Asociacion De Exportadores E Industriales v. United States

Opinion

United States Court of Appeals for the Federal Circuit

______________________

ASOCIACIÓN DE EXPORTADORES E INDUSTRIALES DE ACEITUNAS DE MESA, AGRO SEVILLA ACEITUNAS S. COOP. AND., ANGEL CAMACHO ALIMENTACIÓN, S.L., Plaintiffs-Appellants

ACEITUNAS GUADALQUIVIR, S.L.U., Plaintiff

v.

UNITED STATES, COALITION FOR FAIR TRADE IN RIPE OLIVES, Defendants-Appellees

______________________

2023-1162

______________________

Appeal from the United States Court of International Trade in No. 1:18-cv-00195-GSK, Judge Gary S. Katzmann.

______________________

Decided: May 20, 2023

______________________

MATTHEW P. MCCULLOUGH, Curtis, Mallet-Prevost, Colt & Mosle LLP, Washington, DC, argued for plaintiffsappellants . Also represented by JAMES BEATY, JAMES P. DURLING, DANIEL L. PORTER.

2 ASOCIACIÓN DE EXPORTADORES E INDUSTRIALES v. US

TARA K. HOGAN, Commercial Litigation Branch, Civil Division, United States Department of Justice, Washington , DC, argued for defendant-appellee United States. Also represented by BRIAN M. BOYNTON, PATRICIA M. MCCARTHY, SONIA W. MURPHY; ELIO GONZALEZ, Office of the Chief Counsel for Trade Enforcement and Compliance, United States Department of Commerce, Washington, DC.

RAYMOND PARETZKY, McDermott Will & Emery LLP, Washington, DC, argued for defendant-appellee Coalition for Fair Trade in Ripe Olives. Also represented by DAVID JOHN LEVINE.

______________________

Before PROST, BRYSON, and STARK, Circuit Judges.

BRYSON, Circuit Judge.

Appellants, three organizations of Spanish olive producers (collectively “Asemesa”), appeal from a decision of the Court of International Trade (“the Trade Court”) regarding a countervailing duty imposed on olives imported from Spain. Asemesa argues that an order from the Department of Commerce imposing a countervailing duty on imported olives was contrary to law and that the Trade Court should have overturned the order. The United States and the Coalition for Fair Trade in Ripe Olives argue that Commerce’s factual findings were supported by substantial evidence and that the Trade Court’s decision should be upheld. We affirm.

I

1. Under the Tariff Act of 1930, Congress authorized the Department of Commerce to impose countervailing duties as needed to offset subsidies granted by foreign countries on goods exported to the United States. See Sioux Honey Ass’n v. Hartford Fire Ins. Co., 672 F.3d 1041, 1046– 47 (Fed. Cir. 2012). If, after an investigation, Commerce finds that there was such a subsidy for particular imported

ASOCIACIÓN DE EXPORTADORES E INDUSTRIALES v. US 3

products, the International Trade Commission is required to conduct a parallel investigation to determine whether a domestic industry is being injured, threatened with being injured, or kept from being established by the subsidized imports. If the two agencies both make affirmative findings , Commerce is required to impose “a countervailing duty . . . equal to the amount of the net countervailable subsidy.” 19 U.S.C. § 1671(a).

A foreign government will sometimes subsidize the production of raw agricultural products, which are then processed into finished goods before they are imported into the United States. In such cases, it would be futile for Commerce to impose a duty on the subsidized raw product, which is not the product that is imported, so Commerce is authorized, in certain instances, to impose a duty on the finished product. In particular, Commerce is allowed to impose a countervailing duty on finished agricultural products with subsidized raw ingredients, but only if “the demand for the prior stage product is substantially dependent on the demand for the latter stage product, and the processing operation adds only limited value to the raw commodity.” 19 U.S.C. § 1677–2.

2. The European Union’s Common Agricultural Policy includes subsidies for raw olives. Those subsidies are provided to Spanish farmers through the EU’s “Basic Payment Scheme,” which provides direct subsidies to Spanish olive growers who meet its eligibility requirements.

Olives are rarely sold to consumers in raw form. The majority of olives are processed into olive oil. Even table olives, however, require significant processing. Raw olives are extremely bitter and must be cured to remove that natural bitterness before being consumed as table olives.

Olive varietals can be divided into three biologically distinct categories. “Mill” varietals are those that naturally produce olives suitable for processing into olive oil. “Table” varietals yield olives suitable for eating. “Dual-

4 ASOCIACIÓN DE EXPORTADORES E INDUSTRIALES v. US

use” varietals can produce olives suitable for either application , depending on the manner in which they are cultivated . Mill olives are cultivated according to practices that maximize oil production, whereas table olives are cultivated following practices that maximize size and flavor. Dual-use varietals are cultivated in different ways depending on whether they are intended to produce table olives or mill olives.

3. Following an investigation, Commerce published a preliminary determination in November 2017, in which it found that countervailable subsidies were being provided to producers and exporters of ripe olives from Spain. On July 25, 2018, the International Trade Commission notified Commerce that it had determined that the domestic olive industry was materially injured by the importation of subsidized table olives from Spain. Commerce then imposed a countervailing duty on imported Spanish table olives pursuant to its authority under 19 U.S.C. §§ 1671(a) and 1677–2. Ripe Olives from Spain, 83 Fed. Reg. 37469 (Dep’t of Commerce Aug. 1, 2018).

