Ergon-West Virginia, Inc. v. EPA

980 F.3d 403

Opinion

PUBLISHED

UNITED STATES COURT OF APPEALS FOR THE FOURTH CIRCUIT

No. 19-2128

ERGON-WEST VIRGINIA, INCORPORATED, Petitioner,

v.

UNITED STATES ENVIRONMENTAL PROTECTION AGENCY, Respondent.

------------------------------

NATIONAL BIODIESEL BOARD; PRODUCERS OF RENEWABLES UNITED FOR INTEGRITY TRUTH AND TRANSPARENCY,

Amici Supporting Respondent.

No. 19-2148

ERGON-WEST VIRGINIA, INCORPORATED, Petitioner,

v.

UNITED STATES ENVIRONMENTAL PROTECTION AGENCY, Respondent.

------------------------------

NATIONAL BIODIESEL BOARD; PRODUCERS OF RENEWABLES UNITED FOR INTEGRITY TRUTH AND TRANSPARENCY,

Amici Supporting Respondent.

No. 19-2152

ERGON-WEST VIRGINIA, INCORPORATED, Petitioner,

v.

UNITED STATES ENVIRONMENTAL PROTECTION AGENCY, Respondent.

------------------------------

NATIONAL BIODIESEL BOARD; PRODUCERS OF RENEWABLES UNITED FOR INTEGRITY TRUTH AND TRANSPARENCY,

Amici Supporting Respondent.

On Petition for Review of Final Agency Action of the United States Environmental Protection Agency.

Argued: September 8, 2020 Decided: November 17, 2020

Before NIEMEYER, AGEE and THACKER, Circuit Judges.

Petition for review granted, final agency action vacated, and remanded for further proceedings by published opinion. Judge Agee wrote the opinion, in which Judge Niemeyer and Judge Thacker joined.

3

ARGUED: Jonathan Grant Hardin, PERKINS COIE LLP, Washington, D.C., for Petitioner. Patrick Reinhold Jacobi, UNITED STATES DEPARTMENT OF JUSTICE, Washington, D.C., for Respondent. ON BRIEF: Jeffrey Bossert Clark, Assistant Attorney General, Jonathan D. Brightbill, Principal Deputy Assistant Attorney General, UNITED STATES DEPARTMENT OF JUSTICE, Washington, D.C.; Susan Stahle, Office of the General Counsel, UNITED STATES ENIVRONMENTAL PROTECTION AGENCY, Washington, D.C., for Respondent. Bryan M. Killian, Douglas A. Hastings, MORGAN, LEWIS & BOCKIUS LLP, Washington, D.C., for Amicus National Biodiesel Board. Sandra P. Franco, FRANCO ENVIRONMENTAL LAW LLC, Washington, D.C., for Amicus Producers of Renewables United for Integrity Truth and Transparency.

4

AGEE, Circuit Judge:

As part of the Clean Air Act, the Environmental Protection Agency (“EPA”)

administers a renewable fuel standard program, which requires refineries and other facilities to allocate a certain percentage of their fuel production to renewable fuels. Small refineries may petition to be exempt from the program’s requirements based on a showing that compliance would cause disproportionate economic hardship. Ergon-West Virginia, Inc. sought such an exemption, which the EPA denied. We previously vacated and remanded that decision as arbitrary and capricious. On remand, the EPA again denied Ergon’s petition. In this appeal, Ergon contends the EPA committed the same and related errors as in the initial decision. We have reviewed the record and, although the EPA’s post- remand decision largely cured the problems we previously identified, we conclude that Ergon has come forward with sufficient evidence undermining one aspect of the EPA’s decision. As a result, we again grant Ergon’s petition for review, vacate the EPA’s decision, and remand for further proceedings.

I.

A.

Our prior decision detailed the history of the renewable fuels statutes, so we assume familiarity with that decision and will not belabor their provisions here. See Ergon-West Virginia, Inc. v. EPA, 896 F.3d 600 (4th Cir. 2018) (“Ergon I”). In brief, the Energy Policy Act of 2005 created a renewable fuel standard program (the “RFS Program”) as Section 211(o) of the Clean Air Act. See 42 U.S.C. § 7545(o). The statute directs the EPA

5

Administrator to promulgate annual regulations to ensure that U.S. transportation fuel contains a certain volume of renewable fuels. Id. § 7545(o)(2)(A)(i). After the EPA determines and publishes the percentage of renewable fuel standards every refinery must meet for a particular year, refineries multiply that percentage by the volume of nonrenewable fuel they will produce or import to determine their renewable volume obligation. Id. § 7545(o)(3); 40 C.F.R. §§ 80.1405, 80.1407.

Refineries can satisfy their obligation by: (1) generating a sufficient number of renewables on their own; (2) purchasing a sufficient number of renewable fuel credits from entities that have separated more than their obligation; or (3) combining methods one and two. All renewable fuels are identified by a renewable identification number (“RIN”), which “is a unique number generated to represent a volume of renewable fuel.” 40 C.F.R. § 80.1401. Refineries demonstrate their compliance with the RFS Program by generating their own RINs or purchasing RINs from entities that generate them. Refineries that violate the RFS Program by failing to procure a sufficient number of RINs incur penalties. See 40 C.F.R. §§ 80.1428, 80.1460(c)(1), 80.1463; see also 42 U.S.C. § 7545(o)(5)(B) (stating that “[a] person that generates credits . . . may use the credits, or transfer all or a portion of the credits to another person for the purpose of complying with [the Program]”).

By statute, the RFS Program initially exempted small refineries from compliance.

42 U.S.C. §§ 7545(o)(1)(K), (o)(9)(A); 40 C.F.R. § 80.1441. This initial exemption allowed the EPA and the Department of Energy (“DOE”) time to conduct a study to determine whether compliance “would impose a disproportionate economic hardship on small refineries,” in which case the exemption would be extended automatically. 42 U.S.C.

