned for review of the 2015

rule. Their petitions, which are consolidated in this case,

6

challenge the regulation on multiple fronts. Industry

Petitioners argue that both the legitimacy test and the Verified

Recycler Exclusion exceed EPA’s RCRA authority. Industry

Petitioners also challenge EPA’s treatment of two specific

materials: spent catalysts and off-specification commercial

chemical products. Environmental Petitioners argue that the

Verified Recycler Exclusion is too permissive and that EPA

should have added containment and notification conditions to

the 32 pre-2008 exclusions. We consider these challenges in

turn.

II. Legitimacy Factors

Industry Petitioners first attack EPA’s new legitimacy test.

Before EPA can regulate a hazardous secondary material as

hazardous waste, it must determine that the material has been

“discarded” under 42 U.S.C. § 6903(27). Items recycled

through “immediate reuse in” an “industry’s ongoing

production process,” are not discarded within the meaning of

that section and are outside EPA’s hazardous waste

regulations. See American Mining Congress v. EPA, 824 F.2d

1177, 1183-85 (D.C. Cir. 1987) (“AMC”); see also Ass’n of

Battery Recyclers, Inc. v. EPA, 208 F.3d 1047, 1052-53 (D.C.

Cir. 2000) (explaining that “immediate” in AMC means

“direct,” not instantaneous). But because EPA’s waste disposal

regulations are acknowledged to be very costly to meet, “there

is an incentive for some handlers to claim they are recycling

when, in fact, they are conducting . . . disposal.” Final Rule,

80 Fed. Reg. at 1,719/3. To prevent such evasion, EPA polices

the line “between ‘legitimate’ (i.e., true) recycling and ‘sham’

(i.e., fake) recycling.” Id. at 1,720/1.

Until recently, EPA’s policy on sham recycling existed

chiefly in uncodified guidance, notably a memo issued in 1989

by Sylvia K. Lowrance, Director, EPA Office of Solid Waste

(Apr. 26, 1989) (the “Lowrance Memo”). The memo discussed

7

over a dozen factors for evaluating recycling, all aimed at

determining “whether the secondary material is ‘commodity-

like,’” i.e., is it being handled like a valuable industrial input or

like a worthless industrial byproduct. See id. at 2 &

attachment.

The Final Rule updates and codifies this effort to draw the

distinction between legitimate and sham recycling. It requires

that all recycling of hazardous secondary materials meet a

legitimacy test set forth in 40 C.F.R. § 260.43(a) or else be

labeled “sham” and subjected to full RCRA regulation. 40

C.F.R. § 261.2(g). Like the Lowrance Memo, the rule is rooted

in the assumption that legitimate recycling should involve

some “recognizable benefit,” Final Rule, 80 Fed. Reg. at

1,722/1, independent of merely “avoid[ing] the requirements

of” RCRA regulation, id. at 1,719/3.

To satisfy the legitimacy test for recycling of a particular

material, firms must prevail on all of four factors,

§ 260.43(a)(1)-(4), which are in addition to whatever elements

a specific exclusion might require, see Final Rule, 80 Fed. Reg.

at 1,720/2. First, the hazardous secondary material must

“provide[] a useful contribution to the recycling process.”

§ 260.43(a)(1). Second, “[t]he recycling process must produce

a valuable product or intermediate.” § 260.43(a)(2). Third, the

persons controlling the secondary material must “manage the

hazardous secondary material as a valuable commodity.”

§ 260.43(a)(3). Fourth, “[t]he product of the recycling process

must be comparable to a legitimate product or intermediate.”

§ 260.43(a)(4). Factors 1 and 3 address the process, Factors 2

and 4 the product.

Industry Petitioners do not attack EPA’s authority to

formulate and apply a legitimacy test, nor do they fault EPA’s

premise that legitimate recycling involves “valuable” materials

8

being used for a “recognizable benefit.” Final Rule, 80 Fed.

Reg. at 1,697/3, 1,722/1. At that level of generality, EPA’s

policy seems to be a reasonable method for identifying

materials that are “part of the waste disposal problem” and thus

subject to EPA’s RCRA authority over discarded materials.

Safe Food & Fertilizer v. EPA, 350 F.3d 1263, 1268 (D.C. Cir.

2003). Industry Petitioners instead attack EPA’s planned

means to implement that policy. They complain that

mandating Factors 3 and 4 across all recycling results in EPA’s

“unlawfully regulat[ing] non-discarded materials.” Industry

Pet’rs’ Br. 16 (capitalization omitted).

A. Factor 3

We begin with Factor 3, which requires secondary

materials to be handled as “valuable commodit[ies].” 40

C.F.R. § 260.43(a)(3). Where there is an analogous raw

material, the firm can meet this standard by handling the

secondary material “in an equally protective manner.” Id. If

there is no raw analogue for comparison, EPA requires that the

secondary material be “contained.” Id. “Contained” means

“held in a unit (including a land-based unit . . .) that meets”

multiple enumerated criteria, including that the unit be “labeled

or otherwise ha[ve] a system (such as a log) to immediately

identify the hazardous secondary materials” therein. 40 C.F.R.

§ 260.10 (entry for “Contained”). “[L]and-based unit[s],” id.,

encompass, at least for some materials such as scrap metal,

simply lying on the ground, see Final Rule, 80 Fed. Reg. at

1,721/3, 1,736/2.

EPA previously claimed that any “interdiction in time”

during a secondary material’s trajectory from initial output to

recycling, e.g., for storage, could be considered discard and

thus trip the material into EPA’s RCRA authority. Battery

Recyclers, 208 F.3d at 1052 (internal quotation marks omitted).

We rejected that rule. “To say that when something is saved it

9

is thrown away is an extraordinary distortion of the English

language.” Id. at 1053. Industry Petitioners read that holding

to bar EPA from ever regulating how recycled materials are

contained. Their reading goes too far. EPA can impose a

containment requirement so long as it is such that an inference

of “sham” or illegitimacy would logically flow from a firm’s

non-compliance. And given EPA’s explanation that a material

may be “contained” if it is simply piled on the ground, Final

Rule, 80 Fed. Reg. at 1,721/3, 1,736/2, and meets specific

requirements that petitioners do not challenge as unreasonable

(with one exception, the “labelling” requirement discussed

below), the standard does not on its face appear to ask for

anything beyond what could be expected of firms engaged in

legitimate recycling.

Industry Petitioners express concern about having to label

or log unwieldy molten metals and acidic sludges to satisfy

EPA’s insistence on material being “contained.” But EPA

offers an alternative to labelling in the conventional sense—

provision of “a system (such as a log) to immediately identify

the hazardous secondary materials in the unit.” § 260.10.

Thus, in substance, the requirement is not precisely one of

labeling or logging, but only of assuring that it somehow be

possible for the material to be “immediately identif[iable].” Id.

While doubtless EPA’s language could be interpreted

unreasonably, we cannot see that the requirement itself is

unreasonable.

B. Factor 4

Factor 4 presents more difficulty. EPA explains this factor

as an effort to prevent recyclers from loading products with

hazardous secondary materials that “provide[] no recognizable

benefit to the product,” Final Rule, 80 Fed. Reg. at 1,722/1, and

are simply “along for the ride,” id. at 1,726/2. Although EPA

does not require a material’s “hazardous component[s]”

10

themselves to provide a “useful contribution” to the product,

see id. at 1,723/3 (discussing Factor 1), the agency is concerned

that a purported recycler might “incorporate[] hazardous

constituents into the final product when they were not needed

to make that product effective as a way to avoid proper disposal

of that material, which would be sham recycling,” id. at

1,726/1-2.

The factor sets up two tracks, 40 C.F.R. § 260.43(a)(4)(i)-

(ii), one covering products for which there is an analogue of

undoubted legitimacy, the other addressing products with no

such analogue. EPA refers to these together as the “technical

provisions.” Final Rule, 80 Fed. Reg. at 1,729/1. But as EPA

recognizes that the criteria set forth under these two tracks

don’t draw a satisfactory line between genuine and sham, it

also offers a rather complicated exception—aimed at

preventing products from being labelled a sham when they in

fact pose no “significant human health or environmental risk.”

§ 260.43(a)(4)(iii). But Factor 4’s complex provisions fall

short of the aim. As we shall see, Factor 4 imposes tasks

tangential to disposal vel non (and thus tangential to EPA’s

authority), even when EPA has offered little reason to doubt a

product’s legitimacy.

The second track is the more reasonable of the two. When

there is no analogue, the recycled product will pass if it was

created by looping secondary materials back “to the original

process . . . from which they were generated” or if it meets

“widely recognized commodity standards and specifications.”

§ 260.43(a)(4)(ii)(A)-(B). Those standards or specifications

need not address the hazardous aspects of the product. Final

Rule, 80 Fed. Reg. at 1,728/2-3. And EPA has explained that

compliance with “customer specifications” may suffice for

“specialty” products. Id. at 1,728/1. Although that gloss on

“specifications” appears only in EPA’s discussion of the with-

11

analogue track, the Final Rule offers little indication that the

same word in the no-analogue track is meant to read differently

on this matter. Compare id. at 1,727/3-28/1 (with-analogue),

with id. at 1,728/2-3 (no-analogue). Putting all this together, if

a recycled product, lacking an analogue, fails to satisfy

customer specifications, falls short of relevant commodity

standards, and is not derived from a closed-loop type process,

EPA treats it as discarded (subject to the ultimate exception).

These tests focus largely on the utility of the recycling in

question, a reasonable inquiry when deciding legitimacy. See

id. at 1,728/3 (commodity standards and specifications criteria

mean that “market forces [will] dictate” legitimacy); id. at

1,729/1 (“looping” criterion appropriate because this type of

recycling “conserves the use of raw materials” without adding

new hazards).

The other track in Factor 4’s technical provisions,

applying where the recycled product has an analogue, is more

explicitly tuned to the “along for the ride” metaphor. It requires

that the recycled product exhibit no hazardous “characteristic”

that is absent from the product’s analogue. 40 C.F.R.

§ 260.43(a)(4)(i)(A); see also Final Rule, 80 Fed. Reg. at

1,727/1 (“The characteristics are ignitability, corrosivity,

reactivity, and toxicity.”). This criterion—fenced in as it is by

the definitions of those characteristics, see 40 C.F.R.

§§ 261.21-.24—also seems reasonable: one would expect

analogous products to have similar attributes. But the track

goes on from there. Even if the recycled product and its

analogue share the same hazardous characteristics, the amount

or “levels” of hazardous constituents in the product must be

“comparable to or lower than” its analogue’s.

