Apex Frozen Foods Private Ltd. v. United States

862 F.3d 1337, 2017 WL 2960725, 38 I.T.R.D. (BNA) 2367, 2017 U.S. App. LEXIS 12400

Opinion

CLEVENGER, Circuit Judge.

Plaintiffs appeal the decision of the Court of International Trade (“CIT”) affirming the U.S. Department of Commerce’s (“Commerce”) final results in the eighth administrative review of the anti-dumping duty order on certain frozen warmwater shrimp from India. Apex Frozen Foods Private Ltd. v. United States, 144 F.Supp.3d 1308 (Ct. Int’l Trade 2016); see also Certain Frozen Warmwater Shrimp from India, 79 Fed. Reg. 51,309 (Dep’t Commerce Aug. 28, 2014) (final administrative review). Using the “average-to-transaction” methodology with zeroing, Commerce assessed mandatory respondent Devi Fisheries Limited (“Devi”) with a 1.97 percent duty for entries between February 1, 2012, and January 31, 2012. Using a “mixed alternative” methodology, which blends both the average-to-transac-. tion and average-to-average methodologies, Commerce assessed the second mandatory respondent Falcon Marine Exports Limited/K.R. Enterprises (“Falcon”) with a 3.01 percent duty for the same time period. Non-mandatory respondents (including Apex Frozen Foods Private Limited (“Apex”)) were assessed with a simple-averaged antidumping duty of 2.49 percent.

Plaintiffs include Apex, Devi, Falcon, and other exporters subject to Commerce’s antidumping duties on frozen warmwater shrimp from India (collectively, “Apex”). Apex challenges the methodology used by Commerce to calculate the antidumping duties on a number of grounds related to Commerce’s decision to use the average-to-transaction methodology and zeroing. For the reasons that follow, we affirm the CIT’s decision and sustain Commerce’s results.

Background

I

“Dumping,” in international trade parlance, is a practice where international exporters sell goods to the United States at prices lower than they are sold in their home markets, in order to undercut U.S. domestic sellers and carve out market share. To protect domestic industries from goods sold at less than “fair value,” Congress enacted a statute allowing Commerce to assess remedial “antidumping duties” on foreign exports. 19 U.S.C. § 1673; see also Viet I-Mei Frozen Foods Co. v. United States, 839 F.3d 1099, 1101 (Fed. Cir. 2016) (“The antidumping statute provides for the assessment of remedial duties on foreign merchandise sold in the United States at less than fair market value that materially injures or threatens to injure a domestic industry.”).

“Sales at less than fair value are those sales for which the ‘normal value’ (the price a producer charges in its home market) exceeds the ‘export price’ (the price of the product in the United States).... ” Union Steel v. United States, 713 F.3d 1101, 1103 (Fed. Cir. 2013). Commerce performs this pricing comparison, and the concomitant antidumping duty calculation, using one of three methodologies:

(1) Average-to-transaction [“A-T”], in which Commerce compares the weighted average of the normal values to the export prices (or constructed export prices) of individual transactions.

(2) Average-to-average [“A-A”], in which Commerce compares the weighted average of the normal values to the weighted average of the export prices (or constructed export prices).

(3) Transaction-to-transaction [“T-T”], in which Commerce compares the normal value of an individual transaction to the export price (or constructed export price) of an individual transaction.

Id. (citation omitted).

Previously, Commerce’s general practice was to use the A-T methodology for both investigations and administrative reviews. Id. at 1104. With the adoption of the Uruguay Rounds Agreement Act in 1995, Congress required that the A-A or T-T methods be the presumed defaults for investigations, with the A-T method only to be used in certain circumstances. Id.; see also 19 U.S.C. § 1677f-1(d)(1). Yet “Commerce continued to use average-to-transaction comparisons as its general practice in administrative reviews,” in the absence of any governing statutory authority. Union Steel, 713 F.3d at 1104. Over time, Commerce unified its procedures through regulation, stating, “[i]n an investigation or review, the Secretary will use the average-to-average method unless the Secretary determines another method is appropriate in a particular case,” 19 C.F.R. § 351.414(c)(1) (2012), and began applying the investigations statutory framework to guide its administrative reviews as well.

