OPINION

MUSGRAVE, Senior Judge:

Plaintiffs Gold East Paper (Jiangsu) Co., Ltd. ("Gold East"), Ningbo Zhonghua Paper Co., Ltd., and Global Paper Solutions ("GPS") (hereafter "Plaintiffs" or "APP-China") challenge the Department of Commerce's ("Commerce") final determination in the antidumping investigation of Certain Coated Paper Suitable for High-Quality Print Graphics Using Sheet-Fed Presses from the People's Republic of China, 75 Fed.Reg. 59217 (Dept. Commerce, Sept. 27, 2010) Public Record Doc. ("PR") 360, as amended by Certain Coated Paper Suitable for High-Quality Print Graphics Using Sheet-Fed Presses from the People's Republic of China: Amended Final Determination of Sales at Less than Fair Value and Antidumping Order, 75 Fed. Reg. 70203 (Dept. Commerce, Nov. 17, 2010), PR 369 (collectively, "Final AD Order").

In their second amended complaint, plaintiffs press multiple causes of action, which are addressed below. Second Amended Complaint, dated December 19, 2012, and filed on January 10, 2013, ECF Doc. 114-1, at 9-12; see also Plaintiff's Reply Brief ("Pl's Reply") at 1 (identifying which arguments plaintiff is no longer pursuing). Plaintiffs move for judgment under Rule 56.2, challenging the Final AD Order in five sections of their brief. Respondent Plaintiffs' Brief in Support of Their Motion for Judgment on the Agency Record (Confidential) ("Pl's Br.") at 1-4.

Defendant-Intervenors Appleton Coated LLC, Newpage Corp., SAPPI Fine Paper North America, et al. ("Appleton") cross-move for judgment under Rule 56.2, alleging that Commerce misclassified certain U.S. sales as Export Price transactions, not Constructed Export Price transactions due to the sales' locations and delivery terms. Appleton also claims that Commerce erred in determining the surrogate wage rate for purposes of its Non-Market Economy Normal Value calculations.

The government defends Commerce's findings generally, but does agree to a voluntary remand in order to review the computer programming and to make any changes required to correct any errors found. Due to the multitude of issues involved, the relevant facts are discussed with each issue separately.

JURISDICTION AND STANDARD OF REVIEW

The court has jurisdiction pursuant to 28 U.S.C. § 1581(c). Commerce's final determination will be upheld unless it is found "to be unsupported by substantial evidence on the record, or otherwise not in accordance with law." 19 U.S.C. § 1516a(b)(1)(B)(i). More specifically, when reviewing agency determinations, findings, or conclusions for substantial evidence, the court assesses whether the agency action is reasonable given the record as a whole. Nippon Steel Corp. v. United States, 458 F.3d 1345, 1350-51 (Fed.Cir.2006).

Separately, the two-step framework provided in Chevron, U.S.A., Inc. v. Natural Res. Def. Council, Inc., 467 U.S. 837, 842-45, 104 S.Ct. 2778, 81 L.Ed.2d 694 (1984), governs judicial review of Commerce's interpretation of the antidumping statute. See United States v. Eurodif S.A., 555 U.S. 305, 316, 129 S.Ct. 878, 172 L.Ed.2d 679 (2009) (Commerce's "interpretation governs in the absence of unambiguous statutory language to the contrary or unreasonable resolution of language that is ambiguous"). Commerce's interpretation will not be set aside unless it is "arbitrary, capricious, or manifestly contrary to the statute." Chevron, 467 U.S. at 844, 104 S.Ct. 2778.

ANALYSIS

I. Commerce's Refusal to Use Market Economy Input Prices where Such Inputs did not Constitute More than 33 Percent of Plaintiffs' Purchases

In order to determine the proper antidumping margin, Commerce valued the raw material inputs used in the manufacture of the merchandise. Commerce decided not to use the market-economy ("ME") price paid for those inputs where ME purchases made up less than 33% of those inputs. Instead, Commerce used a price based on the weighted-average ME price plus a surrogate value for the non-market economy ("NME") purchases. Cmt. 18, Issues & Decision Memorandum ("I & D Memo") (Sept. 20, 2010) PR 353 at 46. The record shows that 32.9% of certain disputed inputs were ME purchases. Pl's Br. at 23 (chart).