§ 260.43(a)(4)(i)(B). If the product fails that test, it can still be

legitimate if it “meet[s] widely-recognized commodity

standards and specifications.” Id. Unlike in the no-analogue

track, here the commodity standards and specifications must

12

“specifically address [] hazardous constituents.” Id. Otherwise

EPA will regard the product as discarded (subject to the

ultimate exception).

We have left EPA some leeway in applying the idea that

genuine recyclers cannot include hazardous material just

“along for the ride” in their products. Thus in American

Petroleum Institute v. EPA, 216 F.3d 50 (D.C. Cir. 2000) (“API

I”), we rejected a challenge under “Chevron step one” to a rule

that treated “recovered oil” as discarded if it included “extra

materials . . . that provide no benefit to the industrial process.”

Id. at 58-59 (emphasis added). But we hinted that such a rule

should reasonably avoid “incidentally regulat[ing] oil

containing chemicals [whose presence in the recycled oil was]

not caused by sham recycling (and therefore not discarded).”

See id. at 59.

Judged by that perhaps opaque standard, EPA’s “along for

the ride” metaphor suffers at least one of the usual dangers of

metaphors—imprecision. The record contains examples of

hazardous secondary materials that are beneficially recycled

into valuable products (recognized as such by EPA), even

though those products contain hazardous constituents that do

not, in themselves, contribute to the value of the final product.

See, e.g., Final Rule, 80 Fed. Reg. at 1,721/1-2 (zinc-

containing secondary materials), 1,729/3 (lead-containing

secondary materials). In those cases, even if EPA could

technically say that some small excess of hazardous

constituents has been left in the final product, the mere fact of

their presence would not constitute a reasonable basis for

dubbing the product or the process a sham. After all, it can be

costly to extract tiny amounts of hazardous constituents—

potentially on the order of “parts per million,” see id. at

1,727/2-3—from secondary materials destined for recycling,

and no statute has given EPA authority to compel firms to

13

engage in such extraction where failing to do so imposes no

health or environmental risk. To rule otherwise would be to

disregard the statute’s stated “objective[]” of “encouraging . . .

properly conducted recycling.” 42 U.S.C. § 6902(a)(6).

EPA made this very point in Safe Food to defend its

exclusion for recycled zinc fertilizers even though those

fertilizers could have “considerably higher” contaminant levels

than the corresponding “virgin commercial fertilizer.” 350

F.3d at 1269. After reviewing EPA’s data on the threat posed

by the additional contaminants, we agreed that the excesses of

the contaminant levels that EPA allowed (as consistent with

legitimate recycling) over those in virgin fertilizer samples

“lose their significance when put in proper perspective—

namely, a perspective based on health and environmental

risks.” Id. at 1270.

No such perspective is allowed by the “comparable to or

lower than” standard for products with analogues. That

standard sets the bar at the contaminant level of the analogue

without regard to whether any incremental contaminants are

significant in terms of health and environmental risks. This

problem is reduced, but not eliminated, by firms’ option to

meet “widely-recognized commodity standards and

specifications,” 40 C.F.R. § 260.43(a)(4)(i)(B)—including

“customer specifications” if the product is made-to-order,

Final Rule, 80 Fed. Reg. at 1,728/1. Many products might fail

this alternative, not because they represent sham recycling, but

because the relevant commodity standards or specifications

don’t address the hazardous constituent levels of concern to

EPA. Industry Petitioners contend, and EPA does not

contradict, that such standards usually refer to minimum levels

of desired elements rather than maximum levels of specific

impurities. Doubtless this track will ensnare some sham

recycling, but it does so with a test that is not a “reasonable tool

14

for distinguishing products from wastes.” See Safe Food, 350

F.3d at 1269.

EPA, having recognized some of the shortcomings in

these provisions, created an exception purporting to account for

them. See Final Rule, 80 Fed. Reg. at 1,729/1. A recycler may

avoid the sham label if it “prepare[s] documentation showing

why the recycling is, in fact, still legitimate” and notifies

regulators. 40 C.F.R. § 260.43(a)(4)(iii). The legitimacy “can

be shown” by “lack of exposure from toxics in the product, lack

of the bioavailability of toxins in the product, or other relevant

considerations which show that the recycled product does not

contain levels of hazardous constituents that pose a significant

human health or environmental risk.” Id.

In explaining this exception, EPA has indicated that the

question is whether the recycled product will be used

beneficially in a manner that reasonably protects against the

risks its residual hazardous constituents present. See Final

Rule, 80 Fed. Reg. at 1,729/1-3. Absence of these

circumstances would indicate that the true purpose of the

recycling is disposal. Hence, EPA explained in the rulemaking

that “lead contaminated foundry sand[]” would be sham

recycled when packaged as “children’s play sand” but that the

same material can be legitimately recycled for “mold making

in a facility’s sand loop.” Final Rule, 80 Fed. Reg. at 1,729/2-

3. The sand is (in a sense) equally hazardous in both cases, but

the latter use is legitimate “because . . . there is little chance of

the hazardous constituents being released into the environment

or causing damage to human health”; “there is lead throughout

the foundry’s process” (i.e., the sand isn’t introducing new

hazards); and “there is a clear value to reusing the sand” in that

industry. Id. at 1,729/3. Recyclers can also meet this exception

by analyzing the “increased risk” of their product relative to its

analogues, if any. Id. We read this as saying, in light of EPA’s

15

brief, that a recycler can show its product is legitimate by

documenting that any incremental risk it presents is not

“significant” to health and the environment. See Respondent

Br. 42-43 (citing Safe Food, 350 F.3d at 1269-71).

Contrary to Industry Petitioners’ claims, the general

criteria embodied in the Factor 4 exception seem permissible,

indeed consistent with our ruling in Safe Food. Industry

Petitioners also argue that the exception affords EPA unlimited

discretion to find discard. The language of Factor 4 and its

exception is rather open-ended, so judicial review of EPA’s

subsequent interpretations would normally be highly

deferential, Auer v. Robbins, 519 U.S. 452, 461-62 (1997),

potentially leaving petitioners at the mercy of a different

reading in the future. But we note that Factor 4’s exception is

tuned specifically to “significant human health or

environmental risk[s].” 40 C.F.R. § 260.43(a)(4)(iii). And

EPA has simultaneously provided an explanation of how to

apply the exception along with an example of how a specific

material might pass or fail it. Final Rule, 80 Fed. Reg. at

1,729/2-3 (foundry sand). These aspects of the rulemaking

sufficiently constrict the range of possible interpretations: “[a]n

interpretation at odds with the agency’s expressed intent at the

time of adoption enjoys no judicial deference.” AT&T Corp. v.

FCC, 841 F.3d 1047, 1054 (D.C. Cir. 2016).

The exception nonetheless falls short of saving the rule,

due to the draconian character of the procedures it imposes on

recyclers. See Industry Pet’rs’ Br. 29, 33. To qualify for the

exception just described, a firm must contemporaneously

document how its recycling is “still legitimate,” notify

regulators of that finding, and keep the documents “on-site for

three years after the recycling operation has ceased.” 40 C.F.R.

§ 260.43(a)(4)(iii). Failing any of these steps will make a sham

16

out of what would otherwise have been a legitimate product.

See Final Rule, 80 Fed. Reg. at 1,721/1, 1,735/3-36/1.

EPA is correct that these notice and recordkeeping

mandates will create useful “oversight” and may be correct that

they constitute only a “minimal burden” on recyclers. Id. at

1,730/1, 1,732/1. But paperwork is not alchemy; a legitimate

product will not morph into waste if its producer fails to file a

form (or loses a copy two years later). EPA insists that it can

impose burden-shifting rules even in drawing the line between

what it may and may not regulate. Respondent’s Br. 58. True

enough; but the generality is applicable only if the products

subjected to the burden-shifting are such that it would normally

be reasonable to expect them to qualify as “discarded” in the

absence of affirmative evidence from the recycler. Thus in

American Chemistry Council v. EPA, 337 F.3d 1060 (D.C. Cir.

2003), we affirmed EPA’s decision to put the burden on

regulated entities to initiate a “delisting” process preemptively

to establish that a given “mixture or derivative” of hazardous

waste is not itself hazardous. Id. at 1065. Waste handlers

would evidently have to undertake this process, concededly

“cumbersome,” in advance of any EPA enforcement. Id. But

there EPA had found that “many mixtures of and derivatives

from hazardous wastes are themselves hazardous,” an

inference that those materials’ origin in hazardous waste

renders highly plausible. Id. Further, the rule included

exceptions to “prevent [EPA] from casting too wide a net over”

materials outside its jurisdiction. Id. Compare Dissent at 8.

The same might be said of the no-analogue track and the

hazardous characteristic criterion. But we cannot say the same

for the with-analogue track’s “comparable to or lower than”

test, even as qualified by the exception for products meeting

commodity standards or specifications.

17

Never in the rulemaking does EPA make out why a

product that fails those criteria is likely to be discarded in any

legitimate sense of the term. See Motor Vehicle Mfrs. Ass’n v.

State Farm Mut. Auto. Ins. Co., 463 U.S. 29, 42 (1983) (agency

rules must be “justified by the rulemaking record”).

Environmental Intervenors argue that the necessary

backing for Factor 4 lies in EPA’s report, An Assessment of

Environmental Problems Associated with Recycling of

Hazardous Secondary Materials (Dec. 10, 2014) (the

“Problems Study”). See Respondent-Intv’rs’ Joint Br. 13-14,

16. By its own account, the report was “not exhaustive”; it

restricted itself to 250 “easy to find” instances of

environmental damage associated with recycling. Problems

Study at 4 (identifying sources of “potentially relevant” data

that the study did not exhaust). Compare Dissent at 11. The

study seems to support a proposition, surely indisputable, that

recycling can go awry. Further, the authors claim to have

identified various causal factors, characterized rather vaguely

and clearly overlapping, such as “Improper Disposal of

Residuals,” “Abandoned Materials,” and “Improper

Management of Hazardous Secondary Materials.” Problems

Study at 6-8. But none of these bears any obvious relation to

the “comparable to or lower than” standard of the with-

analogue track. Reading the report liberally, we see around a

dozen instances (out of the 250) involving recycled products

that possibly would have flunked the technical provisions. See

id. app’x 1 at 22-23; 26-27; 45-47; 114-15, 121-22; 128-30,

247-48, 258-59, 298-300, 304, 319-320, 339-40, 404-05, 443-

44. And some of these products could have already been

considered hazardous waste for failing other legitimacy criteria

or for being “placed on the land in a manner that constitutes

disposal,” 40 C.FR. § 261.2(c)(1)(A). See, e.g., Problems

Study app’x 1 at 299 (recycler allegedly “planned to sell []

contaminated ash as fill material to the public”).

