Apex Frozen Foods Private Ltd. v. United States

37 F. Supp. 3d 1286, 36 I.T.R.D. (BNA) 1333, 2014 Ct. Intl. Trade LEXIS 141, 2014 WL 6879122

• Court: United States Court of International Trade
• Decided: December 1, 2014
• Docket: Slip Op. 14-138; Court No. 13-00283
• Status: Published

Opinion

OPINION

GOLDBERG, Senior Judge:

Plaintiffs challenge the final results of an administrative review of an antidump-ing duty order on frozen warmwater shrimp from India. Certain Frozen Warmwater Shrimp from India,. 78 Fed. Reg. 42,492 (Dep’t Commerce July 16, 2013) (final admin, review) (“Final Results ”).. During the review, the U.S. Department of Commerce (“Commerce” or “the agency”) assigned mandatory respondent Apex Frozen Foods Private Ltd. (“Apex”) a 3.49 percent antidumping duty rate for entries between February 1, 2011, and January 31, 2012. Id. at 42,492-93. The agency assigned the same 3.49 percent rate to non-mandatory respondents, including the other plaintiffs on appeal. Id. at 42,494-96. Commerce fixed the rates using an exceptional method to remedy “targeted dumping,” which occurs when foreign exporters sell their goods in the United States for prices that differ significantly by customer, region, or time period.

Apex and others (collectively “Plaintiffs” or “Apex”) claim that Commerce incorrectly applied the targeted dumping method to calculate their antidumping rate. Apex first argues that the agency lacks authority to use the targeting dumping exception in reviews. Next, assuming Commerce could use the targeting dumping method in reviews, Apex contends that the agency misapplied the method to its detriment. Plaintiffs especially oppose the agency’s use of “zeroing” — a hotly debated technique used to calculate antidumping rates — when deploying the targeted dumping exception. But the court rejects these claims and sustains the Final Results in full.

BACKGROUND

For all of its intricacy, this case concerns just one core issue: Commerce’s calculation of Plaintiffs’ weighted-average dumping margin. A weighted-average dumping margin, more commonly known as an antidumping rate, is the amount of duty levied on foreign imports sold in the United States for less than their fair value. See 19 U.S.C. §§ 1677(35)(B), 1675(a)(1)(B) (2006). In this section, the court outlines the agency’s formula for finding antidump-ing rates generally, then discusses the two variations of the formula at issue here. The court next describes the statutory inquiry used to choose between these two variations when targeted dumping has occurred. And finally, the court recounts how Commerce applied the antidumping formula, its two variations, and the targeted dumping inquiry to find Apex’s 3.49 percent rate.

The court begins by outlining the general formula for calculating antidumping rates. The analysis proceeds in four steps. First, Commerce gathers a foreign exporter’s normal values (or sales prices of the subject goods in the exporter’s home country) and export prices (or sales prices of the subject goods in the United States). See id. §§ 1677b(a)(l)(A)-(B), 1677a(a). Second, the agency subtracts the exporter’s export prices from its normal values, yielding a series of “dumping margins.” The court refers to this step as the “comparison step.” See id. § 1677(35)(A). Third, Commerce weight-averages or aggregates the dumping margins into a single, composite margin. The court calls this the “aggregation step.” See id. § 1677(35)(B). And fourth, the agency divides the combined margin by the good’s weight-averaged export price. See id. The quotient is the foreign exporter’s ad valorem antidumping rate.

Over the years, Commerce has used two distinct variations of this process to set antidumping rates in reviews. Before 2012, Commerce typically relied on a version of the formula called the average-to-transactional method (“A-T”). See Calculation of the Weighted-Average Dumping Margin and Assessment Rate in Certain Antidumping Duty Proceedings, 77 Fed. Reg. 8101, 8101 (Dep’t Commerce Feb. 14, 2012) (final modification) (“Final Mod.”). The A-T method differs from other versions of the antidumping formula in its approach to averaging at the comparison step and its treatment of margins at the aggregation step. At the A-T comparison step, the statute requires Commerce to subtract individual export prices from normal values weight-averaged by month. 19 U.S.C. § 1677f — 1(d)(1)(B), (d)(2). This yields a dumping margin for each U.S. sale of the foreign good. The margins may be positive, and indicative of dumping, if normal value exceeds the export price, or negative, and not indicative of dumping, if the export price exceeds normal value.

Next, at the A-T aggregation step, Commerce assigns each negative margin a value of zero, then weight-averages these margins together with positive margins. Final Mod. at 8101. This method of aggregating margins, called “zeroing,” is not required by statute. Nonetheless, Commerce claims that the ambiguous definition of “dumping margin” in 19 U.S.C. § 1677(35)(A) permits it to exclude negative margins from the weighted-average dumping margin. See Timken Co. v. United States, 354 F.3d 1334, 1342 (Fed.Cir.2004). The Federal Circuit has approved zeroing as a reasonable way to aggregate margins under A-T. See id.

The A-T method would not always be Commerce’s favored approach in reviews, however. Between 2007 and 2009, the World Trade Organization (“WTO”) Appellate .Body ruled that the United States could not use zeroing in reviews as a default policy.¹ In response to the WTO rulings, Commerce decided it would no longer apply A-T with zeroing in each review proceeding. The agency would instead apply the average-to-average method (“A-A”), a version of the antidumping formula prescribed as the default approach in investigations, as its new default method in reviews. See Final Mod. at 8102; 19 C.F.R. § 351.414(c)(1) (2014) (requiring AA in reviews unless another method is more appropriate).

The A-A method differs from A-T in two ways. First, the law requires Commerce to subtract weight-averaged export prices from weight-averaged normal values at the A-A comparison step. 19 U.S.C. § 1677f-l(d)(l)(A)(i). Commerce calculates the averages by month, producing twelve separate dumping margins. 19 C.F.R. § 351.414(d)(3). Second, at the AA aggregation step, Commerce combines the twelve margins into one weighted average, including both positive and negative margins in the equation. See Final Mod. at 8102. By computing a net average, AA cancels out dumped margins to the extent there are nondumped margins to offset them. See U.S. Steel Corp. v. United States, 621 F.3d 1351, 1354-55 (Fed.Cir.2010). This offsetting technique, while not required by statute, has been upheld as a reasonable way to aggregate margins under A-A. Id. at 1361-63. Thus A-A produces rates that reflect average dumping over the course of a review period, but not the amount of dumping effected by individual U.S. sales. See U.S. Steel Corp. v. United States, 33 CIT 984, 986, 637 F.Supp.2d 1199, 1205 (2009).