Commerce does not expressly defend the decision, but does defend the 33% policy as an "objective benchmark" which provides "consistency and predictability" in Commerce's determinations. Defendant's Response to Plaintiffs' and Defendant-Intervenors' Separate Motions for Judgment Upon the Administrative Record ("Deft's Br.") at 44. Its brief summarizes the administrative history of the 33% policy and suggests that APP-China could have but failed to prove to Commerce that "case specific facts favor the application of a different threshold".1 In the I & D Memo, Commerce cites to prior cases where it applied the 33% policy, but a review of those cases does not reveal how close to 33% the ME input purchases were. See I & D Memo at 46, n. 140-141.

Under Commerce's regulations, it will "normally" use prices paid by NME producers to ME suppliers to value factors of production in Normal Value calculations. See 19 C.F.R. § 351.408(c)(1). Commerce's policy implementing § 351.408(c)(1) is to use the ME purchase price to value the particular factor where the ME purchases are a "significant" or "meaningful" portion of the whole. See Antidumping Methodologies: Market Economy Inputs, etc., 71 Fed.Reg. 61716, 61716-19 (Dep't Comm. Oct. 19, 2006). Under this policy, the ME inputs constitute a "significant" or "meaningful" portion where they make up more than 33% of purchases of the input. Id., 71 Fed.Reg. at 61717-18. Above the 33% threshold, Commerce will use the ME price to value the entire quantity of inputs. Below that threshold Commerce will use a weighted average of the ME price for the ME portion and a surrogate value for the remainder as it did in this case.

The government argues that Shakeproof Assembly Comp. Div. of Ill. Tool Works, Inc. v. United States, 268 F.3d 1376 (Fed. Cir.2001), which predates Commerce's 33% policy, "upheld Commerce's determination that actual input prices will constitute the `best information available' only if they are found in a `meaningful' quantity." Deft's Br. at 48. That court stated:

Shakeproof, 268 F.3d at 1382. In Shakeproof, Commerce argued that the ME data was more accurate that the surrogate data. Id. The Shakeproof court agreed.

Shakeproof, 268 F.3d at 1382, quoting Lasko Metal Products, Inc. v. United States, 43 F.3d 1442, 1446 (Fed.Cir.1994). The court finds the reasons why Commerce sought to use the ME data in Shakeproof compellingly favor APP-China's position in this case.

The Oxford English Dictionary defines "meaningful" (as applied to data or its presentation), as "accurate and realistic; of practical use."2 Commerce's policy, delineated following the Shakeproof decision, defines 33% as a "meaningful" and "significant" amount. The court cannot understand why, in Commerce's view, purchases of 33% of an input are "meaningful", but purchases of 32.9% are not. There is no meaningful distinction for purposes of determining Normal Value between those two quantities. Commerce itself has said that the 33% rule was not a "rigid, `bright line' threshold". Antidumping Methodologies: Market Economy Inputs, 71 Fed.Reg. at 61718.

The court finds that Commerce's refusal to value the classes of inputs using the ME price paid for 32.9% of those inputs, where it would have done so if APP-China had made ME purchases of 33% of the input, is unreasonable, arbitrary and capricious under the circumstances. This issue is therefore remanded to Commerce to recalculate the affected inputs using the ME prices paid rather than the weighted averages previously used and to recalculate the affected margins accordingly.

II. Commerce's Refusal to Recognize Market Economy Purchases of Inputs from Thailand and Korea

Plaintiffs challenge Commerce's refusal to recognize APP-China's raw material imports from Korea and Thailand as bona fide market economy purchases, or to utilize their prices to value the raw material input. Commerce made this decision because it believed that such inputs may themselves have been sold with the benefit of subsidies. Deft's Br. at 16. Plaintiffs claim that this refusal is contrary to this court's precedent, citing Fuyao Glass Indus. Group Co. v. United States, 29 CIT 109, 2005 WL 280437 (2005) and Sichuan Changhong Elec. Co. v. United States, 30 CIT 1481, 460 F.Supp.2d 1338 (2006).

Under 19 U.S.C. § 1677b(c)(1) and 19 C.F.R. § 351.408(c)(1), Commerce normally will use the price paid to a market-economy producer to value a factor in valuing products exported from a NME such as the PRC. Fuyao Glass and Sichuan Changhong both questioned Commerce's refusal to use raw material values from Korea and Thailand due to Commerce's suspicions that those prices were tainted by possible export subsidies.