4. Asemesa challenged Commerce’s imposition of the duty on Spanish table olives. Asemesa argued that Commerce had failed to show that the market for raw olives was “substantially dependent” on the market for table olives, as required by 19 U.S.C. § 1677–2. At that time, Commerce had defined the prior stage product as all raw olives and had defined the latter stage product as table olives. Employing data from the Spanish government, Commerce found that 8 percent of all Spanish raw olives were ultimately sold as table olives. Based on the evidence before

ASOCIACIÓN DE EXPORTADORES E INDUSTRIALES v. US 5

it, Commerce found that the demand for raw olives was substantially dependent on the demand for table olives. 1 The Trade Court reversed Commerce. Asociación de Exportadores e Industriales de Aceitunas de Mesa v. United States (Asemesa I), 429 F. Supp. 3d 1325 (Ct. Int’l Trade 2020). The court concluded that the evidence that table olives accounted for 8 percent of the demand for raw olives did not show that the demand for raw olives was “substantially dependent” on the demand for table olives. Id. at 1344. The court further held that “Commerce deviated from its past interpretation of ‘substantially dependent,’ which [Commerce] previously found to include most or at least half of the demand of the raw agricultural product.” Id. at 1345. Accordingly, the court remanded the case to Commerce for further analysis. Id. at 1352.

5. On remand, Commerce redefined the market for the prior stage product as the raw olives that the olive industry considers principally suitable for use in the production of table olives, i.e., olives from table olive varietals and dual- use varietals that are cultivated for processing into table olives. Nearly all olives that are cultivated to produce table olives are ultimately processed into table olives. See J.A. 11241 (reporting that 96 percent of such olives were processed into table olives in 2016, the relevant year for purposes of this case).

Once again, the Trade Court rejected Commerce’s analysis . The court reasoned that Commerce’s market definition would “render the requirements of Section 1677–2 largely self-fulfilling.” Asociación de Exportadores e Industriales de Aceitunas de Mesa v. United States (Asemesa II), 523 F. Supp. 3d 1393, 1407 (Ct. Int’l Trade 2021). Although

1 It is undisputed that the second requirement of section 1677–2, that “the processing operation adds only limited value to the raw commodity,” was satisfied in this case.

6 ASOCIACIÓN DE EXPORTADORES E INDUSTRIALES v. US

the Trade Court rejected Commerce’s definition of the relevant market, it agreed with Commerce that the relevant market for the prior stage product need not be all olives grown in Spain. Accordingly, the court remanded the case to Commerce for a second time to correctly define the relevant market for the prior stage product and analyze whether the demand for the prior stage product was substantially dependent on the demand for table olives.

6. Commerce again redefined the relevant market for the prior stage product, this time defining that market as consisting of the olives from varietals that the Spanish government considers suitable for processing into table olives, including dual-use varietals. 2 Those varietals include manzanilla, gordal, carrasqueña, and hojiblanca olives. Cacereña and “other” dual-use varietal olives also fit Commerce ’s new market definition; however, Commerce did not have reliable data on the processing of those varietals, so it excluded them from its analysis. The Spanish government considers manzanilla, gordal, and carrasqueña olives suitable only for processing into table olives. It considers hojiblanca and cacereña olives to be dual-use varietal olives, suitable for use as either table olives or in the production of olive oil.

Relying on data from the Spanish government and the Agencia de Información y Control Alimentarios (the Spanish Food Information and Control Agency, or “AICA”), Commerce calculated that 55.28 percent of all olives from varietals suitable for processing into table olives were indeed sold as table olives. J.A. 62. Commerce adopted the

2 That market definition differs from the market Commerce identified in Asemesa II because that market definition includes all olives from table and dual-use varietals . In Asemesa II, Commerce’s market definition excluded olives from table and dual-use varietals that were cultivated for olive oil.

ASOCIACIÓN DE EXPORTADORES E INDUSTRIALES v. US 7

Trade Court’s interpretation of the “substantially dependent ” provision in section 1677–2 as requiring that more than half of the prior stage product be processed into the relevant finished good. Accordingly, Commerce determined that the demand for olive varietals suitable for processing into table olives was substantially dependent on the demand for table olives, and that a countervailing duty on table olives from Spain was warranted to offset the subsidies provided to Spanish olive growers.

This time, the Trade Court sustained Commerce’s analysis. Asociación de Exportadores e Industriales de Aceitunas de Mesa v. United States (Asemesa III), 589 F. Supp. 3d 1346 (Ct. Int’l Trade 2022).

7. Asemesa now appeals the Trade Court’s determination in Asemesa III. Asemesa argues that Commerce’s interpretation of the statute was contrary to law, and that Commerce’s factual analysis was not supported by substantial evidence. Although our interpretation of section 1677–2 and our analysis of the factual record in this case differ from the Trade Court’s, we agree with that court’s ultimate conclusion on both issues.

II

A

Section 1677–2 was designed to empower Commerce to address attempts to circumvent countervailing duty liability . Enacted as part of the Omnibus Trade and Competitiveness Act of 1988, section 1677–2 authorized Commerce to impose countervailing duties on processed agricultural goods that were not themselves subsidized but were made from subsidized raw products.

Senator Baucus, one of the proponents of section 1677– 2, explained that its purpose was “to fix a glitch in the law.” 133 Cong. Rec. 17,765 (1987). Under the statutory scheme in place prior to the enactment of section 1677–2, the Trade Court had held that Commerce lacked the power to impose 8 ASOCIACIÓN DE EXPORTADORES E INDUSTRIALES v. US

countervailing duties on finished agricultural goods when the producers of those goods benefitted from subsidies received by producers of the raw agricultural products that were used to prepare those goods. See Canadian Meat Council v. United States (Pork from Canada), 661 F. Supp. 622 (Ct. Int’l Trade 1987).

