Mid Continent Steel & Wire, Inc. v. United States , 219 F. Supp. 3d 1326 ( 2017 )

AMENDED OPINION

Kelly, Judge:

This consolidated action comes before the court on motions for judgment on the agency record pursuant to USCIT Rule 56.2. Rule 56.2 Mot. for J. Agency R. PI. Mid Continent Steel & Wire, Inc., Feb. 26, 2016, ECF No. 28 (“Mid Continent Mot.”); PT Pis.’ Rule 56.2 Mot. J. Agency R., Feb. 26, 2016, ECF No. 30 (“PT Mot.”). Plaintiff Mid Continent Steel & Wire, Inc. (“Mid Continent”) and Consolidated Plaintiffs PT Enterprise Inc. et al. (“PT”) challenge various aspects of the Department of Commerce’s (“Department” or “Commerce”) final determination in the antidumping duty (“ADD”) investigation of imports of certain steel nails from Taiwan for the period April 1, 2013 through March 31, 2014. Br. Support Rule 56.2 Mot. J. Agency R. Pl. Mid Continent Steel & Wire, Inc. Confidential Version 17-42, Feb. 26, 2016, ECF No. 27-1 (“Mid. Continent Br.”); Br. Support Pls.’ Rule 56.2 Mot. J. Agency R. Confidential Version 8-45, Feb. 26, 2016, ECF No. 29 (“PT Br.”); see Certain Steel Nails from Taiwan: Final Determination of Sales at Less Than Fair Value, 80 Fed. Reg. 28,959 (Dep’t Commerce May 20, 2015) (final determination of sales at less than fair value) (“Final Results”); Certain Steel Nails From the Republic of Korea, Malaysia, the Sultanate of Oman, Taiwan, and the Socialist Republic of Vietnam: Antidumping Duty Orders, 80 Fed. Reg. 39,-994 (July 13, 2015) (“Order”).

Mid Continent commenced this action pursuant to section 516A of the Tariff Act of 1930, as amended, 19 U.S.C. § 1516a (2012).¹ Summons, Aug. 8, 2015, ECF No. 1; see Compl., Sept. 4, 2015, ECF No. 9. Mid Continent challenges Commerce’s determination that PT’s affiliated producer Pro-Team Coil Nail Enterprise Inc. (“Pro-Team”) is not affiliated with [[ ]] of the [[ ]] tolling companies that do production activities for Pro-Team.² Mid Continent Br. 17-41. PT challenges: (1) several aspects of Commerce’s application of its differential pricing analysis as unreasonable and contrary to law, PT Br. 8-35; (2) Commerce’s treatment of Pro-Team’s costs/expenses relating to production/sales of steam as not supported by, substantial evidence, id. at 35^12; and (3) Commerce’s decision to adjust transfer prices paid for wire drawing and nail making to reflect market prices as not supported by substantial evidence, id. at 42-44.

Defendant, United States (“Defendant”), responds that the court should deny the motions of Mid Continent and PT and sustain Commerce’s Final Results in full. See Def.’s Resp. Opp’n Pis.’ Rule 56.2 Mots. J. Agency R. Confidential Version 9-54, Aug. 3, 2016, ECF No. 47 (“Def.’s Resp.”). Mid Continent filed a response, as defendant-intervenor, in opposition to PT’á motion, see Resp. Br. Mid Continent Steel & Wire, Inc. Opp’n PT Enterprise Inc. et al.’s R. 56.2 Mot. J. Agency R. Confidential Version, Aug. 3, 2016, ECF No. 42 (“Mid Continent Resp. Br.”), and PT filed a response, as defendant-intervenors, in opposition to Mid Continent’s motion. See Def.-Intervenors’ Resp. Br. Opp’n Pl.’s Mot. J. Agency R. Confidential Version, Aug. 3, 2016, ECF No. 48 (“PT Resp. Br.”).

For the reasons that follow, Commerce’s final results are sustained in part and remanded in part. Specifically, the court sustains Commerce’s determinations: (1) that Pro-Team is unaffiliated with the [[ ]] tok lers in question; (2) to use the Cohen’s d test within the differential pricing analysis to determine the existence of a pattern of significant price differences; (3) to use a simple average to calculate the pooled standard deviation in the Cohen’s d test of the differential pricing analysis; (4) to not offset dumped sales with non-dumped sales in calculating the respondent’s anti-dumping duty margin using the average-to-transaction methodology; and (5) to disregard transfer prices paid by Pro-Team to certain affiliated tollers in its calculation of normal value (“NV”). The court remands Commerce’s allocation of expenses associated with Pro-Team’s separate steam line of business for further explanation and consideration consistent with this opinion.

BACKGROUND

On June 25, 2014, in response to a petition filed by Mid Continent, Commerce initiated an antidumping duty (“ADD”) investigation of certain steel nails from six countries, including Taiwan. See Certain Steel Nails from India, the Republic of Korea, Malaysia, the Sultanate of Oman, Taiwan, the Republic of Turkey, and the Socialist Republic of Vietnam: Initiation of Less-Than-Fair-Value Investigations, 79 Fed. Reg. 36,019 (June 25, 2014) (notice of initiation of ADD investigations). Commerce selected Taiwanese exporters PT and its affiliated producer, Pro-Team, and Quick Advance, Inc. and its affiliated producer, Ko’s Nails Inc., as mandatory respondents for the investigation. See Certain Steel Nails from Taiwan: Negative Preliminary Determination of Sales at Less Than Fair Value and Postponement of Final Determination, 79 Fed. Reg. 78,-053, 78,054 (Dep’t Commerce Dec. 29, 2014) (preliminary determination of sales at less than fair value, postponement of final determination) (“Prelim. Results”); Decision Memorandum for the Preliminary Determination in the Antidumping Duty Investigation of Certain Steel Nails from Taiwan at 2, PD 225, bar code 3247845-01 (Dec. 17, 2014) (“Prelim. Decision Memo”);³ see 19 U.S.C. § 1677f-1(c)(2)(B).

On December 29, 2014, Commerce issued its negative preliminary determination. See Prelim. Results, 79 Fed. Reg. at 78,053; Prelim. Decision Memo at 1. Com: merce applied the differential pricing analysis and determined that, although 41.73% of PT’s sales passed the Cohen’s d test, a meaningful difference did not exist in the dumping margins that would result using the standard A-to-A methodology and the alternate mixed methodology. Id at 12, Commerce accordingly applied the standard average-to-average (“A-to-A”) methodology to all of PT’s sales, id,, and preliminarily determined that respondents’ steel nails from Taiwan “are not being, or are not likely to be, sold in the United States at less than fair value.” Id. at 1. Commerce preliminarily assigned PT a weighted-average dumping margin of 0.00%. Prelim. Results, 79 Fed. Reg. at 78,054. Commerce found that PT used certain affiliated tol-lers to produce its steel nails and accordingly Commerce adjusted the transfer prices paid for wire drawing and nail making to reflect the market price. See Prelim. Decision Memo at 14. Commerce also disregarded certain transactions between Pro-Team and its affiliated tollers because it determined that there was a difference between the market price and transfer prices for wire drawing and nail making services. Id. at 14. Commerce also preliminarily determined that PT and the rest of its tollers were unaffiliated. Cost of Production and Constructed Value Calculation Adjustments for the Preliminary Determination—PT Enterprise Inc. at 2, CD 200, bar code 3248421-01 (Dec. 17, 2014). Commerce further found that PT did not have a viable comparison market, so it calculated a constructed value (“CV”) as the basis for NV using the financial statements of PT affiliate Pro-Team. Prelim. Decision Memo at 16.

On May 13, 2015, Commerce issued its final determination. See. Final Results, 80 Fed. Reg. at 28,959. and accompanying Issues and Decision Memorandum for the Affirmative Final Determination in the Less than Fair Value Investigation of Certain Steel Nails from Taiwan, Oct. 16, 2015, ECF No. 17 (“Final Decision Memo”). Commerce continued to use Pro-Team’s financial statements to calculate CV. Final Decision Memo at 10-15. However, Commerce declined to allow subsidy income received by Pro-Team to offset Pro-Team’s general and administrative expenses because Commerce concluded “the subsidy is not related to the general operations of the company as a whole, but instead appears directly related to and intended for the company’s other steam line of business.” Id. at 55. Commerce applied this subsidy income as an offset to Pro-Team’s cost of goods sold instead. Id. Commerce also continued to disregard transactions between Pro-Team and certain affiliated tollers for wire.drawing and nail making services, id. at 53, and continued to find Pro-Team unaffiliated with its other [[ ]] tollers. Id. at 50-53. Commerce determined that 42.27% of PT’s U.S. sales had passed the Cohen’s d test, id. at 19, and accordingly determined to apply the mixed methodology, by which Commerce applies the average-to-transaction (“A-toT”) method to PT’s sales that passed the Cohen’s d test and the A-to-A method to PT’s sales that did not pass the Cohen’s d test to calculate a respondent’s weighted-average dumping margin. Id Commerce determined that a meaningful difference existed between the resultant A-to-A margin and mixed methodology margin, so applied the mixed methodology to calculate PT’s weighted-average dumping margin. Commerce subsequently assigned PT a weighted-average dumping margin ■ of 2.24%. Final Results, 80 Fed. Reg. at 28,-961. The final determination resulted in an ADD order on subject nails from Taiwan. Order, 80 Fed. Reg. at 39,994.