In Fuyao Glass, the court identified three factors to determine whether Commerce had a proper evidentiary basis to believe or suspect that the prices may have been subsidized. On remand, the court ordered Commerce to justify the refusal to accept the raw material prices by demonstrating:

In Sichuan Changhong the court considered Commerce's finding that Korean and Thai inputs may have been subsidized and ordered Commerce to make findings under the Fuyao Glass criteria. Sichuan Changhong, 460 F.Supp.2d at 1350-51. Upon remand, Commerce reopened the record and provided relevant evidence of the alleged subsidy programs.4

In this case, APP-China argued that Commerce must cite evidence establishing that the particular inputs were subsidized in fact. Commerce disagreed.

Cmt. 17, I & D Memo at 44-45 (footnotes omitted).

Commerce tries to distinguish Fuyao Glass because there the court focused on Commerce's statement that it had found that prices were, rather than may have been, subsidized in the countries in question.5 The court does not find Commerce's argument persuasive. The concerns raised in Fuyao Glass and echoed in Sichuan Changhong resonate in this case. Although Commerce need not perform "a formal investigation" whether prices are subsidized, there must be some positive evidence on the record to permit the court to evaluate whether Commerce's decision is supported by substantial evidence. As the record currently stands, there is insufficient evidence to support Commerce's refusal to use the Thai and Korean price data. Therefore, under the Fuyao Glass and Sichuan Changhong precedent, the court orders Commerce on remand to reopen the record and make particularized findings in support of its decision to ignore the Thai and Korean price data (specifically referring to the three criteria from Fuyao Glass), or to reverse its decision not to use such price data and to recalculate the margin accordingly.

III. Commerce's Use of Non-Market Economy Methodology

Gold East argued at Commerce that it was a Chinese "MOE" ("market oriented enterprise"), and that the agency should therefore not have applied NME methodology to calculate Normal Value. Gold East's Request for MOE Treatment (Jan. 21, 2010) at 1-2, PR 107. Gold East also submitted unsolicited information purporting to allow Commerce to apply ME Normal Value methodology. Market Oriented (MOE) Questionnaire Responses, PR 244; Gold East Case Brief at 80, PR 333.

Commerce rejected Gold East's positions. The agency stated in the I & D Memo:

I & D Memo at 30-31. Plaintiffs claim that Commerce should have granted Gold East MOE treatment because it had earlier stated in the Federal Register that it might "grant an individual respondent in China market-economy treatment." See Antidumping Methodologies in Proceedings Involving Certain Non-Market Economies: Market-Oriented Enterprises, 72 Fed.Reg. 60649, 60650 (Dept. of Commerce, Oct. 25, 2007). The government counters that "Commerce considered APP-China's request to be considered for market-economy treatment, and determined that the `available information' did not allow it to calculate normal value under market economy principles." Deft's Br. at 23, citing I & D Memo at 5-10.

Deft's Br. at 28. The court agrees that APP-China's demand that Commerce treat it as an MOE or that it should have used ME methodology was premature, and Commerce's determination on this issue is reasonable.

IV. Targeted Dumping Issues

Commerce concluded that APP-China engaged in "targeted dumping", i.e., that it sold its merchandise at export prices that differed significantly among purchasers.6 APP-China argues that Commerce incorrectly found targeted dumping, and unlawfully applied the targeted dumping remedy to all of its sales.

a. Withdrawn Targeted Dumping Regulation

APP-China's first argument is that Commerce failed to follow a portion of its targeted dumping regulation found at 19 C.F.R. § 351.414(f) (2007), which Commerce withdrew in 2008. Pl's Br. at 27; see also Cmt. 3, I & D Memo at 21-24, citing Withdrawal of Regulatory Provisions Governing Targeted Dumping in Antidumping Duty Investigations, 73 Fed.Reg. 74930 (Dec. 10, 2008) ("Withdrawal Notice"). The withdrawn regulation provided in part that where Commerce found targeted dumping, it would "normally" "limit the application of the average-to-transaction method to those sales that constitute targeted dumping". 19 C.F.R. § 351.414(f)(2) ("Limiting Rule"). APP-China argues that Commerce did not comply with Administrative Procedure Act's ("APA"), 5 U.S.C. § 500, et seq., notice and comment requirements in withdrawing the Limiting Rule, and the withdrawal was ineffective because it did not fall within the exceptions to the APA's notice and comment requirements. Pl's Br. at 27.