In Pork from Canada, Canada subsidized live swine, but not processed pork meats, which were the products imported into the United States. Commerce imposed a countervailing duty on the processed pork in order to offset the Canadian subsidies on live swine. 3 Before the Trade Court, however, the Canadian pork producers successfully argued that Commerce lacked statutory authority to impose a countervailing duty on pork when the subsidy was only on swine.

Section 1677–2 empowered Commerce to combat the circumvention of existing countervailing duty law in that manner. 133 Cong. Rec. 17,765 (characterizing the outcome in Pork from Canada as “disturbing”); see also Pork from Canada, 661 F. Supp. at 629 (proposing that, “[i]f the statutory approach to upstream subsidies [was] inadequate ,” it was up to “Congress to remedy any deficiency”).

Section 1677–2 prescribes the conditions under which Commerce may treat a subsidy on a raw agricultural product as a subsidy on the finished good for countervailing duty purposes. In full text, section 1677–2 provides:

3 Commerce’s theory was that, under the statutory scheme in place prior to the 1988 Act, swine was an input product used in the production of pork, making the subsidy on swine an “upstream subsidy” on pork subject to countervailing duty law. See 19 U.S.C. § 1671(g) (repealed 1988) (providing that Commerce may consider “upstream subsidies ” for countervailing duty purposes).

ASOCIACIÓN DE EXPORTADORES E INDUSTRIALES v. US 9

In the case of an agricultural product processed from a raw agricultural product in which— (1) the demand for the prior stage product is substantially dependent on the demand for the latter stage product, and (2) the processing operation adds only limited value to the raw commodity, countervailable subsidies found to be provided to either producers or processors of the product shall be deemed to be provided with respect to the manufacture, production , or exportation of the processed product.

B

The central question in this case is what it means for the demand for a prior stage product to be “substantially dependent” on the demand for a latter stage product within the meaning of section 1677–2.

1

Asemesa argues that section 1677–2 was meant to codify Commerce’s original approach in Pork from Canada, 50 Fed. Reg. 25097 (Dep’t of Commerce June 17, 1985), and Rice from Thailand, 51 Fed. Reg. 12356-02 (Dep’t of Commerce April 10, 1986), the two cases that led Congress to add section 1677–2 to the Tariff Act. Asemesa cites a statement by Senator Grassley, a proponent of section 1677–2, describing “the rule codified in the proposed amendment” as the rule Commerce applied in Pork from Canada and Rice from Thailand. 133 Cong. Rec. 17765; see also H.R. Rep. No. 100-576, 588 (1988) (“The Senate amendment codifies and clarifies Commerce[’s] practice.”).

In Pork from Canada, Commerce found that the demand for slaughtered and quartered swine is “by far the predominant determinant of the demand for live swine.” 50 Fed. Reg. at 25099. In Rice from Thailand, Commerce stated that “an important criterion is the degree to which 10 ASOCIACIÓN DE EXPORTADORES E INDUSTRIALES v. US

the demand for the prior stage product is dependent on the demand for the latter stage product.” 51 Fed. Reg. at 12358. Commerce explained that “[a]lmost all of the raw agricultural product, paddy or unmilled rice, is dedicated to the production of milled rice,” id., which Commerce regarded as sufficient to justify imposing a countervailing duty on the imported milled rice.

Asemesa’s position is that to be substantially dependent , “‘all or substantially all’ of the demand for the prior stage product must be driven by demand for the latter stage product.” Asemesa Br. 40. Asemesa’s position is essentially that section 1677–2 should be limited to cases in which the degree of dependence is identical to or more extreme than those in Pork from Canada or Rice from Thailand .

Asemesa is correct that those cases provided the incentive for Congress to add section 1677–2 to the Tariff Act. But there is no support for Asemesa’s further proposition that the meaning of “substantially dependent” in the statute requires that the demand for the prior stage product must be, at a minimum, as dependent on the demand for the latter stage products as it was in those two cases.

Asemesa’s position is contrary to the plain language of the statute. Had Congress intended the statute to track the facts of Pork from Canada and Rice from Thailand, it could have parroted the language of those decisions. Instead , Congress’s choice of “substantially dependent” captures the rationale of those decisions while setting a more flexible standard for Commerce to meet.

Senator Grassley’s comment that the statute “codified”

Pork from Canada and Rice from Thailand does not mean that the reach of the statute was confined to the facts of those cases. To begin with, “floor statements by individual legislators rank among the least illuminating forms of legislative history.” NLRB v. SW Gen., Inc., 580 U.S. 288, 307 (2017). But even if we were to assign substantial weight

ASOCIACIÓN DE EXPORTADORES E INDUSTRIALES v. US 11

tothose statements, it is implausible to assume that Senator Grassley used the term “codify” to suggest that section 1677–2 should be limited to the exact circumstances of those cases, when the plain text suggests otherwise. A more reasonable interpretation of Senator Grassley’s comments is that section 1677–2 was meant to create a statutory basis for Commerce to apply countervailing duty principles in cases such as Pork from Canada and Rice from Thailand, but not to confine the application of the statute to circumstances identical to, or more extreme than, in those cases.

