Taranto, Circuit Judge.

Maureen E. Thorson, Wiley Rein, LLP, Washington, DC, argued for plaintiff-appellee. Also represented by Stephanie Manaker Bell, Tessa V. Capeloto, Laura El-Sabaawi, Cynthia Cristina Galvez, Derick Holt, Daniel B. Pickard, Adam Milan Teslik.

John Jacob Todor, Commercial Litigation Branch, Civil Division, United States Department of Justice, Washington, DC, argued for defendant-appellee. Also represented by Jeffrey B. Clark, Jeanne Davidson, Franklin E. White, Jr. ; Paul Keith, Office of the Chief Counsel for Trade Enforcement and Compliance, United States Department of Commerce, Washington, DC.

Gregory S. Menegaz, DeKieffer & Horgan, PLLC, Washington, DC, argued for defendant-appellant. Also represented by James Kevin Horgan, Alexandra H. Salzman.

Before Prost, Chief Judge, Clevenger and Taranto, Circuit Judges.

Taranto, Circuit Judge. Since 2006, importation of diamond sawblades from the People's Republic of China (PRC) has been governed by an antidumping duty order issued by the United States Department of Commerce under 19 U.S.C. § 1673. In 2016, Commerce launched an administrative review, under 19 U.S.C. § 1675, of duties owed on subject merchandise sold to unaffiliated U.S. purchasers from November 1, 2014, through October 31, 2015. In that review, Commerce investigated the dumping margin of Bosun Tools Co., Ltd. (Bosun), an exporter and producer of diamond sawblades from the PRC, which it sends directly to one of its two U.S. importer-affiliates for sale to unaffiliated U.S. purchasers. The second importer-affiliate imports diamond sawblades from a Bosun entity in Thailand (which are not covered by the antidumping duty order). The two importer-affiliates trade between themselves, so both end up selling PRC-originating and Thailand-originating sawblades.

To determine the domestic-price component of the dumping margin calculation, Commerce had to identify which diamond sawblades sold by the Bosun importer-affiliates to unaffiliated U.S. purchasers were from the PRC (not Thailand). Because Bosun's affiliates (and Bosun's overall database) did not record the country of origin on each sale to those purchasers, Bosun supplied country-of-origin information from three sources: (1) the particular product code (which was country-specific for some products); (2) the unit price (which allowed origin identification for some products); and (3), for remaining products, an inference as to origin based on the premise that the importer-affiliates generally sold products in the order they received them (the first-in, first-out, or FIFO, inference).

To calculate Bosun's margin, Commerce used the information Bosun provided, finding it sufficiently verified. The domestic-industry Diamond Sawblades Manufacturers’ Coalition challenged Commerce's determination in the Court of International Trade, which remanded the matter to Commerce for further explanation. Diamond Sawblades Mfrs.’ Coalition v. United States , No. 17-00167, 2018 WL 5281941 (Ct. Int'l Trade Oct. 23, 2018) ( DSMC I ). On remand, Commerce noted problems with some of Bosun's information—perhaps only with the small subset of products for which the FIFO-inference step was used for origin identification—and concluded that it would use "the facts otherwise available" under 19 U.S.C. § 1677e(a), and indeed draw adverse inferences under § 1677e(b), as to the totality of the Bosun-sawblade sales during the period of review. The Trade Court affirmed Commerce's determination. Diamond Sawblades Mfrs.’ Coalition v. United States , 415 F. Supp. 3d 1365, 1369 (Ct. Int'l Trade 2019) ( DSMC II ).

We now conclude that some of the bases on which Commerce invoked § 1677e(a) are unsupported by substantial evidence, while some—which involve only a gap in reliable information—are adequately supported. We also conclude, however, that, in light of the limited bases for applying § 1677e(a), Commerce may have applied that subsection—and hence § 1677e(b), which applies only where subsection (a) applies—too broadly by disregarding all of Bosun's country-of-origin information. It appears that the errors Commerce identified in Bosun's information are limited in their reliability-undermining effect to a defined subset of sold sawblades (the subset of sawblades whose origin Bosun identified only through the FIFO-inference step). If the unreliable information is confined to some or all sawblades within such a defined subset, then there is no substantial evidence to support Commerce's determination that all of the Bosun-supplied origin information was unreliable, and Commerce articulated no supported basis for disregarding the reliable portion of the origin information Bosun supplied. We remand for further proceedings to determine the extent to which unreliability is so confined, and the consequence for Bosun's dumping margin. We leave to the Trade Court the decision whether a further remand to Commerce is needed.

I

A

Under 19 U.S.C. § 1673, Commerce must determine whether merchandise at issue is being sold or is likely to be sold in the United States "at less than fair value," which the statute identifies as "dumping," id. § 1677(34). To make that determination, Commerce must assess the difference between the "normal value" of the goods at issue (reflecting the home-market value) and the "export price or constructed export price" of those goods (reflecting the price at which they are sold into the United States). See id. § 1677b(a) (stating that the determination of the existence of sales "at less than fair value" is to be based on a comparison of "the export price or constructed export price and normal value"); see also id. § 1677a (addressing "export price" and "constructed export price"); id. § 1677b (addressing "normal value"). That difference is the "dumping margin." Id. § 1677(35)(A) (defining "dumping margin"). If Commerce finds dumping, and the International Trade Commission makes specified findings about injury to domestic industries, Commerce is to issue an antidumping duty order that imposes duties to offset the dumping. Id. § 1673.

