Garg Tube Export LLP v. United States, No. 20-00026 (Ct. Int'l Trade 2022)
Annotate this CaseThis opinion or order relates to an opinion or order originally issued on July 9, 2021.
Download PDF
Slip Op. 22-18 UNITED STATES COURT OF INTERNATIONAL TRADE GARG TUBE EXPORT LLP and GARG TUBE LIMITED, Plaintiffs, v. Before: Claire R. Kelly, Judge UNITED STATES, Court No. 20-00026 PUBLIC VERSION Defendant, and WHEATLAND TUBE and NUCOR TUBULAR PRODUCTS INC., Defendant-Intervenors. OPINION AND ORDER [Sustaining in part and remanding in part the U.S. Department of Commerce’s remand redetermination and final results in the 2017-2018 administrative review of the antidumping duty order covering welded carbon steel standard pipes and tubes from India.] Dated: March 11, 2022 Ned H. Marshak and Dharmendra N. Choudhary, Grunfeld, Desiderio, Lebowitz, Silverman & Klestadt LLP, of New York, NY, argued for plaintiffs Garg Tube Export LLP and Garg Tube Limited. Also on the brief was Jordan C. Kahn. Robert R. Kiepura, Trial Attorney, Commercial Litigation Branch, Civil Division, U.S. Department of Justice, of Washington, DC, argued for defendant. Also on the brief were Jennifer B. Dickey, Acting Assistant Attorney General, Brian M. Boynton, Acting Assistant Attorney General, Jeanne E. Davidson, Director, Patricia M. McCarthy, Director, and Franklin E. White, Jr., Assistant Director. Of counsel was Rachel A. Bogdan, JonZachary Forbes, and Shelby M. Anderson, Attorneys, Office of Court No. 20-00026 PUBLIC VERSION Page 2 the Chief Counsel for Trade Enforcement and Compliance, U.S. Department of Commerce, of Washington, DC. Robert E. DeFrancesco, III and Theodore P. Brackemyre, Wiley Rein LLP, of Washington, DC, argued for defendant-intervenor Nucor Tubular Products Inc. Also on the brief were Alan H. Price and Cynthia C. Galvez. Kelly, Judge: Before the court is the U.S. Department of Commerce’s (“Commerce”) redetermination filed pursuant to the court’s order in Garg Tube Export LLP and Garg Tube Ltd. v. United States, 527 F. Supp. 3d 1362 (Ct. Int’l Trade 2021) (“Garg I”) in connection with Commerce’s 2017-2018 administrative review of the antidumping duty (“ADD”) order on welded carbon steel standard pipes and tubes (“CWP”) from India, covering the period of May 1, 2017 to April 30, 2018. Final Results of Redetermination Pursuant to Ct. Remand, Oct. 7, 2021, ECF Nos. 73-1–2 (“Remand Results”); see generally Garg I, 527 F. Supp. 3d 1362; [CWP] From India, 85 Fed. Reg. 2,715 (Dep’t Commerce Jan. 16, 2020) (final results of [ADD] admin. review; 2017–2018) (“Final Results”) and accompanying Issues and Decision Memo., A-533-502, (Jan. 9, 2020), ECF No. 24-5 (“Final Decision Memo.”). On remand, Commerce no longer relies on facts available with an adverse inference for the missing cost of production data. Remand Results at 17–22. Commerce’s Remand Results with respect to the use of neutral facts available are sustained. Also before the court is Commerce’s determination in the Final Results regarding the finding of and subsequent adjustment for a particular market situation (“PMS”). Final Decision Memo. at 3–31; see also Remand Results at 4–10. Because Court No. 20-00026 PUBLIC VERSION Page 3 Commerce’s application of the PMS adjustment to the sales-below-cost test was improper under the statute, the court remanded to Commerce and declined to reach the issues of whether Commerce’s PMS finding was supported by substantial evidence and the reasonableness of the regression model as a methodology for calculating the PMS adjustment. Garg I, 527 F. Supp. 3d at 1371–73. Although Commerce is no longer applying a PMS adjustment to the sales-below-cost test, it continues to find that a cost-based PMS exists in India for hot-rolled coil (“HRC”) and applies a PMS adjustment to the cost of production for sales based on constructed value. Remand Results at 8–10. For the reasons that follow, Commerce’s determination with respect to the finding of a PMS and the corresponding adjustment to Garg’s cost of production are remanded. BACKGROUND The court presumes familiarity with the facts of this case as set forth in its previous opinion ordering remand to Commerce, and now recounts those facts relevant to the court’s review of the Remand Results as well as the Final Results’ finding and subsequent adjustment for a PMS. See generally Garg I, 527 F. Supp. 3d at 1365–70. Commerce conducted an administrative review of the ADD order covering certain CWP from India, for the period of review covering May 1, 2017 through April 30, 2018. See generally Initiation of Antidumping and Countervailing Duty Administrative Reviews, 83 Fed. Reg. 32,270, 32,270 (Dep’t Commerce July 12, 2018); see also id. at 1365. In calculating the dumping margin for Garg, Commerce Court No. 20-00026 PUBLIC VERSION Page 4 relied on facts available with an adverse inference to fill the gap in the record stemming from the refusal of one of Garg’s unaffiliated suppliers to provide the requested cost information. See Final Decision Memo. at 32–41; Garg I, 527 F. Supp. 3d at 1371–73. Commerce found that a PMS existed in India distorting the price of HRC, an input used in CWP, based on the cumulative and collective impact of global steel overcapacity, subsidization of the Indian HRC market by the Government of India (“GOI”), trade interventions by the GOI, and Garg’s nonpayment of antidumping and safeguard duties on imports of HRC on the Indian steel market. Final Decision Memo. at 19; see also [CWP] from India, 84 Fed. Reg. 33,916 (Dep’t Commerce July 16, 2019) (prelim. results of [ADD] admin. review; 2017–2018) (“Prelim. Results”) and accompanying Prelim. Decision Memo. at 19–21, A-533-502, PD 207, bar code 3859225-01 (July 11, 2019) (“Prelim. Decision Memo.”); Memo. Re: Decisions on [PMS] Allegations at 18–27, PD 209, bar code 3859233-01 (July 10, 2019) (“PMS Memo.”).1 Commerce adjusted the cost of production in its sales-below-cost test to account for its PMS finding. Final Decision Memo. at 50. On March 10, 2020, Commerce filed indices to the public and confidential administrative records underlying Commerce’s final determination on the docket at ECF No. 24-1–2. On October 21, 2021, Commerce filed indices to the public and confidential administrative records underlying Commerce’s remand redetermination on the docket at ECF No. 78-2–3. All references in this opinion to documents from the administrative record underlying Commerce’s final determination and remand redetermination are identified by the numbers assigned by Commerce in those indices and preceded by “PD” and “CD” to denote public or confidential documents. 1 Court No. 20-00026 PUBLIC VERSION Page 5 The court remanded Commerce’s decision to rely on an adverse inference when selecting from facts available to fill the gap in Garg’s cost of production data for further explanation or reconsideration because it could not discern how Commerce applied Section 776 of the Tariff Act of 1930, as amended, 19 U.S.C. § 1677e (2018),2 from Commerce’s explanation in the Final Decision Memo. Garg I, 527 F. Supp. 3d at 1371–73 (setting forth contradictory statements from the Final Decision Memo. indicating that Commerce’s rationale may have relied on 19 U.S.C. § 1677e(a) and/or § 1677e(b)). The court also remanded Commerce’s determination to use a PMS adjustment in its sales-below-cost test finding “the statute does not empower Commerce to adjust a respondent’s reported costs to account for a cost-based PMS when Commerce relies on home market or third country market sales to determine normal value.”3 Garg I, 527 F. Supp. 3d at 1370. Commerce filed the Remand Results on October 7, 2021. Remand Results. Under respectful protest,4 Commerce made several revisions to the Final Results. See generally id. First, Commerce clarifies that it is operating under 19 U.S.C. § 1677e(a) to fill the gap in the record and explains that it is no longer using facts Further citations to the Tariff Act of 1930, as amended, are to the relevant provisions of Title 19 of the U.S. Code, 2018 edition. 