18

Thus the study in no way purports to establish that there is

any particular probability, much less a reasonable probability,

that the recycled products exceeding the “comparable to or

lower than” standard will cause damage to health or the

environment. But the quality or relevance of the study makes

no difference in this context, as EPA did not rely on it to justify

its assumption that materials which fail the technical provisions

are “discarded.” The study appears to enter EPA’s Factor 4

discussion only implicitly via the foundry sand example, and

the most EPA inferred from that was that certain recycled

products “may or may not be legitimate, depending on the use.”

Final Rule, 80 Fed. Reg. at 1,729/2-3. That conclusion doesn’t

take us beyond EPA’s bare assertion that “high levels of

hazardous constituents . . . could indicate” discard. Id. at

1,726/1.

In API I, we were satisfied by EPA’s mere “concern[]” that

some test samples had “unexpected” levels of contaminants

(EPA had no evidence that those results were due to

adulteration). 216 F.3d at 58. We stressed, though, that “a

refiner in a specific case” could show that the product was not

adulterated and not discarded. Id. at 59. Thus, the rule

involved at most a rebuttable presumption, which we have said

can “be sustained without an evidentiary showing . . . so long

as the agency articulates a rational basis.” Sec. of Labor v.

Keystone Coal Mining Corp., 151 F.3d 1096, 1101 (D.C. Cir.

1998). But our cases show that here a “rational basis,” id.,

means a reason, grounded in common sense or logic, to

suppose the inference “so probable that it is sensible and

timesaving to assume [its] truth . . . until the adversary

disproves it,” Nat’l Mining Ass’n v. Babbitt, 172 F.3d 906, 912

(D.C. Cir. 1999) (quoting Keystone, 151 F.3d at 1100-01)

(rejecting presumption for which the agency had “not offered

any support, scientific or otherwise”).

19

EPA has not offered a sufficient “rational basis.” Because

a recycler “in a specific case” won’t be able to recover from

failing to file paperwork and failing the technical provisions,

see API I, 216 F.3d at 59, EPA must offer more than timorous

assertions such as “could indicate” and “may or may not be

legitimate,” Final Rule, 80 Fed. Reg. at 1,726/1, 1,729/2-3.

The dissent sees nothing wrong with EPA’s exception

procedure. But our colleague’s view is significantly colored by

an assumption, not made by EPA, that the “comparable to or

lower than” standard is inherently reasonable and may not even

require an exception. Compare Dissent at 3-4, 9, with Final

Rule, 80 Fed. Reg. at 1,729/1. The dissent argues that the

standard is reasonably limited to situations where constituent

levels are “significantly” higher or exceed a “small acceptable

range.” Dissent at 5 (citing Final Rule, 80 Fed. Reg. at

1,727/2). But significant as to what? Acceptable against what

measure? The rulemaking gives no answer, certainly none

linking directly to the “significant human health or

environmental risk” criterion used in the exception.

§ 260.43(a)(4)(iii). Similarly absent is any reference to utility

or market acceptance as embodied in the “commodity

standards” clauses of subparagraphs (i) and (ii). If either of

those perspectives governed the “comparable to or lower than”

standard, why would EPA devote separate provisions to them?

Not even EPA argues that the “comparable to or lower than”

standard is reasonably limited to any such circumstances; we

will not adopt a tortured interpretation to infer that it is. See

generally Final Rule, 80 Fed. Reg. at 1,727/2-3 (explaining

standard via examples of “zinc galvanizing metal” and

“solvent”). Because the “comparable to or lower than”

standard (and, by extension, the with-analogue track) is not

reasonably focused on items that are “part of the waste disposal

problem,” Safe Food, 350 F.3d at 1268, the exception process

must be adequate to offset that fault. It is not.

20

For these reasons Factor 4 is unreasonable as a

requirement applied, through 40 C.F.R. § 261.2(g), to all

hazardous secondary material recycling. (EPA has also written

the legitimacy factors into specific exclusions. See, e.g., 40

C.F.R. § 261.4(a)(23)(ii)(E). Petitioners do not challenge

Factor 4 as applied to those individual exclusions.)

C. Used Oil Recycling

Industry Petitioners also ask us to invalidate EPA’s

legitimacy factors as applied to used oil recycling. This request

misreads EPA’s rules, which exempt used oil from the

legitimacy factors along with all the other “requirements of [40

C.F.R.] parts 260 through 268.” 40 C.F.R. § 261.6(a)(4).

III. Verified Recycler Exclusion

The Final Rule also amended EPA’s stance on

“reclamation,” a type of recycling that occurs when secondary

materials are “processed to recover a usable product, or . . .

regenerated.” 40 C.F.R. § 261.1(c)(4), (7). A dead battery is

reclaimed, for example, by extracting the still-valuable lead

from it. § 261.1(c)(4). The other modes of recycling are

“use[]” and “reuse[],” which occur when “[a] material is . . .

[e]mployed as an ingredient . . . in an industrial process to make

a product” or “[e]mployed . . . as an effective substitute for a

commercial product.” § 261.1(c)(5), (7). In the 1980s, EPA

adopted a rule manifesting its belief that certain hazardous

secondary materials are so “waste-like” that reclaiming them is

equivalent to discard. Hazardous Waste Mgmt. Sys., 50 Fed.

Reg. 614, 619/1 (Jan. 4, 1985). The materials so classified are

spent materials, listed sludges, listed byproducts, and scrap

metal—although EPA has a specific exception for the latter.

See 40 C.F.R. § 261.2(c)(3) & tbl.1. “Listed” means

catalogued by EPA as hazardous in § 261.31 or § 261.32. See

Hazardous Waste Mgmt. Sys., 50 Fed. Reg. at 619/1. Because

21

processing something is hardly akin to throwing it away, we

held that this reclamation rule improperly regulated materials

that were “neither disposed of nor abandoned, but [were]

passing in a continuous stream or flow from one production

process to another.” AMC, 824 F.2d at 1190, 1193.

EPA nonetheless kept the reclamation-equals-discard rule,

apparently on the reasoning that AMC merely “granted the

petition for review” without ordering vacatur. See Revisions to

the Definition of Solid Waste, 72 Fed. Reg. 14,172, 14,176/3-

77/1 (Mar. 26, 2007). Instead EPA sought to “implement the

AMC I opinion” by adding exclusions for specific materials or

processes. See, e.g., Identification and Listing of Hazardous

Waste, 59 Fed. Reg. 38,536, 38,537/1 (July 28, 1994) (adding

exclusion for petroleum-refining secondary materials),

codified as amended at 40 C.F.R. § 261.4(a)(12). Materials-

specific and process-specific exclusions form a large part of the

pre-2008 exclusions discussed in the introduction to this

opinion. See Proposed Rule, 76 Fed. Reg. at 44,139/1-3

(listing pre-2008 exclusions). Further, EPA adopted two

general exclusions, which unlike almost all of the pre-2008

exclusions, depend on whether the recycling is performed by a

third-party. The first general exclusion, the Generator-

Controlled Exclusion, governs reclamation “under the control

of the generator,” § 261.4(a)(23), and is not challenged here.

The other addresses reclamation of materials transferred to and

reclaimed by a third-party, and has come in two successive

editions. EPA adopted the first edition, the Transfer-Based

Exclusion, as part of its 2008 Rule, 73 Fed. Reg. at 64,669/3-

70/1, previously codified at 40 C.F.R. § 261.4(a)(24)-(25)

(2014), and replaced it with the current edition, the Verified

Recycler Exclusion, in the Final Rule, 80 Fed. Reg. at 1,706/3,

codified at § 261.4(a)(24).

22

Under the Transfer-Based Exclusion, the party offloading

the materials (the “generator”) could send them to a reclaimer

that possessed a RCRA permit (or interim status). 40 C.F.R.

§ 261.4(a)(24)(v)(B) (2014). Alternatively, the generator

could send materials to a reclaimer that lacked such a permit or

status, if the generator had made “reasonable efforts to ensure

that [the chosen] reclaimer intends to properly and legitimately

reclaim the hazardous secondary material and not discard it.”

Id. The “reasonable efforts” involved investigating and

“affirmatively answer[ing]” specific questions that the

regulation posed about the reclaimer. Id.

The Verified Recycler Exclusion is quite similar to its

predecessor but makes two changes that Industry Petitioners

challenge. First, the new exclusion requires the generator to

meet special “emergency preparedness” standards in its

custody of the materials before shipment. See 40 C.F.R.

§ 261.4(a)(24)(v)(E) (referring to standards at § 261.400 et

seq.). For example, the generator’s facility must be

“maintained and operated to minimize the possibility of a fire,

explosion, or any unplanned . . . release of hazardous secondary

materials” that “could threaten human health or the

environment.” § 261.410(a). And the generator must (with

some exceptions) have certain emergency preparedness

processes and equipment in place, such as communications and

“fire control” systems. See § 261.410(b)-(f).

Second, the Verified Recycler Exclusion eliminates the

“reasonable efforts” option afforded by the Transfer-Based

Exclusion and requires that generators send their secondary

materials to reclaimers who either have a RCRA permit (or

interim status), as in the Transfer-Based Exclusion, or a RCRA

variance—in effect an EPA (or state-level) approval of a firm

to operate a third-party “reclamation facility.” See 40 C.F.R.

§ 261.4(a)(24)(v)(B); § 260.31(d) (quoted language); see also

23

§ 271.3 (authorizing states to implement RCRA if they meet

certain conditions); Final Rule, 80 Fed. Reg. at 1,695/2

(describing the new rule); id. at 1,715/1, 1,768/2-3 (describing

role of “authorized state[s]”).

The separate Generator-Controlled Exclusion carries the

same emergency preparedness requirements,

§ 261.4(a)(23)(ii)(F), but it significantly does not mandate a

permit, interim status, or variance. It instead asks generators to

maintain a “written description of how the recycling meets all

four [legitimacy] factors.” § 261.4(a)(23)(ii)(E).

Industry Petitioners insist that EPA had no reason, in its

2015 shift to a Verified Recycler Exclusion, to tighten the

conditions of its predecessor. Though EPA disagrees, it

concedes that “withdrawing the transfer-based exclusion”

entirely “would result in hazardous secondary material that is

currently being legitimately recycled and not discarded being

regulated as hazardous waste,” Final Rule, 80 Fed. Reg. at

1,708/3, in effect, regulation in excess of EPA’s authority as

defined in AMC. In this perhaps topsy-turvy universe, all spent

materials, listed byproducts, and listed sludges being reclaimed

are subject to full RCRA control unless affirmatively excluded.

