LA MOLISANA v. UNITED STATES (2025)

La Molisana S.p.A. and Valdigrano Di Flavio Pagani S.r.L. (collectively, “La Molisana”) challenge the final results of the United States Department of Commerce's (“Commerce”) twenty-third administrative review of the antidumping order on certain pasta from Italy. Certain Pasta From Italy: Final Results of Antidumping Duty Administrative Review and Final Determination of No Shipments; 2018–2019, 86 Fed. Reg. 28,336, 28,336–38 (May 26, 2021). The Court of International Trade (“Trade Court”) sustained the final results, concluding that La Molisana had not demonstrated that the alleged flaws in Commerce's model-match methodology were commercially significant. For the following reasons we vacate-in-part, affirm-in-part, and remand for further proceedings.

Background

I

The Tariff Act of 1930, as amended, establishes a remedial regime to combat unfair trade practices. If Commerce “determines that a class or kind of foreign merchandise is being, or is likely to be, sold in the United States at less than its fair value,” it will “impose[ ] upon such merchandise an antidumping duty ․ in an amount equal to the amount by which the normal value exceeds the export price.” 19 U.S.C. § 1673.

When establishing a dumping margin, Commerce must first identify the “foreign like product” that will form the basis for a comparison to the subject merchandise exported to the United States. Pesquera Mares Australes Ltda. v. United States, 266 F.3d 1372, 1375 (Fed. Cir. 2001). “Foreign like product” means merchandise that is “identical in physical characteristics with” the subject merchandise. 19 U.S.C. § 1677(16)(A). Commerce uses a model-match methodology that considers the physical characteristics of the relevant products to create control numbers (“CONNUMs”), which it assigns to the products to categorize them for comparison. See Goodluck India Ltd. v. United States, 11 F.4th 1335, 1338 n.1 (Fed. Cir. 2021). In other words, Commerce compares products sold in the United States (the subject merchandise) and foreign products with the same CONNUMs to determine the dumping margin.

II

Commerce published the antidumping order governing the import of certain Italian pasta in this case in 1996. Notice of Antidumping Duty Order and Amended Final Determination of Sales at Less Than Fair Value: Certain Pasta From Italy, 61 Fed. Reg. 38,547 (July 24, 1996). Relevant here, the disputed physical characteristic used to distinguish and assign CONNUMs to pasta is protein content. Commerce assigns pasta with a protein content of 12.5 percent or more as “1” (premium quality) for its CONNUM code and pasta with less than 12.5 percent as “2” (standard quality).

Commerce began using protein content as a distinguishing physical characteristic of pasta during the twelfth administrative review, covering July 1, 2007, to June 30, 2008. Commerce explained “it is generally accepted that pasta cooking quality can be explained by the differences in gluten and protein composition of the semolina input.” J.A. 2615. Commerce relied specifically on protein content because “protein content can be determined on both the semolina input and the finished product,” but “gluten testing cannot be performed on the finished product.” Id. Commerce determined that “the industry standard for superior semolina is that its protein content must exceed 12.5 percent.” J.A. 2614–15.

In other words, Commerce determined that the protein content of the finished pasta was a proxy for the semolina quality, which is indicative of the quality of the pasta.

III

This appeal concerns the twenty-third administrative review (“AR23”) covering the period from July 1, 2018, to June 30, 2019.

A

In AR23, Commerce determined the protein content of pasta by asking respondents to identify the protein “as stated on the label of the respective product.” J.A. 1258.

La Molisana challenged Commerce's methodology for determining protein content of pasta, arguing that it causes dissimilar goods sold in the United States and Italy to be treated as identical and identical goods sold in the United States and Italy to be treated as dissimilar. Specifically, La Molisana faulted Commerce's methodology for failing to account for: (1) the impact of U.S. Food and Drug Administration (“FDA”) mandated rounding rules on the protein content listed on the label of the U.S. product; (2) the different nitrogen-to-protein conversion factors used in calculating protein content in the United States versus Italy; and (3) evidence that “the 12.5% breakpoint between standard and premium pasta does not reflect current market reality.” J.A. 5–6. Commerce rejected all three arguments during AR23.

