Slip Op. No. 24-102
UNITED STATES COURT OF INTERNATIONAL TRADE
OFFICINE TECNOSIDER SRL,
Plaintiff,
v. Before: Stephen Alexander Vaden, UNITED STATES, Judge
Defendant, Court No. 1:23-cv-00001 (SAV)
and
NUCOR CORPORATION
Defendant-Intervenor.
OPINION
[Remanding Commerce’s Remand Determination.]
Dated: September 17, 2024
Pierce J. Lee, Crowell & Moring LLP, of Washington, DC, for Plaintiff Officine Technosider Srl. With him on the brief was Daniel J. Cannistra.
Augustus J. Golden, Trial Attorney, Commercial Litigation Branch, Civil Division, U.S. Department of Justice, of Washington, DC, for Defendant United States. With him on the brief were Brian M. Boynton, Principal Deputy Assistant Attorney General, Patricia M. McCarthy, Director, Commercial Litigation Branch, Tara K. Hogan, Assistant Director, Commercial Litigation Branch, and Ashlande Gelin, Attorney, Office of the Chief Counsel for Trade Enforcement and Compliance, U.S. Department of Commerce.
Stephanie M. Bell, Wiley Rein LLP, of Washington, DC, for Defendant-Intervenor Nucor Corporation. With her on the brief were Alan H. Price, Christopher B. Weld, and Jeffrey O. Frank. Court No. 1:23-cv-00001 (SAV) Page 2
Vaden, Judge: Before the Court is the U.S. Department of Commerce’s
(Commerce) remand determination for the 2020–21 administrative review of the
antidumping order for steel plate from Italy. At issue is whether Commerce should
depart from its normal annual average cost methodology and instead use an
alternative quarterly cost methodology for Plaintiff Officine Technosider SRL
(Officine). Commerce sought a voluntary remand to reconsider if it should apply the
alternative methodology. On remand, Commerce compared quarterly trends in sales
for the U.S. and home markets with the cost of manufacturing for certain control
numbers and determined that using the alternative methodology is warranted.
However, all U.S. sales took place in one quarter, making it difficult to decipher a
trend for U.S. sales prices. Only one control number shared between the U.S. and
home markets is also among the five highest-selling control numbers in both markets
— further complicating Commerce’s task. And a past Commerce decision points to
another potential way to analyze the data. Because Commerce failed to properly
respond to these shortcomings, its decision lacks substantial evidentiary support and
will be REMANDED for further explanation.
BACKGROUND
I. Procedural Background
This case concerns a challenge to Commerce’s Final Results in the
administrative review of the antidumping order on certain carbon and alloy steel cut-
to-length plate from Italy, covering entries from May 1, 2020, through April 30, 2021 Court No. 1:23-cv-00001 (SAV) Page 3
(the Period of Review). See Certain Carbon and Alloy Steel Cut-To-Length Plate from
Italy: Final Results of Antidumping Duty Administrative Review and Final
Determination of No Shipments; 2020-2021, 87 Fed. Reg. 75,219 (Dep’t of Com. Dec.
8, 2022) (Final Results), and accompanying Issues and Decisions Mem. (IDM), J.A. at
2,305, ECF No. 44. At issue is whether Commerce should use its normal annual
weighted-average cost methodology for the entire Period of Review or an alternative
methodology, which examines costs in shorter periods to minimize potential
distortions, known as the quarterly cost methodology.1 Officine is an Italian producer
and exporter of carbon and alloy steel cut-to-length plates. Pl.’s Mot. for J. on Agency
R. and Supp. Opening Br. (Pl.’s Mot.) at 6, ECF No. 24. Officine challenged
Commerce’s Final Results. See Compl., ECF No. 11. The Court granted Defendant-
Intervenor Nucor Corporation’s (Nucor) Consent Motion to Intervene to support
Commerce’s original determination. Order Granting Intervention, ECF No. 20.
Officine filed its Motion for Judgment on the Agency Record, focusing on
Commerce’s use of the annual weighted-average cost methodology instead of the
quarterly cost methodology. Pl.’s Mot. at 2, ECF No. 24 (identifying five issues, all
dealing with the quarterly cost methodology); see also Certain Carbon and Alloy Steel
Cut-To-Length Plate from Italy: Preliminary Results of Antidumping Duty
Administrative Review and Preliminary Determination of No Shipments; 2020-2021,
1A more detailed explanation of Commerce’s test for applying the quarterly cost methodology is found in the Discussion portion of this opinion. See infra § I (Legal Framework). Court No. 1:23-cv-00001 (SAV) Page 4
87 Fed. Reg. 34,246 (Dep’t of Com. June 6, 2022), and accompanying Prelim. Decision
Mem. at 19, J.A. at 2,289, ECF No. 44 (“[W]e determined that our quarterly cost
methodology is not warranted …. Therefore, we applied our standard methodology
of using annual average costs ….”); IDM at 31, J.A. at 2,335, ECF No. 44 (continuing
to reject requests to use the quarterly cost methodology). Officine contended that
Commerce established a “clear and predictable practice” for when it would employ
the quarterly cost methodology; Officine met both criteria for its use; and Commerce
failed to explain why it declined to use it. Pl.’s Mot. at 14, 18–20, 24–25, ECF No. 24.
It asked the Court to remand the case for Commerce to reconsider its decision. Id. at
34.
After reviewing Officine’s brief, Commerce requested a voluntary remand to
“reconsider … application of the quarterly cost methodology … and, if appropriate,
revise the dumping margin calculation.” Def.’s Mot. for Voluntary Remand at 2, ECF
No. 26 (Def.’s Mot.). Commerce explained that it examines costs averaged over the
entire Period of Review unless there are: (1) “significant cost variations” and (2)
“those variations are linked to changes in sales prices.” Id. at 2 (quoting Rebar Trade
Action Coal. v. United States (RTAC), 45 CIT __, 503 F. Supp. 3d. 1295, 1299 (2021)).