Because EPA chose to retain a rule that improperly treats as

discarded materials that are “no longer useful in their original

capacity though destined for immediate reuse,” AMC, 824 F.2d

at 1185, it has obliged itself to creating sufficient exceptions to

counter that rule’s overbreadth.

Given the parties’ agreement that some general exclusion

for third-party reclamation is necessary, the question before us

is whether EPA acted reasonably in adding emergency

preparedness requirements and in supplanting the reasonable

efforts option with the variance procedure. Specifically, EPA

must show that “the new policy is permissible under the statute,

24

that there are good reasons for it, and that the agency believes it

to be better” than the old one. FCC v. Fox Television Stations,

Inc., 556 U.S. 502, 515 (2009).

Although no party challenged the Industry Petitioners’

standing on this issue, we noted EPA’s assertion in the record

that in the almost seven years under the Transfer-Based

Exclusion no entity had taken advantage of the reasonable

efforts option. See Final Rule, 80 Fed. Reg. at 1,708/1-2,

1,709/1. If in the real world the option drew no takers for seven

years, could its removal really inflict an injury? Wondering if

petitioners’ claim of injury was truly plausible, as required by

our cases, see, e.g., Food & Water Watch, Inc. v. Vilsack, 808

F.3d 905, 913 (D.C. Cir. 2015), we ordered briefing on the

issue.

In their supplemental brief, Industry Petitioners supplied

the explanation: not long after the Transfer-Based Exclusion

was promulgated, “EPA announced that it was seriously

considering repeal,” which “placed the [] exclusion under a

cloud of uncertainty.” Industry Pet’rs’ Supp. Br. 2.

Unpermitted entities chose to wait and see if the reports of the

rule’s imminent demise were true. See id. Accordingly, there

is no apparent reason to doubt that, as Industry Petitioners

insist, EPA’s retention of the reasonable efforts option would

have led some entities to make use of it.

As to Fox’s required justifications for a change in policy,

EPA is quite clear which rule, 2008 or 2015, it “believes [] to

be better.” 556 U.S. at 515. EPA bemoaned that the Transfer-

Based Exclusion allowed third-party reclaimers to operate

without as much oversight as Subtitle C regulation would

require. Final Rule, 80 Fed. Reg. at 1,707/3. This lack of

oversight, EPA believes, “could lead to the potential for an

increased likelihood of environmental” damage, thus justifying

25

the Final Rule’s changes. Id. at 1,708/1; see id. at 1,711/2

(describing 2008 Rule’s “major regulatory gap” from “lack of

oversight and public participation”).

For the remainder of the Fox analysis we address the two

challenged provisions separately.

A. Emergency Preparedness Requirements

First up are the emergency preparedness requirements and

whether their promulgation meets the requirements of showing

consistency with the statute and good reasons for the new rule.

Fox, 556 U.S. at 515. For reasons to qualify as “good” under

Fox, they must be “justified by the rulemaking record.” State

Farm, 463 U.S. at 42. Here EPA’s reasons for its changes

overlap with its statutory justification—to “identif[y]

hazardous secondary materials that are legitimately recycled

and not discarded,” Final Rule, 80 Fed. Reg. at 1,709/2—so we

analyze the two together.

With the emergency preparedness provisions, EPA’s

reasoning is mostly a retread of what we encountered with

Factor 3. As with the handling requirements, it advances the

mandated precautions as an effort to reduce the risk of discard

and to test the generator’s intent to recycle. See id. at 1,710/2.

Here, to be sure, these prophylactic duties go beyond Factor 3’s

in specificity. Compare 40 C.F.R. § 261.410 (emergency),

with § 260.10 (containment). And the inference of “discard”

from feckless preparations is less obvious than such an

inference from lack of containment (as defined by EPA).

But EPA made findings (unchallenged here) that fires and

explosions are a common cause of environmental damage and

that planning against such mischance reflects a generator’s

intent to reduce losses of hazardous secondary materials—

materials that a firm intending genuine reclamation would

26

presumably regard as valuable. See Final Rule, 80 Fed. Reg.

at 1,710/2; Problems Study at 7. EPA also found that the

secondary materials to be recycled under the Verified Recycler

Exclusion (i.e., those materials that are transferred to third

parties and that don’t qualify for other exclusions) are “often”

of negative value to generators, which “typically pay” the

reclaimer to take the materials or receive a payment inadequate

to cover the costs of transfer. See Final Rule, 80 Fed. Reg. at

1,707/2; see also A Study of Potential Effects of Market Forces

on the Management of Hazardous Secondary Materials

Intended for Recycling 3 (Nov. 21, 2006) (the “Market Study”)

(noting that commercial recyclers accept materials “usually for

a fee”). Because generators are likely to view these materials

more as albatross than asset, it is reasonable for EPA to require

additional assurances, beyond those of Factor 3, that the

generator values them as elements of a genuine recycling

effort.

Petitioners do not claim that the preparation requirements

are an unreasonable test of intent, other than to say that they are

“highly prescriptive,” Industry Pet’rs’ Br. 53-54, an epithet that

most readers of the Code of Federal Regulations would likely

apply to every paragraph. In fact the mandated preparations

seem rather basic. If an entity balks at the prospect of keeping

a “telephone” and “[p]ortable fire extinguisher[]” on site,

§ 261.410(b)(2)-(3), it may not really belong in the business of

handling toxic and inflammable secondary materials. And in

practice it may not even have to do that much: EPA stands

ready to waive these and other preparedness requirements

when they’re not necessary. See § 261.410(b), (d), (e).

As we said of the containment requirements, there is some

risk that these mandatory precautions might be read

unreasonably. For example, the obligation “to minimize the

possibility of” accidents might be taken, standing alone, to

27

require all preventive measures no matter the cost.

§ 261.410(a). But we are satisfied that such a reading would

contravene EPA’s explanation in the rulemaking, that the rule

tests whether the generator intends “to reduce potential loss of

valuable hazardous secondary materials.” See Final Rule, 80

Fed. Reg. at 1,710/2.

B. Administrative Approval Requirements and Remedy

Petitioners focus more persuasively on EPA’s abolition of

the reasonable efforts option and its replacement with a

requirement of a variance for third-party reclamation. Under

the Transfer-Based Exclusion, a generator could send materials

to any reclaimer it chose, provided that, after making a

reasonable investigation, it “affirmatively answer[ed]” five

questions about the reclaimer. 40 C.F.R. § 261.4(a)(24)(v)(B)

(2014). These asked if the reclaimer (1) was employing a

legitimate recycling process; (2) had notified regulators of its

operations and its financial stability; (3) had not been the

subject of recent enforcement actions; (4) had adequate skill

and equipment to perform the recycling safely; and (5) had

adequate processes for disposing of any residual wastes

generated during the recycling. Id. The rule required the

generator to have met this obligation “in good faith” and to

have based its analysis for each question on an “objectively

reasonable belief.” 2008 Rule, 73 Fed. Reg. at 64,700/1. A

generator that failed to meet that standard could be liable for a

RCRA violation. Id. at 64,699/3-64,700/1.

The new rule keeps the general framework for evaluating

reclaimers but broadens the inquiry and assigns it to regulators,

not the generator. If the reclaimer lacks a RCRA permit or

interim status, it must secure a regulatory variance under 40

C.F.R. § 260.31(d) from the EPA Administrator or applicable

state regulator. See Final Rule, 80 Fed. Reg. at 1,715/1. And

the questions, transmogrified into criteria for administrative

28

grant, are expanded to include a sixth, requiring the reclaimer

to “address the potential for risk to proximate populations from

unpermitted releases of the hazardous secondary material.”

§ 260.31(d)(1)-(6). EPA asserts that this “additional

oversight” is required “to ensure that [] hazardous secondary

material is legitimately recycled and not discarded.” Final

Rule, 80 Fed. Reg. at 1,709/1. Here again, EPA’s “good

reasons” and its claim for permissibility under the statute

overlap, but not as persuasively as with the emergency

preparation requirements.

Recall that EPA has a Generator-Controlled Exclusion

which is targeted at the same types of material as the Verified

Recycler Exclusion: hazardous secondary materials reclaimed

in a manner that doesn’t qualify for pre-2008 exclusions. EPA

insists that these materials generally have little value as

recycling inputs, a trait from which one can reasonably infer a

greater susceptibility to illegitimate or improper recycling. See

id. at 1,707/1-2; see also EPA, Revisions to the Definition of

Solid Waste Final Rule Response to Comments Document, at

77 (Dec. 10, 2014) (the “Comments Document”)

(acknowledging that “high value” secondary materials are less

likely to be discarded but arguing that EPA has “already

promulgated exclusions for such materials”).

But this risk of discarding low-value materials would

apply whether the reclamation occurs in-house or externally.

And yet while the Generator-Controlled Exclusion and

Verified Recycler Exclusion share some conditions, only the

latter requires an administrative approval. Industry Petitioners

charge that EPA has acted on the basis of an unreasonable

presumption that transfer carries an undue risk of discard. Such

a presumption would contradict our holding in Safe Food that

“[a]s firms have ample reasons to avoid complete vertical

29

integration, firm-to-firm transfers are hardly good indicia of a

‘discard’” under RCRA. 350 F.3d at 1268 (citation omitted).

EPA counters that its reasoning is more nuanced, that it

rests not on transfer alone, but on the confluence of low-value

materials and transfer. These factors combine to form

“perverse incentives . . . to over-accumulate [] hazardous

secondary materials” without recycling them. Final Rule, 80

Fed. Reg. at 1,708/2; see also id. at 1,716/1 (justifying separate

exclusion for transferred “spent solvents” because third-party

reclaimers have “little economic reason to accumulate” these

“higher-value” materials). EPA’s theory is certainly more

clever than Industry Petitioners give it credit for, but EPA fails

to provide sufficient linkage between theory, reality, and the

result reached. See State Farm, 463 U.S. at 43 (“[T]he agency

must examine the relevant data and articulate a satisfactory

explanation for its action including a ‘rational connection

between the facts found and the choice made.’” (quoting

Burlington Truck Lines v. United States, 371 U.S. 156, 168

(1962))).

EPA’s reasoning relies heavily on a theoretical study

predicting that when the value of a recycled product is low, or

the market for it “weak or unstable,” the “acceptance fee”

generators pay when off-loading materials “may be an

important component of the [reclaimer’s] overall revenue.”

