Hyundai Electric v. United States

Case: 21-1009 Document: 62 Page: 1 Filed: 10/04/2021 United States Court of Appeals for the Federal Circuit ______________________ HYUNDAI ELECTRIC & ENERGY SYSTEMS CO., LTD., Plaintiff-Appellant v. UNITED STATES, ABB ENTERPRISE SOFTWARE INC., Defendants-Appellees ______________________ 2021-1009 ______________________ Appeal from the United States Court of International Trade in No. 1:19-cv-00058-MAB, Judge Mark A. Barnett. ______________________ Decided: October 4, 2021 ______________________ RON KENDLER, White & Case LLP, Washington, DC, argued for plaintiff-appellant. Also represented by DAVID EDWARD BOND. KELLY A. KRYSTYNIAK, Commercial Litigation Branch, Civil Division, United States Department of Justice, Wash- ington, DC, argued for defendant-appellee United States. Also represented by BRIAN M. BOYNTON, JEANNE DAVIDSON, LOREN MISHA PREHEIM; DAVID W. RICHARDSON, Office of the Chief Counsel, United States Department of Com- merce, Washington, DC. Case: 21-1009 Document: 62 Page: 2 Filed: 10/04/2021 2 HYUNDAI ELECTRIC v. US MELISSA M. BREWER, Kelley Drye & Warren, LLP, Washington, DC, argued for defendant-appellee ABB En- terprise Software Inc. Also represented by ROBERT ALAN LUBERDA, DAVID C. SMITH, JR. ______________________ Before NEWMAN, REYNA, and HUGHES, Circuit Judges. REYNA, Circuit Judge. Hyundai Electric & Energy Systems Co. appeals a judgment of the U.S. Court of International Trade sustain- ing the U.S. Department of Commerce’s final results in the fifth administrative review of the antidumping duty order on large power transformers from the Republic of Korea. Hyundai challenges Commerce’s decision to cancel verifi- cation on the grounds that the information submitted by Hyundai was unverifiable, Commerce’s reliance on facts otherwise available, and Commerce’s use of an adverse in- ference in selecting from among the facts otherwise availa- ble. For the reasons stated below, we affirm. I The U.S. Department of Commerce (“Commerce”) im- poses antidumping duties on imported products that are sold or likely to be sold in the U.S. at “less than fair value” (“dumping”) when those sales threaten or cause material injury to a U.S. industry. 19 U.S.C. § 1673. In general, to determine whether such products are sold at less than fair value, Commerce undertakes an investigation to ascertain the difference between the “normal value” of the imported goods, i.e., the sales price in the home market, and the price at which the goods are sold in the U.S. Id. §§ 1677(35), 1677b(a). If Commerce determines that a company is sell- ing goods in the U.S. for less than their normal value, and if the U.S. International Trade Commission (“ITC”) deter- mines that such dumping threatens or causes material Case: 21-1009 Document: 62 Page: 3 Filed: 10/04/2021 HYUNDAI ELECTRIC v. US 3 injury to a U.S. industry, 1 Commerce issues an antidump- ing duty order imposing an appropriate antidumping duty rate to remedy the threat or injury. Id. §§ 1673d(a)(1), (b)(1), (c)(2). After Commerce issues such an order, an af- fected party may request an annual administrative review so that Commerce can update dumping margins, if appro- priate, to address continued dumping, if any. See 19 U.S.C. § 1675. Section 1677m governs Commerce’s conduct of admin- istrative reviews and defines, in certain respects, how Com- merce must treat information submitted by an interested party. For example, if an interested party promptly noti- fies Commerce after receiving an information request that it is “unable to submit the information requested in the re- quested form and manner,” and (among other things) pro- poses an alternative form, Commerce must consider the party’s proposal and may modify its requirements to avoid an “unreasonable burden” on the party. Id. § 1677m(c)(1). Commerce must also notify the interested party of a defi- ciency in its response and, if practicable, provide the party an opportunity to rectify the deficiency. Id. § 1677m(d). In certain circumstances, § 1677m prohibits Commerce from declining to consider submitted information even though it does not comply with all of Commerce’s requirements. See id. § 1677m(e). That prohibition applies where the infor- mation is “necessary to the determination” and all of the following requirements are met: (1) the information is submitted by the deadline es- tablished for its submission, (2) the information can be verified, 1 While the respective investigations of Commerce and the ITC are conducted concurrently, this appeal only involves Commerce’s less than fair value investigation. Case: 21-1009 Document: 62 Page: 4 Filed: 10/04/2021 4 HYUNDAI ELECTRIC v. US (3) the information is not so incomplete that it can- not serve as a reliable basis for reaching the appli- cable determination, (4) the interested party has demonstrated