6

§ 7545(o)(9)(A). All told, the DOE conducted two studies, the last of which led to its recommendation that, going forward, small refineries be allowed to apply for a continued exemption because of the continued risk that they would “suffer disproportionate economic hardship from compliance with the RFS program if blending renewable fuel into their transportation fuel or purchasing RINs increases their costs of products relative to competitors to the point that they are not viable” (the “2011 DOE Study”). J.A. 36. After surveying small refineries, the DOE developed a Scoring Matrix composed of two indices—the “Disproportionate Impact Index” and the “Viability Index”—to be used to determine whether a small refinery suffers disproportionate economic hardship and thereby qualifies for the continued exemption for that year.

7

J.A. 71, 74.

8

When a small refinery petitions for an exemption from the RFS Program, the DOE issues a report outlining how the refinery has performed on the Scoring Matrix. The DOE scores each of the subcategories, tallies the total score in each index, and divides the average by 2. If the refinery receives a score greater than 1 in both indices, the DOE recommends that it receive an exemption. As a practical matter, to obtain a high enough score, the refinery must earn “a score equivalent to at least four of the eight metrics for disproportionate impact at the moderate level (5), and a positive value for at least one of the three metrics for the viability index.” J.A. 75.

In December 2016, the EPA issued a memorandum detailing how it evaluates small-

refinery-exemption petitions. It stated that it “considers the findings of the DOE [studies] and a variety of economic factors,” such as “profitability, net income, cash flow and cash balances, gross and net refining margins, ability to pay for small refinery improvement projects, corporate structure, debt and other financial obligations, RIN prices, and the cost of compliance through RIN purchases.” J.A. 556. As a consequence of these criteria, refineries seeking an exemption attach their financial information to their petitions in an effort to prove the economic hardship that would result from compliance.

B.

Ergon is a West Virginia refinery that produces a maximum crude oil capacity of 23,000 barrels per day (well under the small refinery threshold of 75,000 barrels a day). The refinery primarily produces paraffinic lube oils, but transportation fuels are byproducts of that primary production. Of the transportation fuels Ergon produces, two-thirds are

9

diesel and one-third is gasoline. Nearly all of the transportation fuels are sold within a 170- mile radius of the refinery.

Ergon petitioned for a small refinery exemption in 2016. 1 It claimed economic hardship in complying with its RFS Program obligations because, though there is a widespread market for blended gasoline, no such market exists for blended diesel, which is the bulk of Ergon’s transportation fuel production. Ergon also cited its geographic location, asserting that regional customers preferred unblended diesel to blended diesel and that this local market preference further inhibited compliance.

The EPA denied the petition, agreeing with the DOE’s determination that Ergon did not achieve the requisite scores on the Scoring Matrix to qualify for an exemption. Ergon filed a timely petition for review, and this Court vacated and remanded for further proceedings. On remand, the EPA once again denied Ergon’s 2016 petition. Before reaching its decision, the EPA requested that the DOE more thoroughly explain the Scoring Matrix and how it arrived at Ergon’s score, which the DOE provided. The revised DOE Report scored Ergon the same way on both indices.

1 It applied for an exemption in 2014 and 2015 as well, but those petitions were denied and have never been at issue in this Court.

After our remand, Ergon petitioned for exemptions for compliance years 2017 and 2018. The DOE gave Ergon the identical score on the two indices, and the EPA rejected the 2017 and 2018 petitions for substantially the same reasons it had denied Ergon’s 2016 petition. Because the agency decisions are based on the same reasoning, we have consolidated Ergon’s three petitions for review. As a result, our analysis focuses on the 2016 denial, though our decision is applicable to all three agency actions.

10

J.A. 1073.

The EPA expressly adopted the DOE’s expanded explanations for how it arrived at the values assigned to Ergon’s Scoring Matrix. In addition, the EPA offered other economic considerations that led to its decision, which were “related to or in addition to” the DOE

11

Scoring Matrix. J.A. 1079. For example, looking to Ergon’s gross and net refining margins, the EPA pointed out that they both exceeded industry averages, that Ergon had “been competitive and profitable in recent years,” and that it had a “strong positive cash flow from operating activities in 2016” and the preceding years. J.A. 1080–81. In response to Ergon’s argument that compliance is more difficult for refineries that blend diesel rather than gasoline, the EPA observed that refineries had not experienced disproportionate economic hardship due to rising RIN prices because those costs were recouped in higher sales prices for diesel. In addition, it noted that “a loss or reduced profit on one of multiple product lines does not” demonstrate hardship, that Ergon had many options regarding blending fuels, and that it had improperly pointed to its chief competitor as evidence of hardship when the proper focus was hardship compared to the industry as a whole. J.A. 1081; accord J.A. 1079–84. Lastly, the EPA addressed at some length Ergon’s assertion of hardship from having to purchase RINs to satisfy its RFS Program obligations. 2 Ergon filed a timely petition for review of the EPA’s final agency decision. We have jurisdiction under 42 U.S.C. § 7607(b)(1).

2 The EPA also addressed the DOE’s Viability Index scoring decisions and offered additional explanations of other economic factors supporting its adoption of that analysis. Given that Ergon’s arguments focus on the Disproportionate Impact Index, the opinion will not recite those conclusions.

12

II.

A.

The Administrative Procedure Act instructs that, “[t]o the extent necessary to decision and when presented, the reviewing court shall decide all relevant questions of law, interpret constitutional and statutory provisions, and determine the meaning or applicability of the terms of an agency action.” 5 U.S.C. § 706. Furthermore, “[t]he reviewing court shall . . . hold unlawful and set aside agency action, findings, and conclusions found to be . . . arbitrary, capricious, an abuse of discretion, or otherwise not in accordance with law.” Id. § 706(2)(A). During its evaluation of the agency action, “the court shall review the whole record or those parts of it cited by a party, and due account shall be taken of the rule of prejudicial error.” Id. § 706. “The foregoing statutory criteria render our oversight highly deferential, with a presumption in favor of finding the action valid, yet the arbitrary-and-capricious standard does not reduce judicial review to a rubber stamp of agency action.” Friends of Back Bay v. U.S. Army Corps of Eng’rs, 681 F.3d 581, 587 (4th Cir. 2012) (internal quotation marks and citation omitted). The Court conducts this inquiry de novo. Id. “Agency action is arbitrary and capricious if the agency relies on factors that Congress did not intend for it to consider, entirely ignores important aspects of the problem, explains its decision in a manner contrary to the evidence before it, or reaches a decision that is so implausible that it cannot be ascribed to a difference in view.” United States v. F/V Alice Amanda, 987 F.2d 1078, 1085 (4th Cir. 1993).

