Goodluck India Limited v. United States

11 F.4th 1335

• Court: Court of Appeals for the Federal Circuit
• Decided: August 31, 2021
• Docket: 20-2017
• Status: Published

Opinion

Case: 20-2017 Document: 61 Page: 1 Filed: 08/31/2021

United States Court of Appeals for the Federal Circuit ______________________

GOODLUCK INDIA LIMITED, Plaintiff-Appellee

v.

UNITED STATES, Defendant

ARCELORMITTAL TUBULAR PRODUCTS, MICHIGAN SEAMLESS TUBE, LLC, PLYMOUTH TUBE CO. USA, PTC ALLIANCE CORP., WEBCO INDUSTRIES, INC., ZEKELMAN INDUSTRIES, INC., Defendants-Appellants ______________________

2020-2017 ______________________

Appeal from the United States Court of International Trade in No. 1:18-cv-00162-GSK, Judge Gary S. Katzmann. ______________________

Decided: August 31, 2021 ______________________

NED H. MARSHAK, Grunfeld, Desiderio, Lebowitz, Sil- verman & Klestadt LLP, New York, NY, argued for plain- tiff-appellee. Also represented by BRUCE M. MITCHELL; MICHAEL SCOTT HOLTON, JORDAN CHARLES KAHN, Wash- ington, DC. Case: 20-2017 Document: 61 Page: 2 Filed: 08/31/2021

ROBERT ALAN LUBERDA, Kelley Drye & Warren, LLP, Washington, DC, argued for defendants-appellants. Also represented by MELISSA M. BREWER, DAVID C. SMITH, JR. ______________________

Before REYNA, CLEVENGER, and STOLL, Circuit Judges. REYNA, Circuit Judge. Defendants-Appellants appeal the judgment of the United States Court of International Trade affirming a re- mand determination by the United States Department of Commerce in an antidumping duty investigation on U.S. imports of cold-drawn mechanical tubing from India. In the underlying investigation, Commerce rejected Plaintiff- Appellee Goodluck India’s submission of supplemental data and relied on “adverse facts available” under 19 U.S.C. § 1677e(b) for its less-than-fair-value analysis, which resulted in an antidumping margin of 33.8% ad val- orem applicable to Goodluck India’s imports of mechanical tubing. Goodluck India appealed to the Court of Interna- tional Trade, arguing that its submission was a permissible correction of a minor clerical error and that it was entitled to submit supplemental information up to the day of veri- fication. The Court of International Trade agreed with Goodluck India and remanded to Commerce. Commerce, under protest, conducted a new less-than-fair-value analy- sis resulting in a zero-percent antidumping margin for Goodluck India. Defendants-Appellants challenged the re- mand determination, but the Court of International Trade affirmed. Defendants-Appellants now appeal to this court. We hold that Commerce’s initial determination—re- jecting Goodluck India’s supplemental submission on grounds that it constituted new factual information and not a minor or clerical correction of the record, and that the submission was unverifiable as it was submitted on the eve Case: 20-2017 Document: 61 Page: 3 Filed: 08/31/2021

GOODLUCK INDIA LIMITED v. US 3

of verification—is supported by substantial evidence and not otherwise contrary to law. We reverse. BACKGROUND Antidumping Duty Investigation Defendants-Appellants ArcelorMittal Tubular Prod- ucts, Michigan Seamless Tube, LLC, Plymouth Tube Com- pany USA, PTC Alliance Corp., Webco Industries, Inc., and Zekelman Industries, Inc. (together, “Petitioners”) are U.S. domestic producers of cold-drawn mechanical tubing made from carbon and alloy steel (“mechanical tubing”). J.A. 58. In basic terms, mechanical tubing is metal piping sold in various diameters, lengths, and thicknesses suited for var- ious mechanical applications. See generally J.A. 559. On April 19, 2017, Petitioners filed an antidumping duty petition with the United States Department of Com- merce (“Commerce”) on imports of mechanical tubing. J.A. 58. On May 9, 2017, Commerce initiated an anti- dumping duty investigation on mechanical tubing from several countries, including India. See id. (Certain Cold- Drawn Mechanical Tubing of Carbon and Alloy Steel from the Federal Republic of Germany, India, Italy, the Republic of Korea, the People’s Republic of China, and Switzerland: Initiation of Less-Than-Fair-Value Investigations, 82 Fed. Reg. 22,491 (May 16, 2017)). Plaintiff-Appellee Goodluck India Ltd. (“Goodluck”) manufactures mechanical tubing in India, which it sells in both India and the United States. J.A. 2656–64. On June 19, 2017, Commerce selected Goodluck as a respond- ent in the investigation and issued Goodluck a mandatory questionnaire. J.A. 103–213. Relevant here, the question- naire solicited data regarding Goodluck’s sales of mechan- ical tubing in its home market, India (Section B), its sales of mechanical tubing in the United States (Section C), and cost data specific to each product (Section D) applicable during the period of investigation (April 1, 2016, to Case: 20-2017 Document: 61 Page: 4 Filed: 08/31/2021