The investigations statute provides that, in general, antidumping duties are to be calculated using the A-A method — “comparing the weighted average of the normal values to the weighted average of the export prices (and constructed export prices) for comparable merchandise.”¹ 19 U.S.C. § 1677f-l(d)(l)(A)(i). The statute, however, contemplates an exception to this general rule:

The administering authority may determine whether the subject merchandise is being sold in the United States at less than fair value by comparing the weighted average of the normal values to the export prices (or constructed export prices) of individual transactions for comparable merchandise, if—

(i) there is a pattern of export prices (or constructed export prices) for comparable merchandise that differ significantly among purchasers, regions, or periods of time, and

(ii) the administering authority explains why such differences cannot be taken into account using a method described in paragraph (l)(A)(i) or (ii).

19 U.S.C. § 1677f-l(d)(l)(B). In other words, the A-T method can be used, provided two preconditions are met: (1) a pattern of significant price differences, and (2) an inability of the A-A method to “account” for these differences.

The statutory exception exists to address “targeted” or “masked” dumping. Union Steel, 713 F.3d at 1104 n.3. Under the A-A methodology, sales of low-priced “dumped” merchandise would be averaged with (and offset by) sales of higher-priced “masking” merchandise, giving the impression that no dumping was taking place and frustrating the antidumping statute’s purpose. See Koyo Seiko Co. v. United States, 20 F.3d 1156, 1159 (Fed. Cir. 1994). The AT method addresses this concern because, “[b]y using individual U.S. prices in calculating dumping margins, Commerce is able to identify a merchant who dumps the product intermittently — sometimes selling below the foreign market value and sometimes selling above it.” Id. The driving rationale behind the statutory exception is that targeted dumping is more likely to be occurring where there is a “pattern of export prices ... for comparable merchandise that differ significantly among purchasers, regions, or periods of time.” See 19 U.S.C. § 1677f-1(d)(1)(B); Union Steel, 713 F.3d at 1104 n.3; see also H.R. Rep. No. 103-826, pt. 1, at 99 (1994) (“[The exception] provides for a comparison of average normal values to individual export prices ... in situations where an average-to-average ... methodology cannot account for a pattern of prices that differ significantly among purchasers, regions, or time periods, ie., where targeted dumping may be occurring”).

Commerce also devised the practice of “zeroing” when compiling a weighted average dumping margin — “where negative dumping margins (i.e., margins of sales of merchandise sold at nondumped prices) are given a value of zero and only positive dumping margins (i.e., margins for sales of merchandise sold at dumped prices) are aggregated.” Union Steel, 713 F.3d at 1104. Commerce has discontinued its use of zeroing when applying the A-A methodology, but zeroing remains part of Commerce’s calculus when compiling a weighted average dumping margin under the A-T methodology. Id. at 1104-05, 1109 (“Commerce’s decision to use or not use the zeroing methodology reasonably reflects unique goals in differing comparison methodologies .... When examining individual export transactions, using the average-to-transaction comparison methodology, prices are not averaged and zeroing reveals masked dumping.”); see also U.S. Steel Corp. v. United States, 621 F.3d 1351, 1363 (Fed. Cir. 2010).

II

Commerce initiated the eighth administrative review of its antidumping duty covering frozen warmwater shrimp from India (“AR8”) in April 2013 — the review period covered entries of merchandise that occurred between February 1, 2012, and January 31, 2013. Commerce selected Devi and Falcon as mandatory respondents.

Commerce published the final results of AR8 in Au-gust 2014, along with an Issues and Decision Memorandum explaining its methodology and results. By regulation, Commerce typically “use[s] the A-A method unless the Secretary determines another method appropriate in a particular case.” 19 C.F.R. § 351.414(c)(1). Commerce noted that, despite the statutory silence regarding administrative reviews, the “analysis that has been used in [less-than-fair-value] investigations [is] instructive for purposes of examining whether to apply an alternative comparison in this administrative review.” Joint Appendix at 1395. As such, following 19 U.S.C. § 1677f-1(d)(1)(B), Commerce considered (1) whether Devi’s and Falcon’s sales exhibited a pattern of significant price differences among purchasers, regions, or periods of time; and (2) whether “such differences can be taken into account using” the A-A method.

Commerce applied a “differential pricing” analysis² to determine if there was a pattern of significant price differences between Devi’s and Falcon’s purchasers, regions, or periods of time.³ Commerce found that 73:3 percent of Devi’s sales passed the Cohen’s d test (more than 66 percent), therefore theoretically warranting the use of the A-T methodology on all of Devi’s sales. In contrast, Commerce found 65.31 percent of Falcon’s sales passed the Cohen’s d test (between 33 and 66 percent), therefore theoretically warranting the use of the mixed alternative: the A-T methodology for only those sales passing the Cohen’s d test, with the A-A methodology being applied to the non-passing sales.

