Reset A A Font size: Print

United States Court of Appeals,District of Columbia Circuit.

HERMES CONSOLIDATED, LLC, Doing Business as Wyoming Refining Company, Petitioner v. ENVIRONMENTAL PROTECTION AGENCY, Respondent.

No. 14–1016.

Decided: June 02, 2015

Before GARLAND, Chief Judge, and TATEL and SRINIVASAN, Circuit Judges. Eric D. Miller argued the cause for petitioner. With him on the briefs were LeAnn Johnson Koch and William Pedersen. Justin D. Heminger, Attorney, U.S. Department of Justice, argued the cause for respondent. With him on the brief was Sam Hirsch, Acting Assistant Attorney General.

The Environmental Protection Agency administers a renewable fuels program under which oil refineries must satisfy annual obligations concerning production of renewable fuels. Petitioner Wyoming Refining Company operates an oil refinery located in Newcastle, Wyoming. WRC is subject to EPA's renewable fuels program, but obtained an exemption through 2012. WRC unsuccessfully petitioned EPA for an extension of its exemption through 2014. The company now seeks review of EPA's denial.

We reject WRC's various challenges other than those identifying two mathematical errors in EPA's independent analysis of WRC's financial data. EPA concedes those errors, and we are unable to conclude that EPA would have reached the same decision absent its mistakes. We therefore vacate EPA's decision and remand to allow the agency to reevaluate WRC's petition using the correct figures.



In 2005, Congress amended the Clean Air Act to encourage increased use of renewable fuels in the United States. As part of that statutory scheme, Congress prescribed target volumes of renewable fuels for use in each year through 2022. In 2015, for example, the statute calls for consumption of 20.5 billion gallons of renewable fuels. 42 U.S.C. § 7545(o)(2)(B)(i)(I). The statute vests EPA with authority to develop a renewable fuels program to secure satisfaction of the annual benchmarks. Id. § 7545(o)(2)(A)(i).

The statute calls for the Energy Information annually to estimate the total amount of transportation fuel expected to be sold in the United States in the upcoming year. Id. § 7545(o)(3)(A). EPA then divides the renewable-fuels benchmark set out in the statute by the overall fuel estimate provided by DOE, yielding a “volume percentage” requirement for the year. For example, if DOE projects the use of 100 billion gallons of fuel in a year for which the statute requires the use of 20 billion gallons of renewable fuels, the “volume percentage” for that year would be 20%. Obligated parties under the regulations—namely, refineries and importers of fuel—must demonstrate that they meet a pro-rata share of the overarching renewable fuels volume obligations based on that “volume percentage.” See 40 C.F.R. § 80.1406(a). Under a volume percentage of 20%, for example, an obligated party producing 100,000 gallons of fuel in a year would have a renewable fuels volume obligation of 20,000 gallons.

Obligated parties, however, are not necessarily required to produce and blend renewable fuels themselves. Instead, they demonstrate compliance through a system of Renewable Identification Numbers (RINs). Each gallon of renewable fuel produced for use in the United States generates its own RIN. Id. § 80.1426(a). RINs attach to the physical volume of the renewable fuel, but become “separated” from renewable fuel batches upon blending of the renewable fuel into conventional fuel. Id. §§ 80.1426(e), 80.1429(b). After blending, RINs may either be retained by the blending party or sold to other obligated parties. Id. §§ 80.1427(a)(6), 80.1451. As a result, parties who purchase an adequate number of RINs can comply with their renewable fuels obligations without producing or blending renewable fuels themselves. Each year, obligated parties must to submit to EPA a list of RINs in fulfillment of their renewable fuels obligations. A RIN is retired upon submission to EPA. See id. § 80.1427(a)(1).

The mechanics of the RIN system mean that obligated parties incapable of blending must rely disproportionately on the RIN market. Because small refineries generally have more limited blending capacity than larger refineries, they often need to acquire a large number of RINs from the market in order to meet their annual obligations. Congress, aware that small refineries would face greater difficulty complying with the renewable fuels requirements, created a three-tiered system of exemptions to afford small refineries a bridge to compliance.

