Resolute Forest Products, Inc. v. United States Department of Agriculture

187 F. Supp. 3d 100, 2016 U.S. Dist. LEXIS 64439, 2016 WL 2885869

• Court: District Court, District of Columbia
• Decided: May 17, 2016
• Docket: Civil Action No. 2014-2103
• Status: Published

Opinion

MEMORANDUM OPINION

JAMES E. BOASBERG, United States District Judge

For the past several years, Plaintiff Resolute Forest Products, Inc. and the U.S. Department of Agriculture have been locked in a straggle over the latter’s Softwood Lumber Checkoff Order. That Order requires any softwood-lumber domestic manufacturer or foreign importer who produces or imports more than 15 ‘million board feet (15mmbf) per year to pay a mandatory assessment on all softwood lumber shipped above that amount. Checkoff orders such as this are a kind of compulsory marketing program developed by private parties and overseen by the Department in accordance with the Commodity Promotion, Research and Information Act (the CPRIA), 7 U.S.C. §§ 7411-7425. Apparently unhappy that it must pay assessments under the Order, Resolute lodged a failed administrative protest before an ALJ and then subsequently brought suit here, raising four constitutional challenges to the Order and six- alleged violations of the Administrative Procedure Act.

In its September 9, 2015, Memorandum Opinion, this Court dismissed all but one of Plaintiffs APA challenges. See Resolute Forest Products, Inc, v. U.S. Dep’t of Agric., 130 F.Supp.3d 81 (D.D.C.2015). On the sole remaining APA claim (Count V), however, this Court remanded without va-catur to the Department of Agriculture for a reasoned and coherent treatment of its decision to select 15mmbf per year as the threshold amount. Defendants responded with a memorandum and exhibits providing additional explanation for the selection of that figure. See ECF No. 26. Although Defendants’ second explanation was better than its first, it nonetheless raised as many questions as it answered. Unable to reconcile certain discrepancies within the agency’s explanations and the data it presented, the Court remanded again, this time ordering the Department to point tó the underlying data sources relied upon in selecting 15mmbf and to explain the discrepancies the Court identified. See Resolute Forest Products, Inc, v. U.S. Dep’t of Agric., No. 14-2103, 2016 WL 1714312 (D.D.C. Feb. 2,2016). The agency responded again with further exhibits and an additional memorandum. See ECF No. 33.

After all of the back and forth, the same question remains: was the agency’s selection of 15mmbf arbitrary and capricious in violation of the APA? Despite two remand opportunities, Defendants have still not provided a reasonable explanation for selecting that quantity. Nearly every calculation upon which the agency relies has significant mismeasurements or inaccuracies, and many of the agency’s explanations across its original rulemaking process, its briefings, and its two responses to the Court’s remand orders contradict one another. While APA review does not demand perfection from an agency, the Court here must ineluctably conclude that USDA’s promulgation of the Checkoff Order was arbitrary and capricious.

I. Background

Because the Court has already addressed many of the substantive and procedural issues of this case in its earlier Opinion, see Resolute Forest Products, 130 F.Supp.3d 81, it will focus- on those still in contention here.

A. The Softwood Lumber Checkoff Order

The Softwood Lumber Checkoff Order that Plaintiff challenges here grew out of the softwood-lumber industry’s struggles during one of the “worst market[s] in history” after the great recession and the collapse of the housing market at the end of the last decade. See Softwood Lumber Research, Promotion, Consumer Education and Industry Information Order; Proposed Rules, 75 Fed. Reg. 61,002, 61,-005 (Oct. 1, 2010). To prop up the struggling industry, a trade association known as the Blue Ribbon Commission (BRC)— comprising 21 softwood-lumber chief-executive officers and business leaders—submitted its incipient proposal to USDA’s Agricultural Marketing Service. Id. AMS administers marketing orders under the CPRIA, the statute that governs the proposal, approval, and administration of checkoff orders for a variety of commodity products. See 7 U.S.C. § 7412-13. When a proposed order is submitted by “an association of producers” (here, the BRC), the statute instructs the Secretary tó “determine[ ] that a proposed order is consistent with and will effectuate the purpose” of the CPRIA. Id. § 7413(b)(l)-(2). If he so determines, he • then proceeds through the standard notice-and-comment rulemaking process for the proposed order. Id. § 7413(b)(2)-(4).

In addition to typical notice-and-comment rulemaking, however, the CPRIA mandates that the Secretary also obtain the approval of “persons subject to assessments” under the order via a referendum. Id. § 7413(b)(1). The Secretary may conduct said referendum either before finalizing a proposed checkoff order or else within three years of the first assessments taking place in accordance with it. Id. § 7417(b)(2). Crucial to this suit and the present dispute, the Secretary also has the authority to exempt from the order any “de minimis quantity” of the agricultural commodity subject to assessment. Id. § 7415(a)(1). And because eligibility to participate in the referendum depends on being “among persons to be subject to an assessment,” the de minimis quantity also affects who may vote in a given referendum. Id. § 7417(a)(1).

As to the Checkoff Order here, after the Secretary determined that the BRC’s proposal would effectuate the purpose of the CPRIA, AMS announced the proposed rule in the Federal Register, providing notice and seeking comment. See 75 Fed. Reg. at 61,012. The agency announced that the proposed Order would provide for initial assessments of $0.35 per thousand board feet shipped within or imported to the U.S., although it could eventually be increased up to $0.50. Id. The agency also stated that the proposed de minimis quantity exempted from assessment would be 15mmbf per producer or importer per year, with assessments only applying to amounts shipped or imported by a given producer above that threshold in any given year. Id, In determining this assessment price and exemption threshold, the agency also explored what portion of the softwood-lumber industry would pay assessments under the Order and considered several different prices and de minimis quantities. Id. at 61,012-13.

As support for its proposed de minimis quantity, the agency determined that a 15mmbf exemption and an assessment of $0.35 per thousand board feet would “generate sufficient income to support an effective promotion program for softwood lumber.” Id. at 61,013. The agency also noted that the BRC had explored, various de minimis exemption thresholds—including 15 million, 20 million, and 30 million board feet—and concluded that the 15mmbf exemption (“a quantity sufficient to build approximately 1,000 homes,” Resolute For est Products, 130 F.Supp.3d at 102 (internal citation and quotation marks omitted)) would yield “a deduction of 11.3 percent in assessment income” by reducing the total quantity of softwood lumber to be assessed by that percentage. See 75 Fed. Reg. at 61,013. In justifying this exemption quantity, the agency estimated that roughly 61% of domestic manufacturers and about 12% of foreign importers would be subject to the Order. Id.

After the agency issued the initial proposed rule, it followed up with a summary of comments received and provided responses to those comments. See Softwood Lumber Research, Promotion, Consumer Education and Industry Information Order, 76 Fed. Reg. 22,757, 22,770-75 (April 22, 2011). As the majority of comments supported the proposed Order, AMS next ■announced a referendum to approve it, in which all eligible producers and importers could participate. Id. at 22,775. Eligibility required manufacturing and shipping of 15mmbf or more between January 1 and December 31, 2010. Id. After the May 23-June 10, 2011, referendum was conducted, AMS announced that 67% of those voting, a group that collectively shipped 80% of the volume of softwood lumber represented in the referendum, had voted in favor of the Order. See Softwood Lumber Research, Promotion, Consumer Education and Industry Information Order, 76 Fed. Reg. 46,185, 46,188, 46,189 (Aug. 2, 2011). Based on this approval, AMS subsequently put the Checkoff Order into effect.

B. Resolute’s Challenge

Resolute has opposed the Checkoff Order from the beginning. As Plaintiff imported less than 15 million board feet during 2010, it was ineligible to vote in the referendum, see In Re: Resolute Forest Products Petitioner, No. 12-40, 2014 WL 1993757, at *5-6 (U.S.D.A. Apr. 30, 2014), but because it has since begun to import more than that amount, it has had to pay assessments on imports above-that threshold since January 2012. See PL MSJ Reply (ECF No. 21) at 7. Opposing the Checkoff Order, Plaintiff filed a petition with USDA on October 28, 2011, shortly after it went into effect. See Compl., ¶ 81. When Resolute did not prevail administratively, it filed suit before this Court in December 2014.

The grist of Plaintiffs challenge is that AMS violated the Administrative Procedure Act in both the rulemaking and referendum process, id., ¶¶ 149-200, and that the CPRIA unconstitutionally delegates executive and legislative authority to private parties and also’ violates the due-process rights of producers and importers. Id., ¶¶ 123-148. In its September 9, 2015, Opinion, this Court granted summary judgment for the agency on five of Resolute’s six APA challenges. See Resolute Forest Products, 130 F.Supp.3d at 92-100. Because it remanded without vacatur on the sixth APA. claim, the Court, following the doctrine of constitutional avoidance, deferred Resolute’s constitutional challenges for a later date. Id. at 105.