Following the statute, Commerce also determined that the A-A methodology could not “account” for the patterns of price differences in either Falcon’s or Devi’s sales because “the difference[s] in the weighted-average dumping margins computed using the A-to-A method and the appropriate alternative method [were] meaningful.” Joint Appendix at 1389 (footnote omitted); see also id. at 1439 (“In considering this question, the Department tests whether using an alternative method .... yields a meaningful difference in the weighted-average dumping margin as compared to that resulting from the use of the [A-A] method only.”). Specifically, Corn-merce determined that the ultimate margins for Devi and Falcon were zero using the A-A methodology, whereas the margins were 1.97 percent and 3.01 percent, respectively, using the alternative methodologies. Commerce therefore adopted its preliminary findings that, because the calculated margins for both Devi and Falcon “move[d] across the de minimis threshold when calculated using the [A-A] method and an alternative method,” use of the respective alternative methods for each was justified. Id. at 1439.

Consequently, Commerce assessed Devi with a 1.97 percent antidumping duty, calculated using the A-T methodology for all sales; Commerce assessed Falcon with a 3.01 percent antidumping duty, calculated using the mixed methodology, with the AT method applied to sales passing the Cohen’s d test, and the A-A method applied to the remainder. Exporters not selected for individual' review were assigned the simple average of the two rates: 2.49 percent.

Apex filed suit at the CIT, challenging Commerce’s final results. On February 2, 2016, the CIT rejected Apex’s claims and sustained the results of AR8 in full. Apex Frozen Foods, 144 F.Supp.3d 1308. Apex appeals the CIT’s decision to this court. Apex contends that Commerce failed to justify sufficiently its conclusion that the A-A methodology could not “account” for the observed patterns of price differences. Apex also objects to Commerce’s anti-dumping margin calculation for the “mixed” alternative methodology, which was applied to Falcon’s sales.

We have jurisdiction under 28 U.S.C § 1295(a)(5).

Standard of Review

We review Commerce’s actions using the same standard applied by the CIT. Dongtai Peak Honey Indus. Co. v. United States, 777 F.3d 1343, 1349 (Fed. Cir. 2015). As such, we will sustain the agency’s decisions 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). Notwithstanding the CIT’s “unique and specialized expertise in trade law,” we review its decision de novo. Union Steel, 713 F.3d at 1106; see also Novosteel SA v. United States, 284 F.3d 1261, 1269 (Fed. Cir. 2002) (“[W]e also give due respect to the informed opinion of the [CIT].” (internal quotation marks omitted)).

Our review of an agency’s interpretation and implementation of a statutory scheme is governed by the Supreme Court’s holding in Chevron, U.S.A., Inc. v. Natural Resources Defense Council, Inc., 467 U.S. 837, 104 S.Ct. 2778, 81 L.Ed.2d 694 (1984). Under Chevron’s two-part framework, we first ask “whether Congress has directly spoken to the precise question at issue.” Id. at 842, 104 S.Ct. 2778. If yes, “that is the end of the matter,” and we “must give effect to the unambiguously expressed intent of Congress.” Id. at 842-43, 104 S.Ct. 2778. But, “if the statute is silent or ambiguous with respect to the specific issue, the question for the court is whether the agency’s answer is based on a permissible construction of the statute.” Id. at 843, 104 S.Ct. 2778; see also Koyo Seiko, 36 F.3d at 1573 (“In a situation where Congress has not provided clear guidance on an issue, Chevron requires us to defer to the agency’s interpretation of its own statute as long as that interpretation is reasonable.”).

Discussion •

Apex contends that Commerce unlawfully applied the A-T methodology because it failed to explain adequately why the price differences identified by the Cohen’s d test could not be “taken into account” using the A-A methodology, as required by statute. See 19 U.S.C. § 1677f-l(d)(l)(B)(ii). Additionally, assuming it was proper to use the mixed alternative methodology for Falcon’s sales, Apex objects to Commerce’s ultimate antidumping duty calculation under this approach. We address Apex’s arguments in turn.

I

Apex does not challenge the results of Commerce’s application of the Cohen’s d test — sales that illustrate a pattern of significant price differences and that therefore may be evidence of targeting or masked dumping. Rather, Apex contends .that Commerce failed to adhere to the statute’s requirement that “the administering authority explains why such differences cannot be taken into account using” the A-A methodology. 19 U.S.C. § 1677f-l(d)(l)(B)(ii).