3 Where there are sales at less than cost of production, under certain circumstances such sales may be disregarded for the purposes of normal value. See 19 U.S.C. § 1677b(b). “If no sales made in the ordinary course of trade remain, the normal value shall be based on the constructed value of the merchandise.” Id. 4 By adopting a position “under protest,” Commerce preserves its right to appeal. See Viraj Grp., Ltd. v. United States, 343 F.3d 1371, 1376 (Fed. Cir. 2003). 2 Court No. 20-00026 PUBLIC VERSION Page 6 available with an adverse inference and instead relies on neutral facts available. Id. at 17–22. Second, Commerce removes the PMS adjustment to the cost of production in its sales-below-cost test. Id. at 33. Commerce continues to find that a PMS in India distorted the cost of HRC and applies an upward adjustment to Garg’s reported cost of production for sales where normal value was based on constructed value. Id. at 8–9. Garg and Nucor filed comments and replies to the remand redetermination. Pls.’ Comments on Remand Redetermination, Nov. 8, 2021, ECF No. 79 (“Garg’s Remand Comments”); Pls.’ Reply to Comments on Remand Redetermination, Dec. 15, 2021, ECF No. 85 (“Garg’s Reply”); Def.-Intervenor [Nucor]’s Comments on [Remand Results], Nov. 8, 2021, ECF No. 80 (“Nucor’s Remand Comments”); Def.-Intervenor [Nucor’s] Reply Comments on [Remand Results] at 1–4, Dec. 15, 2021, ECF No. 84 (“Nucor’s Reply”). Neither party objects to Commerce’s decision to use neutral facts available to fill in the missing cost of production data before the court. Garg’s Remand Comments at 3; see Nucor’s Remand Comments; Nucor’s Reply. However, the parties disagree about whether Commerce’s decision to remove the PMS adjustment to costs of production from the sales-below-cost test, the existence of a PMS, and the methodology used to quantify the PMS adjustment should be sustained. Compare Garg’s Remand Comments at 3–5; Garg’s Reply with Nucor’s Remand Comments at 2–3; Nucor’s Reply at 1–5. Garg argues that Commerce’s continued finding of a PMS is not supported by substantial evidence and Commerce’s use of the Court No. 20-00026 PUBLIC VERSION Page 7 regression analysis to quantify the PMS adjustment is arbitrary; therefore, a second remand is required. Garg’s Remand Comments at 3–5. Nucor argues that Commerce may adjust costs of production to account for a PMS when Commerce performs the sales-below-cost test. Nucor’s Remand Comments at 2; Nucor’s Reply at 4–5; but see Hyundai Steel Co. v. United States, 19 F.4th 1346 (Fed. Cir. 2021). However, in light of the court’s remand order, Nucor agrees with Commerce’s decision to remove the PMS adjustment in the sales-belowcost test under respectful protest. Nucor’s Remand Comments at 2; see also Nucor’s Reply at 4–5 (suggesting the court reconsider its position because the Court of Appeals for the Federal Circuit has not issued a mandate in Hyundai). Nucor further argues that Commerce’s continued finding of a PMS and the corresponding regression-based adjustment are supported by substantial evidence and otherwise in accordance with law. Nucor’s Reply at 1–4. On December 15, 2021, Defendant filed its reply to the remand comments. Def.’s Resp. to Comments on Remand Redetermination, Dec. 15, 2021, ECF No. 83 (“Def.’s Reply”). Defendant incorporates its previous arguments in support of Commerce’s PMS finding and the resulting adjustment. Id. at 6–7 (citing Def.’s Resp. to Pls.’ Mot. for J. on Agency R. 13–21, 24– 33, Feb. 1, 2021, ECF No. 45). Defendant asks the court to sustain Commerce’s remand redetermination in its entirety. Id. at 8. Court No. 20-00026 PUBLIC VERSION Page 8 JURISDICTION AND STANDARD OF REVIEW This court has jurisdiction pursuant to 19 U.S.C. § 1516a(a)(2)(B)(iii) and 28 U.S.C. § 1581(c) (2018), which grant the court authority to review actions contesting the final determination in an administrative review of an ADD order. The court will uphold Commerce’s determination unless it is “unsupported by substantial evidence on the record, or otherwise not in accordance with law.” 19 U.S.C. § 1516a(b)(1)(B)(i). “The results of a redetermination pursuant to court remand are also reviewed ‘for compliance with the court’s remand order.’” Xinjiamei Furniture (Zhangzhou) Co. v. United States, 968 F. Supp. 2d 1255, 1259 (Ct. Int'l Trade 2014) (quoting Nakornthai Strip Mill Public Co. v. United States, 32 CIT 1272, 1274 (2008)). DISCUSSION I. Application of Facts Available On remand, Commerce relies on neutral facts available to fill the gap created by the missing cost of production data from Garg’s Unaffiliated Supplier5 and articulates its reasoning for its determination. Remand Results at 3, 17–22. No party challenges Commerce’s decision before the court. Def.’s Reply at 7; Garg’s Remand Comments at 3; see Nucor Remand Comments; Nucor Reply. Commerce has complied with the court’s remand instructions and its determination is supported by substantial evidence; therefore, Commerce’s determination is sustained. 5 The Unaffiliated Supplier is [[ ]]. Court No. 20-00026 PUBLIC VERSION Page 9 When Commerce is missing information necessary to make an ADD determination, it must use facts otherwise available to fill the gap in the record created by the missing information. See 19 U.S.C. § 1677e(a); Nippon Steel Corp. v. United States, 337 F.3d 1373, 1380–81 (Fed. Cir. 2003). If a gap exists because a party failed to cooperate to the best of its ability, Commerce may use an adverse inference when selecting facts available to fill the gap. 19 U.S.C. § 1677e(b). However, under 19 U.S.C. § 1677e(a) Commerce may use adverse inferences against a cooperative respondent, if doing so will yield an accurate rate, promote cooperation, and thwart duty evasion. Mueller Comercial De Mexico v. United States, 753 F.3d 1227, 1232–36 (Fed. Cir. 2014). Here, the court remanded Commerce’s decision to use an adverse inference when selecting from facts otherwise available to fill the gap created by the Unaffiliated Supplier’s non-cooperation for reconsideration or additional explanation because the court could not discern if Commerce was acting under 19 U.S.C. § 1677e(a) or (b). Garg I, 527 F. Supp. 3d at 1371–73. To the extent that Commerce chose to rely on 19 U.S.C. § 1677e(a) on remand, the court instructed Commerce to “do more to support its determination” and explain how the decision to use facts available with an adverse inference was appropriate in light of the detracting evidence provided by Garg.6 Garg I, 527 F. Supp. 3d at 1372–73. 6 Garg provided unrebutted evidence that it lacked the market power necessary to (footnote continued) Court No. 20-00026 PUBLIC VERSION Page 10 Commerce clarifies that it is relying on 19 U.S.C. § 1677e(a) to determine the missing cost of production data from the Unaffiliated Supplier because the Unaffiliated Supplier is an interested party as defined by 19 U.S.C. § 1677(9)(A). Remand Results 3, 17–22. Commerce explains that it relies on neutral facts available because it is unable to support a determination that Garg possesses sufficient “market power” or “leverage” to induce the cooperation of the Unaffiliated Supplier with substantial evidence.7 Id. at 17–19. Commerce has complied with the remand order and no party objects to its decision; therefore, its determination is sustained. induce the cooperation of the Unaffiliated Supplier. See Garg’s Suppl. Resp. at Ex. S1-D-2(f) Part-1 at 30, Part-2; Resp. from [Garg] to [Commerce] Re: Section A Questionnaire Resp. at Ex.A-11(b),(d), PDs 33–36, CDs 7-11, bar codes 3754219-01– 04, 3754211-01–05 (Sept. 17, 2018) (“Garg’s Section A Questionnaire Resp.”); see also Garg’s Section A Questionnaire Resp. at Part-1. Garg further demonstrated that it made multiple attempts to induce the cooperation of the Unaffiliated Supplier. Resp. to Commerce Re: Section A–D Suppl. Questionnaire 51–53, PDs 110–34, CDs 69–101, bar codes 3794286-01–25, 3794250-01–33 (Feb. 19, 2019); see also Garg I, 527 F. Supp. 3d at 1366 n.8, 1366–67. 