Thereafter, Commerce typically conducts annual reviews to determine the antidumping duty margin for a given 12-month period for relevant exporters. Id. § 1675. In particular, § 1675(a)(1)(B) states that "[a]t least once during each 12-month period beginning on the anniversary of the date of publication[,] ... [Commerce], if a request for such a review has been received and after publication of notice of such review in the Federal Register, shall ... review, and determine (in accordance with paragraph (2)), the amount of any antidumping duty," and, under subsection (a)(2), "(i) the normal value and export price (or constructed export price) of each entry of the subject merchandise, and (ii) the dumping margin for each such entry." Commerce then "shall determine the individual weighted average dumping margin for each known exporter and producer of the subject merchandise," id. § 1677f–1(c)(1), and may elect to rely on "a sample of exporters, producers, or types of products that is statistically valid based on the information available" or "exporters and producers accounting for the largest volume of the subject merchandise" if "it is not practicable to make individual [determinations] because of the large number of exporters or producers involved in the investigation or review," id. § 1677f–1(c)(2). In the administrative-review context, Commerce's use of the collected information is guided in part by 19 U.S.C. § 1677e. Subsection (a) states:

If—

(1) necessary information is not available on the record, or

(2) an interested party or any other person—

(A) withholds information that has been requested by the administering authority or the Commission under this subtitle,

(B) fails to provide such information by the deadlines for submission of the information or in the form and manner requested, subject to subsections (c)(1) and (e) of section 1677m of this title,

(C) significantly impedes a proceeding under this subtitle, or

(D) provides such information but the information cannot be verified as provided in section 1677m(i) of this title,

[Commerce] shall, subject to section 1677m(d) of this title, use the facts otherwise available in reaching the applicable determination under this subtitle.

Id. § 1677e(a). Where subsection (a) applies, subsection (b) adds that if an additional condition is also met, Commerce "may" draw inferences adverse to an interested party "in selecting from among the facts otherwise available" whose use subsection (a) authorizes:

If [Commerce] finds that an interested party has failed to cooperate by not acting to the best of its ability to comply with a request for information from [Commerce], [Commerce], in reaching the applicable determination under this subtitle, may use an inference that is adverse to the interests of that party in selecting from among the facts otherwise available[.]

Id. § 1677e(b).

Section 1677e(a) refers to four portions of 19 U.S.C. § 1677m. Two of those are referred to in § 1677e(a)(2)(B). One is § 1677m(c)(1), which says that, in certain circumstances, Commerce must consider an interested party's inability to submit requested information "in the requested form and manner" and may modify the requirements to avoid an unreasonable burden on the party. The other is § 1677m(e), which provides that, in a § 1675 review (among other proceedings), Commerce "shall not decline to consider" party-submitted information needed for Commerce's determination, even though the submission does not meet all Commerce-established requirements, if certain conditions keyed to reliability are nevertheless met.

The third subsection of § 1677m to which § 1677e(a) refers (in § 1677e(a)(2)(D) ) is § 1677m(i). That subsection states that verification of information is required in an administrative review like this one if, first , certain interested parties timely request verification and, second , either there was no verification in the previous two administrative reviews or there is good cause for a new verification. Id. § 1677m(i). The final subsection of § 1677m to which § 1677e(a) refers is § 1677m(d), referred to in § 1677e(a) ’s concluding clause—Commerce "shall, subject to section 1677m(d) of this title, use the facts otherwise available ...." Subsection 1677m(d) provides, first , that if a response to an information request "does not comply with the request," Commerce "shall, to the extent practicable, provide ... an opportunity to remedy or explain the deficiency in light of the time limits" set for the proceeding and, second , that if a further submission made in response to the deficiency is untimely or Commerce "finds that such response is not satisfactory," Commerce "may, subject to subsection (e) [quoted supra ], disregard all or part of the original and subsequent responses." Id. § 1677m(d).

Interested parties, including foreign producers or exporters of subject merchandise, importers of such merchandise, and specified domestic trade associations, are allowed to participate in administrative reviews. See 19 U.S.C. § 1677(9)(A), (E) ; 19 C.F.R. § 351.309(c). An interested party that was a party before Commerce may file an action in the Trade Court under 19 U.S.C. § 1516a to challenge Commerce's final determination in an administrative review. 28 U.S.C. § 2631(c) ; id. § 1581(c).

B

Commerce issued an antidumping duty order in 2006 covering "diamond sawblades and parts thereof" (hereafter "sawblades") from the PRC. Final Determination of Sales at Less Than Fair Value and Final Partial Affirmative Determination of Critical Circumstances: Diamond Sawblades and Parts Thereof from the People's Republic of China, 71 Fed. Reg. 29,303 (May 22, 2006). Commerce thereafter conducted annual administrative reviews under 19 U.S.C. § 1675. This case involves the sixth such review, initiated on January 7, 2016. Initiation of Antidumping and Countervailing Duty Admin. Revs., 81 Fed. Reg. 736 (Jan. 7, 2016).