Further, given the context of this appeal, the Court limits its review to whether the EPA’s reliance on the DOE criteria was arbitrary and capricious as opposed to reviewing

13

whether the DOE criteria themselves were arbitrary and capricious. E.g., Dow Agrosciences LLC v. Nat’l Marine Fisheries Serv., 637 F.3d 259, 266 (4th Cir. 2011) (“When a court of appeals reviews the EPA’s reliance on a [report issued by another agency], it would determine only whether the EPA’s reliance was arbitrary and capricious.”). “While the action agency is not required to undertake an independent analysis of another agency’s conclusions, it may not blindly adopt those conclusions. Thus, an action agency’s reliance on a facially-flawed report is arbitrary and capricious.” Ergon I, 896 F.3d at 610 (internal quotation marks and citations omitted).

B.

Ergon argues that the EPA repeated the errors this Court identified in Ergon I by again relying on the DOE’s facially deficient scoring metrics to deny the petition. First, it asserts as arbitrary and capricious the EPA’s reliance on the DOE’s failure to score local market acceptance of E85 and biodiesel (Sections 1(c)(ii) and (iii)) as well as Ergon’s capacity to blend biodiesel and other advanced biofuels (Sections 2(b)(ii) and (iii)). Second, Ergon maintains the EPA acted contrary to the weight of the evidence by ignoring its proof of particularized hardship arising from substantial RIN costs (Section 2(d)). And third, it contends the EPA should not have relied on the DOE’s inconsistent definitions of the term “refining,” which treated lubricant production as an additional economic line of production for purposes of Section 1(b) and yet excluded lubricants for purposes of Section 2(a). Ergon argues that each of these deficient scoring methods negatively impacted its DOE score and contributed to the EPA’s decision to deny its petition. We address each contention in turn.

14

1. Sections 1(c) and 2(b)

As we observed in Ergon I, Section 1(c) “accounts for the refinery’s geographic location by evaluating how likely the local market will accept transportation fuels blended with renewable fuels. Although the category lists subcategories for E10 (a fuel mixture of 10% ethanol and 90% gasoline), E85 (a fuel mixture of 85% ethanol and 15% gasoline), and biodiesel, the last two are reserved for later evaluation.” 896 F.3d at 605–06 (internal quotation marks and citation omitted).

“Section 2(b) evaluates the capacity of the refinery to blend its nonrenewable fuels with renewable fuels, with a lower capacity indicating greater impairment. Although Section 2(b) has subcategories of ethanol, biodiesel, and advanced biofuels, only the first is scored, with the others reserved for later evaluation.” Id. at 606 (internal quotation marks and citation omitted).

a. Ergon I

In its original scoring of Section 1(c), the DOE gave Ergon’s petition a zero for local market acceptance of E10 and provided no score for E85 or biodiesel. The EPA accepted that scoring without comment.

Similarly, the DOE scored only the first subcategory of Section 2(b) (ethanol), for which it gave Ergon a zero, again providing no score for the other two subcategories (biodiesel and advanced biofuel blending). The EPA relied on that score, too, without additional explanation.

In Ergon I, we observed that if the DOE had scored either of the two unscored subcategories in either Section 1(c) or 2(b), Ergon “would have achieved a score greater

15

than 1 and likely earned a small refinery exemption.” Id. at 611. We thus held that the DOE’s treatment of Sections 1(c) and 2(b) was “plainly arbitrary as it treats unfairly those facilities where diesel makes up a substantial percentage of their transportation fuel production.” Id. We observed that the DOE’s “errors [were] readily apparent on the face of the DOE’s report as the index lists ‘[n]ot available’ next to biodiesel in Section 1(c) and ‘not used’ next to biodiesel blending in Section 2(b).” Id. In concluding that the EPA—the agency whose decisionmaking we were, and are, reviewing—acted arbitrarily and capriciously, we stated:

Because the DOE’s recommendation was clearly flawed on its face, a clear error of judgment was made when the EPA relied without explanation on the DOE’s Report for its denial of Ergon’s 2016 waiver petition. In addition, the EPA did not conduct any independent analysis regarding the subject matter of Sections 1(c) and 2(b).

Id. at 611–12 (internal quotation marks and citation omitted). We observed that it was unclear from the EPA’s decision letter how much weight it had given to these Sections and to the DOE’s Report. As such, we could not “determine whether the EPA would have reached the same conclusion had the DOE submitted a proper analysis or had the EPA addressed the DOE’s failure to analyze Sections 1(c) and 2(b).” Id. at 612. We reiterated that the EPA could not “turn a blind eye to errors and omissions apparent on the face of the” DOE’s Report and that, by doing so, “the EPA ignored important aspects of the problem.” Id. (internal quotation marks and citation omitted).

b. On Remand

In response to the EPA’s request for additional explanation of its scoring methods and decisions, the DOE clarified that, although Section 1(c) contains three subcategories

16

of fuels, “only the consumer market acceptance of E10 (subsection 1(c)(i)) has ever been scored for any refinery under the 2011 Study because volumetric data for retail sales is not available at the local level or at the national level for E85 or biodiesel.” J.A. 1035. It observed that the federal government does not collect that data, and that other sources also either failed to collect the full range of data it needed in order to have comparable information to that available for E10. Because it lacked the “retail level data . . . necessary to determine consumer acceptance for a comparison to determine if there is a disproportionate economic hardship,” J.A. 1036, the last two subcategories had never been scored for any exemption petition. Instead, the notation on the index that those two subcategories were “[r]eserved for later evaluation” reflected, in essence, their placeholder positions in case it became possible to consider these factors at some point in the future. J.A. 1035.

In like manner, the DOE amplified its explanation of Section 2(b), including why, although it contained three subcategories, the DOE scored only one of them (ethanol). The DOE observed that its 2011 Study had provided guidance for ethanol that allowed it to score that subcategory uniformly, but it “did not provide explanatory text to inform the evaluation for” the remaining two subcategories (biodiesel and advanced biofuel blending). J.A. 1038. It briefly described the last two subcategories, but noted that insufficient data meant it could not score them. Lastly, it observed that Ergon’s evidence indicated it had not blended other advanced biofuel in 2016, so that subcategory would not have received a score regardless of the DOE’s scoring criteria.