First, the statute granted all small refineries (defined as refineries with crude oil throughput averaging 75,000 barrels or less per day) an exemption from the renewable fuels program through 2011. 42 U.S.C § 7545(o)(9)(A)(i), (o)(1)(K). That blanket exemption gave small refineries time to develop compliance strategies and increase blending capacity. Second, the statute directed DOE to conduct a study “to determine whether compliance ․ would impose a disproportionate economic hardship on small refineries.” Id. § 7545(o)(9)(A)(ii)(I). If DOE determined that any small refinery “would be subject to a disproportionate economic hardship if required to comply with” the renewable fuels program, EPA was required to extend the exemption for that refinery “for a period of not less than 2 additional years.” Id. § 7545(o)(9)(A)(ii)(II). Third, the statute enables a small refinery to initiate an inquiry into disproportionate economic hardship at any time by “petition [ing] the [EPA] Administrator for an extension of the exemption ․ for the reason of disproportionate economic hardship.” Id. § 7545(o)(9)(B)(i). When evaluating a petition for an extension, EPA must consult with DOE and consider the DOE study required by § 7545(o)(9)(A)(ii)(I), as well as “other economic factors.” Id. § 7545(o)(9)(B)(ii).

In 2011, DOE completed the 2011 Small Refinery Exemption Study (the 2011 Study) required by § 7545(o)(9)(A)(ii)(I). DOE concluded that a showing of disproportionate economic hardship “must encompass two broad components: a high cost of compliance relative to the industry average, and an effect sufficient to cause a significant impairment of the refinery operations.” J.A. 26. The 2011 Study also developed a scoring methodology to determine whether a small refinery satisfies those standards. That methodology assigns a score to twelve metrics, which are then used to produce two index scores: a disproportionate impacts index and a viability index. The disproportionate impacts index measures the structural impacts a small refinery would likely face in achieving compliance, while the viability index assesses how compliance would affect the refinery's ability to remain competitive and profitable. If a small refinery receives a score greater than 1 on both indices, it faces disproportionate economic hardship under DOE's standard.

Applying that methodology in 2011, DOE concluded that fourteen small refineries—including WRC—faced disproportionate economic hardship. DOE directed EPA to extend the exemption for two additional years (from 2010 to 2012) for those fourteen refineries pursuant to § 7545(o)(9)(A)(II).

Up through 2012, RINs sold at low prices reflecting the cost of corn ethanol (the most widely used renewable fuel) relative to that of conventional fuel. But beginning in 2013, the nature of the ethanol RIN market changed due to a so-called “ethanol blendwall” or “E10 blendwall.” Conventional engines can handle only a certain percentage (about 10%) of ethanol in fuel. In 2013, the statutory renewable fuels volume requirements exceeded the amount of ethanol that the transportation market could absorb. Because of the ethanol blendwall, RIN prices increased in 2013 and began to fluctuate widely.


WRC is a small refinery that processes about 14,000 barrels of crude oil per day. Its output places it 117th in size out of the 132 refineries in the United States. In 2013, WRC's blending capacity enabled it to satisfy about one-third of its RIN requirements through in-house blending. The company thus primarily relies on the RIN market to achieve compliance. Before 2011, WRC qualified for the blanket small refinery exemption and was not required to comply with the renewable fuels program. WRC then qualified for a two year extension of its exemption pursuant to the 2011 DOE Study, deferring its compliance obligations to 2013.

In August 2013, WRC filed an economic hardship petition under 42 U.S.C. § 7545(o)(9)(B), requesting that EPA extend WRC's hardship exemption for another two years (2013 and 2014). WRC emphasized the financial stress caused by the skyrocketing price of RINs. Pursuant to the statutory directive requiring EPA to “consult” with DOE in evaluating hardship petitions, id. § 7545(o)(9)(B)(ii), EPA submitted WRC's data to DOE and asked DOE to evaluate whether WRC should receive an extension. Applying the methodology established in the 2011 Study, DOE concluded that WRC scored higher than 1 on the disproportionate impacts index but less than 1 on the viability index. Because the viability index fell below the threshold of 1, DOE declined to recommend that EPA extend WRC's exemption.