that it acted to the best of its ability in providing the in- formation and meeting the requirements estab- lished by the administering authority or the Commission with respect to the information, and (5) the information can be used without undue dif- ficulties. Id. Commerce is required to “verify all information relied upon” in making a final determination in an administra- tive review in certain circumstances, i.e., when a specified domestic interested party files a timely verification request and no verification was conducted in the two immediately preceding administrative reviews. Id. § 1677m(i); 19 C.F.R. § 351.307(b)(1)(v). Commerce’s regulations also provide that Commerce will conduct a verification when good cause exists. 19 C.F.R. § 351.307(b)(1)(iv). The regu- lations further set deadlines for the submission of “factual information,” which vary depending on the type of infor- mation. Id. § 351.301(c). For factual information other than the types specified in § 351.301(c)(1)-(4), § 351.301(c)(5) sets a submission deadline of “30 days be- fore the scheduled date of the preliminary results in an ad- ministrative review, or 14 days before verification, whichever is earlier.” Id. § 351.301(c)(5). Section 1677e applies when information requested by Commerce is incomplete or inaccurate. Under that section, Commerce must make determinations based on “facts oth- erwise available” when “necessary information is not avail- able on the record,” id. § 1677e(a)(1), or when a party engages in the following conduct: Case: 21-1009 Document: 62 Page: 5 Filed: 10/04/2021 HYUNDAI ELECTRIC v. US 5 (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 dead- lines 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, id. § 1677e(a)(2). 2 Section 1677e also permits Commerce to draw an ad- verse inference “in selecting from among the facts other- wise available” when “an interested party has failed to cooperate by not acting to the best of its ability to comply with a request for information.” Id. § 1677e(b)(1)(A). In such a case, Commerce is not required to determine a dumping margin as if the interested party had complied. Id. § 1677e(b)(1)(B). Commerce may draw an adverse in- ference from various sources of information, including the petition, a final determination in the underlying investiga- tion, any previous administrative review, or “any other in- formation placed on the record.” Id. § 1677e(b)(2). If Commerce properly draws an adverse inference, then it may “use any dumping margin from any segment of the 2 Section 1677e(a) further specifies that the require- ment to rely on facts otherwise available is subject to the requirements in Section 1677m(d), which requires Com- merce to provide notice and, if practicable, an opportunity to rectify a deficiency in a party’s response to a request for information from Commerce. Case: 21-1009 Document: 62 Page: 6 Filed: 10/04/2021 6 HYUNDAI ELECTRIC v. US proceeding under the applicable antidumping order,” id. § 1677e(d)(1)(B); it may exercise discretion to apply “the highest” dumping margin if warranted based on “the situ- ation that resulted in the administering authority using an adverse inference,” id. § 1677e(d)(2); and it is not required to estimate what the dumping margin would have been if the interested party had cooperated or to demonstrate that the dumping margin selected reflects the alleged commer- cial reality of the interested party, id. § 1677e(d)(3). II A On October 16, 2017, Commerce initiated its fifth ad- ministrative review of the antidumping duty order on large power transformers (“LPTs”) from the Republic of Korea for the period of review (i.e., “POR”) of August 1, 2016, to July 31, 2017. Initiation of Antidumping and Countervail- ing Duty Administrative Reviews, 82 Fed. Reg. 48,051 (Oct. 16, 2017) (J.A. 26). Commerce selected Hyundai Heavy Industries Co. as a mandatory respondent. Hyun- dai Electric & Energy Systems Co. (“Hyundai”) later be- came the successor-in-interest to Hyundai Heavy Industries Co. J.A. 27985 n.1. On December 13, 2017, Commerce issued its initial questionnaire seeking specific information related to Hyundai’s U.S. and home market sales of LPTs during the POR. This case involves two categories of information that Commerce requested from Hyundai, namely product-spe- cific cost information and cost-reconciliation information. Product-Specific Cost Information In Section D of its initial questionnaire, Commerce re- quested information regarding Hyundai’s costs of produc- ing LPTs. See J.A. 205–69. That section specifically sought information about, inter alia, Hyundai’s cost accounting system, including for example “the level of product speci- ficity over which [Hyundai’s] cost accounting system Case: 21-1009 Document: 62 Page: 7 Filed: 10/04/2021 HYUNDAI ELECTRIC v. US 7 normally captures production costs.” J.A. 