1

Commerce initially selected Jiangsu Fengtai Diamond Tool Manufacture Co., Ltd. (Jiangsu), the largest exporter of sawblades from the PRC, along with a second firm, for individual investigation, but on April 27, 2016, Commerce, while retaining Jiangsu for investigation, dropped the initially selected second firm and substituted Bosun—which was the third largest sawblades exporter listed on the initiation notice and which had been selected for individual investigation in three earlier annual reviews. J.A. 56–57. Bosun responded to Section A of Commerce's antidumping questionnaire on May 25, 2016, and Sections C and D on July 1, 2016. Bosun's responses included aggregate data about the quantity and value of its U.S. sales. J.A. 785–93. Bosun explained that it imported sawblades both from the PRC and from Thailand through its U.S.-based affiliated importers; specifically, Bosun Tools, Inc. (Bosun USA) imported only from the PRC; Pioneer Tools, Inc. (Pioneer) imported only from Thailand. J.A. 732. Bosun's responses also noted that its two importer-affiliates sold sawblades between themselves before selling to unaffiliated U.S. customers, thus intermixing the PRC and Thailand sawblades in the affiliates’ hands. J.A. 732. It is undisputed that Bosun's U.S.-sales database, which operated as an omnibus repository for both importers’ sales records, did not record the intra-family sales and did not record the country of origin of sawblades at the time they were sold by the importer-affiliates to unaffiliated U.S. customers (the sales that matter for the annual review). J.A. 732–33, 2886. Bosun accepts that its importer-affiliates did not record the country of origin, on invoices or otherwise, at the time of sale of sawblades to unaffiliated U.S. customers. See DSMC I , 2018 WL 5281941, at *2. Bosun told Commerce that it had derived the origin of sawblades sales by using a multistep method: The product codes for the sawblades often distinguished the country of origin; the unit prices of sales did so for some of the sawblades whose origin was not identified in the first step; and a premise that the importer-affiliates sold on a "first-purchase first-sale" basis allowed an inference as to origin for remaining sales, using the sales dates along with the dates of the affiliates’ receipt of sawblades. J.A. 733–34.

On August 3, 2016, Commerce issued a first supplemental questionnaire asking Bosun to describe how Pioneer "segregated subject merchandise [i.e. , sawblades from the PRC] from diamond sawblades that it purchased from Thailand." J.A. 1141; J.A. 2967 n.82. Bosun responded on September 7, 2016, explaining that Pioneer purchased sawblades only from Bosun's Thai affiliate and from Bosun USA, so the subject merchandise in Pioneer's sales records would be only those products purchased from Bosun USA. J.A. 1155. Bosun also elaborated on its earlier explanation of the method by which it had segregated the subject merchandise, i.e. , identified the PRC-origin sawblades. First, Bosun identified models of sawblades by identifying unique "product codes" assigned to each affiliate; if those codes were not affiliated with Bosun USA, the sale was not of subject merchandise. J.A. 1141–42, 1154–56. Second, Bosun compared the unit purchase price of sawblades whose origin had not been identified based on the product code to the unit purchase price of sawblades whose origin had been so identified. Id. Third, for sawblade sales by affiliates to unaffiliated customers for which the first two steps did not identify the country of origin, Bosun applied what the parties now call a FIFO inference, based on the assumption that Bosun's U.S. affiliates sold their oldest inventory first (and knew the dates of sales and arrivals of inventory). Id.

Commerce issued a second supplemental questionnaire on October 17, 2016, asking that Bosun "provide a key to the product codes of Bosun's subject merchandise" and "explain how you identified these products as produced in China and exported from China, produced in Thailand and exported from Thailand, or produced in China and exported through Thailand." J.A. 2139–40; J.A. 2968 n.84. Bosun timely responded to the second supplemental questionnaire on November 10, 2016, illustrating the already-described process as applied to certain "sales trace[s]"—seemingly the sequence of documents in Bosun's sales database that culminated in the invoice to an unaffiliated U.S. customer. J.A. 2140–42.

Commerce published its preliminary results on December 9, 2016. See Diamond Sawblades and Parts Thereof From the People's Republic of China: Preliminary Results of Antidumping Duty Admin. Rev.; 2014–2015, 81 Fed. Reg. 89,045 (Dec. 9, 2016). On January 17, 2017, Diamond Sawblades requested, as an interested party under 19 C.F.R. § 351.309(c), that Commerce invoke 19 U.S.C. § 1677e(a) and (b) to use adverse inferences in calculating Bosun's export prices (and hence dumping margin). J.A. 2578–87. Diamond Sawblades argued, in particular, that Bosun's information about country of origin was defective and that Bosun did not cooperate to the best of its ability because it and its importer-affiliates did not record the country of origin for each individual sale to an unaffiliated U.S. customer (or therefore have such records to provide to Commerce). J.A. 2579–87.

Commerce issued a letter on April 28, 2017, informing Bosun that it would verify Bosun's questionnaire responses, and asking Bosun to provide a number of "sales-trace package[s]" for sales that Commerce identified. J.A. 2612, 2619. On May 27, 2017, Commerce issued its verification report (2017 Verification Report), which explained that the analysts "recreated [Bosun's] segregation between Chinese and Thai origin products" and "found no discrepancies" in the first two steps of Bosun's method. J.A. 2891–92. The 2017 Verification Report also stated that one of the identified sales traces reported a lower quantity of PRC-originating products than had actually occurred, J.A. 2892–93, a discrepancy that was "a result of the FIFO methodology Bosun used to identify the country of origin," J.A. 2892. The report stated, however, that "[o]ther than the on-site selected sales trace 6, [the analysts] did not find discrepancies in the sales traces that [they] reviewed for sales identification methodology." J.A. 2893.