Yet Commerce did not wholly abandon zeroing when it adopted A-A with offsets as its default method in reviews. As explained earlier, the law provides A-A as a default methodology in investigations. See 19 U.S.C. § 1677f-l(d)(l)(A)(i). But the statute also allows Commerce to use A-T as an exception to A-A to counteract “targeted dumping.” See id. § 1677f-1(d)(1)(B); Uruguay Round Agreements Act, Statement of Administrative Action, H.R. Doe. No. 103-316, vol. 1, at 842-43 (1994), reprinted in 1994 U.S.C.C.A.N. 4040, 4177-78 (“SAA”). Targeted dumping occurs when exporters sell their goods in the United States at dumped prices “to particular customers or regions,, while selling at higher prices to other customers or regions.” SAA at 842. When an exporter’s sales are structured this way, A-A can understate dumping by measuring average dumping over the entire review period, while masking dumping from discrete targeted sales. In contrast, A-T remedies more dumping from targeting by computing a margin for each targeted sale and by disregarding negative values when combining margins into a final rate.

When it adopted A-A as its new default method in reviews, Commerce grafted the targeted dumping exception from investigations into its review procedure. Final Mod. at 8106-07. Now, before applying A-T instead of A-A in reviews, the agency completes a two-step test outlined in statute. The statutory inquiry (the “TD inquiry”) reads as follows:

(B) Exception

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) [i.e., A-A and T-T].

19 U.S.C. § 1677f-l(d)(l)(B).² To paraphrase, Commerce must first decide whether the exporter actually targeted sales by customer, region, or time period. If the exporter targeted sales, then the agency must explain why A-A cannot account for dumping from those sales. Commerce may use A-T to calculate the exporter’s rate if it provides the explanation required.

In this case, Commerce conducted a TD inquiry, then used A-T to set Apex’s rate. At the TD inquiry’s first step, the agency found Apex had targeted some of its sales by time period and customer. See Issues & Decision Mem. (“I & D Memo”) at 10, 14, PD 279 (July 10, 2013). Commerce used the Nails test, a court-approved method for pinpointing targeted sales, to reach this conclusion. See id.; Mid Continent Nail Corp. v. United States, 34 CIT -, -, 712 F.Supp.2d 1370, 1376-79 (2010). No one disputes the finding that Apex targeted some of its sales during the review period.

Next, Commerce analyzed whether A-A could account for the dumping from Apex’s targeted transactions. I & D Memo at 15. The agency first found that Apex’s A-A rate was 0.0 percent. Then Commerce found that Apex’s A-T rate was 3.49 percent. After comparing the two rates, the agency declared that the difference between the rates was “meaningful because it crosses the de minimis threshold,” and that A-A could not adequately account for Apex’s targeting. Id. Commerce then used A-T to calculate the rate, applying the method uniformly to Apex’s targeted and untargeted sales to yield a 3.49 percent result.

In its administrative briefing, Apex contested the agency’s decision to apply A-T to calculate its antidumping rate. See Apex Case Br., PD 265 (Apr. 18, 2013). Apex’s claims were fourfold: (1) Commerce could not use the TD inquiry to decide whether to use A-T instead of A-A, because the statute bars the TD inquiry from reviews, id. at 8-12; (2) the agency failed to explain why A-A could not account for dumping from Apex’s targeted sales, id. at 14-23; (3) Commerce could not apply A-T to all of Apex’s sales, both targeted and untargeted, to calculate Apex’s rate, id. at 25-26; and (4) the agency failed to explain why it zeroed when using A-T, but not when using A-A, id. at 12-14. Commerce rejected all of these claims in the Final Results, and Apex appealed. See Summons, ECF No. 1.

JURISDICTION AND STANDARD OF REVIEW

The court has jurisdiction to hear this appeal under 28 U.S.C. § 1581(c). The court 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)(l)(B)(i).

DISCUSSION

In light of these standards, and after carefully studying each of Apex’s claims, the court sustains Commerce on all points. To begin, Apex has not shown that the statute barred Commerce from conducting a TD inquiry below. Nor has Apex proven that Commerce mishandled the inquiry by inadequately explaining its reliance on AT, by applying A-T to all sales, or by failing to justify its inconsistent usé of zeroing. The court addresses each of these topics in turn.

I. Commerce Had Authority to Conduct the TD Inquiry Below

Plaintiffs first claim that Commerce lacked authority to conduct a TD inquiry in the review below. Pis.’ Mot. for J. on Agency R. 13-19, ECF No. 37 (“Pis.’ Br.”). They note that the statute expressly provides for TD inquiries in investigations, 19 ' U.S.C. § 1677f-l(d)(l)(B), but is silent regarding targeting in reviews, see id. § 1677f-l(d)(2). In Apex’s view, this silence reveals an intent by Congress to withhold the TD inquiry from reviews. See Pis.’ Br. 13-19.

The court disagrees and holds that Commerce has authority to use the TD inquiry in reviews. Although the statute mentions the inquiry only in the context of investigations, the court has declared, in a swelling chorus of cases, that Commerce may apply the TD inquiry in review proceedings. See DuPont Teijin Films China Ltd. v. United States, 38 CIT -, -, 7 F.Supp.3d 1338, 1355-56 (2014) (sustaining TD inquiry in reviews); JBF RAK LLC v. United States, 38 CIT-, -, 991 F.Supp.2d 1343, 1347-49 (2014) (same); CP Kelco Oy v. United States, 38 CIT-,-, 978 F.Supp.2d 1315, 1321-24 (2014) (same); Timken Co. v. United States, 38 CIT-,-, 968 F.Supp.2d 1279, 1286-87 n. 7 (2014) (same). These cases address each of the statutory arguments made in the briefs, and the court need not repeat their reasoning.

In addition to making statutory arguments, Plaintiffs vie that the legislative history of the Uruguay Round Agreements Act (“URAA”) shows Congress’s intent to limit the TD inquiry to investigations. Pis.’ Br. 17-18. Like the statute, the Statement of Administrative Action mentions the inquiry in the context of investigations, but not in the context of reviews. SAA at 843. But the court finds Apex’s argument unpersuasive. When the SAA was drafted, Congress had no reason to consider extending the TD inquiry to reviews. The Uruguay Round agreements, signed by participating countries in 1994, required the United States to use A-A instead of A-T as its default method in investigations. Agreement on Implementation of Article VI of the General Agreement on Tariffs and Trade 1994, art. 2.4.2, Apr. 15, 1994, 1868 U.N.T.S. 201. The agreements did not bar the use of A-T in reviews, however. See SAA at 843. And so, when Congress passed the URAA to comply with the agreements, it had ho need to enact a method for choosing between A-A and A-T in reviews, because A-T remained the default there. See id. at 842-43. Thus the legislature’s silence does not manifest an intent to withhold the TD inquiry from reviews. The silence just as readily betokens Congress’s benign inattention to the matter.