When these criteria are met, Commerce may apply a quarterly cost methodology. Id.
(citing RTAC, 45 CIT __, 503 F. Supp. 3d. at 1299). Commerce believed that an
important part of its required analysis was missing from the record and requested
the Court allow it to consider anew the information Officine submitted. Id. at 1. The Court No. 1:23-cv-00001 (SAV) Page 5
Court granted the unopposed Motion on May 15, 2023. Order Granting Def.’s Mot.
for Voluntary Remand (Remand Order), ECF No. 29.
II. Commerce’s Voluntary Remand and the Present Dispute
Commerce filed its Remand Results on September 12, 2023, and the Parties
submitted comments.2 Final Results of Redetermination Pursuant to Ct. Remand
Order (Remand Results), ECF No. 32; see Def.-Int.’s Comments in Opp’n to Remand
Results (Def.-Int.’s Br.), ECF No. 38; Def.’s Comments on Remand Results (Def.’s Br.),
ECF No. 47. On remand, Commerce reopened the record and requested that Officine
file additional information. Remand Results at 2, ECF No. 32. It requested this data
to determine if Officine’s foreign sales were made below that product’s cost of
production. Id. at 3. Commerce then applied the first part of its test, examining
whether Officine experienced significant changes in the cost of production over the
Period of Review. Id. at 4. It explained that the difference between the highest and
lowest cost of manufacturing during the Period of Review must be at least 25 percent
to be a “significant change.” Id. The agency determined that Officine experienced
such a change. Id.
Next, Commerce examined whether there were linkages between the cost of
manufacturing and sales prices — the second part of its test to determine whether to
depart from its standard methodology and consider the data on a quarterly rather
2 Officine filed a letter in lieu of responsive comments fully supporting Commerce’s determination. See ECF No. 42. Court No. 1:23-cv-00001 (SAV) Page 6
than yearly basis. Id. at 4–5. Commerce “compared weight-averaged U.S. market
and [home] market sales prices, by quarter, with the reported quarterly [cost of
manufacturing] for high volume [control numbers].”3 Id. at 4. It stated that it used
its “standard analysis” to examine two pools consisting of “the five largest sales
volume [control numbers] in the home market and five largest sales volume [control
numbers] in the U.S. market to determine whether there was a reasonable correlation
between changing costs and sales prices from quarter to quarter.” Id. at 10. Officine’s
U.S. sales consisted of only five control numbers, and those sales all occurred in the
same quarter so that “it was not possible to perform the linkage analysis for the U.S.
sales [control numbers].” Draft Remand Calculation Mem. at 2–3, J.A. at 83,020–21,
ECF No. 44. Commerce could only compare Officine’s sales in its home market of
Italy with the cost of manufacturing and look for linkages there. Id. at 3, J.A. at
83,021.
Based on this limited data, Commerce concluded that Officine’s Italian sales
prices and cost of manufacturing showed “a reasonable correlation.” Remand Results
at 4–5, 10, ECF No. 32; id. at 5 n.13 (citing Draft Remand Calculation Mem. at 2–3,
3 “Control number,” often referred to by the contraction “CONNUM,” denotes a unique product based on relevant physical characteristics. To ensure that Commerce is comparing like products in the home and U.S. markets, it asks respondents to sort merchandise according to key differentiating categories with each number in the product’s control number corresponding to physical characteristic groupings particular to the merchandise under review. Xi’an Metals & Minerals Imp. & Exp. Co. v. United States, 45 CIT __, 520 F. Supp. 3d 1314, 1321 n.4 (2021). As a simple shorthand, a reader may substitute “product” any time he reads “control number” or “CONNUM.” Court No. 1:23-cv-00001 (SAV) Page 7
J.A. at 83,020–21, ECF No. 44 and Attachment III, J.A. at 83,383, ECF No. 46). It
found that the quarterly costs of manufacturing and Italian sales prices “mov[ed] in
the same direction for the majority of the [Period of Review].” Id. at 10 n.35.
Commerce determined, “[T]here is linkage between [Officine’s] changing sales prices
and [cost of manufacturing] during the [Period of Review].” Id. at 5. In other words,
as manufacturing costs increased or decreased, the prices Officine charged in Italy
did the same. Because Commerce found that Officine’s data satisfied both elements
of its test, Commerce concluded that it was appropriate to perform its final
computations using the alternative quarterly cost methodology. Id. Commerce
assigned Officine a revised dumping margin of zero percent after applying these
methodological revisions — a dramatic decline from 20.44 percent in the original
determination. Compare id. at 2, with Final Results, 87 Fed. Reg. at 75,220.
Commerce’s flipped determination has effectively flipped the parties in this
case. Although Nucor originally joined this suit as Defendant-Intervenor to help
Commerce support its original determination, Nucor now opposes Commerce’s
remand determination and asks this Court to send it back to Commerce to try again.
Nucor objects to Commerce’s use of the alternative quarterly cost analysis to examine
Officine’s data. It contends that Commerce improperly conducted its two-part test to
determine if quarterly data should be used. Nucor only contests the second step ––
the reasonable linkage analysis. Oral Arg. Tr. at 84:17–20, ECF No. 51 (THE
COURT: “[Y]ou don’t object that the 25 percent threshold has been met? We’re only Court No. 1:23-cv-00001 (SAV) Page 8
talking about the linkages?” MS. BELL: “That’s correct, Your Honor.”). Nucor
argues that Commerce used its “typical approach,” but “the agency’s rote application”
did not consider “specific facts …, including information that undermines the
reasonableness of the agency’s reliance on this approach and ultimate conclusion.”