Commerce argues that the withdrawal sufficiently complied with the APA because two earlier notices had requested comments on related issues. Deft's Br. at 48-9, citing Targeted Dumping in Antidumping Investigations, 72 Fed.Reg. 60651 (Oct. 25, 2007) ("First Comment Request"), and Proposed Methodology for Identifying and Analyzing Targeted Dumping in Antidumping Investigations, 73 Fed.Reg. 26371 (May 9, 2008) ("Second Comment Request").

The APA requires notice of proposed rulemaking (including withdrawal of regulations) to be published in the Federal Register and to include "either the terms or substance of the proposed rule or a description of the subjects and issues involved." 5 U.S.C. § 553(b)(3). This notice must be "sufficiently descriptive to provide interested parties with a fair opportunity to comment and to participate in the rule making." Chocolate Mfrs. Ass'n of U.S. v. Block, 755 F.2d 1098, 1104 (4th Cir.1985) (citations omitted). The APA provides that prior publication of notice and comment shall not be required:

5 U.S.C. § 553(b)(3)(B). The good cause exception is to "be narrowly construed and only reluctantly countenanced." New Jersey v. EPA, 626 F.2d 1038, 1045 (D.C.Cir. 1980).

The Withdrawal Notice and the two Comment Requests curiously do not make reference to each other even though they were all issued within a single year on related topics. The two Comment Requests discuss the methodologies that Commerce will use to determine whether targeted dumping has occurred, while the Limiting Rule restricts Commerce's ability to impose the targeting remedy across all sales. Therefore, the court finds that the First and Second Comment Requests failed to provide interested parties with the adequate notice and comment before Commerce withdrew the Limiting Rule.

Although Commerce gave no notice before it withdrew the Limiting Rule, Commerce argues that the Withdrawal Notice should be deemed adequate under the APA because it provided for post-publication comments. Deft's Br. at 51. However, the APA requires otherwise. Section 553 provides "that notice and an opportunity for comment are to precede rulemaking." See Air Transport Ass'n of America v. Dept. of Transp., 900 F.2d 369 (D.C.Cir. 1990) remanded, 498 U.S. 1077, 111 S.Ct. 944, 112 L.Ed.2d 1033 (1991), vacated as moot, 933 F.2d 1043 (D.C.Cir.1991) (rejecting FAA's contention that its response to comments after promulgation of rule cured any noncompliance with section 553).

The government cites Federal Express Corp. v. Mineta, 373 F.3d 112 (D.C.Cir. 2004), where the agency issued four rules seriatim over the course of less than a year, but only requested comments after each rule was issued. That case does not support Commerce's actions here. In Federal Express, there was an urgency to the rules, which affected airlines' compensation for losses suffered as a consequence of the shutdown of U.S. airspace following the attacks of September 11, 2001. The affected parties commented on each version of the rules. Furthermore, "the agency [] made a `compelling showing,' that it provided `a meaningful opportunity to comment' before the Fourth Final Rule became effective". Federal Express, 373 F.3d at 120. In this case, there was no urgency to Commerce's withdrawal of the Limiting Rule, and Commerce did not make any further statements regarding the Limiting Rule in response to any comments it received from the Withdrawal Notice.

Commerce argued in the Withdrawal Notice that the withdrawal did not require notice and comment under the "good cause" exception. In the Withdrawal Notice, Commerce cites three "good cause" reasons to ignore the APA notice and comment requirements in withdrawing the Limiting Rule.

Withdrawal Notice, 73 Fed.Reg. at 74931. The Withdrawal Notice claims that notice and comment was not required because it was "contrary to the public interest."7 Citing National Customs Brokers and Forwarders Assn. of America v. United States, 59 F.3d 1219 (Fed.Cir.1995) ("NCBFAA"), the government argues that Commerce was justified in the immediate withdrawal of the Limiting Rule. "Immediate withdrawal of the regulation was ... necessary to allow domestic industries to take advantage of their statutory remedies." Deft's Br. at 56. However, in NCBFAA Congress changed the law and Customs acted expeditiously to implement the new requirements via an interim regulation. In this case, Commerce withdrew the Limiting Rule with immediate effect, in order to "implement" a statute that had been in place for 14 years. Cf. Levesque v. Block, 723 F.2d 175, 185 (1st Cir.1983) (generalized interest in fiscal savings or other efficiencies insufficient support for public interest exception).