2

Commerce issued its final Issues and Decision Memorandum on June 6, 2017, recommending that Commerce use Bosun's verified data to calculate the proper antidumping duty margin. J.A. 2942, 2966. Commerce published its final results based on that conclusion (Final Determination) on June 12, 2017. Diamond Sawblades and Parts Thereof From the People's Republic of China: Final Results of Antidumping Duty Admin. Rev.; 2014–2015, 82 Fed. Reg. 26,912 (June 12, 2017). In the memorandum, Commerce explained that it would not apply "the facts otherwise available" under 19 U.S.C. § 1677e(a) or, therefore, draw adverse inferences in selecting from such facts under § 1677e(b). J.A. 2967–70.

Diamond Sawblades challenged Commerce's determination before the Trade Court on June 27, 2017, alleging that Commerce should have applied 19 U.S.C. § 1677e(a) and (b) in calculating Bosun's dumping margin. The Trade Court decided that a remand was warranted on this issue. Quoting Peer Bearing Co.-Changshan v. United States , 766 F.3d 1396, 1400 (Fed. Cir. 2014), and Nippon Steel Corp. v. United States , 337 F.3d 1373, 1382 (Fed. Cir. 2003), the Trade Court concluded that it was unclear if Commerce sufficiently considered precedent to the effect that "the ‘best of its ability’ standard ‘requires the respondent to do the maximum it is able to do,’ inclusive of ‘maintain[ing] full and complete records’ of relevant data." DSMC I , 2018 WL 5281941, at *4 (alteration in original); see also id. at *4–7. The Trade Court considered both of the "best of its ability" statutory provisions quoted above: § 1677m(e) ’s criterion for using information that does not meet all Commerce requirements (the interested party "acted to the best of its ability in providing the information and meeting [Commerce's] requirements"), which plays a role in § 1677e(a)(2)(B) and (indirectly) in the last clause of § 1677e(a) ; and § 1677e(b) ’s precondition to using an adverse inference ("an interested party has failed to cooperate by not acting to the best of its ability to comply with a request for information from" Commerce). DSMC I , 2018 WL 5281941, at *4–8. The Trade Court also noted that Commerce, during verification, had identified errors in Bosun's information regarding some of the sales for which the FIFO step was used to infer origin; and the court concluded that Commerce had not sufficiently explained its determination that the errors were isolated enough not to warrant use of an adverse inference under § 1677e(b). Id. at *7–8. The Trade Court remanded for further explanation. Id. at *8.

3

Commerce issued a Final Remand Redetermination on April 17, 2019. This time, Commerce found § 1677e applicable, disregarded all of Bosun's information about the country of origin for all of its importer-affiliates’ sales during the period of review, and assigned Bosun an antidumping dumping margin of 82.05%. J.A. 3048–51, 3062–63. That figure was the margin Commerce assigned to Jiangsu, the other individually investigated exporter; Jiangsu's margin was itself based on adverse inferences under § 1677e(b). J.A. 3062 (citing J.A. 2945–54).

As to the premises of its new conclusion: Commerce reasoned that it would "resort to the facts otherwise available" under § 1677e(a) because four of the statutorily specified conditions for doing so were met here. J.A. 3046 (relying on § 1677e(a)(1), (a)(2)(B)–(D) ). Commerce noted that it had identified certain errors in Bosun's information during verification, referring specifically (perhaps only—this is unclear) to sales for which "the FIFO methodology" was used. J.A. 3049 & n.37. Commerce disregarded all of Bosun's information about origin for the entirety of the importer-affiliates’ sales, even the sales for which Bosun identified origin based on product type or unit price. J.A. 3049. It stated that "Bosun had the ability to maintain [country-of-origin] records" at the point of sale to unaffiliated U.S. purchasers, "but failed to do so," and that "Bosun is familiar with Commerce's antidumping duty proceedings and should have understood the importance of maintaining adequate country of origin information." J.A. 3047. On that basis, Commerce declared Bosun's information in its entirety " ‘not satisfactory,’ " leaving Commerce with "no reliable information on the country of origin of Bosun's sales." J.A. 3049 (quoting § 1677m(d) ). Finally, having decided to disregard all of Bosun's origin information, Commerce concluded that, in selecting from the information otherwise available, it should use an adverse inference under § 1677e(b) because Bosun flunked the "best of its ability" standard of that subsection when it failed to maintain point-of-sale records. J.A. 3048–49, 3058–59, 3062. For that reason, Commerce used the § 1677e(b) -based Jiangsu margin for Bosun (applying that margin to the imports covered by the administrative review). J.A. 3050, 3062–63.

Bosun challenged Commerce's remand redetermination at the Trade Court. One of Bosun's arguments was that the errors identified during the verification stage all fell into a circumscribed subset of the affiliates’ sales during the period of review, representing less than 2.5% of the total volume of such sales (by unit, not value). J.A. 3075–76. The Trade Court affirmed the Final Remand Redetermination, concluding that the failure to make a record of origin at the point of sale by the importer-affiliates, together with the errors identified by Commerce during verification, supported Commerce's invocation of both § 1677e(a) and § 1677e(b). DSMC II , 415 F. Supp. 3d at 1370–73.