211. It also asked Hyundai to “[i]dentify and quantify” the “differences between the reporting methodology and the normal books and records.” J.A. 217. In other words, Commerce sought information regarding any discrepancies between the cost data reported to Commerce and the cost data actually kept in Hyundai’s normal books and records. Id. In response, Hyundai disclosed that it had shifted costs among projects in the ordinary course of business to show that each LPT project was profitable. J.A. 7687; Appellant’s Br. 30. On May 24, 2018, Commerce issued a supplemental questionnaire. J.A. 16748–50. In question 9, Commerce referenced Hyundai’s cost shifting and requested a detailed disclosure of “the total costs recorded in [Hyundai’s SAP accounting system], the total costs reported to the Depart- ment, and an itemization of the materials and related costs making up the difference” for each sale in the U.S. market and home market. J.A. 16748. Commerce also asked Hyundai to “[e]xplain in detail how [it] was able to identify and quantify the costs that were miss-recorded [sic] in [its] SAP system” and to “show how the adjustments in each project offset each other and reconcile in total.” J.A. 16748–49. In response, Hyundai submitted Attachment SD-16, which included (i) a “Breakdown of Direct Material Cost by Material Type” on an annual basis from 2015 to 2017; (ii) a “Monthly Direct Material Cost” for the same three years; and (iii) a “Breakdown of Direct Material Cost by Project Number” for the month of March 2016, which preceded the period of review. J.A. 16909–12. Hyundai explained that the attachment showed the differences between the LPT projects’ SAP bills of materials and their actual bills of ma- terials. J.A. 16788. Hyundai further explained, “To pre- pare the reconciliation, Hyundai downloaded the BOMs [i.e., bills of materials] from both systems and by computer program was able to trace all materials in the [actual] BOMs to the SAP BOMs.” J.A. 16788–89. Case: 21-1009 Document: 62 Page: 8 Filed: 10/04/2021 8 HYUNDAI ELECTRIC v. US On July 12, 2018, Commerce sent Hyundai a second supplemental questionnaire. J.A. 25047–50. Commerce explained, “You did not provide a response to question 9,” and it listed a schedule of required items: a. Total POR costs recorded in SAP and the total POR costs reported to [Commerce]. Ensure the to- tal POR cost reported to [Commerce] agrees [with Hyundai’s cost of production] file. b. For the difference between the SAP costs and the reported costs . . . itemize each specific material and conversion cost item which make up that dif- ference. For example, identify all parts and raw materials that are included or excluded from other LPTs. c. For all SAP and reported cost itemized material and [conversion] cost differences, show which LPT project the itemized items were shifted to / from in SAP. d. Explain in detail how [Hyundai was] able to identify and [quantify] the costs which were miss- recorded [sic] in SAP. J.A. 25049. Hyundai responded again on July 23, 2018. J.A. 27355–57, 27366–86. This time, Hyundai provided At- tachment 2SD-1, a worksheet that divided the total cost differences by LPT project for reconciliation purposes into six categories: “(1) expenses recorded after the year of cost of goods sold (‘COGS’) recognition for the project; (2) recal- culated silicon steel cost; (3) recalculated other material costs; (4) material costs incurred after the year of COGS recognition; (5) recalculated scrap; and (6) recalculated fixed overhead.” J.A. 27355–56, 27369. For a single cate- gory, “other material costs,” Attachment 2SD-1 purported to show given costs shifted to particular projects and Case: 21-1009 Document: 62 Page: 9 Filed: 10/04/2021 HYUNDAI ELECTRIC v. US 9 described the corresponding types of materials. See J.A. 27370. Regarding silicon steel costs, Hyundai explained that, “[u]nlike all other materials, silicon steel is fungible and it is not possible to trace the projects to and from which sili- con steel cost might have been shifted.” J.A. 27356. Hyun- dai further stated that “actual silicon steel consumption is not recorded on a project basis, and only can be calculated manually by reference to the silicon steel processing re- ports.” Id. In Attachment 2SD-1, Hyundai provided data on shifting of steel costs for one sample LPT project. J.A. 27369–70. Hyundai also referenced earlier-submitted Attachment SD-18, which compared, for one LPT project, the “projected consumption” (calculated by engineers to “achieve the desired electrical properties”) and the “actual consumption” as stated in the steel processing report. J.A. 16789–90, 16925. Hyundai explained that “there can be differences between the core steel purchased for a par- ticular transformer and the [silicon] steel consumed,” and it disclosed the “yield loss” for the sample