Pasta Zara S.p.A. v. United States, No. 20-00023 (Ct. Int'l Trade 2021)

Annotate this Case

The court issued a subsequent related opinion or order on May 4, 2022.

Download PDF

Consol. Court No. 20-00023 Page 2 Elizabeth C. Johnson and David C. Smith, Jr., Kelley Drye & Warren, LLP, of Washington, D.C., argued for Defendant-Intervenors Riviana Foods, Inc. and Treehouse Foods, Inc. With them on the brief was Paul C. Rosenthal. Eaton, Judge: Before the court is the motion for judgment on the agency record of Plaintiffs Ghigi 1870 S.p.A. (“Ghigi”) and Pasta Zara S.p.A. (“Zara”) (collectively, “Ghigi/Zara” or the “company”),1 and Consolidated Plaintiffs Agritalia S.r.L. and Tesa S.r.L. See Pls.’ Mem. Supp. Mot. J. Agency R., ECF No. 33 (“Pls.’ Br.”); Pls.’ Reply, ECF No. 46. By their motion, Ghigi/Zara, Agritalia, and Tesa (collectively, “Plaintiffs”) challenge the final results of the U.S. Department of Commerce’s (“Commerce” or the “Department”) twenty-second administrative review of the antidumping duty order on certain pasta from Italy (“Order”),2 covering the period from July 1, 2017, through June 30, 2018. See Certain Pasta From Italy, 85 Fed. Reg. 2,714 (Dep’t Commerce Jan. 16, 2020) (“Final Results”) and accompanying Issues and Decision Mem. (Jan. 10, 2020) (“Final IDM”), PR 181; see also Certain Pasta From Italy, 84 Fed. Reg. 48,114 (Dep’t Commerce Sept. 12, 2019) (“Preliminary Results”) and accompanying Decision Mem. (Sept. 6, 2019) (“PDM”), PR 156. In particular, Plaintiffs challenge (1) Commerce’s use of adverse facts available and (2) its rejection of arguments, that were raised for the first time after verification, disputing Commerce’s model-match method with respect to protein content and shape. Defendant the United States and Defendant-Intervenors Riviana Foods, Inc. and Treehouse Foods, Inc.3 maintain that Commerce’s use of adverse facts available and its rejection of 1 Ghigi and Zara are affiliated companies that were collapsed into a single entity, Ghigi/Zara, during the underlying review. 2 See Certain Pasta From Italy, 61 Fed. Reg. 38,547 (Dep’t Commerce July 24, 1996) (notice of antidumping duty order). 3 Defendant-Intervenors have withdrawn from this action, effective October 7, 2021. See Order (Oct. 7, 2021), ECF No. 58. Consol. Court No. 20-00023 Page 3 Ghigi/Zara’s post-verification arguments were lawful, and ask the court to sustain the Final Results. See Def.’s Resp. Opp’n Mot. J. Agency R., ECF No. 42; Def.-Ints.’ Resp., ECF No. 43. Jurisdiction is found under 28 U.S.C. § 1581(c) (2018) and 19 U.S.C. § 1516a(a)(2)(B)(iii) (2018). For the reasons that follow, the court remands Commerce’s adverse inferences determination. BACKGROUND On September 10, 2018, Commerce published its notice of initiation of the twenty-second administrative review of the Order. See Initiation of Antidumping & Countervailing Duty Admin. Revs., 83 Fed. Reg. 45,596 (Dep’t Commerce Sept. 10, 2018). Commerce selected two mandatory respondents for individual examination: Ghigi/Zara and Industria Alimentare Colavita S.p.A.4 I. Ghigi’s Reporting of U.S. Payment Date Information On October 12, 2018, Commerce issued an initial questionnaire to Ghigi/Zara. See Dep’t Commerce Initial Quest. (Oct. 12, 2018) (“Initial Quest.”), PR 22. By its questionnaire, Commerce asked for information about the company’s U.S. and home market sales. Among the U.S. sales information requested were the dates on which “payment was received from the customer” for the sales of Ghigi’s U.S. affiliate (data field: PAYDATEU). See Initial Quest. at C-12. In response, Ghigi reported dates and amounts of payments in its U.S. database (the “Original Database”). See Ghigi/Zara Secs. B-D Quest. Resp. (Dec. 10, 2018), CR 38-58, PR 55. Commerce thereafter issued a supplemental questionnaire and asked Ghigi to “[i]nclude the field PAYDATEU with the weighted-average payment in the U.S. sales database.” See Dep’t 4 Industria Alimentare Colavita S.p.A. is not a party in this action. Consol. Court No. 20-00023 Page 4 Commerce Secs. A-C Suppl. Quest. (Feb. 8, 2019) at 17, PR 67. Commerce also instructed Ghigi to “[p]rovide a calculation worksheet for [credit expenses associated with a particular U.S. sale] that identifies each component of the calculation.” Dep’t Commerce Secs. A-C Suppl. Quest. at 17. In other words, Commerce supplemented its original request by asking Ghigi to add weightedaverage payment date information to its U.S. sales database, along with a calculation worksheet for certain credit expenses. In response, Ghigi submitted a revised U.S. sales database (the “Revised Database”). Ghigi/Zara Secs. A-C Suppl. Quest. Resp. (Mar. 14, 2019), CR 76-179, PR 84. After the questionnaire period had concluded, Commerce conducted verification of Ghigi’s U.S. constructed export price sales, pursuant to a request by the petitioners, the DefendantIntervenors.5 See Req. for Verification (Dec. 19, 2018), PR 61. At verification, Ghigi for the first time informed Commerce that “the payment dates in the most recent U.S. sales database [i.e., the Revised Database] [were] incorrect due to a programming error in the U.S. payment date (PAYDATEU) data field.” PDM at 16 (citing Ghigi/Zara Post-Verification Cmts. (July 18, 2019) at 7-9, CR 337-338, PR 140). Commerce found that as a result of the programming error “Ghigi reported the U.S. payment date incorrectly for most of its U.S. sales.” PDM at 16. In its post-verification comments and case brief, Ghigi/Zara argued that Commerce should have used the payment dates reported in the Original Database, instead of the Revised Database, because they were closer to the correct payment dates. See Ghigi/Zara Post-Verification Cmts. at 7-9; see also Ghigi/Zara Case Br. (Oct. 23, 2019) at 32-37, PR 167 (same). Ghigi proposed this 5 In an administrative review, Commerce verifies the factual information on which it will rely in its final determination, if a timely request for verification is made, and no verification was conducted in the last two administrative reviews. See 19 U.S.C. § 1677m(i)(3); see also 19 C.F.R. § 351.307(b)(1)(v) (2019). Consol. Court No. 20-00023 Page 5 plan after verification, and apparently was asking Commerce to either verify its Original Database, or use unverified payment dates. In its decision memoranda, Commerce declined, stating: Commerce’s practice is to rely on the most recently submitted databases as the basis for verification because such data is responsive to Commerce’s most recent supplemental questions. Thus, we . . . find that it is not appropriate to use the payment date information from a prior U.S. sales database. PDM at 