JURISDICTION AND STANDARD OF REVIEW

The court has jurisdiction pursuant to 19 U.S.C. § 1516a(a)(2)(B)(iii) and 28 U.S.C. §■ 1581(c) (2012),⁴ which grant the court authority to review actions contesting the final determination in an administrative review of an antidumping duty order. The court will uphold Commerce’s determination unless it is “unsupported by substantial evidence on the record, or otherwise not in accordance with law.” 19 U.S.C. § 1516a(b)(1)(B)(i).

DISCUSSION

I. Affiliation Between Pro-Team and Certain of its Tollers

Mid Continent challenges as unsupported by substantial evidence Commerce’s determination that PT’s affiliate Pro-Team⁵ is unaffiliated with [[ ]] of its tollers.⁶ Mid Continent Br. 17-42; see Final Decision Memo at 50-53. In particular, Mid Continent argues that Pro-Team’s purportedly unaffiliated tollers are and have been reliant upon Pro-Team, Mid Continent Br. 22-27, 35-37, and that Pro-Team is “legally and operationally in a position to exercise restraint and direction over” the tollers through a close supplier relationship, such that Commerce should have found Pro-Team and the tollers affiliated and accordingly disregarded the transactions between them for purposes of calculating costs of production. Id at 27-35; see 19 U.S.C. §§ 1677(33), 1677b(f)(2). Defendant responds that substantial evidence supports Commerce’s determination that Pro-Team and the [[ ]] tollers are not affiliated. Def.’s Resp. 44-52. For the reasons that follow, Commerce’s determination that Pro-Team is not affiliated with the [[ ]] tollers in question is supported by substantial evidence.

Pursuant to 19 U.S.C. § 1677b(f)(2), Commerce may disregard transactions between persons found to be affiliated for purposes of calculating costs of production, “if, in the case of any element of value required to be considered, the amount representing that element does not fairly reflect the amount usually reflected in sales of merchandise under consideration in the market under consideration.” 19 U.S.C. § 1677b(f)(2). Commerce finds affiliation where, inter alia, some form of control exists between persons or entities.⁷ 19 U.S.C. §§ 1677(33)(F), (G); see SAA, H.R. Rep. No. 103-316, vol. 1 at 838, 1994 U.S.C.C.A.N. at 4,175.⁸ Commerce finds that control exists where one person or entity is “legally or operationally in a position to exercise restraint or direction over the other person” or entity. 19 U.S.C. § 1677(33). To determine if control exists pursuant to 19 U.S.C. § 1677(33), Commerce considers the existence of certain relationships, including “corporate or family groupings; franchise or joint venture agreements; debt financing; and close supplier relationships.” 19 C.F.R. § 351.102(b)(3) (2013).⁹ For the presence of these relationships to evidence control, however, the relationship must have “the potential to impact decisions concerning production, pricing or cost.” Id.

Here, Commerce sought and examined information from Pro-Team and its tollers for indications of control based on close supplier relationships,¹⁰ and determined that the evidence could not support a finding of control. Final Decision Memo at 50-53. Commerce analyzed the nature of the tolling relationships, including the duration of the tolling relationships, whether Pro-Team owned the tollers’ machinery, land, or buildings, and whether Pro-Team and the tollers shared common stock. Id. at 50. Commerce also considered “(i) the terms and provisions of supply agreements; (ii) the relative percentage that tolling services to Pro-Team represented of each of the suppliers’ total sales; (iii) the terms of any financing agreements with the suppliers; and (iv) the overall profitability of the tollers.” Id. Commerce found that many of the tollers operated as tollers prior to doing business with Pro-Team, id. at 51, there was no common stock or familial ownership between the entities, id. at 50-51, and Pro-Team did not share employees, officers, or managers with its tollers. Id. at 50. Commerce also found that: i) there were no contracts or supply agreements between Pro-Team and the unaffiliated tollers “locking the toller into providing services for a specific period of time,” id. at 51; ii) although many of the tollers supplied Pro-Team exclusively, nothing prohibited those tollers from supplying services to others,¹¹ see id. at 51-52; iii) there were no debt financing agreements between Pro-Team and the tollers, id. at 52; and iv) most of the tollers were profitable during 2013.¹² See id. Based on these findings, Commerce determined that Pro-Team and the [[ ]] tollers exhibited “typical economic cooperation” of suppliers and a producer with a “decentralized business model,” and did not evidence “the potential to impact decisions concerning production, pricing or cost” or the ability to “exercise restraint or direction” over the other. See id; 19 C.F.R. § 351.102(b)(3); 19 U.S.C. § 1677(33). Commerce concluded that these relationships did not create affiliation between Pro-Team and its tollers. Final Decision Memo at 53.

Commerce reasonably concluded that Pro-Team’s relationships with its tollers do not rise to the level of control necessary to support a finding of affiliation. The statute, 19 U.S.C. § 1677(33), and the regulation, 19 C.F.R. § 351.102(b)(3), together seek to identify producer-supplier relationships of such an integrated nature that the two entities cannot be said to transact at arm’s length. Commerce sought evidence of conditions that would logically indicate whether either party was able to operate freely from the other, including shared physical space, exclusivity contracts, lack of profitability of the tollers, the presence of control over prices, common ownership, and shared employees. Final Decision Memo at 50-52. Implicit in Commerce’s focus on these conditions is that the presence of these conditions would indicate that one entity functionally controlled the other. Commerce reasonably concluded here, from the absence of most of these conditions, that the tolling relationships are not the type of close supplier relationships that the statute and regulation seek to identify (he., those that lead to unfair transactions with unfair prices). Mid Continent’s argument that the record does not support Commerce’s findings is unavailing. Mid Continent contends that control exists because the tollers did the large majority of their business with Pro-Team, that many of the tollers share facilities with Pro-Team or its affiliated tollers, and that many of the relationships are longstanding. See Mid Continent Br. 21-42. The fact that Pro-Team’s relationships with tollers are long-standing or exclusive is not dis-positive; Commerce found that many of the tollers had been in business before their relationships with Pro-Team, see Final Decision Memo at 51; and that none of the tollers had the “expectation of exclusively providing a particular service to Pro-Team.” Id. at 52.¹³ Commerce’s assessment and ultimate conclusion is reasonable.¹⁴

II. Differential Pricing Analysis

PT challenges three aspects of Commerce’s differential pricing analysis and the resultant application of the alternate A-to-T methodology in this investigation. See PT Br. 8-35. First, PT contends that Commerce’s application of the Cohen’s d test, including the use of the coefficient threshold, in the differential pricing analysis is contrary to the statute, arbitrary, and otherwise unreasonable. Id. at 25-30. Second, PT also challenges as unreasonable, in general and as applied, Commerce’s use of a simple rather than weighted-average to calculate the pooled standard deviation for the Cohen’s d test. Id. at 30-35. Third, PT argues that Commerce’s application of the alternate A-to-T methodology, and specifically its use of “double zeroing,” to calculate PT’s anti-dumping duty margin was contrary to the statute. Id. at 18-25. Each of PT’s arguments with respect to Commerce’s differential pricing analysis are unavailing. For the reasons that follow, Commerce’s determination is supported by substantial evidence and in accordance with law.

A. Exhaustion

As an initial matter, Mid Continent argues that the court should reject PT’s arguments related to Commerce’s differential pricing analysis because PT failed to exhaust its substantial evidence arguments before Commerce.¹⁵ Mid Continent Resp. 11-16, 40. Mid Continent contends that, before Commerce, PT only challenged the differential pricing analysis on legal grounds while, before the court, PT challenges the differential pricing analysis on additional legal grounds and on substantial evidence grounds. Id. at 11-16. PT replies that the exhaustion doctrine is inapplicable here, because PT’s substantial evidence challenges to the differential pricing analysis arose only upon Commerce’s application of the analysis in the final determination and because Commerce was on notice of these arguments as PT raised these challenges “preemptively” in its administrative case brief to Commerce following the preliminary determination. Taiwan Plaintiffs’ Reply Br. Confidential Version 11-13, Sept. 23, 2016, ECF No. 58 (“PT Reply Br.”). For the reasons that follow, the exhaustion doctrine does not bar PT from bringing its challenges to the differential pricing analysis before this Court.

The court must, “where appropriate, require the exhaustion of administrative remedies.” 28 U.S.C. § 2637(d). Commerce’s regulations require parties to submit a case brief containing all their arguments. See 19 C.F.R. § 351.309(c)(2). Exhaustion ensures that the administrative agency maintains the authority to thoroughly perform its functions, including reviewing and responding to any perceived error, before any challenges to its actions are heard by this Court. United States v. L.A. Tucker Truck Lines, Inc., 344 U.S. 33, 37, 73 S.Ct. 67, 97 L.Ed. 54 (1952); Carpenter Tech. Corp. v. United States, 30 C.I.T. 1373, 1374-75, 452 F.Supp.2d 1344, 1346 (2006). Nonetheless, the application of the exhaustion doctrine in trade cases is to be exercised with a measure of discretion by the court. See, e.g., Corus Staal BV v. United States, 502 F.3d 1370, 1381 (Fed. Cir. 2007) (“[Alp-plying exhaustion principles in trade cases is subject to the discretion of the judge of the Court of International Trade.”).