II. Commerce Reasonably Explained Why A-A Could Not Account for Targeting

Even assuming the agency had authority to conduct a TD inquiry below, Apex claims that Commerce mishandled the inquiry. To be clear, the Plaintiffs do not argue that Commerce erred at the first step of the inquiry. The agency found [[Confidential Data Deleted ]] percent of Apex’s U.S. sales were targeted by customer or time period, and Apex has no direct quarrel with this finding. See Pis.’ Reply Br. 5, ECF No. 53 (“Reply Br.”) (calling irrelevant the question of whether targeted dumping occurred); Apex Prelim. Results Calculations (“Prelim. Calcs.”) at 3, CD 228 (Mar. 4, 2013). But Plaintiffs do contest the agency’s analysis at the second step of the TD inquiry, calling it “arbitrary, capricious, and inconsistent.” Pis.’ Br. 19. More specifically, Apex alleges that Commerce inadequately explained why AA could not account for dumping from targeted sales, or why Commerce had to apply A-T instead.

In the I & D Memo, Commerce gave its rationale for applying A-T instead of A-A. It began by finding the difference between Apex’s antidumping rate under A-A and the rate under A-T. The agency explained,

[T]he Department evaluated the difference between the weighted-average dumping margin calculated using the A-to-A method and the weighted-average dumping margin calculated using the A-to-T method. Where there is a meaningful difference between the results of the A-to-A method and the A-to-T method, the A-to-A method would not be able to take into account the observed price differences, and the A-to-T method would be used to calculate the weighted-average margin of dumping for the respondent in question.

I & D Memo at 15. When generating 'the rates, Commerce zeroed when using A-T, but not when using A-A. Then, after finding the antidumping rate under each method, Commerce explained why it found A-A could not account for Apex’s targeted sales:

In this case, Apex’s margin is zero using the A-to-A method and 3.49 percent using the A-to-T method. The Department has concluded that for the purposes of this case, such a difference is meaningful because it crosses the de minimis threshold and warrants the application of the A-to-T method.

Id. In short, Commerce chose A-T to calculate Apex’s rate because A-A yielded a zero margin, and A-T revealed dumping that A-A masked.

Apex challenges four facets of this analysis. First, they argue that the statute required Commerce to compare A-A and A-T on an “apples-to-apples” basis, without zeroing under either method. By zeroing under A-T but not A-A, Commerce ensured that the A-T rate would exceed the A-A rate by a considerable amount. The agency thus perverted step two of the TD inquiry into a useless formality, and guaranteed the use of A-T to set Apex’s rate. See Pis.’ Br. 22-24.

Second, Apex alleges that Commerce failed to explain why the difference between the A-A and A-T rates was “meaningful.” In its analysis, Commerce concluded that the difference was meaningful because it was more than de minimis, or 0.5 percent. I & D Memo at 15. Yet Plaintiffs argue that this 0.5 percent cutoff was “results-oriented” and arbitrary. See Pis.’ Br. 23. As Apex interprets the test, the agency would find a difference of just 0.1 percent to be “meaningful” if the A-T rate was slightly higher than 0.5 percent, and the A-A rate was slightly lower than 0.5 percent. Such a hair-trigger distinction ensured that Commerce would find A-A could not account for Apex’s targeting. Id. at 23-24.

Third, Apex contends that Commerce had to explain whether A-A with monthly-averaging could remedy targeting as well as A-T. Agency regulations hint that A-A with monthly averaging accounts for price anomalies better than A-A with annual averaging. Consequently, Apex alleges that Commerce had to explore the merits of A-A with monthly averaging before choosing A-T to remedy Apex’s targeting. Id. at 24-27.

Finally, Plaintiffs argue that Commerce should have considered seasonal price changes when deciding whether to depart from A-A. Apex suggests it was unnecessary to use the A-T remedy because involuntary variations in demand for shrimp caused the targeting. Id. at 27-28.

The court considers these arguments in turn and finds that each lacks merit.

A. Apex Has Not Shown that the Meaningful Difference Analysis Was Unlawful

At the outset, the court makes one point plain: Apex does not challenge Commerce’s authority to conduct a meaningful difference analysis under § 1677f-l(d)(l)(B)(ii). Plaintiffs concede that significant differences between the A-A and A-T rates may prove that A-A cannot account for targeting. Pis.’ Br. 22-23 (suggesting that statute requires Commerce to contrast different comparison methodologies at TD inquiry’s second step).

Apex’s argument is more subtle than that. As explained previously, the A-A and A-T methods differ in two ways, one statutory, and the other discretionary. By statute, A-A and A-T determine margins at the comparison step using distinct approaches. A-A compares average export prices and normal values to set margins, but A-T compares individual export prices to average normal values to set margins. See 19 U.S.C. § 1677f-l(d)(l)(A)-(B). The A-A and A-T methods also differ in a second way, but this difference is not ensconced in statute. Under A-T, Commerce has chosen to zero negative margins at the aggregation step to remedy transaction-specific dumping, Final Mod. at 8101; but under A-A, the agency offsets positive and negative margins to find average dumping over the course of a review period, see id. at 8102. Apex argues that Commerce erred to compare rates made using A-T with zeroing and A-A without zeroing, because the only statutory difference between the methods concerns averaging at the comparison step. By comparing' a zeroed rate to a nonzeroed rate, Apex continues, Commerce ensured that the A-T rate would exceed the A-A rate, thus guaranteeing the use of the A-T remedy. Pis.’ Br. 22-23.

Yet neither the statute nor logic prohibited the agency’s approach in the review below. It was reasonable, as a legal matter, for Commerce to compare a zeroed AT rate to a nonzeroed A-A rate to decide whether A-A could account for Apex’s targeting.

To begin, the statute does not require that the disparity between rates stem solely from the way A-A and A-T average export prices and normal values. In fact, § 1677f-l(d)(l)(B)(ii) does not compel Commerce to conduct a meaningful difference analysis at all. Rather, the statute simply states that Commerce may apply A-T if “the administering authority explains why such differences [i.e., significant differences in export prices by customer, region, or time period] cannot be taken into account using a method described in paragraph (l)(A)(i) or (ii) [i.e., A-A or TT].” 19 U.S.C. § 1677f-l(d)(l)(B)(ii). This language leaves the form of the explanation to the agency’s discretion, and here, Commerce chose to explain the A-A method’s shortcomings by contrasting Apex’s A-A and A-T rates. The statute does not dictate whether the agency needed to zero one of these rates, both of the rates, or neither. As a consequence, Commerce could compare a zeroed A-T rate to a nonzeroed A-A rate, as long as the comparison demonstrated whether the A-A method could account for targeting. See Chevron, U.S.A., Inc. v. Natural Res. Def. Council, Inc., 467 U.S. 837, 842-43, 104 S.Ct. 2778, 81 L.Ed.2d 694 (1984) (giving deference where statute is ambiguous and interpretation is reasonable).

