Xi'an Metals & Minerals Import & Export Co. v. United States

256 F. Supp. 3d 1346, 2017 CIT 120, 2017 Ct. Intl. Trade LEXIS 121

• Court: United States Court of International Trade
• Decided: September 6, 2017
• Docket: Consol. 15-00109
• Status: Published

Opinion

Opinion & Order

AQUILINO, Senior Judge:

At bar are consolidated complaints invoking 19 U.S.C. §§ 1516a(a)(2)(A)(i)(I) and (B)(iii) and 28 U.S.C. § 1581(c) jurisdiction over the final results of the fifth administrative review (“AR5”) of its anti-dumping-duty order covering certain steel nails from the People’s Republic of China (“PRC”) published by the U.S. Department of Commerce, International Trade Administration (“ITA”) sub nom. Certain Steel Nails from the PRC, 80 Fed.Reg. 18816 (April 8, 2015), PDoc 294. See accompanying final issues and decision memorandum (“IDM”), PDoc 276, covering the period of August 1, 2012 through July 31, 2013.

Moving for judgment on the resultant administrative record of AR5, plaintiff Xi’an Metals & Minerals Import & Export Co., Ltd. raises four issues: (1) the suitability of Thailand as the primary surrogate country,. (2) valuation of its brokerage/handling (“B & H”) and freight costs, (3) adjustment of the weight denominator used in calculating its inland freight and B & H costs, and (4) double counting of SG & A (selling, general, and administrative) labor expenses in the labor rate used.

Also moving for judgment pursuant to USCIT Rule 56.2, consolidated-plaintiffs The Stanley Works (Langfang) Fastening Systems Co., Ltd. and Stanley Black & Decker, Inc. press one minor issue and a much broader matter for relief: (5) correction of a “transcription error” in their factors-of-production (“FOP”) database and (6) various challenges to ITA’s “differential pricing” analysis.

Judicial review of AR5 is governed by the applicable law and by the substantial evidence of record, which has long been defined as “such relevant evidence as a reasonable mind might accept as adequate to support a conclusion.” Consol. Edison Co. v. NLRB, 305 U.S. 197, 229, 59 S.Ct. 206, 83 L.Ed. 126 (1938). See 19 U.S.C. § 1516a(b)(l)(B)(i).

I

The antidumping-duty statute requires the ITA to seek surrogate values (“SVs”) for the factors of production for subject merchandise produced in or exported from a non-market economy (“NME”) country. 19 U.S.C.,§ 1677b(c)(1). The agency selected Xi’an Metals and the Stanley firms as AR5’s mandatory respondents. It sent an-tidumping questionnaires to them, to which they responded in a timely manner. ITA circulated a letter to interested parties inviting comments on surrogate country selection and SV data, to which it received comments and rebuttal comments. It thereafter issued supplemental questionnaires to which Xi’an Metals and Stanley also timely responded.

ITA published notice of the preliminary results of AR5 sub nom. Certain Steel Nails from the People’s Republic of China, 79 Fed.Reg. 58744 (Sept. 30, 2014), PDoc 304. See accompanying preliminary decision- memorandum (“PPM”), PDoc 224. Employing its differential pricing analysis, the agency preliminarily calculated a weighted-average dumping margin of 6.69 percent for Stanley and 72.40 percent for Xi’aii Metals. As part of its analysis, ITA concluded that there was a pattern of export prices for comparable merchandise that differed significantly among purchasers, regions, or time periods. See id. at 17-18. For Stanley, it found that the average-to-average (“A-A”) methodology did not appropriately account for such differences and applied the average-to-transaction (“A-T”) methodology to some Stanley U.S. sales and applied A-A to its other United States sales (reflecting a “mixed” alternative methodology). See id. For Xi’an Metals, ITA concluded that the A-A methodology appropriately accounted for such differences and applied it to calculate that firm’s weighted-average dumping margin. See id. at' 18. The agency also selected Thailand as the primary surrogate country for FOP valuation and surrogate financial ratios in constructing normal value. See PDoc 226.

During the course of its verification of the Stanley United States sales database and FOP, ITA accepted minor corrections that were brought to its attention. In February 2015, the agency requested that Stanley submit new sales and FOP databases to reflect the corrections that were revealed during verification. PDoc 257. Stanley did so timely. Whereafter ITA disclosed to the parties its calculations for AR5. On April. 7, 2015, ITA l’eceived a ministerial error allegation from Stanley that urged the agency to correct a transcription error that Stanley had made in its revised FOP database. ITA declined to do so.

The AR5 final results were published the next day. Based on the differential pricing analysis and the use of Thai SV data, ITA calculated a weighted-average dumping margin of 13,19 percent for Stanley and 72.52 percent for Xi’an Metals. In those results, the agency used the consolidated customer code (field CCUSCODU) in the Stanley margin program after determining that the use. of individual customer codes (field CUSCODU) for the Preliminary Results had been erroneous. See IDM at 45-46. This correction altered the results of the differential pricing analysis, leading ITA to apply the A-T methodology to all of the Stanley U.S. sales.

II

For its AR5 final results, ITA continued to select Thailand as the primary surrogate country. Plaintiff' Xi’an argues the substantial evidence of record shows that that country is unsuitable as a surrogate in this case, that the Thai steel wire rod values are aberrant, and that either the Philippines or Ukraine is a superior primary surrogate country for valuing FOP.

A

Plaintiff Xi’an argues reports compiled by the U.S. Trade Representative in 2011, 2012 and .2013, the U.S. Department of Commerce, and FedEx International Resource Center all constitute substantial evidence, of record showing that Thai customs officials routinely manipulate the entered values of imported merchandise, that, such manipulation is pervasive across all sectors, and that therefore the Thai import data are tainted-. Plaintiff Xi’an further argues that the average Thai import price for steel wire rod (“SWR”) during the POR of $916 per metric ton is not only the highest SWR price of record but exceeds “by far” the benchmarks it provided therefor. Xi’an’s benchmarks included SWR data from the World Bank Global Economic Monitor (“GEM”), world steel prices published by MEPS (International) Ltd., MEPS Asian Market SWR prices, official Thai domestic steel prices, SWR prices for Thai domestic’and export sales from TATA Steel, “UN Comtrade” (i.e., United Nations International Trade Statistics Database) import prices for other countries at a comparable level of economic development as the PRC (including the Philippines and Ukraine), and world market prices published by Asian Metal and Metal Expert.

The AR5 final results explain that, in order to value an input accurately, ITA examines all relevant price information on the record, including any appropriate benchmark data; that in any given case the agency’s current practice is to examine available import data for potential surrogate countries and/or data from the same HTS category for the surrogate country over multiple years to determine if the current data appear aberrational compared to historical values; and that the existence of higher prices alone is not a sufficient basis for concluding that the price data for a particular SV are distorted or misrepre-sentative. On the record for AR5, ITA concluded that none of the datasets suggested by Xi’an Metals serve as reliable benchmark data to determine whether Thai wire rod import data are aberrational 1 , and that Xi-an Metals’ HTS data analysis, submitted to support concluding that the Thai import data for SWR are distorted and should be disregarded because they are higher than export prices, does not permit “an appropriate comparison in order to determine if the data [are] aberrational” because Xi’an’s analysis is at the six-digit HTS level and “does not include any of the 11-digit HTS categories used to value wire rod at the Preliminary Results”. IDM at 16.

Plaintiff Xi’an contends defendant’s reasoning conflates the use of such benchmarks to evaluate the suitability of the average Thai import price with using such benchmarks as SVs in their own right; that the defensive responses 2 of it and the intervenor-defendant do nothing to dispel such “serious deficiencies” in ITA’s choice of Thailand as the primary surrogate country; that because the defendant and ITA acknowledge that Thai customs officials arbitrarily increase some import values the record evidence provides reason to believe or suspect that the import values of the inputs used as SVs for Xi-an Metals’ inputs were manipulated; and that defendant’s claim that the agency “believe or suspect” analysis “hinges on specific and objective evidence on which [ITA] would rely in determining that a country’s surrogate value data were unreliable” is not supported in practice.

Assuming the correctness of its foregoing position, plaintiff Xi’an argues that either the Philippines or Ukraine is superior to Thailand as a primary surrogate country in this case since the data for neither are tainted by manipulation of entered values for imported merchandise. The plaintiff contends the Ukrainian SWR prices from Metal Expert in particular are more specific than the Thai values as to the diameters of the SWR, and ITA has used that source for SWR in past reviews. And plaintiff Xi’an complains that the defendant does not back up its “fall back” argument that ITA “is not required to consider or give weight to any particular criteria in determining what constitutes the best available information on the record” for SVs with reference to substantial evidence when the agency emphasizes certain criteria and “completely ignores” other data quality criteria. XM Reply at 9, referencing Gerald Metals, Inc. v. United States, 132 F.3d 716, 720 (Fed.Cir. 1997) (substantial evidence standard “requires more than mere assertion of ‘evidence which in and of itself justified [the ... determination], without taking into account contradictory evidence or evidence from which conflicting inferences would be drawn’ ”, quoting Universal Camera Corp. v. NLRB, 340 U.S. 474, 487, 71 S.Ct. 456, 95 L.Ed. 456 (1951)).