16 (footnote omitted); see also Final IDM at 12-13. Unable to verify the dates in the Revised Database, Commerce found that the use of facts otherwise available, pursuant to 19 U.S.C. § 1677e(a)(2)(D), was warranted. Additionally, Commerce applied adverse inferences, pursuant to 19 U.S.C. § 1677e(b)(1)(A), when selecting from among the facts available. See PDM at 16-17; see also Final IDM at 13. As adverse facts available, Commerce “applied the longest period between shipment date and payment date for any sale on the record of the review for purposes of imputed credit expenses for Ghigi’s U.S. sales.” PDM at 17; see also Final IDM at 13. II. Ghigi/Zara’s Reporting of Protein Content and Shape A. Protein Content In the initial questionnaire, as it had done in the past, Commerce instructed Ghigi/Zara to report “the percentage of protein in the pasta sold, as stated on the label of the respective product”6 in the United States and Italy.7 See, e.g., Initial Quest. at C-7 (emphasis added). Commerce further 6 Since at least the fifteenth review of the Order, Commerce has relied on the protein percentage as stated on the packaging label of finished pasta products to identify protein content. See Pls.’ Br. attach. 2 at 28 (Issues and Decision Memorandum for Final Results of 2010-2011 Review) (“[T]he package label is a reliable source for the Department to use in identifying the physical characteristics, including protein content.”); see also Certain Pasta From Italy, 78 Fed. Reg. 9,364 (Dep’t Commerce Feb. 8, 2013) (final results of fifteenth administrative review). 7 The protein content in pasta comes from semolina flour—the main input of pasta— which is made from durum wheat. Protein content has been included in Commerce’s pasta model- Consol. Court No. 20-00023 Page 6 instructed Ghigi/Zara to code as a “1” pasta with a protein content of 12.50 percent or more, and a “2” pasta with a protein content of between 10 and 12.49 percent. See Dep’t Commerce Req. for Revised U.S. Sales Database (June 26, 2019), PR 133. The initial questionnaire invited Ghigi/Zara to contact the Commerce official identified on the questionnaire’s cover page if it had any questions: “If you have questions, we urge you to consult with the official in charge named on the cover page.” See Initial Quest. (General Instructions). In its initial Section C response on U.S. sales, Ghigi/Zara coded the protein content of its finished pasta products sold in the United States as a “2,” meaning a protein content of less than 12.50 percent, in line with the protein content of a “standard” pasta product. Ghigi/Zara did not, however, rely on the protein content stated on its products’ U.S. packaging label. Rather, the company converted the protein percentage on the label to reflect the amount of protein measured under Italian protocols,8 and coded the protein content based on that conversion. See, e.g., Ghigi Sales Verification Report (Aug. 1, 2019), PR 146. match method since the twelfth review of the Order. See Certain Pasta from Italy, 75 Fed. Reg. 6,352, 6,353 (Dep’t Commerce Feb. 9, 2010) (final results of twelfth administrative review). In that review, Commerce determined that the dividing line between “premium” and “standard” pasta was a protein content of 12.50 percent. That is, pasta with a protein content of 12.50 percent or more was considered premium, and pasta with a protein content of between 10 percent and 12.49 percent was considered standard. See Final IDM at 7-8 (discussing the 2007-2008 review where Commerce found that “12.5 percent minimum content is an industry standard developed in the Italian market place of pasta manufacturers and semolina sellers”). 8 According to Plaintiffs, “the measurement [of protein content] begins from the observed nitrogen (N) content of the dry pasta”: For the U.S. market, FDA regulations require that the N content be multiplied by 6.25, while the Italian regulation specifies that the protein content equals the N content times 5.7. Thus, a protein content of 12.5% on a U.S. nutrition panel is equivalent to a protein content of 11.4% on an Italian nutrition panel: 12.5%/6.25 * 5.7 = 11.4%. Conversely, protein content of 12.5% on an Italian nutrition panel is equivalent to protein content of 13.7% on a U.S. nutrition panel: 12.5%/5.7 * 6.25 = 13.7%. Consol. Court No. 20-00023 Page 7 Ghigi/Zara did not inform Commerce of the conversion it made in its questionnaire responses. Notwithstanding this omission, the company certified that it had complied with Commerce’s instructions. See, e.g., Pls.’ Br. 13 (“Ghigi/Zara did not flag this as a particular issue, responding ‘Ghigi/Zara report the protein content in accordance with the instruction.’”). At verification, Commerce compared Ghigi/Zara’s protein content coding to the percentages stated on the packaging label of finished pasta products. Commerce thus learned that the company had failed to code according to the label. Subsequently, Commerce instructed Ghigi/Zara to recode the protein content in accordance with the label. See Req. for Revised U.S. Sales Database. Ultimately, in accordance with Commerce’s instruction, Ghigi/Zara reported the protein content of its pasta sold in the United States as a “1,” based on the protein content indicated on the U.S. packaging labels (i.e., without converting the protein content under the Italian protocol), but it did so under protest. See Ghigi/Zara Resp. to Req. for Revised U.S. Sales Database at 1 (July 2, 2019), PR 135 (stating that Ghigi/Zara was “mindful” that the question called for protein content “as stated on the label” but nonetheless “emphatically [stood] by their original reporting,” stating “rote reliance on the ingredient panel of the label gives a misleading result because Italy and the United States have different formulas for measuring the protein content of food”). After verification had closed and the Preliminary Results were issued, Ghigi/Zara argued in its case brief that Commerce’s questionnaire instructions contained a “latent ambiguity” because protein measurement protocols differ between the United States and Italy. See Ghigi/Zara Case Br. at 8. In other words, while under U.S. protocols the pasta the company sold in the United States was premium (protein content of 12.5 percent or more), when the protein percentage was converted Pls.’ Br. 15-16 (cleaned up). Consol. Court No. 20-00023 Page 8 to an Italian measurement, the U.S. product was standard (protein content of less than 12.5 percent). See Ghigi/Zara Case Br. at 1-2. Ghigi/Zara insisted that, absent a conversion of protein content from the U.S. protocol to the Italian protocol, it was impossible to ensure that products with the same protein coding were physically identical, i.e., to permit a comparison of premium U.S. pasta with premium Italian pasta and standard U.S. pasta with standard Italian pasta. Ghigi/Zara also objected to the use of 12.5 percent as the dividing line between premium and standard pasta. See Ghigi/Zara Case Br. at 11, 14. In the Final Results, Commerce declined to accept Ghigi/Zara’s arguments and found that the instructions contained no ambiguity—latent or otherwise. Commerce noted that the meaning of the Department’s direction was plain and required no interpretation: Ghigi/Zara was to report “the percentage of protein content in the pasta sold, as stated on the label of the respective product.” Final IDM at 7 (emphasis added). In rejecting Ghigi/Zara’s other arguments, Commerce relied on its decisions in prior administrative reviews of the Order. First, to address the company’s arguments against the use of the 12.5 percent dividing line between premium and standard pasta, Commerce relied on its decision memorandum from the 2007-2008 review, where the Department first introduced protein content in its model-match method: Our decision to use a minimum protein content of 12.5 percent for premium finished pasta is based on four factors. The first one is that, as stated above, we believe some brands of pasta are produced, marketed, and sold as premium products, distinct from standard products. These premium pasta brands have distinct physical characteristics that are commercially significant. The second factor is that there is not a clearly defined method of identifying premium pasta other than the protein content marked on the packages. The third factor is that there is a clear relationship between the physical characteristics of the semolina used to produce the finished pasta and the finished pasta itself. The [fourth] factor is that 12.5 percent minimum content is an industry standard developed in the Italian market place of pasta manufacturers and semolina sellers. Given these factors, we Consol. Court No. 20-00023 Page 9 believe our approach is reasonable and will contribute to the accuracy of the dumping analysis. Final IDM at 7-8 (footnote omitted); see also Pls.’ Br. attach. 1 cmt. 1 (Issues and Decision Memorandum for Final Results of 2007-2008 Review). Next, with respect to relying on the protein content as stated on the label, Commerce cited its decision memorandum in the 2010-2011 review: [T]he protein content of the finished pasta listed on the package is central to our analysis. . . . [A]ll of the physical characteristics that are basis for our model match criteria are printed on the labels of the finished pasta packages. Buyers and sellers examine this information, as listed on the packaging, in determining which products to purchase and/or sell and the appropriate price. In addition, because pasta is sold through retail chain to individual customers, there are often many different intermediaries involved in the distribution and sale of finished pasta; each of which need to know the relevant information. Furthermore, our reliance upon the information listed on the packaging of the finished product (i.e., the same information that is available to a consumer in the United States) conforms to our statutory obligation to base our price-to-price comparison on a transparent and consistent basis. Thus, relying on the information reported on the packages of finished pasta is appropriate. Final IDM at 8 (footnote omitted); see also Pls.’ Br. attach. 2 cmt. 4 (Issues and Decision Memorandum for Final Results of 2010-2011 Review). Thus, in the Final Results, Commerce restated its view that its use of the packaging label for protein content was an “objective method” to make statutorily required price comparisons on a “transparent and consistent” basis: Commerce based its price-to-price comparisons (i.e., defining the normal value for U.S. sale prices on the sale price(s) of the identical, or alternatively the most similar, product sold in the comparison market) on a transparent and consistent basis by properly selecting the protein content as listed on the packaging label for finished pasta. As we noted in the Pasta 2007-2008 Review, the market reality is that “there is not a clearly defined method of identifying premium pasta other than the protein content marked on the packages.” Thus, Commerce’s reliance on the packaging label is an objective method to achieve a product comparison on a “consistent and transparent” basis because all of the physical characteristics are listed on the product label. Indeed, Ghigi/Zara market and price their sales to U.S. customers based on the specification of the product denoted on the label. Thus, we find unconvincing Ghigi/Zara’s argument that we should base the PROTEINU coding Consol. Court No. 20-00023 Page 10 upon the internal information in its [bill of materials] or on a different measurement protocol for protein content. Final IDM at 8 (footnotes omitted). Accordingly, for the Final Results, Commerce continued to rely on Ghigi/Zara’s revised questionnaire responses, in which the company coded protein content in accordance with the Department’s instructions, instead of its initial responses in which it coded protein content based on percentages converted from the U.S. protocol to the Italian protocol. B. Product Shape Product shape is one of the model match criteria that the Department uses in its comparison of U.S. merchandise and home market products.9 Thus, in the initial questionnaire, Commerce instructed Ghigi/Zara to classify its pasta according to eight categories that are stated in a “shape classification table.” These eight categories broadly distinguish “specialty” long and short cuts from “regular” cuts.10 See Initial Quest., app. III. (emphasis added) (“You are required to classify the pasta types reported in field 3.9 into one of the shape categories specified in field 3.1 in accordance with the questionnaire examples and the attached ‘Classification of Pasta Shapes.’”). It further instructed that “[i]f [the respondent] sold pasta in shapes that do not appear on the attached list, please contact the official in charge.” See Initial Quest., app. III. 9 The pasta model-match method has included “product shape” since the original investigation. See Certain Pasta From Italy, 61 Fed. Reg. 1,344, 1,346 (Dep’t Commerce Jan. 19, 1996) (preliminary determination). 