The court holds that the agency’s comparison reasonably achieved the statute’s purpose. As mentioned above, Congress enacted § 1677f — 1(d)(1)(B) to combat a type of dumping that is difficult to remedy. When exporters target their sales, those sales may disproportionately affect U.S. producers who sell to particular customers or regions or in specific time periods. See SAA at 842. Nonetheless, if exporters counterweight their targeting with above-fair-value sales, then A-A, which averages export prices and offsets negative margins, could yield an understated antidumping rate. See Union Steel v. United States, 713 F.3d 1101, 1108 (Fed.Cir.2013) (holding A-A with offsets “masks individual transaction prices below normal value”). Yet A-T neither averages export prices nor offsets. This approach ensures that the final rate reflects every instance of dumping, even if an exporter balanced its targeting with above-fair-value sales. Id. at 1108-09 (holding A-T with zeroing “reveals individual dumping”). Hence, by comparing Apex’s nonzeroed A-A rate to its zeroed A-T rate, the agency found the precise amount. of dumping — including dumping from targeted sales — that A-A masked. Commerce could then decide whether that dumping was great enough to merit an exceptional remedy. This method fulfills the statute’s aim and deserves deference. See Chevron, 467 U.S. at 842-43, 104 S.Ct. 2778.

Apex also argues that zeroing magnifies the difference between the A-T and A-A rates, thereby guaranteeing the finding that A-A cannot account for targeting. Pis.’ Br. 22-23. But the agency’s approach to zeroing, in itself, did not foreordain the A-T remedy or undermine Commerce’s explanation under § 1677f~l(d)(l)(B)(ii). It is true that A-T rates sometimes exceed A-A rates, because A-T zeroes negative margins that would otherwise offset positive margins under A-A. See, e.g., Union Steel v. United States, 36 CIT-,-, 823 F.Supp.2d 1346, 1361-64 (2012) (appendices showing rates made with zeroing higher than rates made with offsets). Nonetheless, Commerce does not discard its default method whenever the A-T rate exceeds the A-A rate. The agency instead examines whether the rates differ meaningfully before concluding that A-A cannot account for targeting. I & D Memo at 15.³ So even if zeroing widens the gap between the A-A and A-T rates, it does not predestine the A-T remedy or render superfluous the explanation mandated in the statute. The court continues, as before, to uphold Commerce’s reasonable interpretation of § 1677f-l(d)(l)(B)(ii).

Before moving to Apex’s next argument, however, there is another matter the court must confront. Besides arguing that Commerce’s explanation was contrary to law, Apex challenges the conclusion that A-A could not account for targeting as unfounded in substantial evidence.

Under the statute, Commerce must explain “why such differences cannot be taken into account” using A-A before it applies the alternative A-T method. 19 U.S.C. § 1677f-l(d)(l)(B)(ii) (emphasis added). The “differences” mentioned here are the significant differences in export prices “among purchasers, regions, or periods of time” noted in § 1677f-l(d)(l)(B)(i). In other words, Commerce must explain why A-A cannot account for dumping from targeted sales before it deploys the A-T remedy. But as Apex observed in its reply brief, [[Confidential Data Deleted ]] percent of its sales were not targeted. Reply Br. 10. Apex also furnished a document at oral argument showing that [[Confidential Data Deleted ]] percent of its sales were dumped but not targeted. Prelim. Calcs, at Attach. 2; Oral Argument at 28:40. Plaintiffs infer from these data that the difference between the 0.0 percent A-A rate and 3.49 percent A-T rate stems mainly from ordinary dumping, not targeted dumping. See Reply Br. 9-10. If this were true, then the finding that A-A could not account for Apex’s targeting might be untethered from substantial evidence. While the difference between rates would prove that A-A accounts for less dumping than A-T in general, the difference would not necessarily show that A-A cannot account for targeted dumping in particular, as required by statute. See 19 U.S.C. § 1677f-l(d)(l)(B)(ii).

The court declines to consider this argument, however, because it was not exhausted before the agency. By regulation, Apex had to present in its administrative case brief “all arguments that continue in the submitter’s view to be relevant to the Secretary’s ... final results.” 19 C.F.R. § 351.309(c)(2). And by statute, if Apex failed to brief an argument before Com-, merce, the court need not consider the argument on appeal. See 28 U.S.C. § 2637(d). This rule gives the agency the “opportunity to correct its own mistakes,” including fact-specific shortfalls in its analysis, “before it is haled into federal court.” McCarthy v. Madigan, 503 U.S. 140, 145, 112 S.Ct. 1081, 117 L.Ed.2d 291 (1992); see also Blue Field (Sichuan) Food Indus. Co. v. United States, 37 CIT-, -, 949 F.Supp.2d 1311, 1324 (2013) (finding argument regarding reliability of surrogate data not-exhausted where party failed to make argument in case brief and to cite relevant evidence).

Commerce had no such opportunity to correct the alleged flaw in its meaningful difference finding. In its case brief, Apex remarked that “[r]ather than measuring the effect of targeted dumping,” Commerce’s test “in fact measure[ed] mostly the impact of zeroing.” Apex Case Br. 23. But this argument was framed as a legal challenge, not as a substantial evidence challenge. Nowhere in the case brief did Apex cite data to show that most of the dumping in the 3.49 percent rate was untargeted. Instead, Apex illustrated that its A-A and A-T rates would be nearly identical if Commerce zeroed both measures. Id. While this demonstrative, proved that zeroing one rate but not the other was the main reason the rates differed, the illustration did not reveal whether the A-T rate captured dumping mainly from untargeted sales. Because the case brief gave Commerce no notice of Apex’s substantial evidence argument, the argument was not exhausted. See Trust Chem Co. v. United States, 35 CIT-, -, 791 F.Supp.2d 1257, 1268 n. 27 (2011).⁴

Even if Apex exhausted its substantial evidence argument, the court still would not consider it. The Court’s rules require plaintiffs to raise any argument on appeal in their lead briefs. See USCIT Rule 81(J) (“A reply brief or memorandum must be confined to rebutting matters contained in the brief of a responding party”). This requirement gives defendants a chance to respond to plaintiffs’ arguments, and ensures that any matters raised are sufficiently developed to permit the court to rule. See Novosteel SA v. United States, 284 F.3d 1261, 1273-74 (Fed.Cir.2002). But here, Apex’s briefing afforded neither of these benefits to the defendant or the court. In its initial brief, Apex argued (as it did below) that Commerce unfairly compared zeroed to nonzeroed rates. Pis.’ Br. 22-23. It was not until the reply brief, however, that Apex explicitly alleged that the A-T rate reflected mostly dumping from untargeted sales. Reply Br. 9-10. And even then, Plaintiffs did not offer clear evidence to prove the point. Although Apex argued that [[Confidential Data Deleted ]] percent of its sales were not targeted, these data merely implied, but did not show, that the A-T rate embraced other than targeted dumping. Id. Only at oral argument did Apex supply record evidence to suggest that most of the dumping was not targeted — and this over an objection that the evidence had not been cited in the briefs. Oral Argument 32:20. Because Apex raised its substantial evidence argument at the eleventh hour, the argument was waived.