In its remaining APA challenge (now before the Court), Resolute alleged that the agency acted arbitrarily and capriciously in selecting the 15mmbf “de minim-is quantity” under the CPRIA. See PI. Opp./MSJ (ECF No. 15) at 25. Plaintiff especially took issue with the agency’s original legal argument that any exemption quantity that would “generate sufficient income to, support an effective promotion program” would be a- permissible de minimis quantity because it was “impossible for [AMS] to know the total volume” of softwood lumber produced and shipped. See Def. MSJ (ECF No. 13) at 24 (citation and internal quotation marks omitted). Resolute argued that AMS lacked discretion to designate any amount whatsoever as the de minimis quantity and asserted that the Service could not substantiate’its reasons for selecting 15mmbf as the de minimis quantity. See PI. Opp./MSJ at 26-27. In essence, it concluded, “AMS accepted the 15 million board foot exemption given to it by the BRC because that threshold was calculated by the BRC to hit the revenue targets that the BRC desired.” Id. at 27.

The Court shared Plaintiffs concern about the agency’s argument that it was “impossible” to know the amount of softwood lumber to be assessed, particularly where considerable record evidence suggested that total volumes of softwood lumber produced and shipped were readily available and, indeed, were relied upon .in determining the 15mmbf exemption. See Resolute Forest Products, 130 F.Supp.3d at 101 (“At least two documents in the Joint Appendix submitted by the parties suggest such figures were obtainable or had been obtained.”). The Court, accordingly, remanded without vacatur to the agency to supply additional explanation as to the data that supported a 15mmbf exemption threshold, as well as the underlying rationale in selecting such a threshold. Id. at 103-05. Defendants returned -several months later with a memorandum from Rex A. Barnes, AMS Associate Administrator, discussed in greater detail below. See First Remand Notice (ECF No. 26), Exh. A.

In the course of examining Barnes’s explanation and attached exhibits, the Court was still unable to understand how the sources of data the agency purported to rely upon yielded the estimates it had provided during rulémaking. Heeding the maxim of “if at first you don’t succeed, try, try, try again,” the Court remanded without vacatur a second time, ordering the agency to provide reassurance that, inter alia, “some verifiable source of data accurately depicted the softwood-lumber market and supported the selection of 15 million board feet as the appropriate de minimis quantity.” Resolute Forest Products, 2016 WL 1714312, at *3. The agency responded with a memorandum from Charles W. Parrott, Deputy Administrator of the Specialty Crops Program, as well as additional exhibits. See Notice (ECF No. 33), Exh. 1. This, too, proved unsatisfactory to Resolute. See PI. Second Remand Response (ECF No. 35). In any event, with this additional information in hand— the agency’s two remand memoranda and attached exhibits—the Court may finally rule on Resolute’s remaining APÁ challenge.

II. Legal Standard

In the typical case, summary judgment may be granted if “the movant shows that there is no genuine dispute as to any material fact and the movant is entitled to judgment as a matter of law.” Fed. R. Civ. P. 56(a); see also, Anderson v. Liberty Lobby, Inc., 477 U.S. 242, 247-48, 106 S.Ct. 2505, 91 L.Ed.2d 202 (1986); Holcomb v. Powell, 433 F.3d 889, 895 (D.C.Cir.2006). A fact is “material” if it is capable of affecting the substantive outcome of the litigation. See Liberty Lobby, 477 U.S. at 248, 106 S.Ct. 2505; Holcomb, 433 F.3d at 895. A dispute is “genuine” if the evidence is such that a reasonable jury could return a verdict for the nonmoving party. See Scott v. Harris, 550 U.S. 372, 380, 127 S.Ct. 1769, 167 L.Ed.2d 686 (2007); Liberty Lobby, 477 U.S. at 248, 106 S.Ct. 2505; Holcomb, 433 F.3d at 895.

Although styled Motions for Summary Judgment, the pleadings in this case more accurately seek the Court’s review of an administrative decision. Challenges under the CPRIA proceed under the Administrative Procedure Act’s familiar “arbitrary and capricious” standard of review. See 7 U.S.C. § 7418(b)(1); 5 U.S.C. § 706(2)(A). Because of the limited role federal courts play in reviewing such administrative decisions, the typical Rule 56 summary-judgment standard does not apply to the parties’ dueling motions on Resolute’s APA claims. See Sierra Club v. Mainella, 459 F.Supp.2d 76, 89-90 (D.D.C.2006). Instead, in APA cases, “the function of the district court is to determine whether or not ... the evidence in the administrative record permitted the agency to make the decision it did.” Id. (internal citations omitted). Summary judgment thus serves as the mechanism for deciding, as a matter of law, whether an agency action is supported by the administrative record and is otherwise consistent with the APA standard of review. See Bloch v. Powell, 227 F.Supp.2d 25, 31 (D.D.C.2002) (citing Richards v. INS, 554 F.2d 1173, 1177 (D.C.Cir.1977)).

The APA requires courts to “hold unlawful and set aside agency action, findings, and conclusions” that are “arbitrary, capricious, an abuse of discretion, or othérwise not in accordance with law.” 5 U.S.C. § 706(2)(A). Under this “narrow” standard of review—which appropriately encourages courts to defer to the agency’s expertise—an agency is required to “examine the relevant data and articulate a satisfactory explanation for its action including a rational connection between the facts found and the choice made.” Motor Vehicle Mfrs. Ass'n of United States, Inc, v. State Farm Mut. Auto. Ins. Co., 463 U.S. 29, 43, 103 S.Ct. 2856, 77 L.Ed.2d 443 (1983) (citation and internal quotation marks omitted). “In reviewing, agency action under that standard, a court is not to substitute its judgment for that of the agency,” GameFly, Inc. v. Postal Regulatory Comm’n, 704 F.3d 145, 148 (D.C.Cir.2013) (citation and internal quotation marks omitted), nor to “disturb the decision of an agency that has examine[d] the relevant data and articulate[d] ... a rational connection between the facts found and the choice made,” Americans for Safe Access v. DEA, 706 F.3d 438, 449 (D.C.Cir.2013) (internal quotation marks and citation omitted). On the other hand, where the agency has not provided a reasonable explanation for its actions, “[t]he reviewing court should not attempt itself to make up for such deficiencies: We may not supply a reasoned basis for the agency’s action that the agency itself has not given.” State Farm, 463 U.S. at 43, 103 S.Ct. 2856 (citation and internal quotation marks omitted). A court should nevertheless “uphold a decision of less than ideal clarity if the agency’s path may reasonably be discerned.” Id. (quoting Bowman Transp. Inc, v. Arkansas-Best Freight System, 419 U.S. 281, 286, 95 S.Ct. 438, 42 L.Ed.2d 447 (1974)).

More specific to Resolute’s remaining APA challenge here—a challenge to the Secretary’s interpretation of an ambiguous statutory term—“[w]hen a court reviews an agency’s construction of the statute which it administers, it is confronted with two questions.” Chevron, U.S.A., Inc, v. Natural Res. Def. Council, Inc., 467 U.S. 837, 842, 104 S.Ct. 2778, 81 L.Ed.2d 694 (1984). “First, applying the ordinary tools of statutory construction, the court must determine ‘whether Congress has directly spoken to the precise question at issue. If the intent of Congress is clear[,] ... the court, as well as the agency, must give effect to the unambiguously expressed intent of Congress.’ ” City of Arlington, Tex. v. FCC, — U.S. —, 133 S.Ct. 1863, 1868, 185 L.Ed.2d 941 (2013) (quoting Chevron, 467 U.S. at 842-43, 104 S.Ct. 2778). However, “if the statute is silent or ambiguous with respect to the specific issue, the question for the court is whether the agency’s answer is based on a permissible construction of the statute.” Chevron, 467 U.S. at 843, 104 S.Ct. 2778. This latter analysis is colloquially known as “Chevron step two.” Vill. of Barrington, Ill. v. Surface Transp. Bd., 636 F.3d 650, 660 (D,C.Cir.2011) (“At Chevron step two we defer to the agency’s permissible interpretation, but only if the agency has offered a reasoned explanation for why it chose that interpretation.”).

III. Analysis

The Court now turns to the heart of Resolute’s remaining APA challenge: that the agency’s selection of 15mmbf as the de minimis quantity exempted was arbitrary and capricious. See PL Opp./MSJ at 25-26. The first step in considering a challenge such as this is to assess the agency’s interpretation of the statute itself. Because the Court has already found the statutory term “de minimis quantity” ambiguous, see Resolute Forest Products, 130 F.Supp.3d at 102-103, it resumes its analysis at Chevron step two: given the ambiguity in the statute, has the agency offered a permissible construction of “de minimis quantity”?

This question, in turn, implicates two separate issues. The Court must first assess whether the agency considered appropriate criteria in determining a viable de minimis quantity to be exempted. Satisfied that it did so, the Court next considers the agency’s explanation and evidence supporting its selection of 15mmbf as de minimis in light of the agency’s identified criteria.