March 31, 2017). Id.; see also J.A. 58 (defining period of investigation). The questionnaire required Goodluck to create a con- trol number (“CONNUM”) 1 for each product identified in its submitted sales and cost databases. J.A. 142, 169. It also required Goodluck to use the same CONNUM for any “products with identical physical characteristics reported” across all of its submitted database files. Id. When Commerce issued the questionnaire, it had not yet determined which physical characteristics would com- prise the CONNUMs, so it left those fields blank in its in- structions. See id. On July 6, 2017, Commerce issued a letter filling in those blanks, identifying which product characteristics should be used in forming each CONNUM. J.A. 440–52 (Letter from the Department, CONNUM Let- ter, dated July 6, 2017 (“July Letter”)). 2 The July Letter directed Goodluck to report, among other things, wall thickness information for its products in two questionnaire fields—Fields 2.5 and 3.5. Id. Field 2.5

1 See Union Steel v. United States, 823 F. Supp. 2d 1346, 1349–50 (Ct. Int’l Trade 2012) (“A ‘CONNUM’ is a contraction of the term ‘control number,’ and is simply Commerce[’s term] for a unique product (defined in terms of a hierarchy of specified physical characteristics deter- mined in each antidumping proceeding). All products whose product hierarchy characteristics are identical are deemed to be part of the same CONNUM and are regarded as ‘identical’ merchandise for purposes of the price compar- ison. The hierarchy of product characteristics defining a unique CONNUM varies from case to case depending on the nature of the merchandise under investigation.” (cita- tion omitted)). 2 The July Letter was withdrawn and reissued the following day due to an error. J.A. 439. Case: 20-2017 Document: 61 Page: 5 Filed: 08/31/2021

GOODLUCK INDIA LIMITED v. US 5

called for a nominal wall thickness value, in millimeters. J.A. 443. Field 3.5, in turn, called for a two-digit code that corresponded to a range within which the nominal wall thickness fell. J.A. 450. The July Letter set forth nine codes to be used in Field 3.5 as follows:

Id. Thus, for example, if a product had a nominal wall thickness of 1.5 mm, then Goodluck was meant to enter “1.5 mm” in Field 2.5 and “03” in Field 3.5. J.A. 443, 450. Shortly after Commerce issued the July Letter, Peti- tioners wrote to Commerce arguing that the ranges set forth in the July Letter were too broad to accurately cap- ture cost and expense differences. J.A. 457 (Letter to the Department, re: Petitioners’ Comments on the Department’s Release of Product Matching Criteria and Request for Ex- pansion of Certain Criteria Fields, dated July 12, 2017). In particular, Petitioners asked Commerce to create more ranges for use in Field 3.5 of the questionnaire. J.A. 459–60. Interested parties, including Goodluck, were invited to comment on or rebut Petitioners’ request for more ranges and the need for more particularized wall thickness information. Goodluck did not comment or raise any rebuttal. J.A. 566 (Letter to All Interested Parties, re: Revised Product Characteristics, dated August 7, 2017 (“August Letter”)). Case: 20-2017 Document: 61 Page: 6 Filed: 08/31/2021

On August 7, 2017, Commerce issued another letter with an updated coding chart that included fourteen ranges (instead of nine) for use in Field 3.5, as follows:

J.A. 566, 577. On August 25, 2017, Goodluck submitted its initial re- sponses to questionnaire Sections B–D. 3 J.A. 585. In those responses, Goodluck confirmed that it reported wall thick- ness codes according to the fourteen ranges specified in Commerce’s August Letter. J.A. 613. On November 15, 2017, Commerce issued a Prelimi- nary Determination, tentatively assigning Goodluck a zero- percent antidumping duty rate based on its questionnaire responses. J.A. 1810–14 (Certain Cold-Drawn Mechanical Tubing of Carbon and Alloy Steel From India: Preliminary