The question, as always, is whether substantial evidence of record supports ITA determination(s). This court is unpersuaded herein that it does not, or that the agency has not considered all available evidence. Defendant’s logic is weak at points 3 , but ITA’s determination has substantial support on the record, and in toto, plaintiff Xi-an essentially asks for substitution of judgment for that of the agency, a request in conflict with the teaching of Consolo v. Fed. Mar. Comm’n, 383 U.S. 607, 619-20, 86 S.Ct. 1018, 16 L.Ed.2d 131 (1966). See, e.g., Matsushita Elec. Indus. Co. v. United States, 750 F.2d 927, 933 (Fed.Cir. 1984) (“the possibility of drawing two inconsistent conclusions from the same evidence does not prevent an administrative agency’s finding from being supported by substantial evidence”) (quoting same). Cf. Elkay Manufacturing Co. v. United States, 40 CIT -, -, 180 F.Supp.3d 1245, 1255 (2016) (“record evidence of manipulation of [Thai] customs values does not rise to such a level that [ITA] was left with no choice but to foreclose any use of Thai import data to determine [an SV] for a production input”), appeal filed, No. 16-2637 (Fed.Cir. Sept. 14, 2016).

B

Plaintiff Xi’an also argues that a military coup in Thailand should have triggered a “reason to believe or suspect” standard of pricing distortion in that country’s economy. It placed 43 pages of articles on the record detailing the massive political unrest and protests that rocked Thailand between 2011 and 2014, culminating in military-controlled government. The review period (“POR”) at issue in this appeal is August 1, 2012 through July 31, 2013. The plaintiff claims the materials covering Thailand’s political and economic turbulence attest that the 2011 elections were never accepted as legitimate, that the military coup was an undemocratic and complete takeover of the country, and that it was reasonable to conclude that a free-market economy could-not properly-function in the absence of the free flow of information or impartial rule of law. Plaintiff Xi’an argues the issue should be remanded for ITA to explain how the military coup does not meet the lenient “reason to believe or suspect” standard, because it is counter-intuitive, if not hypocritical, that- the agency would reject the PRC economy based on state control but then select the an alternative country under"' military dominance as the “free-market” surrogate for the PRC where “better” alternative countries were presented on the record for which no such distortions were alleged.

The burden is on the plaintiff, however, to provide for the record evidence to support its argument. The AR5 final results explain ITA’s

disagree[ment] with Xi’[a]n Metals’ argument that the Thai military coup renders Thai import data to be unrepresentative and unreliable. ... [I]t is the Department’s practice to focus on several criteria, including whether thé SV data are contemporaneous) publicly available, tax and duty exclusive, representative ' of a broad market average, and specific. Xi’[a]n Metals has neither provided any evidence on the record as to why the military coup affects'; the criteria considered by the Department nor how specific inputs are affeeted.

IDM at 13.

Plaintiff Xi’an’s arguments here do not persuade as to the incorrectness of ITA’s position on the' subject. See, e.g., NMB Singapore Ltd. v. United States, 557 F.3d 1316, 1319 (Fed.Cir. 2009) (ITA’s explanations need not be perfect, only “reasonably discernible”).

C

Plaintiff Xi’an also complains that an NME respondent has no “ability” to select the primary surrogate country. The defendant counters this is contrary to the statute because the surrogate country must be deemed “appropriate by the administering authority.” Xi’an replies that it could select the home market by' applying all of the statutory criteria, including economic comparability and significance of production, see 19 U.S.C. § 1677b(c)(l)(B), and that the defendant has no answer'to the problem of how an NME respondent can comply with the remedial nature of the anti-dumping laws if it has no way to estimate its costs when it sets the price for export to the United States. Plaintiff Xi’an states that it is simply pointing out that the NME respondent is severely disadvantaged vis-á-vis market economy respondents if it is not permitted to assert a surrogate country meeting all the criteria and host to reasonably reliable data for the valuation of its factors.

The court appreciates this concern and can concur that a rational producer would not chose the highest available steel costs when lower domestic, regional, and economically comparable sources are available 4 , but the argument is one that conflicts with what has long been the case: “It is [ITA], not the respondent, that determines what information is to be provided” for a particular proceeding. Ansaldo Componenti, S.p.A. v. United States, 10 CIT 28, 37, 628 F.Supp. 198, 205 (1986). Accord Essar Steel Ltd. v. United States, 34 CIT 1057, 1072-73, 721 F.Supp.2d 1285, 1296-99 (2010), aff'd in relevant part, 678 F.3d 1268 (Fed.Cir. 2012); NSK, Ltd. v. United States, 20 CIT 361, 367, 919 F.Supp. 442, 447 (1996); Ñachi-Fujikoshi Corp. v. Unit ed States, 19 CIT 914, 920, 890 F.Supp. 1106, 1111 (1995); Tianjin Mach. Import and Export Co. v. United States, 16 CIT 931, 936, 806 F.Supp. 1008, 1015 (1992); Chinsung Indus. Co. v. United States, 13 CIT 103, 705 F.Supp. 598 (1989); Timken Co. v. United States, 11 CIT 786, 804, 673 F.Supp. 495, 513 (1987); Smith-Corona Group Consumer Prods. Div., SCM Corp. v. United States, 713 F.2d 1568, 1577 n. 26 (Fed.Cir. 1983), cert. denied, 465 U.S. 1022, 104 S.Ct. 1274, 79 L.Ed.2d 679 (1984). As the final selection of an appropriate surrogate country occurs post-closing of the review period, it may be that such selection is not susceptible to the kind of predictability plaintiff Xi’an desires, but it is, nonetheless, susceptible to the burden of persuasion borne by interested parties.

D

Plaintiff Xi’an also asserts that ITA’s surrogate brokerage and handling (“B & H”) is unreliable and unreasonably high. It prays for remand consistent with Since Hardware (Guangzhou) Co. v. United States, 38 CIT -, 977 F.Supp.2d 1347 (2014), which barred the agency from relying on the hypothetical weight postured in Doing Business reports. The plaintiff argues that, for the denominator calculating B & H and inland freight, costs, ITA should use either the maximum cargo load for a 20-foot container or Xi’an Metals’ own average cargo load instead of the weight of 10,000 kilograms from the relevant Doing Business report.

The defendant responds that Xi’an Metals failed to exhaust this specific argument before ITA during the administrative process. It also responds that it would be inappropriate to use Xi’an Metals’ own average cargo load because the value must come from a selected surrogate country, and that substantial evidence supports the use of the 10,000 kilogram denominator in any event because the relationship between costs and quantity is maintained in reliance upon the Doing Business report and results in an accurate per-unit cost. See also Def-Int’s Resp. at 18-20.

Pursuant to 28 U.S.C. § 2637(d), the court “shall, where appropriate, require the exhaustion of administrative remedies” in civil actions arising from ITA’s antidumping — and countervailing-duty determinations. The doctrine of exhaustion is that “no one is entitled to judicial relief for a supposed or threatened injury until the prescribed administrative remedy has been exhausted.” Sandvik Steel Co. v. United States, 164 F.3d 596, 599 (Fed.Cir. 1998), quoting McKart v. United States, 395 U.S. 185, 193, 89 S.Ct. 1657, 23 L.Ed.2d 194 (1969). It is well-settled that “[a] reviewing court usurps the agency’s function .when it sets aside. an agency determination upon, a ground not theretofore presented and deprives the [agency] of an opportunity to consider the matter, make its ruling, and state, the reason for its action.” Unemployment Compensation Comm’n of Alaska v. Aragon, 329 U.S. 143, 155, 67 S.Ct. 245, 91 L.Ed. 136 (1946) (“UCCA”); accord Yantai Oriental Juice Co. v. United States, 27 CIT 1709, 1719 (2003). “Simple fairness to those who are engaged in the tasks of administration, and to litigants,' requires as a general rule that courts should not topple over administrative decisions unless the administrative body not only has erred but has erred against objection made at the time appropriate under its practice.” United States v. L.A. Tucker Truck Lines, Inc., 344 U.S. 33, 37, 73 S.Ct. 67, 97 L.Ed. 54 (1952). See also Metz v. United States, 466 F.3d 991, 999 (Fed.Cir. 2006).

Thus, a party must present all arguments to ITA at the time it is addressing an issue. E.g., Mittal Steel Point Lisas Ltd. v. United States, 548 F.3d 1375, 1383-84 (Fed.Cir. 2008). A party’s obligation to exhaust its administrative remedies applies equally to overall issues as well as to individual arguments. Rhone Poulenc, Inc. v. United States, 899 F.2d 1185, 1191 (Fed.Cir. 1990). By failing to raise this argument until now, Xi’an Metals deprived ITA of the “opportunity to consider the matter, make its ruling, and state the reason for its action.” UCCA, 329 U.S. at 155, 67 S.Ct. 245.