7 Commerce maintains that Mueller does not require Commerce to demonstrate that Garg has “market power” or “leverage” over the Unaffiliated Supplier. Remand Results at 17; but see Mueller, 753 F.3d at 1234–36 (explaining that for Commerce to induce the cooperation of a non-cooperating supplier through a cooperating party, there must be substantial evidence that the cooperating party has leverage over the non-cooperating supplier). Commerce complains that it is “troubled by the implications of the Court’s opinion” because it is difficult for Commerce to obtain the evidence necessary to support a finding of Garg’s “market power” or “leverage” to induce the Unaffiliated Supplier’s cooperation without the cooperation of the Unaffiliated Supplier. Remand Results at 20–21. However, Mueller explains that the statutory scheme allows adverse inferences against a cooperative respondent, who has leverage, to compel the cooperation of the uncooperative party. (footnote continued) Court No. 20-00026 PUBLIC VERSION II. Page 11 PMS and the Sales-Below-Cost Test Commerce removed the PMS adjustment from its sales-below-cost test. Remand Results at 3, 8–9. Garg and Defendant argue that this removal complies with the court’s remand instructions and should be sustained. Garg’s Remand Comments at 3–4; Garg’s Reply Comments at 1–2; Def.’s Reply at 5–7. Nucor argues that the statute permits Commerce to apply a PMS adjustment to the cost of production during the sales-below-cost test and urges the court to reconsider its position because the issue is currently on appeal. Nucor’s Remand Comments at 2– 3; Nucor’s Reply at 4–5. The court’s remand instructed Commerce to remove the PMS adjustment from the sales-below-cost test. Garg I, 527 F. Supp. 3d at 1369–71. Furthermore, in Hyundai Steel, the Court of Appeals found that 19 U.S.C. § 1677b(b) does not authorize the application of a PMS adjustment to the cost of production in the sales-below-cost test. 19 F.4th 1346 (Fed. Cir. 2021). Commerce’s determination complies with the court’s remand order is sustained. III. Particular Market Situation Determination The court remands Commerce’s PMS determination for reconsideration or additional explanation. A finding of a PMS in the constructed value context requires Mueller, 753 F.3d at 1232–36. Lastly, Commerce argues that it does “not have the privilege of time in accumulating evidence of a ‘long-term’ relationship between Garg [] and [the Unaffiliated Supplier]” because this is the first review in which Commerce reviewed Garg individually. Remand Results at 21. Yet providing evidence of leverage is not confined to demonstrating a long-term relationship. See Mueller, 753 F.3d at 1234–36. Court No. 20-00026 PUBLIC VERSION Page 12 Commerce to identify what unique fact or set of facts in the market prevents a respondent’s reported “cost of materials and fabrication or other processing” from “accurately reflecting the cost of production in the ordinary course of trade.”8 See 19 U.S.C. § 1677b(e), (f)(1). Commerce relies on the cumulative and collective impact of multiple market phenomena to support its PMS determination. See Final Decision Memo. at 19. Commerce must demonstrate the existence of each market phenomenon and support its determination with substantial evidence. See 19 U.S.C. § 1677b(e). Commerce must then explain how these market phenomena create a PMS, i.e., a unique set of facts distorting the cost of materials and fabrication or other processing in the country. Id. Lastly, Commerce must demonstrate that this PMS renders the cost of materials and fabrication inaccurate. Id. Although Commerce demonstrates the existence of each market phenomenon that it contends contributes to a PMS in the Indian HRC market, it fails to explain how, when combined, these market phenomena give rise to a unique set of facts distorting the cost of materials and fabrication or other processing of any kind, such that Garg’s reported costs do not accurately reflect the cost of production in the ordinary course of trade. See Final Decision Memo. at 3–31. “For the purposes of [19 U.S.C. § 1677b(b) and (e)] . . . Costs shall normally be calculated based on the records of the exporter or producer of the merchandise, if such records are kept in accordance with the generally accepted accounting principles of the exporting country (or the producing country, where appropriate) and reasonably reflect the costs associated with the production and sale of the merchandise.” 19 U.S.C. § 1677b(f)(1)(A). 8 Court No. 20-00026 PUBLIC VERSION Page 13 To determine “whether subject merchandise is being sold, or is likely to be sold at less than fair value” Commerce compares the export price or constructed export price with the normal value of the subject merchandise. 19 U.S.C. § 1677b(a). Commerce may determine the normal value of the subject merchandise using one of several methodologies. See 19 U.S.C. § 1677b(a)(1)-(5). Commerce may use the constructed value of the subject merchandise if the normal value cannot be determined under paragraph (1)(B)(i) or (ii). 19 U.S.C. § 1677b(a)(4); see 19 U.S.C. § 1677b(a)(1)(B)(i), (ii). Constructed value is the sum of: (1) the cost of materials and fabrication or other processing of any kind employed in producing the merchandise, during a period which would ordinarily permit the production of the merchandise in the ordinary course of trade; (2)(A) the actual amounts incurred and realized by the specific exporter or producer being examined in the investigation or review for selling, general, and administrative expenses, and for profits, in connection with the production and sale of a foreign like product, in the ordinary course of trade, for consumption in the foreign country [; and] (3) the cost of all containers and coverings of whatever nature, and all other expenses incidental to placing the subject merchandise in condition packed ready for shipment to the United States. 19 U.S.C. § 1677b(e). However, “for the purposes of paragraph (1)” if Commerce determines that “a particular market situation exists such that the cost of materials and fabrication or other processing of any kind does not accurately reflect the cost of production in the ordinary course of trade, [Commerce] may use another calculation methodology under this part or any other calculation methodology.” § 1677b(e). 19 U.S.C. Court No. 20-00026 PUBLIC VERSION Page 14 Neither the statute nor the legislative history defines a PMS. The legislative history to the Trade Preferences Extension Act of 2015 (“TPEA”), which added the PMS language to 19 U.S.C. § 1677b(e), offers no examples of what may constitute a PMS. The phrase “particular market situation” existed prior to TPEA in 19 U.S.C. § 1677b(a)(1)(B) and (C) which inquire whether a “particular market situation prevents proper comparison of normal value with export price.” § 1677b(a)(1)(B)(ii)(III), (C)(iii). 