The Trade Court entered a final judgment on December 16, 2019. Bosun timely appealed. We have jurisdiction under 28 U.S.C. § 1295(a)(5). II

A

Bosun challenges the Trade Court's decision in DSMC I insofar as it remanded the matter to Commerce for additional explanation. We reject this challenge.

"We review decisions of the Court of International Trade that remand decisions of the Commission for further explanation (based on an inability to evaluate on the basis of the record before the court) with the more deferential abuse-of-discretion standard." Diamond Sawblades Mfrs.’ Coalition v. United States , 612 F.3d 1348, 1356 (Fed. Cir. 2010). "In reviewing the trial court's discretion, this court examines its reasons for remand for any legal error." Id. at 1359 (internal quotation marks omitted). Remands are common, and they serve an important function—to ensure the adequacy of agency explanation that is crucial to judicial review, including review of whether substantial evidence exists for the premises of Commerce's exercise of discretion. See , e.g. , CP Kelco US, Inc. v. United States , 949 F.3d 1348, 1355 (Fed. Cir. 2020) (four remands by Trade Court for further explanation); Mid Continent Steel & Wire, Inc. v. United States , 941 F.3d 530, 537–38 (Fed. Cir. 2019). We see no abuse of discretion in the remand in the present matter.

The Trade Court in DSMC I expressed reasonable uncertainty about whether Commerce had properly considered the two "best of its ability" standards regarding a person's supply of information—the one in 19 U.S.C. § 1677e(b) ; and the one in 19 U.S.C. § 1677m(e) (a subsection referred to directly in § 1677e(a)(2)(B) and indirectly through the concluding phrase of § 1677e(a) ’s reference to § 1677m(d), which refers to § 1677m(e) ). For purposes of assessing the DSMC I remand, Bosun has shown no material legal error in the Trade Court's view of this court's precedents, which explain, while focusing on § 1677e(b), that the "best of [a person's] ability," in proper circumstances, may be tested by reference to the person's pre-questionnaire recordkeeping. See Peer Bearing , 766 F.3d at 1400 ; Nippon Steel , 337 F.3d at 1382. Nor has Bosun shown an abuse of discretion on the particular facts. Specifically, we see no abuse of discretion in the Trade Court decision to remand for a fuller explanation from Commerce of its initial judgment that the standards did not apply, even in part, to Bosun's recordkeeping, given that country of origin was not recorded at the point of sale to the first unaffiliated U.S. purchaser.

The Trade Court also reasonably sought additional explanation from Commerce about the ramifications of the errors Commerce identified in verifying Bosun's submissions. Noting Commerce's examination of four sales traces, the court stated: "Given the maximum sample size of four sales traces, Commerce's conclusion that the error [in Bosun's FIFO methodology] was ‘isolated’ and did not affect other sales is not sufficiently explained." DSMC I , 2018 WL 5281941, at *7. The Trade Court suggested that the errors all involved sales whose origin Bosun had used the FIFO step to identify, but that fact left a question about why Commerce had not applied § 1677e to any of the Bosun sales, not even the full subset of sales for which Bosun had relied on the FIFO step. The Trade Court reasonably concluded: "[E]ven if the FIFO step was applied only in the last resort, Commerce has yet to explain its conclusion that the error discovered at verification was not replicated in other sales, which were not reviewed at verification, to which the FIFO step applied." Id. at *8.

In short, the Trade Court's DSMC I remand to Commerce for further explanation was not an abuse of discretion. B

Bosun also challenges the Trade Court's affirmance, in DSMC II , of Commerce's Remand Redetermination. We agree in part with this challenge.

On this appeal from the Trade Court, we carefully consider that court's informed opinion, US Magnesium LLC v. United States , 839 F.3d 1023, 1027 (Fed. Cir. 2016) (citing Diamond Sawblades , 612 F.3d at 1356 ), but we must apply the same standard of review in considering the challenges to Commerce's actions as the standard that was applicable in the Trade Court, Apex Exports v. United States , 777 F.3d 1373, 1377 (Fed. Cir. 2015). For a final determination under 19 U.S.C. § 1675, we consider whether Commerce's decision is "unsupported by substantial evidence on the record, or otherwise not in accordance with law." 19 U.S.C. § 1516a(b)(1)(B)(i) ; see Ta Chen Stainless Steel Pipe, Inc. v. United States , 298 F.3d 1330, 1335 (Fed. Cir. 2002). We generally decide legal issues de novo, CP Kelco US , 949 F.3d at 1356 ; and here, there is no invocation of deference under Chevron U.S.A., Inc. v. Natural Resources Defense Council, Inc. , 467 U.S. 837, 104 S.Ct. 2778, 81 L.Ed.2d 694 (1984), and no legal issue we decide for which such deference would make a difference. We review factual determinations, including determinations of facts relevant to application of 19 U.S.C. § 1677e(a) and (b), for substantial-evidence support, which is "such relevant evidence as a reasonable mind might accept as adequate to support a conclusion," considering "the record as a whole, including evidence that supports as well as evidence that ‘fairly detracts from the substantiality of the evidence.’ " Nippon Steel , 337 F.3d at 1379 ; see also Universal Camera Corp. v. NLRB, 340 U.S. 474, 487–88, 71 S.Ct. 456, 95 L.Ed. 456 (1951) ; Consol. Edison Co. v. NLRB , 305 U.S. 197, 229, 59 S.Ct. 206, 83 L.Ed. 126 (1938) ; In re Gartside , 203 F.3d 1305, 1312 (Fed. Cir. 2000).