None of the limited exceptions to the exhaustion requirement apply here. See Corus Staal BV v. United States, 30 CIT 1040 n. 11, 1050, 2006 WL 2056401 n. 11 (2006) (identifying exceptions for pure legal questions, futility in raising argument at agency level, denial of access to confidential record, intervening judicial interpretation), aff'd, 502 F.3d 1370 (Fed.Cir. 2007). First, the pure question of law exception does not apply to arguments concerning the new factual analysis that plaintiff Xi’an now posits for the first time. See Mittal Steel, 548 F.3d at 1384 (pure question of law exception does not apply when the argument relies on unique facts of the case). ITA did not have the opportunity to analyze, in the first instance, Xi’an Metals’ contentions pertaining to the Doing Business report and its preference for instead using those B & H costs incurred by Thai exporters of frozen freshwater shrimp. Second, it would not have been futile for Xi’an Metals to present the analysis of the factual information to the agency during the underlying administrative proceeding. The futility exception to the exhaustion doctrine is narrow: parties must demonstrate that they “would be required to go through obviously useless motions in order to preserve their rights”, Corus Staal, 502 F.3d at 1379 (internal quotations and citations omitted), and plaintiff Xi’an’s argument does not satisfy that standard. Third, there has been no intervening judicial decision that might excuse the absence of Xi’an Metals’ argument at the administrative level. Fourth, plaintiff Xi’an does not allege any untimely access to the confidential record. Thus did it fail to exhaust.

E

Plaintiff Xi’an’s last claim is that ITA made two labor classification errors: (1) it included staff labor costs in the selling, general, and administrative (“SG & A”) expenses, reasoning that the respondents, did not report labor hours associated with the selling and administrative staff; and (2) from the financial statements of the Thai company L.S. Industries Co. (“LSI”) used for calculation of surrogate financial ratios 5 ITA accounted for various line items such as “welfare” and “social security and compensation” as SG & A-type labor costs despite the fact that the Thailand Nátional Statistics Office (“NSO”) statistics used to calculate labor SV includes such benefits in the reported labor rate. See IDM at 19-20. ITA decided that it would adhere to how the surrogate financial statements themselves classified these items. See id. at 20.

And yet, in the calculation of surrogate financial ratios, it is the agency’s practice to avoid double-counting labor costs that are included among SG & A by “adjusting] the surrogate financial ratios when the available record information — in the form of itemized indirect labor costs— demonstrates that labor costs are overstated.” Antidumping Methodologies in Proceedings Involving Non-Market Economies: Valuing the Factor of Production: Labor, 76 Fed.Reg. 36092, 36093-94 (June 21, 2011) (“Labor Methodologies”). Stated differently, ITA looks to the surrogate financial statements on the record, and if they “include disaggregated overhead and [SG & A] expense items that are already included in the [record data used to value labor], [it] will remove these identifiable costs items.” Id. at 36094. See, e.g., Certain Frozen Warmwater Shrimp From the Socialist Rep. of Vietnam, 67 Fed.Reg. 56158 (Sept. 12, 2011), and accompanying issues and decision memorandum (“I & D memo”) at cmt. 5.B.

The defendant maintains that ITA followed practice during the administrative proceeding by treating labor-related costs in its financial ratio calculations in the same manner that the surrogate company disaggregates labor costs, explaining that under ITA’s FOP methodology for calculating normal value, labor expenses capture the labor cost only for manufacturing, which is obtained by multiplying a respondent’s reported direct and indirect labor hours to manufacture subject merchandise by the surrogate labor rate (e.g., the Thai NSO labor rate). The defendant contends that the Thai NSO 2007 labor data, used to calculate the labor SV, were derived from an average remuneration paid for persons engaged in various “manufacturing and non-manufacturing activities” 6 and that, contrary to plaintiff Xi’an’s argument, it does not follow that the labor expenses calculated using the NSO labor rate capture all labor expenses. Further, the defendant contends, the respondents did not report labor hours associated with selling and administrative staff, as staff labor costs would normally be expected to be included among the SG & A expenses. Concluding, the defendant argues the SG & A labor expenses in each surrogate company’s financial statement should therefore be included in the numerator of the SG & A ratio associated with that company, and therefore the-SG & A labor expenses listed in LSI’s financial statements should be classified under the SG & A expensés and included in the respective numerator of the SG & A ratio calculation, an outcome ITA ensured by including the “Salary and Bonus” line item from LSI’s “Total Cost of Management”- in the SG & A buildup in the financial ratio calculations. See Defs Resp. at 82-83, referencing IDM at 19-20. See also Def-Int’s Resp. at 20-22.

Plaintiff Xi’an counters that defendant’s (and intervenor-defendant’s similar) explanation merely restates ITA’s position from the IDM rather than confronting the facts and logic of their argument, which it contends amounts to a waiver of any surrogate labor cost defense 7 ; and. that the Thai NSO 2007 labor data do indeed cover “all” labor expenses, including overtime, benefits, vacation pay, and the range of executive, administrative, and production labor, regardless of the exactitude of “various manufacturing and non-manufacturing activities”. XM Reply at 19, referencing Pet’s SV Submission at Ex. 9, PDocs 158-160. Plaintiff Xi’an further argues, it does not follow from the fact that respondents are not required to report hours of administrative or non-production labor (see NME questionnaire) that those costs have not been counted, and also that it is indisputable that the labor rate ITA now relies upon pursuant to Labor Methodologies is an inflated- rate intended to account for those very expenses.

It is apparent from the IDM at page 19 that ITA’s reasoning was informed by Elkay Manufacturing Co. v. United States, 38 CIT -, -, 34 F.Supp.3d 1369, 1375-84 (2014) (rejecting argument that the NSO labor rate “failed to capture any SG & A labor costs”, but also rejecting conclusion that “double-counting” of SG & A labor expenses required the specific downward adjustments made in that case, i.e., the record “lack[ed] substantial evidence to support [ITAJ’s conclusion that the rate [it] applied to the hours of production labor reported by the investigated respondents overstated the value of those labor hours to such an extent as to justify the specific, compensatory adjustments ... made to the SG & A/interest expense ratios”). See also Elkay Manufacturing Co., supra, 40 CIT at -, 180 F.Supp.3d at 1257-59 (sustaining ITA’s revised decision not to remove identifiable SG & A labor items from its calculated SG & A expense ratios). The problem here, however, appears similar to that which was recently considered in Yingqing v. United States 8 . But, the source and labor rate ITA has deliberately chosen pursuant to Labor Methodologies apparently includes all types and forms of labor as well as labor benefits, and, in that announcement of new methodology, the agency recognized that it would be .over-counting the labor rate for production labor and specifically indicated therein that. the financial ratios would have to be adjusted so labor was, not double-counted; the implicit remedy would be to move all labor costs explicitly incorporated in the SG & A source and rate chosen to the ratio denominators.

In this case, by' not removing the various line items such as “welfare” and “social security and compensation” that are presumptively included already in the Thai NSO rate, the SV for labor is inflated, which requires correction initially via the court’s grant of the pertinent part of plaintiff Xi’an’s motion for agency reconsideration. On remand therefor, if ITA continues to select a source and rate that includes all labor positions and benefits, it needs to ensure that all forms of labor costs on the financial statements are in the “materials-labor-energy” (or “MLE”) denominator of the ratios in accordance with its Labor Methodologies, but whatever course it chooses will need to obviate the double counting 9 that is manifest in the AR5 final results.

III

The first claim of the Stanley plaintiffs’ Rule 56.2 motion is that ITA arbitrarily refused to correct a transcription error in their February 17, 2015 post-verification FOP database (specifically the omission of a zero in the tenth or one-hundredth decimal place in- field “V-DLCROD”), which they claim resulted in the FOP for low-carbon SWR being overstated by almost nine percent, and directly resulted in an erroneous increase in their dumping margin of 3.09 percentage points — about 30 percent higher than it would have been absent the error. ITA had directed them to submit the post-verification database to implement minor FOP corrections that it had accepted at verification, including corrections to the variance rate for SWR. PDoc 257. The purpose of the revised database was thus to ensure that the AR5 final results would be based on verified data, and the minor corrections should have reduced the Stanley dumping margin from the Preliminary Results.

The Stanley plaintiffs contend ITA did not issue its request for the revised FOP database until eight weeks after verification, and the agency initially afforded only two days in which to prepare and submit the revised database,- a deadline subsequently extended over a President’s Day weekend, which relatively short deadline “certainly contributed to Stanley’s computer programmer inadvertently omitting a zero to the right of the decimal point in the field for concerning low-carbon SWR.” Stanley Reply at 3. The consolidated-plaintiffs further explain that it was. not possible to have identified the error in their administrative case brief because the revised database was submitted on the same day as that brief. Id. at 4.

Whatever the excuse, the error occurred, and it.is manifest. ITA’s ministerial error memorandum, PDoc 297, and the defendant imply Stanley had an opportunity to bring the error to ITA’s attention in the time period between the case brief and the AR5 final results, to which the Stanley reply is that the timely submission of a ministerial error allegation is the only available procedure for correcting a clerical error in a submission made concurrently with a case brief. See 19 U.S.C. § 1675(h) and 19 C.F.R. § 351.224(e) (contemplating that final results are only “final” subject to correction of ministerial errors). Stanley did so. CDoc 327. But ITA rejected the ministerial error allegation by stating that, generally, “ministerial errors include only those errors that are produced by the Department. The Department will only correct a respondent’s error when that error is ‘so egregious and so obvious’ that failing to correct the error would be arbitrary and capricious,” PDoc 297 at 4. ITA then concluded the error was “neither so égrégious nor so obvious as to be characterized as a ministerial error.” Id.