On January 31, 2013, EPA issued a decision denying WRC's request for extension of its hardship exemption. After setting forth DOE's method for reviewing hardship petitions, EPA explained that it would “consider all information submitted by a petitioner” but that “DOE's evaluation of [WRC's] survey [ ] [would be] the primary factor in EPA's determination.” J.A. 322. “DOE has expertise in evaluating economic conditions at U.S. refineries,” EPA observed, “and DOE used its expertise to develop a survey form and assessment process to identify when disproportionate economic hardship exists in the context of the renewable fuel standard program.” Id. EPA therefore would “accord considerable deference to DOE's analysis of disproportionate economic hardship in deciding whether or not to grant a petition for extension.” Id. After summarizing the data submitted by WRC, EPA incorporated DOE's recommendation into its decision, observing that “DOE's evaluation indicates that the disproportionate impacts index ․ exceeds the hardship threshold, but the viability index ․ does not.” J.A. 329–31. EPA then performed a “qualitative[ ]” review to “ascertain if the information [submitted by WRC] is consistent with the finding of no disproportionate economic hardship.” J.A. 331. Concluding that WRC's financial data was “indeed consistent with that finding,” EPA denied the petition. Id. WRC now petitions this court for review of EPA's decision.


We first address WRC's challenge to EPA's interpretation of the statutory term “disproportionate economic hardship.” 42 U.S.C. § 7545(o)(9)(B). EPA construed that term in conformity with DOE's scoring indices, and it therefore required WRC to show that compliance both would impose disproportionate economic effects and would pose some threat to the viability of the refinery. WRC contends that the statute requires EPA “to grant exemptions when a small refinery faces disproportionate economic hardship—that is, a hardship that is out of proportion to that faced by larger refineries.” Pet'r's Br. 24. Consideration of a viability index, WRC argues, is inconsistent with that statutory mandate. We disagree.

We consider WRC's statutory argument under the two-step Chevron framework. Under the first step, if “Congress has directly spoken to the precise question at issue,” the agency “must give effect to the unambiguously expressed intent of Congress .” Chevron, U.S.A., Inc. v. Natural Res. Def. Council, Inc., 467 U.S. 837, 842–43, 104 S.Ct. 2778, 81 L.Ed.2d 694 (1984). WRC contends that EPA's reliance on a viability index is invalid at Chevron step one because it contradicts the plain language of § 7545(o)(9)(B).

The statute, however, contains no definition of the term “disproportionate economic hardship.” See 42 U.S.C. § 7545(o)(1). Congress instead gave EPA general guidance on the evaluation of economic hardship petitions under § 7545(o)(9)(B). In particular, Congress required EPA to consult with DOE and to “consider the findings of the [2011 Study] and other economic factors” when evaluating petitions. Id. § 7545(o)(9)(B)(ii). The statute gives no further instruction and identifies no particular economic factors or metrics to be considered. That sort of statutory silence about the particular factors that an agency must consider conveys “nothing more than a refusal to tie the agency's hands.” Monroe Energy, LLC v. EPA, 750 F.3d 909, 915 (D.C.Cir.2014). As long as EPA consults with DOE and considers the 2011 Study and “other economic factors,” EPA retains substantial discretion to decide how to evaluate hardship petitions. We therefore reject WRC's Chevron step-one challenge.

Alternatively, WRC contends that EPA's reliance on a viability index should be rejected at the second step of the Chevron framework. At Chevron step two, we must satisfy ourselves that EPA's method of evaluating “disproportionate economic hardship” is “based on a permissible construction of the statute.” Chevron, 467 U.S. at 843. We conclude that it is.