As noted above, Commerce’s justification for why the A-A methodology was unable to account for the price differences was based on its “meaningful difference” test, which simply compared the ultimate antidumping duties that would be applied under the A-A methodology versus the alternative methodologies — the pure A-T methodology for Devi’s sales, and the mixed methodology for Falcon’s sales. Because the margins for both Devi and Falcon “move[d] across the de minimis threshold” — going from below 0.5 percent with the A-A methodology to above 0.5 percent with the alternative methodologies — Commerce concluded that there was a meaningful difference between the rates and that using an alternative methodology was warranted.⁴ Joint Appendix at 1439.

Apex takes issue with several aspects of Commerce’s meaningful difference test as a mechanism for satisfying the statute.

A

First, Apex challenges Commerce’s use of all sales when conducting its meaningful difference analysis for Devi and Falcon, instead of only those sales found to have passed the Cohen’s d test. Apex argues that including all sales is in direct contravention of the statute, which says Commerce must explain “why such differences cannot be taken into account using” the AA methodology. See 19 U.S.C. § 1677f-1(d)(1)(B)(ii) (emphasis added). According to Apex, “such differences” refers to the prior subsection’s reference to a “pattern of export prices ... that differ significantly among purchasers, regions, or periods of time,” i.e., targeted sales. § 1677d-l(d)(l)(B)(i). Apex argues that applying the Cohen’s d test yields “two pools of sales, one pool of all targeted sales and another pool of all non-targeted sales. The [meaningful difference] test which follows must then be conducted on ‘such differences,’ which in this case are differences related to the targeted sales.” Apex Opening Brief at 35. Apex reasons that, by using the entirety of Devi’s and Falcon’s sales in the meaningful difference analy-ses, Commerce ran afoul of Congress’s statutory directive and that we are obligated, under Chevron step one, to reverse. See Chevron, 467 U.S. at 842-43, 104 S.Ct. 2778 (“If the intent of Congress is clear, that is the end of the matter; for the court, as well as the agency, must give effect to the unambiguously expressed intent of Congress.”).

We disagree that the statutory language that Apex re-lies on decides the “precise question at issue.” See id. at 842, 104 S.Ct. 2778 (emphasis added). Under a plain reading of the statute, the use of “such differences” does not, in itself, manifest Congress’s intent to dictate how Commerce is to make the determination whether the A-A methodology can account for potential targeted or masked dumping. See id. at 843 n.9, 104 S.Ct. 2778 (explaining that courts are to use “traditional tools of statutory construction” to determine whether “Congress had an intention on the precise question at issue”). Chevron step one asks if Congress has already spoken unambiguously on the course of conduct the agency is to follow — we are not convinced Congress has expressed any intent whatsoever as to the matter at hand. Therefore, we reject Apex’s argument that this issue may be resolved as a matter of Chevron step one.⁵

“[I]f the. statute is silent or ambiguous with respect to the specific issue, the question for the court is whether the agency’s answer is based on a permissible construction of the statute.” Id. at 843, 104 S.Ct. 2778. An agency’s reasonable interpretation “is ‘given controlling weight unless [it is] arbitrary, capricious, or manifestly contrary to statute.’ ” PSC VSMPO-Avisma Corp. v. United States, 688 F.3d 751, 763-64 (Fed. Cir. 2012) (alteration in original) (quoting Chevron, 467 U.S. at 843-44, 104 S.Ct. 2778). Apex maintains that, even if Congress has not expressly spoken to the question before us, Commerce’s implementation of the statutory scheme is arbitrary, capricious, and clearly unreasonable and should be set aside. See Changzhou Wujin Fine Chem. Factory Co. v. United States, 701 F.3d 1367, 1374 (Fed. Cir. 2012). We disagree.

By statute, Commerce must explain why an observed pattern of price differences “cannot be taken into account using” the A-A methodology. 19 U.S.C. § 1677f-l(d)(l)(B)(ii). As already established, the statute is silent on how Commerce is to perform this analysis or even what it means for the A-A methodology to take “account” of price differences. Faced with a broad delegation of authority, Commerce devised its meaningful difference test, in which antidumping rates — as they would ultimately be applied for the A-A methodology versus an alternative — are compared, across all sales.

We find Commerce’s provided rationales in support of its meaningful difference analysis to be reasonable. First, we agree that the difference in the actual antidumping rates that would be assessed — below de minimis when calculated with the A-A methodology; above de minimis when calculated with an alternative methodology — indeed informs the question of whether the A-A methodology can adequately account for a pattern of significant price differences “because A-A masked the dumping that was occurring as revealed by the A-T calculated margin.” See Apex Frozen Foods, 144 F.Supp.3d at 1333 n.24; see also id. at 1334 (“It is reasonable for Commerce to judge whether AA is able to account for the price differences by assessing its ability to do so against all sales, as it would ultimately need to be able to do so when calculating the dumping margin.”).