Def.-Int.’s Br. at 4, ECF No. 38. It first argues that the linkage test is unreliable
because Commerce was unable to analyze U.S. sales and only focused its analysis on
sales in Officine’s Italian home market. Id. (citing Draft Remand Calculation Mem.
at 2, J.A. at 83,020, ECF No. 44). Nucor also argues Commerce’s home market
analysis should have focused on those products sold in both the U.S. and Italy –– as
opposed to the five largest-selling home market control numbers. Id.; Final
Calculation Mem. at 1, J.A. at 83,396, ECF No. 44. Because of Commerce’s myopic
focus, Nucor asserts “there is evidence of a correlation between cost and price for only
half of the relevant data” –– the home market sales –– “and no evidence of a
correlation for the other half” –– the U.S. market sales. Def.-Int.’s Br. at 4, ECF No.
38. For both these objections, Nucor claims that Commerce failed to “meaningfully
discuss or consider the record before it and how the specific facts of this case support
its cost methodology.” Id. at 5.
Commerce responds that it used its “standard practice of examining the top
five selling [control numbers]” in the U.S. and home markets. Def.’s Br. at 10, ECF
No. 47. It acknowledges it did not conduct a linkage analysis for the U.S. sales but
contends that it could still find a reasonable linkage between changes in home market Court No. 1:23-cv-00001 (SAV) Page 9
sales prices and changes in the cost of manufacturing. Id. at 9–10. Responding to
Nucor’s objection that Commerce should have analyzed those products that had sales
in both the U.S. and Italian markets, Commerce retorts: (1) It relied on its
established test for applying the quarterly cost methodology; (2) Nucor’s approach is
impractical because it requires running the linkage analysis after the margin
calculation; and (3) even under Nucor’s approach, Commerce could still find
reasonable linkages in the home market control numbers that overlap with the U.S.
market control numbers. See id. at 10–11 (citing Remand Results at 11, ECF No. 32);
id. at 13 (citing Final Calculation Mem. at 2, J.A. at 83,397, ECF No. 44).
The Court held oral argument on March 27, 2024. ECF No. 49. The parties
debated whether Commerce could find a reasonable linkage based on home market
sales data alone and if Commerce’s explanation is supported by substantial evidence.
See, e.g., Oral Arg. Tr. at 25:23–25, 29:17–20, 37:9–14, ECF No. 51. No party
identified a prior case or administrative review where Commerce conducted a linkage
analysis solely based on quarterly home market data. Id. at 14:25–15:11, 50:19–51:8,
52:16–23. The Court also sought to discern how Commerce’s linkage analysis works
under normal circumstances, i.e., when there are multiple quarters of sales data for
control numbers sold in both markets. Id. at 6:6–16, 7:6–13 (discussing the linkage
analysis in an “optimal situation”). The case is now ripe for decision. Court No. 1:23-cv-00001 (SAV) Page 10
JURISDICTION AND STANDARD OF REVIEW
This Court has jurisdiction under 28 U.S.C. § 1581(c), which grants authority
to review actions contesting final determinations in antidumping reviews. The Court
must set aside Commerce’s remand determination if it is found to be “unsupported
by substantial evidence on the record, or otherwise not in accordance with law ….”
19 U.S.C. § 1516a(b)(1)(B)(i). “[T]he question is not whether the Court would have
reached the same decision on the same record[;] rather, it is whether the
administrative record as a whole permits Commerce’s conclusion.” New Am. Keg v.
United States, 45 CIT __, 2021 Ct. Intl. Trade LEXIS 34, at *15 (Mar. 23, 2021). When
reviewing agency determinations, findings, or conclusions for substantial evidence,
the Court assesses if the agency’s action is reasonable given the record as a whole.
Nippon Steel Corp. v. United States, 458 F.3d 1345, 1350–51 (Fed. Cir. 2006); see also
Universal Camera Corp. v. NLRB, 340 U.S. 474, 488 (1951) (“The substantiality of
evidence must take into account whatever in the record fairly detracts from its
weight.”). The Federal Circuit describes “substantial evidence” as “such relevant
evidence as a reasonable mind might accept as adequate to support a conclusion.”
DuPont Teijin Films USA, LP v. United States, 407 F.3d 1211, 1215 (Fed. Cir. 2005)
(quoting Consol. Edison Co. v. NLRB, 305 U.S. 197, 229 (1938)). Additionally,
“results of a redetermination pursuant to court remand are … reviewed for
compliance with the … remand order.” Ellwood City Forge Co. v. United States, 47 Court No. 1:23-cv-00001 (SAV) Page 11
CIT __, 2023 Ct. Intl. Trade LEXIS 113, at *7 (July 24, 2023) (internal quotation
marks omitted).
DISCUSSION
At issue is the adequacy of Commerce’s explanation for why its standard test
for applying a quarterly cost methodology is reliable under nonstandard facts.
Commerce normally calculates an annual weighted-average cost for the Period of
Review but may apply an alternative quarterly cost methodology if it finds (1)
significant changes in the cost of manufacturing during the Period of Review and (2)
evidence of linkage between changes in the cost of manufacturing and sales prices.
No party disputes that the first criterion is met. Although the quarterly cost
methodology may be an established alternative to the expected method, Commerce
must explain why the data present here warrant its use. Commerce’s rote application
of its test to a non-traditional dataset in the face of specific objections from Nucor
does not suffice. Commerce did not adequately explain: (1) why focusing solely on
Italian sales is a reliable indicator of linkage for U.S. sales, (2) why it chose not to
follow its precedent in Ferrovanadium from Korea and focus its analysis on products
jointly sold in both the Italian and U.S. markets, and (3) how it analyzed the data it
did examine to determine there was proper linkage between the cost of
manufacturing and the sales price. The Court does not hold — and Nucor does not
argue — that Commerce cannot find the necessary linkage exists when there is only
one quarter of U.S. sales data. However, Commerce must set forth a new Court No. 1:23-cv-00001 (SAV) Page 12
determination supported by substantial evidence to explain why its test is reliable in
this case.