17

The EPA reviewed the DOE Scoring Matrix, agreed with that analysis, and provided additional reasons for denying Ergon’s petition related to the factors these sections address. Specifically, it explained why it agreed with the DOE that the “appropriate data” did not exist to score the second and third subcategories of Section 1(c), not just for Ergon’s petition, but for any petition filed to date. J.A. 1076. It further agreed that the DOE lacked information about the retail sales volume of E85, which would be necessary to “measure ‘local market customer acceptance.’” Id. The EPA explained that wholesale sales data was not a reliable measure for this inquiry because E85 is often sold at that level, but then blended with other fuels “at the point of retail sale,” so consumers are not in fact purchasing E85. Id. Similarly, the EPA observed that “available biodiesel data” is “unsuitable” for deriving a score for this subcategory because the data covers “large, multi-state geographical regions” that do not adequately tailor the data to assess local market acceptance. J.A. 1076–77. In sum, it concluded that insufficient data existed that would allow a comparison across refineries to reach “any conclusions about disproportionate impact.” J.A. 1078.

The EPA also addressed Section 2(b) in its expanded analysis, discussing what it viewed as inconsistencies in Ergon’s petition and rejecting some of its assertions about local market acceptance. For example, it noted that Ergon complained of a lack of market acceptance due to the local market preferring unblended fuel and argued that the colder climate in the region made the refining process for its crude oil more difficult. But despite these asserted barriers, Ergon had “invested more than $1 million to install renewable fuel blending equipment,” sold its blended diesel in the region, and was planning “a second

18

blending project which would allow the refinery to blend its diesel production with up to 5% biodiesel and sell this blended fuel” regionally. J.A. 1079–80 (internal quotation marks omitted). The EPA concluded “[i]t makes no sense for [Ergon] to make additional investments to expand its biodiesel blending capabilities if it anticipates being unable to sell that product” and the local market apparently “accepts renewable fuel blends well enough to encourage” that expansion. J.A. 1080.

c. Analysis

In its petition for review, Ergon asserts that the EPA neglected to conduct any independent analysis of the local market acceptance of E85 or biodiesel for Section 1(c) or for biodiesel and advance biofuel blending for Section 2(b). In its view, the EPA “merely parrot[ed] DOE’s claims,” and thus relied on the DOE Report’s facially deficient scoring. 3 Opening Br. 35. In sum, it argues that because Ergon I already held that (1) relying on a facially deficient Scoring Matrix was arbitrary and capricious, and (2) obtaining a different score on any of these subcategories likely would have resulted in a score that entitled it to an exemption, the post-remand decision requires vacatur, too, because the EPA did not fix the identified problems. Further, it contends the DOE’s and EPA’s conclusions are faulty because sufficient data of the relevant categories exists to provide a score.

We disagree with Ergon’s understanding of our prior decision and with its interpretation of the EPA’s post-remand decision on this issue. Contrary to Ergon’s

3 Ergon challenges only the decision to not to score the subcategories discussed above, which it claims would have resulted in a favorable overall score. It does not challenge the EPA’s reliance on Ergon receiving scores of zero on the other subcategories of Sections 1(c) and 2(b).

19

suggestion, our prior decision did not directly critique the DOE Scoring Matrix or conclude that the DOE should have given Ergon a different score as to any section. Our holding was narrower, focusing on the EPA’s failure to adequately explain its reasons for denying Ergon’s petition because it appeared to have relied on the DOE Scoring Matrix even though there was readily observable existence of subcategories favorable to Ergon that the DOE had left unscored. For example, we determined that the EPA made a “clear error of judgment” by relying on the DOE’s Scoring Matrix “without explanation” and without “conduct[ing] any independent analysis regarding” the decision not to score Sections 1(c) and 2(b). Ergon I, 896 F.3d at 611 (internal quotation marks and citation omitted). And we concluded the EPA’s decision did not permit the Court to “determine whether the EPA would have reached the same conclusion had the DOE submitted a proper analysis or had the EPA addressed the DOE’s failure to analyze” those Sections. Id. at 612 (emphasis added). In short, nothing in our decision suggests that the EPA was prohibited from relying on the DOE’s Scoring Matrix or that the EPA could only redress the issues the Court identified by requiring the DOE to rescore Ergon’s petition. To the contrary, we plainly and repeatedly indicated that the reason the EPA’s decision was arbitrary and capricious was because of the inadequacy of the explanation.

Our review of the record on remand confirms that the EPA redressed the deficiencies we previously identified in its analysis of Sections 1(c) and 2(b). As recounted earlier, the EPA first directed the DOE to provide an additional explanation for its Scoring Matrix and the basis for its recommendations to the EPA. This was entirely consistent with the statutory mandate that the EPA make its decisions about exemption petitions “in

20

consultation with the Secretary of Energy.” 42 U.S.C. § 7545(o)(9)(B)(ii); see also Lion Oil Co. v. EPA, 792 F.3d 978, 982 (8th Cir. 2015) (“EPA did not arbitrarily use DOE’s scoring decision. Rather, EPA did as Congress directed[.]”). Although the EPA is the acting agency that makes the final decision as to exemption petitions, it functions well within its statutory directive to rely on the DOE’s expertise so long as a basis for its decision is apparent in the record. That basis was not present in the earlier decision, but it is here.

And that more robust explanation begins with the DOE’s own significantly more detailed explanation of the Scoring Matrix and how it scored Ergon compared to the streamlined review process it had previously provided. Specifically, for Sections 1(c) and 2(b), the DOE has now explained at some length why the unscored subcategories are included on the Scoring Matrix. It indicated that the unscored subcategories were, in effect, placeholders established in its 2011 Study for possible future consideration, but that they had never been considered in any review of an exemption petition to date. And it also explained why informative data relevant to the sections was unavailable, making it impractical to assess the two subcategories left unscored.