First, Commerce rejected La Molisana's assertion that the FDA's requirement to report grams of protein per pasta serving rounded to the nearest gram leads to inaccurate comparisons between subject merchandise and foreign like products because pasta with less than 12.5 percent protein content will sometimes be reported on the label as having a protein content that equates to greater than 12.5 percent due to the required rounding up. “Commerce explained that it did not consider the differences caused by the different rounding standards to be commercially significant such that they would be a basis to alter the coding for the protein content physical characteristic.” J.A. 8247. Commerce then turned to La Molisana's second argument: that Commerce should adjust its model-match methodology to account for the different nitrogen conversion factors used by the United States and Italy in calculating the number of grams of protein. Commerce again rejected La Molisana's argument, explaining it had “considered and rejected the claims regarding rounding and nitrogen conversion factors in prior reviews and in doing so has repeatedly emphasized the importance of transparency and consistency.” J.A. 8248. Commerce concluded:

Given that we have found “there is not a clearly defined method of identifying premium pasta other than the protein content marked on the packages,” we do not see a basis to find the discrepancy in protein measurement standards between the U.S. and Italian markets as commercially significant when the market perception of premium pasta or non-premium pasta relies on information readily available to consumers, namely the packaging label associated with the pasta in the marketplace.

J.A. 8249.

Commerce next considered La Molisana's third argument, that the “12.5 percent breakpoint between standard and premium pasta is not reflective of either the U.S. or Italian pasta market.” Id. In support of this argument, La Molisana relied on a report prepared by counsel for another mandatory respondent that presented price and protein content information for a sample of pasta products sold in one food retail chain in Italy and four food retailers in a suburb of Washington, D.C. (“Market Report”). The Market Report concluded that the “true breakpoint between standard and premium pasta is 13.5 percent protein content,” J.A. 8249, citing: (1) the Bologna Grain Exchange's decision to redefine superior semolina as having 13.5 percent or greater protein content, and (2) the survey data purportedly showing pasta with a protein content of 12.5 percent is standard pasta with the minimum accepted protein among supermarket sellers. See J.A. 8249–50. Commerce rejected the conclusions drawn in the Market Report because it surveyed only “four supermarkets in a small geographic region of the United States,” and La Molisana made no attempt “to address the potential for manipulation in choice of purchases or to support a claim that these purchases are reflective of the entire U.S. market for pasta products.” Id. Commerce also rejected the Bologna Grain Exchange's selection of a new 13.5 percent protein breakpoint because “Commerce relied on the breakpoints of three separate Italian commodity exchanges” when it adopted the 12.5 percent protein breakpoint. J.A. 8250. Commerce explained “a single exchange's breakpoint is not sufficient evidence of an industry-wide change in standards from semolina and thus that it is not a compelling reason to change the instructions for reporting protein content.” Id.

Commerce thus concluded that La Molisana had not presented compelling reasons for it to change its model-match methodology for reporting protein content and declined to do so.

B

La Molisana appealed to the Trade Court, asking that court to “remand th[e] matter to Commerce with instructions to adjust its model-match method for coding protein content and revise La Molisana's and Valdigrano's rates accordingly.” La Molisana S.p.A. v. United States, 633 F. Supp. 3d 1266, 1270 (Ct. Int'l Trade 2023). The Trade Court sustained Commerce's final results, concluding that substantial evidence supported Commerce's determinations.

First, the Trade Court found no error in Commerce's conclusion that “differences in Italian and U.S. protein measurement standards and rounding rules were not commercially significant.” Id. at 1275. In so holding, the court explained “[i]t is unrebutted that consumers rely on packaging information when making pasta purchasing decisions, and that coding for protein content based on the nutrition label fosters transparency and consistency in CONNUM-building.” Id. At the same time, the court acknowledged “that Commerce's reliance on the finding that customers make purchasing decisions based on information found on a pasta product's packaging departs from the relevant inquiry, which focuses on the physical characteristics of the product, not its packaging.” Id. at 1275 n.8. Nevertheless, the Trade Court concluded La Molisana “failed to demonstrate that Commerce's conclusion—that values other than those readily available to consumers on the packaging label were not commercially significant—was unreasonable.” Id. at 1275.