Second, Commerce explained its view that considering all sales is actually necessary to achieve the overall aim of § 1677f-1(d)(1)(B), which is to address masked dumping. Specifically, Commerce stated in its final Issues and Decision Memorandum:

Higher-priced sales and lower-priced sales do not operate independently; all sales are relevant to the analysis. Higher- or lower-priced sales could be dumped or could be masking other dumped sales — this is immaterial in the Cohen’s d test and the question of whether there is a pattern of prices that differ significantly, because this analysis includes no comparisons with [normal values]. By considering all sales, both higher-priced and lower-priced, the Department is able to analyze an exporter’s pricing behavior and to identify whether there is a pattern of prices that differ significantly.... Where the evidence indicates that the exporter is engaged in a pricing behavior which creates a pattern, there is cause to continue with the analysis to determine whether masked dumping is occurring.

Joint Appendix at 1412. We understand Apex to be challenging Commerce’s position on this point, but we cannot say that the methodology Commerce has chosen to implement Congress’s statutory scheme is unreasonable, even where its justification may be, as the CIT found, “less than ideal.” See Apex Frozen Foods, 144 F.Supp.3d at 1333 n.24; see also PSC VSMPO-Avisma, 688 F.3d at 764 (“This court has recognized that the antidumping statute reveals tremendous deference to the expertise of the Secretary of Commerce in administering the antidumping law. Antidumping and countervailing duty determinations involve complex economic and accounting decisions of a technical nature, for which agencies possess far greater expertise than courts.” (quoting Fujitsu Gen. Ltd. v. United States, 88 F.3d 1034, 1039 (Fed. Cir. 1996))); cf. FCC v. Fox Television Stations, Inc., 556 U.S. 502, 513-14, 129 S.Ct. 1800, 173 L.Ed.2d 738 (2009) (“[A] court is not to substitute its judgment for that of the agency, and should uphold a decision of less than ideal clarity if the agency’s path may reasonably be discerned.” (internal quotation marks and citations omitted)).

Apex, however, raises two specific counterarguments, as to why Commerce’s implementation of the statute is unreasonable. According to Apex, the statute contemplates a “two-stage process”: Commerce only needs to consider the entirety of an exporter’s sales when ascertaining a pattern of price differences; but when performing the meaningful difference analysis, Commerce “need not consider all sales again.” Apex Opening Brief at 38. Moreover, Apex draws a distinction between the meaningful difference analysis, which goes to the threshold question of whether an alternative methodology other than A-A is appropriate, and the ultimate remedy — i.e., the weighted-average anti-dumping margin calculation. Whereas it may be reasonable to consider all sales when calculating a final antidumping duty with the A-T methodology, Apex argues it is not reasonable to do so at the threshold “account” stage.

We see no merit to Apex’s first argument that Commerce, after considering all sales in conducting its “pattern” analysis, should not consider all sales in its meaningful difference analysis. See Apex Opening Brief at 38 (“Logically, there is no need to consider ‘all sales’ ... during the second stage.... ”). To the extent Apex is arguing that Commerce’s meaningful difference test is unreasonable because it is inconsistent with the statute’s text, Apex’s argument rests on an artificially rigid reading of the statute that we find unsupported. At a minimum, even if Apex presented a plausible interpretation of the statute, it does not necessarily follow that Commerce’s differing interpretation would be unreasonable or impermissible. See Chevron, 467 U.S. at 843 n.11, 104 S.Ct. 2778 (“The court need not conclude that the agency construction was the only one it permissibly could have adopted to uphold the construction, or even the reading the court would have reached if the question initially had arisen in a judicial proceeding.”). And as to whether Commerce acted arbitrarily or capriciously, Apex’s own argument seems to suggest that, while it did not need to consider all sales, Commerce nonetheless could consider them. Thus, Apex’s argument fails.