2

The Trade Court interpreted section 1677–2, as applied to this case, to mean that the demand for raw olives would be substantially dependent on the demand for table olives only if table olives accounted for “at least half” of the market for raw olives from table and dual-use varietals. Asemesa I, 429 F. Supp. 3d at 1345. We do not agree with the Trade Court that the statute imposes a test requiring that at least 50 percent of the prior stage product be processed into the latter-stage product for section 1677–2 to apply.

The statutory term “substantially dependent” is general in nature, indicating that Congress intended to delegate the question of whether particular facts satisfy the statute’s requirements to Commerce. “Congress . . . may confer substantial discretion on executive agencies to implement and enforce the laws.” Gundy v. United States, 139 S. Ct. 2116, 2130 (2019). 4 By using nonspecific

4 Gundy addressed a challenge to agency rulemaking under the nondelegation doctrine, whereas this case concerns agency adjudication. Statutory interpretation, however , is key to nondelegation cases, Gundy, 139 S. Ct. at 2123 (“[A] nondelegation inquiry always begins . . . with 12 ASOCIACIÓN DE EXPORTADORES E INDUSTRIALES v. US

statutory language, Congress invokes its “ability to delegate power under broad general directives.” Id. Here, Congress ’s use of the term “substantially dependent,” as opposed to specifying a minimum percentage, reflects “an expression of its well-considered judgment as to the degree of administrative authority which it was necessary to grant.” Lichter v. United States, 334 U.S. 742, 784 (1948) (addressing a statute instructing agency to determine whether contracts resulted in “excessive profits,” but not specifying what qualified as “excessive”). 5 In United States v. Zenith Radio Corp., a leading case dealing with countervailing duties, our predecessor court adopted the same rationale. 562 F.2d 1209 (CCPA 1977), aff’d, 437 U.S 443 (1978). The court held that Congress’s use of the terms “bounty” and “grant,” which were “broad but not ambiguous,” demonstrated “Congress’[s] intent to provide a wide latitude, within which the Secretary of the Treasury . . . may determine the existence or non-existence of a bounty or a grant.” Id. at 1216 (crediting, in particular, Congress’s “refusal to define the words ‘bounty,’ ‘grant,’ or ‘net amount’”). The court added:

Not without reason has Congress refrained from spelling out either the precise criteria for determining what shall constitute a bounty or grant and what shall

statutory interpretation.”), a principle that applies whether the delegation is of rulemaking or adjudicative authority .

5 Justice Scalia made the point succinctly in his dissenting opinion in Mistretta v. United States, 488 U.S. 361, 417 (1989) where he wrote that “a certain degree of discretion , and thus of lawmaking, inheres in most executive or judicial action, and it is up to Congress, by the relative specificity or generality of its statutory commands, to determine —up to a point—how small or how large that degree shall be.”

ASOCIACIÓN DE EXPORTADORES E INDUSTRIALES v. US 13

not, or the calculations to be followed in determining net amount. . . . “In the assessment of a countervailing duty, the determination that a bounty or grant is paid necessarily involves judgments in the political, legislative or policy spheres.”

Id. at 1217 (quoting United States v. Hammond Lead Prods., Inc., 440 F.2d 1024, 1030 (CCPA 1971)).

Applying the same principle, we have held that similarly general language used in a related provision of the antidumping statute committed to Commerce’s discretion the question of whether particular facts satisfy the statute. See Nation Ford Chem. Co. v. United States, 166 F.3d 1373, 1377 (Fed. Cir. 1999) (“While § 1677b(c) provides guidelines to assist Commerce in this process, this section also accords Commerce wide discretion in the valuation of factors of production in the application of those guidelines.”); Magnesium Corp. of Am. v. United States, 166 F.3d 1364, 1372 (Fed. Cir. 1999) (holding that the “broad statutory mandate” gave Commerce “broad discretion”); accord Keller Trucking, Inc. v. United States, 567 F.2d 147, 149 (D.C. Cir. 1977) (interpreting an adjudicative determination as being “within the realm of the expertise and discretion of the [agency]” due to “the imprecise terms of the statute” at issue).

As with the broad statutory mandate at issue in Nation Ford and Magnesium Corp., Congress’s use of the term “substantially dependent” in section 1677–2 gives Commerce considerable discretion in determining whether particular facts meet that standard. Congress’s use of more general language indicates its understanding that assessing dependence, for purposes of section 1677–2, is a holistic determination. It further shows that Congress delegated the task of making that determination to Commerce , based on the circumstances of each case.

The government urges us to apply the Chevron doctrine in this case, see generally Chevron U.S.A., Inc. v. Nat. Res.

14 ASOCIACIÓN DE EXPORTADORES E INDUSTRIALES v. US

Def. Council, Inc., 467 U.S. 837, 844–45 (1984), and to defer to Commerce’s interpretation of section 1677–2. Because we regard the term “substantially dependent” as general but not ambiguous, we believe this case is more properly viewed as one involving implied delegation of adjudicative authority to the agency rather than deference to the agency’s interpretation of an ambiguous statute. 6

3

The relevant dictionary definitions of “substantial” are “[i]mportant, essential, and material; of real worth and importance ,” Black’s Law Dictionary 1728 (11th ed. 2019), and “something of moment: an important or material matter , thing, or part,” Webster’s Third New International Dictionary of the English Language 2280 (1998 ed.). Thus, the natural reading of the statutory text is that the demand for the prior stage product is “substantially dependent” on the demand for a latter stage product if the demand for the latter stage product has a real, material, or important effect on the demand for the prior stage product.