However, in light of the Stanley presentment,'it is difficult to fathom how their ministerial error could have been concluded otherwise, especially given its impact on their overall dumping margin (a 43.5 percent change from the Preliminary Results). In short, ITA must be ordered on remand to make the correction. See, e.g., NTN Bearing Corp. v. United States, 74 F.3d 1204, 1208 (Fed.Cir. 1995) (it is incumbent on ITA to correct such errors, as it has a “duty to determine dumping margins ‘as accurately as possible’ ”), quoting Rhone Poulenc, Inc. v United States, 899 F.2d 1185, 1191 (Fed.Cir. 1990). See also Brother Indus., Ltd. v. United States, 15 CIT 332, 341, 771 F.Supp. 374, 384 (1991) (“court-ordered amendments of ministerial errors are not destructive of the ITA’s ability to manage its proceedings”).

The remainder of the Stanley motion focuses on ITA’s targeted dumping analysis of its sales, i.e, by “purchasers, regions, or periods of time.” 19 U.S.C. § 1677f-1(d)(1)(B). See, e.g., Mid Continent Nail Corp. v. United States, 38 CIT -, n. 3, 999 F.Supp.2d 1307, 1311 n. 3 (2014).

Section 1677f-1(d) of Title 19, U.S.C. directs “in general” that ITA “shall” calculate dumping margins using the A-A or transaction-to-transaction (“T-T”) price comparison methods, see id., subsection (1)(A), but where the record establishes the existence of a pattern of export prices that differ significantly among customers, regions, or time periods and why such differences cannot be accounted for using the A-A method is explained, ITA “may” calculate dumping margins using a different methodology such as the A-T method. See 19 U.S.C. § 1677f-1(d)(1)(B). When ITA uses that method, it reverts to “zeroing” 10 but does not ignore non-dumped sales when it uses the A-A method. The Statement of Administrative Action (“SAA”) accompanying the Uruguay Round Agreements Act explains that Congress intended “targeted dumping” to comprise “situations [in which] an exporter may sell at a dumped price to particular customers or regions, while selling at higher prices to other customers or regions.” The SAA explicitly links ITA’s use of the A-T method to “targeted dumping”:

New Section 777A(d)(l)(B) provides for a comparison of average normal values to individuals export prices ... in situations where an average-to-average or transaction-to-transaction methodology cannot account for a pattern of prices that differ significantly among purchasers, regions, or time periods) i.e., where targeted dumping may be occurring.

SAA at 843.

Consistent with the SAA, ITA promulgated a targeted dumping regulation, 19 C.F.R. § 351.414(f). See Antidumping Duties; Countervailing Duties, 62 Fed. Reg. 27296, 27373-76 (May 19, 1997). Its salient elements are:

1. Targeted dumping must be determined through the use of “standard and appropriate statistical techniques.”

2. The A-T comparison is used only for those specific sales that comprise targeted dumping.

3. “Normally,” targeted dumping will be pursued only in response to an allegation by a petitioner that includes supporting factual information and an explanation as to why the A-T comparison could not take into account any alleged price differences.

The current 11 test of targeted dumping, differential pricing, purports to examine differences in a respondent’s prices among individual purchasers, geographic regions, and quarterly time periods. It is performed at the level of individual product control numbers (CONNUMs) and net of adjustments to gross U.S. selling price. ITA does not require any allegation or factual support in differential pricing; rather, ITA now perfonns differential pricing by rote in every proceeding.

ITA analyzes prices for each CONNUM by dividing them into a series of “test groups” (each comprising prices to a specific purchaser, region, or calendar quarter) and “base groups” (comprising the remaining purchasers, regions, or calendar quarters). Prices to every purchaser, region, and calendar quarter are serially analyzed as a test group and then recycled into the base group of prices for that CONNUM. Differential pricing then entails three elements.

In the first element, ITA employs “Cohen’s d statistic” to measure the “effect size” between each test group and its relevant base group. The agency describes effect size as a descriptive measure of the “magnitude” of the difference between two groups, which the Stanley plaintiffs contend infra is gross oversimplification.

ITA calculates the Cohen d statistic as the difference between the weighted average net prices of the test and base groups divided by the “pooled” standard deviation of the net prices of the two groups. The pooled standard deviation is calculated as the square root of the sum of the square of the base group’s standard deviation plus the square of the test group’s standard deviation, divided by two. The resulting coefficients are labeled as “small”, “medium”, or “large”.

Notably, ITA ignores whether a test group’s weighted-average price is higher or lower than the base group’s weighted-average price. A “large” Cohen’s d coefficient is 0.8 or greater, which means that the weighted-averages of the base group and the test group differ by 0.8 standard deviations. The agency deems all sales that meet or exceed the 0.8 Cohen d coefficient to have “passed” that threshold, thereby satisfying the statute’s requirement that “significant” price differences exist as a precondition to using the A-T method.

In the second element, called the “ratio” test, ITA stratifies the percentage of a respondent’s sales that “pass” the Cohen d test. If the value of a respondent’s passing sales account for 66 percent or more of the value of its total sales, then the agency uses the A-T method with zeroing for all sales. If the Cohen d test “pass” rate is 33 percent or less, then ITA uses the A-A method for all sales. If the Cohen d test “pass” rate falls between 33 percent and 66 percent, then the agency uses the A-T method with zeroing for sales that “pass” the Cohen d test and the A-A method for the remaining sales. ITA deems this stratification, of CDT “pass” rates to establish whether a “pattern” of significant price differences exists.

In the third element, culled the “meaningful, difference” test, ITA calculates, the respondent’s dumping margin in three ways. First, it uses the A-A method for all sales. Second, it uses a “mixed” method in which the A-T method with zeroing is applied only to sales that have “passed” the Cohen d test while the A-A method is applied to the remaining sales. Third, ITA applies the A-T method with zeroing to all sales. Depending on the results of the “ratio” test, the margin'resulting from'either the second or third method is compared to the margin resulting from the A-A method for all sales. The agency deems a “meaningful difference” to exist between thé two calculations if the margin using-the A-T (or “mixed”) method (1) generates a 25 percent relative change in the dumping margin compared to the A-A method, or (2) generates a dumping margin that crosses the de minimis threshold when compared to the A-A method, ITA deems the existence of a “meaningful difference” sufficient to explain why it cannot account for a pattern of significant price differences using the A-A method.

As an initial matter, the Stanley plaintiffs raise again the issue of ITA’s “abrupt” withdrawal of its 1997 targeted dumping regulation pursuant to Withdrawal of the Regulatory Provisions Governing Targeted Dumping in Antidumping Duty Investigations, 73 Fed.Reg, 74930 (Dec. 10, 2008). See 19 C.F.R. § 351.414(f) (2007).

Mid Continent Nail Corp. v. United States, 846 F.3d 1364 (Fed.Cir. 2017), indeed held that withdrawal to have been unlawful and not harmless in accordance with the Administrative Procedures Act. Defendant’s explanation is that, whereas the statute places certain restrictions on ITA selection of a comparison methodology for purposes of investigations, it does not do so for purposes of administrative reviews such as the one at bar. Def s Resp. at 23-24, referencing 19 U.S.C. § 1677f-1(d)(1)(B), SAA at 842-43. JBF RAK LLC v. United States, 790 F.3d 1358, 1364 (Fed. Cir. 2015), held that to be true, and cases since have consistently deferred to that interpretation. E.g., Fine Furniture (Shanghai) Ltd. v. United States, 40 CIT -, — , 182 F.Supp.3d 1350, 1364 (2016); Nan Ya Plastics Corp., Ltd. v. United States, 39 CIT -, -n. 3, 128 F.Supp.3d 1345, 1349 n. 3 (2015); Apex Frozen Foods Private Ltd. v. United States, 38 CIT -, -, 37 F.Supp.3d 1286, 1293 (2014), aff'd, 862 F.3d 1322 (Fed.Cir. 2017); CP Kelco Oy v. United States, 38 CIT -, -, 978 F.Supp.2d 1316, 1320 (2014); Timken Co. v. United States, 38 CIT -, - & n. 7, 968 F.Supp.2d 1279, 1286 & n. 7 (2014). Further, although ITA typically cites to 19 U.S.C. § 1677f~l(d)(l)(B) for “guidance”, it does not consider that provision “binding” legal authority since its statutory authority to select a comparison methodology in reviews is derived from a different provision, 19 U.S.C. § 1677f-l(d)(2), which does not place restrictions on ITA’s choice of comparison methodology. As such, the agency has discretion 12 to apply A-T methodology in reviews notwithstanding the circumstances surrounding the withdrawal of the pre-2008 targeted dumping regulation.

The defendant contends ITA properly-applied A-T methodology during AR5. It found that the value of Stanley sales passing the Cohen d test accounted for more than 66 percent of the value of total Stanley United States sales and also found a meaningful difference between the weighted-average dumping margins calculated using the A-A methodology and an alternative comparison methodology based on the A-T method. See Stanley Final Results Analysis Memo at 3. See also IDM at 36. Specifically, when comparing the Stanley weighted-average dumping margin calculated pursuant to the A-A method and an alternative comparison method based on the A-T method, that margin rose above the de minimis threshold. Such a difference in the weighted-average dumping margins has been held to satisfy the statutory requirement that ITA explain why the A-A method cannot account for such differences. See Apex Frozen Foods, supra, 38 CIT at -, 37 F.Supp.3d at 1295-96. Cf. Golden Dragon Precise Copper Tube Grp., Inc. v. United States, 39 CIT -, Slip Op. 15-89 (2015) at 15-16, 2015 WL 4927515 at *8 (“the significance of the ‘effect size’ ... in and of itself ‘explains why such differences cannot .be taken into account’ using A-A methodology”).