The DOE’s thorough explanation of the Scoring Matrix and its rationale when scoring Ergon provided a sounder framework for the EPA to explain its own reasoning in making the final agency decision as to Ergon’s petition. See Appalachian Voice v. State Water Control Bd., 912 F.3d 746, 753 (4th Cir. 2019) (reiterating that a final agency decision is not arbitrary or capricious when it examines “[t]he relevant data and articulate[s] a satisfactory explanation for its action including a rational connection between the facts found and the choice made” (internal quotation marks and citation

21

omitted)). When the EPA adopted the DOE’s reasoning, it was not simply parroting a conclusion and outcome, it was adopting a substantive analysis of the Scoring Matrix. So even the EPA’s adoption of the reasoning is based on a better-analyzed DOE recommendation. Moreover, though it’s true that the EPA briefly recited the DOE’s conclusions and summarily expressed that it “agrees with DOE’s” evaluation, that was not the end of its analysis. J.A. 1076–77, 1078–79. As reflected in the earlier summation of the EPA’s decision, the agency also independently addressed and defended its decision based on a variety of other economic factors after explaining why it rejected some of Ergon’s arguments in favor of its petition. In sum, the EPA has now addressed the DOE’s Scoring Matrix, explained the basis for relying on a score that left certain subsections unscored, and independently analyzed the question before it. This part of its analysis was neither arbitrary nor capricious and is not a basis for reversing and remanding the final agency decision.

2.

Next, Ergon challenges the EPA’s handling of Section 2(d), which considers “whether the refinery generates revenue by selling RINs or must purchase RINs in the market.” Ergon I, 896 F.3d at 606.

a. Ergon I

In its original recommendation and Scoring Matrix, the DOE did not give Ergon a score for Section 2(d), observing broadly that it “has not scored this category for any hardship petition evaluations.” Id. at 607. In Ergon I, we observed that the “failure to score this section is apparent on the face of the DOE’s Report, and that failure negatively

22

impacted Ergon’s petition.” Id. at 612. However, we concluded that the EPA’s final decision “was not arbitrary and capricious in and of itself as the EPA did not rely on that factor in its determination.” Id. To the contrary, the EPA observed that in the time since the DOE’s 2011 Study, it had concluded that refineries did not experience a disproportionate economic hardship from having to purchase RINs because the prices fluctuate such that “no net cost of compliance [results] for the refinery.” Id. This “specific response address[ed] why it . . . implicitly disregard[ed] the scoring of the factor in the DOE’s” recommendation as to Section 2(d). Id.

However, we did conclude that the EPA’s analysis of this factor was arbitrary and capricious because it “ignored specific evidence suggesting that [RIN prices] had a negative effect” on Ergon. Id. at 613. We pointed to the EPA’s “solitary statement” addressing Ergon’s hardship evidence related to this factor, as well as its focus on evidence applicable to “the refining industry as a whole,” instead of Ergon’s evidence that “its refinery cannot pass the RIN costs on to purchasers because of the local market’s low acceptance of blended diesel.” Id. We held that the EPA had acted arbitrarily and capriciously by engaging in a “cursory consideration and fail[ing] to address Ergon’s specific evidence regarding RIN costs.” Id. Accordingly, we granted the petition for review on this separate ground and remanded for the EPA “to properly address Ergon’s petition with regard to RIN costs.” Id.

b. On Remand

In its EPA-directed additional analysis, the DOE acknowledged Section 2(d)’s purpose “to determine whether RIN purchases or sales constitute a net revenue or cost,”

23

but explained that it was “not scored in the 2011 Study because there was a lack of consistency among responders.” J.A. 1040 (internal quotation marks omitted). It asserted that without “information available to compare a refinery’s RIN cost/revenue with an industry average obtained from study of refineries’ data (rather than a study of national price data),” it could not determine if a disproportionate economic hardship existed. Id. For that reason, it had not scored Section 2(d) when considering any refinery’s petition.

In its discussion of the Scoring Matrix, the EPA agreed with the DOE’s decision not to score Section 2(d) given that the “DOE had no baseline against which to compare a small refinery’s RIN costs/revenues for the purpose of” the exemption. J.A. 1079. Later, the EPA expounded on its reasons for rejecting Ergon’s claim that the increased cost of RINs had decreased its profitability. It noted that Ergon’s arguments did “not account for the fact that [its] competitors do bear . . . compliance costs, either through RIN or renewable fuel purchases.” J.A. 1082. Specifically, it observed that in considering exemption requests, it considered “each refinery’s individual situation and its conditions relative to the industry average” rather than as compared to “any one competitor,” as Ergon tended to advocate. Id. It also noted that its research indicated “that the market has adjusted to treat RFS [Program] compliance costs like any other operating costs . . . . As with any other operating cost, when RIN prices increase, the market prices for the obligated fuels (unblended gasoline and diesel) rise to enable obligated parties to recoup their cost of RFS [Program] compliance.” J.A. 1083. For that reason, the EPA concluded Ergon’s “status as a price taker is immaterial, as the price [Ergon] is being forced to take [by purchasing RINs] already includes the cost of RFS [Program] compliance.” Id.

24

c. Analysis

On appeal, Ergon argues that the EPA failed to heed Ergon I’s admonition to analyze the specific evidence of its hardship, and instead arbitrarily and capriciously relied on the DOE’s decision to assign a score of zero to Section 2(d). Ergon asserts it should have received a score of ten because it had to purchase RINs, and it maintains that— contrary to the DOE’s representation, which the EPA accepted—no comparative data should have been necessary to score this “binary analysis of whether RFS [Program] compliance is a net cost to the refinery.” Opening Br. 40. Ergon maintains that the EPA “merely regurgitated DOE’s explanation” and “recycle[d] its old refrain that Ergon can supposedly recoup its RIN costs by passing them through to purchasers of its fuel.” Opening Br. 41. Lastly, Ergon points to evidence that it had to purchase RINs as proof “that RINs were a substantial net cost to its refinery.” Id.

Ergon’s argument lacks merit. At the outset, to the extent Ergon directly challenges the DOE’s decision to assign it a score of zero and explanation of what Section 2(d) is designed to address, we cannot consider these arguments. As we noted in Ergon I, Ergon challenged the EPA’s denial of its exemption petition, so our review is limited to

whether the EPA’s reliance on the DOE’s Report is arbitrary and capricious.