To be sure, the fact that a large percentage of a prior stage product is processed into a given latter stage product is strong evidence that the demand for the prior stage product substantially depends on the demand for the latter stage product. The Trade Court may be right that the fact that about 50 percent of the prior stage product was processed into the latter stage product is evidence of substantial dependence in this case, while 8 percent is not. Such a pure numerical test, however, is not what the statute calls

6 This case also does not involve the situation, separately discussed by the Court in Chevron, in which Congress has made “an express delegation of authority to the agency to elucidate a specific provision of the statute by regulation.” 467 U.S. at 843–44; see United States v. Mead Corp., 533 U.S. 218, 226–27 (2001).

ASOCIACIÓN DE EXPORTADORES E INDUSTRIALES v. US 15

for. The percentage of prior stage product processed into the latter stage product is just one factor in evaluating whether the demand for one product is “substantially dependent ” on the demand for another. The principal task under the statute—and one that Congress has assigned to Commerce by use of the broad term “substantially dependent ”—is to determine whether the demand for the latter stage product has a real, material, or important effect on the demand for the prior stage product.

Commerce adopted essentially that interpretation of the statute in its preliminary determination in this case. Ripe Olives from Spain, 82 Fed. Reg. 56218 (Dep’t of Commerce November 28, 2017). In the Nov. 21, 2017, Issues and Decision Memorandum (“Preliminary Memo”) accompanying that determination, Commerce explained that substantial dependence focuses on “the nature of the raw product and the market” rather than on “a specific minimum threshold.” Preliminary Memo at 16. As an example, Commerce cited a past determination in which it found the demand for fresh shrimp to be substantially dependent on the demand for frozen shrimp because “one quarter of the fresh shrimp market would collapse” if frozen shrimp did not exist. Id. (citing Shrimp from China, 78 Fed. Reg. 50391-01 (Dep’t of Commerce Aug. 19, 2013)).

Following the first remand from the Trade Court, Commerce complied with the Trade Court’s construction of section 1677–2 but expressed its continuing disagreement with that construction. Commerce reaffirmed the position it took in its Preliminary Memo, explaining that “if the demand for table olives were to cease, a sizeable sector of the raw olives market . . . would be negatively impacted.” J.A. 156. Although it ultimately applied the Trade Court’s interpretation of the statute, Commerce maintained that the term “substantially dependent” does not contemplate a numerical “minimum threshold of demand.” J.A. 157.

16 ASOCIACIÓN DE EXPORTADORES E INDUSTRIALES v. US

While we disagree with the Trade Court’s “at least half”

interpretation of section 1677–2, which Commerce applied under protest, our disagreement does not affect the outcome of this case. The Trade Court’s interpretation was more restrictive than Commerce’s more flexible interpretation, which we consider to be correct. Commerce found that the demand for raw olives was substantially dependent on the demand for table olives under both interpretations of the statute. Commerce’s findings therefore satisfy section 1677–2(1).

III

Aside from its statutory interpretation arguments, Asemesa raises three separate challenges to Commerce’s factual analysis. First, it argues that Commerce misconstrued the raw olive market by failing to credit evidence showing the extent of the use of table varietal olives for olive oil production and mill varietal olives for table olive production . Second, it argues that Commerce committed various analytical mistakes in calculating the 55.28 percent figure underlying Commerce’s “substantial dependence ” finding. Third, it argues that Commerce should have relied on the varietal-specific data from the AICA, rather than data from the Government of Spain, which did not include a varietal-by-varietal breakdown. None of Asemesa’s factual arguments renders Commerce’s findings “unsupported by substantial evidence” or the product of prejudicial error. Ta Chen Stainless Steel Pipe, Inc. v. United States, 298 F.3d 1330, 1335 (Fed. Cir. 2002).

A

Asemesa’s first factual argument relates to Commerce ’s definition of the market for the prior stage product. Commerce defined the prior stage product in this case as the “table and dual-use raw olive varietals that are biologically distinct from other raw olive varietals,” i.e., those varietals the Government of Spain considers fit for table olive production. J.A. 55. Commerce defined the latter stage

ASOCIACIÓN DE EXPORTADORES E INDUSTRIALES v. US 17

product as table olives. Id. The main olive varietals that satisfy Commerce’s total market definition are manzanilla, gordal, carrasqueña, hojiblanca, and cacereña olives, which accounted for 95% of the entire table olive production during the 2015 to 2016 investigation period. J.A. 56. The remaining 5% are “other dual-use varietals.”

Commerce’s characterization of the market assumes that nearly all the pure table olive varietals (manzanilla, gordal, and carrasqueña) are processed into table olives, and that effectively all olives from the pure “mill olive” varietals are processed into olive oil. But the record contains at least anecdotal evidence that some mill olives were processed into table olives and that some olives grown for sale as table olives were used to make olive oil. J.A. 11721–22, 11241.

Asemesa’s evidence does not “repudiate” Commerce’s characterization of the market, as Asemesa argues. Asemesa Br. 47. The fact that some olives from the mill varietals were processed for table use is not inconsistent with Commerce’s characterization. Without evidence about how much cross-use existed between pure table and pure mill varietals, it was not unreasonable for Commerce to assume that such cross-use was negligible. Similarly, although the Spanish government’s data showed that some olives grown for processing into oil were ultimately processed for table use and vice versa, Commerce reasonably assumed that such cross use was attributable to dual-use varietals. It is plausible that olives from dual-use varietals cultivated to produce mill olives could be repurposed into table olives, but that those olives from pure mill varietals ordinarily could not. The fact that the Government of Spain categorizes olive varietals as mill, table, and dual use is itself evidence that the Spanish olive market is divided accordingly.