The court finds that none of Commerce's reasons in support of immediate revocation (without prior notice and comment) rise to the level required. That Commerce improvidently enacted rules without adequate experience of how they would work, that the rules apply to ongoing investigations, and the rules could deny relief to domestic industries, do not rise to the level required for it to avoid the APA's requirements. Indeed, those justifications could apply to almost any rule promulgated by the agency.

This situation is closely analogous to that found in Citibank, Federal Sav. Bank v. FDIC, 836 F.Supp. 3 (D.D.C.1993), where the FDIC withdrew without providing notice or comment a rule regarding disposition of reserves held by FDIC on behalf of member banks. Several years later, Citibank sued to have the rule enforced, and the court agreed.

Citibank, 836 F.Supp. at 7. Because Commerce failed to provide notice and comment before withdrawing the Limiting Rule, and the agency failed to provide adequate cause to qualify under the exceptions to the notice and comment requirements, the court finds that the repeal of the regulation was invalid, and the Limiting Rule is still in force. Commerce's decision to apply the targeted dumping remedy to all of APP-China's sales failed to comply with applicable law.

Commerce must, on remand, reconsider its application of the targeted dumping remedy under the Limiting Rule. Assuming the finding of targeted dumping remains positive after reconsideration of the other issues addressed in this opinion, Commerce must limit application of the targeted dumping remedy to the targeted sales, or provide an adequate explanation why the situation is not a "normal" one before applying the remedy to all APP-China sales.

b. Targeted Dumping Testing

APP-China argues that Commerce's test for targeted dumping was unsupported by substantial evidence. "Commerce created a test so complex that Commerce itself failed to apply its own test correctly." Pl's Br. at 40. APP-China alleges that Commerce incorrectly applied the test derived from the decision in Mid Continent Nail Corp. v. United States, 712 F.Supp.2d 1370, 1377-78 (CIT 2010) ("Nails test"). In the first part of the Nails test, Commerce analyzed averages of prices, rather than individual prices, and calculated an average price and a standard deviation based on customer specific average prices.

APP-China argues that this violates 19 U.S.C. § 1677f-1(d)(1)(B)(i), because the statute distinguishes between "export prices" and "weighted average export prices", and Commerce used an average where the statute references only "export prices". See e.g., 19 U.S.C. § 1677a(a). Commerce should be required to analyze targeted dumping using individual transaction prices rather than averages. Pl's Br. at 41. APP-China alleges that "the prices to the alleged target did not pass the first part of the [Commerce] test when using actual prices rather than constructed averages." Id. at 42 (emphasis in original).

Commerce responds that it applied the Nails methodology accurately by using weighted average prices. Deft's Br. at 69. In the Second Comment Request, Commerce explained the Nails test. In the first step of the test, "[t]he calculation of the standard deviation value would be done product by product ... using period of investigation[]-wide average prices (weighted by sales value) for each allegedly targeted customer and each distinct non-targeted customer." Second Comment Request, 73 Fed.Reg. at 26372.

The court agrees that Commerce correctly applied this portion of the Nails test, as described in the Second Comment Request. The statute does not require otherwise, despite plaintiffs' statutory construction arguments. Therefore, the court upholds this portion of Commerce's determination as within its discretion in interpreting the statute involved.

c. Programming Errors

Under the second part of the Nails test, Commerce found "significant" price discrepancies whenever a non-targeted customer had higher-than-average prices. APP-China argues that Commerce's computer code did not correctly compare the average price to the next-higher average price to a non-targeted customer as required by the Nails test. Pl's Br. at 45, citing I & D Memo at 22. Instead, the program searched for the lowest weighted average price to a non-targeted customer with a larger than average price gap. Because Commerce actually never compared the alleged targeted price to the "next higher" price, as it was supposed to, APP-China argues that Commerce incorrectly identified targeted dumping in 8 of 8 examples used to justify the targeted dumping finding. Pl's Br. at 46.