10 Each of the eight categories in the shape classification table has its own code for reporting purposes: long cut pasta (e.g., linguine or spaghetti) was coded as 1; specialty long cuts (e.g., capellini or fioccini) were coded as 2; nested/folded/coiled was coded as 3; and lasagna was coded as 4. Short cuts (e.g., fagiolini, medium shells) were coded as 5; specialty short cuts (e.g., mezzanelli, pasta mista) were coded as 6; soupettes (e.g., ditali, corallini) were coded as 7; and combinations of shapes were coded as 8. See Initial Quest., app. III. Consol. Court No. 20-00023 Page 11 Relevant here, the table categorized fusilli and cavatappi as “specialty” short cuts. In its questionnaire response, however, Ghigi/Zara reported its fusilli and cavatappi pasta as simply “short cuts,” coded as a “5”. See Ghigi/Zara Sec. A Quest. Resp. (Nov. 16, 2018), PR 36-46. As with protein content, Ghigi/Zara did not inform Commerce that it had departed from the classification table when coding for shape. Commerce only discovered Ghigi/Zara had done so at verification. See Final IDM at 10. After verification had closed and the Preliminary Results were issued, Ghigi/Zara argued in its case brief that its “original reporting was correct.” Ghigi/Zara Case Br. at 28. The company maintained that it was “correct to classify fusilli and cavatappi as normal [regular] short cuts rather than specialty short cuts, based on [its] production throughput rate (at multiple plants), [its] merchandising, and [its] pricing.” Ghigi/Zara Case Br. at 32. In the Final Results, Commerce rejected Ghigi/Zara’s original coding and applied the shape codes set forth in the classification table: [W]e found at verification that Ghigi misreported fusilli and cavatappi as short cuts. Commerce’s longstanding practice is to require respondents to report pasta shape codes based on the pasta shape classification table if the shapes are already listed on that table. This practice has been approved by the Court [in La Molisana S.p.A. v. United States, No. 16-00047, 2018 WL 3089242 (CIT June 21, 2018) (not reported in Federal Supplement), aff’d 784 Fed. Appx. 780 (Fed. Cir. 2019) (mem.)]. Fusilli and cavatappi are both listed as specialty short cuts in the shape classification table. Final IDM at 10 (footnotes omitted). Additionally, Commerce rejected Ghigi/Zara’s argument that its quicker line speeds justified classification of fusilli and cavatappi as short cuts. Commerce noted that it “has previously rejected . . . attempts to reclassify pasta shapes based on companyspecific throughput rates”: In Pasta 2013-2014 Review, Commerce determined to reject La Molisana’s similar attempt to replace the well-established shape classification in the model match methodology with a system based on company-specific line speeds. Specifically, Consol. Court No. 20-00023 Page 12 Commerce required La Molisana to report its pasta sales in accordance with the existing shape classification table because this methodology is reasonable and pursuant to the requirement of [19 U.S.C. § 1677(16)(A)] that classifications be based on a product’s “physical characteristics.” In that review, Commerce further explained that La Molisana’s contentions were meritless because: (1) line speed is not the defining factor in determining pasta shape under Commerce’s methodology; and (2) Commerce has no practice of permitting respondents to re-classify existing pasta shapes based upon company-specific line speeds. In La Molisana, the CIT sustained Commerce’s application of its model-matching methodology, which required La Molisana to report product shapes in conformity with the existing identities and categories of shapes on Commerce’s pasta shape list. The CIT also rejected La Molisana’s argument that company-specific line speeds are a sufficient reason to depart from the list for shapes that are already on the list. The [Federal Circuit] affirmed the CIT’s holding in La Molisana. Accordingly, consistent with the Preliminary Results, we continue to re-classify fusilli and cavatappi as specialty short cuts, consistent with the instructions in the Initial Questionnaire and Commerce’s longstanding practice. Final IDM at 10-11 (footnotes omitted). The Department had thus considered arguments similar to those now made by Ghigi/Zara in the eighteenth administrative review—four years prior to this review—and found them wanting. In the Final Results, Commerce rejected Ghigi/Zara’s coding of its fusilli and cavatappi pasta as short cuts, and instead treated those shapes as specialty short cuts, coded as a “6,” as provided in the table. Based on the application of adverse facts to Ghigi/Zara with respect to Ghigi’s payment dates, Commerce calculated a weighted-average dumping margin of 91.76 percent for Ghigi/Zara, and an all-others rate of 44.56 percent that was applied to Agritalia and Tesa. See Final Results, 85 Fed. Reg. at 2,714. Plaintiffs timely commenced this action to challenge Commerce’s adverse facts available determination, and its rejection of Ghigi/Zara’s post-verification arguments against the Department’s method of coding for protein content and shape. Consol. Court No. 20-00023 Page 13 STANDARD OF REVIEW The court will sustain a determination by Commerce 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). LEGAL FRAMEWORK Where Commerce determines that a gap in the factual record exists because necessary information is missing, the statute directs the use of “facts otherwise available,” to supply the needed facts. See 19 U.S.C. § 1677e(a). Under the statute, a gap may be found to exist where an interested party provides the requested information, “but the information cannot be verified” in accordance with § 1677m(i).11 See 19 U.S.C. § 1677e(a)(2)(D). Where Commerce has determined that the use of facts available is warranted, it may apply an adverse inference when selecting from among the facts available if it makes the requisite additional finding that an “interested party has failed to cooperate by not acting to the best of its ability to comply with a request for information from the [Department].” Id. § 1677e(b)(1). “To the best of its ability” means “one’s maximum effort.” Nippon Steel Corp. v. United States, 337 F.3d 1373, 1382 (Fed. Cir. 2003). 11 Under § 1677m(i), Commerce “shall verify all information relied upon in making . . . a final determination in a review under section 1675(a)” if “(A) verification is timely requested by an interested party,” and “(B) no verification was made under this subparagraph during the 2 immediately preceding reviews and determinations under section 1675(a) . . . of the same order, finding, or notice, except that this clause shall not apply if good cause for verification is shown.” 