EPA's decision to incorporate the 2011 Study's methodology into its evaluation—including the viability index—is entirely reasonable. The statute, as noted, requires EPA to consult with DOE and “consider the findings of the [2011 Study] and other economic factors” in evaluating an economic hardship petition. 42 U.S.C. § 7545(o)(9)(B)(ii). EPA interpreted the term “disproportionate economic hardship” in conformity with the 2011 Study because “[t]he basis for any grant of an exemption extension by EPA in response to an individual petition is the same as the basis of evaluation in the [2011 Study]—disproportionate economic hardship.” J.A. 322. EPA's rationale accords with the well-established presumption that “a given term is used to mean the same thing throughout a statute.” Mohamad v. Palestinian Auth., –––U.S. ––––, ––––, 132 S.Ct. 1702, 1708, 182 L.Ed.2d 720 (2012).

Even considered on its own terms, EPA's interpretation of the phrase “disproportionate economic hardship” is wholly reasonable. DOE concluded, and EPA agreed, that the relative costs of compliance alone cannot demonstrate economic hardship because all refineries face a direct cost associated with participation in the program. Of course, some refineries will face higher costs than others, but whether those costs impose disproportionate hardship on a given refinery presents a different question. DOE determined that the best way to measure “hardship” entailed examining the impact of compliance costs on a refinery's ability to maintain profitability and competitiveness—i.e., viability—in the long term. EPA adopted DOE's understanding, and that choice lies well within the agency's discretion.


WRC next contends that we should vacate EPA's decision because DOE changed its scoring practice without adequate notice or explanation. We are unpersuaded.

Under DOE's methodology for evaluating economic hardship petitions, a refinery must score above 1 on both the viability index and the disproportionate impacts index in order to demonstrate “disproportionate economic hardship.” The viability index in turn reflects three component scores. Those scores measure: (i) whether compliance costs would eliminate efficiency gains to the refinery; (ii) whether individual special events would adversely affect the refinery; and (iii) whether compliance costs would likely lead to shutdown of the refinery. The three scores are added together and divided by 6 to produce a final viability index value. Thus, a score of 10 on any one component would secure a score exceeding 1 on the viability index.

In 2011, DOE awarded one of only two scores—0 (no impact) or 10 (impact)—on the first component metric (the one concerning efficiency gains). For that year, DOE gave WRC a score of 10 on that metric, guaranteeing a viability score of greater than 1 and ultimately leading to WRC's receipt of a hardship exemption. When evaluating WRC's 2013 petition, however, DOE assigned WRC a score of 5 (moderate impact) for the efficiency-gains metric and a 0 on the other two metrics, resulting in a viability index value of less than 1. In a footnote, EPA explained that DOE “already used intermediate scores of 5 in some ․ metrics, and believes it is also appropriate to use intermediate scores [for the efficiency-gains metric] to more accurately characterize the impacts of compliance costs.” J.A. 321. WRC raises both substantive and procedural challenges to EPA's reliance in 2013 on an intermediate score for the efficiency-gains metric.

As to substance, WRC contends that the addition of an intermediate score was arbitrary and capricious because there was no explanation for the change in scoring practice. We disagree. Judicial review of a “change in agency policy is no stricter than our review of an initial agency action.” White Stallion Energy Ctr. LLC v. EPA, 748 F.3d 1222, 1235 (D.C.Cir.2014) (citing FCC v. Fox Television Stations, Inc., 556 U.S. 502, 514–16, 129 S.Ct. 1800, 173 L.Ed.2d 738 (2009)). An agency must “provide reasoned explanation for its action,” which normally requires “that it display awareness that it is changing position.” Fox Television, 556 U.S. at 515 (emphasis omitted). Here, EPA acknowledged the change and explained that DOE added an “intermediate score[ ]” in order to “more accurately characterize the impacts of compliance costs ․ on a refinery.” J.A. 321 n. 4. WRC contends that EPA provided a description of the change rather than an explanation of it. But a change is not invalid merely because it is readily explained. The change at issue here fits in that category: as EPA explained, the addition of an intermediate score to the efficiency-gains metric allows for more nuanced and accurate characterization of the impact of compliance costs. That is a reasonable explanation for the change.