APP-China next alleges that Commerce failed to determine the average price gap for the entire group of non-targeted customers. Pl's Br. at 47-8. This is a critical part of determining whether the targeted price gap represented an aberration. Instead, Commerce determined the average price gap for only customers with higher average prices. This approach increased the size of the price gap significantly and switched the result of the test from negative to positive. Id. According to APP-China, Commerce also ignored data from non-targeted customers with average prices below the price to the alleged targeted customer. Pl's Br. at 49. Commerce used data from all the non-targeted customers during the first step in its targeting analysis — the "pattern" test. But Commerce excluded lower-priced non-targeted customers in analyzing whether the prices differed significantly. Finally, APP-China argues that Commerce used methodologies derived from a small sample, and then applied the results to the entire data set. Commerce used data relating to shipments of only 1,005 metric tons (out of 53,809 MT total shipped). Pl's Br. at 50, citing its Complete Output exhibit.

Pl's Br. at 51 (emphasis in original, citation omitted).

Appleton argues that the program contains another error "that artificially enhanced the `gap' between targeted and non-targeted prices so as to make targeting more difficult to find." Appleton Supplemental Brief, ECF Doc. 80 at 6.

Commerce requests a voluntary remand to "examine the calculation program, and if appropriate, to correct the alleged errors and reconsider the finding that export prices differed significantly among purchasers." Deft's Br. at 75.

The court agrees with the parties and remands to Commerce with instructions to review each of the computer programming issues identified above, to recalculate the margins after correcting any errors found, or to explain why it believes the errors do not exist. See SKF USA, Inc. v. United States, 254 F.3d 1022, 1027-31 (Fed.Cir. 2001) (explaining circumstances where voluntary remand requests will be granted).

V. Double-Counted Rebate

APP-China argues that Commerce erred in double-counting a rebate in the calculation of its net U.S. price. Plaintiffs contend that its revised U.S. sales database included data for the "Program for Growth" rebate twice. Commerce responds that APP-China's submissions did not adequately demonstrate that APP-China was entitled to the adjustment. Deft's Br. at 85. Appleton argues that APP-China based its "double-count" claim on tardy information which was unreliable. Appleton's Resp. Br. at 55-56.

APP-China presented rebate information in two documents, one of which was an untranslated document which Commerce could not read. Deft's Br. at 86. Commerce denied the requested adjustment because it could not determine how the rebate amount was calculated. The court agrees with Commerce that administrative exhaustion principles apply in this instance. Gerber Food (Yunnan) Co. v. United States, 601 F.Supp.2d 1370, 1379 (CIT 2009). APP-China could have submitted the information to Commerce in a more timely, accurate and legible manner. The court finds that APP-China failed to meet its burden to show that it was entitled to an adjustment before Commerce, and this aspect of Commerce's decision is supported by substantial evidence.

VI. Export Price vs. Constructed Export Price Analysis

APP-China classified its sales as export price ("EP") transactions, which Commerce accepted. 19 U.S.C. § 1677a(a). According to Appleton, because some of the goods were delivered in the U.S. on a DDP (delivered duty-paid) or DDU (delivered duty-unpaid) basis, the sales should have been found to have been made in the U.S. Appleton Br. at 5, 13. If the sales were made in the U.S., they should have been classified as Constructed Export Price ("CEP") transactions and downward adjustments to the CEP prices made under 19 U.S.C. § 1677a(d). 19 U.S.C. § 1677a(b). In response to these arguments, Commerce held that:

The government defends Commerce's decision by arguing that the date of the invoice (which establishes the price and quantity) is critical under the terms of the statute and its regulations, especially where that date is prior to the date of importation. Deft's Br. at 89, citing 19 C.F.R. § 351.401(i). But an invoice date prior to import may indicate either an EP or CEP sale for purposes of 19 U.S.C. § 1677a, because both EP and CEP sales can be made prior to importation.

The parties argue that the decisions in AK Steel, Corus Staal and Nucor should settle the issue of whether the sale of the goods was made inside or outside the U.S. But while those decisions are instructive, the factors driving each of those cases differ from this case. In AK Steel Corp. v. United States, 226 F.3d 1361 (Fed.Cir. 2000), our appellate court stated that:

AK Steel, 226 F.3d at 1368. However, the holding in A.K. Steel that the first unaffiliated sale is a CEP sale when it is made by an exporter's U.S. affiliate does not resolve the issue here. Unlike in AK Steel, Commerce decided that GPS, the U.S. importer, did not make the sales in question. I & D Memo at 35. Rather, Commerce found the sales were made by GEHK, Gold East's affiliate.