As discussed above, Commerce issued a negative preliminary determination in this investigation after preliminarily applying the standard A-to-A methodology to all of PT’s sales. See Prelim. Results, 79 Fed. Reg. at 78,054; Prelim. Decision Memo at 1, 12. PT nonetheless included challenges to Commerce’s differential pricing analysis in its administrative case brief, noting that it was raising the arguments should the differential pricing analysis result in the final determination “in different percentages,! ■ • • or in Commerce] deciding that the differences in prices cannot be taken into account using the preferred A-A methodology.” Admin. Case Br. PT/Pro-Team, Antidumping Duty Investigation of Certain Steel Nails from Taiwan at 8, CD 262, bar code 3267614-01 (Mar. 31, 2016) (“PT Admin. Case Br.”), PT acknowledges that its arguments before the agency were unsupported by facts, but emphasizes it did not put forth specific facts because facts to support the challenges “did not exist yet” at the time of submission of its case brief to Commerce. PT Reply Br. 12-13 (emphasis in original). When Commerce did reach a different result in the final determination, after applying the mixed methodology and assigning PT a weighted-average dumping margin of 2.24%, Final Results, 80 Fed. Reg. at 28,961, PT challenged the differential pricing analysis before this court on substantial evidence grounds, supported by the facts of the case. See generally PT Br. 6-35.

PT’s substantial evidence challenge and the facts to support it arose in the final determination, after Commerce applied a different methodology and reached a materially different result for PT than it had reached in the preliminary determination. Barring PT from bringing this claim now would leave PT without the opportunity to be heard on this substantial evidence challenge. Additionally, a party cannot be required to raise a particular challenge preemptively before the facts support it. Any such preemptive substantial evidence challenge would be necessarily speculative, illogical, and useless. Nonetheless, PT did raise relevant arguments in its administrative case brief to notify Commerce of its objections to the differential pricing analysis. The court finds that, under these eir-cumstances, Commerce was on sufficient notice of PT’s arguments, despite that the arguments did not include specific factual challenges.

B. Cohen’s d coefficient threshold

PT challenges as contrary to law, unreasonable, and arbitrary Commerce’s use of the Cohen’s d test in its differential pricing analysis to assess whether there are significant price differences among a particular group of U.S. sales. PT Br. 25-30. Defendant responds that the Cohen’s d test is in accordance with law and reasonable, generally and as applied to PT’s sales in this investigation. See Def.’s Resp. 23-26. For the reasons that follow, PT’s challenges are unavailing.

Section 19 U.S.C. §§ 1677f-1(d)(1)(B)(i) and (ii) provide that Commerce may use the alternate A-to-T methodology to calculate weighted-average dumping margins where (i) Commerce finds a pattern of export prices for comparable merchandise that differ significantly among purchasers, regions, or periods of time, and (ii) Commerce explains why such differences cannot be taken into account using the standard A-to-A methodology.¹⁶ See 19 U.S.C. §§ 1677f-1(d)(1)(B)(i)—(ii).

As no statute or regulation guides Commerce in determining whether a pattern of significant price differences exists, Commerce is afforded broad discretion to select a methodology to make that determination. See Fujitsu General Ltd., 88 F.3d 1034, 1039 (Fed. Cir. 1996); Torrington Co. v. United States, 68 F.3d 1347, 1351 (Fed. Cir. 1995). In situations involving complex and technical methodological choices, Commerce has broad discretion and the court' need only address whether Commerce’s methodological choice is reasonable. See Motor Vehicle Mfrs. Ass’n of U.S. v. State Farm Mut. Auto. Ins. Co., 463 U.S. 29, 48-49, 103 S.Ct. 2856, 77 L.Ed.2d 443 (1983) (“[A]n agency must cogently explain why it has exercised its discretion in a given manner.”); see Smith-Corona Group v. United States, 713 F.2d 1568, 1571 (Fed. Cir. 1983), cert. denied, 465 U.S. 1022, 104 S.Ct. 1274, 79 L.Ed.2d 679 (1984); Fujitsu Gen. Ltd., 88 F.3d at 1039 (granting Commerce significant deference in determinations “involving] complex economic and accounting decisions of a technical nature”); Ceramica Regiomontana, S.A. v. United States, 10 C.I.T. 399, 404-405, 636 F.Supp. 961, 966 (1986), aff'd, 810 F.2d 1137, 1139 (Fed. Cir. 1987).

Commerce determines whether a pattern of significant price differences exists among purchasers, regions, or periods of time with the differential pricing analysis. Final Decision Memo at 16; Prelim. Decision Memo at 10. The first stage of the differential pricing analysis is the Cohen’s d test, which Commerce uses to measure the degree of price disparity between two groups of sales. Prelim. Decision Memo at 11; see Final Decision Memo at 20-21, 24-26. In the Cohen’s d test, Commerce calculates the number of standard deviations by which the weighted-average net prices of U.S. sales for a particular purchaser, region, or time period (the “test group”) differ from the weighted-average net prices of all other U.S. sales of comparable merchandise (the “comparison group”).¹⁷ See Prelim. Decision Memo at 11; Final Decision Memo at 25. The result of this calculation is a coefficient. See Prelim. Decision Memo at 11; Final Decision Memo at 25. To arrive at the coefficient, Commerce divides the difference in the means of the net prices of the test' group and comparison group by the pooled standard deviation.¹⁸ Prelim. Decision Memo at 11; Def.’s Resp. 14. The coefficient is the number of standard deviations by which the weighted-average of the comparison group and the test group differ. See Def.’s Resp; 14; PT Br. 15. Commerce then uses a threshold to qualify the extent of the difference measured: a group of sales with a coefficient equal to or greater than 0.8 is said to “pass” the Cohen’s d test, which signifies to Commerce that a pattern of price differences exists within that group of sales. See Prelim. Decision Memo at 11.

Commerce explained that the Cohen’s d test identifies relative difference, rather than absolute difference, which allows Commerce to accurately assess the significance of price differences within a group. Final Decision Memo at 25. • Commerce explained that a statistical tool which measures relative difference “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.” H. (emphasis in original; internal quotation and citation omitted). Commerce explained that it finds the Cohen’s d test to be useful for determining, per the statute, whether significant price differences exist within a certain group of sales, precisely because this test captures relative difference. See id. Commerce thus finds the Cohen’s d test to be a “reasonable tool” for effectuating its statutory directive to assess whether there is a pattern of significant price differences among a particular group of U.S. sales. Id. at 26.

The court agrees that this methodology is reasonable. The significance of differences between prices will depend on the prices themselves. A test which calculates absolute difference would not afford Commerce an accurate picture in all instances of the significance of price differences within a group. It is reasonable for Commerce to instead use a test, such as the Cohen’s d test, to evaluate the relative differences in the export prices between two groups in order to assess whether that difference is significant in accordance with the statutory directive of 19 U.S.C. § 1677f-1(d)(1)(B)(i). Commerce has sufficiently explained why its use of-the Cohen’s d test is a reasonable method by which to determine whether a pattern of significant price differences exists pursuant to 19 U.S.C. § 1677f-1(d)(1)(B)(i), and the court finds Commerce’s methodological choice to make this determination using the Cohen’s d test, including its fixed thresholds, to be a reasonable exercise of Commerce’s discretion.

PT contends that Commerce should dispel with the 0.8 coefficient threshold and institute a more fluid test that considers case-by-case circumstances, in order to avoid arbitrary results based on insignificant differences deemed significant by application of a “rigid” threshold. See PT Br. 25-28. PT argues that Commerce’s “use of these arbitrary thresholds without any further analysis of the relative or actual price differences at issue in a case, within the context of the industry being analyzed,” does not comport with the statute, id. at 28, and that the test as applied here resulted in arbitrary and unreasonable results because there were instances in which “demonstrably insignificant price differences” passed the 0.8 threshold so were determined to be significant.¹⁹ See id. at 28-29.

PT’s argument misunderstands the Cohen’s d test. As Commerce explained, the test gauges relative differences, Final Decision Memo at 25-26, and measurement of relative differences is necessarily case-by-case:

The purpose of identifying a pattern of prices which differ significantly is to establish that the exporter’s .pricing behavior has created conditions under which dumping may be masked .... This is a determination made on the facts on the record, Le., the U.S. prices which exhibit the exporter’s pricing behavior[.] ... Without such a fact-based approach[,] ... the Department’s dumping calculations would be fraught with subjectivity with no regard to transparency and predictability.

Id. at 20. It is discernible from this explanation that Commerce uses the Cohen’s d test because the test allows a case-by-case analysis of whether the price differences between groups of sales are significant. The test’s fixed 0.8 threshold allows Commerce to assess the significance of relative differences with a consistent benchmark.²⁰ See id. at 30. Without the consistent threshold, Commerce would have no way to assess the significance of relative differences which differ in every case. Commerce is best positioned to identify what constitutes a “significant” difference in a certain case. See Fujitsu Gen. Ltd., 88 F.3d at 1039 (noting that Commerce is given significant deference in determinations “involving] complex economic and accounting decisions of a technical nature”). Commerce sufficiently explained its methodology and that methodology appears tailored to achieve the statutory directive. This test, including its use of the 0.8 threshold, is reasonable and within Commerce’s discretion to determine whether significant price differences exist.