In addition, despite Apex’s urging to the contrary, there is no basis (statutory or otherwise) for demanding a distinction between the meaningful difference analysis and the ultimate margin calculation. Nowhere is Commerce instructed how to perform a threshold “account” determination or that it must be different from the remedial margin calculation. A meaningful difference test is not even required under the statute. And, as we have already determined, Commerce has explained why a comparison of the ultimate antidumping rates sheds light on whether the A-A methodology can account for price differences — an explanation the CIT found adequate and reasonable, as do we. See Apex Frozen Foods, 144 F.Supp.3d at 1333 n.24 (“The court can discern from Commerce’s explanation that A-A cannot account for the pattern of significant price differences because A-A masked the dumping that was occurring as revealed by the A-T calculated margin. Thus, the meaningful difference between the margins demonstrated that AA is not equipped to uncover the mandatory respondents’ dumping.”).

We affirm Commerce’s decision to analyze all of Devi’s and Falcon’s sales in conducting its meaningful difference analysis as a reasonable exercise of its delegated authority.

B

Second, Apex objects to Commerce’s uneven use of zeroing in its meaningful difference analysis. As already noted, when looking at whether there was a meaningful difference between the A-A methodology and the A-T alternatives, Commerce compared the antidumping margins as they would be ultimately calculated in practice. Commerce does not use zeroing when applying the A-A methodology, but does use zeroing with the A-T methodology. See generally Union Steel, 713 F.3d at 1104-09. Apex contends that, contrary to the goal of the statute, “Commerce is simply measuring differences in [antidumping] margins caused by zeroing, rather than measuring whether A-A can account for masked dumping attributed to targeted sales.” Apex Opening Brief at 40. Apex repeats many arguments discussed already in the context of Commerce’s use of all sales. Apex argues that the disparate use of zeroing is contrary to language of the statute, which requires Commerce to determine whether A-A can account for significant price differences, “not differences in calculation methodologies attributable to zeroing.” Id. Apex also argues that, regardless of how zeroing is applied at the ultimate remedy stage, it should be applied evenly at the threshold meaningful difference analysis. Finally, Apex contends that, when zeroing is used consistently, the differences between the A-A methodology and the A-T alternatives are “miniscule,” demonstrating that there is no meaningful difference between the methodologies, except due to the distortive effects of zeroing. Apex Opening Brief at 43.

Much of our analysis from the previous discussion applies with equal force to the question now presented. As we held before, the statutory text of 19 U.S.C. § 1677f-l(d)(l)(B)(ii) does not illustrate a clear Congressional directive to Commerce. Certainly it does not demand whether Commerce is to use zeroing in any particular fashion. Therefore, we merely assess whether Commerce’s reading of the statute was permissible and whether its implementation was otherwise arbitrary, capricious, or unreasonable. See Chevron, 467 U.S. at 843-44, 104 S.Ct. 2778; Koyo Seiko, 36 F.3d at 1573.

We hold that Commerce’s meaningful difference analysis — comparing the ultimate antidumping rates resulting from the A-A methodology, without zeroing; and the A-T methodology, with zeroing — was reasonable. Apex argues Commerce can only measure masked dumping by zeroing on both sides or not at all (“a true ‘apples-to-apples’ comparison”). Apex Opening Brief at 48. But, as we stated above, nothing in the statute demands inventing a two-part analysis as Apex suggests — one calculation for the meaningful difference test and a different calculation for the ultimate remedy. Commerce’s methodology compares the A-A and A-T methodologies, as they are applied in practice, and in a manner this court has expressly condoned. See Union Steel, 713 F.3d at 1109 (“Commerce’s decision to use or not use the zeroing methodology reasonably reflects unique goals in differing comparison methodologies.”); Apex Frozen Foods, 144 F.Supp.3d at 1335 (“The zeroing characteristic of A-T is inextricably linked to the comparison methodology and its effect in the meaningful difference analysis does not render the approach unreásonable.”). Apex’s proposal for the meaningful difference analysis would require artificial comparators — either the A-T methodology without zeroing, or the A-A methodology with zeroing. We think, in light of Commerce’s contrary practices and our precedent, Apex’s preferred approach would provide a skewed perspective. At the very least, we cannot say that Commerce’s meaningful difference analysis is unreasonable — intuitively, an analysis that compares the methodologies as they would ultimately be applied “makes sense.” See Commerce Brief at 48.