WRC's procedural challenge asserts that DOE's decision to use an intermediate score (and EPA's adoption of that decision) required notice and comment rulemaking. That argument turns on WRC's contention that the addition of an intermediate score for the efficiency-gains metric fundamentally altered the operation of the scoring matrix. According to WRC, under the preexisting system, any impact on efficiency, no matter how insubstantial, would produce a score of 10 on the efficiency-gains metric (which, in turn, would result in a viability index exceeding 1). Whereas any impact on efficiency once qualified WRC for a hardship exemption, WRC claims, it must now demonstrate significant hardship. Consequently, WRC asserts that the addition of the intermediate score did not merely round out the existing scoring method, but instead worked a substantive change of a kind requiring notice and comment.

The 2011 Study belies WRC's understanding of the old system. The efficiency-gains metric was never understood to require a score of 10 in the event of any impact on efficiency, regardless of its magnitude. According to the 2011 Study, the efficiency-gains metric assesses whether “the totality of factors ․ would reduce the profitability of the firm enough to impair future efficiency improvements.” J.A. 59 (emphasis added). If a refinery were to face “significant constraints on efficiency improvements,” it might be placed “at risk” even though reductions in profitability would not lead to “immediate shutdown.” Id. The 2011 Study thus indicates that the efficiency-gains metric aimed to protect against “significant” effects on efficiency, a position entirely consistent with DOE's decision to use an intermediate score to denote “moderate impact” in its 2013 evaluation. The 2011 Study also recognized that “[r]efineries that receive a[n] extension of their exemption” could take steps to “reduc[e] the impact” of future compliance costs. Id. As a result, “refineries that currently score high” on the efficiency-gains metric would “likely see a reduction in the scoring of this category in the future.” Id. DOE's award of an intermediate score when evaluating 2013 petitions therefore was fully consistent with the 2011 Study. It follows that DOE did not substantively change the efficiency-gains metric in the way WRC suggests.

Even assuming that DOE's addition of an intermediate score amounted to a substantive modification, WRC points us to no authority suggesting that the decision to make available a more refined score within an already-existing metric requires notice-and-comment procedures. We see no basis for creating such a rule here. Nothing in § 7545(o)(9)(B) compelled DOE to apply the exact same methodology—in every particular—that it had used in 2011. Instead, the statute merely called for EPA to consult with DOE and to “consider” the results of the 2011 Study when evaluating individual hardship petitions. 42 U.S.C. § 7545(o)(9)(B)(ii). EPA asked DOE to examine WRC's petition, and both EPA and DOE certainly “considered” the 2011 Study in doing so. Congress placed no limits on how DOE should provide its consultation to EPA under § 7545(o)(9)(B)(ii), and DOE's consultation did not purport to establish rights or obligations of WRC. As a result, we find no reason to conclude that DOE was obligated to engage in notice-and-comment procedures before adding a finer gradation within a preexisting scoring range.


We next consider WRC's contention that EPA erroneously considered (or failed to consider) various economic factors when reviewing WRC's petition. While § 7545(o)(9)(B) calls for EPA to consider “other economic factors” in evaluating hardship petitions, the statute contains no requirement to consider any particular factors. “In the absence of any express or implied statutory directive to consider particular factors,” EPA retains broad discretion to choose which “economic factors” it will (and will not) consider. Monroe Energy, 750 F.3d at 915. EPA acted within its discretion here.


WRC first asserts that EPA failed adequately to consider the high cost of purchasing RINs relative to producing them by blending. WRC's argument amounts to a substantive disagreement with the manner in which EPA chose to account for RIN costs in its review. In particular, WRC disagrees with the RIN price estimates EPA used in evaluating WRC's petition. We see no inadequacy in EPA's consideration of the cost of purchasing RINs. To be sure, EPA declined to rely on WRC's initial RIN estimates; but that was because WRC averaged only two months of RIN prices to produce an “average” substantially exceeding normal RIN prices. EPA instead reasonably relied on an updated estimate submitted by WRC in October, 2013. In reaching its final decision, moreover, EPA noted that “RIN prices have declined significantly” since WRC submitted its initial hardship petition and projected that RIN prices would continue to decrease. J.A. 328 n. 13. EPA therefore adequately considered the cost of purchasing RINs in its decision.