19 U.S.C. The Statement of Administrative Action to the Uruguay Round Agreements Act (“URAA”), which added these provisions, does give some clue as to what a PMS may include. The Agreement does not define “particular market situation,” but such a situation might exist where a single sale in the home market constitutes five percent of sales to the United States or where there is government control over pricing to such an extent that home market prices cannot be considered to be competitively set. It also may be the case that a particular market situation could arise from differing patterns of demand in the United States and in the foreign market. For example, if significant price changes are closely correlated with holidays which occur at different times of the year in the two markets, the prices in the foreign market may not be suitable for comparison to prices to the United States. Uruguay Round Agreements Act, Statement of Administrative Action, H.R. Doc. No. 103–316, vol. 1, at 822 (1994), reprinted in 1994 U.S.C.C.A.N. 4040, 4162 (“SAA”). The SAA examples involve unique facts in a given country that render the home market price unsuitable for comparison to the U.S. price. The TPEA, in providing Commerce the alternative to use any other methodology to calculate costs under 19 U.S.C. § 1677b(e)(1), adopts both a Court No. 20-00026 PUBLIC VERSION Page 15 comparative requirement and a causal requirement. See 19 U.S.C. § 1677b(e). The statute provides “if a particular market situation exists such that the cost of materials and fabrication or other processing of any kind does not accurately reflect the cost of production in the ordinary course of trade.” Id. The use of the causal phrase “such that” requires that in addition to finding unique market phenomena, Commerce must demonstrate that those market phenomena prevent the cost of materials and fabrication from accurately reflecting the cost of production. Commerce must therefore identify what unique facts render the cost of materials and fabrication an inaccurate reflection of the cost of production in the ordinary course of trade.9 Commerce may choose the appropriate methodology to identify what unique facts render the cost of materials and fabrication an inaccurate reflection of the cost of production, so long as it comports with its statutory mandate and provides a reasoned explanation supported by substantial evidence. See Ceramica Regiomontana, S.A. v. United States, 10 CIT 399, 404–05 (1986) (citing Chevron Ordinary course of trade “means the conditions and practices which, for a reasonable time prior to the exportation of the subject merchandise, have been normal in the trade under consideration with respect to merchandise of the same class or kind.” 19 U.S.C. § 1677(15). Commerce shall consider sales and transactions to be outside of the ordinary course of trade where “the particular market situation prevents a proper comparison with the export price or constructed export price.” 19 U.S.C. § 1677(15)(C). However, the language of 19 U.S.C. § 1677(15)(C) does not negate 19 U.S.C. § 1677b(e)’s causal requirement. To suggest otherwise would render the inclusion of “such that the cost of materials and fabrication or other processing of any kind does not accurately reflect the cost of production” meaningless. 19 U.S.C. § 1677b(e). 9 Court No. 20-00026 PUBLIC VERSION Page 16 U.S.A. Inc. v. Natural Resources Defense Council, 467 U.S. 837, 843 (1984); Fujitsu Gen. Ltd. v. United States, 88 F.3d 1034, 1039 (Fed. Cir. 1996); Universal Camera Corp. v. Nat’l Labor Relations Bd., 340 U.S. 474, 488 (1951) (substantial evidence “must take into account whatever in the record fairly detracts from its weight”). The evidence must be sufficient such that a reasonable mind might accept the evidence as adequate to support its conclusion while considering contradictory evidence. See Consol. Edison Co. of New York v. N.L.R.B., 305 U.S. 197, 229 (1938); see also Suramerica de Aleaciones Laminadas, C.A. v. United States, 44 F.3d 978, 985 (Fed. Cir. 1994). Commerce identifies various market phenomena to support its determination that a PMS exists in the Indian HRC market. Final Decision Memo. 19–33. In light of the latitude afforded by the statute, Commerce reasonably concluded that the global steel overcapacity crisis is one of several market phenomena it could consider as contributory to a PMS in India.10 Id. at 19–27. Commerce explains that the global steel overcapacity crisis is well documented, existed prior to the period of review, and continued throughout the period of review, creating serious market distortions Contra NEXTEEL Co. v. United States, 355 F. Supp. 3d 1336, 1350 (Ct. Int'l Trade 2019); NEXTEEL Co. v. United States, 450 F. Supp. 3d 1333,1340–41 (Ct. Int'l Trade 2020); Dong-A Steel Co. v. United States, 475 F. Supp. 3d 1317, 1337 (Ct. Int'l Trade 2020); Seah Steel Corp. v. United States, 513 F. Supp. 3d 1367, 1396 (Ct. Int'l Trade 2021); Husteel Co. v. United States, 426 F. Supp. 3d 1376, 1392 (Ct. Int'l Trade 2020), aff'd sub nom. Hyundai Steel Co. v. United States, 19 F.4th 1346 (Fed. Cir. 2021). 10 Court No. 20-00026 PUBLIC VERSION Page 17 resulting in negative effects and downward pressure on the global steel market.11 PMS Memo. at 19 n.124; id. at 20–27. Commerce supports its finding of the ongoing global overcapacity crisis with record evidence.12 Final Decision Memo. at 26 (citing PMS Memo. at 19); see generally [PMS] Allegation, Resubmitted Market Viability Allegation, & Supp. Info., PDs 72–99, CDs 33–61, bar codes 3785365-01–28, 378532201–29 (Dec. 21, 2018) (“PMS Allegation”)). Although a global phenomenon cannot be considered unique, nothing in the statute precludes Commerce from finding that a global market condition contributes to a PMS in India if it were to combine with other market phenomena unique to India. In addition to the overcapacity crisis, Commerce relied on GOI subsidies as a contributing market phenomenon to the PMS in the Indian HRC market. Final Commerce lists “significant price suppression, displaced markets, unsustainable capacity utilization, negative financial performance, shutdowns, and lay-offs” as negative effects of the global overcapacity crisis. PMS Memo. at 19 n.124 (citing Circular Welded Non-Alloy Steel Pipe from the Republic of Korea, 82 Fed. Reg. 57,583 (Dep’t Commerce Dec. 6, 2017) (Prelim. Results of [ADD] Admin. Review; 2015-2016) and accompanying Prelim. Decision Memo. at 14). 12 Garg disputes the evidence Commerce relies on to identify global capacity as a market phenomenon contributing to its PMS determination, arguing that steel overcapacity and suppression of HRC prices did not exist during the period of review in either the Indian market or the global market. Pls.’ Br. at 9–17; Garg’s Remand Comments at 4. In support of its argument, Garg proffered evidence that the steel market had recovered. Letter Re: Rebuttal Resp. to Pet’rs’ [PMS] Allegations: Admin. Review of the [ADD] Order on [CWP] from India at 69, Exs. 2A, 2B, 3A, 3B, 3C, 17, 22, C.R. 102-06, P.R. 142-46 (Mar. 4, 2019) (“PMS Rebuttal”). Commerce rejected Garg’s argument, explaining that evidence of short-term recovery does not suggest that the distortive effects of the overcapacity crisis have been mitigated in their entirety. PMS Memo. at 20–21; Final Decision Memo. at 20–23. 11 Court No. 20-00026 PUBLIC VERSION Page 18 Decision Memo. at 27–30. Commerce explains “the GOI’s subsidization of large Indian HRC producers permeates the entire Indian HRC market” contributing additional downward pressure on Indian HRC prices. Id. at 29. The decision to include GOI subsidies as one of the market phenomena is informed by Commerce’s findings in Circular Welded-Carbon-Quality Steel Pipe from India, 77 Fed. Reg. 66,468 (Oct. 12, 2012) (final affirmative countervailing duty deter.) and accompanying Issue and Decision Memo. at 21–22 (countervailing the provision of HRC for less than adequate remuneration to producers of carbon quality steel pipe and Tube in India)13 and the PMS Allegation. Final Decision Memo. at 27–28; see PMS Memo. at 4, 4 n.29; PMS Allegation at 18–27 (listing the programs that Commerce has found to benefit Indian HRC producers). Commerce reasonably explained that it used the existence of subsidies in the Indian market as evidence that distortions were present in the market. See Final Decision Memo. at 27–30. Commerce believes both the global overcapacity crisis and the subsidies provided by the GOI depress the price of HRC in the Indian market and that the GOI attempted to remedy the distortion caused by the global overcapacity crisis through the enactment of trade remedies. Id. at 21–31; PMS Memo. at 22. Commerce points to evidence of an influx of cheap steel imports into the Indian market and explains Although Commerce found a countervailable subsidy, no countervailing duty order was issued because the U.S. International Trade Commission did not find that the subsidy materially injured the domestic industry. PMS Allegation at 23. 