I. Legal Framework
This case involves antidumping duties, which are imposed on merchandise
“sold in the United States at less than its fair value.” 19 U.S.C. § 1673(1).
Antidumping duties equal “the amount by which the normal value exceeds the export
price … for the merchandise.” Id. § 1673(2)(B). That amount is the “dumping
margin.” 19 U.S.C. § 1677(35)(A). Normal value is the home market price, and the
export price is the U.S. price. See Nagase & Co. v. United States, 47 CIT __, 628 F.
Supp. 3d 1326, 1331 (2023) (citing Koyo Seiko Co. v. United States, 258 F.3d 1340,
1342 (Fed. Cir. 2001)). Commerce calculates normal value “based on home market
sales … made ‘in the ordinary course of trade.’” RTAC, 45 CIT __, 503 F. Supp. 3d.
at 1301–02 (quoting 19 U.S.C. § 1677b(a)(1)(B)(i)). However, Commerce disregards
home market sales made at prices below the cost of production. Id. (citing 19 U.S.C.
§§ 1677b(b)(1), 1677(15)(A)). Thus, Commerce needs to determine the cost of
production in order to choose appropriate home market sales against which to
compare sales in the U.S. market.
The statute sets no time period Commerce must examine to calculate the cost
of production and consequently no time period “over which a respondent’s various
costs must … be averaged.” Pastificio Lucio Garofalo, S.p.A. v. United States, 35 CIT
630, 633 (2011) (citing 19 U.S.C. § 1677b(b)), aff’d sub nom. Pastificio Lucio Garofalo, Court No. 1:23-cv-00001 (SAV) Page 13
S.P.A. v. United States, 469 F. App’x 901 (Fed. Cir. 2012). Commerce’s “normal
practice is to calculate an annual weighted-average cost for the [Period of Review].”
Remand Results at 3, ECF No. 32; see also Def.-Int.’s Br. at 3, ECF No. 38. But
Commerce may “depart from its normal practice,” RTAC, 45 CIT __, 503 F. Supp. 3d
at 1302, and use “alternative cost averaging period[s]” –– usually quarterly –– when
its normal method “would distort the dumping analysis due to significant cost
changes.” See Pastificio, 35 CIT at 634; see also Oral Arg. Tr. at 46:7–9, ECF No. 51
(MR. GOLDEN: “[T]his is a methodology not to determine the dumping margins, but
to determine what data gives … the accurate dumping margin.”).
Commerce uses the alternative quarterly methodology if (1) there are
significant changes in the cost of manufacturing during the Period of Review and (2)
“record evidence indicates that sales made during the shorter cost-averaging periods
[can] be reasonably linked with [the cost of manufacturing] during the same shorter
cost-averaging periods.” Remand Results at 3, ECF No 32; see also Def.-Int.’s Br. at
3, ECF No. 38. In cases such as here where the Period of Review is twelve months,
Commerce has established a set threshold for the changes in the cost of
manufacturing to be “significant.” The difference between the highest and lowest
costs of manufacturing during the Period of Review must be a minimum of 25 percent.
Remand Results at 4, ECF No 32 (citing e.g., Stainless Steel Plate in Coils from
Belgium: Final Results of Antidumping Duty Administrative Review, 73 Fed. Reg.
75,398 (Dep’t of Com. Dec. 11, 2008), and accompanying IDM at cmt. 4). No party Court No. 1:23-cv-00001 (SAV) Page 14
disputes that the significance test is met. Oral Arg. Tr. at 15:18–16:8, 84:17–20, ECF
No. 51.
Commerce next analyzes if the changes in sales prices and the cost of
manufacturing are “reasonably linked.” Remand Results at 4, ECF No 32 (citing e.g.,
Stainless Steel Sheet and Strip in Coils from Mexico: Final Results of Antidumping
Duty Administrative Review, 75 Fed. Reg. 6,627 (Dep’t of Com. Feb. 10, 2010), and
accompanying IDM at cmt. 6). Absent a surcharge or similar pricing mechanism that
provides a direct linkage, Commerce examines the data to determine if “changes in
selling prices reasonably correlate to changes in unit costs.” Certain Welded Carbon
Steel Pipe and Tube from Turkey: Notice of Final Results of Antidumping Duty
Administrative Review (Tube from Turkey), 76 Fed. Reg. 76,939 (Dep’t of Com. Dec.
9, 2011), and accompanying IDM at 4; see also Oral Arg. Tr. at 8:3–16, ECF No. 51
(MR. GOLDEN: “[Commerce is] looking for … a linkage, a correlation, something
that reasonably shows that sale prices and cost of manufacturing are linked …. [A]re
these numbers … trending together …? When [both numbers] go up … do they …
seem to go up at about the same rate?”). Commerce examines sales and cost of
manufacturing data for two pools of control numbers, i.e., products: the home market
and U.S. market. Commerce typically analyzes cost and price trends for the five most
frequently sold home market control numbers and the five most frequently sold U.S.
control numbers. Tube from Turkey, IDM at 4; see SeAH Steel Corp. v. United States,
34 CIT 605, 612 (2010) (analyzing quarterly average price and cost changes for the Court No. 1:23-cv-00001 (SAV) Page 15
five largest U.S. and home market control numbers). But see Ferrovanadium from
the Republic of Korea: Final Determination of Sales at Less Than Fair Value
(Ferrovanadium), 82 Fed. Reg. 14,874 (Dep’t of Com. Mar. 23, 2017), and
accompanying IDM at 24 n.109 (identifying Commerce’s standard test for using a
quarterly cost methodology but deviating when only one control number was sold in
the U.S. market).