Bosun challenges Commerce's choice to disregard the entirety of its origin information, and to turn to "the facts otherwise available" for all of the period-of-review U.S. sales, as an application of § 1677e(a) not supported by substantial evidence. Bosun Opening Br. at 24–26, 42–46. We agree with Bosun's challenge in part. We address the preconditions to use of "the facts otherwise available" and then the premise for Commerce's decision to disregard all of Bosun's origin information as unreliable.

1

Under 19 U.S.C. § 1677e(a), Commerce "shall" resort to "the facts otherwise available" (subject to § 1677m(d) ) when any of five preconditions are met. The first precondition, stated by itself in paragraph (1), is not tied to the conduct of any interested party or other person: it is simply that "necessary information is not available on the record." Id. § 1677e(a)(1). The other four preconditions, stated in paragraph (2), are tied to the conduct of "an interested party or any other person": Each specifies conduct of a person that triggers the directive to Commerce to "use the facts otherwise available." Id. § 1677e(a)(2)(A)–(D). In its Remand Redetermination in this matter, Commerce invoked four of the five preconditions for such use. J.A. 3046. We find substantial-evidence support as to two of them.

The one that Commerce did not rely on is § 1677e(a)(2)(A), which requires that a person have "withh[e]ld[ ] information that has been requested" by Commerce. 19 U.S.C. § 1677e(a)(2)(A). Commerce did not find that Bosun withheld requested information.

Commerce found that § 1677e(a)(2)(C) applies. That provision applies here only if Bosun "significantly impede[d] the proceeding." Id. § 1677e(a)(2)(C). Although Commerce so found, that finding is not supported by substantial evidence. Commerce has not identified a withholding or misrepresentation of information that lengthened or otherwise impeded the proceeding. Cf. Ad Hoc Shrimp Trade Action Comm. v. United States , 802 F.3d 1339, 1355 (Fed. Cir. 2015). Indeed, Commerce has not identified any additional effort it had to expend because of Bosun's reporting method that it would not have expended if point-of-sales records had been kept. For example, Commerce did not find, and the record supplies no basis for finding, that Commerce would have accepted such records without verification under 19 U.S.C. § 1677m(i) or with a less burdensome verification effort than the one Commerce actually expended.

Commerce also found that § 1677e(a)(2)(B) applies here. For that provision to apply, the record must support a finding that Bosun "fail[ed] to provide [necessary] information by the deadlines for submission of the information or in the form and manner requested, subject to subsections (c)(1) and (e) of section 1677m of this title." 19 U.S.C. § 1677e(a)(2)(B). But the record does not contain substantial evidence to support such a finding.

Commerce did not find that Bosun missed a deadline in providing requested information. Cf. Dongtai Peak Honey Indus. Co. v. United States , 777 F.3d 1343, 1355–56 (Fed. Cir. 2015). Commerce did find that Bosun failed to provide information "in the form and manner requested," but that finding is not supported by substantial evidence. Commerce has not identified any language in its requests to Bosun that specified a particular form or manner of the country-of-origin information Bosun should submit. See generally J.A. 732, 1140–41, 2140. Commerce's initial questionnaire asked that Bosun provide "a chart for reporting the sales quantity and value," which Bosun provided. J.A. 65, 84–85. Commerce asked in questionnaire Section C that Bosun "prepare a separate computer data file containing each sale made during the [period of review]," which Bosun provided. J.A. 735, 784–91. The 2017 Verification Report addressing Bosun's information similarly recognized that Bosun provided the pre-selected "sales trace packages" Commerce requested, J.A. 2889, and also provided Microsoft Excel spreadsheets (i.e. , a "computer data file," as requested) that assisted Commerce's inquiry, J.A. 2890. Although the Trade Court spoke of Commerce having requested "direct" origin information, DSMC II , 415 F. Supp. 3d at 1371, Commerce has supplied no evidentiary support for that characterization.

For those reasons, essential requirements for applicability of § 1677e(a)(2)(B) are not met here. It is immaterial, in this circumstance, whether the "subject to subsections (c)(1) and (e) of section 1677m" phrase is met. That phrase merely obliges Commerce to consider excusing deadline or form-or-manner violations in certain situations. It has no application when there is no evidence-supported deadline or form-or-manner violation in the first place, as here.

Commerce found applicable two other triggers for the use of the facts otherwise available under § 1677e(a). It found that "necessary information is not available on the record," 19 U.S.C. § 1677e(a)(1), and, referring to information requested by Commerce, that Bosun "provide[d] such information but the information cannot be verified," id. § 1677e(a)(2)(D). Substantial evidence supports the finding as to the second and, a fortiori , as to the first. During verification Commerce found problems in several of Bosun's origin identifications; and putting to one side the important question of how much of Bosun's overall origin information those problems render unreliable, we think it clear that Commerce could reasonably find the problems sufficient to deem unreliable at least a portion of Bosun's information, i.e. , the portion resorting to the FIFO step for identifying the origin of particular U.S. sales. See J.A. 3049, 3061; see also DSMC II , 415 F. Supp. 3d at 1371. As to those sales, the evidence supports a finding that the information submitted could not be adequately verified and that, as a result, origin information about those sales was missing. Commerce therefore properly found § 1677e(a) to apply in this matter.