A. Permissible Interpretation of “De Minimis Quantity”

The Court begins by considering ■ the agency’s interpretation of “de minimis quantity” under the CPRIA. As a reminder, Defendants’ initial summary-judgment pleadings maintained that because it was “impossible” to know the total quantity of softwood lumber produced—despite evidence to the contrary in the agency’s own rulemaking notices—the Secretary’s selection of “an/’ de minimis quantity was permissible under the CPRIA. Compare Def. Reply at 23 (“‘[fit’s impossible for us to know the total volume’ of softwood lumber”), with 75 Fed. Reg. at 61,003 (“According to USDA’s Forest Service, for 2007-2008, total output (production) of softwood lumber by U.S. sawmills averaged about 29.5 billion board feet annually.”), and id. at 61,004 (“According to U.S, Department of Commerce, Census Bureau, Foreign Trade Statistics data, imports of softwood lumber from 2007 through 2009 averaged about 13 billion board feet annually.”) (citation omitted). Given the implausibility of the agency’s interpretation—in light of the plain meaning of “de minimis” and the appearance of evidence in its rule-making notices suggesting it was possible to obtain total quantity estimates—the Court remanded “for a reasoned and coherent treatment of the decision to select a 15 million-board-feet-per-year exemption as the ‘de minimis quantity’ exemption in accordance with” the CPRIA. See Resolute Forest Products, 130 F.Supp.3d at 105.

In response to this Order, Defendant provided a memorandum from Rex A. Barnes, Associate Administrator, AMS. Recognizing the problematic nature of its initial litigation position at summary judgment, the agency’s memorandum provides a more thorough account of the general criteria it asserts are appropriate in selecting a “de minimis quantity” in accordance with 7 U.S.C. § 7415(a)(1)., The agency has not had a prior occasion to articulate how it determines a “de minimis quantity” to be exempted from a proposed checkoff order, nor has a court previously endorsed a particular interpretive approach, so this is a question of first impression.

As the agency noted in its rulemaking notice, “[T]he 1996 Act does not define the term de minimis and USDA is not limited to using the definition of de minimis as specified in another law or, agreement. The de minimis quantity is defined for a particular program and industry.” 76 Fed. Reg. at 22,772. Because the CPRIA “provides no set methodology or formula for computing a de minimis quantity,” the Barnes Memorandum explains that USDA considered several factors in selecting a threshold, including (1) an estimate of the total quantity of the particular agricultural commodity (both quantity assessed and quantity exempted); (2) free-rider implications of a particular quantity; (3) the impact of such a limit on small businesses; and (4) the available funding to. support a viable program operating at that exemption threshold. See Barnes Mem. at 3.

From the vantage point of Chevron step-two analysis, the question is whether the agency’s proposed construction of the ambiguous term—“de minimis quantity”— is a permissible interpretation. These general factors were not articulated in quite this fashion in the agency’s notice of the proposed rulemaking, its response to comments, and in the final regulation implementing it. Given that Chevron deference is owed to “the administrative official and not to appellate counsel,” Bowen v. Georgetown Univ. Hosp., 488 U.S. 204, 212, 109 S.Ct. 468, 102 L.Ed.2d 493 (1988) (citation and internal quotation marks omitted), “we give no deference to agency ‘litigating positions’ raised for the first time on judicial review.” Vill. of Barrington, Ill., 636 F.3d at 660. In this case, however, it was legal counsel’s position— that it was impossible to know the total quantity of softwood lumber—that the Court found not credible, and the explanation of considerations regarding the selection of a de minimis quantity come from a member of the agency (Rex A. Barnes of AMS), not from legal counsel.

Consideration of the agency’s arguments on the first remand regarding its approach to interpreting the ambiguous term is also perfectly acceptable insofar as courts “frequently remand matters to agencies while leaving open the possibility that the agencies can reach exactly the same result as long as they ... explain themselves better or develop better evidence for their position.” Nat’l Treasury Employees Union v. Fed. Labor Relations Auth., 30 F.3d 1510, 1514 (D.C.Cir.1994). The agency’s more robust explanation is entirely the product of this Court’s first remand order for a fuller account of the 15mmbf-exemption selection criteria, and so the Court may consider these factors in assessing whether the agency’s choice of the de minimis quantity was supported by substantial evidence. After all, “the usual rule is that, with or without vacatur, an agency that cures a problem identified by a court is free to reinstate the original result on remand.” Heartland Reg’l Med. Ctr. v. Leavitt, 415 F.3d 24, 29-30 (D.C.Cir.2005); see also FEC v. Akins, 524 U.S. 11, 25, 118 S.Ct. 1777, 141 L.Ed.2d 10 (1998) (noting that, after remand, agency “might later, in the exercise of its lawful discretion, reach the same result for a different reason” than one rejected by reviewing court) (citing SEC v. Chenery Corp., 318 U.S. 80, 63 S.Ct. 454, 87 L.Ed. 626 (1943)). “Therefore, the proper focus for this Court’s inquiry is whether the [challenged agency action] upon remand is sustainable for the reasons stated in [the agency’s] supplemental determination and in light of the administrative record as a whole.” Bean Dredging, LLC v. United States, 773 F.Supp.2d 63, 79 (D.D.C.2011).

It is also worth noting that many— though not all—of the considerations identified on first remand were already more or less identified in the agency’s notices. See, e.g., 75 Fed Reg. at 61,013 (considering “the economic impact of the proposed Order on affected entities”); 76 Fed. Reg. at 22,772 (“15 million board feet would be appropriate because such a level would still provide the Board with resources to have a program that could be successful.”); id. (“[T]his level would exempt small operations that would otherwise be burdened by the assessment”). Given that the “de minimis quantity is defined for a particular program and industry,” id., the Court concludes that this case-by-case, context-specific approach, drawing on the selection criteria identified, is “a permissible construction of the statute.” Chevron, 467 U.S, at 843, 104 S.Ct. 2778. The agency’s general approach to selecting a de minimis quantity, then, was perfectly permissible.

B. Sufficiency of Evidence

Although the agency’s approach to determining a “de minimis quantity” was a plausible interpretation of the statute, Resolute’s APA challenge also asserts that the agency’s decision to choose 15mmbf was not supported by evidence in the administrative record. In other words, even if USDA’s construction of an ambiguous statutory term is permissible, “the agency must [also] examine the relevant data and articulate a satisfactory explanation for its action including a ‘rational connection between the facts found and the choice made.’ ” State Farm, 463 U.S. at 43, 103 S.Ct. 2856 (quoting Burlington Truck Lines v. United States, 371 U.S. 156, 168, 83 S.Ct. 239, 9 L.Ed.2d 207 (1962)).

Given the circuitous path this case has traveled—through the original cross-motions for summary judgment and the two remand orders—the Court begins its discussion by identifying precisely what may be considered record evidence relied upon by the agency during the promulgation of the Checkoff Order. It then turns to assessing USDA’s explanations in its first-remand response memorandum to determine whether the evidence before the agency—coupled with the criteria it states were considered—provides the minimal support necessary to justify the selection of 15mmbf. This memorandum, while clarifying USDA’s reasons for selecting 15mmbf, left the Court with concerns regarding its methodological approach and numerical estimates. It accordingly remanded again, this time ordering the agency to provide specific primary sources and clarification as to the estimates USDA purported to have relied upon. The Court concludes by assessing the agency’s second remand memorandum in response to the latter order-.

1. Administrative Record Evidence

As the agency included new attachments and exhibits as part of its responses to the Court’s two remand orders, the Court must’ first discuss their admissibility and what documents it will consider in determining whether the agency provided a “rational connection-between the facts found and the choice made.” Americans for Safe Access, 706 F.3d at 449 (internal quotation marks and citation omitted).

In contrast to most federal-agency rule-making, the CPRIA leaves open the possibility for private-industry groups to come to the agency and propose potential marketing orders. See 7 U.S.C. § 7413(b)(l)(B)(i) (A checkoff order “may be ... submitted to the Secretary by ,. an association of producers of the agricultural commodity.”). As a result, in this instance it was the Blue Ribbon Commission that came to USDA with the proposal for a checkoff order. The Secretary’s obligation was then to “determine[] that a proposed order is consistent with and will effectuate, the purpose of’ the CPRIA. Id. § 7413(b)(2). So satisfied, the Secretary then “publishes] the proposed order in the Federal Register and give[s] due notice and opportunity for public comment .... ” Id. This the Secretary did, publishing a notice of the proposed rulemaking and seeking comments from concerned parties regarding the Checkoff Order. See 75 Fed. Reg. at 61,002. Sixth months later, the Secretary responded to those comments and announced the final Checkoff Order and referendum to ratify it. See 76 Fed. Reg. at 22,757-22,775. After ratification of the proposed Order by eligible voters, AMS published a notice announcing its implementation. See 76 Fed. Reg. at 46,-185.

These notices appear to have relied heavily on the submissions of the BRC— which proposed the Checkoff Order—in particular its report, “BRC Proposal for a National Research and Promotion Program For Softwood Lumber,” see Letter from Jack Jordan, BRC Chairman, to Robert C. Keeney, Deputy Administrator, USDA, AMS (Feb. 16, 2010), Attach. B (“Overview, Justification, and Objectives for a National Research and Promotion Program For Softwood Lumber”) (“BRC Proposal”) (AR1353-AR1364), as well as the BRC’s “20 Myths and Facts About the Softwood Lumber Check-off’ (“20 Myths”) (AR0061-AR0065), a pamphlet circulated to softwood-lumber industry participants. While neither of these documents was cited in the agency’s notices, they were included in the Joint Appendix and form the core of the agency’s administrative record.