For each control number, Commerce examines the quarterly average prices
and cost of manufacturing for reasonable linkage. See, e.g., Tube from Turkey, IDM
at 4. Its analysis is holistic. See Ferrovanadium, IDM at 25 (“[L]inkage does not
require direct traceability between specific sales and their specific production costs,
but rather relies on whether there are elements which would indicate a reasonably
positive correlation between the underlying costs and the final sales prices charged
by a company.”). Elements indicating reasonable linkage may include: (1) the
relative magnitude of the changes in the price and cost; (2) whether, from quarter-to-
quarter, the prices and costs moved in the same direction; and (3) whether the
respective slope lines for the quarterly prices and costs consistently trended together.
See id. Commerce explained at oral argument that, although it examines the entire
pool of ten control numbers, it starts the analysis with the control numbers
representing the most sales on a dollar-value basis because they are the best
indicators of linkage. Oral Arg. Tr. at 13:6–11, ECF No. 51 (“[I]f we see trends in the
largest value sales, but [not] … in the smaller control numbers …, Commerce would Court No. 1:23-cv-00001 (SAV) Page 16
not be dissuaded [from finding a reasonable linkage].”). Commerce can also still find
a “reasonable linkage” when an element cuts against doing so. See, e.g., Tube from
Turkey, IDM at 4 (noting that the sales price and the cost of pipe did not always trend
together from quarter to quarter but that the magnitude of changes and consistent
slope line trends still indicated reasonable linkage).
Commerce has never addressed if the linkage analysis is reliable when
Commerce can only analyze home market sales. Oral Arg. Tr. at 14:25–15:13, ECF
No. 51 (admitting that no party cited “a case where [Commerce] was forced to deal
with a situation … where … one of the markets only had one quarter of data”); id. at
50:19–51:8, 52:16–23. And when there is only one quarter of sales data, Commerce
cannot examine normal elements in its linkage analysis such as the magnitude of
price changes quarter-to-quarter, the direction of those changes, or if the slope line
moves in the same direction as costs. See Draft Remand Calculation Mem. at 2–3,
J.A. at 83,020–21, ECF No. 44; Remand Results, Nucor’s Comments, at 8, ECF No.
32 (“[A] price trend cannot be calculated from a single data point.”). Here there is
only one quarter of U.S. sales data. The question before the Court is what this means
for Commerce’s linkage analysis. Draft Remand Calculation Mem. at 2–3, J.A. at
83,020–21, ECF No. 44. Court No. 1:23-cv-00001 (SAV) Page 17
II. Commerce’s Quarterly Cost Methodology Analysis in the Present Case
Nucor challenges Commerce’s determination to use its alternative quarterly
cost methodology to analyze Officine’s data. Under an optimal situation, Commerce
examines two pools of five control numbers each in the home market and U.S. market
to determine if there is a reasonable linkage between price and cost of manufacturing.
Remand Results at 10, ECF No. 32. But see Ferrovanadium, IDM at 25. In this case,
the U.S. pool contains only one quarter’s worth of sales data. Draft Remand
Calculation Mem. at 2–3, J.A. at 83,020–21, ECF No. 44. Nucor agrees that
Commerce followed its typical test but argues that its “rote application here failed to
take into account the specific facts of this record, including information that
undermines the reasonableness of the agency’s reliance on this approach and
ultimate conclusion.” Def.-Int.’s Br. at 4, ECF No. 38. The company raises two
objections based on (1) the absence of usable U.S. sales data in the linkage analysis
and (2) the limited overlap between U.S. market control numbers and home market
control numbers. Id. at 4–5. The Court takes each objection in turn.
Nucor cites to a past case where Commerce changed its analysis when the data
created questions about whether the normal test was appropriate. Id. at 6–7
(discussing Ferrovanadium). In Ferrovanadium, a respondent produced three
products during the period of review but sold only one in the U.S. market.
Ferrovanadium, IDM at 24 n.109. The lone U.S. control number was also the highest Court No. 1:23-cv-00001 (SAV) Page 18
sales volume control number in the home market. Id. at 24. Commerce based its
linkage analysis on that control number only. Id. at 25.4 It compared quarterly prices
and the cost of manufacturing, found a correlation, and determined that there was
reasonable linkage. Id. The decision did not address whether the other home market
control numbers factored into Commerce’s analysis, id. at 21–26, nor did the decision
explain what percentage of sales the one shared control number represented in the
home market. Id. The decision does say that the one shared product had the largest
sales in both markets. Id. at 25 n.109.
Nucor maintains that Ferrovanadium is relevant to both its objections because
Commerce (1) modified its test when faced with a non-standard data set and (2)
focused on the overlap between home market and U.S. market control numbers in its
linkage analysis — neither of which it did here. Oral Arg. Tr. at 66:14–67:12, ECF
No. 51; see also Def.-Int.’s Br. at 7, ECF No. 38 (“Commerce considered the specific
facts of [Ferrovanadium] and took a modified approach to address a unique factual
situation. This is what Nucor has asked for, and Commerce has failed to do, here.”).
The company argues that Commerce’s failure to address Ferrovanadium and how the
non-standard data set may have affected the reliability of the linkage analysis
4In Ferrovanadium, Commerce also conducted the significant cost change analysis based on the single shared control number. Ferrovanadium, IDM at 24. Because no party contests Commerce’s determination that the significant change test is met, the Court only focuses on the second prong, the reasonable linkage analysis. Oral Arg. Tr. at 15:18–16:8, 84:17–20, ECF No. 51. Court No. 1:23-cv-00001 (SAV) Page 19
renders the determination unsupported by substantial evidence. Oral Arg. Tr. at
65:5–10, ECF No. 51.