PT also argues that Commerce’s reliance on the Cohen’s d test in this case is contrary to 19 U.S.C. § 1673b(b)(3), which instructs Commerce to consider price differences of less than 2% not to constitute actionable dumping. See PT Br. 29-30. PT emphasizes that, here, “there are [[]] instances where the price differences identified are less than 2% of the average price for the CONNUM,” id., and contends that “[i]t is hard to imagine the statutory directive that would consider a 2% price difference not to be indicative of dumping but sufficient to uncover targeted dumping.” Id at 30. Section 19 U.S.C. § 1673b(b)(3) is inapposite here, as the provision instructs Commerce in determining weighted-average dumping margins. The Cohen’s d test does not measure dumping; it is a tool to measure price differences pursuant to 19 U.S.C. § 1677f-1(d)(1)(B)(i). PT’s argument is unpersuasive.

C. Use of a simple average to calculate the pooled standard deviation

PT argues that it is unreasonable for Commerce to use a simple average rather than a weighted-average to calculate the pooled standard deviation for the Cohen’s d test, contending that a simple average is distortive and internally inconsistent.²¹ See PT Br. 30-35. PT proposes that Commerce instead weight the averages by observations or sales quantities to ensure that every transaction is weighted equally, regardless of the number of sales in the group. Id. at 35. Defendant responds that Commerce’s practice of using a simple average for the pooled standard deviation is reasonable. See Def.’s Resp. 26-29. PT has not demonstrated that Commerce’s use of a simple average is unreasonable, in general or as applied to PT’s sales in this investigation.

As mentioned above, to calculate the Cohen’s d coefficient, Commerce divides the difference in the weighted-averages of the net prices of the test group and the comparison group by a pooled standard deviation of the comparison and test groups. Prelim. Decision Memo at 11; see Final Decision Memo at 25. The pooled standard deviation is derived using a simple average of the variances in the net prices within the test and comparison groups.²² Final Decision Memo at 28-29. Commerce explained that it uses a simple average for the pooled standard deviation to calculate an average of the comparison and target group’s pricing behaviors that equally reflects both groups: “[b]y using a simple average, the respondent’s pricing practices to each group [are] weighted equally, and the magnitude of the sales to one group does not skew the outcome.” Id. At oral argument, the Government further explained that Commerce uses a simple average in the pooled standard deviation “to accord equal weight to pricing behavior to the test group and the comparison group.” Confidential Oral Arg. Tr. 56, Mar. 2, 2017, ECF No. 82 (“Oral Arg. Tr.”). Given this objective, weight-averaging by sales quantity “would inappropriately move the pooled standard deviation toward the pricing behavior of either the test or comparison group, depending on which had more weight.” Id.

It is apparent that the parties disagree as to the purpose of the pooled standard deviation within the Cohen’s d test. PT’s position is that the pooled standard deviation is used to discern “whether specific price changes in the test group are significantly different from the rest” of the price changes overall for that CONNUM, Oral Arg. Tr. 60; PT Br. 30, 33, so the average used should reflect the overall average of the pricing behavior among all sales in the test and comparison groups, regardless of the size of each group. See Oral Arg. Tr. 60. PT argues that a weighted-average would account for size differences between the test and comparison groups and ensure that each transaction is accorded equal weight in the overall analysis. See PT Br. 30. Commerce does not dispute that each transaction is not equally weighted in the simple average, but explains that the objective of the pooled standard deviation is to calculate an average of the two groups’ prices so weighting both groups equally is important. See Final Decision Memo at 28-29. Commerce explains that the pooled standard deviation should reflect the overall average of the pricing behavior in the test and comparison groups, regardless of the number of transactions or quantity of sales within each group. See id Commerce explains that a simple average ensures that the two groups’ average pricing behaviors are weighted equally. Id.; see also Oral Arg. Tr. 56.

It is discernible from Commerce’s explanations that Commerce views the pooled standard deviation as an average reflective of the respondent’s average pricing behavior for these two groups, rather than an average reflective of all of the individual prices. A simple average does ensure that “equal weight of pricing behavior [is given] to the test and comparison groups.” Oral Arg. Tr. 57. The standard deviation is intended to serve as a “yardstick” by which to measure the difference in a certain group of sales from the overall spread of difference in the test and comparison groups. Id. at 52; Final Decision Memo at 26. One logical way to create this yardstick is to average the pricing behaviors of the two groups together. Although PT’s alternate weighted-average approach may be another reasonable way of achieving this yardstick, using a simple average to achieve it is also reasonable.

PT’s contention that the simple average achieves a distorted standard deviation is therefore unavailing. PT argues that the simple average is distortive because it gives more power to the pricing of sales in lower-quantity groups, PT Br. 30-33; see also PT Admin. Case Br. 20-22; Oral Arg. Tr. 49, which distorts the standard deviation given “[t]he point of a pooled standard deviation is to come up with a coefficient that is representative of the price changes that are going on within that time.” Oral Arg. Tr. 49. PT’s argument of distortion is founded on a misunderstanding of the objective of the pooled standard deviation. Using a simple average, certain sales have more effect over the outcome than other sales. However, a different outcome is not a distortion; a distortion requires a different result due to an unreasonable practice. Here, as Commerce’s practice is reasonable within the objective of this stage of the test and achieves a coefficient representative of the average price changes between the two groups, the use of a simple average cannot be said to lead to distortive results here.²³

Finally, PT contends that the use of a simple average in the pooled standard deviation was unreasonable as applied in this case. See PT Br. 34-35. Specifically, PT presents calculations which it argues demonstrate that using an average weighted by sales or quantity in the standard deviation would result in a de minimis dumping margin for PT, while using a simple average resulted in a non-de minimis.dumping margin. See id. Calculations demonstrating this potential impact to this respondent of using a different methodology are insufficient to demonstrate that Commerce’s use of the simple average is unreasonable. PT has not presented evidence demonstrating that the use of a simple average to calculate the pooled standard deviation is unreasonable as applied here. Although understandably PT would prefer the methodology that leads to a de minim-is margin for PT, calculations resulting in a de minimis margin, without more, do not demonstrate that the methodology which did resulted in a non-de minimis margin is unreasonable or distortive.

D. “Double Zeroing”

PT argues that Commerce’s application of the mixed methodology, and specifically its use of “double zeroing” when aggregating the A-to-A and A-to-T results, was contrary to the statute. PT Br. 18-25. Defendant responds that its application of the mixed methodology in this case was consistent with the statute and reasonable because allowing negative A-to-A results to offset positive A-to-T results would defeat the purpose of the methodology. Def.’s Resp. 18-23. Commerce’s application of the mixed methodology is in accordance with law and is reasonable in this case.

Commerce applies the alternate A-to-T methodology based on the percentage of U.S. sales found to pass the Cohen’s d test.²⁴ See Prelim. Decision Memo at 11. When the value of a respondent’s U.S. sales passing the Cohen’s d test accounts for more than 33 percent but less than 66 percent of the value of the respondent’s total U.S. sales, Commerce applies a “mixed methodology” to calculate the weighted-average dumping margin, using the standard A-to-A method (with offsetting dumped sales with non-dumped sales) to calculate the dumping margin for the sales that did not pass the Cohen’s d test and using the alternate A-to:T method (without offsetting dumped sales with non-dumped sales) to calculate the dumping margin for the sales that did pass the Cohen’s d test. H.; Final Decision Memo at 26, Commerce then aggregates the results of the two calculations to arrive at a weighted-average dumping margin for all of the respondent’s U.S. sales. Prelim. Decision Memo at 11; Final Decision Memo at 26. When aggregating the two calculations, Commerce does not offset the dumped sales by the non-dumped sales from either the A-to-A or A-to-T calculation. See Final Decision Memo at 26. PT refers to this practice as “double zeroing.” See PT Br. 19.

Commerce explained that zeroing the non-dumped sales is required to ensure that the masked dumping uncovered by the A-to-T method is preserved when the margin is aggregated with the A-to-A results in the mixed methodology, just as the uncovered masked dumping is preserved when the A-to-T method is applied to all of a respondent’s sales. See Final Decision Memo at 26-27. Commerce explained that allowing the non-dumped sales to instead offset the dumped sales would allow the non-dumped sales to “reduce or completely negate” the masked dumping uncovered by A-to-T, rendering the alternate method provided by the statute useless. Id. at 26. Thus, disallowing offsets when aggregating in the mixed methodology allows Commerce to preserve the statutory alternate remedy while also applying the A-to-T methodology proportionately based on the degree of masked dumping identified. This methodological choice to not offset when aggregating the margins in the mixed methodology is reasonable, as offsetting the dumped sales with the non-dumped sales when aggregating would make it impossible for Commerce to both preserve unmasked dumping uncovered with A-to-T and apply a proportionate remedy.