In addition to these documents, both the BRC and the agency heavily relied on a 2009 U.S. Forest Service research report. See Henry Spelter, David McKeever & Daniel Toth, Profile 2009: Softwood Sawmills in the United States and Canada, FPL-RP-659 (Oct. 2009) (ECF No. 33, Exh. A) (“Profile 2009”)). This Profile 2009 report was cited both in the BRC’s own report, see BRC Proposal at 4 (AR1353), and in the agency’s notices in the Federal Register. See, e.g., 75 Fed. Reg. at 61,003 nn. 1, 3 & 6; id. at 61,004 nn. 7-8 & 10; id. at 61,012 nn. 14 & 16; id. at 61,013 n. 17. Because this document contains statistics on .the number of sawmills and total softwood-lumber production capacity for all U.S. and Canadian softwood-lumber companies, it was central to both the BRC’s proposal and the agency’s decisionmaking process, and is thus front and center in the dispute between the parties here. The Court therefore will consider this document as part of the record.

On top of these documents, as part of its response to the Court’s second remand order, Defendants provided additional exhibits to explain the calculations upon which the agency relied during rulemak-ing, Resolute contends that these documents may not be considered part of the administrative record, for “USDA never requested and was never granted leave to expand or supplement the record, and USDA never provided for the record data to substantiate its conclusions.” PI. Second Remand Resp. (ECF No. 35) at 4. While it is true that these exact materials were not submitted as part of the Joint Appendix, the Court disagrees that it may take no consideration of them whatsoever. Most of the additional exhibits provided by USDA in both remands help to explain the conclusions drawn from the documents that were extensively cited in the agency’s Federal Register notices, and where “the raw data itself is at issue and was directly considered, analyzed, or manipulated by the agency in the course of reaching its decision, that raw or underlying data is ‘properly considered part of the administrative record.’ ” Univ. of Colorado Health at Mem’l Hosp. v. Burwell, 151 F.Supp.3d 1, 23, No. 14-1220, 2015 WL 6911261, at *14 (D.D.C. Nov. 9, 2015) (quoting Ctr. for Biological Diversity v. U.S. Bureau of Land Mgmt., 2007 WL 3049869, at *4 (N.D.Cal. Oct. 18, 2007)). After all, any materials an agency considered “either directly or indirectly” must be considered part of the administrative record. See Marcum v. Salazar, 751 F.Supp.2d 74, 78 (D.D.C.2010). As the Court’s two remand orders specifically pointed to the BRC’s Proposal and the Forest Service’s Profile 2009 and ordered. the agency to explain how it used the data contained therein in developing its estimates reported in the Federal Register, the Court will consider exhibits attached to the remand memoran-da to the degree they, shed light on :the agency’s underlying rationale. To do otherwise would undermine the very purpose of the Court’s two remand orders.

Finally, the agency provided several new documents as exhibits to. its two remand memoranda that were not previously part of the administrative record. While the agency can provide additional attachments to explain how it came to the' decision it did, the Court nevertheless must still rely only on evidence contained in the extant administrative record that supports the agency’s rationale and selection at the time it made the. decision. See Prairie State Generating Co. LLC v. Sec’y of Labor, 792 F.3d 82, 98-94 (D.C.Cir.2015) (“[T]he ‘focal point’ in arbitrary-and-caprieious review is ‘the administrative record already in existence’ ”) (quoting Camp v. Pitts, 411 U.S. 138, 142, 93 S.Ct. 1241, 36 L.Ed.2d 106 (1973)); see also Ass’n of Private Sector Colleges & Universities v. Duncan, 681 F.3d 427, 441 (D.C.Cir.2012) (“In evaluating an agency’s decisionmaking, our review is fundamentally deferential ... [b]ut we are limited to assessing the record that was actually before the agency.”) (citation and internal quotation marks omitted). The Court will, however, consider these documents to the degree that they shed light on how the agency considered evidence elsewhere contained in the extant administrative record.

2. First Remand Explanation

Having addressed questions concerning evidence in the record, the Court now pivots to an assessment of the agency’s first remand explanation. As a reminder, the Barnes Memorandum explains that USDA considered several factors in selecting its de minimis threshold, including (1) free-rider implications of a particular quantity; (2) an estimate of the total quantity of the particular agricultural commodity (both quantity assessed and quantity exempted); (3) the impact of such a limit on small businesses; and (4) the available funding to support a viable program operating at that exemption threshold. See Barnes Mem. at 2. The Court will discuss each of these considerations and the evidence Defendants cite to support selecting 15mmbf as the de minimis quantity, as well as Resolute’s objections.

a. Free Riders

The first—and perhaps most straightforward—claim is that the agency took free riders into consideration when selecting the de minimis quantity. Barnes explains that “[i]n approving the proposed exemption threshold of 15[mmbf], USDA took into consideration the potential impact of free riders on an effective checkoff program for softwood lumber.” Id. at 5. Rather than pointing to manifest evidence of this in the Federal Register, however, the agency cites only to the BRC’s statements in its proposal that “ ‘free riders within the industry have taken advantage of the voluntary nature of the programs, frustrating enthusiasm and support for fundraising among the paying players.’ ” Id. (quoting BRC Proposal at 10 (AR1359) (emphasis omitted)). The agency’s Barnes Memorandum further emphasizes the free-rider concerns raised in the BRC’s “leaflet advocating approval of the checkoff order,” which states that the exemption’s “impact would be ‘de minimis as far as free riders [are] concerned.’ ” Id. at 5 (quoting 20 Myths at 3 (AR0929)). As Resolute rightly points out in response, “USDA does not cite any Federal Register notice to show that USDA considered free riders and agreed with the BRC about the impact of ■the exemption” as to that consideration. See PL First Remand Reply (ECF No. 28) at 12 (emphasis added). This is only the first of several problems with Defendants’ explanation on remand.

b. Estimates of Total Quantity Assessed and Exempted

Defendants’ second factor was the total quantity of softwood lumber that would be assessed as well as the portion exempted from assessment as de minimis. As USDA largely points to the BRC’s estimates to substantiate this, id. at 3-5, the Court begins there. In proposing the Checkoff Order, the Blue Ribbon Commission settled on a 15mmbf “de minimis exemption for all producers and importers.” BRC Proposal at 10 (AR1359). To justify this selection, the BRC provided estimates of the percentage. of total softwood-lumber production capacity that would be excluded from assessment at this exemption level. Estimating that a total of 664 companies in the United States and Canada had an approximate total production capacity of 74.9 billion board feet of softwood lumber in 2007, the BRC then estimated the share of production capacity it expected would be exempted based on several different de minimis quantities. Id. at 11 (AR1360). It concluded that exempting producers whose production capacity was “[ujnder 16mmbf ’ per year would result in 257 companies representing 2.5% of total capacity being fully exempted. Id. Despite the fact that both the BRC and the agency rely on this estimate as a chief justification for the de minimis quantity, the estimate itself is inexplicably listed as “[ujnder 16mmbf” per year, not under 15mmbf. The Court is uncertain whether this is a transcription error, as everywhere else the agency treats these estimates as if they measure an exemption of 15mmbf, not 16mmbf.

This discrepancy aside, the BRC also reported that exempting the first 15mmbf in production capacity for all companies (including those with greater than 15mmbf annual production capacity) would expand the amount not assessed from 2.5% to 11.3% of total softwood-lumber production capacity. Id. The proposal went on to state that “[tjhe BRC believes that this proposal will meet both criteria, on the one hand be acceptable to the industry, and on the other mount a program of sufficient size and scope to achieve meaningful results in the marketplace.” Id. at 10 (AR1359).

Turning now to USDA’s decisionmaking process, the Barnes Memorandum states that USDA “[c]oncurr[ed] with the BRC that companies that produced under 15[mmbf] annually equated to about 2.5% of the industry’s total assessable volume.” Barnes Mem. at 4. On this basis, “USDA concluded that the adoption of the proposed exemption threshold of 15[mmbf] was appropriate because 2.5% of the total assessable volume of softwood lumber is a ‘de minimis quantity’ of that commodity and because the use of that threshold would not result in a substantial amount of uncollected assessments.” I&

As Resolute points out, see PI. First Remand Reply (ECF No. 28) at 3 & n. 2, this is a blatant contradiction of the evidence provided in the administrative record at the time the agency announced the Checkoff Order. As the agency stated in the Federal Register when it issued the notice of proposed rulemaking,

Regarding exemption levels, the BRC explored projected assessment income at exemption levels of 15, 20, and 30 million board feet. With a 15 million board foot exemption, the BRC projected a deduction of 11.3 percent in assessment income. Table 4 below shows the BRC’s projected income levels at various assessment options in light of the proposed 15 million board foot exemption.