2

Having permissibly concluded that there were (limited) bases for applying the command to use "the facts otherwise available" under § 1677e(a), Commerce had to determine, under the "otherwise" language, which facts constituted "other[ ]" facts that had to be disregarded. Commerce ultimately concluded that it would disregard all of Bosun's origin information. J.A. 3049. The basis on which Commerce did so, however, leaves a significant question about substantial-evidence support, and the answer to that question seems consequential, because the record appears to suggest that there is no sufficient support for disregarding more than 2.5% of the U.S. sales of Bosun's affiliates.

Commerce did not decide that the "otherwise" phrase, without more, itself demands, or should be interpreted to demand, disregard of all information of any person whose conduct comes within one of the § 1677e(a) preconditions, even if the only applicable preconditions are § 1677e(a)(1) and (a)(2)(D) and most of the information that person supplied is verified and not otherwise soundly deemed unreliable. Nor did Commerce advance a categorical position of that sort when it relied on § 1677m(d), to which § 1677e(a) refers in its concluding phrase. Subsection 1677m(d) states that, in certain circumstances involving a person's submission attempting to cure an earlier failure to "comply with [an information] request," if Commerce finds the submission "not satisfactory," it "may, subject to subsection (e), disregard all or part of the original and subsequent responses." 19 U.S.C. § 1677m(d). We may assume arguendo that the provision applies here. When Commerce invoked the provision, by deeming Bosun's submissions unsatisfactory, it did not say that it was applying § 1677m(d) ’s "disregard" clause based on a policy decision to disregard all of the Bosun-supplied origin information no matter how limited the basis was for finding Bosun's information not satisfactory, i.e. , even if the reliability-undermining effect of any deficiency was cabined.

Commerce likewise asserted no such position based on § 1677m(e). That provision, which does not directly apply here because there is no supported finding under § 1677e(a)(2)(B), is only of indirect relevance to § 1677e(a) in this matter based on § 1677e(a) ’s statement that Commerce's use of the facts otherwise available is "subject to section 1677m(d)"—which itself authorizes disregard of information only "subject to [ § 1677m(e) ]." Section 1677m(e) on its face is only a requirement that Commerce sometimes use party-supplied information that, in the absence of the requirement, Commerce could or must disregard; its language is not a directive to expand the amount of party-supplied information Commerce must disregard. Regardless, Commerce did not, under this clause, adopt a position that it would disregard all reliable information submitted by a person that includes some unverifiable information within § 1677e(a)(2)(D), whenever the person did not act to the best of its ability in keeping records.

Such rationales, had Commerce adopted them, would raise serious questions in a case like this one, where only § 1677e(a)(1) and (a)(2)(D) undergird application of § 1677e(a), about conformity to the statutory policies that must guide any agency's exercise of discretion. See , e.g. , Judulang v. Holder , 565 U.S. 42, 55, 132 S.Ct. 476, 181 L.Ed.2d 449 (2011) (discussing "may" authority in 8 U.S.C. § 1182(c) (1994) ). We have explained that "[a]n overriding purpose of Commerce's administration of antidumping laws is to calculate dumping margins as accurately as possible." Yangzhou Bestpak Gifts & Crafts Co. v. United States , 716 F.3d 1370, 1379 (Fed. Cir. 2013) (citing Rhone Poulenc, Inc. v. United States, 899 F.2d 1185, 1191 (Fed. Cir. 1990) ); see also Mid Continent Steel , 941 F.3d at 542. More particularly, we have often focused, when applying § 1677e(a), on the subsection's role as a command to Commerce "to fill a gap in the record" when information is missing or compromised. Zhejiang DunAn Hetian Metal Co., Ltd. v. United States , 652 F.3d 1333, 1348 (Fed. Cir. 2011) ; see also Nippon Steel , 337 F.3d at 1381. To the extent that information supplied is reliable, i.e. , not in fact tainted by its supplier's conduct, "gap" seems an inapt characterization. And the authoritative Statement of Administrative Action, see 19 U.S.C. § 3512(d), states that subsection 1677e(a) pertains to situations "where requested information is missing from the record or cannot be used because, for example, it has not been provided, it was provided late, or Commerce could not verify the information," and when needed information is missing, Commerce "must make [its] determinations based on all evidence of record, weighing the record evidence to determine that which is most probative of the issue under consideration," H.R. Doc. No. 103-316, vol. 1, at 869 (1994), reprinted in 1994 U.S.C.C.A.N. 4040, 4179. That explanation, on its own, suggests an information-specific consideration of probativeness rather than any blanket disregard of all information supplied by a person whenever some of the information supplied by that person is unreliable.

Notably, this is not a case involving withholding of information, failure to meet timing, form, or manner requirements, or significant impeding of a proceeding, under § 1677e(a)(2)(A), (B), and (C). Such situations implicate a policy of cooperation with Commerce that is evident on the face of those statutory provisions, as it is evident on the face of § 1677e(b). The relevance of such a policy is not as facially evident at the § 1677e(a) stage where, as here, applying § 1677e(a) involves only missing or unverifiable information, under § 1677e(a)(1) and (a)(2)(D). We need not go further than note the facial difference between these and the other preconditions for application of § 1677e(a). In particular, we do not pursue a full statutory analysis or, therefore, conclude that Commerce is statutorily precluded from doing more than filling in gaps in reliable information when applying § 1677e(a) even when the only preconditions are § 1677e(a)(1) and (a)(2)(D).