3 Though not relevant for purposes of this appeal, Goodluck submitted its initial Section A response in July 2017 and its supplemental Section A responses in September 2017. J.A. 525–65, 580–84, 1208–28. Goodluck also submitted responses to supplemental Sections B–D in October 2017. J.A. 1275–76, 1320–21. Case: 20-2017 Document: 61 Page: 7 Filed: 08/31/2021

GOODLUCK INDIA LIMITED v. US 7

Affirmative Determination of Sales at Less Than Fair Value, in Part, Postponement of Final Determination, and Extension of Provisional Measures, 82 Fed. Reg. 55,567 (Nov. 22, 2017)). 4 On November 22, 2017, Commerce sent Goodluck a Sales Verification Agenda outlining the plan for verifying the data in Goodluck’s responses to questionnaire Sec- tions B and C. J.A. 1816. Per standard procedure, Com- merce warned Goodluck that new information would only be accepted at verification if “(1) the need for that infor- mation was not evident previously; (2) the information makes minor corrections to information already on the rec- ord; or (3) the information corroborates, supports, or clari- fies information already on the record.” Id. (emphasis added). On November 27, 2017, Commerce sent Goodluck a Cost Verification Agenda, outlining the plan for verifying the data in Goodluck’s Section D responses. J.A. 1833–34. Again, Commerce warned Goodluck that only the correc- tion of “minor errors” would be allowed at verification. J.A. 1834. Commerce also specified, “Minor errors are mi- nor mistakes in addition, subtraction, or other arithmetic function, minor data entry mistakes, clerical errors result- ing from inaccurate copying, duplication, or the like, and minor classification errors. Minor errors do not include items such as methodology changes.” J.A. 1836 n.1. On December 14, 2017, the first day of cost verification, Goodluck wrote to Commerce to identify and correct

4 On January 3, 2018, Commerce revised Goodluck’s rate to 4.2% based on a clerical error that Petitioners flagged. J.A. 1787–92, 2847–49 (Certain Cold-Drawn Me- chanical Tubing of Carbon and Alloy Steel From India: Amended Preliminary Determination of Sales at Less Than Fair Value, 83 Fed. Reg. 1021 (Jan. 9, 2018)). Case: 20-2017 Document: 61 Page: 8 Filed: 08/31/2021

682 misreported values in its Section B database, calling them “minor corrections.” J.A. 1867–68 (Letter to the De- partment, Goodluck Verification Minor Corrections, dated December 14, 2017); J.A. 1873–74 (identifying Goodluck’s coding errors); J.A. 2655 (counting 682 sales affected by misreported CONNUMs). Specifically, Goodluck explained that it prepared its Section B, Field 3.5 responses using the nine wall thickness codes provided in the July Letter, not the fourteen codes provided in the August Letter. J.A. 1873–74. Goodluck also explained that this mistake resulted in errors in its Section D database, which relied on CONNUMs already created for the Section B responses. Id. On January 17, 2018, Commerce issued its Cost Verifi- cation Report, which acknowledged Goodluck’s coding er- rors but noted that “[c]orrections of these errors would cause changes to the reported physical characteristics of 24 CONNUMs and the addition of 13 CONNUMs.” J.A. 2633. On February 7, 2018, Commerce issued its Sales Verification Report, which further noted: “[T]he values in Field 3.5 do not correspond to the Field 2.5 Nominal Wall Thickness. As a result, 682 [values] in the home database are affected by this issue.” J.A. 2655. On February 15, 2018, Goodluck filed its administra- tive case brief in Commerce’s investigation. 5 J.A. 2770–72. Commerce rejected that case brief, however, because it found that the brief contained new factual information— e.g., “corrected worksheets” and a new database reflecting changes in Goodluck’s reporting of wall thicknesses. J.A. 2725–26 (Letter from the Department, re: Rejection of New Factual Information, dated Feb. 20, 2018). Commerce

5 Petitioners also submitted their administrative case brief on February 15, 2018. J.A. 2689–712. Case: 20-2017 Document: 61 Page: 9 Filed: 08/31/2021