provided in At- tachment SD-18. J.A. 16789–90. With respect to the re- maining four categories, Hyundai disclosed aggregate cost data. Cost-Reconciliation Information In its initial questionnaire, Commerce asked Hyundai to provide worksheets, similar to the sample Commerce provided, “that illustrate how the costs reported on the fi- nancial statements reconcile to the general ledger or trial balance, to the cost accounting system (i.e., the source used to derive the reported costs), and to the reported costs.” J.A. 216–18. Hyundai responded by providing a worksheet called WS2 in Attachment D-20 that identified nine cate- gories of costs and, for each category, distinguished be- tween “Subject Merchandise” and “Non-subject Merchandise.” J.A. 8033; see also J.A. 8. Case: 21-1009 Document: 62 Page: 10 Filed: 10/04/2021 10 HYUNDAI ELECTRIC v. US Commerce issued a supplemental questionnaire that requested Hyundai to “[d]iscuss how [it] separated cost of sales on tab WS2 between MUC and non-MUC” 3 and to “[d]emonstrate and provide supporting documentation for the MUC and non-MUC breakout for [transformers].” J.A. 16750. Hyundai responded by providing Attachment SD-23, which showed the same information as that pro- vided in Hyundai’s initial response. J.A. 17076. Subsequently, after Commerce issued its preliminary results, Hyundai submitted a case brief in which it clarified for the first time that the line item for non-MUC for trans- formers included the cost of manufacturing for “1) non-sub- ject merchandise; 2) third-country sales; 3) U.S. shipments that did not enter the United States during the POR; and, 4) home market shipments made outside the POR and win- dow periods.” J.A. 28309. Hyundai did not separately identify these reconciliation items in its questionnaire re- sponses. B Commerce issued its preliminary results on August 31, 2018, assigning Hyundai a 60.81 percent ad valorem anti- dumping margin, the same margin assigned in the previ- ous administrative review. J.A. 27985–8008. Commerce explained that it used an adverse inference in selecting from the facts otherwise available because Hyundai “had failed to cooperate by not acting to the best of its ability to comply with a request for information to reconcile reported costs at the individual LPT project-level to its normal rec- ords.” J.A. 27998. Commerce found that “[t]he missing in- formation [wa]s necessary for Commerce to analyze Hyundai’s section D responses and to calculate a margin.” 3 “MUC” refers to merchandise under consideration, and “Non-MUC” refers to merchandise not under consider- ation. Case: 21-1009 Document: 62 Page: 11 Filed: 10/04/2021 HYUNDAI ELECTRIC v. US 11 Id. Specifically, regarding product-specific costs, Com- merce explained that Hyundai “failed to provide part-spe- cific itemized cost differences.” J.A. 28002–03. In submitting Attachment 2SD-1, Hyundai “only provided the cost differences in aggregate” and averred that “it [wa]s not possible to trace” cost differences for silicon steel, the larg- est material input. J.A. 28003. Commerce also explained, regarding cost reconcilia- tion, that Hyundai had “failed to provide its cost reconcili- ation in the format requested” and failed to adjust the cost of production figures from fiscal year cost of goods sold to period-of-review cost of goods sold. Id. Commerce con- cluded that, despite having “many opportunities,” Hyundai “failed to provide support for the cost differences or an ac- curate cost reconciliation” and therefore “Commerce was left with unreliable cost data.” J.A. 28004. Commerce also stated that “the information submitted by the established deadline cannot be verified,” id., and shortly thereafter it sent Hyundai a letter confirming that it had decided not to conduct a verification. J.A. 28097. On April 12, 2019, after the parties had submitted their case briefs following Commerce’s preliminary results, Com- merce published its final results and an accompanying is- sues and decision memorandum. J.A. 28295–321. Commerce again assigned Hyundai a dumping margin of 60.81 percent ad valorem, J.A. 28320, and it “continue[d] to find that Hyundai failed to provide the information as requested, or to sufficiently address its manipulation of transformer costs, within its own normal books and rec- ords,” J.A. 28301. Commerce first addressed the reliability of Hyundai’s product-specific costs and found that Hyundai inade- quately responded to the initial questionnaire by “only identif[ying] the cost difference in aggregate for each [LPT] project” and by “fail[ing] to fully distinguish each quantity and value difference between its SAP[] costs and the costs Case: 21-1009 Document: 62 Page: 12 Filed: 10/04/2021 12 HYUNDAI ELECTRIC v. US reported to Commerce by cost type (i.e., raw materials, di- rect labor, etc.).” J.A. 28304. It further found that