18 ASOCIACIÓN DE EXPORTADORES E INDUSTRIALES v. US

B

Asemesa’s second factual argument relates to Commerce ’s calculations. Commerce calculated the percentage of olives from table and dual-use varietals that are processed into table olives in what can be characterized as an exercise in estimation based on limited available data.

Asemesa challenges two aspects of Commerce’s calculation . First, Asemesa argues that Commerce improperly counted as table olives those hojiblanca varietal olives that are grown for mill but are sold as table olives. Second, Asemesa challenges Commerce’s treatment of cacereña and “other” dual-use varietal olives, arguing that Commerce should not have excluded those varietals from its analysis, and in any event that Commerce did not implement that exclusion correctly. Neither of those challenges warrants a remand.

1

Asemesa first argues that Commerce incorrectly counted 71,814 tons of hojiblanca olives as table olives, even though they were grown for processing into oil. Commerce counted them as it did because they were ultimately processed into table olives. Asemesa’s argument is that the farmers’ intentions are what matter, not how the olives are ultimately used. Accordingly, Asemesa argues that Commerce should have counted those 71,814 tons as mill olives.

Commerce found the ultimate use to which the olives were put to be the most probative indicator of demand in particular segments of the olive industry. Asemesa has not pointed to any reason to believe that the original intentions of Spanish olive farmers would provide a better measure of demand. We therefore conclude that Commerce was not wrong to treat the relevant inquiry as focusing on what percentage of olives from suitable varietals were ultimately processed into table olives.

Case: 23-1162 Document: 52 Page: 19 Filed: 05/20/2024

ASOCIACIÓN DE EXPORTADORES E INDUSTRIALES v. US 19

2

Asemesa’s challenge to Commerce’s treatment of cacereña olives is more complicated and requires more explanation . We ultimately conclude that Commerce’s calculations were flawed, but not in a way that prejudiced Asemesa.

Commerce’s analysis focused on the percentage of raw olives from table or dual-use varietals that depend on the market for table olives. That percentage is equal to the volume of table olives derived from the relevant varietals divided by the total volume of olives from those varietals, which is shown by the expression below:

𝑚𝑚𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡 + 𝑔𝑔𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡 + 𝑞𝑞𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡 + ℎ𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡 + 𝑐𝑐𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡 + 𝑜𝑜𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡 𝑚𝑚 + 𝑔𝑔 + 𝑞𝑞 + ℎ + 𝑐𝑐 + 𝑜𝑜

The letters m, g, q, h, c, and o in that expression stand for the volumes of manzanillas, gordales, carrasqueñas (“q”), hojiblancas, cacereñas (“c”), and “other” dual-use olives, respectively . 7 The letters with “table” subscripts represent the amounts of those varietals that were used as table olives .

The Spanish government publishes data on the aggregate volume of olives grown for the purpose of producing table olives. It also publishes data on the aggregate volume of olives that are ultimately used as table olives across all varietals. J.A. 11241. 8 Using these aggregate values

7 All “volumes” in this case are measured in tons.

Although “ton” is a unit of mass, “volume” is typically used to describe the amount of an agricultural product, even though the product may be measured by weight.

8 The total volume of table olives, 𝑇𝑇𝑢𝑢𝑠𝑠𝑠𝑠𝑠𝑠 𝑎𝑎𝑎𝑎 𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡 , is the sum of two published values: (1) the olives grown for table and processed into table olives, and (2) the olives grown for

Case: 23-1162 Document: 52 Page: 20 Filed: 05/20/2024

20 ASOCIACIÓN DE EXPORTADORES E INDUSTRIALES v. US

instead of individual varietal volumes and assuming that no pure table varietal olives were grown for mill, Commerce ’s expression can be simplified to:

𝑇𝑇𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢 𝑎𝑎𝑎𝑎 𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡 𝑇𝑇𝑔𝑔𝑔𝑔𝑔𝑔𝑔𝑔𝑔𝑔 𝑓𝑓𝑓𝑓𝑓𝑓 𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡 + ℎ𝑔𝑔𝑔𝑔𝑔𝑔𝑔𝑔𝑔𝑔 𝑓𝑓𝑓𝑓𝑓𝑓 𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚 + 𝑐𝑐𝑔𝑔𝑔𝑔𝑔𝑔𝑔𝑔𝑔𝑔 𝑓𝑓𝑓𝑓𝑓𝑓 𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚 + 𝑜𝑜𝑔𝑔𝑔𝑔𝑔𝑔𝑔𝑔𝑔𝑔 𝑓𝑓𝑓𝑓𝑓𝑓 𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚

where 𝑇𝑇 represents the total volume across all varietals.