GOODLUCK INDIA LIMITED v. US 9

instructed the parties not to reference this rejected mate- rial in their rebuttal briefs. J.A. 2726. 6 On February 21, 2018, Goodluck submitted a redacted case brief that was similarly rejected on March 7, 2018, again for containing new factual information. J.A. 2764–65. On March 8, 2018, Goodluck submitted a second redacted case brief, removing some language re- garding the missing CONNUMs and wall thickness errors. J.A. 2766–67. Commerce’s Final Determination On April 16, 2018, Commerce published its Final De- termination, assigning Goodluck an antidumping duty rate of 33.8%. J.A. 2835–37 (Certain Cold-Drawn Mechanical Tubing of Carbon and Alloy Steel from India: Final Affirm- ative Determination of Sales at Less Than Fair Value, 83 Fed. Reg. 16,296 (Apr. 16, 2018) (“Final Determina- tion”)); see also J.A. 2805–25 (Issues and Decision Memo- randum for the [Final Determination], dated Apr. 9, 2018). Commerce found that Goodluck’s 682 misreported val- ues, which resulted in 24 misreported CONNUMs and 13 unreported CONNUMs, were not due to “clerical” or “minor” errors. See J.A. 2655–56, 2812. Commerce rea- soned that Goodluck’s mistakes “entailed more than copy- ing, duplicating, or the like,” as coding wall thicknesses entailed “analyzing the nominal tube wall thickness[es] and assigning the corresponding codes as directed by Com- merce.” J.A. 2812. Commerce also noted that “this sys- temic error render[ed] the entire dumping calculation inaccurate[] because the control number is fundamental to Commerce’s calculation, as it controls the allocation of costs and determines the product matches between U.S. and home markets.” J.A. 2811. Moreover, Commerce

6 The parties submitted their rebuttal briefs on Feb- ruary 23, 2018. J.A. 2728–51, 2753–63. Case: 20-2017 Document: 61 Page: 10 Filed: 08/31/2021

found that “any attempts to correct these errors would in- volve both extensive SAS programming and complex calcu- lations to Goodluck’s cost database,” and it was “impossible to assess whether such a large-scale revision [wa]s appro- priate.” J.A. 2816. Thus, Commerce determined that “Goodluck’s cost and home market sales databases [we]re unreliable” for calculating an estimated dumping margin. J.A. 2809–10. Additionally, Commerce found that Goodluck “failed to cooperate by not acting to the best of its ability to comply with Commerce’s requests for information” because “the scope of the errors and omissions identified at verifica- tion . . . [we]re the result of both inattentiveness and care- lessness.” J.A. 2817; see Nippon Steel Corp. v. United States, 337 F.3d 1373, 1382–83 (Fed. Cir. 2003) (“While the [‘best of ability’] standard does not require perfection, and recognizes that mistakes sometimes occur, it does not con- done inattentiveness, carelessness, or inadequate record keeping.”). Consequently, Commerce based its margin cal- culation for Goodluck on all facts available, using an ad- verse inference. J.A. 2818; see 19 U.S.C. § 1677e. Commerce therefore relied on the “highest dumping mar- gin contained in the petition” and assigned Goodluck an antidumping duty rate of 33.8%. J.A. 2818. CIT Action Goodluck challenged Commerce’s Final Determination before the United States Court of International Trade (“CIT”). J.A. 39; see Goodluck India Ltd. v. United States, 393 F. Supp. 3d 1352, 1361 (Ct. Int’l Trade 2019) (“Good- luck I”). The CIT reversed, finding that Commerce abused its discretion by not accepting Goodluck’s corrected infor- mation, as Goodluck’s coding errors “could have been ad- dressed through a ‘straightforward mathematical adjustment’ even though the ‘effect of these mistakes was compounded’ by how Commerce used the incorrect CON- NUMs.” Goodluck I, 393 F. Supp. 3d at 1364 (quoting NTN Case: 20-2017 Document: 61 Page: 11 Filed: 08/31/2021