Market Study at 18; see also id. at 3. EPA asserts that this

incentive leads “commercial third party recyclers to maximize

the amount of hazardous secondary material they can accept to

increase profits,” thus creating risks of “over-accumulat[ion]”

and “discard.” Final Rule, 80 Fed. Reg. at 1,752/1. But having

found that some types of recycling are typified by transfers of

materials low or unstable in value, see Market Study at 88-89,

and having surmised that those conditions could lead to

“market failure,” id. at 3, the study disclaims any analysis of

30

whether such failures actually occur and to what degree:

“limitations on the availability and quality of data prevented us

from conducting [] empirical tests,” id. at 43.

EPA is free to rely on theoretical or model-based

approaches, as long as that reliance is reasonable in context.

As our dissenting colleague points out, Dissent at 10-11, we

long ago recognized that “[r]easoned decisionmaking can use

an economic model to provide useful information about

economic realities, provided there is a conscientious effort to

take into account what is known as to past experience and what

is reasonably predictable about the future.” American Public

Gas Ass’n. v. FPC, 567 F.2d 1016, 1037 (D.C. Cir. 1977). And

more recently, as our colleague also points out, Dissent at 11,

we deferred to EPA’s use of particle-trajectory modeling when

the agency found it to be “particularly illuminating,” noted that

it was “more precise” in some cases than historical data, and

“took reasonable steps to account for [its] limitations,”

Mississippi Comm’n on Environmental Quality v. EPA, 790

F.3d 138, 166-71 (D.C. Cir. 2015). Thus what we seek is some

indication of a reasonable concurrence between model and

reality. Here the Market Study cautions that its hypothesized

“sources of market failure,” e.g., skewed incentives leading to

discard and environmental damage, “do not necessarily

correlate directly to observable characteristics of the firm or

market.” Market Study at 48-49. Thus, the study offers EPA

reasons (based on seemingly sensible notions of market actors’

incentives) to think that the incidence of discard might be

somewhat higher in the presence of specific characteristics

(e.g., low-value materials and third-party transfer) than in their

absence. But it offers no data to support the view that the

increased incidence actually exists nor to show how great the

increase is. That type of information (or a sufficient

explanation for its absence) is quite important in cases such as

this, where EPA is determining that an activity nominally

31

outside of its jurisdiction should be banned absent regulatory

pre-approval. Thus EPA’s reliance on the study, standing

alone, does not provide a sufficient basis for the administrative

approval provisions.

EPA claims to have the necessary support in its Problems

Study, a document whose faults we’ve already discussed. Of

the study’s 250 instances of recycling gone awry, 238 involved

third-party recycling as opposed to on-site recycling. Problems

Study at 8. Based on these “easy to find” cases, id. at 4, EPA

inferred that discard could occur under the old Transfer-Based

Exclusion unless “additional oversight” was imposed, Final

Rule, 80 Fed. Reg. at 1,708/2. But far from confirming the

Market Study’s assumptions, the Problems Study is even more

tentative in its treatment of third-party recycling. It cautions

that the greater proportion of problems at off-site recyclers

might be because “on-site recycling is simply a less common

practice.” Problems Study at 8. (The study made no effort to

explain how the proportions of on-site and off-site reclamation

in the examples reviewed matched those of the real world.)

From the study, EPA concluded that “the vast majority of

environmental damages—approximately 94%—occur at off-

site commercial recyclers.” Final Rule, 80 Fed. Reg at 1,699/2.

But by focusing only on recycling gone wrong, that statistic

tells us nothing about such episodes’ overall likelihood in any

particular setting. Compare Dissent at 11-12. The dissent

offers a helpful example: imagine that 94% of plane crashes are

associated with Airline A; can we say that this airline is less

safe than its competitors? Dissent at 12. Before we can land

at that conclusion, we’d need first to know something about the

distribution of flights among airlines. If Airline A performs

94% of the set of flights that happened to be studied, its crash-

percentage would seem decidedly average. Compare Problems

Study at 8 (noting that on-site recycling may be “less

32

common”). Or maybe Airline A flies only the most dangerous

routes. Context gives clues. The Problems Study leaves us

grasping.

After commenters attacked EPA’s interpretation of this

study, EPA responded that because most recyclers lack any

duty to notify regulators, the agency does not have access to

better data about recycling practices. Final Rule, 80 Fed. Reg.

at 1,740/3-41/1-2. Indeed, we commonly “defer to an agency’s

decision to proceed on the basis of imperfect scientific

information, rather than to invest the resources to conduct the

perfect study.” Cablevision Sys. Corp. v. FCC, 649 F.3d 695,

717 (D.C. Cir. 2011) (internal quotation marks omitted). But

limited data do not justify unlimited inferences. Agency

reliance on imperfect information makes sense only where that

information supports the agency action.

The Market Study and Problems Study at most support a

belief, carried over from the Transfer-Based Exclusion, that

third-party reclaimers present distinct risks compared to on-site

reclaimers. These risks would accordingly justify special

conditions, such as the variance criteria for which the

Generator-Controlled Exclusion has no analogue. The first

five of these criteria, which petitioners do not challenge, seem

properly focused on whether the third-party reclaimer has the

inclination and ability to recycle legitimately. See 40 C.F.R.

§ 260.31(d)(1)-(5). (We cannot readily say the same of the

sixth, discussed below.)

But the imposition of a requirement of advance

administrative approval cannot be justified merely on the

differences that EPA has identified between on-site and third-

party reclamation. EPA must explain why the risk that

purported third-party recyclers will in reality “discard” the

materials is so high that reclamation under the Verified

33

Recycler Exclusion may only proceed on the basis of prior

agency approval. On this key aspect of third-party reclamation,

EPA’s Problems and Market studies say nothing useful.

EPA invokes yet another study, An Assessment of Good

Current Practices for Recycling of Hazardous Secondary

Materials (Nov. 22, 2006). This analysis, performed before

adoption of the Transfer-Based Exclusion, discussed the extent

to which generators voluntarily audited their third-party

recyclers to ensure that “their materials are not mishandled.”

Id. at 7. The study found that “auditing is being practiced by

many responsible companies” but that “small generators do not

audit as regularly as larger customers” and that smaller

generators’ audits may not be as thorough. Id. at 20. In 2008,

EPA evidently did not find much alarm in this data; it made the

reasonable efforts option available for small and large

generators alike. By 2015, EPA was less sanguine about the

study’s results, warning that “many smaller generators would

not have the technical expertise or resources to” adequately

assess third-party reclaimers. See Final Rule, 80 Fed. Reg. at

1,711/3 & n.17. EPA is free to reasonably revise its

interpretation of that study, but even this updated reasoning

cannot support the Final Rule. EPA admits in the rulemaking

that “many large companies do conduct environmental audits

of recycling facilities.” Id. at 1,711/3. A risk that some smaller

generators would misapply the reasonable efforts option does

not explain why EPA should treat larger generators as prone to

making inadequate assessments.

Along with their challenge to the variance procedure,

Industry Petitioners also claim that the sixth variance criterion

is, in substance, vague and unreasonable. This criterion

involves something of a “cumulative” nuisance standard; it

requires third-party reclaimers to account for how any

“unpermitted releases” from their facilities might combine with

34

“other nearby potential stressors” to create “risk[s] to

proximate populations.” 40 C.F.R. § 260.31(d)(6). The more

environmental problems there already are in an area—such as

“other industrial facilities, landfills, transportation-related air

emissions, poor housing conditions (e.g., lead-based paint),

leaking underground tanks, pesticides, and incompatible land

uses”—the less appropriate it might be for the reclaimer to add

yet another stress. See Final Rule, 80 Fed. Reg. at 1,714/3-

15/1.

Thus the criterion assumes discard, i.e., behavior regulable

under RCRA, and seeks to constrain its environmental impact,

rather than testing for discard’s existence. It identifies one of

the many problems related to waste disposal, but not whether

the reclaimer is actually contributing to the waste disposal

problem. Were we dealing with materials that were lawfully

identified as hazardous waste, this test might be valid for some

purposes. But the Verified Recycler Exclusion covers

materials that might be labeled waste only because of a

reclamation-equals-discard rule that EPA has all but conceded

is overbroad. Id. at 1,708/3. This criterion therefore cannot

stand as a means of identifying discard.

As for remedy, Industry Petitioners ask that we keep the

Verified Recycler Exclusion in place while removing its

objectionable provisions. They seek this remedy because not

all of the Final Rule’s changes were to their detriment.

Whereas the Transfer-Based Exclusion disqualified spent

catalyst generators from relying on it, 40 C.F.R.

§ 261.4(a)(24)(iii) (2014) (spent catalysts referenced as K171

and K172), the Verified Recycler Exclusion removed that bar.

As at least one of petitioners’ members is a spent catalyst

generator, an unalloyed return to the Transfer-Based Exclusion

would be for it a hollow victory.

35

We will “sever[] and affirm[] [] a portion of an

administrative regulation” only when we can say without any

“‘substantial doubt’ that the agency would have adopted the

severed portion on its own.” New Jersey v. EPA, 517 F.3d 574,

584 (D.C. Cir. 2008) (internal quotation marks omitted). Thus

we have severed provisions when “they operate[d] entirely

independently of one another.” Davis Cty. Solid Waste Mgmt.

v. EPA, 108 F.3d 1454, 1459 (D.C. Cir. 1997). Here, though,

we are not sure that EPA’s regulatory and deregulatory efforts

were wholly independent. The rulemaking shows that EPA

entertained two different options for removing the spent

catalyst bar: first as part of the plan to repeal the Transfer-

Based Exclusion entirely and replace it with “alternative

Subtitle C regulat[ions]” for which “spent catalysts would be

eligible,” Proposed Rule, 76 Fed. Reg. at 44,141/3 & n.54;

second as part of the Verified Recycler Exclusion that EPA

adopted, Final Rule, 80 Fed. Reg. at 1,738/1. At no point in

the record does EPA propose keeping the Transfer-Based

Exclusion and repealing its spent catalyst disqualifier.

Would EPA have so proposed had it known the Verified

Recycler Exclusion would be vacated? There is some evidence

pointing in that direction, but doubts remain. EPA explained

that its spent catalyst decision was due in large part to changes

to the “contained” standard at 40 C.F.R. § 260.10; these

revisions addressed the risk of fire that originally led EPA to

bar spent catalysts. See Final Rule, 80 Fed. Reg. at 1,738/1.