Be that as it may, it misses the Stanley point that the regulation expressly limits the A-T methodology “to those sales that constitute targeted dumping”, and it is this “limiting rule” that Gold East Paper (Jiangsu) Co. v. United States, 37 CIT -, -, 918 F.Supp.2d 1317, 1327 (2013), and Mid Continent both held still in effect at the times in question. The AR5 final results violate this rule by applying the A-T methodology to all Stanley sales. Remand, for the purpose of properly applying it, is therefore necessary.

The Stanley plaintiffs argue that ITA’s use of the Cohen d test is unlawful because it (i) was allegedly designed for a context dissimilar to that being analyzed by the agency in a differential pricing analysis; (ii) is arbitrary in terms of its classification of effect sizes; (iii) is unreasonable when the entire data population is available; and (iv) fails to measure statistical significance.

(i)

THeir claim is that Dr. Cohen’s d test was created for'psychological research and used as a tool in the behavioral sciences and should not apply in a'matter like this. The defendant responds that ITÁ uses the test to analyze' a respondent’s pricing behavior, see IDM at 31, and that the economics of pricing behavior is, in fact, a subset within the ambit of behavioral science. It is an accepted statistical test, employed by ITA to discern a pricing pattern, and the Stanley position neither persuades that the agency’s use of it was unreasonable nor demonstrates unlawfulness thereof.

(ii)

The Stanley plaintiffs contend ITA’s classification method for the Cohen d test effect size is arbitrary. By way of background, after ITA determines Cohen’s d coefficient, it establishes a threshold to determine whether that is significant. See PPM at 16-17. The defendant explains that the agency adheres to the three different fixed thresholds Dr. Cohen deduced (small, medium, and large) because they allow ITA to determine the “significance” (or meaningfulness) of the differences between prices to a particular purchaser, region, or time period as well as the prices of comparable merchandise to all other purchasers, regions, or time periods in an efficient and predictable way, and are generally accepted thresholds for the d test. See id. at 34-35 (citing and quoting David Lane et ah, “Effect Size,” Section 2, “Difference Between Two Means” (stating that the guidelines suggested by Dr. Cohen as to what constitutes small, medium, and large effect size “have been widely adopted”)). ITA generally uses the “large” threshold (i.e. Cohen’s d coefficient above 0.8) as the threshold for passing the d test, because the “large” threshold provides the strongest support for the differences being meaningful. Id. at 36-37.

Substantial evidence of record herein supports the use of Dr. Cohen’s d test and the threshold demarcations he intuited, along with the caveats he enunciated, since the record evinces that his test gained awareness, acceptance, and use among scientists within various disciplines of the self-professed community of “experts”, see, e.g., id., and the Stanley arguments do little to contradict or counteract this fact. Therefore, because ITA used widely accepted thresholds, provided a rational explanation as to which threshold to employ, and selected a threshold for the Cohen d coefficient which has real world, practical meaning consistent with the statute, its use of the threshold is not arbitrary. Cf. Cosco Home & Office Prods. v. United States, 28 CIT 2043, 2049-50, 350 F.Supp.2d 1294, 1299-1300 (2004) (holding ITA’s interpretation of 19 U.S.C. § 1675(a)(1) and amendment of its regulations reasonable); Mitsubishi Heavy Indus., Ltd. v. United States, 21 CIT 1227, 1233-35, 986 F.Supp. 1428, 1434-35 (1997) (50 percent test).

(iii)

The Stanley plaintiffs challenge ITA’s application of the ratio test, claiming that it does not explain how the three thresholds thereof 13 satisfy the statutory requirements. ITA explained that it uses the ratio test to complete its determination as to whether there exists a pattern of prices that differ significantly by purchaser, region, or period of time. See PPM at 17. This is necessary because, even though the sales for one or more groups of comparable merchandise for specific purchasers, regions, or time periods may pass the Cohen d test, it does not necessarily follow that, in relation to the total volume of a respondent’s export sales, there is sufficient evidence that there exists a pattern of prices that differ significantly. See IDM at 37-38. Pursuant to 19 U.S.C. § 1677f-1(d)(1)(B), ITA “may determine” whether sales were made at less than fair value using the alternative method when subsections (i) and (ii) of the provision are satisfied, but the statute is silent as to how ITA may determine whether those subsections are thus and such. See id. The agency in this matter lawfully exercised its discretion in filling the gaps of determining how the A-T method could be considered as an alternative methodology. See id. See also JBF RAK, supra, 790 F.3d at 1364.

(iv)

The Stanley plaintiffs assert that the statute requires ITA to measure “sta tistical significance,” which the Cohen d test does not measure. This assertion underlies many of the Stanley arguments, but “statistical significance” is irrelevant where, as here, the agency has a complete set of data to consider.

The statute provides that ITA may apply an alternative comparison methodology if it finds “a pattern of export prices (or constructed export prices) for comparable merchandise that differ significantly among purchasers, regions, or periods of time[.]” 19 U.S.C. § 1677f-l(d)(l)(B)(i) (emphasis added). Additionally, the SAA states:

New Section 777A(d)(l)(B) provides for a comparison of average normal values to individual export prices ... in situations where an [A-A] or [T-T] methodology cannot account for a pattern of prices that differ significantly among purchasers, regions, or time periods, i.e., where targeted dumping may be occurring.

SAA at 843 (emphasis added). Neither the statute nor the SAA defines the term “significantly”, but the consolidated-plaintiffs contend its plain meaning is “statistically significant.” ITA interprets otherwise.

Statistical significance only takes on relevance when “determining] from a sample (i.e., the data at hand) of a larger population an estimate of what the actual values (e.g., the mean or variance) of the larger population may be”. IDM at 34. Here, ITA has the entire population of the respondents’ sales in the U.S. market; therefore, “ ‘statistical significance’ is not a relevant consideration.” Id. The agency calculates the Cohen d coefficients to determine whether differences in prices for comparable merchandise among purchasers, regions, or time periods are significant, and those calculations are based upon all of the United States sales that Stanley reported for the POR, not merely a sample, and thus form the entire population of U.S. sales of subject merchandise. See id. Sampling error does not exist when there are complete data for analysis.

The Stanley plaintiffs state that the purpose of the Cohen d test is to “make reasonable queries as to how big an intervention effect may be when only a sample is available.” It would be more accurate to state, however, that the Cohen d coefficient measure of effect size “quantifies the size of the difference between two groups, and may therefore be said to be a true measure of the significance of the difference” based on complete information, not samples. See id. at 33 (citations omitted). Accordingly, once again, the “statistical significance” of ITA’s calculations is not relevant to its analysis, and requiring the agency to measure statistical significance here, where it has incorporated all of the respondents’ data in the analysis, would be inappropriate 14 .

As noted above, Congress did not use the word “statistical” or any variation thereof when'it drafted the statute. And as ITA stated in the IDM, and as Stanley has argued elsewhere, Congress acts intentionally when it drafts statutory language. Simply put, if Congress had wanted ITA to measure “statistical significance,” it would have included the word “statistical” 15 . In applying the Cohen d test, ITA fulfills the statutory requirement to measure whether there exists a pattern of prices that differ “significantly”, and the d test enables the agency to quantify, in a simple and transparent approach, whether prices differ significantly among purchasers, regions, or time periods.

(v)

The Stanley plaintiffs claim that several other aspects of ITA’s differential pricing analysis contravene congressional intent. However, mere disagreement with its approach, where' the statute is silent, is not a sufficient basis for the court to overturn the agency’s reasoning. See Mid Continent Nail Corp. v. United States, 34 CIT 512, 519, 712 F.Supp.2d 1370, 1376-77 (2010) (“[generally, courts lack an ‘independent authority to tell the [agency] how to do its' job’ when a statute does not specify ‘any Congressionally mandated procedure or methodology for assessment of the statutory tests’”), quoting U.S. Steel Group v. United States, 96 F.3d 1352, 1362 (Fed.Cir. 1996). The statute does not specify the particular analysis or approach that ITA must use, and in the absence of showing challenged aspects of agency analysis unreasonable, the court will defer to its discretion.

(vi)

The final step of ITA’s differential pricing analysis examines whether the AA methodology can account for a pattern of prices that differ significantly by determining whether there exists a meaningful difference in the weighted-average dumping margins calculated using that methodology and an appropriate alternative comparison methodology. See IDM at 36. The Stanley plaintiffs argue that the AR5 final results do not explain why the difference in the pattern of prices cannot be accounted for with the A-A method. But their argument fails to persuade that ITA’s explanation therein as to why that approach cannot account for pricing differences was unreasonable. See id.