As for the breakpoint between standard and premium pasta, the Trade Court concluded it “cannot fault Commerce for finding that the Market Report was insufficient to support Plaintiffs’ claim that the 12.5% breakpoint is out of step with current industry-wide standards because no serious argument can be made that the report is representative of the entire industry either in the United States or in Italy.” Id. at 1274. The court explained that when “Commerce has reconsidered its model-match criteria in the past, it has stated that for data to be ‘industry-wide’ it must be public, published information” and, “[i]n contrast, the report here was prepared for presentation to Commerce, and bears no indicia of having been publicized or published to or by the industry at large.” Id. Addressing the Bologna Grain Exchange's new definition, the court agreed there is “nothing in the report that indicates that this single Bologna exchange represents the entire market or even a large portion of it.” Id. The court further noted that this grain exchange “was one of three Italian exchanges considered by Commerce [in adopting the 12.5% breakpoint] ․ [and] no evidence from the other two exchanges is included.” Id. The Trade Court thus concluded, “Commerce reasonably found the Market Report insufficient to support a change in the standard-to-premium breakpoint from 12.5% to 13.5%.” Id.

La Molisana appeals. We have jurisdiction pursuant to 28 U.S.C. § 1295(a)(5).

Discussion

I

We review decisions of the Trade Court de novo, applying anew the same standard it used. Sunpreme Inc. v. United States, 946 F.3d 1300, 1308 (Fed. Cir. 2020) (en banc). Under that standard, we must uphold the agency's determinations unless they are “unsupported by substantial evidence on the record, or otherwise not in accordance with law.” 19 U.S.C. § 1516a(b)(1)(B)(i); see also Sunpreme, 946 F.3d at 1308.

Under the Tariff Act, Commerce must compare merchandise that is identical in physical characteristics:

(16) Foreign like product

The term “foreign like product” means merchandise in the first of the following categories in respect of which a determination for the purposes of part II of this subtitle can be satisfactorily made:

(A) The subject merchandise and other merchandise which is identical in physical characteristics with, and was produced in the same country by the same person as, that merchandise.

19 U.S.C. § 1677(16)(A) (emphasis added). We have previously held that products may be “considered to be identical despite the existence of minor differences in physical characteristics, if those minor differences are not commercially significant.” Pesquera Mares, 266 F.3d at 1384.

Commerce will not modify its model-match methodology unless “compelling reasons” exist to do so. See SKF USA, Inc. v. United States, 537 F.3d 1373, 1377–80 (Fed. Cir. 2008). In reviewing Commerce's determination on whether compelling reasons exist to modify its methodology, we apply the substantial evidence standard, asking whether the model-match methodology is consistent with 19 U.S.C. § 1677(16)(A). It follows that compelling reasons to modify the existing model-match methodology would exist when that methodology results in comparison of subject merchandise and foreign like products that have commercially significant physical differences. This is consistent with the Trade Court's approach below, where it stated that compelling reasons are present when “ ‘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.” La Molisana, 633 F. Supp. 3d at 1272 (quoting Manchester Tank & Equip. Co. v. United States, 483 F. Supp. 3d 1309, 1315 (Ct. Int'l Trade 2020)).

With this legal framework in mind, we turn to the issues raised on appeal.

II

We address La Molisana's first two arguments together: (1) FDA-mandated rounding rules result in different protein levels displayed on U.S. product labels compared to identical products in Italy; and (2) comparison of products without adjusting for the different nitrogen-to-protein conversion factors used in calculating protein content in the United States versus Italy results in “identical product[s] [being] treated differently purely because of scalar differences.” Appellants’ Br. 22. We find both arguments persuasive for the same reason: the alleged sources of error result in inaccuracies in the comparison of identical products. The statute requires the foreign like product be “identical in physical characteristics” to the subject merchandise in the United States, and here Commerce's model-match methodology does not satisfy this requirement. 19 U.S.C. § 1677(16)(A) (emphasis added).