Plaintiffs rejoin that their fact-specific argument was implicit in their submissions both to Commerce and to the court. According to Plaintiffs, their claim against the application of the A-T method to all sales — both targeted and untargeted— makes sense only if some untargeted sales were dumped. See Pis.’ Supp. Br. 6-7, ECF No. 65; see also Pis.’ Br. 28-34.⁵ Were there no dumping among Apex’s un-targeted sales, then A-A and A-T would yield the same null rate for those sales. This would have left Apex without an incentive to dispute the global application of A-T. Yet because Apex did dump some of its untargeted sales, Plaintiffs reason that A-T yielded a higher rate than A-A for those sales; thus Apex had cause to restrict the A-T method to targeted sales only. Apex concludes that its argument to limit A-T to targeted sales put Commerce on notice that some of its dumped sales were not targeted. Id.

The court is not convinced. Even if Apex’s argument hinted that some untar-geted sales were dumped, this would not have alerted Commerce to a flaw in its meaningful difference results. Apex moved to limit A-T to targeted sales at the TD inquiry’s remedy phase, but the meaningful difference test comes a step earlier, when Commerce considers whether A-A can account for targeting. See Pis.’ Br. 21-24 (meaningful difference argument), 28-34 (TD remedy argument); see also I & D Memo at 15 (describing sequence of the TD inquiry). The agency would not naturally infer from an argument made at the remedy step that a conclusion made at an earlier step was wrong. And furthermore, even if Apex had implied its substantial evidence challenge more overtly, Commerce need not respond to arguments that are. merely implicit in the briefing. To exhaust the argument below and to preserve the argument on appeal, Apex had to state its challenge in reasonably explicit terms and provide some evidence in support. See Nat’l Ass’n of Clean Air Agencies v. EPA 489 F.3d 1221, 1231 (D.C.Cir.2007) (holding agency need not answer arguments merely implied in submissions). Because Apex did neither of these things until its reply brief, the court cannot consider the argument now.

Based on the record and arguments be-, fore it, the court holds that the comparison of a zeroed A-T rate to a nonzeroed A-A rate under § 1677f-l(d)(l)(B)(ii) complies with law.

B. Commerce’s Meaningful Difference Threshold Was Not Arbitrary

Next, and in a similar vein, Apex challenges Commerce’s evaluation of the difference between the A-A and A-T rates. In the I & D Memo, the agency noted that Apex’s A-A rate was 0.0 percent, and that the A-T rate was 3.49 percent. The agency then “concluded that for purposes of this case,” the difference between rates was “meaningful because it crosses the de minimis threshold and warrants the application of the A-to-T method.” I & D Memo at 15.

Apex argues that it was arbitrary for Commerce to apply the A-T remedy simply because the A-T rate was more than de minimis. Pis.’ Br. 23-24. As Plaintiffs interpret the test, Commerce would apply A-T if an exporter’s rate were 0.45 percent under A-A and 0.55 percent under A-T, because in this hypothetical, the A-T rate exceeds the de minimis threshold of 0.5 percent. But to apply A-T based on a 0.1 percentage-point difference in rates might be unreasonable. Hence, in Plaintiffs’ view, for Commerce to use A-T, simply because Apex’s 3.49 percent rate exceeds 0.5 percent, was “arbitrary and results-oriented.” Id. at 23.

The court finds Commerce’s reliance on the de minimis standard was not arbitrary, as Apex alleges. Defined as 0.5 percent in 19 C.F.R. § 351.106(c)(2), the de minimis threshold represents the minimum weighted-average dumping margin needed to levy antidumping duties in reviews. That is, the agency does not impose duties at all if it finds that an exporter’s rate is less than or equal to 0.5 percent. The threshold is small by design, because reviews aim “to counteract as much dumping behavior” as possible. Union Steel, 36 CIT at -, 823 F.Supp.2d at 1359.

Using this regulation as a guide, Commerce reasonably concluded that A-A could not account for dumping from targeted sales. When it calculated Apex’s rate using A-A, Commerce found no dumping. I & D Memo at 15. Had the agency relied strictly on A-A to fix Apex’s rate, § 351.106(c)(2) would have precluded Commerce from imposing antidumping duties. Yet when Commerce calculated the rate using A-T, it found a rate of 3.49 percent. Id. This figure is nearly seven times greater than the minimum threshold needed to impose duties. The rates thus offered a stark choice. Commerce could fully counteract Apex’s targeted dumping by applying A-T, or it could counteract none of Apex’s targeted dumping by applying A-A. Because A-T would yield some duties but A-A would yield none, Commerce reasonably decided that the difference between the rates was “meaningful,” and that A-A could not account for dumping from targeted sales. Cf. Beijing Tianhai Indus. Co. v. United States, 38 CIT -, -, 7 F.Supp.3d 1318, 1331-32 (2014) (holding explanation invalid where Commerce provided no basis to conclude that A-A could not account for targeting).⁶

Even so, Apex argues that this case mirrors Kelco, which was remanded for Commerce to explain a different, ill-defined de minimis test. See Reply Br. 7. In Kelco, Commerce examined whether plaintiff had targeted its sales under § 1677f-l(d)(l)(B)(i). Kelco, 38 CIT at-, 978 F.Supp.2d at 1319-20. The agency found targeting, and concluded that because targeted sales comprised more than a de min-imis share of total sales, it would consider whether A-A accounted for targeting under § 1677f-l(d)(l)(B)(ii). Id. Yet Commerce never defined how many sales must be targeted to qualify as more than de minimis. The court held that it was arbitrary to omit this explanation from the analysis. Id. at -, 978 F.Supp.2d at 1327-29.

Here, Apex challenges a different test and on different grounds. Unlike Kelco, which considered a de minimis test at the TD inquiry’s first step, this case examines the use of a de minimis threshold at the TD inquiry’s second step, § 1677f-l(d)(l)(B)(ii). Furthermore, unlike Kelco’s plaintiff, Apex does not claim that Commerce failed to define the de minimis threshold used here. As shown by their brief, Plaintiffs clearly understood that an-tidumping rates are more than de minimis if they exceed 0.5 percent. See Pis.’ Br. 23-24 (equating de minimis standard with 0.5 percent threshold from regulation). Apex instead argues that Commerce could not apply A-T simply because the difference between its A-A and A-T rates was greater than 0.5 percent. Id. But this argument faults Commerce for the analysis it rendered, not for analysis that it neglected to give. Kelco is inapposite, and the court sustains Commerce’s use of the de minimis threshold to decide whether A-A could account for targeted sales.