75 Fed Reg. at 61,013 (emphasis added). Resolute is correct that the agency never once cited the 2.5% exemption estimate in its notices in the Federal Register. It is difficult to credit the agency on first remand when it states that it concluded that “2.5% of the total assessable volume of softwood lumber is a ‘de minimis quantity1 of that commodity_” Barnes Mem. at 4. If so, why did the agency report that 11.3% of quantity was exempted rather than 2.5%?

Deepening the Court’s frustration is the fact that the Barnes Memorandum does not clarify how the agency (or the BRC) arrived at either the 2.5% or the 11.3% estimate. Both statistics are cited without any explanation as to their origin or source. And while both appeared to derive from the BRC’s proposal, that proposal was not cited by the agency in its notice of proposed rulemaking, and the BRC proposal itself does not identify its source of these estimates. See BRC Proposal at 11 (AR1360). Even after the first remand order the Court was thus still unable to understand precisely what percentage of softwood lumber the agency thought would be exempted from assessment when it promulgated the Checkoff Order. As discussed below, this same methodological problem plagues the .agency’s estimate of companies exempted from the Checkoff Order, which in turn necessitated a further remand order from the Court. .

c. Impact on Small Companies

Another factor the agency states it “considered in approving the proposed exemption threshold ... was the impact that-the exemption would have on small companies.” Barnes Mem. at 5. In part,. this. is because the agency was “required to examine the impact of the proposed rule on small entities” under the Regulatory Flexibility Act, 5 U.S.C. §§ 601-12. See 75 Fed. Reg. at 61,012. As defined by regulations promulgated under the RFA, small softwood-lumber entities are those that “hav[e] annual receipts of no more than $7 million,” which the agency roughly translated as meaning manufacturers “who ship[] less than 25[mmbf] per year_” 75 Fed. Reg. at 61,012. Drawing on data from the American Lumber Standard Committee (ALSO), the agency estimated that “363 domestic manufacturers, or about 61 percent [of 595],” were small entities that shipped less than 25mmbf per year. Id. at 61,012 & n. 15; see also Parrott Mem. at 8-9 (“Data obtained from the [ALSC] provided the ostensible basis for these sentences”). As for the foreign-importer data, the agency stated that it relied on “Customs data” suggesting that “there were about 883 importers of softwood lumber annually. About 798 importers, or about 90 percent, imported less than” 25mmbf per year and were thus small entities as defined by the RFA. See 75 Fed. Reg. at 61,012.

While a helpful starting point, this explanation did not actually address the impact of the 15mmbf exemption on these small entities. Although the agency claimed that “USDA has performed this initial RFA analysis regarding the impact of the proposed rule on small entities,” 75 Fed. Reg. at 61,014, it nowhere stated what the impact of the 15mmbf exemption would be on companies shipping less than 25mmbf, such as the number of companies that ship between 15mmbf and 25mmbf per year (and would therefore pay assessments under the Checkoff Order), and how these companies might be affected by the assessments.

Instead, the agency provided estimates of the impact of the 15mmbf exemption on companies shipping less than 15mmbf per year, which it believed numbered 232 out 595 domestic manufacturers. Id Combining these 232 domestic manufacturers with the estimated 780 out of 883 foreign companies that imported less than 15mmbf per year, the agency determined that a total of 1,012 producers out of 1,478 would be exempt from assessments under the proposed Order. Id. at 61,015. The agency, however, never justified why 15mmbf was a reasonable proxy for a small company, as opposed to the 25mmbf definition of small entity as defined by the RFA, In response to comments, the agency merely stated that it “concur[red] with this exemption level because this level would exempt small operations that would otherwise be burdened by the assessment,” 76 Fed. Reg. at 22,772, never distinguishing between 15mmbf and 25mmbf. As a result, under the agency’s own (and only) definition of small entity—the 25mmbf measure used for its RFA analysis—many such small entities would, presumably, be “burdened by the assessment.” Yet USDA provided no discussion as to how many such companies would be affected or the extent of the burden.

Even more troubling, prior to the second remand order, the Court also had reason to doubt the integrity of USDA’s estimate that 232 out of 696 domestic manufacturers ship less than 15mmbf per year because the denominator for this estimate appeared spurious. As Resolute argues, there is a wide disparity between the estimates provided by the agency in the Federal Register and those offered by the BRC, which USDA purported to rely on. See First Remand Reply (ECF No. 28) at 8. The BRC’s proposal, which did not offer separate estimates for domestic and foreign entities, suggested that with a 15mmbf exemption, approximately 257 combined domestic and foreign entities would be exempted out of a total of 664. See BRC Proposal at 11 (AR1360), These numbers are not even close to the USDA combined estimates of 1,012 out of 1,478. The contrasting figures are puzzling because it appears that both USDA and the BRC derived their estimates from the same source of data—the Forest Service Profile 2009 report. Id. n. 14 (citing Profile 2009); BRC Proposal at 4 (AR1353) (stating that “data and much of the information in this application has been compiled from” Profile 2009).

Even more problematic, the Profile 2009 report does not measure the number of softwood-lumber entities; it only provides estimates for the number of North American sawmills. Given the confusion over just what USDA was measuring, the Court- examined the Profile 2009 report itself, as it explained in its second remand order. See Resolute Forest Products, 2016 . WL 1714312, at *2-3. USDA had claimed that its estimate of 595 domestic manufacturers was-a “number [that] represents separate business entities; one business entity may include multiple sawmills.” 75 Fed. Reg. at 61,012. Yet the agency cited Profile 2009 as the source of this information, see id. at n. 14, and that document makes clear' that the ■ estimates measure “past and current capacity of sawmills”—not entities. See Profile 2009 at 15. To confirm this, the Court itself averaged the number of U.S. sawmills in 20 tables listed in the appendix of the Profile 2009 report for the years 2007, 2008, and 2009, and arrived at the same number that USDA cited in the Federal Register, 595. The problem, of course, is that Profile 2009 reported 595 as the number of sawmills, whereas the agency reported 595 was the number of entities. Compare Resolute Forest Products, 2016 WL 1714312, at *4 tbl. 1, with 75 Fed Reg. at 61,012 and 76 Fed. Reg. at 22,767.

Prior to the second remand order, then, the agency had provided neither a coherent analysis of the impact of the 15mmbf exemption on “small entities” nor a reliable source of data for its estimates concerning the number of softwood-lumber entities exempted from assessment. The Court will return to this issue after summarizing its second remand order below,

d. Sufficient Revenue

The last factor the agency points to in the Barnes Memorandum is “whether a checkoff order that contained [the 15mmbf exemption] threshold would generate enough income to support a viable and effective research and promotion program for softwood lumber.” Barnes Mem. at 6. Drawing again on estimates provided by the BRC, see BRC Proposal at 10-11 (AR1360-61), the agency “found that ‘the [proposed exemption] and the initial $0.35 per thousand board foot assessment rate’ would generate ‘between $12.4 and almost 19 million [per year] ... with shipment levels ranging from 40 to 60 billion board feet.’ Barnes Mem. at 6 (quoting 76 Fed Reg. at 22,773). The Barnes Memorandum goes on to state that “[a]greeing with the BRC that ‘$20 million is an ideal threshold for an effective program ,USDA approved the proposed exemption.” Id. at 6-7 (quoting 76 Fed. Reg. at 22,767).

Resolute objects to this explanation, arguing that USDA improperly relied not on data for shipments in 2010 but instead on production capacity as of 2007. See First Remand Reply (ECF No. 28) at 5. Given the substantial differences between these measurements and the years in question, this could drastically alter the amount of revenue expected to be generated under the Checkoff Order. Plaintiff contends that “USDA was supposed to rely on shipment data from 2010,” which was the “representative period” under 7 U.S.C. § 7417(a)(1). Id. at 4. It also alleges that the “BRC stated, without justification or explanation, that production capacity was being used in this analysis as a proxy for shipments.” Id. at 5 (citation and internal quotation marks omitted). Résolute’s objection is effectively two challenges: one to the year of measurement (2007 vs. 2010), the other to the type of quantity measured (capacity vs. shipments). The Court tackles each of these grievances in turn.

i. 2007 vs. 2010

The statutory provision concerning the “representative period” provides no instruction as to how to measure that period, stating only that an optional referendum must include as participants those “persons subject to an assessment” who “engaged in” the “production” or “importation” of the commodity “during a representative period determined by the Secretary.” 7 U.S.C. § 7417(a)(l)(A)-(B). For the purposes of determining participants in the referendum, that period was calendar year 2010, the most recent year for which data was available. In announcing the referendum on April 22, 2011, USDA stated that eligible participants would include all those who “have domestically manufactured and/or imported 15 million board feet of more of softwood lumber during the representative périod from January 1 through December 31, 2010.” 76 Fed. Reg. at 22,757. Such a determination appears to be eminently reasoned and appropriate.