Hyundai inadequately responded to Commerce’s supplemental questionnaire because “Hyundai again identified only the total POR cost differences” and, for one sample month out- side the period of review, “Hyundai provided a table show- ing the difference between each project’s SAP[] BOM and the [actual] BOM, and not between SAP[] and the reported costs.” J.A. 28304. Commerce likewise found that Hyundai inadequately responded to its second supplemental questionnaire by providing the requested level of detail for “only one of the six categories of cost, i.e., other materials, that it identified as being manipulated.” J.A. 28305. Regarding the silicon steel category, Commerce explained that “Hyundai failed to demonstrate and support how each project’s reported sil- icon steel consumption quantities and per-unit input val- ues were calculated, that they truly represent actual consumption, and how the per-unit input valuations dif- fered from those recorded in SAP[].” Id. “Hyundai simply attributed the difference in quantities between the silicon steel processing report and the engineering calculations to yield losses”; however, Commerce rejected that attribution because “[y]ield losses are typically based on the difference between the consumption for the job and the actual amount in the final product, not between consumption at a prelim- inary processing stage and theoretical quantities.” J.A. 28306. Commerce further found that Hyundai had failed to show cost differences as requested for the remain- ing four categories of costs identified by Hyundai. Id. Commerce also rejected Hyundai’s argument that the information provided was sufficient because “Commerce has previously relied on the very same information which Commerce now considers unreliable.” Id. Commerce ex- plained that in the earlier proceedings “Hyundai claimed that it was stopping the practice, however the shifting re- occurred in this segment.” Id. Further, Commerce found Case: 21-1009 Document: 62 Page: 13 Filed: 10/04/2021 HYUNDAI ELECTRIC v. US 13 reason in this administrative review to “take a closer look at [Hyundai’s] continuing practice and [its] attempt to cor- rect the manipulation” because, in the earlier proceedings, unlike these, “Hyundai indicated the manipulation was limited to select parts of the SAP[] system only.” J.A. 28307. Commerce then turned to Hyundai’s cost reconcilia- tion. Id. It found that Hyundai provided a cost reconcilia- tion in response to Commerce’s initial questionnaire that “a) did not comply with the format requested and b) did not provide requested details.” J.A. 28309. Specifically, Com- merce had asked Hyundai to “[l]ist each category of non- MUC separately” and reiterated that request in a supple- mental questionnaire. Id. In the reconciliations Hyundai provided, however, Hyundai “did not provide details on each category of non-MUC” but instead “included a single line titled ‘Non-MUC from Transformer’ as a reconciling item with no explanation or support.” Id. Commerce found it was not until Hyundai’s case brief after the preliminary results that Hyundai explained that the single reconciling item included “1) non-subject merchandise; 2) third-coun- try sales; 3) U.S. shipments that did not enter the United States during the POR; and, 4) home market shipments made outside the POR and window periods.” Id. Com- merce rejected as “nonsensical” Hyundai’s argument that details on these [non-MUC] items are not relevant because Commerce would ultimately exclude them.” J.A. 28310. Commerce stated that it “routinely analyze[s] costs ex- cluded from reporting and request[s] supporting docu- ments and detailed explanations of why the cost is appropriate to exclude.” Id. Commerce also explained that, in a case such as this “where the respondent admits to manipulating its normal books and records, and the ex- cluded costs include LPTs sold to third countries and mer- chandise made at the same facilities, it was even more crucial for Commerce to identify the detailed reconciling categories and related costs.” Id. Case: 21-1009 Document: 62 Page: 14 Filed: 10/04/2021 14 HYUNDAI ELECTRIC v. US Commerce next found that Hyundai had not acted to the best of its ability, and thus an adverse inference was warranted. J.A. 28312. Commerce explained that “Hyun- dai failed to provide the basic information necessary to per- form the dumping calculations as described in the preceding comments and to substantiate what the actual costs were for its transformers.” J.A. 28317. Hyundai’s failure to provide the basic information, Commerce found, prevented Commerce from calculating an accurate anti- dumping margin and from reversing the effects of Hyun- dai’s cost shifting. Id. Commerce found that Hyundai’s failures to disclose the requested information rendered ver- ification “meaningless,” and it rejected Hyundai’s argu- ment