EPA also removed the spent catalyst disqualifier from the

Generator-Controlled Exclusion, which is generally less

restrictive than the Verified Recycler Exclusion. See id. These

facts suggest that EPA might have removed the disqualifier

absent the other changes in the Verified Recycler Exclusion.

But when commenters attacked EPA’s proposal to remove the

spent catalyst bar and advocated a more stringent approach,

EPA responded that, to fulfill the goal of allowing only

36

legitimate recycling, there was no need to impose the suggested

“additional conditions.” Comments Document at 265-66.

EPA’s answer assumed that that the new Verified Recycler

Exclusion and the new containment standard were together

sufficient to regulate transferred spent catalysts. We cannot

clearly infer what EPA would have done absent that exclusion.

The only changes in the Verified Recycler Exclusion that

we can sever without any “substantial doubt” are the

emergency preparedness requirements, 40 C.F.R.

§ 261.4(a)(24)(v)(E), which are as we explained lawful, and an

expanded containment requirement, § 261.4(a)(24)(v)(A),

which was not challenged. These new provisions address some

of EPA’s perceived “regulatory gaps” in the Transfer-Based

Exclusion, Final Rule, 80 Fed. Reg. at 1,706/3, and they do not

depend on any vacated portions of the Verified Recycler

Exclusion. On remand, EPA can of course renumber its rules

as necessary to accommodate the returning Transfer-Based

Exclusion provisions.

EPA has not commented on the requested remedy,

probably because the remedy section in Industry Petitioners’

opening brief was quite confusing, and their desire to sever and

affirm was made evident only in their reply. If EPA, or any

party, wishes to disabuse us of our substantial doubt with a

petition for rehearing, we will of course reconsider as

necessary. See MD/DC/DE Broadcasters Ass’n v. FCC, 253

F.3d 732, 740 (D.C. Cir. 2001) (citing Virginia v. EPA, 116

F.3d 499, 500-01 (D.C. Cir. 1997)).

Having concluded that the Verified Recycler Exclusion is

unreasonable, we need not address Environmental Petitioners’

argument that the exclusion is too lenient.

IV. Remaining Challenges by Industry Petitioners

37

Industry Petitioners have two remaining challenges. The

first is that EPA cannot subject spent catalysts to the Verified

Recycler Exclusion. The second is that EPA cannot treat off-

specification commercial chemical products as secondary

materials. The first is rendered moot by our restoration of the

Transfer-Based Exclusion, and no more needs to be said about

it here. The second is also outside our jurisdiction, but for

reasons requiring more explanation.

During the rulemaking, a commenter asked EPA to

confirm that commercial chemical products are not “hazardous

secondary material[s]” as that class is defined in 40 C.F.R.

§ 260.10. Comments Document at 313. EPA answered, much

to Industry Petitioners’ chagrin, that “a commercial chemical

product listed in 40 CFR 261.33 could be considered a

hazardous secondary material if it is off-specification or

otherwise unable to be sold as a product.” Id. at 314; see

Industry Pet’rs’ Br. 58-65. The question and EPA’s answer

concern an issue that is antecedent to the Final Rule’s

definition of discarded hazardous waste. The rule identifies

when secondary materials become waste as a result of being

sham recycled, but that delineation necessarily builds on prior

law and regulations governing when materials are secondary.

We cannot assess EPA’s statement on that subject unless we

can find the issue within our original jurisdiction, which is

limited to actions by EPA “promulgating” regulations, etc. 42

U.S.C. § 6976(a)(1).

Tellingly, the comment and EPA’s response are

interpreting provisions in 40 C.F.R. § 260.10 and § 260.33 that

were left untouched by the Final Rule. See Comments

Document at 313-14. Because of the limits on our jurisdiction,

we cannot entertain the claim unless EPA’s statement was

more than just an interpretation of a prior rule; it must interpret

part of the Final Rule or be itself an effective “legislative rule.”

38

See Cement Kiln Recycling Coal. v. EPA, 493 F.3d 207, 226

(D.C. Cir. 2007). Industry Petitioners’ allegation, though, is

that EPA’s response abandoned a prior policy, embodied

largely in guidance materials, without properly recognizing

that change. Industry Pet’rs’ Br. 64-65. Such a challenge is

properly before the district court, not this tribunal (Industry

Petitioners make no claim of pendent jurisdiction). See 42

U.S.C. § 6976(a)(1). We express no opinion on when EPA

may consider commercial chemical products to be secondary

materials.

V. Challenges by Environmental Petitioners

Environmental Petitioners challenge EPA’s approach to

the pre-2008 exclusions. As noted above, before 2008, EPA

had promulgated 32 exclusions from the definition of solid

waste—that is, it had exempted 32 different materials,

products, or processes from Subtitle C regulation. In its

Proposed Rule, EPA proposed subjecting facilities that

qualified for these exclusions to four new requirements, three

of which are relevant here: legitimacy, containment, and

notification. 76 Fed. Reg. at 44,138/3-39/1-2. The proposed

legitimacy condition set forth the factors that facilities had to

satisfy in order to prove they are engaged in legitimate, rather

than sham, recycling. Under the proposed containment

condition, facilities had to store all hazardous secondary

materials in units that meet certain safety, quality, and labeling

criteria. Id. at 44,140/1. And the proposed notification

condition obligated regulated parties periodically to submit

information to EPA so that the agency could monitor

compliance. Id. at 44,140/1-2. EPA based these conditions on

a study of environmental damage cases involving hazardous

waste (an earlier version of the Problems Study) and EPA’s

finding that most of cases in that study were associated with

secondary materials exempted under a pre-2008 exclusion. Id.

at 44,138/1-2.

39

In the final rule, however, EPA opted to apply only the

legitimacy condition to all pre-2008 exclusions and deferred a

decision about whether to do the same with containment and

notification. Specifically, EPA stated that it was “deferring

action on applying the contain[ment] [and notification]

standard[s] to the pre-2008 exclusions and exemptions until [it

could] more adequately address commenters’ concerns.” Final

Rule, 80 Fed. Reg. at 1,766/2-3. Commenters had raised

unanticipated objections, EPA explained, regarding the

difficulties of implementing a universal containment provision

and the burdens imposed by a notification requirement. Id.

Environmental Petitioners take issue with EPA’s decision

to defer action on containment and notification. Drawing on

language from the Proposed Rule, they argue that EPA

fundamentally changed its position without explanation:

whereas the agency originally viewed containment and

notification as “minimum requirements necessary to define

when recycled hazardous secondary materials are not

discarded,” 76 Fed. Reg. at 44,138/3-39/1, it ultimately

determined that containment and notification conditions were

expendable. This unexplained reversal, Environmental

Petitioners contend, was arbitrary and capricious.

We need not—indeed cannot—reach the merits of this

challenge. RCRA’s judicial review provision vests this court

with exclusive power to review “action[s] of the Administrator

in promulgating any regulation, or requirement under this

chapter or denying any petition for the promulgation,

amendment or repeal of any regulation under this chapter.” 42

U.S.C. § 6976(a)(1). This provision gives us jurisdiction over

only “three types of actions by EPA: promulgation of final

regulations, promulgation of requirements, and the denial of

petitions for the promulgation, amendment or repeal of RCRA

regulations.” API I, 216 F.3d at 68; see Molycorp, Inc. v. EPA,

40

197 F.3d 543, 545 (D.C. Cir. 1999) (characterizing 42 U.S.C.

§ 6976(a)(1) as “a limitation on our jurisdiction”). Critically

here, we have held that “[a] decision by an agency to defer

taking action is not a final action reviewable [under RCRA].”

API I, 216 F.3d at 68; see also American Portland Cement

Alliance v. EPA, 101 F.3d 772, 777 (D.C. Cir. 1996). Because

EPA expressly stated that it was deferring action on applying

containment and notification conditions to the pre-2008

exclusions, we lack jurisdiction to review Environmental

Petitioners’ claim.

Environmental Petitioners resist this straightforward

jurisdictional analysis. Citing Montana v. Clark, 749 F.2d 740

(D.C. Cir. 1984), and Appalachian Power Co. v. EPA, 208 F.3d

1015 (D.C. Cir. 2000), they argue that we may review EPA’s

decision to defer. But neither of these cases construes RCRA’s

judicial review provision. See Appalachian Power Co., 208

F.3d at 1020-22 (interpreting the Clean Air Act’s judicial

review provision); Clark, 749 F.2d at 744 (interpreting the

Administrative Procedure Act). And even if they did, those

cases are easily distinguished. Whether we have authority to

review an agency’s express rejection of a request to amend

longstanding regulations, Clark, 749 F.2d at 744, is irrelevant

where, as here, EPA has merely deferred—rather than

rejected—a particular action. Moreover, although “[t]he fact

that a law may be altered in the future has nothing to do with

whether it is subject to judicial review at the moment,”

Appalachian Power Co., 208 F.3d at 1022, we lack jurisdiction

to review EPA’s deferred action not because EPA could change

its mind down the road, but because it has yet to make up its

mind in the first place.

Alternatively, Environmental Petitioners contend that we

have jurisdiction over their challenge because EPA “reopened”

comment on the pre-2008 exclusions and then declined to

41

revise them. Environmental Pet’rs’ Br. 43. The reopener

doctrine “permits a plaintiff to bring an otherwise-stale

challenge . . . . when an agency has considered substantively

changing a rule but ultimately declined to do so.” Mendoza v.

Perez, 754 F.3d 1002, 1019 n.12 (D.C. Cir. 2014).

Environmental Petitioners’ reopener argument falters for a

simple reason: the doctrine has no applicability to this case

because EPA never considered changing the substance of the

pre-2008 exclusions. As it stated in the Proposed Rule, EPA

was “not reopening comment on any substantive provisions of

the regulatory exclusions or exemptions,” but rather was

proposing legitimacy, containment, and notification

requirements “as means to better enforce the regulations.” 76

Fed. Reg. at 44,138/3.

Of course, nothing in our conclusion forecloses judicial

review of EPA’s inaction once and for all. Environmental

Petitioners may petition EPA to promulgate a rule imposing

containment and notification conditions and, if their petition is

denied, seek review in this court. See 42 U.S.C. § 6976(a)(1)

(granting jurisdiction to review denials of rulemaking

petitions). We conclude only that Environmental Petitioners

are barred from obtaining review in the manner they now seek.