The reasonableness of this methodology does not depend on the presence of a pattern established for every sale, contrary to PT’s argument. See PT Br. 18-25. PT focuses on the pattern to suggest that only sales within the pattern may be subject to the remedy, including not offsetting non-dumped sales during the aggregation stage in the mixed methodology. See id. at 25. However, Commerce’s test uses the existence of the pattern within the sales as a trigger for applying the alternate method, and Commerce then uses the ratio test to decide what remedy to apply; not allowing offsets in the aggregation step preserves that remedy. The statute provides Commerce with the A-to-T methodology as a tool to capture masked dumping and thereby calculate more accurate margins than would be possible using A-to-A. See 19 U.S.C. § 1677f-1(d)(1)(B); Yangzhou Bestpak Gifts & Crafts Co., Ltd. v. United States, 716 F.3d 1370, 1379 (Fed. Cir. 2013) (“An overriding purpose of Commerce’s administration of antidumping laws is to calculate dumping margins as accurately as possible.”). Commerce’s decision to effectuate the statute by applying the mixed methodology, without offsets during the aggregation stage, to both preserve the masked dumping uncovered with A-to-T and achieve a proportionate remedy is reasonable.

III. G & A Expense Ratio

PT challenges Commerce’s calculation of the General and Administrative (“G & A”) expense ratio within its constructed value calculation. PT Br. 35-42. PT contests Commerce’s determination to include in the calculation of PT’s G & A expense ratio certain costs and expenses generated in the production and sale of steam, which Commerce concluded is a separate line of business unrelated to the production of subject merchandise by PT affiliate, Pro-Team. See id. at 38-39; see also Final Decision Memo at 55. PT notes Defendant’s description of how Commerce calculated PT’s G & A expense ratio in its brief before the court is not consistent with Commerce’s actual calculation.²⁵ PT Reply Br. 15. Alternatively, PT argues that, if Commerce includes costs related to Pro-Team’s separate steam production, subsidy income should offset G & A expenses. Id. at 39-42. Defendant counters that Commerce followed its normal practice of allocating company-wide costs to cost of goods sold (“COGS), Def.’s Resp. 35-36, and Commerce acted reasonably and consistently with its practice in allocating the subsidy income attributable to Pro-Team’s separate steam line of business. Id. at 39-40. Commerce fails to state or explain how its cost allocation methodology could result in allocating certain steam-related costs to G & A expenses, when it claims to have allocated all those costs to COGS. On remand, Commerce must explain how it allocates different costs to the respective components G & A expense ratio and explain why its determination is supported by the record evidence or reconsider its determination.

Commerce calculates a respondent’s dumping margin by determining “the amount by which the [NV] exceeds the export price ... of the subject merchandise.” 19 U.S.C. § 1677(35)(A). NV should normally be calculated based on “the price at which the foreign like product is first sold ... for consumption in the exporting country, in the usual commercial quantities and in the ordinary course of trade.” 19 U.S.C. § 1677b(a)(1)(B)(i). However, when, as here, Commerce determines that the respondent does not have viable home or third-country market sales, the statute directs that Commerce may use a constructed value to calculate a NV for respondent (he., CV).²⁶ See 19 U.S.C. § 1677b(a)(4); 19 U.S.C. § 1677b(a)(1)(B)-(C); see also Prelim. Decision Memo at 13 (stating that Commerce used CV as the basis for NV because PT did not have a viable comparison market).

In calculating constructed value, Commerce is required to include selling, general, and administrative expenses. See 19 U.S.C. § 1677b(e)(3)(1)-(3). The statute does not define selling, general, and administrative expenses. However, G & A expenses are generally understood to mean “expenses which relate to the activities of the company as a whole rather than to the production process.” Torrington Co. v. United States, 25 C.I.T. 395, 431, 146 F.Supp.2d 845, 885 (2001) (internal quotations omitted). The court affords Commerce significant deference in developing a methodology for determining this component of CV because it is a determination “involv[ing] complex economic and accounting decisions of a technical nature.” Fujitsu Gen. Ltd. v. United States, 88 F.3d at 1039. However, Commerce’s “must cogently ■ explain why it has exercised its discretion in a given manner.” State Farm, 463 U.S. at 48-49, 103 S.Ct. 2856.

Commerce states that its practice is to calculate and allocate G&A expenses based on company-wide G&A costs, which are the expenses that “relate to the general operations of the company as a whole” and not specific products and processes.²⁷ Final Decision Memo at 56 (citing Issues and Decision Memorandum for the Final Results of the Antidumping Duty Administrative Review on Narrow Woven Ribbons with Woven Selvedge from Taiwan at 29, A-583-844, (Apr. 6, 2015), available at http://ia.ita.doc.gov/frn/summary/ taiwan/2015-08436-1.pdf (last visited Mar. 20, 2017) (“Ribbons with Woven Selvedge from Taiwan I & D”). Therefore, according to Commerce’s stated practice, the numerator for the G&A expense ratio is the respondent’s expenses attributable to general operations of the company and the denominator is the respondent’s company-wide COGS. See id. Thus, the G & A expense ratio, expressed as an equation is as follows:

[[Image here]]

To compute the per-unit amount of G & A expense, the G & A expense ratio is multiplied by the total costs of manufacture for each product assigned a control number by Commerce.²⁸ See PT Enterprise Section D Response at 26, CD 60-62, bar code 3228714-01-03 (Sept. 16, 2014) (“PT Sec. D. Resp”).

Here, Commerce relied on Pro-Team’s financial statements to calculate the G&A expense ratio in its preliminary determination. See Prelim. Decision Memo at 16. In its initial questionnaire response, Pro-Team calculated its G & A expense ratio by excluding all expenses related to its production and sale of steam from both G & A and COGS.²⁹ See PT Sec. D. Resp. at 21, 34, Exs. D-15, D-16. Following comments from interested parties, Commerce asked Pro-Team to recalculate G & A to include steam related production costs and expenses, and Pro-Team complied. See PT Enterprise Supplemental Section D Response at 11-12, Exs. SD-21, SD-24, CD 79-⅜4, bar codes 3237002-01-05 (Oct. 21, 2014). Pro-Team offset the G & A expenses generated in the production of steam by certain other income. Id. at Ex. SD-24. These adjustments resulted in an increase in Pro-Team’s G&A expense ratio for Commerce’s preliminary determination.³⁰ See Analysis of Data Submitted by PT Enterprise Inc. in Certain Nails from Taiwan at Attach. 2, CD 194, bar code 3248230-01 (Dec. 14, 2014).

In its final determination, Commerce found that some portion of the other income Pro-Team had offset against G & A expenses represented an energy subsidy for steam provided by the government of Taiwan to promote energy production. See Final Decision Memo at 56 (citing PT Cost Verification Rep. at 22). Commerce did not allow the income attributable to this subsidy to offset Pro-Team’s G & A expenses because Commerce concluded “the subsidy is not related to the general operations of the company as a whole, but instead appears directly related to and intended for the company’s other steam line of business.” Id Therefore, Commerce applied this income as an offset to COGS denominator, not the G & A expense numerator. Id.

Commerce does not specify exactly how it allocates costs and expenses attributable to a separate line of business, except to say that “[t]he costs associated with the steam line of business were properly included in the denominator of the G & A expense ratio calculation (i.e,, COGS).”³¹ Final Decision Memo at 55. Defendant states that Commerce allocates all costs and expenses attributable to Pro-Team’s steam business, including research and development (“R & D”) expenses and depreciation expenses attributable to steam production, to COGS (ie., the denominator of the G & A expense ratio). See Def.’s Resp. 32, 37-38. PT correctly points out that Defendant’s statement “does not accurately reflect the manner in which Commerce actually calculated COGS and G & A.” Pl.’s Reply Br. 15, Addendum, Sept. 23, 2016, EOF No. 58; see generally Verification of the Cost Response of PT Enterprise Inc. in the Antidumping Duty Investigation of Certain Steel Nails from Taiwan at 20, Ex. CVE-9, CD 261, bar code 3265886-01 (Mar. 19, 2015) (“PT Cost Verification Rep.”) (demonstrating that Commerce allocated certain expenses, including R & D, depreciation, and other expenses attributable to Pro-Team’s steam line of business, to G & A expenses, and other expenses to COGS). Commerce’s statement of its practice does not address why certain costs, specifically R & D and depreciation, were allocated to G & A while others were allocated to COGS when Commerce claims its practice is'to allocate all expenses not attributable to the administration of the company as a whole, including expenses attributable to the production of non-subject merchandise, to COGS.³² On remand, Commerce must explain its methodology for allocating costs associated with Pro Team’s separate steam business and explain why that methodology is reasonable or reconsider its determination. Commerce does not explain why it allocates certain types of expenses related to steam production to G & A and others to COGS.³³

In the alternative, PT argues that, if Commerce includes costs related to Pro-Team’s separate steam production in G & A, Commerce’s practice must require the income from the subsidy benefiting Pro-Team’s steam production to offset those G & A expenses attributable to its steam production. PT Br. 3£M2. PT contends it would be inconsistent to offset all subsidy income to COGS if some costs attributable to steam are allocated to G & A. Id. Defendant counters that Commerce offset COGS (he., the denominator) by the amount of government subsidy for the production of steam because Commerce determined that the subsidy is unrelated to the general operation of the company. Def.’s Resp. Br. 39 (citing Final Decision Memo at 56).