Commerce, however, lacked varietal-by-varietal data for the volumes of hojiblanca, cacereña and other dual-use olives grown for mill. Commerce had data on the total production volume and acreage of hojiblancas from which it could estimate the volume of hojiblancas grown for mill, but it lacked corresponding data for both cacereña and the “other” category of dual-use varietal olives. Commerce therefore sought to omit cacereña and other dual-use varietals from its calculation. J.A. 58. Modified by those omissions , Commerce’s revised expression was:

𝑇𝑇′𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢 𝑎𝑎𝑎𝑎 𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡 ′ 𝑇𝑇𝑔𝑔𝑔𝑔𝑔𝑔𝑔𝑔𝑔𝑔 𝑓𝑓𝑓𝑓𝑓𝑓 𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡 + ℎ𝑔𝑔𝑔𝑔𝑔𝑔𝑔𝑔𝑛𝑛 𝑓𝑓𝑓𝑓𝑓𝑓 𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚

where 𝑇𝑇′ denotes the total volume of raw olives from relevant varietals processed as table olives, excluding cacereña and “other” dual-use varietal olives. Put differently, 𝑇𝑇′ is the volume of manzanilla, gordal, carrasqueña and hojiblanca olives. Based on various assumptions, Commerce calculated that the volume of 𝑇𝑇′ processed into table olives

mill but processed into table olives. In 2016, the relevant harvest year, those numbers were 492,244 and 90,404 tons respectively. 𝑇𝑇𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢 𝑎𝑎𝑎𝑎 𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡 therefore equals 582,648 tons.

Commerce assumed that effectively all olives grown for table were from the varietals the Government of Spain considers suitable for table olive production, and that effectively all olives grown for mill but processed into table olives were from dual-use varietals.

ASOCIACIÓN DE EXPORTADORES E INDUSTRIALES v. US 21

was 564,058 tons. 9 The Spanish government reports that the total volume of olives grown for table in 2016 was 511,122 tons. J.A. 11241. Lastly, Commerce estimated the volume of hojiblancas grown for mill to be 509,304 tons based on other available data regarding the yield rate and acreage of hojiblancas dedicated to each use. 10

9 The Government of Spain reports the total volume of olives sold as table olives, but that number includes cacereña and other dual-use varietal olives that Commerce intended to exclude from its analysis. Therefore, Commerce had to estimate the volume of cacereña and other dual-use varietal olives to subtract from the numerator. The Spanish government’s data reports that the total volume of dual-use varietal olives grown for mill but used for table in 2016 was 90,404 tons. J.A. 11241. The AICA data reports the varietal-by-varietal breakdown of dual-use varietal olives grown for table use, J.A. 11643, which can be converted to a percentage breakdown of those varietals: 79.44% hojiblanca, 12.41% cacereña, and 8.16% other. By assuming that the same varietal breakdown applied to dual-use varietal olives grown for mill but used for table, Commerce calculated that 18,590 of the 90,404 tons of dual-use varietal olives grown for mill but used for table were cacereña or other dual-use varietal olives and that the remaining 71,814 tons were hojiblancas. Commerce therefore found that the volume of 𝑇𝑇′ processed into table olives is 𝑇𝑇𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢 𝑎𝑎𝑎𝑎 𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡 minus 18,590, or 582,648 minus 18,590, which equals 564,058 tons.

10 Commerce had data on the total production and total hectares in cultivation for both table olives and mill olives . From the data, Commerce calculated industry average yield rates for both olives grown for table and olives grown for mill, which it assumed to be representative of the same yield rates for hojiblancas. For olives grown 22 ASOCIACIÓN DE EXPORTADORES E INDUSTRIALES v. US

Beginning with the above expression and substituting values yields the percentage that Commerce found to satisfy the “substantially dependent” requirement of section 1677–2:

564,058

= 55.28%

511,122 + 509,304

Asemesa takes issue with Commerce’s 55.28 percent figure on two grounds. First, Asemesa argues that Commerce did not properly exclude cacereña and “other” dual- use olives from the numerator of the expression because the Spanish government’s estimate of the total table olives, on which Commerce based its numerator, included olives from those varietals. Commerce did not exclude olives from those varietals grown for table. Second, Asemesa argues that Commerce could have and should have included cacereña and “other” dual-use varietal olives in its analysis. Commerce’s decision not to do so skewed the results in Commerce’s favor. Asemesa is correct on both issues; however , neither issue makes a material difference to the outcome of this case.

for table, Commerce calculated that 511,122 tons divided by 160,400 hectares equaled 3.19 tons per hectare; and for olives grown for mill, Commerce calculated that 6,571,428 tons divided by 2,243,700 hectares equaled 2.93 tons per hectare. J.A. 11892 (relying on the Spanish government’s data). Dividing the total volume of hojiblancas as reported in the AICA data by the yield rate for table olives, Commerce found that there would have needed to be 91,176 acres of hojiblancas dedicated to table olive production to achieve that volume. Given that there was a total of 265,000 hectares of hojiblancas in cultivation, the remaining 173,824 hectares were dedicated to mill olive production . And at the calculated yield rate of 2.93 tons per hectare, those acres would yield 509,304 tons of olives.

ASOCIACIÓN DE EXPORTADORES E INDUSTRIALES v. US 23

Although Commerce removed cacereña and other dual-

use varietal olives grown for mill from its analysis, see supra , note 9, Commerce neglected to remove cacereña and other dual-use varietal olives grown for table. The Spanish government’s data on total table olives considers olives to be “table olives” if they were grown with that intention. J.A. 10704. The table olive figures Commerce relied on, represented by 𝑇𝑇 or 𝑇𝑇′ in the above expressions, therefore include cacereña and other dual-use varietal olives grown for table. Commerce did not make any adjustment to remove cacereña and other dual-use olives grown for table from the numerator of its expression.