Moreover, like the CIT, we find it immaterial whether the A-A and A-T margins would be nearly identical if zeroing were applied evenly or not at all. See Apex Frozen Foods, 144 F.Supp.3d at 1335 (“While [Apex] may be correct that the AT and A-A margins would be nearly identical if one were to either eliminate zeroing or zero on both sides of the comparison, that fact does not present an arguable issue.... ”). The notion that Commerce’s chosen methodology is unreasonable because it only measures the effects of zeroing is misplaced. Notwithstanding some controversy surrounding the use of zeroing, see Union Steel, 713 F.3d at 1104, differences revealed by zeroing are not inconsequential or to be ignored, as Apex seems to suggest. “In [A-A] comparisons, ... Commerce examines average export prices; zeroing is not necessary because high prices offset low prices within each averaging group. When examining individual export transactions, using the [A-T] comparison methodology, prices are not averaged and zeroing reveals masked dumping.” Id. at 1109. In other words, the effects of zeroing are precisely what 19 U.S.C. § 1677f-l(d)(l)(B) seeks to address. Apex argues that the justifications for zeroing are only relevant to the “remedy phase,” but, for the reasons already given, we reject a clear division between the “account” analysis and the “remedy” calculation.

While Commerce’s methodology may indeed be “results-oriented,” we cannot say that it preordains the use of an AT alternative methodology or that it is unreasonable. Apex’s submitted approach may offer another reasonable alternative, but “[w]hen a statute fails to make clear ‘any Congressionally mandated procedure or methodology for assessment of the statutory tests,’ Commerce ‘may perform its duties in the way it believes most suitable.’” See JBF RAK LLC v. United States, 790 F.3d 1358, 1363 (Fed. Cir. 2015) (quoting U.S. Steel Grp. v. United States, 96 F.3d 1352, 1362 (Fed. Cir. 1996)). We agree that Commerce’s chosen methodology reasonably achieves the overarching statutory aim of addressing targeted or masked dumping.

II

Apex finally argues that, even if Commerce were justified in determining that an alternative methodology should be applied, Commerce’s calculation of the “mixed” antidumping margin for Falcon was flawed.

As mentioned briefly already, 65.31 percent of Falcon’s sales passed the Cohen’s d test. Consequently, following its differential pricing analysis, Commerce applied the A-T methodology (with zeroing) to those sales passing the test, and the A-A methodology (without zeroing) for sales that did not pass, resulting in two antidumping margins: an A-T margin and an A-A margin. In this case, the A-A margin for Falcon’s sales was negative. In order to arrive at a final, weighted-average antidumping margin under this mixed alternative methodology, Commerce aggregated the two margins, but set the negative A-A margin to zero, rather than allowing it to offset the positive A-T margin. Apex argues that it was arbitrary, capricious, or otherwise unreasonable to use zeroing a second time, at the aggregation step, after already using zeroing to derive the initial A-T anti-dumping margin. According to Apex, this practice of “double zeroing” defeats the purpose of the mixed alternative methodology by undermining the A-A portion, which does not use zeroing. Apex contends the use of double zeroing resulted in a much higher (two-fold) ultimate antidump-ing duty for Falcon’s sales because “significant negative” A-A margins were zeroed, rather than offsetting positive margins. Apex Opening Brief at 54.

Critically, Apex has not challenged the mixed alternative methodology itself- — just Commerce’s chosen means of administering it. At first glance, Apex’s complaint is not entirely without merit. Commerce discontinued the practice of zeroing in the AA methodology context. See U.S. Steel Corp., 621 F.3d 1351. Zeroing the negative A-A margins would appear to “defeat the purpose” of using the A-A methodology in the mixed calculation at all, as Apex suggests. Yet Apex’s solution — that negative margins be aggregated with positive margins to offset and dampen the final, weighted average antidumping duty — runs into a similar paradox, wherein Commerce would effectively be performing “double offsetting” and “re-masking” masked dumping revealed by the A-T methodology-

This tension is the result of Commerce’s decision to merge the A-A and AT methodologies into its mixed alternative approach. “[T]he [A-A and A-T] comparison methodologies compute dumping margins in different ways and are used for different reasons.” Union Steel, 713 F.3d at 1104. It is therefore unsurprising that, in seeking to combine the two methodologies to arrive at a single antidumping rate, Commerce would be forced to subordinate the policy goals of one to the other. As explained by the CIT:

Commerce had the option to aggregate the two calculated margins by either providing for or not providing for offsets where there was negative dumping in the sales subject to A-A. Commerce has made the discretionary decision not to provide for offsets to calculate the weighted-average dumping margin for a respondent whose dumping has been assessed using more than one comparison method.

Apex Frozen Foods, 144 F.Supp.3d at 1336. It is our role merely to assess whether Commerce’s methodological choice was reasonable. Like the CIT, we find that it was. Having already concluded that the preconditions for applying the statutory exceptions were satisfied, 19 U.S.C. § 1677f-l(d)(l)(B), Commerce chose to maximize and preserve the extent of uncovered masked dumping. This decision was consistent with the overall statutory purpose.