As explained in the AR5 final results (and again above), if the difference in the weighted-average dumping margins calculated using the A-A method and an appropriate alternative comparison method is meaningful, then that fact is indicative of whether that method cannot account for such differences and therefore an alternative method would be appropriate.. See id. More precisely, a meaningful difference between the results of the A-A methodology and an appropriate alternative (A-T in this instance) exists if: (1) there is a 25 percent relative change in the weighted-average dumping margins between the AA methodology and the appropriate alternative where both are above the de minim-is threshold, or (2) the resulting weighted-average dumping margins move across that threshold. See id.

ITA found that a meaningful difference exists because the Stanley weighted-average dumping margin did move across that threshold upon a comparison of the two methods. See id. This threshold is reasonable because comparing the weighted-average dumping margins calculated using the two methods allows ITA to quantify the extent to which the A-A method cannot take into account different Stanley pricing. And ITA’s determination that the A-A methodology cannot account for the . difference in the pattern of prices in similar circumstances has been upheld in court. E.g., Samsung Elecs. Co. v. United States, 39 CIT -, -, 72 F.Supp.3d 1359, 1368 (2015) (holding that ITA reasonably explained that “the A-to-A method does not take into account such price differences because there is a meaningful difference in the weighted average dumping margins when calculated using the A-to-A method and the A-to-T method” and that Samsung’s margin had moved across the de minimis threshold (citations omitted; emphasis in original)); Apex, supra, 38 CIT at -, 37 F.Supp.3d at 1299-1300 (holding that ITA reasonably concluded that the A-A methodology could not account for targeting where plaintiffs margin crossed the de minimis threshold), aff'd, 862 F.3d at 1323-24. The Stanley argument does not persuade that this is ah unreasonable approach to fulfilling the statute’s aim of combating masked dumping.

In AR5, ITA concluded that the A-A methodology could not account for the difference in the pattern of prices once a meaningful difference existed between that methodology and the A-T approach when the Stanley weighted-average dumping margin moved above the de minimis threshold. And, as in Apex and Samsung, Stanley does not show that-the meaningful difference was immaterial.

(vii)

The Stanley plaintiffs allege that ITA use of the Cohen d test is ‘biased toward finding prices that differ significantly, leading it to overuse the A-T rheth-od. The argument appears to conflate passing the d test with application of the A-T comparison methodology, which requires that ITA find not only that a pattern of prices that differ significantly exists but also that the A-A methodology cannot account for such differences. Each of these provisions requires a separate analysis, with distinct results, and both must be satisfied to apply an alternative comparison methodology. Moreover, Stanley citations to instances when respondents’ sales passed Cohen’s d test without discussing whether ITA applied an alternative comparison methodology, Stanley Brief at 39-40, illustrate only that the respondents’ pricing behavior exhibited certain significant differences in prices. See IDM at 37 (stating that both requirements under the statute must be satisfied before applying an alternative comparison methodology and that the Stanley analysis is concerned with and limited to'only the first of the two requirements). These instances do not show whether ITA applied an alternative comparison methodology or whether it found that the respondents sold subject merchandise at less than normal value.

The Stanley ‘plaintiffs also fail to appreciate the difference between sales found to be at significantly different prices as opposed to whether ITA has applied an alternative comparison methodology to address masked dumping. They connect high rates of sales passing Cohen’s d test to dumping. A high passing rate, however, does not mean that the A-A methodology cannot account for such differences (i.e., whether or not dumping even exists or is being masked). As ITA explained, “[b]oth requirements of -section 1677f-l(d)(l)(B) of the Act must be satisfied before [it] has the option of applying an alternative comparison method in less-than-fair-value investigations.” IDM at 37-38. As such, even if a large proportion of U.S. sales pass the d test, ITA does not automatically apply the A-T method. Id. It must also consider whether the A-A method can account for such differences and, if the standard comparison methodology can account for such differences, ITA will not apply an alternative methodology. See id.

In other words, a finding that there exists a pattern of prices that differ significantly means only that ITA will consider whether the standard comparison methodology can account for the differences. Subject merchandise can be sold in the United States market at significantly different prices yet none of the sales are priced at less than normal value (i.e., there is no dumping); in such a situation, the A-A method will be able to account for the differences, and that method will be used to calculate any weighted-average margin. A firm can also make those same U.S. sales at significantly different prices among purchasers, regions, .or time periods at prices which are all less than normal value (i.e., all sales are dumped); in such a situation, the A-A method also will be able to account for such differences, and thus, that method can, again, be used. Thus, even if there is a high Cohen’s d pass rate, it is meaningless without consideration of whether the A-A method can account for the differences. See id. at 38; 19 U.S.C. § 1677f-1(d)(1)(B)(ii).

(viii)

ITA reiterated the importance of both lower and higher priced sales in masked dumping, noting that “higher priced sales are equally capable as lower priced sales to create a pattern of prices that differ significantly.” IDM at 38. The Stanley plaintiffs disagree that “high” and “low” priced sales are appropriate considerations when conducting Cohen’s d test.

They argue that ITA may not find that higher priced sales pass that test and are part of a pattern of prices that differ significantly, but “high” and “low” are relative terms, and they concede that the statute is silent as to this issue, providing only that the agency must determine whether a pattern of prices that differ significantly exists. The statute does not specify whether ITA may or may not consider prices that differ because they are higher or lower. See 19 U.S.C. § 1677f-1(d)(1)(B) (alternative methodology may be applied if (i) “there is a pattern of export prices ... for comparable merchandise that differ significantly among purchases, regions, or periods of time, and (ii) [ITA] explains why such differences cannot be taken into account” using the A-A methodology).

Finding no explicit statutory support, the Stanley plaintiffs look to the SAA, which they interpret to mean that targeting and dumping are linked in the statute and, thus, ITA is only authorized to consider dumped prices. But the SAA does discuss both “dumped prices” and “higher prices”, as Stanley itself notes: “[t]he SAA explains that ‘targeted dumping’ comprises ‘situations [in which] an exporter may sell at a dumped price to particular customers or regions, while selling at higher prices to other customers or regions.’ ” Stanley Brief at 42, quoting SAA at 842 (emphasis deleted). Thus, the SAA acknowledges that “targeted dumping” includes sales which have been made at both “dumped” (or lower) prices as well as higher prices, and high priced sales will offset lower priced sales, “either implicitly through the calculation of a weighted-average sale price for a [United States] averaging group, or explicitly through the granting of offsets when aggregating the A-to-A comparison results, that can mask dumping”. IDM at 38. In other words, higher and lower priced sales do not operate independently: in theory at least all sales are relevant to the analysis, and nothing in the statute or the SAA precludes ITA from reviewing both higher and lower priced sales. See id. at 38-39.

(⅛)

Additionally, the Stanley plaintiffs challenge the calculation of the measure which ITA uses to gauge the effect size, i.e., the Cohen d coefficient. To calculate the effect size, it uses the “pooled standard deviation,” which is based on the distribution of the prices between the test and comparison groups, because it “reflects the dispersion, or variance, of prices within each of the two groups.” IDM at 36. The consolidated-plaintiffs contend that the use of a pooled standard deviation leads to a bias for finding high Cohen d pass rates. See Stanley Brief at 39 (“ITA incorrectly calculated the pooled standard deviation in the Cohen d statistic — generating an upward bias in the ‘pass’ rate — by giving equal weight to the squared standard deviations of the ‘target’ and ‘comparison’ price groups despite clear evidence that the target groups were much smaller in volume and the standard deviations of the target and comparison groups were not equal”).

But once again, there is no statutory directive with respect to how ITA determines whether a pattern of prices that differ significantly exists, let alone how to calculate the pooled standard deviation of the Cohen d coefficient. See Certain Frozen Warmwater Shrimp From the Socialist Republic of Vietnam, 80 Fed.Reg. 55328 (Sept. 15, 2015), and accompanying I & D memo at 27. ITA has generally relied on a reasonable and predictable approach by using a simple average when determining the pooled standard deviation. E.g., id. By giving equal weight to the test and comparison groups, ITA balances the importance of the exporter’s pricing behavior to a given purchaser, region, and time period, and the exporter’s pricing behavior to other purchasers, regions, and time periods. This implies that the magnitude of the sales to one group does not skew the outcome. See id.

Furthermore, as discussed above, even when a majority of a respondent’s sales pass the Cohen d test, this does not end ITA’s analysis in determining whether to apply an alternative methodology when calculating its weighted-average dumping margin. See IDM at 38. The agency must also consider and explain why the A-A comparison method cannot account for such differences, in order to satisfy both requirements under section 1677f-1(d)(1)(B) of the Act, and only then does ITA consider the application of the A-T method. Id.

The Stanley plaintiffs attempt to validate their claim on the supposed bias of the Cohen d test by pointing to the outcomes of 150 preliminary determinations in which a differential pricing analysis was employed. See Stanley Administrative Case Brief at 33-34 and Addendum A. 16 However, the Stanley data and analysis fail to establish (1) that a bias exists among those preliminary determinations and (2) how any potential bias would be attributable to ITA’s calculation of the pooled standard deviation based on a sim- pie average of the variances of the test and comparison groups. See IDM at 37.