The FDA requires that labels for food sold in the United States shall include “[a] statement of the number of grams of protein in a serving, expressed to the nearest gram, except that if a serving contains less than 1 gram.” 21 C.F.R. § 101.9(c)(7). The impact of the FDA's protein per serving rounding requirement can be seen in the illustrative table from La Molisana's brief:

Appellants’ Br. 17. Notably, the Government does not contest the accuracy of this chart. See generally Appellee's Br. 20–24. As shown, pasta with an actual protein content of 6.51 grams and an actual protein percentage of 11.63 percent would list a protein content of 7 grams on the label, resulting in a protein percentage of 12.50 percent, which is coded as premium pasta, causing the pasta to be compared against foreign like (Italian) premium pasta. Based on its actual protein content, this pasta is standard pasta that should be compared against foreign like (Italian) standard pasta. By relying on the listed protein content, which is rounded up under FDA rounding requirements, Commerce fails to compare products “identical in physical characteristics” in violation of the requirements of 19 U.S.C. § 1677(16)(A).

Commerce reasoned “that it did not consider the differences caused by the different rounding standards to be commercially significant” because it had previously determined that “slight differences [between actual protein content and the label content] are not readily apparent to customers of the finished product and, therefore, are not commercially significant.” J.A. 8247. The Trade Court agreed, noting that “there was no evidentiary basis to conclude that differences in ․ rounding rules, which is not information readily available to consumers, mattered in the marketplace.” La Molisana, 633 F. Supp. 3d at 1273. But, as the Trade Court correctly acknowledged, Commerce's approach “departs from the relevant inquiry, which focuses on the physical characteristics of the product, not its packaging.” Id. at 1275 n.8 (emphasis added).

Moreover, Commerce's conclusion that the allegedly “slight” differences in protein content caused by rounding are not commercially significant is belied by its own prior statements. Commerce explicitly determined that the amount of protein in pasta is commercially significant when it chose to designate the amount of protein in pasta as a proxy for pasta quality. Commerce “verified that physical differences exist” between various pastas based on the different wheat (i.e., semolina) qualities used. J.A. 2610. Commerce determined that (1) “the cost of the highest grade of semolina is materially more than that of the lowest grade”; (2) the quality differences in semolina were reflected in pasta prices; and (3) that quality differences in semolina are “commercially significant and an appropriate criterion for product matching.” Id. Commerce then looked to published industry standards for determining quality differences in semolina and examined definitions for superior semolina. It noted that, at that time, the Milan Grain Exchange, Bologna Grain Exchange, and Milan Commodities Exchange all defined superior semolina as having a minimum protein content of 12.5 percent. Commerce concluded that, accordingly, “the industry standard for superior semolina is that its protein content must exceed 12.5 percent.” J.A. 2614–15. Commerce also recognized that “protein content can be determined on both the semolina input and the finished product.” J.A. 2615. Commerce thus recognized that differences in pasta protein content indicate differences in semolina quality and are thus commercially significant and an appropriate criterion for product matching. Accordingly, given Commerce's own critical distinction between pasta with 12.5 percent or more protein versus pasta with under 12.5 percent protein, Commerce cannot now assert that the differences in protein content caused by rounding in the United States as identified by La Molisana are not commercially significant. In other words, because Commerce itself designated protein content as commercially significant and a proxy for pasta quality, it cannot reasonably rely on protein calculation methods that undisputedly inject inaccuracies and characterize those inaccuracies as commercially insignificant.

Commerce's and the Trade Court's reliance on the goals of transparency and consistency fares no better. Commerce attempts to justify its “reliance on the packaging label [by explaining it] 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.” J.A. 8248; La Molisana, 633 F. Supp. 3d at 1275 (emphasizing that “coding for protein content based on the nutrition label fosters transparency and consistency in CONNUM-building”). While we agree that transparency and consistency are reasonable goals, the statute requires accuracy in comparison of the subject merchandise to like products “identical in physical characteristics.” 19 U.S.C. § 1677(16)(A). Commerce's goals of transparency and consistency cannot override the statutory requirement of identicality in physical characteristics. As we have recognized, “[a]n overriding purpose of Commerce's administration of antidumping laws is to calculate dumping margins as accurately as possible.” Yangzhou Bestpak Gifts & Crafts Co. v. United States, 716 F.3d 1370, 1379 (Fed. Cir. 2013) (citing Rhone Poulenc, Inc. v. United States, 899 F.2d 1185, 1191 (Fed. Cir. 1990)); see also Borlem S.A.-Empreedimentos Industriais v. United States, 913 F.2d 933, 937 (Fed. Cir. 1990) (“The law does not require, nor would it make sense to require, reliance on data which might lead to an erroneous result.”). In other words, accuracy is a key consideration.