Apex argues, without citation, that Commerce “was arbitrary, capricious, and unreasonable for automatically zeroing during the aggregation phase, and without consideration of the facts and any impact on purported ‘masking.’ Commerce must consider the evidence to understand the extent of any ‘masking’ on the targeted ... sales.” Apex Opening Brief at 56. It is not apparent on what authority Apex rests its challenge to Commerce’s methodological choice. Moreover, Apex seems to misunderstand the judiciary’s role when reviewing agency action in circumstances such as this. “When a challenge to an agency construction of a statutory provision, fairly conceptualized, really centers on the wisdom of the agency’s policy, rather than whether it is a reasonable choice within a gap left open by Congress, the challenge must fail. In such a case, federal judges — who have no constituency — have a duty to respect legitimate policy choices made by those who do.” Chevron, 467 at 866; see also PSC VSMPO-Avisma, 688 F.3d at 764 (“In examining Commerce’s approach, we must be mindful that as the ‘master of antidumping law,’ Commerce is entitled to substantial deference in its choice of ... methodology.” (quoting Thai Pineapple Pub. Co. v. United States, 187 F.3d 1362, 1365 (Fed. Cir. 1999))).

Commerce’s decision to preserve the maximum amount masked dumping by zeroing the negative A-A margin was a reasonable exercise of its delegated authority, to which we defer.

Conclusion

For the foregoing reasons, we affirm the decision of the CIT, and Commerce’s final results in AR8 are sustained.

AFFIRMED

Costs

No costs.

1

. The statute also supports using the T-T method, but the parties are in agreement that the T-T method is not at issue here. 19 U.S.C. § 1677f-l(d)(l)(A)(ii).

2

. A high-level summary of the differential pricing analysis is sufficient for our purposes, as the parties do not dispute the use and results on appeal. First, Commerce uses a statistical test referred to as the "Cohen's d" test, "a generally recognized statistical measure of the extent of the difference between the mean of a test group and the mean of a comparison group.” Joint Appendix at 1438. The Cohen’s d test yields a coefficient that may be situated within fixed thresholds: small, medium, or large. "The large threshold provides the strongest indications that there is a significant difference between the means of the test and comparison groups....” Id. As such, targeted test groups "pass” the Cohen's d test if they yield coefficients equal to or exceeding the "large” threshold.

Second, Commerce considers the ratio of the sales in the targeted groups found to have passed the Cohen’s d test to the exporter’s total sales. If the "passing" sales make up 33 percent or less of the exporter’s total sales, the results suggest that an alternative methodology is not justified and the traditional A-A methodology for all sales is adequate. If the passing sales make up 66 percent or greater, the results support the application of the alternative A-T methodology to the entirety of the exporter’s sales. Finally, if the passing sales make up between 33 and 66 percent, the results support a "mixed” alternative methodology, wherein the A-T methodology is applied only to those sales found to have passed the Cohen’s d test, but the A-A methodology is still used for sales not passing the test.

3

. In previous administrative reviews, Commerce applied what was known as the Nails test to assess exporters' pricing differences. See Mid Continent Nail Corp. v. United States, 712 F.Supp.2d 1370, 1376-79 (Ct. Int’l Trade 2010); see also Apex Frozen Foods Private Ltd. v. United States, No. 15-2085, slip op. at 8 & n.2 (Fed. Cir. July 12, 2017) (discussing the Nails test used in Commerce’s seventh administrative review ("AR7”) of certain frozen warmwater shrimp from India). Commerce explained its reasoning for the change in methodology in Differential Pricing Analysis; Request for Comments, 79 Fed. Reg. 26,720 (May 9, 2014). The propriety of Commerce’s change to its differential pricing analysis is not at issue on appeal.

4

. "[Commerce] will treat as de minimis any weighted-average dumping margin ... that is less than 0.5 percent ad valorem, or the equivalent specific rate.” 19 C.F.R. § 351.106(c). In other words, Commerce disregards anti-dumping margins that are less than 0.5 percent.

5

. We also note, again, that the statutory framework of 19 U.S.C. § 1677f-l(d)(l), by its terms, only applies to Commerce’s investigations, and not administrative reviews. Indeed, § 1677Í-I (d)(2) specifically contemplates the continued use of the A-T methodology in reviews, without elaborating on the appropriate circumstances for doing so. As such, although Commerce has elected to follow the investigations framework for its reviews as well, we will defer to a reasonable agency interpretation, given that Congress did not enact the statute to deal with the issue we face.