19 U.S.C. § 1677m(i). Consol. Court No. 20-00023 Page 14 DISCUSSION I. Commerce’s Use of Adverse Facts Available with Respect to Ghigi’s U.S. Payment Dates Is Neither Supported by Substantial Evidence nor in Accordance with Law In the Final Results, Commerce found that the use of adverse facts available was required because it could not verify Ghigi’s final submitted payment dates for U.S. sales—that is, the dates reported in the Revised Database did not match the dates on the invoices for each of the sales traces that Commerce performed at verification: We disagree with Ghigi’s argument that Commerce’s application of partial [adverse facts available] is unwarranted because Ghigi’s [Original Database] has accurate payment dates and thus, Commerce should use its information to calculate antidumping duty margins. At the time of commencement of verification, Ghigi’s [Revised Database] was the most recent version of the U.S. sales database. Accordingly, the verifiers performed their verification procedures on the [Revised Database], and they did not conduct verification procedures on [the Original Database], which was the initial US sales submission. As indicated in the Preliminary Results, Commerce’s practice is to rely on the most recently submitted databases as the basis for verification because such data is responsive to Commerce’s most recent supplemental questions. Thus, we find it is not appropriate to use the payment date information from [the Original Database], which was the prior U.S. sales database. As for the [Revised Database], its payment dates do not match payment dates listed in the sales documentation for the U.S. transactions examined at verification. Pursuant to [19 U.S.C. § 1677e(a)(2)(D), i.e., providing for the use of facts available where information cannot be verified], we find that a determination based on the facts otherwise available is warranted because the information on payment data was not verifiable. Accordingly, we find that the application of partial adverse inferences under [19 U.S.C. § 1677e(b)(1)(A), i.e., providing for the use of adverse inferences where a respondent fails to cooperate to the best of its ability] is warranted, as it applies to Ghigi’s U.S. payment date field. Final IDM at 13 (footnotes omitted). As an initial matter, Plaintiffs maintain that Commerce had no reason to use facts otherwise available, let alone with an adverse inference, because instead of the Revised Database, it could have used Ghigi’s Original Database: Consol. Court No. 20-00023 Page 15 At the outset of verification, [Ghigi’s U.S. affiliate] informed Commerce of the programming error that garbled the [Revised Database] payment dates and proposed use of the [Original Database] dates, but Commerce chose to verify the [Revised Database] dates instead. It is simply counterfactual for Commerce to assert that “information on payment data was not verifiable” when [Ghigi’s affiliate] presented a perfectly usable set of payment dates in [Original Database]. Furthermore, the 17 sales for which Commerce reviewed payment dates constituted a reasonable sample against which to test the accuracy of the [Original Database] payment dates, and this test proved the accuracy of the [Original Database] dates, as explained in the July 19 post-verification comments . . . . Thus, the record does not justify application of [adverse facts available] under 19 U.S.C. § 1677e, because there is no gap in the record that must be filled with facts available and there is no justification for finding that Ghigi failed to cooperate to the best of its ability. Pls.’ Br. 48 (citing Ghigi/Zara’s Post-Verification Cmts.). Here, the court must determine if Commerce properly used facts available, a finding that turns on (1) whether Commerce’s practice of relying on the most recently submitted database as the basis for verification is reasonable; and (2) whether Ghigi’s proposal that Commerce consider the Original Database came too late. As to the first consideration, Commerce’s stated justification for its practice is sufficiently clear: its “practice is to rely on the most recently submitted databases as the basis for verification because such data is responsive to Commerce’s most recent supplemental questions.” Final IDM at 13 (emphasis added). Based on the record, it was not unreasonable for Commerce to decline to consider the Original Database after finding the dates in the Revised Database unverifiable. Commerce issued a supplemental questionnaire asking Ghigi to add payment date information to the Original Database. It is uncontested that in the process of compiling the Revised Database, in response to that request, Ghigi made a programming error which garbled the dates, and in turn Consol. Court No. 20-00023 Page 16 prevented the verifiers from matching them to the original invoices.12 Commerce cannot be said to err when it uses a respondent’s own responses to the Department’s last-in-time requests. Nor could it be said to err by not sua sponte reverting to the Original Database. Ghigi/Zara did not propose using the Original Database until after verification. As to the second consideration, as noted, Plaintiffs did not make their arguments in favor of using the Original Database until after verification. Importantly, Ghigi failed to inform Commerce of the errors until verification had already commenced. By that point, though, it was too late, for “[v]erification represents a point of no return.” Goodluck India Ltd. v. United States, 11 F.4th 1335, 1343 (Fed. Cir. 2021). “The purpose of verification is to test information provided by a party for accuracy and completeness.” Id. at 1343-44 (citation omitted) (cleaned up). If Commerce had accepted Ghigi’s proposal, it would be agreeing either to forego verification entirely with respect to that database, or to conduct another verification. On these facts, it was not unreasonable for Commerce to decline to use the unverified Original Database. Thus, the court finds that Commerce’s inability to verify the Revised Database created a gap in the record that justified the use of facts available, pursuant to 19 U.S.C. § 1677e(a)(2)(D) (stating that Commerce shall use facts available where an interested party “provides [information requested by Commerce] but the information cannot be verified”). Next, the court turns to whether Commerce’s use of an adverse inference when selecting from among the facts available was reasonable. Where Commerce determines that the use of facts available is warranted, it may apply adverse inferences to those facts when replacing a party’s information only if it makes the requisite 12 While it is Commerce’s practice to permit correction of minor errors, here Commerce determined the