We turn now to the nitrogen conversion factors. We agree with La Molisana that failing to account for the different nitrogen conversion factors in the United States versus Italy creates inaccuracies in the comparison of identical products. The grams of protein in a product are calculated by multiplying the determined nitrogen content (nitrogen units) by a nitrogen-to-protein conversion factor. In the United States, the nitrogen conversion factor, as set by FDA rules, is 6.2510, while the Italian conversion factor is 5.7110, as set by European Union standards. J.A. 8245. The following example demonstrates the impact of the different conversion factors on the resulting protein percentage:

In this example, physically identical products would be classified differently in the United States versus Italy based on the protein displayed on the packaging. As shown above, Commerce fails to account for these scalar differences, which results in its failure to compare like products based on “physical characteristics.” Accordingly, Commerce has violated § 1677(16)(A).

As with the FDA-mandated rounding rules, Commerce's only justification for declining to adjust its model-match methodology to account for the different nitrogen conversion factors is “the importance of transparency and consistency.” J.A. 8248. Commerce explained it did not “see a basis to find the discrepancy in protein measurement standards between the U.S. and Italian markets as commercially significant when the market perception of premium pasta or non-premium pasta relies on information readily available to consumers, namely the packaging label associated with the pasta in the marketplace.” J.A. 8249. This reasoning misses the mark because (1) it elevates the goals of transparency and consistency over the “identical in physical characteristics” requirement set forth in § 1677(16)(A), and (2) these goals are not lost if Commerce performs a mathematical conversion such that the subject merchandise can be compared with like foreign products.

We hold that a compelling reason to modify the model-match methodology exists because Commerce's current methodology fails to adhere to the statutory requirement to compare goods on the basis of identical physical characteristics. For the reasons stated above, we conclude that Commerce's determinations regarding rounding and nitrogen conversion factors are not supported by substantial evidence. Therefore, we vacate the Trade Court's judgment in these respects and remand for further proceedings consistent with this opinion.

III

Turning now to the final issue, La Molisana argues that Commerce erred in determining that the Market Report and Bologna Grain Exchange definition of “superior” semolina does not constitute a compelling reason to modify the percentage breakpoint for distinguishing between standard and premium pasta from 12.5 percent to 13.5 percent. We are not persuaded by La Molisana's argument. Based on the record evidence, a reasonable fact finder could accept that La Molisana failed to present compelling evidence that the current 12.5 percent breakpoint is inconsistent with industry standards.

First, Commerce reasonably explained that the Market Report relied on pasta purchased from only “four supermarkets in a small geographic region of the United States,” and La Molisana made no attempt “to address the potential for manipulation in choice of purchases or to support a claim that these purchases are reflective of the entire U.S. market for pasta products.” J.A. 8249–50. We cannot say that Commerce erred in finding that this survey data did not constitute a compelling reason to change its approach. The survey was limited to four stores in an affluent Washington, D.C. metro area, and a reasonable mind could accept that such limited survey data is not reflective of the entire U.S. pasta market.

Nor can we say that Commerce erred by not adopting a 13.5 percent breakpoint based on La Molisana's citation to a single commodity exchange's adoption of that breakpoint. La Molisana's Market Report included a screenshot of the Bologna Grain Exchange's website, which now defines “ ‘superior’ semolina as having 13.5 percent or greater protein content.” J.A. 8250. La Molisana only provided evidence from one of the commodity exchanges that Commerce considered in adopting the 12.5 percent protein breakpoint. Commerce explained “[t]he plain meaning of ‘industry-wide’ connotes an entire industry or, at the very least, predominance or prevalence within an industry ․ [and] a single exchange's breakpoint is not sufficient evidence of an industry-wide change in [semolina] standards.” Id.

Because a reasonable mind could conclude that La Molisana failed to present compelling evidence of a change in the protein breakpoint, we affirm the judgment of the Trade Court on this issue.

Conclusion

We have considered La Molisana's remaining arguments and do not find them persuasive. For the foregoing reasons, we vacate-in-part, affirm-in-part, and remand for further proceedings consistent with this decision.

VACATED-IN-PART, AFFIRMED-IN-PART, AND REMANDED

Costs

No costs.

Stoll, Circuit Judge.