A. Commerce failed to address whether its usual linkage test remains reliable
here, whether the absence of usable U.S. market data undermines its determination,
and whether this case warrants a non-standard methodology, as in Ferrovanadium.
See Bowman Transp., Inc. v. Ark.-Best Freight Sys., Inc., 419 U.S. 281, 285–86 (1974)
(“[W]e may not supply a reasoned basis for the agency’s action that the agency itself
has not given ….”). The substantial evidence standard requires that Commerce
“articulate [a] rational connection between the facts found and the choice made.”
Burlington Truck Lines v. United States, 371 U.S. 156, 168 (1962). Further, “[t]he
substantiality of evidence must take into account whatever in the record fairly
detracts from its weight.” Universal Camera Corp., 340 U.S. at 488. Nucor does not
contend that Commerce can never conduct a linkage analysis using only home market
quarterly data. Oral Arg. Tr. at 60:10–14, ECF No. 51; id. at 21–24 (THE COURT:
“You haven’t made the strong form argument [that Commerce] can never do that.
You’ve just said, here, I’m not convinced; where’s your explanation?” MS. BELL:
“Correct, Your Honor.”). Instead, it attacks the adequacy of Commerce’s explanation.
Remand Results at 10–11, ECF No. 32 (“Nucor further argues that Commerce is
unable to conduct a trend analysis given that [Officine] only made U.S. sales during Court No. 1:23-cv-00001 (SAV) Page 20
the third quarter of the [Period of Review] and thus, none of the [control numbers]
sold in the U.S. support a finding of linkage between costs and prices.”).
First, Commerce repeatedly argues that it used its “standard” practice to
determine that the quarterly cost methodology is warranted. See, e.g., Remand
Results at 10, ECF No. 32. This is misleading. Commerce’s normal practice is to
examine the data over the entire period of review. Id. at 3. Officine is asking for an
exception from the normal methodology so that the data can be examined quarterly
instead of annually. Second, Commerce has failed to explain why it is confident that
its test shows reasonable linkage between the cost of manufacturing and sales prices
in both the home market and U.S. market when it only analyzed home market data.
Commerce ideally examines cost of manufacturing and sales data from ten control
numbers — five from the U.S. market and five from the home market — for multiple
quarters, giving more weight to higher sales volume control numbers. Id. at 10, ECF
No. 32; Oral Arg. Tr. at 13:6–11, ECF No. 51. No party has identified a prior case
where Commerce found a reasonable linkage based solely on home market sales data.
Oral Arg. Tr. at 14:25–15:13, 50:19–51:8, 52:16–23, ECF No. 51 (acknowledging this
to be the case). Thus, Commerce cannot claim it has used its standard methodology
when it has failed to analyze the home market data as part of its analysis.
When asked where Commerce addressed the absence of U.S. sales data,
counsel directed the Court to this passage: Court No. 1:23-cv-00001 (SAV) Page 21
We also find that, while Commerce may not be able to complete a trend analysis for U.S. prices due to the timing of [Officine’s] sales, this limitation is not dispositive in light of the affirmative evidence of linkage between costs and prices in the home market. Moreover, it is not that there is no correlation regarding the U.S. prices, but rather that the data points we have do not allow us to perform the analysis. Commerce analyzes the data points of the top five [control numbers] in the comparison and U.S. markets, in accordance with our practice, to the extent that the data points are available.
Remand Results at 12, ECF No. 32 (emphasis added); see also Oral Arg. Tr. at 75:10–
22, ECF No. 51. The cited passage merely states that (1) Commerce could not
complete a trend analysis for U.S. prices, (2) this fact is “not dispositive” given other
linkages in the home market data, and (3) Commerce will use the normal ten control
number approach “to the extent that the data points are available.” Remand Results
at 12, ECF No. 32. Commerce’s nod to Nucor’s objection — that the lack of data is
“not dispositive” — fails to explain why the absence of data apparently analyzed by
Commerce in every other application of its test does not undermine the results. Cf.
Def.-Int.’s Br. at 3, ECF No. 38. It fails to link the “facts found” with the “choice
made.” See Burlington Truck Lines, 371 U.S. at 168.
Commerce also points to data and documents found in the record and asks the
Court to trust the agency that it has examined the data and determined the results
were reliable. See Oral Arg. Tr. at 32:13–34:11, ECF No. 51 (drawing the Court’s
attention to information in Officine’s supplemental questionnaire); id. at 35:9–11
(MR. GOLDEN: “[T]he data Commerce uses, and the data in the record –– and this
may sound ridiculous –– is the next approximately 350 pages …. It starts on [J.A. Court No. 1:23-cv-00001 (SAV) Page 22
at] 83,024.”); id. at 38:3–5 (MR. GOLDEN: “[T]he Decision Memorandum cites to
these documents, and these documents contain the mathematics and the code and
the data.”). But “[t]he Court reviews answers Commerce actually gave for substantial
evidentiary support.” Bonney Forge Corp. v. United States, 46 CIT __, 560 F. Supp.
3d 1303, 1312 (2022); see 19 U.S.C. § 1516a(b)(1)(B)(i). “It does not draft answers
Commerce never gave from the available record information before the Department.”
Bonney Forge, 36 CIT ___, 560 F. Supp. 3d at 1312; accord Motor Vehicle Mfrs. Ass’n
v. State Farm Mut. Auto. Ins. Co., 463 U.S. 29, 43 (1983) (“[W]e may not supply a
reasoned basis for the agency’s action that the agency itself has not given.”).