And because we dispose of their challenge by concluding that

we are without statutory jurisdiction, we have no reason to

address Industry Intervenors’ contention that Environmental

Petitioners lack Article III standing. See Sinochem

International Co. Ltd. v. Malaysia International Shipping

Corp., 549 U.S. 422, 431 (2007) (holding that “there is no

mandatory ‘sequencing of jurisdictional issues’” and that “a

federal court has leeway ‘to choose among threshold grounds

for denying audience to a case on the merits’” (quoting

Ruhrgas AG v. Marathon Oil Co., 526 U.S. 574, 584-85

(1999))).

42

VI. Conclusion

The Final Rule is upheld in part and vacated in part as

consistent with this opinion. Briefly put: Factor 3 is upheld;

Factor 4 is vacated insofar as it applies to all hazardous

secondary materials via § 261.2(g); the Verified Recycler

Exclusion is vacated except for its emergency preparedness

provisions and its expanded containment requirement; and the

Transfer-Based Exclusion is reinstated. As a consequence of

the latter, the removal of that exclusion’s bar on spent catalysts

is vacated, subject, as we noted above, to such arguments as

parties may raise supporting a different outcome.

So ordered.

TATEL, Circuit Judge, dissenting from Parts II.B and III.B: In

the mid-1970s, as industrial and technological developments

spurred the national economy, the United States faced “a rising

tide of scrap, discarded, and waste materials.” 42 U.S.C.

§ 6901(a)(2). This mounting waste caused “serious financial,

management, intergovernmental, and technical problems,” id.

§ 6901(a)(3), and posed a grave threat “to human health and

the environment,” id. § 6901(b)(5). In response, Congress

passed the Resource Conservation and Recovery Act (RCRA),

42 U.S.C. §§ 6901–6992k, a comprehensive scheme “to

regulate hazardous wastes from cradle to grave in accordance

with . . . rigorous safeguards and waste management

procedures,” Chicago v. Environmental Defense Fund, 511

U.S. 328, 331 (1994). Through RCRA, and central to this case,

Congress sought to prevent environmental harm by ensuring

that hazardous waste was “properly managed in the first

instance thereby reducing the need for corrective action at a

future date.” 42 U.S.C. § 6902(a)(5).

Congress gave the Administrator of the Environmental

Protection Agency (EPA) broad authority to effectuate this

goal. See id. § 6912. Selected by the President and confirmed

by the Senate for his or her expertise in environmental issues,

the Administrator may promulgate “such regulations as are

necessary to carry out his [or her] functions.” Id. § 6912(a)(1).

The judiciary, by contrast, has a limited role under RCRA.

When reviewing rules issued by the Administrator, the courts,

lacking environmental expertise and political accountability,

are bound by two fundamental principles of judicial restraint.

First, because RCRA provides for review “in accordance

with” the Administrative Procedure Act, id. § 6976(a), a

reviewing court’s task is to ask only whether the rule is

“arbitrary, capricious, an abuse of discretion, or otherwise not

in accordance with law,” 5 U.S.C. § 706(2)(A). As the Supreme

Court has made clear, once a court is satisfied that EPA is

acting within its delegated authority, the “scope of [judicial]

2

review under the ‘arbitrary and capricious’ standard is narrow.”

Motor Vehicle Manufacturers Association of the United States

v. State Farm Mutual Automobile Insurance Co., 463 U.S. 29,

43 (1983). Courts are “not to ask whether a regulatory decision

is the best one possible or even whether it is better than the

alternatives.” FERC v. Electric Power Supply Association, 136

S. Ct. 760, 782 (2016). This is especially true where, as here,

agency action involves “a high level of technical expertise,”

Marsh v. Oregon Natural Resources Council, 490 U.S. 360,

377 (1989) (quoting Kleppe v. Sierra Club, 427 U.S. 390, 412

(1976) (internal quotation mark omitted)), and “predictive

judgments about areas that are within the agency’s field of

discretion,” BNSF Railway Co. v. Surface Transportation

Board, 526 F.3d 770, 781 (D.C. Cir. 2008) (quoting Wisconsin

Public Power, Inc. v. FERC, 493 F.3d 239, 260 (D.C. Cir.

2007)).

Second, when reviewing facial challenges to a rule—again

as here—courts are required to assess the rule’s validity across

a broad spectrum of applications; they are not to imagine

whether the rule might be arbitrary in “uncommon particular

applications,” which, of course, can be challenged later should

they arise. EPA v. EME Homer City Generation, L.P., 134

S. Ct. 1584, 1609 (2014). As Congress well knew when it

authorized pre-enforcement facial review of RCRA rules, see

42 U.S.C. § 6976(a)(1), the fact that a petitioner—or for that

matter a judge—“can point to a hypothetical case in which the

rule might lead to an arbitrary result does not render the rule

‘arbitrary or capricious,’” American Hospital Association v.

NLRB, 499 U.S. 606, 619 (1991).

In this case, EPA promulgated a rule defining when

hazardous materials qualify as “discarded” and thus may be

subjected to RCRA’s rigorous protections. The court never

questions the Administrator’s statutory authority to issue the

3

Final Rule, but nonetheless invalidates two of its critical

features: Factor 4 of the legitimacy test, which distinguishes

genuine from sham recycling; and the verified recycler

exclusion, which ensures that companies claiming to recycle

hazardous waste in fact do so. In reaching this result, the court

displays a level of scrutiny that I believe conflicts with the

APA’s highly deferential standard of review and with the

principles governing judicial review of facial challenges to

rules. As a result, the court has deprived the public of two

safeguards that the Administrator, exercising her statutory

authority under RCRA, reasonably believed were needed to

protect “human health and the environment.” 42 U.S.C.

§ 6901(b)(5). I respectfully dissent.

I.

Factor 4 of the legitimacy test targets sham recyclers that

incorporate hazardous materials into recycled products in order

to avoid proper recycling or disposal. It does so by requiring

that the product of a recycling process “be comparable to a

legitimate product or intermediate.” 40 C.F.R. § 260.43(a)(4).

This approach makes sense: as the Administrator explained,

“high levels of hazardous constituents” in an allegedly recycled

product “could indicate that the recycler incorporated

hazardous constituents into the final product when they were

not needed to make that product effective.” 80 Fed. Reg. 1,726.

The Final Rule offers recyclers three alternative avenues for

demonstrating compliance with Factor 4.

First, subparagraph (i) addresses recycled products that

have raw analogues. Such products satisfy Factor 4 if they (A)

“do[] not exhibit a hazardous characteristic . . . that analogous

products do not exhibit” and (B) contain comparable

concentrations of hazardous constituents or hazardous-

constituent levels that meet widely used commodity standards.

40 C.F.R. § 260.43(a)(4)(i). In my view, this subparagraph

4

rationally effectuates Factor 4’s general approach. EPA

inferred that if a recycled product contains more hazardous

constituents or properties than its raw analogue, sham recycling

has occurred. 80 Fed. Reg. 1,727. Why else would those

hazardous constituents or properties be present? By way of

example, EPA pointed to paint made from recycled hazardous

materials. If such paint contains significant amounts of

cadmium (a hazardous constituent), but the same type of paint

made from raw materials contains no cadmium, such a

disparity “could indicate that the cadmium serves no useful

purpose and is being passed through the recycling process and

discarded in the product.” Id.

We validated an almost identical technical judgment by

the Administrator in Safe Food and Fertilizer v. EPA, 350 F.3d

1263 (D.C. Cir. 2003). Under the rule in that case, certain

recycled materials were deemed non-discarded when (1)

market participants treated them “more like valuable products

than like negatively-valued wastes” and (2) “the [products]

derived from the recycled [materials were] chemically

indistinguishable from analogous commercial products made

from virgin materials.” Id. at 1269. In essence, this rule

exempted materials from regulation based on their compliance

with criteria that, like Factors 3 and 4, assess whether recyclers

treat materials as valuable commodities and generate products

chemically indistinguishable from analogous products. We

held that these two factors, in conjunction, represented a

“reasonable tool for distinguishing products from wastes.” Id.

As to the “identity principle”—subparagraph (i)’s

counterpart—the court reasoned that where a recycled product

is “indistinguishable in the relevant respects” from the

analogous “virgin” product, it is “eminently reasonable” to

treat both as “products rather than wastes.” Id.

5

In spite of Safe Food, this court concludes that

subparagraph (i) is too “imprecis[e]” to be reasonable. Maj. Op.

at 12. In its view, some legitimately recycled products may

contain “some small excess of hazardous constituents,” and the

presence of those hazardous materials “would not constitute a

reasonable basis for dubbing the product or the process a

sham.” Id. But subparagraph (i) does not simply target products

with “some small excess of hazardous constituents.” Rather, it

targets products with significantly more hazardous constituents

or properties than an analogous raw product, i.e., beyond “a

small acceptable range” of difference. 80 Fed. Reg. 1,727. The

Administrator explained: “If a product produced with

hazardous secondary material exhibited a characteristic of

hazardous waste that an analogous product did not exhibit, this

would be an indication that sham recycling could be occurring

as a significant hazardous constituent or characteristic would

be in the product only as a result of the recycling of the

hazardous secondary material.” Id. (emphasis added).

Perhaps the presumption underlying subparagraph (i) does

suffer from some “imprecision.” Maj. Op. at 12. Yet because

Industry Petitioners have mounted a facial attack on the Final

Rule, this court has no authority to conjure up “hypothetical

case[s] in which the rule might lead to an arbitrary result.”

American Hospital Association, 499 U.S. at 619. Where, as

here, the Administrator’s presumption of sham recycling based

on elevated levels of hazardous constituents is reasonable

across most applications, we must uphold it. Id. If someday the

Administrator applies the rule to a recycler in an arbitrary and

capricious manner—for instance, as the court fears, by

selecting an unreasonably “small acceptable range of

difference,” see Maj. Op. at 19—that recycler “may bring a

particularized, as-applied challenge to the [rule],” EME Homer

City Generation, 134 S. Ct. at 1609.

6

The court’s analysis of subparagraph (i) suffers from a

second defect. Whether the presence of hazardous constituents

provides sufficient evidence of sham recycling is exactly the

type of technical judgment that RCRA delegates to the

Administrator. Of course, the Administrator “must examine the

relevant data and articulate a satisfactory explanation for its

action.” State Farm, 463 U.S. at 43. The court, however, never

questions the Administrator’s compliance with these two

requirements. Instead, it second guesses the Administrator’s

“predictive judgments,” BNSF Railway Co., 526 F.3d at 781,

about a matter—the precise level of hazardous constituents

needed to demonstrate sham recycling—that “requires a high

level of technical expertise” to which “we must defer,” Marsh,

490 U.S. at 377.

Subparagraph (ii), which applies when a recycled product

has no raw analogue, offers recyclers a second way to show

compliance with Factor 4. These products qualify as legitimate

if they “meet[] widely recognized commodity standards and

specifications” or if “[t]he hazardous secondary materials

being recycled are returned to the original process . . . from

which they were generated.” 40 C.F.R. § 260.43(a)(4)(ii).