In EPA's independent evaluation of WRC's financial data, the agency observed that “WRC perceived that sufficient funds were available in 2012 for it to make a substantial discretionary dividend payment.” J.A. 332. The funds used to pay those discretionary dividends, EPA reasoned, “could have been used to pay for [compliance] projects.” J.A. 331. WRC argues that discretionary dividends paid in 2012 have “little relevance to whether [WRC] would face a disproportionate economic hardship from regulatory compliance in 2013.” Pet'r's Br. 43. And even if those dividend payments were relevant, WRC contends it should not have been faulted for failing to prepare for unforeseeable increases in RIN prices. We disagree.

Although it was exempt from the renewable fuels program in 2012, WRC remained an obligated party covered by the statute. See 40 C.F.R. § 80.1406(a). And EPA reasonably expected WRC to make preparations to comply with its 2013 obligations during the company's five-year exemption period. The discretionary dividend payment indicated that WRC elected to distribute profits to its owners rather than use profits to prepare for approaching compliance obligations. Allowing small refineries to perpetuate that manner of self-inflicted hardship would conflict with the terms of the statute, which contemplate a “[t]emporary exemption” for small refineries with an eye toward eventual compliance with the renewable fuels program for all refineries. 42 U.S.C. § 7545(o)(9)(A) (emphasis added). EPA reasonably considered the compliance efforts made (and not made) during WRC's five-year exemption in evaluating WRC's petition for a further extension of its exemption.


WRC next claims that EPA erred in assessing the refinery's cash flow by looking to WRC's net income rather than its adjusted Earnings Before Interest, Taxes, Depreciation, and Amortization (EBITDA). WRC further contends that “EPA should have adjusted EBITDA to account for the unavoidable cash outlays of capital expenditures, loan principal repayments, and interest payments.” Pet'r's Br. 5152. But EPA did consider EBITDA (in addition to net income) in evaluating WRC's finances. For instance, the agency noted that WRC “will again be profitable in 2013,” citing both net income and EBITDA. J.A. 332. And while WRC may believe that an adjusted EBITDA would provide a better measure of cash flow, it cannot succeed in its challenge unless it demonstrates that EPA's measure was unreasonable. As WRC itself acknowledged in correspondence with EPA, however, “EBITDA ․ is the standard basis for evaluating the economic health of a refining company.” J .A. 138; see Pet'r's Br. 51. EPA therefore acted reasonably when it chose to consider unadjusted EBITDA in evaluating WRC's petition.


While most refineries are corporations, WRC is an LLC and is therefore a pass-through entity for tax purposes. WRC contends that EPA should have accounted for income taxes paid by the unit holders of WRC's holding company. Failure to account for income taxes paid by unit holders, WRC argues, caused EPA to overstate WRC's net income relative to other refineries. But the pertinent statutory text requires EPA to consider whether the small refinery itself faces disproportionate economic hardship. See 42 U.S.C. § 7545(o)(9)(A)-(B). In that light, it was reasonable for EPA to confine its evaluation to the finances of the refinery without considering taxes paid by third parties.


WRC finally challenges EPA's decision based on two miscalculations in EPA's independent analysis of WRC's financial data. EPA concedes the two errors, but argues that we should nevertheless deny WRC's petition because the errors were harmless. This court will affirm an agency's decision despite errors when “it is clear that ․ the agency would have reached the same ultimate result” had the errors not been made. Salt River Project Agric. Improvement & Power Dist. v. United States, 762 F.2d 1053, 1061 n. 8 (D.C.Cir.1985). Although WRC bears the burden of establishing that the errors were prejudicial, that is not “a particularly onerous requirement.” Jicarilla Apache Nation v. U.S. Dep't of Interior, 613 F.3d 1112, 1121 (D.C.Cir.2010). “Often the circumstances of the case will make clear to the appellate judge” that an error was prejudicial, “and nothing further need be said.” Shinseki v. Sanders, 556 U.S. 396, 410, 129 S.Ct. 1696, 173 L.Ed.2d 532 (2009). Here, because the conceded errors significantly alter important figures in EPA's independent analysis of WRC's financial data, we cannot conclude with sufficient certainty that the agency would have made the same decision absent its errors.