Commerce may not state its conclusion, cite reams of data without explanation, and
say “Trust us. We’re right.” By doing so, the agency creates a textbook example of
failing to “articulate [a] rational connection between the facts found and the choice
made.” Burlington Truck Lines, 371 U.S. at 168. Likewise, where an explanation is
provided at oral argument but not in the agency’s actual decision, it cannot be
considered. Compare Oral Arg. Tr. at 54:24–55:22 (Officine’s counsel asserting that
Commerce addressed nonstandard factual datasets in Officine’s prior reviews but
admitting that those prior reviews are neither on the record nor discussed in the
Remand Results), with Burlington Truck Lines, 371 U.S. at 168 (“The courts may not
accept appellate counsel’s post hoc rationalizations for agency action[.]”).
Commerce has also failed to address why it opted to use its typical test here
when it deviated in Ferrovanadium. The Government argues that Ferrovanadium is Court No. 1:23-cv-00001 (SAV) Page 23
distinguishable because it involved multiple quarters of sales data in both markets
— albeit with a smaller pool of control numbers. Oral Arg. Tr. at 75:10–22, ECF No.
51; see also Remand Results at 12, ECF No. 32 (“Commerce still followed its
established practice in [Ferrovanadium] of analyzing the largest sales volume
[control numbers] in the pool …; the pool simply happened to be a pool of one [control
number] ….”). But that explanation is both mistaken and beside the point.
Commerce’s claim that there “simply happened to be a pool of one” control
number in Ferrovanadium is incorrect. Remand Results at 12, ECF No. 32. The
Ferrovanadium respondent produced three total control numbers — two sold
exclusively in the home market and one sold in both the home and U.S. markets. Id.
at 11–12. Despite there being three total control numbers, Commerce focused solely
on the control number shared between both markets for its linkage analysis. Id.
Thus, as Nucor notes, faced with a non-standard data set, Commerce deviated from
its test and focused on the one product sold in both markets. Here, Commerce has
done the opposite. Faced with another non-standard data set, it ignored the limited
U.S. market data entirely and made its conclusion based solely on data from the
Italian home market. Despite never having done that before, Commerce claims it is
merely following its normal methodology, so all is well. Unfortunately for Commerce,
it is not.
If Commerce wishes to continue applying the quarterly methodology, it must
explain (1) why it believes home market sales data alone allows it to render a Court No. 1:23-cv-00001 (SAV) Page 24
conclusion about whether sales prices are reasonably linked to the cost of production,
(2) why it believes ignoring U.S. market data is or is not a departure from past
practice, and (3) why whatever test it applies can be trusted to produce reliable
results. Because Commerce failed to answer any of these questions or acknowledge
that its analysis deviated from its normal methodology, its decision is unsupported
by substantial evidence and must be REMANDED.
B.
In a final attempt to save its determination from being remanded, Commerce
asks the Court to consider its alternative holding where Commerce claims that, even
if it applied Nucor’s preferred analysis, the result would be the same. See id. at 8
(summarizing Nucor’s argument that “Commerce’s linkage analysis should … focus
on the home market [control numbers] that matched to U.S. sales in the margin
calculations.”).5 Commerce claims it would still find the necessary linkage to allow
5 The parties bracketed information identifying (1) the number of overlapping control numbers between the U.S. and Italian markets (three), (2) the number of control numbers sold in the U.S. market that also appear in the top five highest-selling Italian market control numbers (one), and (3) the number of overlapping control numbers for which Commerce could find a linkage under Nucor’s proposal to consider all overlapping control numbers between the two markets (two out of three). See, e.g., Final Calculation Mem. at 1–2, J.A. at 83,396– 97, ECF No. 47. However, the parties waived any confidentiality claim by referring to these facts in open court. Compare CVB, Inc. v. United States, 48 CIT __, 681 F. Supp. 3d 1314, 1317–19 (2024) (refusing to redact information for similar reasons), with Fed. Cir. R. 25.1(c)(1) (“Material will lose its status … if and when it … has appeared in a filing without being marked confidential.”), Oral Arg. Tr. at 24:4–7, ECF No. 51 (the Court summarizing Nucor’s argument that Commerce should primarily care about control number “A” instead of examining the top five highest-selling control numbers); id. at 23:24–24:1 (Officine’s counsel stating that three U.S. control numbers were also sold in the home market); id. at 24:12–16 Court No. 1:23-cv-00001 (SAV) Page 25
it to use the quarterly cost methodology instead of reviewing the data on an annual
basis. Because Commerce’s alternative suffers from the same flaw as its primary
finding — lack of an adequate explanation on the record — the Court cannot sustain
Commerce’s determination on this alternative ground.
Commerce examined the five highest-selling control numbers in the home
market and the U.S. market.6 Id. at 10. Officine’s home market sales consisted of
more than five control numbers, but Commerce limited its review to the five highest-
selling control numbers. Compare Officine Suppl. Questionnaire Resp., Ex. RD-3,
J.A. at 83,015, ECF No. 44 (identifying the universe of Officine’s home market control
numbers), with Attachment III, J.A. at 83,383, ECF No. 46 (comparing changes in
sale price and manufacturing cost for the five highest-selling home market control
numbers). Nucor argues that Commerce should focus its analysis on three
overlapping control numbers –– present in both the home market and U.S. market —
which the Court referred to at oral argument as control numbers “A,” “B,” and “C.”