13 Court No. 20-00026 PUBLIC VERSION Page 19 that in response to the unfairly traded HRC, the GOI attempted to “level the [domestic] playing field” by imposing a variety of safeguard and antidumping duties.14 PMS Memo. at 6 n.40 (listing GOI safeguard duty orders), 22, 22 n.138–40; see Final Decision Memo. at 29; See also PMS Allegation at Ex. 63. Commerce also points to evidence on the record that Garg and two of its suppliers were exempt from paying safeguard and antidumping duties.15 Final Decision Memo. at 31 (relying on PMS Memo. at 26–27). Commerce explains that although Garg was exempt from paying antidumping and safeguard duties, nonpayment of antidumping and safeguard duties can be evidence of a PMS. Id. at 30–32 (relying on PMS Memo. at 25–27). Commerce explains that when safeguard duties are implemented to remedy price distortions, nonpayment of safeguard duties, that should be paid but for an exemption, evidences that the distortive effect of global overcapacity still exists in the market because the exemptions prevent the purpose of the safeguard duties from being fulfilled. See id. at 30 (relying on PMS Memo. at 2526.) With respect to the nonpayment of antidumping duties, Commerce explains that Petitioners state that during the period of review there were antidumping duty orders covering HRC imports from Brazil, China, Indonesia, Japan, Korea, and Russia into India. PMS Memo. at 6. Additionally, safeguard measures for imports of HRC into India were in effect “for nearly all of the [period of review].” Id. 15 Garg reported to Commerce that it was exempt from paying antidumping or safeguard duties on the imported HRC used to produce the subject merchandise during the period of review because Garg Tube Export LLP is located in a special economic zone and Garg Tube Limited imports HRC under the Advance Authorization Scheme. Id. at 25. 14 Court No. 20-00026 PUBLIC VERSION Page 20 nonpayment of antidumping duties is relevant to its PMS analysis because it creates another entry point for the distortive impacts of global overcapacity into the market.16 See Final Decision Memo. at 31 (relying on PMS Memo. at 26–27). Therefore, the trade remedies enacted by the GOI evidence distortions and the exceptions to those remedies evidence that the distortions persist in the Indian HRC market.17 Id. (relying on PMS Memo. at 26–27). In sum, Commerce supports with substantial evidence its determination that a variety of factors exist in the Indian market that affect the price of HRC. Despite identifying market phenomena which might distort the price of HRC, Commerce fails to explain how the cumulative and collective impact of the market For example, Commerce states that all of Garg’s import purchases on the record from China were made below the non-injurious price set in the GOI’s antidumping order. Id. Commerce believes that the existence of an antidumping duty order covering HRC imports from China demonstrates that the Chinese market is distorted due to the overcapacity crisis. Id. Because Garg was exempt from paying antidumping duties, the corrective measures of the antidumping duty order were not effectuated; therefore, the distortive effects of overcapacity entered the market. Id. at 26–27. 17 Garg argues that Commerce misconstrued Garg’s non-payment of safeguard and antidumping duties. Pls.’ Br. at 27–29. Garg misunderstands the core of Commerce’s argument. Commerce explains that it does not base its PMS determination on “whether a respondent’s specific purchase prices of HRC were distorted and, thus, outside of the ordinary course of trade.” Final Decision Memo. at 29. Instead, Commerce bases its PMS determination on “whether prices reflected in the entire Indian HRC market, as a whole, are distorted.” Id. In light of this explanation, it is reasonably discernable that Commerce points to Garg’s non-payment of safeguard and antidumping duties as evidence that the trade remedies put in place to cure the distortive impact of the overcapacity crisis can be circumvented, calling into question their effectiveness. See id. at 29–31. 16 Court No. 20-00026 PUBLIC VERSION Page 21 phenomena upon which it relies are unique to the Indian market and therefore constitute a PMS. Commerce states that “the GOI actively pursued measures, such as subsidization and trade remedies, all aimed at supporting the domestic steel producers and their ambitions for capacity expansions, a scenario of further distortions that is unique to India.” Final Decision Memo. at 23 (quoting PMS Memo. at 21). Yet, Commerce itself acknowledges that the overcapacity crisis impacted markets globally and suppressed steel prices. Id. at 20. A global market condition is not a unique market phenomenon. The existence of trade remedies and subsidies are not unique market phenomena, nor are the exceptions to the imposition of trade remedies. Also absent from Commerce’s Final Decision Memo. is an explanation of how the alleged PMS distorts the cost of production so that Garg’s reported costs of production are not accurate in the ordinary course of trade. Commerce’s explanation implies that it believes the distortions created by the market phenomena automatically render the market unsuitable for comparison. See id. at 19–32. The statute requires that the existence of a PMS must be “such that” it renders the cost of fabrication and materials inaccurate. See 19 U.S.C. § 1677b(e). In the explanation currently before the court Commerce has identified market phenomena which affect the market price of the HRC inputs and conclusively states that a PMS exists. Id. at 21–32. On remand, Commerce must either reconsider its determination or explain how the combination of each of the market phenomenon Court No. 20-00026 PUBLIC VERSION Page 22 upon which it relies creates a PMS in the Indian HRC market such that “the cost of materials and fabrication or other processing of any kind does not accurately reflect the cost of production in the ordinary course of trade” and support its explanation with substantial evidence. IV. The Regression Analysis Although the statute permits Commerce to choose any reasonable methodology to quantify a PMS adjustment, any adjustment must be supported by substantial evidence. Here, Commerce fails to explain its determination in light of detracting evidence on the record. Therefore, should Commerce continue to find a PMS on remand, Commerce must either reconsider or further explain its PMS adjustment consistent with this opinion. Where Commerce identifies a PMS such that the cost of materials and fabrication are not accurate 19 U.S.C. § 1677b(e) permits Commerce to use any other calculation methodology to quantify the impact of the PMS on the costs of materials and fabrication. 