What is left of Resolute’s challenge is the lag between the year of data relied upon for the initial proposal and the year used for referendum-eligibility purposes. The Court thus now considers the reasonableness of the delay between the year relied on for developing the estimates (2007) and the referendum “representative period” (2010). Because. the agency was required to undergo notice-and-comment rulemaking before implementing the referendum, some delay between the time of the BRC’s proposal to USDA and the final implementation of the referendum was all but inevitable. Cf. N. Mariana Islands v. United States, 686 F.Supp.2d 7, 15-16 (D.D.C.2009) (finding 18 months reasonable period for agency to undergo notice- and-comment rulemaking). After all, notice-and-comment rulemaking was not the first step in the process here; the BRC had to first gather research on the utility and feasibility of the proposed Checkoff Order, engage. USDA in getting the Secretary’s approval, and assist in the formulation of the proposed rule. The record suggests this time period was lengthy; as of Febru ary 2010, the BRC seemed to indicate it had already worked with AMS for the prior two years on the proposed Checkoff Order. See Jordan Letter at 2 (AR1351) (expressing appreciation for AMS’s assistance “over the past two years”). Some amount of delay is therefore reasonable between the initial data gathering required to develop a proposed rule and the final rule issued after notice-and-comment rule-making. Nor was the agency ignoring substantially more recent data; when it issued the notice of proposed rulemaking, it appears that the latest year for which complete data was available was 2008. See BRC Proposal at 11 (AR1360) (providing only estimated as opposed to actual softwood-lumber consumption data for the year 2009). While Resolute is correct in recognizing that the difference between 2007 and 2010 was probably significant considering the effects the recession had on the softwood-lumber market, if this were the only problem with the agency’s data, USDA would likely be on firm footing.

ii. Production Capacity vs. Shipments

Resolute’s objection to the time period of the estimates gains traction, however, when considered alongside its complaint about substituting production capacity for shipments in selecting the 15mmbf de min-imis quantity. The Court shares Plaintiffs concern about the agency’s unaccounted use of production capacity in place of actual shipments, given the potentially vast differences between these measures. Resolute charges that “[t]he BRC stated, without justification or explanation, that production capacity was being ‘used in this analysis as a proxy for shipments.’ ” Reply at 5 (quoting BRC Proposal at 11 (AR1360)). Technically speaking, the BRC did provide some explanation: “Given current market conditions this table is ‘relatively’ correct, but doesn’t take into account recent temporary and permanent closures, reduced production, and possible omissions or double counting due to subsidiary relationships. Efforts were made to eliminate these.” BRC Proposal at 11 (AR1360).

This explanation nevertheless falls far short of a justification for the choice, particularly when the very same' page of the BRC Proposal makes clear just how stark the differences were between production capacity and shipments: while in 2009 actual U.S. consumption of softwood lumber was estimated to be only 31.9bbf, the estimates used to justify the 15mmbf exemption measured nearly 75bbf in production capacity in 2007—well over double actual consumption. Id. Here, the year in question made a substantial difference: the BRC’s data for 2007, 52.7bbf in shipments vs. nearly 75bbf in capacity, shows a far smaller gap between capacity and shipments than in 2010, confirming the significance of Resolute’s concern that the pre-recession data was significantly outdated. Id. Worse still, while the BRC was at least transparent about the difference between production capacity and actual shipments, nowhere in the agency’s notices did USDA make clear that its estimates regarding the 15mmbf exemption were based on production capacity, not actual shipments. It instead merely opaquely referenced—without citation—the BRC’s estimates. This measurement is also troubling because it treats all of U.S. and Canadian softwood lumber as a common market. Yet it is conceivable, if not probable, that much of Canadian softwood lumber remains in Canada and is not imported into the United States, which means that some additional portion of that production capacity would never turn into actual shipments to the United States.

3. Second Remand Explanation

Resolute’s arguments concerning use of 2007 capacity data vs. 2010 shipment data, confusion over whether 2.5% or 11.3% of softwood lumber would be exempted from assessment, and discrepancies in the estimates of the number of companies exempted and those that were eligible to participate in the referendum left the Court scratching its head, uncertain as to whether any of the data cited by either the BRC or USDA was likely to have been correct (let alone supportive of the 15mmbf de minimis exemption). While an agency’s “decision of less than ideal clarity” does not necessarily constitute one that is arbitrary or capricious, “the agency’s path [must nonetheless be] reasonably be discerned,” “including a rational connection between the facts found and the choice made.” State Farm, 463 U.S. at 43, 103 S.Ct. 2856 (citation and internal quotation marks omitted). The agency’s response to the first remand fell far short of this, with discrepancies implicated in nearly every pertinent estimate the agency provided in its notice of proposed rulemaking. The Court, as a result, was assured neither that the data supported the agency’s decision nor that it was accurate. Unable to discern the agency’s path,-the Court once more remanded the matter, this time specifically ordering that the agency provide:

1. An account of the BRC’s “Actual U.S. Consumption 2003-2009” estimate on page 11 of its BRC Proposal for a National Research and Promotion Program For Softwood Lumber (AR1360), and verification of this estimate based on its underlying source or sources;

2. An account of the BRC’s “Impact of Exemption on Check-off Participation: Capacity Removed from Assessment” estimate on page 11 of the same document, and verification of this estimate [that 257 companies representing 2.5% of capacity would be exempted] based on its underlying source or sources; and

3. Verification via underlying data of the estimates provided in the agency’s notice of proposed rulemaking (Softwood Lumber Research, Promotion, Consumer Education and Industry Information Order, 76 Fed. Reg. 22,757 (Apr.- 22, 2011) concerning the number and percentage of softwood-lumber market participants exempted from the checkoff order at the 15-million-board-feet threshold.

Resolute Forest Products, 2016 WL 1714312, at *4.

Defendants once again responded with a memorandum, this time from Charles W. Parrott, Deputy Administrator of the Specialty Crops Program of AMS. As this document provided responses to the Court’s specific requests in its second remand order, the Court will assess the explanations in light of the difficulties identified above.

a. Actual U.S. Softwood Lumber Consumption

As to the estimates of actual softwood-lumber consumption, Parrott responds that Stephen M. Lovett, who then worked for the BRC and prepared the estimates in the BRC Proposal, drew on data supplied by Random Lengths, “a firm that ‘provides the forest products industry with unbiased, consistent and timely reports of market activity and prices, related trends, issues, and analyses,’ ” Parrott Mem. at 2 (quoting id., Exh. B (About Random Lengths)), and from Forest Economic Advisors (FEA), which “describes itself as a firm that ‘brings modern econometric techniques to the forest products industry.’ ’’ Id. (quoting id, Exh. C (About Forest Economic Advis-ors)). Random Lengths, in turn, advised the agency in response to the second remand order that it obtains figures like those drawn on by Lovett from “industry associations, like the Western Wood Products Association, and industry analysts, like FEA.” Id. (citation and internal quotation marks omitted).

Although Lovett cannot precisely replicate the calculations he made in 2010, USDA provided a similar estimate based on data available to the Court, drawing on the. Forest Service’s Profile 2009 report. The agency pointed to “Table 4-United States softwood lumber end-use by market, 2003-2009” of the report, see Profile 2009 at 3, as a close approximation of the data included in “Actual U.S. Consumption 2003-2009,” BRC Proposal at 11 (AR1360). Because the Forest Service’s Profile 2009 measured total end use in cubic meters, the BRC converted this measure into billion board feet for its calculations. 1 Thus for the calendar year 2008, the estimate of 99.0 million cubic meters cited in Profile 2009 converts to approximately 41.9 billion board feet, slightly off of the 42.7bbf estimated by the BRC in its proposal to AMS in 2010. While the difference between 42.7bbf and 41.95bbf is not zero, given the differences and variations in underlying reporting sources for the softwood-lumber market, a difference of less than 2% is not itself alarming.

• - Even if this data seems generally reliable, as discussed earlier, the 15mmbf exemption threshold was set not on the basis of data about actual consumption, but on the basis of data bout production capacity. See PI. Second Remand Response at 7 (“USDA relied (if at all) on lumber production capacity data, not on lumber consumption data .... ”). And even the BRC’s own data recognizes how vast the differences were between production capacity and actual shipments. As noted above, the BRC’s proposal stated that actual U.S. consumption of softwood lumber was estimated to be only 31.9bbf in 2009, while the production capacity was estimated at nearly 75bbf, well over twice the consumption figure. See BRC Proposal at 11 (AR1360). Such a huge disparity undermines the credibility of either the 2.5% or 11.3% estimate as the actual quantity of shipped softwood lumber that would be exempted from assessment.

b. Impact of Exemption Estimates

The Court’s second remand order also requested that the agency provide “[a]n account of the BRC’s ‘Impact of Exemption on Check-off Participation: Capacity Removed from Assessment’ estimate ... and verification of this estimate based on its underlying source or sources.” Resolute Forest Products, 2016 WL 1714312, at *4. In the Parrott Memorandum, the agency explains that “Lovett prepared the impact-of-exemption estimate,” relying on an earlier version of the Profile 2009 report (see Second Notice, Exh. E (Henry Spelter, David McKeever & Matthew Alderman, Profile 2007: Softwood Sawmills in the United States and Canada, FPL-RP-644 (Oct. 2007))), as well as “a draft of Profile 2009 that Mr. Lovett obtained” from the authors of what would eventually become the final published Profile 2009 report because “he wanted to use the most recent data available.” Parrott Mem. at 4. The agency also explains that “[t]he updates to the data that Mr. Lovett used consisted of information that he obtained ... regarding mill closures and mills not in operation because of the economic downturn that began in 2007.” Id. This explanation does not seem to square with Lovett’s ultimate estimates included in the BRC’s proposal, for we are told that “2007 Capacity [was] used in this analysis as a proxy for ‘shipments.’ ” BRC Proposal at 11 (AR1360). If so, then what happened after 2007 would be irrelevant to these estimates. This is yet another instance in which the agency’s explanation is not on all fours with the evidence available in ¡-the administrative record.