In Corus Staal BV v. United States, 502 F.3d 1370 (Fed.Cir.2007), the exporter argued the sales were EP because they were made prior to importation, but Commerce found they were CEP sales made after importation because the orders were filled from U.S. stock. Corus Staal, 502 F.3d at 1376. The case turned on whether the sale predated importation and the court agreed with Commerce that it did not. Corus Staal, 502 F.3d at 1377. Corus Staal's holding was controlled by the statute because if the sale was made after importation then it could not be an EP sale. 19 U.S.C. § 1677a(a) (EP sales are made "before the date of importation"). Corus Staal's holding as a result does not resolve the issues in this case.

Nucor Corp. v. United States, 612 F.Supp.2d 1264, 1275 (CIT 2009), upheld Commerce's finding of an offshore sale where title transferred to the U.S. buyers overseas, among other factors. Commerce disagreed with an allegation that title transferred in the U.S. Id. "[T]he sales agreement was signed in Turkey by [exporter] personnel, the invoice was issued by an entity in Turkey (i.e., the producer/exporter) to an entity in the United States (i.e., the U.S. customer), and it was concluded outside the United States." Id., quoting Decision Memo at 66-67. The court agreed with Commerce that the sales should be classified as EP because they occurred outside the U.S. Id. at 1282-83. Unfortunately, as with Corus Staal, the structure of the transactions in Nucor differed significantly from those in this case.

Moreover, the record evidence provides only a limited indication where the sale occurred under the statute and caselaw. Commerce cited the following facts in its determination:

I & D Memo at 35. The record reveals that in these transactions, payment was made by overseas customers directly to GEHK or the U.S. affiliate GPS (which forwarded payment to GEHK). See Attachment 1 to Appleton Reply Br. (sales flow charts). It appears that GEHK was not the producer or exporter of the merchandise, which shipped directly from Gold East to customers in the U.S. Id. GEHK also purchased raw materials which were sent directly to Gold East. Id. But there is no discussion of contract terms regarding passage of title or risk of loss, other than the use by the parties of DDP and DDU terms. Cf. Nucor, 612 F.Supp.2d at 1273-1275 (discussing factors relevant to determination of locus of sale). Also absent from the record is a discussion of where and by whom the negotiations for the sales were held. There is no mention of whether there was an overall sales agreement covering the sales in question, and if so, where it was made. Commerce does not discuss the importance of the fact that payments went from the U.S. customers to GEHK. While Commerce found that the sales were not made by GPS, that does not necessarily mean that the sales were not made in the U.S. due to other factors.

Even if the record fully supported Commerce's determination that the sales were made outside the U.S., that factor is not dispositive of the EP/CEP classification, because the statute also distinguishes between sales made by the producer and those made by another party on behalf of the producer. Under 19 U.S.C. § 1677a, an EP sale is the first sale (or agreement to sell) "by the producer or exporter of the subject merchandise" outside the United States to an unaffiliated purchaser in the U.S. 19 U.S.C. § 1677a(a). A CEP sale is the first sale (or agreement to sell) in the U.S. "by or for the account of the producer or exporter ... or by a seller affiliated with the producer or exporter" to an unaffiliated purchaser in the U.S. 19 U.S.C. § 1677a(b) (emphasis added). In this case, the first sale was made by an affiliated seller (GEHK) which was not the producer/exporter.9 Under the statute, if GEHK was the seller but not the producer or exporter of the merchandise in question, the sale cannot be an EP sale. By the same token, however, if the sale was made outside the U.S., it cannot be a CEP sale. 19 U.S.C. § 1677a(b).

This structural distinction is discussed in AK Steel. There, the fact that the affiliate was deemed the seller was critical to the court's conclusion that the sale could not be deemed an EP sale. As stated in AK Steel, "a sale made by a U.S. affiliate or another party other than the producer or exporter cannot be an EP sale." AK Steel, 226 F.3d at 1374 (emphasis added).

AK Steel, 226 F.3d at 1370-71. The Statement of Administrative Action also distinguishes between sales by the seller and sales for the account of the producer. The latter are classified as being CEP sales without reference to the location of the sale. Statement of Administrative Action accompanying Uruguay Round Agreements Act (URAA), Pub.L. No. 103-465, tit. II, 108 Stat. 4809, H.R. Doc. No. 103-316 at 823 (1994) ("SAA"). It states,

Id. at 822-23. The SAA thus classifies as CEP sales those made "for the account of" the producer or those made by the producer's affiliate in the U.S.