We need not confront questions raised about a blanket policy of that sort because Commerce did not announce (or therefore explain) such a policy. Instead, Commerce justified its disregard of all of Bosun's information based on its determination that the defects in Bosun's origin-identifying methodology left Commerce with "no reliable information on the country of origin of Bosuns sales." J.A. 3049. That premise asserts a reliability problem with all the Bosun information. We assess Commerce's decision to disregard all of Bosun's information on the basis Commerce gave for that decision. See SEC v. Chenery Corp. , 332 U.S. 194, 196, 67 S.Ct. 1760, 91 L.Ed. 1995 (1947).

We conclude that Commerce has not satisfactorily explained why substantial evidence supports its determination of unreliability of all of Bosun's origin information. Commerce has not explained why, as a general matter, records other than point-of-sale records are categorically less reliable than point-of-sale records (both of which may require verification)—or, therefore, why the entirety of Bosun's three-step origin-identification process is "not satisfactory" just because it does not involve point-of-sale records. Nor has Commerce identified a methodological problem with the first two steps of Bosun's identification process. See J.A. 2891–92. And neither in its Remand Redetermination decision, e.g. , J.A. 3049, 3060, nor in its brief in this court, has Commerce provided a comprehensible explanation for why, if so, the errors found in Bosun's submissions have a reliability-undermining effect outside the category of sales to unaffiliated U.S. purchases whose origin Bosun identified through the FIFO inference. The language used by Commerce, especially at J.A. 3049 & n.37, can easily be understood as limited to the FIFO-inference step.

This deficiency in Commerce's explanation appears to matter considerably to the outcome of this proceeding. Bosun has argued that any absence of or taint on origin information lies entirely within the category of sales for which Bosun relied on the FIFO inference—a category that Bosun asserts, without apparent contradiction, involves less than 2.5% of the sales during the period of review. Bosun Opening Br. at 38–39. The government's evidentiary argument for a broader gap or taint is distinctly limited. In its Remand Redetermination, Commerce noted some problems identified during verification, seemingly limited to the FIFO-inference step, J.A. 3049, and it stated, in a footnote, "that Bosun's errors in reporting physical characteristics ‘affected three out of sixteen transactions identified at verification, and related to multiple product characteristics,’ " id. at 3049 n.37 (quoting DSMC I , 2018 WL 5281941, at *8 ). Whether that statement even refers to an effect beyond the FIFO-inference step is not apparent; still less clear is an evidentiary basis for a finding to that effect. The text of the paragraph in which the footnote appears suggests that any taint is confined to the FIFO-inference category of sales. See id. at 3049 ("For this reason, we now find that, based on our verification process, Bosun's supplemental responses explaining the FIFO methodology were not satisfactory, because Bosun could have maintained adequate country of origin information for its products in the first place.").

We are not persuaded, on the briefing and other materials presented to us, that there is a supported basis for finding the Bosun-supplied information unreliable outside the category of sales for which origin was identified using only the FIFO-inference step (rather than the two earlier steps). But we also are not confident that there is no such basis. We think that a remand is advisable for the parties to address this focused issue, which may have substantial consequences for the bottom-line result of the proceeding.

3

Neither Commerce nor the Trade Court misinterpreted our holdings in Nippon Steel or Peer Bearing regarding the "best of its ability" standard of § 1677e(b). Before applying that standard, however, the appropriate threshold determination under § 1677e(a) requires that Commerce determine what are "the facts otherwise available." It is only for "selecting from among the facts otherwise available," as properly determined under § 1677e(a), that § 1677e(b) authorizes Commerce to use an adverse inference. 19 U.S.C. § 1677e(b) ; see Zhejiang DunAn , 652 F.3d at 1346 ("As these two subsections make clear, Commerce first must determine that it is proper to use facts otherwise available before it may apply an adverse inference.").

Nor did Commerce or the Trade Court misinterpret the governing precedent that, on the facts that properly come within § 1677e(a) and hence § 1677e(b), the "inference" that Commerce "may use" in "selecting from among the facts otherwise available" must "be a reasonably accurate estimate of the respondent's actual rate, albeit with some built-in increase intended as a deterrent to non-compliance." F.lli de Cecco di Filippo Fara S. Martino S.p.A. v. United States , 216 F.3d 1027, 1032 (Fed. Cir. 2000) ("[T]he purpose of section 1677e(b) is to provide respondents with an incentive to cooperate, not to impose punitive, aberrational, or uncorroborated margins."); see also Essar Steel Ltd. v. United States , 678 F.3d 1268, 1276 (Fed. Cir. 2012) ; Gallant Ocean (Thailand) Co. v. United States , 602 F.3d 1319, 1324 (Fed. Cir. 2010). If, on remand, it is determined that, as currently appears, there is no basis for Commerce to disregard under § 1677e(a) the Bosun-supplied origin information for the sales to unaffiliated U.S. customers during the period of review outside the category of sales analyzed via the FIFO methodology, a redetermination of how § 1677e(b) applies to this matter will be needed.

III

For the foregoing reasons, we affirm in part, reverse in part, and vacate in part, and we remand for further proceedings in accordance with this opinion.

The parties shall bear their own costs.

AFFIRMED IN PART, REVERSED IN PART, VACATED IN PART, AND REMANDED