Defendant contends that Commerce’s practice permits income from a subsidy to offset G & A expenses (i.e., the numerator of the G & A expense ratio) only where the subsidy benefits the company as a whole, not where Commerce specifically finds that a subsidy only benefits a particular line of business. See id. (citing Final Decision Memo at 56). Although Commerce cites this same statement of its practice, Final Decision Memo at 56 (citing Certain Pasta From Italy, 64 Fed. Reg. 6,615, 6,626-27 (Dep’t Commerce Feb. 10, 1999) (final results and partial rescission of antidumping duty administrative review) (treating grants for equipment purchases and loan-restitution payments as offsets to total G & A expenses where Commerce found they relate to the company’s general operations)), the court cannot assess the reasonableness of this practice for offsetting subsidies subsidy income generally or as applied here until Commerce has clarified its practice of allocating all costs not attributable to company-wide administrative expenses, including those relating to non-subject merchandise, to COGS. Only once Commerce has clarified the apparent inconsistencies in its allocation of costs in this investigation and explained the reasonableness of its cost allocation methodology can the court assess if allocating all non-administrative company-wide expenses to COGS is reasonable and supported by the record here. The court defers consideration of this issue.

IV. Transfer Prices Paid by Pro-Team to Tollers for Wire Drawing and Nail Making Services

PT argues that Commerce’s rejection of transfer prices paid by Pro-Team to affiliated tollers for wire drawing and nail making services performed for Pro-Team is not supported by substantial evidence because the prices Pro-Team paid to affiliated and unaffiliated tollers were substantially similar. PT Br. 42-44. Defendant responds that Commerce reasonably disregarded transfer prices paid to affiliated tollers because it determined that the average amount paid to unaffiliated tollers (i.e., average market price) is greater than the average amount paid to affiliated tol-lers (he., average transfer price). Defi’s Resp. 42-44. Commerce’s determination is supported by substantial evidence.

In the calculation of CV, the statute permits Commerce to disregard transactions “directly or indirectly between affiliated persons” if those transactions do not fairly reflect an arm’s-length price. See 19 U.S.C. § 1677b(f)(2). If Commerce disregards affiliated party transactions and no other transactions are available for consideration, the determination of the amount shall be based on the information on the record as to what an arm’s-length price would have been if the transaction had occurred between unaffiliated persons. Id. The statute does not define what it means for affiliated party transactions to not fairly reflect an arm’s-length transaction. See id.

In order to determine whether affiliated party transactions fairly reflect the market price for such transactions, Commerce’s practice is to compare the average transfer prices for affiliated tollers with the average market prices for unaffiliated tollers. See, e.g., Polyethylene Retail Carrier Bags From Thailand, 76 Fed. Reg. 12,700 (Dep’t Commerce Mar. 8, 2011) (final results of antidumping duty administrative review) and accompanying Issues and Decision Memorandum for the Antidumping Duty Administrative Review of Polyethylene Retail Carrier Bags from Thailand for the Period of Review August 1, 2008, through July 31, 2009 at 19, A-549-821, (Mar. 1, 2011), available at http://ia.ita.doc.gov/frn/ summary/thailand/2011-5267-l.pdf (last visited Mar. 20, 2017) (“Polyethylene Retail Carrier Bags from Thailand I & D”). Where transfer prices (i.e., affiliated party transactions) are lower than those made at arm’s length prices (he., unaffiliated party transactions) on a weighted-average basis, Commerce may disregard those transactions as not fairly reflecting the amount usually reflected in sales of merchandise under consideration in the market under consideration. See id.; see generally 19 U.S.C. § 1677b(f)(2).

Here, Commerce reasonably disregarded transactions for low carbon wire drawing and steel nail making services performed by Pro-Team’s tollers because it found the average market price for such services performed by unaffiliated tollers was higher than those of affiliated tollers.³⁴ Final Decision Memo at 53. Commerce did not disregard affiliated party transactions for coating and wire strip collating services performed by tollers because it found the average market price for coating and wire strip collating services performed by affiliated tollers was higher than the price paid to unaffiliated tollers.³⁵ See id.

PT argues that Commerce lacked substantial evidence to disregard the transfer prices paid to affiliated tollers for low carbon wire drawing and steel nail making services because the prices paid to affiliated tollers and those paid to unaffiliated tollers are substantially similar.³⁶ PT Br. 42-43. PT does not contest Commerce’s findings that Pro-Team paid a higher weighted-average price to unaffiliated tol-lers than it did to affiliated tollers for both services. See id. The record therefore indicates that the weighted-average of the prices paid to affiliated tollers was lower than that paid to unaffiliated tollers for such services. See Final Decision Memo at 53; PT Verification Rep. at 19-20. PT does not argue that Commerce’s practice is unreasonable, and Commerce’s determination to follow its practice is supported by record evidence here.

In the alternative, PT argues that even if the weighted-average prices paid by Pro-Team to its affiliated wire drawing and nail making tollers were lower, Commerce abused its discretion in applying its transactions disregarded practice here because the price differences were too small to justify rejecting them. See PT Br. 44. A determination by Commerce is an abuse of discretion if it is clearly unreasonable, arbitrary, or fanciful, is based on an erroneous conclusion of law, rests on clearly erroneous fact findings, or follows from a record that contains no evidence on which Commerce could rationally base its decision. See Gerritsen v. Shirai, 979 F.2d 1524, 1529 (Fed. Cir. 1992). Although the statute does not require Commerce to disregard transactions that do not reflect non-arm’s-length transactions between affiliated purchasers, the statute permits Commerce to do so where the transaction does not fairly reflect an arm’s-length transaction. See 19 U.S.C. § 1677b(f)(2).

PT does not demonstrate why it is unreasonable or arbitrary for Commerce conclude that even slightly lower weighted-average prices paid to affiliated tollers do not fairly reflect the amount usually reflected in such transactions. Nor does PT point to affiliated party transactions for other services where the weighted-average transfer price was lower than for unaffiliated party transactions but Commerce opted not to exclude such prices. The fact that Commerce elected not to exclude transactions to unaffiliated tollers for coating or to affiliated tollers for wire strip collating does not demonstrate Commerce’s determination is arbitrary. See Final Decision Memo at 53. For coating and wire strip collating services performed for tollers, the weighted-average prices for those services performed by affiliated tollers was higher than for the same services performed by unaffiliated tollers. See PT Cost Verification Report at 19-20. Commerce excluded these transactions because the weighted-average prices paid to unaffiliated tollers is higher. See Final Decision Memo at 53; PT Cost Verification Report at 19-20.. It is reasonably discernible that Commerce determined that higher prices paid to unaffiliated tollers reflect an arms-length transaction because affiliates that are not pricing their services at arms-length would reflect lower pricing. See Final Decision Memo at 53; PT Cost Verification Report at 19-20. PT points to no reason why it is irrational or arbitrary for Commerce to conclude that slightly lower weighted-average prices paid to affiliated parties reflect transactions that have not occurred at arm’s length. Therefore, Commerce’s determination to disregard transactions to affiliated tollers performing low carbon wire drawing and steel nail making services is reasonable and not an abuse of discretion.

CONCLUSION

For the foregoing reasons, the U.S. Department of Commerce’s final determination in the investigation of the antidumping duty order covering certain nails from Taiwan is sustained in part and remanded in part.

The final determination is sustained with respect to Commerce’s determinations: (1) that Pro-Team is unaffiliated with the [[ ]] tollers in question; (2) to use the Cohen’s d test within the differential pricing analysis to determine the existence of a pattern of significant price differences; (3) to use a simple average to calculate the pooled standard deviation in the Cohen’s d test of the differential pricing analysis; (4) to not offset dumped sales with non-dumped sales in calculating the respondent’s anti-dumping duty margin using the average-to-transaction methodology; and (5) to disregard transfer prices paid by Pro-Team to certain affiliated tollers in its calculation of normal value (“NV”). The final determination is remanded with respect to Corn-merce’s allocation of expenses associated with Pro-Team’s separate steam line of business.

In accordance with , the foregoing, it is hereby

ORDERED that Commerce’s allocation of expenses associated with Pro-Team’s separate steam line of business for further explanation and consideration consistent with this opinion. On remand, Commerce must:

(1) Explain its methodology for allocating costs associated with Pro Team’s separate steam business and explain why that methodology is reasonable or reconsider its determination, and

(2) Explain how its cost allocation methodology in this case conformed with its practice or reconsider its determination; and it is further

ORDERED that Commerce shall file its remand determination with the court within 45 days of this date; and it is further

ORDERED that Plaintiff shall have 30 days thereafter to file comments on the remand determination; and it is further

ORDERED that Defendant shall have 15 days thereafter to file a reply to comments on the remand determination.

1 . Further citations to the Tariff Act of 1930, as amended, are to the relevant provisions of Title 19 of the U.S. Code, 2012 edition.

2 . The court consolidated Mid Continent’s challenge with an action filed by PT, an individual exporter of steel nails, and Pro-Team Coil Nail Enterprise Inc. (“Pro-Team”), Uni-catch Industrial Co., Ltd., WTA International Co., Ltd., Zon Mon Co., Ltd., Hor Liang Industrial Corporation, President Industrial Inc., and Liang Chyuan Industrial Co., Ltd., producers of steel nails (collectively "PT”). See Order, November 19, 2015, ECF No. 20.

3 . On October 16, 2015, Defendant submitted indices to the public and confidential administrative records, which identify the documents that comprise the public and confidential administrative records to the Commerce's final determination. The indices to the public and confidential administrative records to Commerce’s final determination can be located at ECF No. 17. All further references to the documents from The administrative records are identified by the numbers assigned by Commerce in these administrative records.