See Dow AgroSciences LLC v. Nat’l Marine Fisheries Serv., 637 F.3d 259, 266–67 (4th Cir. 2011) (“When a court of appeals reviews the EPA’s reliance on a [report issued by another agency], it would determine only whether the EPA’s reliance was arbitrary and capricious); City of Tacoma v. FERC, 460 F.3d 53, 75 (D.C. Cir. 2006) (“Accordingly, when we are reviewing the decision of an action agency to rely on [another agency’s report], the focus of our review is quite different than when we are reviewing a [report]

directly. In the former case, the critical question is whether the agency action’s reliance was arbitrary and capricious, not whether the [report] itself is somehow flawed.”).

25

896 F.3d at 610; see also id. (stating that we could not say that the EPA’s reliance on the DOE’s scoring was arbitrary and capricious because the arguments went to “the DOE’s scoring methodology and are not apparent on the face of the DOE’s report”).

What is more, our review of the EPA’s decision indicates—as was true of the original decision—that it largely “did not rely on [Section 2(d)] in its determination” because it had concluded in the time since the DOE’s 2011 Study that the data did not support using this category as a basis for determining hardship, regardless of its presence on the Scoring Matrix. Compare id. at 612, with J.A. 1081. Yet again, the EPA expressly stated its disagreement with the 2011 DOE Study’s inclusion of this factor as evidence of hardship, finding “as a general principle” that

a refinery does not experience disproportionate economic hardship simply because it may need to purchase a significant percentage of its RINs for compliance from other parties, even though RIN prices have increased since the DOE study, because the RIN prices lead to higher sales prices obtained for the refineries’ blendstock, resulting in no net cost of compliance for the refinery.

J.A. 1081. Thus, while the EPA agreed with the DOE’s decision not to score Section 2(d), it’s equally clear that a refinery’s “RINs net revenue or cost” was not something the EPA deemed to be an objective measure of a disproportionate economic hardship.

Further, to the limited extent Ergon challenges the adequacy of the EPA’s rejection of its specific evidence of hardship arising from its purchase of RINs, the EPA’s decision belies that assertion. Considering the EPA’s analysis as a whole, it cured the problem requiring remand in Ergon I. There, we held the EPA had “failed to squarely address Ergon’s petition with regards to RIN costs” and instead “cited generally to an industry-

26

wide study and a nonspecific nationwide trend” as the sole basis for its conclusions. 896 F.3d at 613. Here, the EPA thoroughly discussed Ergon’s purported evidence of hardship, explained why it rejected Ergon’s arguments, and set out other factors that led it to reach an opposite conclusion. For instance, the EPA explained that the mere fact that Ergon had to spend money to purchase RINs to comply with its obligations is not, in itself, evidence of a particular hardship. It considered what led to Ergon’s heightened operating costs apart from the need to purchase RINs, and in particular, the “specialty products” that are its primary purpose, “a business model very different from that of most U.S. refineries.” J.A. 1080. It addressed Ergon’s profitability, capital investments, and net incomes over the span of 2016 and the preceding years. And it looked at a host of other factors as well, including Ergon’s contention that while other refineries could pass along their compliance costs, it could not. In particular, the EPA concluded that Ergon’s arguments failed to account for other refineries incorporating the costs associated with RFS Program compliance into other aspects of their production that allowed them to manufacture and blend compliant renewable fuels, which were not necessarily as visible as a one-time compliance method through the purchase of RINs. And it explained why Ergon’s evidence that it had to purchase RINs while a specific competitor refinery did not fail to consider hardship across the industry. At bottom, the EPA’s post-remand discussion of Ergon’s evidence connected the dots left unaddressed in its original decision. Ergon has failed to show any basis for vacating the final agency decision arising from the EPA’s discussion of hardship arising from the purchase of RINs.

27

3.

Lastly, Ergon contends the EPA acted arbitrarily and capriciously by continuing to rely on the DOE’s decision to define and treat the concept of “refining” differently for purposes of Sections 1(b) and 2(a). “Section 1(b) considers a refinery’s business lines other than refining and marketing—in particular upstream operations such as exploration and development that are less correlated with refining—which may insulate the refinery from the volatility of refining margins.” Ergon I, 896 F.3d at 605 (internal quotation marks and citation omitted). Section 2(a) reflects the refinery’s “relative refining margin—essentially its refining revenue minus its refining costs, or refining profit—compared to the three-year industry average.” Id. at 606.

a. Ergon I

In its initial report, the DOE scored both Sections 1(b) and 2(a) as zeros, and Ergon challenged the EPA’s reliance on that decision on appeal. We agreed that Ergon pointed out an “apparent contradiction” in how the DOE defined “refining” for purposes of these Sections. Id. at 610. Specifically, “[i]n Section 1(b) (the ‘other business lines besides refining and marketing’ factor), the DOE separated Ergon’s refining from its lube oil production [and] considered the latter as an ‘other business line[] besides refining and marketing,’” while “in Section 2(a) (the ‘relative refining margin’ factor), the DOE treated Ergon’s lube oil production as refining for purposes of the relative refining margin measure[.]’” Id. But we held that because “these arguments go to the DOE’s scoring methodology and are not apparent on the face of the DOE’s Report,” “we cannot say that the EPA’s reliance on the DOE’s scoring of these factors was arbitrary and capricious.” Id.

28

b. On Remand

The DOE used the same contradictory definitions of “refineries” in Sections 1(b)

and 2(a), and scored Ergon the same way on remand. Because it is central to our resolution of Ergon’s petition, we’ll focus on the explanation of Ergon’s other lines of business in Section 1(b).

When explaining what it considered as part of the Section 1(b) score, the DOE said it considers “both the refinery level and the corporate business level businesses.” J.A. 1034. It further represented that it had “always considered an applicant’s additional lines of business, in particular upstream operations,” and that it “has consistently considered lubricant manufacture at a refinery as another line of business and has applied this interpretation for all petitioners.” Id. And it indicated that if a refinery has other lines of business, the refinery receives a score of zero, but that if it does not have them, it receives a ten. Relying on Ergon’s production of lubricants and its parent company’s other lines of business, the DOE again gave Ergon a score of zero for Section 1(b).