GOODLUCK INDIA LIMITED v. US 11

Bearing Corp. v. United States, 74 F.3d 1205 (Fed. Cir. 1995)). The CIT also noted: “[T]his is not a situation where a company was unresponsive, provided fraudulent infor- mation, or clearly ignored Commerce’s instructions; rather, Goodluck believed it had reported the correct information in accordance with Commerce’s instructions—and largely did so—but made a mistake.” Id. at 1366 n.11. Reasoning further that “[c]lerical errors are by their nature not errors in judgment but merely inadvertencies,” the CIT found that Goodluck’s coding errors were clerical. See id. at 1368 (quoting NTN Bearing, 74 F.3d at 1208). Thus, the CIT held that Goodluck’s coding errors were “correctible im- porter mistake as opposed to untimely new factual infor- mation” and remanded for Commerce to consider the corrected information. Id. at 1370. 7 On December 23, 2019, Commerce issued its Final Re- sults of Redetermination Pursuant to Court Remand (“Re- mand Results”) under respectful protest, assigning Goodluck a revised antidumping duty rate of 0%. J.A. 3384. Petitioners challenged that determination be- fore the CIT, and the CIT sustained the Remand Results. J.A. 27–30; see Goodluck India Ltd. v. United States, 439 F. Supp. 3d 1366, 1367 (Ct. Int’l Trade 2020) (“Goodluck II”). Now, Petitioners appeal to this court. We have jurisdiction pursuant to 28 U.S.C. § 1295(a)(5). STANDARD OF REVIEW We review decisions by the CIT de novo, reviewing final determinations by Commerce under the same standard

7 The CIT also directed Commerce to “explain why it . . . departed from its general practice for calculating cash deposit offset rates in this case.” Goodluck I, 393 F. Supp. 3d at 1363. Goodluck does not raise this issue on appeal. Case: 20-2017 Document: 61 Page: 12 Filed: 08/31/2021

applied by the CIT. ABB, Inc. v. United States, 920 F.3d 811, 820 (Fed. Cir. 2019). Accordingly, we affirm Com- merce’s rulings unless they are “unsupported by substan- tial evidence on the record, or otherwise not in accordance with law.” Id.; 19 U.S.C. § 1516a(b)(1)(B)(i). We consider whether “the administrative record contain[s] substantial evidence to support” Commerce’s decision and whether that decision was “rational.” Matsushita Elec. Indus. Co. v. United States, 750 F.2d 927, 933 (Fed. Cir. 1984). DISCUSSION Petitioners argue that (1) the CIT improperly substi- tuted its judgment to find that Goodluck’s submitted cor- rections were not “minor”; (2) substantial evidence supports Commerce’s finding that Goodluck’s submitted corrections were not “minor”; and (3) the CIT legally erred by relying on NTN Bearing. For reasons stated below, we reverse. Commerce has discretion to accept or reject corrective information on a case-by-case basis. Deacero S.A.P.I. de C.V. v. United States, 353 F. Supp. 3d 1303, 1307 (Ct. Int’l Trade 2018) (citing Timken U.S. Corp. v. United States, 434 F.3d 1345, 1353 (Fed. Cir. 2006)). For example, as in investigations, Commerce has discretion to establish and enforce time limits for submitting information in an admin- istrative review. Reiner Brach GmbH & Co. KG v. United States, 206 F. Supp. 2d 1323, 1334 (Ct. Int’l Trade 2002) (“Commerce clearly cannot complete its own work unless it is able at some point to freeze the record and make calcu- lations and findings based on that fixed and certain body of information.” (internal quotation marks and citation omit- ted)); see also NTN Bearing, 74 F.3d at 1207 (“[I]t is within the discretion of [Commerce] to promulgate appropriate procedural regulations.”). Commerce’s discretion in estab- lishing and enforcing its procedures, in particular the cor- rection of clerical errors, is grounded in the trade statute, 19 U.S.C. § 1673d(e) (“The administering authority shall Case: 20-2017 Document: 61 Page: 13 Filed: 08/31/2021