19 U.S.C. § 1677b(e). The chosen methodology must be reasonable, and the determination must be supported by substantial evidence. See Vicentin S.A.I.C. v. United States, 404 F. Supp. 3d 1323, 1342 (Ct. Int’l Trade 2019). Substantial evidence is “such relevant evidence as a reasonable mind might accept as adequate to support a conclusion.” Universal Camera, 340 U.S. at 477 (quoting Consol. Edison Co., 305 U.S. at 229). “The substantiality of evidence must take into account whatever in the record fairly detracts from its weight.” Id. at 488. In Court No. 20-00026 PUBLIC VERSION Page 23 providing its explanation Commerce must articulate a “rational connection between the facts found and the choice made.” Burlington Truck Lines, Inc. v. United States, 371 U.S. 156, 168 (1962). Commerce calculated the PMS adjustment using an ordinary least squares regression model provided in petitioners’ PMS submissions with certain modifications by Commerce (the “OLS Regression Model”).18 See Final Decision Multiple regression models estimate the relationship between explanatory variables and a dependent variable, holding all other variables equal. Final Decision Memo. at 64; Jeffrey M. Wooldridge, Introductory Econometrics A Modern Approach 68 (South-Western Cengage Learning 5th ed.) (2013) (“Wooldridge, Econometrics”); Letter from Petitioners Re: Revised PMS Valuation Methodology 1, PDs 152–64, CDs 107–26, bar codes 3810691-01–13, 3810640-01–20 (Mar. 22, 2019) (“Revised PMS Methodology Memo.”). Multiple regression models use a variety of estimators, i.e., rules “for combining data to produce a numerical value for the population parameter” to estimate relationships. Wooldridge, Econometrics at 757, 848. Ordinary least squares is an estimator that obtains estimates by “minimizing the sum of squared residuals.” Id. at 854. The OLS Regression Model estimated the relationship between the domestic price of HRC, the dependent variable; and uneconomic capacity, iron ore prices, scrap metal pieces, aluminum prices, country specific exchange rates and country specific gross fixed capital formation (“GFCF”), the explanatory variables, for the selected countries from 2008 to 2017, using the following equation: ln(HRC á á á á á á á á á where “i” is the country, “t” is the year,“ ” is a countryspecific dummy variable, and “ ” is the error term. See Revised PMS Methodology Memo. at Ex. 1.1 p. 1; Final Decision Memo. at 65. A dummy variable “represents whether, in each time period, a certain event has occurred.” Wooldridge, Econometrics at 357. The error term is “the variable in a . . . multiple regression equation that contains unobserved factors which affect the dependent variable. The error term may also include measurement errors in the observed dependent or [explanatory] variables.” Id. at 848. Commerce cites to the Petitioners’ “model 18 (footnote continued) Court No. 20-00026 PUBLIC VERSION Page 24 Memo. at 65; Letter from Petitioners Re: Revised PMS Valuation Methodology 18– 22, Exs. 1.1, 1.7, PDs 152–64, CDs 107–26, bar codes 3810691-01–13, 3810640-01–20 (Mar. 22, 2019) (“Revised PMS Methodology Memo.”). The OLS Regression Model attempts to calculate the 2017 counterfactual HRC price using domestic HRC prices from eight countries, including India, assuming a utilization rate of 80%.19 Final Decision Memo. at 65. Rather than relying on the OLS Regression Model’s counterfactual HRC price to quantify the PMS adjustment, Commerce took the estimated regression coefficient for uneconomic capacity20 calculated by the OLS Regression Model and multiplied it by the “the percent reduction in uneconomic iteration number 11” at Exhibit 1.7a in the Regression Analysis, see Final Decision Memo. at 65 n.329, however, such exhibit does not exist. The court believes ]] prices instead Commerce relies on model iteration 11: “The OLS model using [[ of import AUVs for all countries (restricted to the 8 countries with available prices)” contained in exhibit 1.7 of the Revised PMS Methodology Memo. Revised PMS Methodology Memo. at Ex. 1.7 p. 1–2. Additionally, Commerce does not explain whether model iteration 11 accurately measures the relationship between HRC prices and the explanatory variables. 19 Commerce has determined that a capacity utilization rate of 80% is sufficient for sustaining profitable operations and the operational efficiency of the steel industry. Final Decision Memo. at 66. 20 Uneconomic capacity is “the amount of steel capacity in a given year in excess of the largest possible quantity of steel that may be demanded in that year (i.e., global capacity minus the highest global production ever experienced prior to that year).” Revised PMS Methodology Memo. at 12–13. The OLS Regression Model calculated an estimated regression coefficient for uneconomic capacity of -0.4338 meaning that “a 10 percent decrease in uneconomic capacity will result in a 4.338 percent increase in domestic Indian HRC prices.” Final Decision Memo. at 65; see also id. at Ex. 1.7 p. 1. Court No. 20-00026 PUBLIC VERSION Page 25 capacity that is required to reduce overall production capacity to the ‘implied capacity’ level,21 resulting in a 10.30% increase in Indian HRC prices.” Id. 65–66. Commerce’s decision to quantify a PMS adjustment using an ordinary least squares regression model is reasonable. Commerce asserts that ordinary least squares regression models are an acceptable means for quantifying PMS adjustments if the model includes a sufficient number of explanatory variables and data points accounting “for all relevant categories of factors from a price determination standpoint (i.e. supply and demand)” and minimizes the endogeneity problem,22 through proxies23 if necessary. Id. at 64. Commerce supports its assertion with record evidence recognizing ordinary least squares regression models as “the best unbiased estimator for determining a linear relationship between variables.” Final Decision Memo. at 64, 64 n.327 (citing Revised PMS Methodology Memo. at Ex. 1.1 p. 5 and Jeffrey M. Wooldridge, Econometrics at 101–02); see also Revised PMS The implied capacity level is the 2017 production amount divided by a capacity utilization rate of 80%. Final Decision Memo. at 65. 22 A linear regression suffers from an endogeneity problem if one of the explanatory variables is correlated with the error term. See Wooldridge, Econometrics at 86–87. 23 Occasionally when attempting to estimate and quantify the relationship between a dependent variable and explanatory variables using a multiple linear regression one or more of the explanatory variables may be unobservable, impossible to quantify, or the data may be missing. Id. at 308–09. In those cases, a proxy variable may be used in place of the explanatory variable that the creator of the linear regression would like to control for. Id. A proxy variable is a variable correlated with the unobservable explanatory variable. Id. 21 Court No. 20-00026 PUBLIC VERSION Page 26 Methodology Memo. at 6 (explaining that ordinary least squares is unbiased and more efficient than other estimators “provided the necessary assumptions are met”). Commerce’s reliance on the regression coefficient for uneconomic capacity of the counterfactual HRC price is also reasonable. Commerce explains that the regression coefficient for uneconomic capacity directly relates to the global overcapacity crisis,24 which it asserts is the primary cause of the alleged PMS. See Final Decision Memo. at 66. However, the counterfactual HRC price is dependent on five explanatory variables unrelated to the global overcapacity crisis. See id. Commerce concluded that the regression coefficient for uneconomic capacity captures the impact of the global overcapacity crisis on Indian HRC prices better than the counterfactual HRC price. Id. 19 U.S.C. § 1677b(e) allows Commerce to select a calculation methodology of its choosing, bound by the principles of reasonableness and substantial evidence. Commerce articulates a rational connection between its decision to use an ordinary least squares regression model to quantify a PMS adjustment via the regression coefficient for uneconomic capacity and the record evidence. Final Decision Memo. at 63–66. For those reasons, Commerce’s choice of calculation methodology is reasonable. Commerce explains that the OLS Regression Model’s uneconomic capacity variable “imperfectly” attempts to capture the “weight” that “global excess capacity, driven by Chinese excess capacity” places on the market. PMS Memo. at 29. Nonetheless, the uneconomic capacity variable is “consistent with expert, third-party research [confirming] a link between capacity and price.” Id. 24 Court No. 20-00026 PUBLIC