The EPA, in turn, “agree[d] with DOE’s scoring of 0.” J.A. 1076; accord J.A. 1074– 76. In so doing, it quoted the DOE’s representation about “consistently consider[ing] lubricant manufacture at a refinery as another line of business and [applying] this interpretation for all petitioners.” J.A. 1075. It also relied on Ergon’s parent company’s other lines of business.

When scoring Section 2(a), the DOE looks to a refinery’s “relative refining margin”

(revenue minus costs) compared to the three-year industry average. The DOE did not eliminate Ergon’s lube oil production when making this calculation, a decision that treated

29

it as “refining” for purposes of this section even though it was excluded for purposes of 1(b)’s definition of “refining.” J.A. 1037–38. Based on the values used, the DOE concluded that Ergon’s three-year average net refining margin for 2013 to 2015 was “well over the industry average,” so it scored this metric as a zero. J.A. 1038.

The EPA agreed with the DOE’s approach, and rejected Ergon’s argument that the definitions were inconsistent. J.A. 1077–78.

c. Analysis

On appeal, Ergon challenges the EPA’s “knowing reliance” on the DOE’s contradictory treatment of the terms in Sections 1(b) and 2(a). It asserts that the EPA acted arbitrarily and capriciously in continuing to accept the DOE’s approach despite being alerted to this inconsistency in Ergon I. Moreover, Ergon contends that the concept of “refining” should be treated in the same way when scoring both sections and that, if it had been, it would have received a score in the range the DOE has said would lead it to recommend an exemption.

Relatedly, Ergon contends the EPA acted arbitrarily and capriciously in relying on the DOE’s definition of “refining” in Section 1(b) to exclude other refined petroleum product lines such as lube oils. It asserts that this definition does not comport with how the DOE’s 2011 Study and the U.S. Energy Information Administration define “refining.” Ergon observes that its “transportation fuel and lube oils are petroleum products that are refined from crude oil,” so the lube oils should be considered part of its refining operations rather than being deemed an upstream or downstream operation. Opening Br. 45. And it asserts that because “every refinery makes petroleum products other than transportation

30

fuels,” the relied-on definition of refinery “effectively read[s] [Section] 1(b) out of existence.” Opening Br. 46–47.

Lastly, in its Reply Brief, Ergon pointed to a claim of proof that “a similarly situated small refinery with the same lines of business as Ergon—including substantial lubricant production—received a score of ‘10’ for [Section 1(b)] for four straight years from 2014 through 2017.” Reply Br. 11. Ergon attached, under seal, an affidavit from the other refinery’s Chief Financial Officer confirming both the statement and the accuracy of excerpted EPA decision documents reflecting how it had scored that refinery on Section 1(b). 4 Ergon argued that this evidence showed that “[e]ven a cursory review of the decision documents in EPA’s possession would have revealed that [its] statement [about uniformly defining ‘refinery’ for all exemption petitions] is demonstrably false.” Id. 5 The EPA contends its reliance on the DOE was not arbitrary and capricious on the merits, and urges us not to take into account Ergon’s supplemental materials. In support of the second argument, it raises three points. First, it notes that the Court’s review is limited to the record before the EPA at the time of its final decision and contends Ergon should not

4 Simultaneous with the filing of its Reply Brief, Ergon moved for leave to file the supplemental material as an attachment. By order dated July 17, 2020, the Court granted Ergon’s motion.

5 We typically do not allow parties to raise new issues in a Reply Brief. Cavallo v.

Star Enter., 100 F.3d 1150, 1152 n.2 (4th Cir. 1996). But we exercise our discretion to do so here because: (1) Ergon is presenting an additional reason to support an existing argument rather than raising an entirely new issue; and (2) we have no basis for questioning its representation that the reason it had not made this argument and alerted us to the supplemental materials earlier arose from the timing of obtaining and coordinating the scope of permission to rely on confidential business information about a competing small refinery.

31

be permitted to supplement the administrative record on appeal. Second, it maintains that the evidence about the other refinery does not contradict the EPA’s position that its consistent practice has been not to score lubricants as another line of business for purposes of scoring Section 1(b) because, at most, it demonstrates a departure granted to one refinery based on factors unique to that refinery. Third, it suggests that the Court could rely on the merits of the EPA’s additional grounds for relying on the DOE’s scoring decision (namely, Ergon’s parent corporation’s lines of business), because the EPA no longer relies in this case on the ground that it had consistently defined “refinery” in Section 1(b) to mean the same thing when considering exemption petitions. Consistent with its last argument and after Ergon tendered the supplemental materials, the EPA filed an amended Response Brief that removed that ground as a basis for defending the EPA’s final decision.

We have reviewed the administrative record and the supplemental materials and conclude that remand is appropriate, limited to this one aspect of the EPA’s decisionmaking process. The EPA is correct that we are ordinarily restricted to the administrative record that the agency had before it when it made its decision. See Camp v. Pitts, 411 U.S. 138, 142 (1973) (“[T]he focal point for judicial review should be the administrative record already in existence, not some new record made initially in the reviewing court.”); Sanitary Bd. of Charleston, W. Va. v. Wheeler, 918 F.3d 324, 334 (4th Cir. 2019) (“A party challenging an agency bears a special burden of demonstrating that the court should reach beyond the record, either to examine information that should have been before the agency but was not, or to introduce extra-record evidence that the agency actually relied on that was omitted from the administrative record.”). But this case does not

32

present the scenario where a petitioner seeks to augment the record with more support in favor of their original petition to the agency. To the contrary, the supplemental materials go to the agency’s purported methodology in reviewing an exemption petition, and consists of documentation that was—and has been—squarely in the agency’s possession and knowledge.