The Stanley data fail to demonstrate a bias in ITA’s application of the Cohen d test. They show that 113 of the 150 cases cited involved a sufficient percentage of sales value passing the d test to consider the application of an alternative comparison methodology. See Stanley Administrative Case Brief at Addendum A. Of these, ITA applied the A-T method in only 50 of the determinations. From Stanley’s own data, accordingly, there does not appear to exist a bias in the agency’s application of the differential pricing analysis including Cohen’s d test based on the use of a simple average in determining the pooled standard deviation. Only one-third of the cases to which Stanley cites resulted in the application of an alternative comparison methodology, representing less than one-half of the cases in which there existed a pattern of prices that differ significantly pursuant to the Cohen d and ratio tests.

The Stanley argument, to wit, “the conclusion that two companies targeted all of their sales underscores” the unreasonableness of differential pricing because “it makes no economic sense for any one company to ‘target’ the majority of its sales,” Stanley Brief at 40, and because “if all sales are ‘targeted,’ then none can be,” Stanley Administrative Case Brief at 33, expresses a misappreciation of how ITA determines the existence of a pattern of export prices that differs significantly among purchasers, regions, or time periods. The focus is not on “targeting” and economic decision-making, but on the difference between export prices. 17 While Stanley pointed to a single case where all of the • respondent’s sales prices differed significantly, there are also 16 cases in the data where none of the sales prices did so, indicating that ITA’s approach is not unreasonable and does not exhibit a bias. In other words, the phenomenon to which Stanley points as proof of bias is controverted by its opposite, i.e., that no sales pass the Cohen d test. Accordingly, Stanley’s own data indicate that, if anything, there is a tendency against finding a pattern of prices that differ significantly across purchasers, regions, or time periods,

(x)

The Stanley plaintiffs press a number of additional arguments, none of which is persuasive.

First, their argument that the “meaningful difference” element of the Cohen d test has the perverse effect of allowing a respondent to avoid the A-T method with zeroing if all its sales are dumped, gaining a lower margin than if only some of its sales are dumped, lacks merit. If all of a respondent’s sales are dumped, then there is no zeroing because there are no sales that are not dumped, and the weighted-average dumping margin calculated using the A-A and A-T method is identical. If only some of a respondent’s sales are dumped, the calculated weighted-average dumping margin will be reduced, reflecting the fact that there is less dumping overall, regardless of whether or not zeroing is applied to the non-dumped sales. Accordingly, it is unclear how the respondent would gain a lower dumping margin if -all of its sales were dumped. Stanley erroneously associates the possible use of zeroing with always reducing the weighted-average dumping margin, and draws an unsupportable conclusion.

Second, the Stanley plaintiffs argue that ITA’s approach is mechanical and rote, contrary to Congress’s intent that an analysis to detect masked dumping be conducted on a case-by-case basis. But the agency does examine whether the statutory requirements have been satisfied on a case-by-case basis. It reviews the individual pricing behavior of each respondent when it conducts a differential pricing analysis. Its analysis begins by examining the extent to which a respondent’s sales pass the Cohen d test, and whether a group of sales passes that test is measured relative to the “pooled standard deviation” discussed above, which specifically reflects the pricing behavior of each individual respondent. Then, ITA determines whether the differences in respondent’s prices, based bn purchaser, time period, and region, can be accounted for using the A-A methodology, which is directly'related to the respondent’s dumping in the U.S. market and whether such dumping is masked. See IDM at 41 (“[o]n a case-by-case basis, [ITA] also considers the factual information and arguments on the record for each segment of a proceeding”).

Furthermore, ITA considers arguments from parties in each segment of a proceeding concerning whether its approach should be modified. See PPM at 17 (“[interested parties may present arguments and justification in relation to the above-described differential pricing approach used in these preliminary results, including arguments for modifying the group definitions used in this proceeding”). For example, in the 2011-2012 administrative review of copper tubing from the PRC, the agency modified the time periods used in the Cohen d test. See Seamless Refined Copper Pipe and Tube From the PRC, 79 Fed.Reg. 23324 (April 28, 2014), and accompanying I & D memo at 13-14.

The defendant contends that not only does ITA review the specific circumstances of a respondent, it continues to expand its experience and alter its method as it applies the methodology, see IDM at 41, and that the agency reviewed Stanley sales to determine which passed the Cohen d test, compared Stanley weighted-average dumping margins ’ calculated. using' the A-A methodology and the mixed alternative methodology to determine whether the significant price differences based on purchaser, period, and region, could be accounted for by the A-A methodology, and found that the A-A methodology did not account for such differences. See IDM at 30-31, To the extent ITA’s application of the differential pricing analysis' was tailored to Stanley, it was not mechanical; and even if the Cohen d test itself may be inferred mechanistic, that does not,’ ipse dixit, make it unlawful, or else all calculations would be.

Third, the Stanley plaintiffs challenge ITA’s continued use of sales that have been found to pass the Cohen d test in the base group of other comparisons. But as stated in the Preliminary Results, the purpose of that test is “to evaluate the extent to which the net prices to a particular purchaser, region, or time period differ significantly from the net prices of all other sales of comparable merchandise.” PPM at 16. Simply because certain sale prices are part of a test group in one instance and part of a comparison group in other instances does not constitute double counting; agency dumping analysis includes all information and data on the record, and selectively including or. excluding certain sales is not supported by the statute.

Furthermore, the inclusion of sales that “pass” the Cohen d test in base groups for other test groups does not cause sales to “pass” that otherwise would not. The Stanley assertion to the contrary is refutable through the use of a hypothetical scenario:

[T]here are two purchasers, A and B, which purchase the subject merchandise at average prices of 10 and 20, respectively. Based on the Cohen’s d Test, when testing purchaser A, the weighted-average price to purchaser B will be the comparison group, and the difference in the two prices between purchaser A and purchaser B, i.e., 10, is found to pass the Cohen’s d Test. Then, when purchaser B is the test group, purchaser A will be the comparison group, and the sales to purchaser B will also be found to pass the Cohen’s d Test.

IDM at 41-42. If the weighted-average price to purchaser A differs significantly from the weighted-average price to purchaser B, the weighted-average price to purchaser B also differs significantly from the weighted-average price to purchaser A. The Stanley suggestion (that once ITA finds that the weighted-average price to purchaser A differs significantly from the weighted-average price to purchaser B, the sales prices to purchaser A should be excluded henceforth from the analysis) appears illogical, as it would result in no comparison being made for the weighted-average price to purchaser B because sales to purchaser A would not be allowed to be a basis for comparison. Further, if purchaser B’s sales were tested first, purchaser A’s sales would not be tested for the same reason, and such an approach would lead to arbitrary and unpredictable results that would depend upon the order in which purchasers, regions, or time periods were examined.

Fourth, the Stanley plaintiffs contend that the Cohen d test prevents respondents from refraining from engaging in “targeted dumping.” They claim that high pass rates for that test make it difficult to “avoid being found ‘guilty’ of targeted dumping.” But that test alone does not determine whether ITA will apply an alternative comparison methodology. See IDM at 37-38.

Lastly, the consolidated-plaintiffs challenge agency use of net prices rather than gross prices when examining if a pattern of prices differs significantly. Their specific contention is that ITA fails to account for circumstances of sale that cause net prices to vary, and that such circumstances are exogenous factors that do not affect a respondent’s pricing behavior but are beyond its control because of differences in selling circumstances. The defendant contends ITA uses net prices to address all circumstances of sale, deducting the associated expenses from the reported gross unit prices which are used in the Cohen d test and that the suggestion that circumstances of sale do not affect the pricing behavior of a respondent is misleading. The defendant explains that respondents will generally account for costs such as freight, packing, and direct selling expenses in their pricing decisions, and that, when a dumping margin is calculated, it is based on net prices. For that reason, the defendant continues, ITA deems it appropriate to examine whether there is a “differ significantly” pattern based on net prices, because such examination later informs agency margin calculation. See IDM at 37 (“[ITA] finds that it is appropriate and reasonable that its examination of a pattern of prices that differ significantly to be based on net prices rather than gross prices, as net prices are the basis used to calculate dumping margins and determine a respondent’s amount of dumping”). As this appears to implement the intent of the statute and the regulations, where the purpose of a differential pricing analysis is to de termine whether the A-A comparison methodology is the appropriate tool with which to measure a respondent’s dumping in the U.S. market, see 19 C.F.R. § 351.414(c)(1), this court cannot fault defendant’s rationale.

In sum, with the exception of section III.C, supra, the Stanley plaintiffs have not established that ITA’s utilization of its differential pricing analysis was out of order. See Apex Frozen Foods Private Ltd. v. United States, 862 F.3d 1322 and 862 F.3d 1337 (Fed.Cir. 2017), passim.

IV

In view of the foregoing, the motions of the plaintiff and the consolidated-plaintiffs for judgment on the agency record 18 can be granted only to the extent of remand to ITA for reconsideration of the issues of (1) the two labor classification matters, as discussed in section II.E, supra, (2) the apparent omission, in the Stanley February 17, 2015 post-verification factor of production database, of a zero in the tenth or one-hundredth decimal place in field “VJDLCROD”, as discussed in section III.A above, and (3) the application of the limiting rule, as discussed in section III.C, supra.

The results of this remand shall be filed on or before November 30, 2017, with any comments thereon due within 30 days of the filing thereof.

So ordered.