What Commerce should have done, instead, is to use the Spanish government’s raw data for the numerator and estimate the additional volume of cacereñas grown for mill that must be included in the denominator. Doing so would have been a matter of arithmetic because Commerce had already assumed that all olives from table olive varietals are processed into table olives and that different dual-use varietals are processed into table olives and olive oil at the same rate. Including cacereña and other dual-use olives in its analysis would have required no new assumptions or factfinding and would have captured the entire market as Commerce defined it. Our analysis uses Commerce’s data and assumptions and corrects its arithmetic.

Commerce had already calculated the varietal breakdown of dual-use varietal olives. See supra, note 9. It also had already assumed that dual-use varietal volume is proportionately allocated between table and mill on a varietal- by-varietal basis. J.A. 59–60. Applying that proportionality assumption to the 509,304 tons of hojiblancas grown for mill would yield the following expression:

79.44% 20.57% =

509,304 𝑐𝑐𝑔𝑔𝑔𝑔𝑔𝑔𝑔𝑔𝑔𝑔 𝑓𝑓𝑓𝑓𝑓𝑓 𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚 + 𝑜𝑜𝑔𝑔𝑔𝑔𝑔𝑔𝑔𝑔𝑔𝑔 𝑓𝑓𝑓𝑓𝑓𝑓 𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚

Case: 23-1162 Document: 52 Page: 24 Filed: 05/20/2024

24 ASOCIACIÓN DE EXPORTADORES E INDUSTRIALES v. US

That expression can be solved for the volume of cacereña and other dual-use olives grown for mill, which is 131,877 tons.

Commerce’s expression without the simplifying assumption excluding cacereña and “other” dual-use varietal olives was:

𝑇𝑇𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢 𝑎𝑎𝑎𝑎 𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡 𝑇𝑇𝑔𝑔𝑔𝑔𝑔𝑔𝑔𝑔𝑔𝑔 𝑓𝑓𝑓𝑓𝑓𝑓 𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡 + ℎ𝑔𝑔𝑔𝑔𝑔𝑔𝑔𝑔𝑔𝑔 𝑓𝑓𝑓𝑓𝑓𝑓 𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚 + 𝑐𝑐𝑔𝑔𝑔𝑔𝑔𝑔𝑔𝑔𝑔𝑔 𝑓𝑓𝑓𝑓𝑓𝑓 𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚 + 𝑜𝑜𝑔𝑔𝑔𝑔𝑔𝑔𝑔𝑔𝑔𝑔 𝑓𝑓𝑓𝑓𝑓𝑓 𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚

The total volumes of table olives reported by the Government of Spain, 𝑇𝑇 in the expression above, already include cacereña and “other” dual-use varietal olives. Substituting values and simplifying the above expression yields:

564,058

= 48.95%

511,122 + 509,304 + 131,877 Although Commerce erred in its treatment of cacereña and “other” dual-use varietal olives, the error did not have a significant effect on the percentage calculation. Either way, roughly half of all olives from the relevant varietals are ultimately processed into table olives. Commerce’s finding, that such a high percentage indicates that the demand for raw olives substantially depends on the demand for table olives, remains valid after correcting for this minor calculation error. Because, contrary to the Trade Court, we have construed the statute as not requiring “at least half” of the demand for raw olives to depend on demand for table olives, any error Commerce made by excluding cacereña and “other” dual-use varietal olives did not prejudice Asemesa and does not warrant a remand for further proceedings. See 5 U.S.C. § 706 (“[D]ue account shall be taken of the rule of prejudicial error.”); Shinseki v. Sanders , 556 U.S. 396, 406 (2009) (characterizing section 706 of

ASOCIACIÓN DE EXPORTADORES E INDUSTRIALES v. US 25

the Administrative Procedure Act as an “administrative law harmless error rule.”) (cleaned up). 11

C

Asemesa’s third factual argument is that Commerce should have relied on the AICA data rather than the Spanish government’s data, because only the AICA separated its findings by varietal. Commerce did rely on the AICA data for certain purposes, such as to calculate the portion of the market attributable to different varietals. See supra, note 9; J.A. 11892. Commerce chose to use the Spanish government’s data over the AICA data for some applications because the Spanish government’s analysis focused on how olives are used—not how olives are grown.

Even if we were to agree that Commerce should have relied on the AICA data in place of the Spanish government ’s data, Asemesa has not identified how doing so would have changed the result. In particular, Asemesa has not stated what the percentage of raw olives from the relevant varietals that are processed into table olives would have been if Commerce had credited the AICA data. Commerce chose to rely on the Spanish government’s data instead of the AICA data for certain purposes, and that choice was a reasonable one. This court will not “reweigh” the evidence when Commerce makes a rational decision regarding which set of data to credit. Downhole Pipe &

11 Affirmance in this case does not run afoul of the rule in Securities & Exchange Commission v. Chenery Corp., 318 U.S. 80, 94 (1943), because it is clear that the agency would have reached the same result in this case absent the calculation errors we have identified. See Mass. Trs. of E. Gas & Fuel Assocs. v. United States, 377 U.S. 235, 248 (1964); Oracle Am., Inc. v. United States, 975 F.3d 1279, 1290 (Fed. Cir. 2020); In re Watts, 354 F.3d 1362, 1370 (Fed. Cir. 2004).

26 ASOCIACIÓN DE EXPORTADORES E INDUSTRIALES v. US

Equip., L.P. v. United States, 776 F.3d 1369, 1376 (Fed. Cir. 2019).

*****

Because Commerce’s findings satisfy the statutory requirements of section 1677–2 and are supported by substantial evidence, we sustain the Trade Court’s decision.

AFFIRMED