In short, the evidence Ergon relies on to call into question the EPA’s grounds for denying its petition are EPA decision documents in another case. The EPA should have known whether it had previously approved the DOE’s apparent use of different criteria to score Section 1(b) than the criteria used to score Ergon’s petition. And it—along with the DOE—both had access to their respective decisionmaking documents about all small refinery exemption petitions, including the ones excerpted in the supplemental materials. But Ergon did not. Moreover, Ergon had no basis upon which to compel confidential documents about the disposition of other petitions, nor would it ordinarily have any cause to know that such contradictory statements existed in those documents. The nature of this evidence, and the doubt it casts with respect to the EPA’s grounds of decision in Ergon’s case, makes it appropriate for us to consider these documents despite their not being part of the administrative record. See Sanitary Bd. of Charleston, 918 F.3d at 334.

Having determined that it’s appropriate to consider the supplemental materials, we briefly discuss how they affect Ergon’s petition for review. Because the supplemental materials were filed under seal, we will not detail their contents except to note that they support Ergon’s representation that the DOE has not scored every refinery that produces lubricants the same way that it scored Ergon in Section 1(b). We are especially concerned

33

because this type of contradictory evidence would not normally be something we—or a refinery—would have access to when considering a final agency decision. We rely on agencies to accurately and correctly represent how they reach their decisions, particularly when they make such sweeping statements as the ones made in this case representing that it applied the same standards for Ergon’s Scoring Matrix that it has for every other exemption petition. Suffice it to say, the supplemental materials do call into question the veracity of the DOE and the EPA’s broader representations in reviewing Ergon’s petition about how Section 1(b) is scored and how the EPA determines whether a refinery has other business lines that might impact its refinery’s burden of complying with the RFS Program.

Contrary to the EPA’s attempt to recharacterize the statements in the record, we also note that the supplemental materials directly call into question both the DOE and the EPA’s unequivocal representations during the agency proceedings. The DOE broadly explained that it had “always considered an applicant’s additional lines of business, in particular upstream operations,” and that it had “consistently considered lubricant manufacture at a refinery as another line of business and ha[d] applied this interpretation for all petitioners.” J.A. 1034 (emphases added). The EPA expressly reiterated the same in issuing its final decision. In short, the record belies the EPA’s current effort to recharacterize these prior statements as something short of categorical representations about how it had scored Section 1(b) for all petitioners. As we have previously recognized, an agency acts arbitrarily and capriciously when it “explains its decision in a manner contrary to the evidence before it.” F/V Alice Amanda, 987 F.2d at 1085. And that is how it would appear the EPA has acted in this case.

34

Nor can the EPA remedy the apparent inconsistency by retracting its appellate reliance on this part of the agency’s rationale for its final decision. To be sure, the DOE offered two principal explanations for why it scored Section 1(b) as it did, and the parent- corporations line of reasoning is not implicated by the supplemental materials, which relate solely to whether lubricants have been consistently deemed other lines of business for purposes of this category. But in making its final decision, the EPA expressly relied on both grounds, with no indication that either would have been independently sufficient. Because we are tasked with reviewing whether the EPA’s decision was arbitrary or capricious, it follows that we cannot remove parts of that decision from the whole and still conclude that it constituted “reasoned decisionmaking.” Judulang v. Holder, 565 U.S. 42, 53 (2011). Put another way, we review the agency’s reasoning as of the time it made its decision, and one part of that reasoning has now been undermined by the EPA’s own prior decision documents. To excise that ground and conclude that the EPA’s decision was not arbitrary or capricious would be to substitute our judgment for the EPA’s, which we cannot do. Citizens to Preserve Overton Park, Inc. v. Volpe, 401 U.S. 402, 413–14 (1971); see also Hermes Consol., LLC v. EPA, 787 F.3d 568, 571 (D.C. Cir. 2015) (“[W]e are unable to conclude that EPA would have reached the same decision absent its mistakes.”).

In view of the supplemental materials, we conclude that part of the EPA’s basis for accepting the DOE’s reasoning as to Section 1(b) has been reliably called into question. Accordingly its decision was arbitrary and capricious. See Roe v. Dep’t of Defense, 947 F.3d 207, 220 (4th Cir. 2020) (“Agency action is arbitrary and capricious when the agency . . . entirely failed to consider an important aspect of the problem, offered an explanation

35

for its decision that runs counter to the evidence before the agency, or is so implausible that it could not be ascribed to a difference in view or the product of agency expertise.” (internal quotation marks and citation omitted)). We take no view on the merits of Ergon’s exemption petition on remand, how the DOE should score Section 1(b), or what weight the EPA should give to the DOE’s approach to Section 1(b). Further, by remanding we express no view on how the EPA should act to cure the flawed analysis that leads us to grant Ergon’s petition, vacate the agency’s decision, and remand. It’s possible that the DOE or the EPA could explain why it’s appropriate to apply a different standard in some cases. Or they may reach a different result regarding Ergon’s petition upon further review of the approaches previously taken when applying Section 1(b). Alternatively, the DOE or the EPA may be able to harmonize or distinguish the decisions at issue in the supplemental materials from the evidence supporting Ergon’s petition. For now, however, we are constrained to conclude that the EPA acted arbitrarily and capriciously in denying Ergon’s petition. We therefore vacate and remand for further proceedings consistent with this opinion.

One final observation is necessary. As discussed above, Ergon’s arguments as to Section 1(b) include challenges not only to the scoring of this one category of the DOE’s Scoring Matrix, but also to the apparently contradictory definitions of “refinery” used in

36

Section 1(b) and (2)(a). Because of the threshold problem with the rationale provided for the Section 1(b) scoring, we do not reach this secondary issue at this time. 6

III.

For these reasons, we grant Ergon’s petition for review, vacate the EPA’s decision, and remand the case for further proceedings consistent with this opinion.

PETITION FOR REVIEW GRANTED, FINAL AGENCY ACTION VACATED, AND REMANDED FOR FURTHER PROCEEDINGS

6 We find this course to be the proper remedy here and reject Ergon’s argument that the EPA’s decision on remand warrants the irregular relief of awarding it an exemption and ordering the EPA to issue it RINs equaling the monetary value of the amount it has spent on RINs to comply with its RFS Program obligations in 2016, 2017, and 2018. That novel theory of relief finds no support in our case law and nothing on this record suggests that such an extraordinary remedy would be appropriate under the circumstances presented here. Instead, we follow our ordinary course upon concluding that the agency has acted arbitrarily or capriciously by vacating the agency decision and remanding for further proceedings.