In part to shore up doubt, the agency states that “[a]t thé request of USDA, [Paul] Jannke of FEA has prepared ... two impact-of-¿xemption estimates using data” from “individual sawmill capacity from Appendix C to Profile 2Q07, adjusted for mills known by FEA to have closed in 2008,” as well as an .estimate drawing on “data on individual sawmill capacity from the Appendix to Profile 2009.” Parrott Mem, at 5. Neither of these estimates is particularly helpful, however, as both simply rely on the same data without explaining the BRC’s method that converted 897 sawmills identified in the Profile 2009 report, see Resolute Forest Products, 2016 WL 1714312, at *4 tbl. 1, into approximately 629 companies, 264 (or 243) of which supposedly had production capacity of less than 15mmbf. See Parrott Mem. at 5; see also BRC Proposal at 11 (AR1360). It simply defies logic that the agency has failed on multiple remands to explain precisely how it derived its estimates for the' number of companies excluded and included under the Checkoff Order, and it strongly suggests that USDA' never ¿dually knew them.

This raises a related problem with another of USDA’s stated reasons for selecting 15mmbf as the de minimis quantity: generating sufficient revenue for an effective checkoff order. The substitution" of capacity for shipments raises serious doubts as to whether the Checkoff Order would in fact raise the revenue both the BRC and the agency stated it must raise to be successful. In its notice of proposed rulemaking, the agency affirmed the BRC’s conclusion that “an exemption threshold of 15[mmbf] was appropriate and would generate sufficient income to support an effective promotion program for softwood lumber.” 75 Fed. Reg. at 61,013. This conclusion presumably drew on the BRC’s proposal, which stated that “$20 million (from asséssments) is the threshold for an effective program that can move the needle.” 'BRC Proposal at 10 (AR1369). The BRC estimated that “an initial assessment fate of $0.35/mbf ...' would raise sufficient funds for a $20 million program.” Id, The BRC’s own chart, however, recognized that with a 15mmbf exemption threshold, the Checkoff Order would either require shipments of 60bbf to yield $21 million at an assessment rate of $0.35/mbf or else necessitate upping the assessment rate to $0.50/mbf to yield $20 million on 40bbf in shipments. Id. at 10-11 (AR1359-60). If actual shipments in 2009 were 31.9bbf, however, the Checkoff Order would have yielded far less than the $14 million that was estimated at 40bbf in assessments, id. at 10 (AR1359), itself an amount far lower than what the BRC suggested was necessary for an effective marketing campaign. Given these issues with the underlying data, it is difficult to understand how the agency could have concluded that the 15mmbf exemption “would generate enough income to support a viable and effective research and promotion program for softwood lumber.” Barnes Mem. at 6. This, of course, is only one of the defects in the data the agency claims supported the 15mmbf exemption..

c. Estimates of Companies Exempted and Total Companies

The Court also ordered the agency to clarify seemingly contradictory estimates of the number and percentage of exempted softwood-lumber producers and exporters included in the Federal Register notices to ensure that “some verifiable source of data accurately depicted the softwood-lumber market.” Resolute Forest Products, -2016 WL 1714312, at *3. Prior to the second remand order, the agency had never been able to provide a coherent account of the estimates used to assess the number of softwood-lumber companies that would be exempt from the Checkoff Order. The Par-rott Memorandum, unfortunately, falls short as.well. As the Court explained in its second remand order, the 595 domestic “manufacturers” that USDA cited in its notice of proposed rulemaking appears instead to be a three-year average (2007-2009) from Profile 2009 estimates for the number of sawmills in the U.S. Id. at *2-3. Whereas USDA stated that this “number represents separate business entities [where] ... one business entity may include multiple sawmills,” 75 Fed. Reg. at 61,012, the Profile 2009 report on which that estimate was based clearly specifies that its count consists of sawmills, not business entities. See Profile 2009 at 15 (“The following maps and tables show past and current capacity of sawmills and the availability of timber, by county, in the vicinity of these mills .... ”).

The agency retorts in the Parrott Memorandum that “[bjecause the industry was in a state of flux, USDA considered it reasonable to use the figure 595 .... [but] should have explained, however, that the Forest Service figures were for sawmills _” Parrott Mem. at 8. It further defends that “USDA had no data on how many sawmills were individual business entities or were part of a group of sawmills making up one business entity .... Therefore, USDA treated each domestic manufacturer (sawmill) as a separate entity in its analysis.” Id. This explanation is extraordinary given that the agency expressly characterized the estimate as measuring entities, see 75 Fed. Reg. at 61,012 (“This number represents separate business entities; one business entity may include multiple sawmills”), and then relied on that measure to determine the number of companies that would be exempted under the proposed Checkoff Order. See, e.g., id. at 61,013 (“Of the 595 domestic manufacturers, it is estimated that about 232, or 39 percent, ship less than 15[mmbf] per year and would thus be exempt from paying assessments under the proposed Order.”); 76 Fed. Reg. at 22,772 (“USDA concurs with this exemption level because this level would exempt small operations that would otherwise be burdened by the assessment.”).

The Parrott Memorandum also reveals that the agency never really knew how many companies ship less than 15mmbf: “[25mmbf] per year is the lowest number of board feet for which [the American Lumber Standard Committee] segregated shipment data. Having no individual company shipment data to use for U.S. entities ... USDA referred in these sentences to shipments of 25[mmbf] per year rather than shipments of 15[mmbf] per year.” Id. This is incredible considering the agency’s repeated contention that it justified the 15mmbf number on the basis of the number of companies that would be exempted from assessments. In reality, it had no reliable data whatsoever concerning domestic entities shipping less than 15mmbf per year. The agency’s explanation that it lacked, such data, furthermore, in no way justifies falsely portraying its estimates as being those of entities rather than sawmills. As Senator Daniel Patrick Moynihan once said, “Everyone is entitled to his own opinion, but not his own facts.”

Even worse, however, the agency then incorrectly transmuted the number of entities shipping less than 25mmbf—363, according. to the ALSO—for the number shipping more than 15mmbf. It appears to have subtracted 363 (entities that ship less than 25mmbf) from the 595 (total sawmills) to conclude—arbitrarily—that “about 232, or. 39 percent, ship less than 15[mmbf] per year and would thus be exempted from paying assessments_”76 Fed. Reg. at 22,767. The agency has no explanation for this astounding error, instead simply acknowledging it in a footnote on remand. See Parrott Mem. at 9 n. 3. In sum, the agency substituted 15mmbf for 25mmbf, sawmills for entities, and production capacity for shipments, without being transparent. about any of these substitutions. To garnish this plate of errors, it then got its basic arithmetic backwards.

Defendants’ data on foreign importers is hardly more assuring. As a reminder, the agency relied on “Customs data” that suggested that “there were about 883 importers of softwood lumber annually.” 75 Fed. Reg. at 61,012. The agency further stated that “[a]bout 798 importers, or about 90 percent, imported” so little softwood lumber as to be considered small entities. Id. The notice of proposed rulemaking later stated that 780 out of 883 importers shipped less than 15mmbf, and so only 103 foreign importers would pay assessments under the Order. Id. at 61,013. In contrast to most of the other estimates it discussed, the agency provided no citation .as to the specific source of that estimate. Yet, despite the Court’s express instruction in its second remand order to provide “[vjerifieation via underlying data of the estimates provided in the agency’s notice of proposed rulemaking ... concerning the number and percentage of softwood-lumber market participants exempted from the checkoff order,” Resolute Forest Products, 2016 WL 1714312, at *4, the agency failed to provide any additional support for the claim. Instead, the Parrott Memorandum merely repeats the agency’s conclusory statement in its notice. Compare Parrott Mem. at 9 (“These sentences are based on information obtained by USDA from Customs and Border Protection (CBP) for the years 2007-2009. CBP is the sole source of information available to USDA concerning importers of record.”), with 75 Fed. Reg. at 61,012 (“[A]ccording to Customs data, it is estimated that, between 2007 and 2009, there were about 883 importers ... ”). After additional opportunities to substantiate its estimates, - that is not good enough.

The absence of the underlying data is especially galling considering that the number of Canadian sawmills derived from the Forest Service’s Profile 2009—an average of roughly 349, see Resolute Forest Products, 2016 WL 1714312, at *4 tbl. 1— is far smaller than the 883 importers cited by the agency. Because USDA itself stated that “imports from Canada ... comprisfe] about 92 percent of total imports,” 75 Fed. Reg. at 61,004, it seems incredible that 883 separate entities import softwood lumber into the U.S. despite the existence of only 349 Canadian softwood-lumber sawmills in total.