GOODLUCK INDIA LIMITED v. US 13

establish procedures for the correction of ministerial errors in final determinations within a reasonable time after the determinations are issued . . . .”). Commerce’s discretion has limits. See Goodluck I, 393 F. Supp. 3d at 1358; see also Borlem S.A.-Empreedi- mentos Industriais v. United States, 913 F.2d 933, 937 (Fed. Cir. 1990) (“Congress’ desire for speedy determina- tions on dumping matters should not be interpreted as au- thorizing proceedings that are based on inaccurate data.”). Commerce abuses its discretion, for instance, if it departs from a consistent practice without reasonable explanation. See Consol. Bearings Co. v. United States, 348 F.3d 997, 1007 (Fed. Cir. 2003). Commerce can also abuse its discre- tion by “refusing to accept updated data when there [i]s plenty of time for Commerce to verify or consider it.” Pa- pierfabrik August Koehler SE v. United States, 843 F.3d 1373, 1384 (Fed. Cir. 2016) (collecting cases); see, e.g., NTN Bearing, 74 F.3d at 1207–08 (holding Commerce abused its discretion by refusing, at the preliminary results stage, to accept information correcting reporting errors); Fischer S.A. Comercio v. United States, 700 F. Supp. 2d 1364, 1370–71 (Ct. Int’l Trade 2010) (same); Timken, 434 F.3d at 1353 (explaining that “Commerce is free to correct any type of importer error” if the request is timely and justi- fied). Relevant here, the untimely submission of corrective information at verification results in “a tension between fi- nality and correct result.” Timken, 434 F.3d at 1353 (citing NTN Bearing, 74 F.3d at 1208). In such a case, Commerce must determine whether the need for finality outweighs the need for accuracy, or vice versa. See generally Civ. Aer- onautics Bd. v. Delta Air Lines, Inc., 367 U.S. 316, 321–22 (1961) (“Since these policies are in tension, it is necessary to reach a compromise in each case . . . .”). To that end, Commerce’s typical practice is to accept corrective infor- mation at verification only for “minor corrections to infor- mation already on the record.” J.A. 1816; see 19 C.F.R. Case: 20-2017 Document: 61 Page: 14 Filed: 08/31/2021

§§ 351.301, 351.302(d)(1)(i); 19 U.S.C. § 1673d(e). A minor correction is one that rectifies “minor mistakes in addition, subtraction, or other arithmetic function, minor data entry mistakes, clerical errors resulting from inaccurate copying, duplication, or the like, [or] minor classification errors.” J.A. 1836 n.1 (“Minor errors do not include items such as methodology changes.”); see also 19 C.F.R. § 351.224(f). This practice, as applied at verification, strikes an appro- priate balance between finality and accuracy. And, im- portantly, it is within Commerce’s discretion to decide which interest outweighs the other on a case-by-case basis. See Micron Tech., Inc. v. United States, 117 F.3d 1386, 1396 (Fed. Cir. 1997) (“Congress has implicitly delegated to Commerce the latitude to derive verification procedures ad hoc.”); Am. Alloys, Inc. v. United States, 30 F.3d 1469, 1475 (Fed. Cir. 1994) (“[T]he statute[s] give[] Commerce wide latitude in its verification procedures.”). We hold that Commerce acted within its discretion in rejecting Goodluck’s revised submissions on the day of ver- ification because substantial evidence supports Com- merce’s determination that Goodluck’s revisions were not minor. Goodluck’s revisions were a systemic change to the entire reported database. The revisions were not singular, such as a missing word or an error in arithmetic. The rec- ord reflects that Goodluck’s coding errors resulted in 24 misreported CONNUMs and 13 unreported CONNUMs, thereby resulting in misreported CONNUMs for 682 sales in Goodluck’s home market database. J.A. 2810. It also appears generally undisputed that Goodluck’s errors “ren- der[ed] the entire dumping calculation inaccurate[] be- cause the control number is fundamental to Commerce’s calculation, as it controls the allocation of costs and deter- mines the product matches between U.S. and home mar- kets.” J.A. 2811. It was therefore rational for Commerce to find that “any attempts to correct these errors would in- volve both extensive SAS programming and complex Case: 20-2017 Document: 61 Page: 15 Filed: 08/31/2021