1 . Specifically, ITA concluded: that the World Bank GEM data on the record are unclear as to which countries (which could include NME countries) were used to calculate the steel rod prices, but more critically do not make any distinction for carbon content, which is one of the most important physical characteristics of that input; that the "MEPS data suffer[] from similar deficiencies” in that none of the countries covered thereby are at the same level of economic development as the PRC nor do the data distinguish carbon content; and that none of the countries covered by the Asia Metal Market prices are potential surrogate countries meeting ITA’s surrogate country criteria. See IDM at 15-16.

2 . To wit: that the reports cited by Xi’an Metals refer only to general concerns about certain practices by Thailand’s Customs Department and fail to address the specific raw material inputs consumed by respondents in this case; that the Thai SWR import values on which ITA relied cannot be concluded aberrant, as the existence of higher prices alone does not necessarily indicate that price data are distorted or misrepresentative; and that Xi’an Metals fails to identify record evidence that materially undermines the integrity of the SWR values upon which ITA relied.

3 . Plaintiff Xi’an argues, for example, that defendant’s contention that Ukraine’s Metal Expert prices are unusable because they do not satisfy ITA’s criteria of being exclusive of taxes and duties is disingenuous when the defendant states in a previous sentence that the Metal Expert prices are actual transactions that include 20% VAT and a 4% mark-up charged to intermediate traders who buy the material from domestic producers and sell to warehouses, and the calculation of the price free of such taxes and duties is a simple mathematical computation which ITA has performed in the past.

4 . Cf. Sigma Corporation v. United States, 117 F.3d 1401, 1408 (Fed.Cir. 1997) (finding, problematic the rationale that a casting producer in the surrogate country would choose to pay the highest combination of prices for pig iron plus-freight).

5 . See Final Surrogate Value Submission and Pre-Preliminary Comments (Aug. 19, 2014) at Exhibit SV-1, PDoc 208.

6 . Def s Resp. at 82.

7 . Á la Calgon Carbon Corp. v. United States, 40 CIT -, 145 F.Supp.3d 1312, 1321 (Jan. 20, 2016) ("[t]he government and petitioners, in their response briefs, chose not to address the merits of CAC’s arguments, which were raised by CAC in its opening, brief supporting its CIT Rule 56.2 motionQ] [a]ny argument, therefore, defending ITA’s selection of a $2.42 per kilogram rate to Shanxi DMD, is waived, as CAC claimed in its reply brief”), citing United States v. Great American Ins. Co. of New York, 738 F.3d 1320, 1328 (Fed.Cir. 2013) ("[i]t is well established that arguments that are not appropriately developed in a party’s briefing may be deemed waived”) (add’l citations omitted). See also Nan Ya Plastics Corp. v. United States, 810 F.3d 1333, 1346-47 (Fed.Cir. 2016), quoting id.

8 . 40 CIT -, -, 195 F.Supp.3d 1299, 1309-11 (2016) (discussing reliance upon Thai 2007 NSO data and LSI’s financial statements and remanding for explanation of why items such as “Employee welfare cost” and “Subsidy of Social Security Fund and Workmen Compensation Fund”, which ITA had previously recognized in Certain Steel Nails from the PRC, 79 Fed.Reg. 19316 (April 8, 2014), and accompanying I & D memo at cmt. 2, as indirect labor expenses of the type covered by the 2007 NSO data which therefore necessitated adjustment of the surrogate financial ratios to avoid double counting, had not been treated similarly in the review under consideration in Yingqing).

9 . Prior to Labor Methodologies, ITA’s approach had been to select a source for the production labor rate that included only production labor, but the "mixed method” adopted by ITA. here double counts the respondent’s labor cost- by saddling production labor with a rate embedded with administrative and executive labor and then charging for that administrative and executive labor a second time by leaving those costs in the factory overhead and SG & A ratio numerators. Cf. Yingqing, supra.

10 . This refers to the practice of not using transactions with U.S. selling prices above normal value to offset transactions with U.S. selling prices below normal value. See, e.g., Timken Co. v. United States, 38 CIT -, -, 968 F.Supp.2d 1279, 1281-82 (2014). ITA abandoned “zeroing” in administrative reviews in 2012, Antidumping Proceedings: .Calculation of the Weighted-Average Dumping Margin and Assessment Rate in Certain Antidumping Duty Proceedings; Final Modification, 77 Fed.Reg. 8101 (Feb. 14, 2012), and . Stanley complains that the continued act of “zeroing” generates higher calculated dumping margins (Stanley claims the use of A-T with zeroing raised its margin from zero to 13.19 percent). The court observed in dicta nearly twenty years ago that comparisons based on the A-A method appear to "allow higher prices to cancel out some amount of dumping” and also that "transaction-specific price comparisons are statistically biased toward a dumping finding”, Borden, Inc. v. United States, 22 CIT 233, 235-40, 4 F.Supp.2d 1221, 1224-28 (1998), citing How the GATT Affects U.S. Antidumping and Countervailing Duty Policy, 33-35, 66 (Congressional Budget Office 1994), but to that point "zeroing” had long been understood to be a not-improper philosophic, not mathematic, interpretation of how dumping is best determined under U.S. law — at least until certain members of appellate panels of the World Trade Organization began to surprise these United States in opining what had originally been negotiated and "agreed to” when the Antidumping “Agreement” was signed.

11 . ITA's approach has evolved over at least five distinct tests to determining the presence of targeted dumping: (1) the "pasta test”, announced in 1998 in response to the Borden decision, supra; (2) the "P/2" test (Notice of Final Determination of Sales at Less Than Fair Value: Coated Free Sheet Paper from the Republic of Korea, 72 Fed.Reg. 60630 (Oct. 25, 2007)); (3) the "Nails I” test (Certain Steel Nails from the People's Republic of China: Final Determination of Sales at Less Than Fair Value and Partial Affirmative Determination of Critical Circumstances, 73 Fed.Reg. 33977 (June 16, 2008)); (4) the "Nails 11” test (Polyethylene Retail Carrier Bags from Taiwan: Final Determination of Sales At Less Than Fair Value, 75 Fed.Reg. 14569 (March 26, 2010)); and (5) differential pricing (Xanthan Gum from the People's Republic of China: Final Determination of Sales At Less Than Fair Value, 78 Fed.Reg. 33351 (June 4, 2013), and accompanying I & D memo).

12 . But as a further threshold matter, the Stanley plaintiffs complain ITA initiated dif ferential pricing without an allegation that they had engaged in targeted dumping. And cf. Diamond Sawblades Manufacturers' Coalition v. United States, 39 CIT -, Slip Op. 15-116 (Oct. 21, 2015), at 5-6 & n.4, 2015 WL 6387113, at *2 & n.4 (ITA rejecting a targeted dumping allegation as untimely and declining to self-initiate on the ground that the targeted dumping provision applies by its express terms to agency investigations not administrative reviews). .The consolidated-plaintiffs do not press the point to one of unlawfulness herein, however.

13 . Thirty-three percent or less; 33-66 per-c'ent; and 66 percent or more.

14 . A number of Stanley arguments continue to press an interpretation of the' term “significant” that is at variance with what the statute requires. For example, the consolidated-plaintiffs raise concerns with regard to accounting for random events and Type I error (i.e., a "false positive” leading to incorrect rejection of a true null hypothesis), and they also express concerns about whether Cohen’s d test tests a statistical hypothesis, which is necessary when measuring for statistical significance. But as indicated above, there are no random estimates of actual statistical measures because ITA’s analysis relies on complete information to perform such calculations. See IDM at 34. Because the agency has the complete population of Stanley United States sales, none of the resulting calculations evince random errors because of sampling, and because there is no sampling or randomness, all issues related to Type I errors, which are errors that occur because of sampling, are moot.

15 . The Stanley definition of "significant” is "1. a) having or expressing a meaning, b) full of meaning; 2, important; momentous; 3. having or conveying a special or hidden meaning; suggestive; 4. of or pertaining to an observed departure from a hypothesis too large to be reasonably attributed to chance”, Stanley Br., p. 32, citing Webster’s New World Dictionary of the American Language at 1325 (1980) (emphasis omitted), and that proffered definition supports ITA’s understanding of being tasked by statute to find a meaningful difference between the average price of a test group and the average price of a comparison group, which the Cohen d test accomplishes. See IDM at 32, 36-37.

16 . The Stanley brief at bar cites an expanded data set covering 209 respondents through September 2015. See p. 40, Addendum B. The defendant requests that this expanded data set be ignored, as it was not submitted to ITA during the administrative process and is therefore not part of the record for this administrative review per 19 U.S.C. § 1516a(b)(l)(B)(i). It is so ordered.

17 . For example, consider two purchasers, A and B. If the prices to purchaser A are found to differ significantly from the prices to purchaser B, then'it follows that the prices to purchaser B differ significantly from the prices to purchaser A. Here, it is reasonable to conclude that all prices differ significantly. Similarly, if the prices to purchaser A do not differ significantly from the prices to purchaser B, then it follows that the prices to purchaser B do not differ significantly from the prices to purchaser A. Here, it is reasonable to conclude that none of the prices differ significantly.

18 . The quality of the papers submitted in support, as well as of those presented in opposition, obviated any need to burden the parties with oral argument, and their motion therefor, for the record, is thus hereby denied.