C. Commerce Did Not Need to Explain the Impact of Monthly Averaging in Reviéws

Plaintiffs do not only dispute Commerce’s affirmative findings, however. They also fault the agency for perceived omissions in its § 1677f — 1 (d)(1)(B)(ii) analysis. Apex argues, for instance, that Commerce had to explain whether A-A with monthly averages could account for targeting before selecting the A-T remedy. See Pis.’ Br. 24-27. A regulation allows Commerce to use A-A with monthly instead of annual averages if normal values and export prices differ significantly over a period of investigation or review. 19 C.F.R. § 351.414(d)(3). To Apex, this means A-A with monthly averages might unmask “distortions and fluctuations” from targeted sales just as well as A-T. Pis.’ Br. 26.

The court rejects this argument. By comparing Apex’s A-A and A-T rates, Commerce already showed that A-A with monthly averages could not account for targeted dumping. The agency’s default method for setting antidumping rates in reviews, as mentioned before, is A-A with offsets and monthly averaging of export prices. Final Mod. at 8102. Commerce found a 0.0 percent rate when it applied this method to Apex’s sales. See I & D Memo at 15. Then Commerce applied AT, a method that neither averages export prices nor offsets negative margins, and this yielded a 3.49 percent rate. See id. The disparity between the rates demonstrated that A-A with monthly averaging accounted for none of the dumping from Apex’s targeting. See supra Discussion II.B. This comparison and explanation sufficed to carry Commerce’s burden under § 1677f-l(d)('l)(B)(ii). Whether A-A with monthly averages generally captures more dumping than A-A with annual averages was irrelevant to the matter.

D. Commerce Did Not Need to Account for Seasonality in Its Targeting Analysis

Apex next challenges another perceived omission in Commerce’s § 1677f-1(d)(1)(B) analysis. According to Plaintiffs, Commerce had to consider seasonal changes in demand for frozen shrimp before imposing the A-T remedy. Pis.’ Br. 27-28. In an agency hearing, Apex noted that demand for shrimp falls in the fourth quarter of each calendar year, depressing U.S. prices below normal levels and creating a false image of targeting. See Hr’g Tr. at 19-20, PD 275 (June 17, 2013). Plaintiffs imply that Commerce could not use A-T to remedy targeting caused by natural moves in supply and demand, and not by intentional acts.

The court again disagrees, because the statute does not require Commerce to decide why sales were targeted before imposing the A-T remedy. Two recent cases note that the statute is silent regarding exporters’ intent to target their U.S. sales. JBF RAK, 38 CIT at-, 991 F.Supp.2d at 1355; Borusan Mannesmann Boru Sanayi Ve Ticaret A.S. v. United States, 38 CIT-,-, 990 F.Supp.2d 1384, 1389 (2014). The agency thus had discretion to examine intent and seasonality—or not—in its § 1677f—1(d)(1)(B) analysis. And furthermore, though it may seem fair to impose the A-T remedy only when exporters intentionally target their sales, the court must not lay an analytical “burden on Commerce that is not required or suggested by the statute.” Borusan, 38 CIT at -, 990 F.Supp.2d at 1389 (quoting Viraj Grp. v. United States, 476 F.3d 1349, 1358 (Fed.Cir.2007)). The court would breach this rule if it forced Commerce to consider seasonality where the law does not demand it.

Apex also suggests that Commerce was bound to consider seasonality as a matter of agency practice. Pis.’ Br. 27-28. This argument fails too, because Apex has not shown that Commerce uniformly considers seasonality in its TD inquiry. See Ranchers-Cattlemen Action Legal Found v. United States, 23 CIT 861, 884-85, 74 F.Supp.2d 1353, 1374 (1999) (holding agency actions become binding when practice produces reasonable reliance). Indeed, Apex cites just one seventeen-year-old proceeding where Commerce considered seasonality, and this in the context of determining normal values. See Certain Fresh Cut Flowers from Colombia, 62 Fed. Reg. 53,287, 53,295-97 (Dep’t Commerce Oct. 14, 1997) (final admin, review). This single proceeding, which did not concern targeted dumping, did not form a binding practice. Commerce did not act arbitrarily by refusing to consider seasonality below.

III. Commerce Reasonably Applied the A-T Remedy to All Sales

After finding targeting among Apex’s sales, and after deciding that A-A could not account for the targeting, Commerce used the A-T method to calculate Apex’s antidumping rate. In doing so, Commerce applied A-T not only to targeted sales, but also to Apex’s untargeted sales. See I & D Memo at 15. Plaintiffs claim this was error. In their view, the statute clearly requires Commerce to deploy A-T only against targeted sales. Pis.’ Br. 29. Apex also contends that the agency promised to limit A-T to targeted sales when it adopted its investigation procedures in reviews. Id. at 29-34.

Neither argument persuades. First, the statute does not restrict the A-T method to targeted sales once targeting is found. It is true, as a general matter, that the agency must use A-A to calculate exporters’ antidumping rates in investigations: 19 U.S.C. § 1677f-l(d)(l)(A) invokes the unequivocal “shall” when ordering Commerce to use A-A or T-T in ordinary cases. But even so, if the agency identifies targeting that default methods cannot explain, the statute permits Commerce to “determine whether the subject merchandise is being sold in the United States at less than fair value” using A-T. 19 U.S.C. § 1677f-l(d)(l)(B) (emphasis added). Notice the statute does not expressly confine A-T to targeted sales. Instead, it permits Commerce to apply A-T to “the subject merchandise,” regardless of whether the merchandise was sold in a pattern of targeting. Given this ambiguous language, Commerce could apply A-T to all sales or to targeted sales only, provided that the choice was reasonable. See Chevron, 467 U.S. at 842-43, 104 S.Ct. 2778.

The agency’s decision to apply A-T to all sales was reasonable here. Commerce explained in the I & D Memo that applying A-T to all sales was “more consistent with the Department’s approach to selection of the appropriate comparison method” than applying A-T only to targeted sales. I & D Memo at 18. Because Commerce applies A-A to all sales under § 1677f-l(d)(l)(A), the agency found it equally fair to apply A-T to all sales under § 1677f-l(d)(l)(B). But more importantly, Commerce noted that applying A-T to all sales

eliminates the masked dumping by exposing 1) any implicit masking within the weighted-average U.S. sales price by basing the comparison on the transaction-specific U.S. sales price rather than the weighted-average U.S. sales price, and 2) any explicit masking between individual comparison results by not providing offsets for negative comparison results.

Id. at 19. More succinctly, Commerce applied A-T across the board to reveal dumping hidden by sales that were neither targeted nor dumped. This ’ approach served the statute’s purpose and warrants appropriate deference.⁷

Second, and notwithstanding any ambiguity in the statute, Apex argues that the law still bound Commerce to limit A-T to targeted sales. Pis.’ Br. 29-34. Pursuant to a regulation called the “Limiting Rule,” Commerce long restricted A-T to targeted sales when it used the TD exception in investigations. See 19 C.F.R. § 351.414(f)(2) (2008). And although the agency withdrew the Limiting Rule in 2008, see Withdrawal of the Regulatory Provisions Governing Targeted Dumping in Antidumping Duty Investigations, 73 Fed.Reg. 74,930-31 (Dep’t Commerce Dec. 10, 2008), a series of cases later invalidated the withdrawal, see Mid Continent Nail Corp. v. United States, 38 CIT-,-, 999 F.Supp.2d 1307, 1321-23 (2014); Gold East Paper (Jiangsu) Co. v. United States, 37 CIT -, -, 918 F.Supp.2d 1317, 1327-28 (2013). Consequently, Apex argues that Commerce had to confine A-T to targeted sales below, because the agency adopted its investigation procedures in reviews at a time when the Limiting Rule still controlled in investigations. See Pis.’ Br. 31-32; Reply Br. 13-14.