errors were not minor because they affected the majority of the payment dates for the company’s U.S. sales. See Final IDM at 12. Consol. Court No. 20-00023 Page 17 additional finding that a party has “failed to cooperate by not acting to the best of its ability to comply with a request for information from the [Department].” Id. § 1677e(b)(1); see also Nippon Steel, 337 F.3d at 1381 (citation omitted) (cleaned up) (“[S]ubsection (b) permits Commerce to use an inference that is adverse to the interests of a respondent in selecting from among the facts otherwise available, only if Commerce makes the separate determination that the respondent has failed to cooperate by not acting to the best of its ability to comply. The focus of subsection (b) is respondent’s failure to cooperate to the best of its ability, not its failure to provide requested information.”). Importantly, here the use of facts available requires a finding of missing information because the information on the record was unverifiable. The application of an adverse inference requires a finding with respect to a respondent’s behavior. Although the use of facts available was clearly warranted here, the application of adverse inferences “in selecting from among the facts otherwise available” was not. See 19 U.S.C. § 1677e(b)(1)(A). The problem with the Final Results is that Commerce based its finding that the application of an adverse inference was warranted on the same facts that it found justified its use of facts available: “Pursuant to [19 U.S.C. § 1677e(a)(2)(D)], we find that a determination based on the facts otherwise available is warranted because the information on payment data was not verifiable. Accordingly, we find that the application of partial adverse inferences under [19 U.S.C. § 1677e(b)(1)(A)] is warranted, as it applies to Ghigi’s U.S. payment date field.” Final IDM at 13 (emphasis added). This finding, however, only recites that information was missing because it was unverifiable. It says nothing about Ghigi’s behavior. As courts have explained in numerous decisions, the determination to use facts available is a separate determination from the application of adverse inferences. Each determination must be made separately, and each must be explained separately. See, e.g., Nippon Steel, 337 F.3d at Consol. Court No. 20-00023 Page 18 1381. Commerce’s single, conclusory assertion is inadequate to satisfy the statute because it does not explain the reasons for the application of an adverse inference and indeed seems to be based on Commerce’s inability to verify the information on payment data. See Final IDM at 13. In the Final Results, the Department failed to satisfy the statutory requirement that it make a determination as to whether a party failed to cooperate to the best of its ability. Accordingly, the Final Results are remanded for Commerce to determine whether Ghigi failed to cooperate to the best of its ability and, if the Department continues to find that it did, explain its adverse inference determination with reference to record evidence. If Commerce is unable to explain its determination on remand, it may not use an adverse inference when selecting from among the facts otherwise available. II. Commerce Properly Rejected Ghigi/Zara’s Post-Verification Arguments Disputing the Instructions for Reporting Protein Content and Pasta Shape The court next turns to whether Commerce erred when it rejected Ghigi/Zara’s post- verification arguments against the Department’s method of coding for protein content and pasta shape. The court finds no error here. The “basic purpose” of the antidumping statute is to “determin[e] current margins as accurately as possible.” Rhone Poulenc, Inc. v. United States, 899 F.2d 1185, 1191 (Fed. Cir. 1990). The burden of creating the administrative record lies with the interested parties; through questionnaires, Commerce asks for the information that it deems necessary to make its margin determinations. See BMW of N. Am. LLC v. United States, 926 F.3d 1291, 1295 (Fed. Cir. 2019). Gathering information during the questionnaire phase of a proceeding is often an iterative process, with Commerce issuing supplemental questionnaires and inviting respondents to contact the official assigned to their case if they have any questions, e.g., with respect to reporting Consol. Court No. 20-00023 Page 19 instructions. See 19 C.F.R. § 351.301(c)(1); see also Initial Quest. (General Instructions) (“If you have questions, we urge you to consult with the official in charge named on the cover page.”). From the outset, respondents are directed to comply fully with Commerce’s questions: Your response to the questionnaire should include all of the information requested. It is essential and in your interest that the Department receive complete information early in the proceeding to ensure a thorough and accurate analysis and to provide all parties the fullest opportunity to review and comment on your submission and the Department’s analysis. Initial Quest. (General Instructions). The time for fact-finding is limited by Commerce’s regulations. See 19 C.F.R. § 351.301 (setting forth the time limits for submitting factual information). Thus, there comes a time when Commerce ceases to gather information through questionnaires, and, under certain circumstances, undertakes to verify the information on the record. The antidumping statute provides that Commerce “shall verify all information relied upon in making . . . a final determination in a review under section 1675(a)” if “(A) verification is timely requested by an interested party,” and “(B) no verification was made under this subparagraph during the 2 immediately preceding reviews and determinations under section 1675(a) . . . of the same order, finding, or notice, except that this clause shall not apply if good cause for verification is shown.” 