Defendants also appear to have introduced new errors in the Parrott Memorandum, in which it is claimed that

USDA estimated that 335 .entities domestically shipped or imported 15[mmbf| or more annually and, therefore, would pay assessments under the program (232 U.S. manufacturers and 103 importers) and 1,143 entities domestically shipped or imported less than 15[mmbf] annually and would be exempt from paying assessments (363 U.S. manufacturers and 780 importers). Given the uneer-tainty in the industry at the time with mills closing or not operating, USDA’s estimate proved to be remarkably accurate. The 335-estimate of assessment payers was very close to the number of entities (311) that were found eligible to vote in the 2011 referendum ....

Parrott Mem. at 10. Dismayingly, the agency seems once again to have transmuted its own incorrect figures. Parrott claims that 232 U.S. manufacturers were estimated to pay assessments and 363 would be exempt, but the agency’s notice in the Federal Register stated just the opposite. See 75 Fed. Reg. at 61,013 (“[I]t is estimated that. about 232 ... ship less than 15[mmbf] per year and would thus be exempt from paying assessments ... [and] about 363 domestic manufacturers ... would pay assessments —”) (emphasis added). The Parrott Memorandum thus should have said that the total number of estimated entities paying in was 466. See id. (“Thus, about 363 domestic manufacturers and 103 importers would pay assessments under the Order.”) This 466 estimate—which itself is based on completely spurious estimates, as discussed above—is itself not close to 311 at all. The only thing remarkable about the agency’s estimates is that, even after two remands, USDA still manages to introduce new basic computational errors into its calculations in an effort to shore up its shoddy data.

In sum, the little data the agency presented in its rulemaking notices was patently misrepresentative, and after two remands it has not provided a more reliable source. The agency still has not been able to offer a coherent explanation for its estimate that “about 363 domestic manufacturers and 103 importers would pay assessments under the Order.” Id. No source—the agency, the ALSO, or the Forest Service’s Profile 2009 report— seems to identify how many domestic manufacturers produce less than 15mmbf per year. Lacking reliable data, the Court has no way whatsoever to assess the impact of the 15mmbf exemption on small entities, and it casts doubt on whether the agency even had its eyes on the road as it steered the proposed Checkoff Order through notice and comment.

¾* $ *

As the Court has thoroughly expounded above, the agency’s explanation of its selection of 15mmbf as the de minimis quantity exempted raises a litany of problems. Its reliance on production capacity estimates from 2007 for a rule assessing actual shipments and implemented nearly four years later undermines the agency’s ability to rely on estimates regarding the percentage of softwood lumber removed from assessment. Given that actual shipments were estimated to be less than half of production capacity during this period, it also strongly calls into question whether the Checkoff Order could produce the revenue both the BRC and USDA stated were necessary to run an effective marketing campaign. Worse still, the agency has gone back and forth as to whether it relied on 2.5% or 11.3% of production capacity as the “de minimis” quantity. Its contradictions suggest the agency is either uncertain about why it made its decision, or else is simply making it up as it goes along.

In all probability, of course, neither estimate is likely to represent the actual quantity of shipments excluded from assessment under the Order. Nor does the Court have any way to verify whether this is true: despite two chances on remand, the agency has not'provided an adequate explanation for how it transmuted data from the Forest Service’s Profile 2009 report on production capacities for sawmills into data on shipments by entities. Nor has it provided the underlying U.S. Customs data it purports to have used to estimate the number of foreign importers. The agency’s problems do not end there, however. On first remand the Barnes Memorandum states that USDA considered “the impact of program requirements on small businesses,” Barnes Mem. at 2, but essentially all of those data seem faulty, contradictory, or unsubstantiated, and the agency’s Parrott Memorandum on second remand could not resolve them. The agency’s claim that it considered the “free rider implications” of a 15mmbf exemption is not substantiated by any indication of this whatsoever in its rulemaking notices.

The record, in sum, simply contains too many misstatements, unsubstantiated (or incorrect) estimates, and ■ statements con tradicted by the agency’s subsequent litigation positions to support the selection of 15mmbf as the de minimis quantity. It is no rejoinder that the BRC had better estimates and a clearer understanding of the measurements in question. While the CPRIA contemplates cooperation between the agency and industry groups in proposing and implement checkoff orders, the Secretary remains obligated under the statute to “determinen that a proposed order is consistent with and will effectuate the purpose” of the CPRIA, 7 U.S.C. § 7413(b)(2), and this must—at a minimum—require an independent verification that there was a “rational connection between the facts found and the choice made.” Americans for Safe Access, 706 F.3d at 449 (internal quotation marks omitted). Given the record in this case, no reliable evidence suggests the agency verified (or even could verify) a rational connection between the estimates and the BRC’s selection of 15mmbf as the de min-imis quantity. The agency, furthermore, is required to “give due notice” about the proposed order in the Federal Register, see 7 U.S.C. § 7413(b)(2), and “due notice” surely requires reasonably accurate (and certainly not blatantly misleading) data to substantiate its decision and provide interested commentators with the opportunity to assess the proposed rule.

As cited above, “an agency rule [is] arbitrary and capricious if the agency .,. offered an . explanation 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.” State Farm, 463 U.S. at 43, 103 S.Ct. 2856. None of the relevant evidence provided by USDA during rulemaking could reasonably be relied upon to conclude that 15mmbf would be a de minimis quantity because none of the statistics cited can be reasonably relied upon to measure what they purport, And where an agency has relied on incorrect or inaccurate data or has not made a reasonable effort to ensure that appropriate data was relied upon, its decision is arbitrary and capricious and should be overturned. See, e.g., Native Vill. of Point Hope v. Jewell, 740 F.3d 489, 502-03 (9th Cir.2014) (overturning agency’s deterr mination as arbitrary and capricious after finding agency assumptions were made based on contradictory estimates and without rational basis in record); Kentucky Riverkeeper, Inc, v. Rowlette, 714 F.3d 402, 410 (6th Cir.2013) (overturning as arbitrary and capricious agency’s permit reauthorization where agency relied on inappropriate estimates to gauge impact of reauthorization); Sierra Club v. EPA, 671 F.3d 955, 965-66 (9th Cir.2012) (overturning as arbitrary and capricious agency’s action where it failed to consider newer “data [that] told a different story than .,. earlier data” that agency had actually relied upon and where agency had failed to provide an adequate explanation for its reliance on outdated data).

In short, “a court must be satisfied from the record that ‘“the agency ... examine[d] the relevant data and articulate[d] a satisfactory explanation for its action.’ ” Islander E. Pipeline Co., LLC v. McCarthy, 525 F.3d 141, 151 (2d Cir.2008) (quoting State Farm, 463 U.S. at 43, 103 S.Ct. 2856). After all, “[t]he requirement that' agency action not be arbitrary, or capricious includes a requirement that the agency adequately explain its result.” Public Citizen, Inc, v. FAA, 988 F.2d 186, 197 (D.C.Cir.1993). This standard “mandat[es] that an agency take whatever steps it needs to provide an explanation that will enable the court to evaluate the agency’s rationale at the time of decision.” Dickson v. Sec’y of Def., 68 F.3d 1396, 1404 (D.C.Cir.1995) (quoting Pension Benefit Guaranty Corp. v. LTV Corp., 496 U.S. 633, 654, 110 S.Ct. 2668, 110 L.Ed.2d 579 (1990)). The Court has given the agency-multiple chances to provide that explanation, and it has fallen short each time. Without any reliable data to support the selection of 15mmbf as the de minimis quantity exempted, that decision cannot be characterized as anything other than arbitrary and capricious.

Finally, what of Resolute’s constitutional challenges? Because the Court has found the Checkoff Order arbitrary and capricious as promulgated, it need not reach Resolute’s constitutional challenges to the CPRIA, both facial and as applied. See Spector Motor Serv. v. McLaughlin, 323 U.S. 101, 105, 65 S.Ct. 152, 89 L.Ed. 101 (1944) (“If there is one doctrine more deeply rooted than any other in the process of constitutional adjudication, it is that we ought not to pass on questions of constitutionality ... unless such adjudication is unavoidable.”); see also Resolute Forest Products, 130 F.Supp.3d at 105.

IY. Conclusion

On the basis of the contradictory, conflicting, and misstated estimates described above, the Court concludes that the agency’s selection of 15mmbf as the de minimis quantity was arbitrary and capricious and that, accordingly, the Checkoff Order was promulgated unlawfully. The Court in the accompanying Order will set a hearing to discuss the appropriate next steps concerning the remedies sought by Plaintiff.

1 . Throughout this Opinion, the Court uses the ratio of 2,36 cubic meters per 1000 board feet (or 1:423) to convert between these two measures. See 75 Fed. Reg. at 61,010 (“One cubic meter is equal to 423,776001 board feet.”); see also Parrott Mem. at 2-3.