The court concedes that subparagraph (ii) is reasonable,

see Maj. Op. at 10–11, and for good reason. The Final Rule

describes the agency’s efforts to address commenters’ concerns

that in many cases of legitimate recycling “there may not be an

analogous product with which a facility can compare the

product of the recycling process.” 80 Fed. Reg. 1,728. In

response to these concerns, as well as other comments

supporting an approach focused on commodity standards and

closed-loop recycling, the Administrator carved out “recycling

processes that [are] designed to use a specific hazardous

secondary material to make a useful product and processes that

7

always incorporate[] a hazardous secondary material back into

the generating process during manufacturing.” Id.

Finally, subparagraph (iii)—a catchall for recyclers unable

to comply with subparagraphs (i) or (ii)—allows recyclers to

demonstrate legitimacy by showing either a “lack of exposure

from . . . or bioavailability of . . . toxics” in the product.

40 C.F.R. § 260.43(a)(4)(iii). Even if they fail to make either

showing, moreover, recyclers can still demonstrate legitimacy

by pointing to any “other relevant considerations” showing that

the product does not “pose a significant human health or

environmental risk.” Id. To make these showings, recyclers

must “prepare documentation,” including a “certification

statement that the recycling is legitimate,” which “must be

maintained on-site for three years after the recycling operation

has ceased.” Id.

Although the court acknowledges that subparagraph

(iii) reasonably draws the line between recycling and discard

through a perspective based on health and environmental risks,

Maj. Op. at 15 (citing Safe Food, 350 F.3d at 1269–70), it

nonetheless concludes that subparagraph (iii) “falls short of

saving the rule, due to the draconian character of the

procedures it imposes on recyclers,” namely, the requirement

to prove legitimacy by preparing and maintaining

“paperwork,” id. at 15–16.

For their part, however, Industry Petitioners never argue

that the rule’s paperwork obligations are too rigorous. This is

understandable. If subparagraph (iii) qualifies as draconian,

then so too would countless other run-of-the-mill requirements

that entities file applications and keep certificates on hand: like

those for pilots, see 14 C.F.R. § 61.3; id. § 61.123, elevator

operators, see D.C. MUN. REGS. tit. 12, § 3010A–3011A, and

businesses selling alcohol, see D.C. CODE § 25-401; id. § 25-

8

711, just to name a few. Not even the procedures for gaining

and maintaining admission to the District of Columbia Bar

would pass muster, as they require candidates to prepare a

character and fitness application and certify completion of a

mandatory course on professional conduct. See D.C. COURT OF

APPEALS R. 46; D.C. BAR BYLAWS, R. 2.

In any event, the court’s conclusion runs headlong into

precedent. In American Chemistry Council v. EPA, 337 F.3d

1060 (D.C. Cir. 2003), we considered a challenge to an EPA

rule that presumed certain mixtures and derivatives of waste

were “hazardous” and thus subject to regulation, yet permitted

regulated entities to show otherwise. Upholding this rule, we

concluded that the Administrator acted reasonably in

“[p]lacing the burden upon the regulated entity to show the lack

of a hazardous characteristic.” Id. at 1065. This burden-shifting

approach, we determined, alleviated unmanageable

administrative obligations for the agency and comported with

RCRA’s command to “err on the side of caution.” Id. at 1065–

66.

Subparagraph (iii) works just like the rule we approved in

American Chemistry Council. If a recycler is unable to satisfy

subparagraph (i) or (ii), it is a presumptive sham recycler.

Subparagraph (iii) then allows the recycler to prove otherwise

by making the requisite showings through documentation. If

anything, the rule here is more lenient than the one in American

Chemistry Council because subparagraph (iii) provides for a

“self-implementing certification process,” 80 Fed. Reg. 1,730,

rather than a “cumbersome . . . delisting process,” American

Chemistry Council, 337 F.3d at 1065.

According to the court, the Final Rule is unlike the one in

American Chemistry Council because the Administrator never

demonstrated that recyclers failing to meet subparagraph (i) are

9

presumptively discarding. Maj. Op. at 16–17. At bottom, then,

the court’s critique of subparagraph (iii) traces back to its

conclusion that subparagraph (i) (and only subparagraph (i))

does not reasonably distinguish legitimate from sham

recycling. But contrary to the court’s view, EPA cogently

explained why subparagraph (i) is reasonable across most

applications, adding subparagraph (iii) only given the

possibility that “there may still be instances where recycling is

legitimate, but is unable to meet” subparagraph (i) or (ii). 80

Fed. Reg. 1,729. Subparagraph (iii) thus serves as a catchall

provision designed to give industry even more “flex[ibility],”

id., not as a tacit acknowledgment that subparagraph (i) is

deficient, contra Maj. Op. at 14. Rather than “substitute [its]

own judgment for that of [EPA],” this court should defer to the

agency’s technical and policy decisions. Electric Power Supply

Association, 136 S. Ct. at 782.

II.

The key difference between the verified recycler exclusion

and its predecessor—the transfer-based exclusion—is that the

new rule shifts oversight of off-site recyclers from the industry

to the Administrator. 80 Fed. Reg. 1,709. Whereas before waste

generators audited off-site recyclers to ensure their legitimacy,

now the Administrator or a state authority issues a variance

confirming that a recycler’s practices are sound. Id. at 1,695.

The court never questions the Administrator’s authority to

promulgate this rule. Instead, invoking a single line from Safe

Food—“firm-to-firm transfers are hardly good indicia of a

‘discard,’” 350 F.3d at 1268—the court concludes that the

Administrator had no basis for finding that transferred

hazardous materials “carr[y] an undue risk of discard,” Maj.

Op. at 28.

10

Safe Food, however, held only that transferred materials

are not automatically discarded simply because they are sent

off-site. As we explained, although “we have never said that

RCRA compels the conclusion that material destined for

recycling in another industry is necessarily ‘discarded,’” the

statute “does not preclude application of RCRA to such

materials if they can reasonably be considered part of the waste

disposal problem.” Safe Food, 350 F.3d at 1268. The verified

recycler exclusion is consistent with Safe Food: it defines

transferred materials as discarded if—and only if—the off-site

recycler receiving the materials fails to meet certain criteria,

which carefully discern whether allegedly recycled materials

“can reasonably be considered part of the waste disposal

problem.” Id.

This approach finds ample support in the administrative

record. When designing the verified recycler exclusion, the

Administrator relied on multiple sources, including a report on

market forces in the recycling industry and a study of the

environmental problems associated with recycling hazardous

secondary materials. 80 Fed. Reg. 1,707. The first of these, the

market study, concluded that off-site commercial recyclers,

which generate revenue primarily by receiving hazardous

materials, have “economic incentives to accumulate waste

beyond their ability to deal with it.” Id. The second report, the

problems study, found that of 208 cases in which hazardous

waste recycling led to serious environmental damage, 94

percent were attributable to “off-site third-party recyclers.” Id.

In the court’s view, neither study justifies the rule.

Although not impugning the market study on its merits, the

court rejects it as lacking empirical analysis. But no rule of

administrative law bars agencies from relying on studies that

use economic models to assess market incentives. In fact, EPA

often relies on theoretical models—that is, studies without

11

corroborating “data,” Maj. Op at 30—and our court has long

held that “[r]easoned decisionmaking can use an economic

model to provide useful information about economic realities.”

American Public Gas Association v. FPC, 567 F.2d 1016, 1037

(D.C. Cir. 1977); see also Mississippi Commission on

Environmental Quality v. EPA, 790 F.3d 138, 171 (D.C. Cir.

2015) (“EPA’s application, interpretation and modification of

[predictive] modeling [to set emissions standards] plainly fall

‘within its technical expertise’ and thus we owe it ‘an extreme

degree of deference.’” (quoting ATK Launch Systems, Inc. v.

EPA, 669 F.3d 330, 338 (D.C. Cir. 2012))).

At any rate, the problems study provides plenty of

empirical support for the conclusion that off-site recycling

leads to discard. It surveyed cases since 1982 in which

recyclers contaminated the environment by discarding

hazardous waste, poisoning soil and groundwater “with

remediation costs in some instances in the tens of millions of

dollars.” 80 Fed. Reg. 1,707. To identify these cases, EPA

reviewed scores of sources, including the Superfund National

Priorities List, national and state databases, comments from at

least three different rulemakings, media reports, and

information gleaned from contacts in EPA regional offices and

state agencies. See EPA OFFICE OF RESOURCE CONSERVATION

AND RECOVERY, AN ASSESSMENT OF ENVIRONMENTAL

PROBLEMS ASSOCIATED WITH RECYCLING OF HAZARDOUS

SECONDARY MATERIALS 4 (2014). This thorough canvassing

revealed that a full 94 percent of cases involving serious

environmental damage could be attributed to off-site recycling.

The court condemns the problems study for “focus[ing]

only on recycling gone wrong.” Maj. Op. at 31. As a result, the

court reasons, the study “tells us nothing” about the relative

risks of off-site recycling or the total damage caused by off-site

recyclers. Id. But this focuses on the wrong question. As the

12

Administrator recognized, the salient question is not what

percentage of all off-site recycling damages the environment,

but rather what portion of serious damage from hazardous

waste disposal is caused by off-site recyclers. The core issue

here is whether EPA may target the very companies (off-site

recyclers) most responsible for environmental damage. Given

the agency’s statutory obligation to prevent environmental

harm from discarded hazardous waste, I see no reason why it

cannot. Accordingly, that some off-site recycling is safe or that

serious environmental damage is relatively unusual is beside

the point.

Consider this issue in a different context. If there were 208

plane crashes and 94 percent were linked to one carrier, it

would be eminently reasonable for an agency tasked with

preventing plane crashes to require that carrier to demonstrate

that its practices were safe, no matter how many flights the

carrier completed or what percentage of total flights it

performed. Contra Maj. Op. at 31–32. No one would argue that

it was unreasonable to regulate the carrier because only a small

percentage of its total flights crashed. Yet this court’s approach

would yield just that result.

In the end, the fundamental problem with the court’s

conclusion—that the Administrator needs more proof that off-

site recycling is unsafe before requiring a variance—is that the

court decides for itself a policy question Congress left to the

Administrator. RCRA envisions a careful balance of authority

between EPA and this court. Today the court upsets that

balance.