19 U.S.C. § 1677m(i); see also 19 C.F.R. § 351.307 (setting out, inter alia, procedures for verification of information). It has been said, as noted supra, that “[v]erification represents a point of no return.” Goodluck, 11 F.4th at 1343. That is, “verification is not intended to be an opportunity for submission of new factual information.” Ghigi Sales Verification Agenda (May 16, 2019) at 2, PR 111 (noting that “[n]ew information will be accepted at verification only when: (1) the need for that information was not evident previously; (2) the information makes minor corrections to information already on the record; or (3) the information corroborates, supports, or clarifies Consol. Court No. 20-00023 Page 20 information already on the record”). Verification is thus the culmination of an orderly process of information-gathering. “The purpose of verification is to test information provided by a party for accuracy and completeness.” Goodluck, 11 F.4th at 1343-44 (citation omitted). Although, there have certainly been cases where Commerce has accepted new information at verification, no sound reason has been provided to do so in this review. Here, after the questionnaire period and verification were complete, Ghigi/Zara argued for the first time that Commerce’s instructions on coding for protein content were ambiguous. See Ghigi/Zara Case Br. at 1. It further argued that notwithstanding the classification table’s shape categories, it was “correct to classify fusilli and cavatappi as normal short cuts rather than specialty short cuts, based on [its] production throughput rate (at multiple plants), [its] merchandising, and [its] pricing.” Ghigi/Zara Case Br. at 32. While not necessarily required to do so because they were made so late in the proceeding, Commerce addressed Ghigi/Zara’s claims. In doing so, Commerce did not credit these arguments in the Final Results. As to protein content, it found that its instructions to report protein content “as stated on the label” of the finished pasta product were plain and unambiguous and comported with its past practice: [T]he market reality is that “there is not a clearly defined method of identifying premium pasta other than the protein content marked on the packages.” Thus, Commerce’s reliance on the packaging label is an objective method to achieve a product comparison on a “consistent and transparent” basis because all of the physical characteristics are listed on the product label. Final IDM at 8 (footnote omitted); see also Pls.’ Br. attach. 2 at 28 (Issues and Decision Memorandum for Final Results of 2010-2011 Review) (“[T]he package label is a reliable source for the Department to use in identifying the physical characteristics, including protein content.”). Consol. Court No. 20-00023 Page 21 As to the shape classification table, Commerce cited case law where this Court has rejected similar arguments to those advanced by Plaintiffs and a prior determination rejecting arguments similar to those advanced by Ghigi/Zara. See infra at 11-12 (quoting Final IDM at 10-11). There is no serious dispute that the label is a useful source of protein content information to differentiate between premium and standard pasta. See Pls.’ Br. 29 (“[Plaintiffs] do[] not disagree with Commerce’s differentiation between standard and premium pasta based on protein content as expressed on the label. [Their] only disagreement is with Commerce’s failure to account for the difference in protein-testing protocol as between [the United States and Italy].”). With respect to the shape classification table, this Court has sustained Commerce’s rejection of proposed modifications to the table that were based on company-specific data, rather than evidence of industry-wide commercial practices.13 See, e.g., La Molisana, 2018 WL 3089242, at *4 (rejecting plaintiff’s argument for reclassification of its pasta based on company-specific line 13 The Federal Circuit has upheld the view that products may be considered identical, “despite the existence of minor differences in physical characteristics, if those minor differences are not commercially significant.” See Pesquera Mares Australes Ltda. v. United States, 266 F.3d 1372, 1384 (Fed. Cir. 2001) (interpreting 19 U.S.C. § 1677(16)(A)). Determining whether a physical difference between products is commercially significant, i.e., one “that merits distinguishing between identical and similar products,” is a fact-intensive inquiry. See Manchester Tank & Equip. Co. v. United States, 44 CIT __, __, 483 F. Supp. 3d 1309, 1316, 1317 (2020) (“Considering the record as a whole, Commerce has supported with substantial evidence its decision to accept as commercially significant the distinction between zinc and non-zinc coatings because zinc coating requires unique production processes, is specifically requested by customers, and leads to price variations.”). This inquiry may involve a consideration of whether and to what extent the industry at large treats the difference as significant. See Pesquera Mares, 266 F.3d at 1384, 1385 (cleaned up) (affirming Commerce’s finding that “the differences between superpremium and premium salmon do not warrant separate classification in an antidumping analysis,” where Commerce relied on record evidence of the “commercial practice of the world’s largest salmon farming countries whose salmon industries also exported to” the third country market, Japan). Relying on industry-wide data, instead of a smaller, company-specific dataset, avoids the risk of manipulation of sales information by the respondent. See id. at 1385 (“Indeed, if Commerce were to limit itself to consideration of the small volume of ‘premium’ sales of the particular exporter, it would risk market manipulation for antidumping purposes.”). Consol. Court No. 20-00023 Page 22 speeds and upholding Commerce’s method as reasonable); Prodotti Alimentari Meridionali, S.r.l. v. United States, 27 CIT 547, 549-550 (2003) (not reported in Federal Supplement) (rejecting the plaintiff’s argument that the distinction between two types of short cuts was meaningless based on similar production speeds and machinery). Plaintiffs’ real argument is that the model-match method requires modification with respect to protein content and shape to ensure a comparison of like products when Commerce makes its dumping determination. For Plaintiffs, the method for reporting protein content should be changed to account for differences in the ways that the United States and Italy measure protein content. Likewise, they argue that the shape classification table, at least with respect to specialty short cuts, is outdated and ought to be revised to account for advances in technology. The problem for Ghigi/Zara is that (1) it decided to modify the model-match method with respect to protein content and shape on its own without alerting Commerce to its activities,14 and (2) it did not raise the argument in favor of its preferred methods for calculating protein content and for classifying shapes during the questionnaire phase, when Commerce could have considered its claims, sought evidence of industry-wide changes, and, importantly, made findings on the proposed modifications to the methods based on substantial record evidence. Rather than contact Commerce with a question about the meaning of the purportedly ambiguous reporting instructions when it received the initial questionnaire, or propose alternate methods of reporting protein content 14 The twenty-second administrative review is not the first time Ghigi/Zara decided not to disclose its alternative methods for reporting protein content and shape to Commerce. As Plaintiffs note in their brief, Ghigi/Zara seems to have failed to bring its disputes about the modelmatch method to Commerce’s attention in the past two reviews in which it participated as a mandatory respondent. See Pls.’ Br. 13 n.4 (“There is no reason to believe that the companies’ approach to the protein coding changed over the course of these reviews.”). Ghigi/Zara’s alternative methods apparently went undetected until this review. Notably, verification was not conducted in either of those reviews. Consol. Court No. 20-00023 Page 23 or shape, Ghigi/Zara responded to the questionnaires in its own special way, “correcting” what it found to be flaws or ambiguities in the instructions, without alerting Commerce. See Pls.’ Br. 5 (arguing, with respect to protein content, that “Ghigi’s original reporting of PROTEINU was reasonable and correct, and the final results should restore Ghigi’s original designation”) & 44 (arguing, with respect to pasta shape, “[i]f Commerce’s coding causes products that are factually dissimilar to be matched, then the resulting margins are inaccurate. Correcting such an inaccuracy is not manipulation; it is correction”). Then, despite having used alternative reporting methods for protein content and shape, Ghigi/Zara certified that it had complied with Commerce’s instructions. Indeed, Ghigi/Zara advanced its arguments regarding protein content and shape only after its departures from Commerce’s questionnaire instructions came to light. In particular, Commerce learned for the first time at verification that Ghigi/Zara reported the protein content of its U.S. pasta sales, not according to the percentage on the packaging, but by converting the U.S. percentages to an Italian measurement. It also learned at verification that even though fusilli and cavatappi are shapes that have been identified as specialty short cuts in the classification table, Ghigi/Zara coded them as regular short cuts. By proceeding in this way, Plaintiffs deprived themselves of the opportunity to make their case before Commerce that a change to the method is warranted, and prevented Commerce from properly considering their proposed changes, and making findings accordingly.15 Changes to 15 For example, in New World Pasta, the respondent pasta producer (Ferrara) proceeded correctly by initially raising its concern that Commerce should add a product-matching criterion for die-type in defining the “foreign like product” during the questionnaire phase: In the antidumping review at issue here, Commerce originally chose four criteria to use in identifying the foreign like product: pasta shape, wheat type, presence of additives, and presence of enrichment. . . . Ferrara, in answering its first questionnaire, requested that a fifth criterion be added, representing the type of die used to extrude the pasta. . . . In response to a supplemental questionnaire Consol. Court No. 20-00023 Page 24 Commerce’s model-match method are not undertaken lightly. This Court and the Federal Circuit “have looked for ‘compelling reasons’ when Commerce modifies a model-match methodology in a review after having used that methodology in previous segments of the proceeding.” Manchester Tank & Equip. Co. v. United States, 44 CIT __, __, 483 F. Supp. 3d 1309, 1315 (2020); see also SKF USA, Inc. v. United States, 537 F.3d 1373, 1380 (Fed. Cir. 2008); Koyo Seiko Co. v. United States, 31 CIT 1512, 1517-18, 516 F. Supp. 2d 1323, 1331-32 (2007), aff’d 551 F.3d 1286 (Fed. Cir. 2008); Fagersta Stainless AB v. United States, 32 CIT 889, 894-95, 577 F. Supp. 2d 1270, 1276-77 (2008). “‘Compelling reasons’ require the agency to provide ‘compelling and convincing evidence that the existing model-match criteria are not reflective of the merchandise in question, that there have been changes in the relevant industry, or that there is some other compelling reason’ requiring the change.” Manchester Tank, 44 CIT at __, 483 F. Supp. 3d at 1315 (quoting Fagersta, 32 CIT at 894, 577 F. Supp. 2d at 1277). So too must respondents provide a compelling reason to change the model-match criteria by presenting their proposed modifications first to Commerce during the administrative process, and only if they fail to convince Commerce, by bringing their requesting explanation of why a fifth criterion should be added, Ferrara provided Commerce's verification, from the review conducted a year previously, that the surface texture of bronze-die and Teflon-die pastas were noticeably different. New World Pasta Co. v. United States, 28 CIT 290, 307, 316 F. Supp. 2d 1338, 1353 (2004) (emphasis added) (cleaned up); see also Prodotti Alimentari Meridionali, S.r.l. v. United States, 26 CIT 749 (2002) (not reported in the Federal Supplement) (remanding the plaintiff pasta producer’s challenge to Commerce’s separation of certain pasta shapes in the classification table, at Commerce’s request, so that it could reconsider its analysis); Prodotti Alimentari Meridionali, 27 CIT at 548 (sustaining Commerce’s shape classification method after remand, rejecting the plaintiff’s argument that “Commerce’s application of that methodology . . . [was] flawed”); Fagersta Stainless AB v. United States, 32 CIT 889, 889, 577 F. Supp. 2d 1270, 1273 (2008) (“During the course of the review, Plaintiff Fagersta Stainless AB . . . requested that Commerce modify its existing model-match methodology by adding an additional product criterion. Commerce rejected this request on the basis that Fagersta had not demonstrated that there were ‘compelling reasons’ to do so. Plaintiff challenges this determination [before the Court].”). Consol. Court No. 20-00023 Page 25 arguments to this Court. Ghigi/Zara has failed to do so. The court finds no error in Commerce’s adherence to its long-standing model-match method with respect to protein content and pasta shape. CONCLUSION AND ORDER For the foregoing reasons, it is hereby ORDERED that the Final Results are remanded for Commerce to reconsider and explain its adverse inference determination in a manner that satisfies the requirements of 19 U.S.C. § 1677e(b) and to support its redetermination with substantial record evidence; and it is further ORDERED that if, on remand, Commerce is unable to explain its determination, it may not use adverse inferences when selecting from among the facts otherwise available. /s/ Richard K. Eaton Judge Dated: 1RYHPEHU 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.