4 . Further citations to Title 28 of the U.S. Code are to the 2012 edition.

5 . Commerce determined that PT and Pro-Team are affiliated pursuant to 19 U.S.C. § 1677(33)(F) because PT and Pro-Team "shar[e] a common shareholder,” who "owns a sufficient percentage of PT and Pro-Team to establish control over both companies.” Prelim. Decision Memo at 7. The finding that PT and Pro-Team are affiliated has not been challenged here.

6 .PT indicated to Commerce that Pro-Team uses the services of tolling companies to produce its nails. See, e.g., PT Enterprise Section D Response at 3, CD 61-61, bar code 3228714-01 (Sept. 16, 2014) ("[All production process activities] are done on a tolling basis. Pro-team [organizes all the production. As such, besides providing the raw material, collating materials and packing materials, Pro-team also incurred] labor, water, electricity and other common factory overhead expenses in its normal course of business.”).

7 . Specifically, relevant here, this control may be in the form of "two or more persons [or entities] directly or indirectly controlling, controlled by, or under common control with, any person [or entity]” or "any person [or entity] who controls any other person [or entity] and such other person [or entity].” 19 U.S.C, §§ 1677(33) (F), (G).

8 . The Statement of Administrative Action accompanying the Uruguay Round Agreements Act (“SAA”) explains that control may also exist within corporate groupings, see SAA, H.R. Rep. No. 103-316, vol. 1 at 838, 1994 U.S.C.C.A.N. at 4,175, so "persons” here is considered to also apply to "entities,” Specifically, the SAA explains that an emphasis on control was included in the statutory definition of affiliated persons due to the Administration’s position

that including control in the definition of “affiliated” will permit a more sophisticated analysis which better reflects the realities of the marketplace. The traditional focus on control through stock ownership fails to address adequately modem business arrangements, which often find one firm "operationally in a position to exercise restraint or direction” over another even in the absence of an equity relationship. A company may be in a position to exercise restraint or direction, for example, through corporate or family groupings, franchises or joint venture agreements, debt financing, or close supplier relationships in which the supplier or buyer becomes reliant upon the other.

SAA, H.R. Rep. No. 103-316, vol. 1 at 838, 1994 U.S.C.C.A.N. at 4,175.

9 . Further citations to Title 19 of the Code of Federal Regulations are to the 2013 edition.

10 . The other factors Commerce may consider to find control under 19 C.F.R. § 351.102(b)(3)—corporate or family groupings; franchise or joint venture agreements; debt financing—are not applicable to the relationship between Pro-Team and its tollers. See Final Decision Memo 50-52.

11 . Commerce found that, to the contrary, the record "indicates that there are at least 136 producers/exporters of nails in Taiwan from which the U.S. and comparison-market customers can purchase nails, thereby eliminating any notion of dependence on Pro-Team by its unaffiliated tollers.” Final Decision Memo at 52.

12 . This finding was significant to Commerce, as "[t]he fact that these suppliers were profitable suggests that Pro-Team lacks the ability to control completely the prices at which it purchases services from its tollers.” Final Decision Memo at 52.

13 . Mid Continent further contends that Commerce's determination conflicts with prior practice which, according to Mid Continent, requires Commerce to give great weight to the percentage of the business provided by the tollers to Pro-Team. Mid Continent Br. 20, 24, citing Solid Fertilizer Grade Ammonium Nitrate From the Russian Federation, 65 Fed. Reg. 1139, 1143 (Dep't Commerce Jan. 7, 2000) (preliminary determination of sales at less than fair value). The cited case does not demonstrate the existence of such a practice. Rather it demonstrates only that Commerce has previously determined that "a close supplier relationship may occur when a majority of sales are made to one customer,” Solid Fertilizer Grade Ammonium Nitrate From the Russian Federation, 65 Fed. Reg. 1139, 1143 (Dep’t Commerce Jan. 7, 2000), which is not inconsistent with Commerce's finding in the instant case.

14 . In the conclusion of its moving brief, Mid Continent requests in the alternative that the court remand to Commerce its determination regarding at least [[ ]] of Pro-Team’s [[ ]] purportedly unaffiliated tollers:

To the extent that the Court finds that the evidence does not support [a finding of affiliation] for all [[]] tollers, Mid Continent respectfully submits that the Court should remand the Final Determination to Commerce to find that the record contains substantial evidence to establish that the following [[]] purportedly unaffiliated tollers are affiliated with Pro-Team pursuant to 19 U.S.C, §. 1677(33)(G), and the transactions between these tollers and Pro-Team should be evaluated pursuant to 19 U.S.C. § 1677b(f)(2): [[]].

Mid Continent Br. 42. Mid Continent highlighted facts (taken from Pro-Team's submissions to Commerce) detailing the conditions of each of these tollers' tolling relationship with Pro-Team. Id. at 7-14. These facts include: the nature of the tolling activities provided by each toller; each toller's profit margin; whether the toller or Pro-Team owns the tolling machinery and space; whether the toller has its own employees; the percentage of its business that each toller conducts with Pro-Team; the duration of the tolling relationship; and the tolling fee charged to Pro-Team. IcL Mid Continent also highlighted these facts in its case brief to Commerce. See Petitioners Administrative Case Brief at 57-70, CD 264, PD 285, bar code 3268085-01 (Mar. 31, 2015). Although Commerce did not specifically address any of the [[]] tollers in the Final Decision Memo, Commerce referenced certain.facts highlighted in Mid Continent’s case brief regarding these tollers, Final Decision Memo at 48, n.218, 219, 222, and cited Pro-Team's data for all of the [[ ]] tol-lers, which includes data for these [[ ]] tollers. Final Determination Memo at 50-52; Cost of Production and Constructed Value Calculation Adjustments for the Preliminary Determination—PT Enterprise Inc., Department of Commerce, Attach. 1, CD 200, bar code 3248421-01 (Dec. 17, 2014). Mid Continent does not make explicit why it argues in the alternative for at least these [[ ]] tollers to be found affiliated, though it appears that these [[ ]] tollers conduct the highest percentages of their business with,Pro-Team (with each of them doing either [[ ]]% or [[ ]]% of its business with Pro-Team, while each of the other [[ ]] tollers conducts much lower percentages of their business with Pro-Team). See PT Enterprise Supplemental Section D Response, Ex. SD-12, CD 79-84, PD 139, bar codes 3237002-01-05 (Oct. 21, 2014). Nothing in the details highlighted by Mid Continent undermines Commerce's determination that these [[ ]] tollers are’ unaffiliated.

15 . Defendant does not argue exhaustion on this point. See generally Def.’s Resp.; Confidential Oral Arg. Tr. 63-64, Mar. 2, 2017, ECF No. 82.

16 . In an ADD investigation, Commerce ordinarily determines whether the subject merchandise is being sold in the United States at less than fair value by using the A-to-A methodology. See 19 U.S.C. § 1677f-1(d)(1)(A)(i). Under A-to-A, Commerce compares the weighted-average of the normal value of the merchandise to the weighted-average of the export prices (or constructed export prices) for comparable merchandise. See id. Although the transaction-to-transaction methodology (“T-to-T”), which is "a comparison of the normal values of individual transactions to the export prices of individual transactions,” is also a statutorily preferred method (under 19 U.S.C. § 1677f-1(d)(1)(A)(ii)), Commerce’s regulations provide that T-to-T will be employed only in rare cases, "such as when there are very few sales of subject merchandise and the merchandise sold in each market is identical or very similar or is custom-made.” 19 C.F.R. § 351.414(c)(2).

17 . As Commerce explained,

Purchasers are based on the reported customer codes reported by [mandatory respondents]. Regions are defined using the reported destination (i.e., State) and are grouped into regions based upon standard definitions published by the U.S. Census Bureau. Time periods are defined by the quarter within the [period of investigation] being examined based upon the reported date of sale. For purposes'of analyzing sales transactions by purchaser, region, and time period, comparable merchandise is considered using the product control number and any characteristics of the sales, other than purchaser, region, and time period, that [Commerce] uses in making comparisons between [export price] 'and NV for the individual [antidumping duty] margins. Prelim. Decision Memo at 10-11. To calculate a Cohen’s d coefficient for a particular test group (all sales of the comparable merchandise to a specific purchaser, region, or time period), the test group and comparison group (all other sales of the comparable merchandise) must each have at least two observations and the sales quantity for the comparison group must account for at least five percent of the total sales quantity of the comparable merchandise. See Prelim. Decision Memo at 11.

18 . The pooled standard deviation is derived using the simple average of the variances in the net prices within the test and comparison groups. See Final Decision Memo at 28-29; Def.’s Resp. 14.

19 . For example, PT notes that, in one case, a price difference of $0.0029 was found to be significant. See PT Br. 29.

20 . Commerce explained how the Cohen's d test identifies the significance of price differences:

The Cohen’s d coefficient measures the difference in the weighted-average prices between the test group and the comparison group relative to the distribution of prices within each group (i.e,, the variance or standard deviation). As a result, if prices within the test and comparison groups differ by only small amounts, then the variance within each group is small and there only needs to be a proportionally small difference in the weighted-average prices between the test group and the comparison group to identify a significant difference. Likewise, if there would be a wide dispersion of prices within either the test group or the comparison group, then a difference between the weighted-average prices between the test group and the comparison group would have to be correspondingly larger for the Cohen’s d test to identify this difference to be significant.