VERSION Page 27 Commerce’s application of its selected methodology via the OLS Regression Model is unsupported by substantial evidence because Commerce does not adequately address record evidence detracting from the validity of the OLS Regression Model. Both Commerce and Petitioners rely upon Wooldridge, Econometrics to explain that the validity and efficiency of an ordinary least squares linear regression for time series data are informed by whether the input data returns results which satisfy the Gauss-Markov assumptions. See Final Decision Memo. at 63–64; Revised PMS Methodology Memo. passim, Ex. 1.6 (excerpt of Wooldridge, Econometrics); Wooldridge, Econometrics at 349–355. For ordinary least squares regressions based on time series data, the Gauss-Markov assumptions require the model to be linear in parameters, have no perfect collinearity between the explanatory variables,25 have an error term with a population mean of zero, have no serial correlation, have a When explanatory variables are correlated in a multiple regression model, multicollinearity is present in the model. See Wooldridge, Econometrics at 853. The Gauss-Markov assumptions for multiple linear regressions using time series data prohibit perfect multicollinearity, i.e., one explanatory variable is an exact linear function of one or more other explanatory variables, id. at 854, but some level of multicollinearity is expected in multiple linear regression models. id. at 94–95. Since the presence of multicollinearity does not violate one of the Gauss-Markov assumptions, the magnitude of an improper correlation is not well defined. Id. at 95. Multicollinearity can be problematic because as the correlation between two or more explanatory variables gets stronger, it becomes harder for the model to estimate the relationship between the dependent variable and each explanatory variable. See id. at 94–96. 25 Court No. 20-00026 PUBLIC VERSION Page 28 constant variance of the error term over time, and have normality.26 See Wooldridge, Econometrics at 349–55, 374. Garg challenges Commerce’s use of the OLS Regression Model arguing that it is unsupported by substantial evidence because it does not satisfy the Gauss-Markov assumptions which guarantee that ordinary least squares is the best linear unbiased estimator for the data set chosen. See Pls.’ Br. at 32–36 (“In sum, the OLS regression model did not constitute a sufficiently accurate and undistorted methodology needed to compute the PMS adjustment”); see also Wooldridge, Econometrics at 354. Commerce fails to explain whether and to what extent a significant reduction in the number of countries selected impacts the OLS Regression Model. Commerce explains that in response to comments made by the parties, it made three changes to the PMS adjustment calculation used in the preliminary results. Final Decision Memo. at 65. One of those changes was a reduction in the number of countries analyzed from 38 countries to eight. Id. A significant reduction in the sample size without further explanation calls into question whether the eight remaining If the model is linear in parameters, has no perfect collinearity between the explanatory variables, and has an error term with a population mean of zero, the model is procedurally unbiased, however bias may still be introduced through the sample selected. See Wooldridge, Econometrics at 87–88, 353. If the first five GaussMarkov assumptions are met, no alternative unbiased estimator will be better than the ordinary least squares estimator. See id. at 354. An OLS regression model has normality if the error term “is independent of the explanatory variables . . . and is normally distributed with zero mean and variance.” Wooldridge, Econometrics at 118, 355. 26 Court No. 20-00026 PUBLIC VERSION Page 29 countries are a representative sample of the population or cherry-picked to favorably manipulate the results of the OLS Regression Model. See Wooldridge, Econometrics at 324–26 (explaining that nonrandom sampling and sample size reduction can lead to biased results). Without further explanation, the court cannot determine if Commerce’s decision is reasonable. Commerce fails to address evidence proffered by Garg disputing the inverse relationship between uneconomic capacity and HRC prices. See PMS Rebuttal at 81– 83, Ex. 28 pp. 65–66; Pls.’ Br. at 34–35. Commerce argues that an “inverse relationship between steel overcapacity and HRC prices is an empirical fact” asserting Garg “offers no formable evidence that disputes the inverse relationship between uneconomic capacity and HRC prices.” Final Decision Memo. at 68–69. Yet, as part of its PMS Rebuttal, Garg submitted a study contradicting the inverse relationship between global steel overcapacity and HRC prices. Pls.’ Br. at 34–35; PMS Rebuttal Ex. 28. Commerce does not address this study in either the Final Decision Memo. or the PMS Memo. See Final Decision Memo.; PMS Memo. Commerce’s use of the regression coefficient for uneconomic capacity to calculate the PMS adjustment stems from its belief that the global overcapacity crisis is the primary cause of the alleged PMS in India. Final Decision Memo. at 66. In order to continue relying on the regression coefficient for uneconomic capacity Commerce must address the evidence submitted by Garg either by explaining why it disregarded the evidence or why its determination is reasonable in light of the evidence. Court No. 20-00026 PUBLIC VERSION Page 30 Nor does Commerce adequately respond to arguments proffered by Garg that the OLS Regression Model is skewed. Garg proffers evidence suggesting the OLS Regression Model is unstable and argues that some of the variables selected by Commerce are over inclusive, while other “critical” explanatory variables are omitted completely. Garg Tube’s PMS Comments at 3–4, Exs. 1–2, PD 237 CD 162–163 bar code 3879948-02 (Aug. 16, 2019) (“PMS Rebuttal Comments”); Garg Tube’s Admin. Case Br. at 65–70, PD 240, CD 164, bar code 3884484-01 (Aug. 27, 2019) (“Garg’s Agency Br.”). Garg’s argument suggests that due to the problems it identifies with the variables, the OLS Regression Model does not satisfy the Gauss-Markov assumptions for unbiasedness of the estimator, thus the OLS Regression Model cannot be the best estimator for the relationship between the explanatory variables and the dependent variable and any results it produces are of questionable validity. See Garg’s Agency Br. at 65–70. Commerce relies on gross fixed capital formation (“GFCF”) as a proxy variable for the explanatory HRC demand variable, arguing that GFCF is an appropriate proxy variable for demand because “it is indisputable that conditions that lead to changes in investment goods also lead to changes in demand for HRC.”27 Final Decision Memo. at 69. Related to the use of GFCF as a proxy for the demand variable, Garg argues that Commerce omitted critical explanatory variables from the OLS Garg’s argument that GFCF is an overly inclusive proxy for macroeconomic demand asks the court to reweigh the evidence. See Pls.’ Br. at 35. 27 Court No. 20-00026 PUBLIC VERSION Page 31 Regression Model. Garg’s Agency Br. at 68–70 (“Since the OLS regression model entirely fails to account for such fiscal/monetary/taxation factors, it is an oversimplified and incomplete model,” providing the sub-prime mortgage crisis as an example). Implicit in Garg’s argument is that the omission of fiscal, monetary, and taxation explanatory variables leads to omitted variable bias, a violation of the Gauss-Markov zero conditional mean assumption. See Garg’s Agency Br. at 65-70; Wooldridge, Econometrics at 350–51. The violation of this Gauss-Markov assumption calls into question the validity of the regression model. See Wooldridge, Econometrics at 88–94. Omitted variable bias occurs when a variable is correlated with one or more explanatory variables and is a determinant of the dependent variable but is omitted from the regression model. Id. If a variable meets both conditions, it is considered a relevant variable. Id. The exclusion of a relevant variable introduces bias28 into a regression model, distorting all other regression coefficients. See id. (explaining the impact of omitted variable bias on a multiple regression model). Omitted variable bias can be corrected by including the omitted variable in the regression.29 See id. at 88–91. Garg’s argument has merit. Commerce argues that demand for HRC is Bias is “[t]he difference between the expected value of an estimator and the population value that the estimator is supposed to be estimating.” Id. at 845. 