The court rejects these arguments. By its own terms, the Limiting Rule applies only to investigations. See 19 C.F.R. § 351.414(f)(l)-(2) (2008) (stating agency must limit scope of A-T remedy under paragraph (f)(1), which permits A-T in “antidumping investigation[s]”). “Commerce, like other agencies, must follow its own regulations,” Torrington Co. v. United States, 82 F.3d 1039, 1049 (Fed. Cir.1996), but it need not extend those regulations beyond their terms.

Furthermore, although the Limiting Rule technically applied to investigations in 2012, Commerce did not believe that the Rule bound its conduct when it grafted its investigation procedures into reviews. The initial decision invalidating the Limiting Rule’s withdrawal, Gold East Paper, was not decided until June 17, 2013, over a year after Commerce published the Final Modification. 37 CIT at -, 918 F.Supp.2d at 1317;- Final Mod. at 8101. And Apex has identified not one instance between 2012 and the Final Results when Commerce limited A-T to targeted sales in a review. See Pis.’ Br. 29-34. Because it had never before recognized the Limiting Rule in reviews, Commerce was not bound to apply the Rule as a matter of agency practice. See Ranchers-Cattlemen, 23 CIT at 884-85, 74 F.Supp.2d at 1374.

This holding renders moot Apex’s citation to Chang Chun Petrochemical Co. v. United States, 37 CIT -, -, 906 F.Supp.2d 1369, 1378-81 (2013). In Chang Chun, Commerce refused to follow the Limiting Rule in an investigation, even though the agency was bound by regulation to observe it. The court held that Commerce could not apply A-T to all sales unless it explained why plaintiffs case justified an exception from the Rule. Id. at -, 906 F.Supp.2d at 1380-81. In a similar vein, Apex argues that Commerce failed to justify applying A-T universally, for instance, by showing that segregating Apex’s targeted and untargeted sales would be unduly burdensome. See Pis.’ Br. 32-33; see also Chang Chun Petrochem. Co. v. United States, 37 CIT-, -, 953 F.Supp.2d 1300, 1305 (2013) (approving limitation of A-T to targeted sales where Commerce found it could segregate targeted and untargeted sales). Yet here, the Limiting Rule never controlled Commerce’s conduct in reviews. Commerce had no need to explain why Apex’s case merited an exception from the Limiting Rule, because the agency was not bound to observe the Rule in the first place.

IV. Commerce’s Explanation of Zeroing Accorded with Law

Finally, Plaintiffs allege that Commerce failed to justify its approach to zeroing when calculating Apex’s rate. As the court discussed previously, Commerce does not use zeroing to aggregate margins under A-A, the default method in reviews. The agency does use zeroing, however, when applying the A-T method to remedy targeting. Plaintiffs allege that Commerce had to explain why it interprets the statute to allow zeroing when targeted dumping is a concern, but not otherwise. In their view, Commerce’s explanation on the record inadequately addressed the inconsistency. See Pis.’ Br. 34-39.⁸

The agency’s rationale for its zeroing policy withstands this assault. As mentioned before, the statute does not say whether Commerce must zero or provide offsets at the aggregation step of the anti-dumping formula. Accordingly, in its discretion, Commerce has long zeroed when using A-T, and with the Federal Circuit’s blessing. See Timken, 354 F.3d at 1342. Commerce has also provided offsets instead of zeroing when using A-A, and again with the Federal Circuit’s approval. See U.S. Steel, 621 F.3d at 1361-62. Nevertheless, the agency cannot apply zeroing inconsistently without justification. In Dongbu Steel Co. v. United States, 635 F.3d 1363, 1371-73 (Fed.Cir.2011), the Federal Circuit found that Commerce behaved arbitrarily when it failed to explain why it used A-T with zeroing in reviews and A-A with offsets in investigations. The appellate court reached the same result in JTEKT Corp. v. United States, 642 F.3d 1378, 1384-85 (Fed.Cir.2011).- Both cases remanded so Commerce could provide the explanation required, if such an explanation were possible.

In Union Steel, 713 F.3d at 1101, Commerce provided the justification the Federal Circuit sought. Union Steel, like its predecessors, confronted whether, the agency adequately defended its decision to use zeroing in reviews but not investigations. Commerce gave three reasons why it interpreted the statute’s silence differently in each proceeding. First, Commerce explained that it zeroed in reviews but not investigations because the two proceedings served different purposes. Investigations assess whether to impose anti-dumping orders in the first place, thus justifying the use of broad averages and offsets under A-A; reviews, by contrast, set “the final rates to be used for actual assessment,” requiring greater accuracy and necessitating individual dumping margins and zeroing. Union Steel, 36 CIT at -, 823 F.Supp.2d at 1359. Second, Commerce explained that A-A and A-T average normal values and export prices differently, justifying zeroing under A-T but not A-A. Union Steel, 713 F.3d at 1108-09. And last, Commerce offered that it had abandoned zeroing in investigations, but not reviews, pursuant to WTO law. Id. at 1109-10. The Federal Circuit found these reasons sufficient to sustain the agency’s proceeding-specific approach to zeroing. Id. at 1109-10.

A three-part rationale similar to that in Union Steel explains why Commerce zeroes to remedy targeting, but not otherwise. In the I & D Memo, the agency said that it uses zeroing inconsistently to combat different types of dumping. See I & D Memo at 13. In general, Commerce relies on A-A without zeroing in reviews to remedy ordinary dumping. See Final Mod. at 8104. But when exporters engage in targeted dumping, A-A without zeroing may mask less-than-fair-value sales by averaging export prices at the comparison step and offsetting at the aggregation step. Commerce expanded on this concept when discussing why it applied A-T to all of Apex’s sales at the remedy phase:

The [Federal Circuit] has explained that “masked” dumping occurs ... when “profitable sales serve to ‘mask’ sales at less than fair value.” An Indian exporter, who competes with U.S. producers, could gain U.S. customers either by dumping to all customers at once or by dumping to a specific customer (or customers). In the latter scenario, the Indian exporter uses “profitable” (ie., non-dumped) sales to mask its dumped sales to a particular customer by compensating for its dumped sales to one customer with its profitable sales to other customers.