Final Decision Memo at 30.

21 . PT claims that the use of simple averages in this stage of the differential pricing analysis is inconsistent with Commerce’s use of weighted-averages elsewhere in the differential pricing analysis, see PT Br. 31-32, and contends that "there is a reason why [Commerce] use[s] weight averages in a lot of other places in this test and it doesn't make sense why they wouldn’t use it” in this stage of the test as well. Confidential Oral Arg. Tr. 49, Mar. 2, 2017, ECF No. 82. Defendant acknowledges that Commerce generally uses weighted-averages. Irk at 40. But Commerce is not required to use a weighted-average in this part of the test simply because a weighted-average is used at another part of the test. Each stage of the test has a different objective, and Commerce should utilize the methodology it determines best achieves the objective of each, regardless of the methodology used at a different stage. The relevant inquiry is whether Commerce acted reasonably to achieve the objective of each stage.

22 . PT explains this practice in more detail:

The standard deviation of each group assesses how much prices fluctuate within each group and is based on a weighted-average by quantity of the prices within each group. The higher the standard deviation, the more prices within the group fluctuate. The lower the standard deviation, the more stable the prices are. The Department then calculates what it refers to as the pooled variance or pooled standard deviation which is essentially a simple average of the standard deviations of the test group and the [comparison] group.

PT Br. 14. PT noted that square roots are used in the calculation of the standard deviation so that price changes above the mean do not cancel out price changes below the mean, and vice versa. See id. at 14, n.6.

23 . Commerce also explains that weight-averaging is inferior to simple averaging because it would allow respondents to potentially manipulate the data. Final Decision Memo at 29. PT argues that it is speculative that a respondent could or would manipulate by observations and that, regardless, it would not be possible to manipulate by quantity of sales. PT Br. 33-34. Defendant conceded at oral argument that it would not be possible to manipulate by quantity of sales. Oral Arg. Tr. 40-41. (Defendant-Intervenor disagreed, contending that it would also theoretically be possible for a respondent to manipulate its reporting of quantity. See Oral Arg. Tr. 41-43). Commerce’s argument that weight-averaging is inferior due to the possibility of manipulation therefore seems problematic, but the court finds it is a secondary argument subsumed within the court’s determination that the use of a simple average here is reasonable.

24 . Commerce makes this determination through conducting its “ratio test”:

The "ratio test” assesses the extent of the significant price differences for all sales as measured by the Cohen's d test. If the value of sales to purchasers, regions, and time periods that pass the Cohen’s d test accounts for 66 percent or more of the value of total sales, then the identified pattern of [export prices] that differ significantly supports the consideration of the application of the A-to-T method to all sales as an alternative to the A-to-A method. If the value of sales to purchasers, regions, and time periods that pass the Cohen’s d test accounts for more than 33 percent and less than 66 percent of the value of total sales, then the results support consideration of the application of an A-to-T method to those sales identified as passing the Cohen’s d test as an alternative to the A-to-A method, and application of the A-to-A method to (hose sales identified as not passing the Cohen’s d test. If 33 percent ór less of the value of total sales passes the Cohen's d test, then the results of the Cohen's d test do not support consideration of an alternative to the A-to-A method.

Prelim. Decision Memo at 11.

25 . In its reply brief, PT points out that Commerce did not allocate all expenses attributable to steam to cost of goods sold, as Defendant says Commerce did. PT Reply 15 (citing Def.’s Resp. 38). Rather, PT claims that Commerce allocated research and development and depreciation costs to G & A expenses, not to cost of goods sold. Id. (citing id. at Addendum). PT argues that, to the extent steam expenses are properly included, they should all be allocated to COGS to be consistent with Commerce’s practice, as described by Defendant. See id. at 17, n 7.

26 . The statute provides that CV of imported merchandise is equal to the sum of: (1) the cost of materials of fabrication or other processing of any land in producing the merchandise; (2) some representation of the amounts incurred and realized for selling, general, and administrative expenses and for profits in connection with the production and sale of merchandise for consumption in the foreign country; and (3) packing and other expenses incidental to placing the subject merchandise in condition packed and ready for shipment to the United States. 19 U.S.C. § 1677b(e)(1)-(3). If actual data are not available for selling, general, and administrative expenses, then Commerce may calculate selling, general and administrative expenses based on: (1) the actual amounts incurred and realized by the specific exporter Or producer for selling, general, and administrative expenses in connection with the production and sale, for consumption in the foreign country, of merchandise of the same general category of products as the subject merchandise; (2) the weighted-average of the actual amounts actually incurred and realized by other exporters or producers subject to the investigation; or (3) based on any other reasonable method. 19 U.S.C. § 1677b (e) (2) (B) (i)-(iii).

27 . Defendant restates Commerce's practice in materially identical terms. See Def.'s Resp. 8.

28 . In its Section D questionnaire, Commerce defines G&A expenses as

those period expenses which relate indirectly to the general operations of the company rather than directly to the production process. G&A expenses include amounts incurred for general R & D activities, executive salaries and bonuses, and operations relating to your company’s corporate headquarters.

PT Enterprise Section D Response at 26, CD 60-62, bar code 3228714-01-03 (Sept. 16, 2014). Commerce does not define COGS either in its questionnaire or in its final determination. See id.; Final Decision Memo.

29 .The resulting G&A ratio calculated by Pro-Team was [[ ]]%. PT Sec. D. Resp. at Ex. D-16.

30 . The resulting G & A ratio increased to [[ ]]%. See Analysis of Data Submitted by PT Enterprise Inc. in Certain Nails from Taiwan at Attach. 2, CD 194, bar code 3248230-01 (Dec. 14, 2014).

31 . Because PT does not have a viable home or third-country market, Commerce concluded that it lacked a comparison market for selling expenses to use in its CV calculations. Prelim. Decision Memo at 16. Commerce used PT’s financial statements to calculate the selling, general, and administrative expenses component of CV. Id.; see also Final Decision Memo at 55; 19 U.S.C. § 1677b(e)(2)(B)(iii).

32 .Defendant states that Commerce has a practice of including R & D activities in COGS where these expenses are not considered to be part of a company's general expenses in the calculation of its G &■ A expense ratio, Def.’s Resp. Br. 37. However, neither Defendant nor Commerce cites any practice that either states explicitly that R & D expenses attributable to non-subject merchandise are allocated to COGS or explaining why it is reasonable to do so.

33 . Missing from Commerce's analysis and Defendant's brief is an explanation why R & D and depreciation expenses not attributable to production of subject merchandise (i.e., cost of manufacture) and that cannot be attributed to company-wide administration (i.e., G & A) are allocated to COGS. The proceedings cited by Defendant, focus on the propriety of including certain expenses in G & A ■ (he., the numerator of the G & A expense ratio), but they do not address why it is reasonable to include all other company-wide expenses in COGS (he., the denominator of the G & A expense ratio). See Issues and Decision Memorandum for the Antidumping Duty Administrative Review on Certain Frozen Warmwater Shrimp from India—August 4, 2004, through January 31, 2006 at 15-18, A-533-850, (Sept. 5, 2007), available at http:// ia.ita.doc.gov/fm/summary/india/E7-18006-l. pdf (last visited Mar. 20, 2017) (addressing the propriety of including expenses attributable R & D, supply, support and chain management, compensation, and other voluntary retirement schemes in the numerator of the G & A expense ratio); Issues and Decision memorandum for the Final Determination in the Antidumping Duty Investigation of Certain Lined Paper Products from India at 9-10, A-533-843, (Aug. 8, 2006), available at http:// ia.ita.doc.gov/fm/summary/india/E6-12811 -1. pdf (last visited Mar. 20, 2017) (addressing why G & A expenses need to reflect company-wide administrative costs and why certain expenses attributable to non-subject merchandise should not be allocated to G & A, but not why it is reasonable to allocate those expenses to COGS).

34 . Commerce compared tolling services purchased by Pro-Team from affiliated and unaffiliated suppliers, which included wire drawing, nail making, threading, coating, wire strip collating. See PT Enterprise Cost Verification Rep at 19-20. Commerce determined that for wire drawing of low carbon steel, the average transfer price paid to affiliated tollers was [[ ]] while the average price paid for such services purchased from unaffiliated tollers was [[]]. Id. at 19. For steel nail making services, the average price paid to affiliated tollers was [[]] while the average price paid to unaffiliated tollers was [[]]. Id.

35 . For coating services, the average price paid to affiliated tollers was [[]] while the average price paid to unaffiliated tollers was [[ ]]. PT Enterprise Cost Verification Report at 19. For wire strip collating services, the average price paid to affiliated tollers was [[]] while the average price paid to unaffiliated tollers was [[ ]]. Id. at 20.

36 . Specifically, PT argues that, for wire drawing of low carbon steel, the affiliated party transfer price of [[]] is "only [[]] percent less than the [[]] price paid to the unaffiliated company responsible for [[ ]] percent of wire drawing.” PT Br. 42, For nail making, PT argues "the prices paid to affiliated tollers ( [[ ]] and [[ ]]) were greater than prices paid to . two unaffiliated tollers ([[]]) and merely [[ ]] percent less than the average. Ich