that it should conduct verification to accept new information that would establish the accuracy of its data and resolve the issues stemming from its cost shifting, J.A. 28313. C On May 8, 2019, Hyundai sought judicial review in the U.S. Court of International Trade (“CIT”). Hyundai chal- lenged certain aspects of Commerce’s final determination, including its (1) cancellation of verification, (2) application of facts otherwise available, and (3) use of an adverse in- ference. Hyundai Elec. & Energy Sys. Co. v. United States, 466 F. Supp. 3d 1303, 1307 (Ct. Int’l Trade 2020). On Au- gust 4, 2020, the CIT issued a decision sustaining Com- merce’s final results in their entirety. Id. The CIT first determined that substantial evidence supported Com- merce’s decisions to rely on facts otherwise available and cancel verification. Id. at 1309–18. The CIT pointed to Commerce’s findings that Hyundai had failed to provide adequate information on product-specific costs and cost reconciliation. Id. Regarding Hyundai’s product-specific cost disclosures, the CIT noted Commerce’s finding that Hyundai had only provided adequate product specific cost information for one Case: 21-1009 Document: 62 Page: 15 Filed: 10/04/2021 HYUNDAI ELECTRIC v. US 15 of the six cost categories identified by Hyundai, namely other material costs. See id. at 1310–13. The CIT also pointed to Commerce’s finding that Hyundai had not tracked the shifting of silicon steel costs from one project to another and had not properly accounted for the differences between the amounts reported in the silicon steel pro- cessing reports and the engineering documents. Id. at 1313–14. The CIT explained, as Commerce had found, Hyundai had also failed to adequately report product-spe- cific costs on the four remaining cost categories identified by Hyundai; instead, Hyundai had provided sample and aggregate data. Id. at 1314–15. Regarding Hyundai’s cost-reconciliation disclosures, the CIT rejected Hyundai’s arguments that it had provided information satisfying Commerce’s requests and that Com- merce did not ask Hyundai for the level of detail that Com- merce contends it did. Id. at 1316–17. The CIT also determined that substantial evidence supported Commerce’s use of an adverse inference. Id. at 1318–20. The CIT rejected Hyundai’s argument that Com- merce, in determining that Hyundai had failed to comply to the best of its ability, improperly overlooked the limita- tions of Hyundai’s cost accounting system. Id. at 1318–19. The CIT reasoned that, although Hyundai did not ade- quately report its cost-reconciliation and product-specific costs, that information “had to be available to Hyundai if it had accurately recaptured all costs—and indeed, in limited instances, Hyundai provided discrete samples detailing the adjustments for short periods of time and for limited cate- gories of expenses.” Id. at 1319. Hyundai appealed. We have jurisdiction under 28 U.S.C. § 1295(a)(5). III We apply the same standard of review applied by the CIT. Dupont Teijin Films USA, LP v. United States, Case: 21-1009 Document: 62 Page: 16 Filed: 10/04/2021 16 HYUNDAI ELECTRIC v. US 407 F.3d 1211, 1215 (Fed. Cir. 2005); SNR Roulements v. United States, 402 F.3d 1358, 1361 (Fed. Cir. 2005). Ac- cordingly, we uphold a determination by Commerce unless it is “unsupported by substantial evidence . . . or otherwise not in accordance with law.” 19 U.S.C. § 1516a(b)(1)(B)(i); Dupont, 407 F.3d at 1215; SNR Roulements, 402 F.3d at 1361; see also Fujitsu Gen. Ltd. v. United States, 88 F.3d 1034, 1038 (Fed. Cir. 1996). Substantial evidence means “such relevant evidence as a reasonable mind might accept as adequate to support a conclusion.” Matsushita Elec. In- dus. Co. v. United States, 750 F.2d 927, 933 (Fed. Cir. 1984) (quoting Universal Camera Corp. v. NLRB, 340 U.S. 474, 477 (1951)). A Commerce’s decision to rely on facts otherwise availa- ble was supported by substantial evidence and not contrary to law. Section 1677e instructs Commerce to rely on facts otherwise available when, for example, “necessary infor- mation is not available on the record.” 19 U.S.C. § 1677e(a)(1). Commerce explained that information per- taining to both product-specific costs and reconciliation was missing from the record and prevented it from under- standing Hyundai’s cost shifting and determining an anti- dumping margin. Regarding Hyundai’s product-specific costs, Commerce itemized the specific information it needed from Hyundai in the second supplemental ques- tionnaire. In response, Hyundai identified six categories of costs but only provided the requested level of detail for a single category, other materials. With respect to the silicon steel category, Hyundai failed to provide the details requested. Instead, it ex- plained that it was “not possible to trace” cost shifting for silicon steel. Hyundai