29 The choice to include a relevant explanatory variable requires an analysis of the trade-off between bias and variance. Id. at 98. Introduction of the relevant variable may reduce bias while increasing the variance of the explanatory variables. Id. at 94–98. Variance of the explanatory variables is an important consideration because a larger variance results in a less precise estimation and less accurate hypotheses tests. Id. at 94. 28 Court No. 20-00026 PUBLIC VERSION Page 32 captured by the GFCF explanatory variable and admits that fiscal, monetary, and taxation policies “lead to changes in the demand for HRC.” Final Decision Memo. at 69. Commerce’s admission demonstrates a relationship between fiscal, monetary, and taxation policies and demand for HRC. It is unclear to the court how Commerce isolates the impact of fiscal, monetary, and taxation policies on the explanatory HRC demand variable. Furthermore, Commerce asserts that India’s fiscal policies such as safeguard duties and antidumping duties were put in place to alter the price of domestic HRC, the dependent variable. Id. at 23. Commerce’s explanation suggests fiscal, monetary, and taxation polices may be relevant variables and if so, failure to control for them introduces bias into the OLS Regression Model. The court cannot determine if Commerce’s decision to exclude variables for fiscal, monetary, and taxation policies is reasonable because Commerce does not explain whether the variables are relevant or whether their omission introduced an unacceptable amount of bias into the regression. The OLS Regression Model is designed to examine how the price of HRC is impacted by the identified explanatory variables. Id. at 65. Commerce seeks to isolate the impact that uneconomic capacity has on the price of HRC using the OLS Regression Model in order to calculate the appropriate PMS adjustment. Id. Yet, if a relevant variable has been omitted from the OLS Regression Model, the identified explanatory variables will suffer from distortion. See Wooldridge, Econometrics at 88–91. On remand, Commerce must explain what impact, if any, the exclusion of these variables has on the OLS Court No. 20-00026 PUBLIC VERSION Page 33 Regression Model.30 To the extent that the exclusion of these variables introduces bias into the OLS Regression Model, Commerce must either include them or provide an adequate explanation for their exclusion. If the variables are not relevant, Commerce must explain why. Finally, Garg submitted evidence demonstrating that changing the time period analyzed by the regression significantly impacts the relationship between the explanatory variable and dependent variable.31 PMS Rebuttal Comments at 3–4, Exs. 1–2 (“the coefficient for uneconomic capacity . . . was negative for the 2008-17 and 2009-17 runs, but then abruptly switched to a positive sign for the runs from 2010-17 and onward”). Garg’s argument suggests that the explanatory variables time and uneconomic capacity are highly correlated. Pls.’ Br. at 36; see also Final Decision Memo. at 54 (arguing that the explanatory variables are “highly correlated” with one another). In response, Commerce explained that because the relationship between The OLS Regression Model is by necessity a simplification of the observed economy, limited by the availability of economic data and Commerce’s resources. See Final Decision Memo. at 63–64. The court’s remand order does not require Commerce to control for every possible explanatory variable. It only requires Commerce to control for relevant explanatory variables or explain to the court why the exclusion of a relevant explanatory variable does not result in a significantly less accurate PMS adjustment calculation. 31 The impact of the years regressed is best defined as an influential observation. An observation is influential “if dropping it from the analysis changes the key OLS estimates by a practically ‘large’ amount.” Wooldridge, Econometrics at 326–27. Although ordinary least squares regressions are sensitive to influential observations, there is cause for concern when slight modifications of the sample result in large changes to the estimates. See id. 30 Court No. 20-00026 PUBLIC VERSION Page 34 uneconomic capacity and domestic prices is constant over time, the inclusion or exclusion of data from 2008 and 2009 does not change the overall nature of the relationship. Final Decision Memo. at 69. However, Garg provides evidence in its PMS rebuttal comments demonstrating that the overall nature of the relationship does change with the inclusion or exclusion of the 2008 and 2009 data. PMS Rebuttal Comments at 3–4, Exs. 1–2; see also Garg’s Agency Br. at 86–Br. (explaining the relevance of the regression output in the PMS Rebuttal Comments.); Pls.’ Br. at 36 (same). Depending on the time period examined, the direction of the relationship between uneconomic capacity and HRC price changes. PMS Rebuttal Comments at 3–4, Exs. 1–2; see also Garg’s Agency Br. at 86–87. The directional change in the relationship depending on the time period examined suggests a correlation between time and uneconomic capacity. Commerce must either reconsider its determination or explain its determination and support it with substantial evidence. Commerce explains that the inclusion of the 2008 and 2009 data is necessary to demonstrate the origins of the overcapacity crisis and avoid the degrees of freedom problem, but it does not address Garg’s argument regarding the correlation between the explanatory variables. See Final Decision Memo. at 69. Until Commerce addresses Garg’s correlation argument, the court cannot say if Commerce’s decision to use the OLS Regression Model is reasonable in light of the evidence detracting from its determination. Court No. 20-00026 PUBLIC VERSION Page 35 CONCLUSION For the foregoing reasons, the court sustains Commerce’s determinations to remove the PMS adjustment from the sales-below-cost test and rely on neutral facts available for the missing cost of production data. Commerce’s PMS determination and adjustment are remanded. In accordance with the foregoing, it is ORDERED that Commerce’s remand redetermination is remanded for further explanation or reconsideration consistent with this opinion; and it is further ORDERED that Commerce shall file its remand redetermination with the court within 90 days of this date; and it is further ORDERED that the parties shall have 30 days thereafter to file comments on the remand redetermination; and it is further ORDERED that the parties shall have 30 days to file their replies to the comments on the remand redetermination; and it is further ORDERED that the parties shall have 14 days thereafter to file the Joint Appendix; and it is further ORDERED that Commerce shall file the administrative record within 14 days of the date of filing its remand redetermination. /s/ Claire R. Kelly Claire R. Kelly, Judge Dated: March 11, 2022 New York, New York
Some case metadata and case summaries were written with the help of AI, which can produce inaccuracies. You should read the full case before relying on it for legal research purposes.
This site is protected by reCAPTCHA and the Google Privacy Policy and Terms of Service apply.