Def.-Int.’s Br. at 4, ECF No. 38; Oral Arg Tr. at 5:8–23, 23:24–24:4, ECF No. 51; Final
Calculation Mem. at 2, J.A. at 83,397, ECF No. 44. Control number “A” is the only
control number sold in both the Italian and U.S. markets that is also among the five
(Officine’s counsel stating that Commerce looked at all three overlapping control numbers and found a reasonable linkage for the majority). 6 The U.S. market only has five control numbers; thus, the pool of U.S. control numbers
consists of all U.S. market control numbers. See Officine Suppl. Questionnaire Resp., Ex. RD-3, J.A. at 83,016, ECF No. 44 (identifying the universe of Officine’s U.S. market control numbers). Court No. 1:23-cv-00001 (SAV) Page 26
largest-selling Italian market control numbers. Def.-Int.’s Br. at 4, ECF No. 38
(“[O]nly one of the five largest-selling home market [control numbers is relevant] to
the margin calculation, making Commerce’s analysis of the five largest-selling home
market [control numbers] of limited value.”); Oral Arg. Tr. at 24:4–7, ECF No. 51.
Control numbers “B” and “C” are sold in both the U.S. and Italian markets but are
not among the top five sales volume control numbers in Italy. Nucor notes that
Commerce limited its review in Ferrovanadium to the single, shared control number
and disregarded home market control numbers not also sold in the U.S. market. Def.-
Int.’s Br. at 6–7, ECF No. 38.
Commerce responds that, even under Nucor’s approach, it “still would find that
Nucor’s analysis is questionable” because it could find sufficient evidence of linkage
for two of the three overlapping control numbers Nucor identified. Final Calculation
Mem. at 2, J.A. at 83,397, ECF No. 44; Oral Arg. Tr. at 24:14–15, ECF No. 51 (MR.
LEE: “Commerce looked at all three [shared control numbers]. And for a majority of
those[,] Commerce found a correlation.”). Commerce asserts that, although control
number “B” does not support finding linkage, the two other overlapping control
numbers –– “A” and “C” –– together match to a large majority of U.S. sales and
demonstrate linkage in the majority of the quarters during the Period of Review.
Final Calculation Mem. at 2, J.A. at 83,397, ECF No. 44; see also Oral Arg. Tr. at
24:12–16, 45:7–11, ECF No. 51. Although Commerce acknowledges that it could
deviate from its “normal” approach and look at only the overlapping control numbers, Court No. 1:23-cv-00001 (SAV) Page 27
Commerce declined to do so because deviation is inconsistent with its normal linkage
analysis. Remand Results at 11, ECF No. 32 (stating that Nucor’s approach is
impractical because it requires analyzing linkages after running the margin
calculation).
Merely identifying a fallback argument cannot save Commerce. If Commerce
believes there is more than one way to look at the data, it must state the alternatives,
explain how it analyzes the available data, and offer a reasoned explanation. Should
Commerce wish to rest a future determination on an analysis of the overlapping
control numbers, it needs to state what potential flaws caused it to abandon its
“normal” linkage analysis, explain how Ferrovanadium informs its decision, and
describe how any alternative analysis leads to trustworthy results. Because those
explanations are not found in the Issues and Decisions Memorandum, the Court may
not sustain Commerce’s determination on this alternative basis.
CONCLUSION
Commerce chose to deviate from its normal practice of using an annual
weighted-average cost methodology and instead used a quarterly cost methodology to
analyze sales during the Period of Review. Although its test for applying the
quarterly cost methodology may be based on past practice, this case presents an
atypical data set. Commerce failed to explain why its test for applying a quarterly
cost methodology is adequate to address a situation where there is only one quarter
of U.S. sales data. It may not ignore an objection without a reasoned explanation Court No. 1:23-cv-00001 (SAV) Page 28
because “an agency’s statement of what it ‘normally’ does or has done before … is not,
by itself, an explanation of ‘why its methodology comports with the statute.’” See CS
Wind Viet. Co. v. United States, 832 F.3d 1367, 1377 (Fed. Cir. 2016) (quoting SKF
USA Inc. v. United States, 263 F.3d 1369, 1383 (Fed. Cir. 2001)).
Once it found itself in court, Commerce gave many potential answers and cited
hundreds of pages of exhibits. But the necessary analysis and explanation linking
that information to Commerce’s decision are absent in the Remand Results. Because
it is only explanations contained within the Remand Results that count, the Court
REMANDS this matter for further consideration by Commerce. On remand,
Commerce may reopen the record to accept evidence from past reviews, provide
further explanation to bolster its current analysis, or choose an alternative pathway
in the vein of Ferrovanadium. Whichever options Commerce selects, it should link
the facts it has found to the choices it makes to explain why the results of its linkage
analysis engender confidence. Accordingly, it is hereby:
ORDERED that the Remand Results are REMANDED to Commerce for
further explanation consistent with this opinion. It is further:
ORDERED that Commerce shall conduct a new analysis to determine if use
of its quarterly cost methodology is warranted;
ORDERED that Commerce shall file its Remand Determination with the
Court within 120 days of today’s date; Court No. 1:23-cv-00001 (SAV) Page 29
ORDERED that Defendant-Intervenor shall have 30 days from the filing of
the Remand Results to file its brief;
ORDERED that Defendant shall have 30 days from the date of Defendant-
Intervenor's filing to submit a response;
ORDERED that Plaintiff shall have 21 days from the date of Defendant's
filing to submit its response; and
ORDERED that Defendant-Intervenor shall have 14 days from the date of
Plaintiffs filing to submit any reply; it is also
ORDERED that Plaintiff shall have 14 days from the date that Defendant-
Intervenor submits its reply to file the Joint Appendix. Motions for Oral Argument,
if any, shall be due 21 days after the filing of the Defendant-Intervenor's reply.
SO ORDERED.
Stephen Alexander Vaden, Judge
Dated: J,yfwJ . . . 17, l..,z.'( New York, New York