also attributed discrepancies be- tween projected consumption and actual consumption to “yield losses.” But as Commerce pointed out, Hyundai’s comparison of the projected consumption to the silicon steel Case: 21-1009 Document: 62 Page: 17 Filed: 10/04/2021 HYUNDAI ELECTRIC v. US 17 processing reports would not result in a yield loss figure. And for the four remaining cost categories, Commerce ob- served that Hyundai had provided aggregate-level infor- mation that did not satisfy Commerce’s request. As for Hyundai’s cost reconciliations, Hyundai pro- vided the same single line item twice, and only after Com- merce’s preliminary results did Hyundai articulate what that line item included. Commerce’s determination that necessary information was missing from the record and its decision to rely on facts otherwise available were supported by substantial evidence. Hyundai argues that Commerce did not actually re- quest details on each category of non-MUC for purposes of cost reconciliation. Appellant’s Br. 35. We are not per- suaded. In its supplemental questionnaire, Commerce asked Hyundai to “[d]iscuss how you separated cost of sales on tab WS2 between MUC and non-MUC” and to “[d]emon- strate and provide supporting documentation for the MUC and non-MUC breakout for [transformers].” J.A. 16750. By their plain terms, these requests seek more detail than just the “category” of non-MUC as Hyundai contends. Hyundai also contends that it in fact satisfied Com- merce’s requests to fully demonstrate Hyundai’s cost shift- ing. Appellant’s Br. 37. But the record belies Hyundai’s argument. While there is no doubt that Hyundai provided certain information relating to its cost-shifting, we are not persuaded that Hyundai disclosed information that satis- fied Commerce’s requests. Indeed, Hyundai provided the level of detail that Commerce requested with respect to one of the six cost categories, namely “other materials,” that Hyundai identified in its response to Commerce’s second supplemental questionnaire. Hyundai’s repeated disclo- sure of partial, aggregate, or sample information rather than complete and itemized information establishes that Commerce’s decision to rely on facts otherwise available Case: 21-1009 Document: 62 Page: 18 Filed: 10/04/2021 18 HYUNDAI ELECTRIC v. US was reasonable and supported by substantial evidence. See 19 U.S.C. § 1677e(a)(1). B Commerce’s decision to cancel verification was also supported by substantial evidence and not contrary to law. Section 1677m(e) provides that Commerce is not obligated to conduct verification when, for example, the information cannot be verified, the information is so incomplete as to be unreliable, or the interested party has not acted to the best of its ability to meet Commerce’s requirements. 19 U.S.C. § 1677m(e). Such is the case here because Hyundai failed to provide the information necessary for Commerce’s anal- ysis despite being given multiple opportunities to do so. Where necessary information is absent, Commerce need not conduct a verification in an attempt to obtain the miss- ing information. AMS Assocs., Inc. v. United States, 719 F.3d 1376, 1380 (Fed. Cir. 2013) (concluding that Com- merce did not err in declining to conduct verification where, “[w]ithout verifiable information on those matters, Aifudi was necessarily unable to carry its burden”); Qingdao Sea- Line Trading Co. v. United States, 766 F.3d 1378, 1386 (Fed. Cir. 2014) (“Commerce was unable to verify the index because Sea-line did not provide the correct source of the data.”). Indeed, as the CIT has explained, consistent with Commerce’s objective to verify the accuracy and complete- ness of submitted factual information under 19 C.F.R. § 351.307(d), Commerce typically accepts new information at verification under limited circumstances, i.e., “only when: (1) the need for that information was not evident previously; (2) the information makes minor corrections to information already on the record; or (3) the information corroborates, supports, or clarifies information already on the record.’” Jinko Solar Co. v. United States, 229 F. Supp. 3d 1333, 1356 (Ct. Int’l Trade 2017). Hyundai does not persuasively show that these circumstances are present. Case: 21-1009 Document: 62 Page: 19 Filed: 10/04/2021 HYUNDAI ELECTRIC v. US 19 Hyundai argues that Commerce erred in finding Hyun- dai’s cost information unverifiable because, in the past, Commerce conducted verifications on submitted infor- mation similar to that submitted by Hyundai in this case. Appellant’s Br. 29. We are not persuaded. We have re- jected the notion that “Commerce is forever bound by its past practices.” Jiaxing Bro. Fastener Co. v. United States, 822 F.3d 1289, 1299 (Fed. Cir. 2016). Instead, “each ad- ministrative review is a separate exercise of Commerce’s authority that allows for different conclusions