While DOE's recommendation serves as the “primary” factor in EPA's determination, J.A. 322, EPA also conducted an independent, “qualitative” review of WRC's financial data, J.A. 331. That independent review aimed to determine whether WRC's financial data supported a finding of no disproportionate economic hardship. EPA concluded that the information submitted by WRC was “consistent” with such a finding. J.A. 331. Two key factors EPA considered in reaching that conclusion were (i) that WRC's projected 2013 net income was “significantly greater” than its projected RIN costs, and (ii) that WRC's net refining margins compared favorably to those of other refineries that petitioned for a hardship exemption. J.A. 331–32. EPA now concedes it made substantial mathematical errors in calculating both WRC's projected 2013 net income and its net refining margins.

With regard to the first error, EPA sought to exclude “hedge impacts” (here, gains and losses realized from investment positions in crude oil) from the company's net income. As WRC points out, however, EPA made a computational error when it added, rather than subtracted, WRC's hedge impacts in calculating net income excluding hedges. EPA concluded that compliance would not “significantly impact” WRC's viability because “WRC projected ․ that they will have a [2013] net income ․ significantly greater than [the] projected 2013 purchased RIN cost.” J.A. 331 (emphasis added). EPA now concedes that it accidentally overstated WRC's projected 2013 net income, and that the correct net income figure was less than half of the figure EPA relied on in its decision. EPA contends that we can nonetheless sustain its decision because, even using the correct figure, the projections indicated that WRC would be profitable in 2013. But given that EPA's error resulted in a substantial overstatement of net income, we are unable to conclude with adequate certainty that EPA still would have regarded WRC's net income as “significantly greater” than projected RIN costs.

EPA also urges us to deem its error harmless because the error did not infect DOE's recommendation to EPA. We cannot accept that argument. It is uncontested that EPA retained ultimate and independent authority to grant or deny economic hardship petitions under § 7545(o)(9)(B). According to EPA's decision, its independent analysis aimed to determine whether “the information submitted by WRC” was “consistent with the finding of no disproportionate economic impact.” J.A. 331. Had EPA concluded that WRC's financial information was inconsistent with that finding, it presumably would have granted WRC's petition notwithstanding DOE's recommendation. Although EPA considered several factors in the course of concluding that WRC's information was “indeed consistent” with DOE's recommendation, id., “[w]hat weight [EPA] gave to those [factors] is impossible to discern.” PDK Labs. Inc. v. DEA, 362 F.3d 786, 799 (D.C.Cir.2004). Accordingly, we cannot conclude that EPA “would have reached the same ultimate result” had it correctly calculated WRC's projected 2013 net income. Salt River Project, 762 F.2d at 1061 n. 8.

EPA also acknowledges a second error in its analysis. As EPA explains, it “should have accounted for realized hedge impacts in determining [WRC's] net refining margins and in comparing its average refining margin[s] to those of other small refineries.” Resp't Br. at 68. As a result of that second error, EPA significantly overstated WRC's average net refining margin per barrel for 2012 and 2013. In light of our conclusion that EPA's first error cannot be considered harmless, we have no occasion to apply harmless-error analysis to EPA's second error: the agency must redo its analysis in any event based on the first error. In doing so, EPA presumably would also correct the second error by incorporating correct net refining margins.

For the foregoing reasons, we vacate EPA's decision and remand for further consideration consistent with this opinion.

So ordered.

Opinion for the Court filed by Circuit Judge SRINIVASAN.