I & D Memo at 19; see also Def.’s Resp. in Opp. to Pis.’ R. 56.2 Mot. for J. on Agency R. 31-32, ECF No. 49. In other words, targeted dumping poses a challenge that ordinary dumping does not: It disproportionately impacts specific segments of the market (i.e., regions, customers, or time periods), but hides the effect with above-fair-value sales to other segments. To remedy this problem, Commerce applies A-T with zeroing instead of A-A, which exposes “any implicit masking ... by basing the comparison [of normal values and export prices] on the transaction-specific U.S. sales price rather than the weighted-average U.S. sales price.” I & D Memo at 19. A-T also reveals “any explicit masking between individual comparison results by not providing offsets for negative comparison results.” Id.; see also U.S. Steel 621 F.3d at 1363 (recognizing use of zeroing for “alleviating concerns of targeted or masked dumping”). In light of the differences between targeted and ordinary dumping, the court finds it was reasonable to deploy zeroing to address one type of dumping but not the other.

Commerce did not rest its explanation on these grounds alone, however. It also noted that differences inherent in the A-T and A-A methods justify zeroing under one method but not the other. See I & D Memo at 13. As noted in Union Steel, the statute requires Commerce to compare average export prices to average normal values when creating dumping margins under A-A. By averaging export prices, A-A implicitly offsets low export prices against high export prices. Thus it is also appropriate to offset positive and negative values when combining A-A margins into a single rate. Union Steel 713 F.3d at 1108-09. Under A-T, by contrast, the agency creates margins by comparing average normal values to individual transaction prices. This process yields a dumping margin for each sale of subject merchandise during a review period. By refusing to offset positive and negative values when aggregating margins, A-T ensures a remedy for each dumped transaction. See id. In sum, the agency’s approach to zeroing under A-A and A-T is reasonable because “the comparison methodologies compute dumping margins in different ways and are used for different reasons.” Id. ■ at 1109.

Third and finally, Commerce justified its approach to zeroing as a prudent response to WTO rulings. I & D Memo at 13-14. In the 2012 Final Modification, the agency abandoned A-T with zeroing as its default method in reviews-in favor of A-A without zeroing. Commerce made the change pursuant to several WTO Appellate Body decisions, which found that zeroing as a default in reviews was inconsistent with international law. See Final Mod. at 8101 n. 4. Yet these decisions did not outlaw zeroing entirely. Although it condemned zeroing as a default approach in review proceedings, the WTO did not forbid zeroing for use in exceptional circumstances. See id. at 8106-07. As a result, Commerce chose to continue zeroing on a case-by-case basis, and as needed to remedy targeted dumping. I & D Memo at 13-14. The court finds it was reasonable, and within Commerce’s discretion, to discontinue zeroing only to the extent required by WTO law. See Union Steel, 713 F.3d at 1110.

CONCLUSION

After carefully considering the parties’ briefs and the administrative record, the court sustains. Commerce’s Final Results in all respects. The court will enter judgment accordingly.

1 . As listed in the Final Modification, these decisions include Appellate Body Report, United States — Laws, Regulations and Methodology for Calculating Dumping Margins ("Zeroing”), WT/DS294/AB/R (May 14, 2009); Appellate Body Report, United States — Continued Existence and Application of Zeroing Methodology, WT/DS350/AB/R (Feb. 4, 2009); Appellate Body Report, United States — Final Anti-Dumping Measures on Stainless Steel from Mexico, WT/DS344/AB/R (Apr. 30, 2008); and Appellate Body Report, United States — Measures Related to Zeroing and Sunset Reviews, WT/DS322/AB/R (Jan. 9, 2007). See Final Mod. at 8101 n. 4.

2 . Commerce rarely uses the transaction-to-transaction or "T-T" method to calculate an-tidumping rates. Final Mod. at 8102. The agency did not appear to consider whether TT could not account for Apex’s targeted sales in this case, but Apex does not contest this omission.

3 . Apex argues that Commerce's measure of the “meaningful difference” between the rates was arbitrary. The court considers and rejects this argument in the next section. See infra Discussion II.B.

4 . None of the traditional exceptions to the exhaustion requirement apply here. See Ta Chen Stainless Steel Pipe, Ltd. v. United States, 28 CIT 627, 645 n. 18, 342 F.Supp.2d 1191, 1206 n. 18 (2004) (listing exceptions).

5 .. The court addresses the merits of this claim below. See infra Discussion III.

6 . Perhaps it would be arbitrary if Commerce found a "meaningful difference” between a 0.45 percent A-A rate and a 0.55 percent A-T rale. See Pis.’ Br. 23. But as the Government correctly notes, this illustration does not mean the analysis below was similarly absurd. Def.'s Resp. in Opp. to Pis.’ R. 56.2 Mot. for J. on Agency R. 28-29, ECF No. 49. In this case, the difference between the A-A and A-T rates was 3.49 percentage points, a disparity nearly thirty-five times greater than Apex’s hypothesized 0.1 percentage point difference. It was not arbitrary, given this difference between rates, to conclude that A-A could not account for targeting.

7 . Apex notes that Commerce now applies A-T uniformly only if targeted sales comprise more than a certain fraction of the total value of an exporter's sales. See Pis. Br. 33-34 (citing Xanthan Gum from the People’s Republic of China, 78 Fed.Reg. 33,351 (Dep't Commerce June 4, 2013) (final admin, review) and accompanying I & D Memo at cmt. 3). This change of course does not invalidate the policy disputed here, however. If an - agency practice reasonably complies with statute, that practice is not rendered invalid simply because the agency replaces it with an equally defensible policy. Cf. Nat’l Cable & Telecomm. Ass’n v. Brand X Internet Servs., 545 U.S. 967, 981, 125 S.Ct. 2688, 162 L.Ed.2d 820 (2005) (holding new policy not invalid simply because policy is inconsistent with past practice).

Apex also contends that it was "unduly punitive” to apply A-T to all sales "regardless of the amount of targeting that may have occurred.” Pis.' Br. 30; see also Antidumping Duties; Countervailing Duties, 61 Fed.Reg. 7308, 7350 (Dep’t Commerce Feb. 27, 1996) (req. for rants.). Yet Plaintiffs did not explain why the ratio of targeted to untargeted sales mattered until the reply brief, where Apex first argued that its A-T rate reflected mainly untargeted sales. See Reply Br. 9-10. Because Apex neither exhausted this argument below nor raised it properly on appeal, the court does not consider it. See supra Discussion II.A.

8 . This appears to be an issue of first impression. Though courts have addressed whether Commerce may zero inconsistently in investigations and reviews, none have decided whether the agency may apply zeroing disparately in a single proceeding. See Beijing Tianhai, 38 CIT at-, 7 F.Supp.3d at 1337 (reserving decision on zeroing in TD context); Mid Continent Nail, 38 CIT at -, 999 F.Supp.2d at 1320, 1323-24 (same).