based on dif- ferent facts in the record.” Qingdao, 766 F.3d at 1387. Here, Commerce articulated sound reasons for seeking more detailed information regarding Hyundai’s cost-shift- ing in this administrative review than in prior reviews, in- cluding its observation that cost shifting had a larger impact on this administrative review. J.A. 28306–07. Such concerns support the reasonableness of Commerce’s re- quests for a greater amount of detail in this administrative review. C Commerce’s decision to use an adverse inference in se- lecting from among the facts otherwise available is also reasonable and supported by substantial evidence, and not contrary to law. The statement of administrative action on the Uruguay Round Agreements Act provides that “the purpose of the adverse inference provision is to encourage future cooperation and ensure that a respondent does not obtain a more favorable antidumping rate by failing to co- operate.” Mukand, Ltd. v. United States, 767 F.3d 1300, 1307 (Fed. Cir. 2014) (citing H.R. Rep. No. 103–316, at 200 (1994), reprinted in 1994 U.S.C.C.A.N. 4040, 4199). An ad- verse inference is warranted where an interested party fails to act to the best of its ability in responding to Com- merce’s request. 19 U.S.C. § 1677e(b)(1)(A); Nippon Steel Corp. v. United States, 337 F.3d 1373, 1382 (Fed. Cir. 2003). The “best of its ability” standard requires an inter- ested party to “put forth its maximum effort to provide Case: 21-1009 Document: 62 Page: 20 Filed: 10/04/2021 20 HYUNDAI ELECTRIC v. US Commerce with full and complete answers to all inquiries in an investigation.” Nippon, 337 F.3d at 1382; see also Mukand, 767 F.3d at 1306. The standard “does not con- done inattentiveness, carelessness, or inadequate record keeping.” Nippon, 337 F.3d at 1382. “An adverse inference may not be drawn merely from a failure to respond, but only under circumstances in which it is reasonable for Commerce to expect that more forthcoming responses should have been made . . . .” Id. at 1383. We have held that an adverse inference may be appro- priate where an interested party has been notified of a de- fect in its questionnaire response yet continues to provide a defective response. Maverick Tube Corp. v. United States, 857 F.3d 1353, 1361 (Fed. Cir. 2017) (“Borusan had already failed to provide the information requested in Com- merce’s original questionnaire, and the supplemental ques- tionnaire notified Borusan of that defect. § 1677m(d) does not require more.”). Hyundai did so here when, in response to Commerce’s second supplemental questionnaire, it only provided the requested level of detail for one out of six cost categories of product-specific cost information. It also did so when it twice provided the same single line item for non- MUC with respect to transformers in its responses pertain- ing to cost reconciliation. Given these circumstances, Com- merce’s determination that Hyundai did not act to the best of its ability in responding to Commerce’s requests is sup- ported by substantial evidence. Hyundai contends that it acted to the best of its ability in responding to Commerce’s requests. Hyundai states that it engaged in a “comprehensive effort to provide [Com- merce] with” cost reconciliation information. Appellant’s Br. 48. Hyundai also contends that it could not have been more forthcoming in providing Commerce with product- specific cost tracing given the nature of its accounting. Id. at 49. We are not persuaded. To the extent that the short- comings of Hyundai’s responses are attributable to its rec- ord keeping, that alone does not avoid an adverse Case: 21-1009 Document: 62 Page: 21 Filed: 10/04/2021 HYUNDAI ELECTRIC v. US 21 inference. Nippon, 337 F.3d at 1382. That is all the more true where, as here, Commerce clearly and repeatedly re- quested the information and identified the defects in Hyun- dai’s responses, and the information that was ultimately missing from the record was foundational to Commerce’s ability to perform the antidumping duty calculations in a sound manner. See, e.g., Mukand, 767 F.3d at 1307 (“Prod- uct-specific information is a necessary element in the dumping analysis, and it is standard procedure for Com- merce to request product-specific data in antidumping in- vestigations. It was thus reasonable for Commerce to expect from Mukand more accurate and responsive an- swers to the questionnaire.”). IV We hold that Commerce’s determinations to rely on facts otherwise available, to cancel verification, and to draw an adverse inference in selecting from among the facts otherwise available are supported by substantial evi- dence and otherwise not contrary to law. We therefore af- firm the CIT’s decision sustaining Commerce’s final results. We have considered Hyundai’s remaining and ar- guments and find them unpersuasive. AFFIRMED COSTS No costs.