GOODLUCK INDIA LIMITED v. US 15

calculations to Goodluck’s cost database.” J.A. 2816. Such corrections are not “minor.” See J.A. 2812. The record belies Goodluck’s argument that it should be excused because it acted to the best of its ability. Sub- stantial evidence supports Commerce’s finding that Good- luck “failed to cooperate by not acting to the best of its ability to comply with Commerce’s requests for infor- mation.” J.A. 2817. Despite receiving clear instructions in the August Letter, Goodluck failed to update its Field 3.5 responses to reflect the new coding ranges ordered by Com- merce. Notably, Goodluck was aware of the August Letter instructions, as Goodluck represented to Commerce that it coded wall thicknesses according to the fourteen ranges set forth in that letter. J.A. 613. This evidence supports Com- merce’s conclusion that Goodluck’s errors were “the result of both inattentiveness and carelessness.” J.A. 2817. Thus, Commerce did not abuse its discretion in applying all facts available with an adverse inference. See Nippon Steel, 337 F.3d at 1380–84; 19 U.S.C. § 1677e(b). The cases relied on by Goodluck do not compel a differ- ent outcome. Those cases—namely, NTN Bearing, Fischer, and Timken—all stand for the proposition that Commerce cannot reject corrective information at a preliminary deter- mination stage (where there are no finality concerns), pro- vided that the corrections are otherwise justifiably necessary. See Papierfabrik, 843 F.3d at 1384 (discussing NTN Bearing, 74 F.3d at 1207–08, and Timken, 434 F.3d at 1353). Here, in contrast, Goodluck submitted its revised databases at verification. Verification represents a point of no return. The purpose of verification is “to test infor- mation provided by a party for accuracy and completeness.” Micron Tech., 117 F.3d at 1396 (quoting Bomont Indus. v. United States, 733 F. Supp. 1507, 1508 (Ct. Int’l Trade 1990)). At that stage, Commerce enjoys “broad discretion” to promulgate and enforce its procedural rules. Stupp Corp. v. United States, 5 F.4th 1341, 1350–51 (Fed. Cir. 2021) (“Short of a showing that Commerce’s enforcement of Case: 20-2017 Document: 61 Page: 16 Filed: 08/31/2021

its procedural rules is so haphazard or unreasonable as to be arbitrary or capricious[,] . . . Commerce’s failure to ap- ply those rules with Procrustean consistency in every case does not deprive it of the authority to enforce those rules in any case.”); see also 19 U.S.C. §§ 1673d(e), 1677e; 19 C.F.R. §§ 351.301, 351.302(d)(1)(i). Lastly, we agree with Petitioners that the CIT improp- erly substituted its judgment for that of Commerce when it determined, for instance, that Goodluck’s reporting errors “could have been addressed through a straightforward mathematical adjustment.” Goodluck I, 393 F. Supp. 3d at 1364 (internal quotation marks and citation omitted). Under the substantial evidence standard of review, a re- viewing court “must affirm [Commerce’s] determination if it is reasonable and supported by the record as a whole, even if some evidence detracts from [Commerce’s] conclu- sion.” Hitachi Metals, Ltd. v. United States, 949 F.3d 710, 716 (Fed. Cir. 2020) (quoting Altx, Inc. v. United States, 370 F.3d 1108, 1121 (Fed. Cir. 2004)). “Even if it is possible to draw two inconsistent conclusions from evidence in the record, such a possibility does not prevent Commerce’s de- termination from being supported by substantial evi- dence.” Am. Silicon Techs. v. United States, 261 F.3d 1371, 1376 (Fed. Cir. 2001). In this case, the CIT noted: “[Goodluck] necessarily has the correct information on hand, but inadvertently reports the wrong information instead and thus seeks to correct that mistake. . . . It is thus unclear . . . what renders Good- luck’s error here a failure to follow instructions rather than a correctible error.” Goodluck I, 393 F. Supp. 3d at 1366 n.11. But notwithstanding the CIT’s own observations, the record clearly supports a finding that Goodluck failed to follow instructions. The information that Goodluck “inad- vertently” miscoded was addressed by Commerce on two occasions prior to verification, and Goodluck raised no ob- jection when given the chance to rebut Petitioners’ request for expanded wall thickness criteria. Nor, apparently, was Case: 20-2017 Document: 61 Page: 17 Filed: 08/31/2021

GOODLUCK INDIA LIMITED v. US 17

Goodluck incentivized by those factors to revisit its submis- sions to ensure compliance and consistency with Com- merce’s request for data. Moreover, despite receiving the August Letter and affirmatively representing to Commerce that its reported data complied with the reporting criteria, Goodluck failed to code product wall thicknesses as in- structed until the eleventh hour, when it attempted to sub- mit hundreds of revisions at the verification door. This record shows that Goodluck knew, or had reason to know, of its reporting errors. More importantly, it supports Com- merce’s conclusion that Goodluck failed to follow instruc- tions and did not merely commit a minor error. The CIT cannot impose its own contrary finding over a determina- tion by Commerce that is supported by substantial evi- dence. CONCLUSION Commerce’s determination to reject Goodluck’s revi- sions to the record is supported by substantial evidence and is otherwise not contrary to law. The judgment of the CIT is reversed. The action is remanded for further proceedings consistent with our ruling. REVERSED AND REMANDED COSTS No costs.