The statute, through § 1677a(d), provides for downward adjustments in CEP prices to account for the presumed additional costs attributable to sales by affiliates, but the downward adjustment is not restricted to only U.S. affiliates. Commerce's analysis failed to recognize the incongruity of classifying as EP sales that were made by the affiliate GEHK. Therefore, Commerce's finding that the sales were EP sales is not in accordance with law and is unsupported by substantial evidence on the record.

In addition, there is significant evidence on the record that conflicts with Commerce's finding that the sales were not made in the U.S., including the DDP and DDU nature of some transactions. Coupled with the relative paucity of record information about several factors relevant to the finding of the location of the sale, and the limited discussion of this issue in the I & D Memo, the court finds that the determination that the sales were made outside the U.S. is unsupported by substantial evidence. This issue is remanded to Commerce with instructions to reopen the question of whether the sales were EP or CEP. Commerce shall review the record to determine if there is further evidence of where the sales were made. Commerce shall also provide a fuller analysis of why the sales should be deemed EP or CEP, including a discussion of how the fact that the sales were made by an affiliate should affect that determination.

VII. Surrogate Wage Rate Analysis

Appleton argues that Commerce used a flawed methodology to determine the surrogate wage rate for the People's Republic of China. Appleton Br. at 6. Commerce initially used a traditional regression methodology in the Preliminary Determination. When that methodology was overturned in Dorbest v. United States, 604 F.3d 1363, 1371 (Fed.Cir.2010), Commerce used a so-called "bookend" methodology for the Final Determination.

I & D Memo, Comment 30 at 65, PR 353. Commerce found 23 countries with significant exports of comparable merchandise between 2007 and 2009. Id. at 66. Commerce further refined the list by identifying 15 countries with the necessary wage data. Id. at 67-68.

Commerce, according to Appleton, used "a broad average rate developed from earnings for all workers in all industries in multiple countries whose only demonstrated comparability to China was country-level per capita income." Appleton Br. at 14. Commerce rejected Appleton's insistence that it use only data from an Indian paper company's financial statements because "wage data from a single country does not constitute the best available information for purposes of valuing the labor input due to the variability that exists between wages and GNI." Cmt. 30, I & D Memo at 68. "[Commerce] has a longstanding and predictable practice of selecting economically comparable countries on the basis of absolute GNI, and nothing in Petitioners' submissions undermines the reasonableness of that practice." Id. at 65.

The bookends approach has subsequently been invalidated and been replaced with one which looks to wages in the primary surrogate country. Appleton Reply at 10, citing Antidumping Methodologies in Proceedings Involving Non-Market Economies: Valuing the Factor of Production: Labor, 76 Fed.Reg. 36902, 36093 (Dep't Comm. June 21, 2011). The government argues that the court should not impose the new primary surrogate methodology because the Antidumping Methodologies announcement by its terms was not made retroactive. Deft's Br. at 92, n. 16, citing Grobest & I-Mei Indus. (Vietnam) Co. v. United States, 815 F.Supp.2d 1342, 1359-60 n. 23 (CIT 2012).

The court agrees with the holding in Grobest, where the court upheld Commerce's decision to use averaged data rather than specific data.

Grobest, 815 F.Supp.2d at 1360. For similar reasons, the court declines Appleton's invitation to overturn Commerce's reasonable decision to use the then-applicable "bookends" approach. As stated by the government, "Appleton's argument is nothing more than an invitation for the court to substitute its judgment regarding the best surrogate data source for that of Commerce." Deft's Br. at 94.

Appleton also argues that Commerce erred in its selection of countries used in the "bookends" approach because it did not choose countries bracketed equally around China's data. But the countries selected had gross national incomes both above and below that of China, unlike the situation in Dorbest, where the court remanded for reconsideration because Commerce selected bookend countries that all had GNIs below that of China. Dorbest Ltd. v. United States, 755 F.Supp.2d 1291, 1298 (CIT 2011). The court has reviewed the record and finds that this aspect of Commerce's decision was reasonable under the circumstances.

CONCLUSION

Based upon the record developed before Commerce and upon the papers and proceedings before this court, for the reasons set forth above, the court remands this action to Commerce for action consistent with this opinion. Commerce shall prepare a preliminary analysis of the issues and submit it to the parties no later than 90 days from the date of this opinion. Within 30 days of the preliminary analysis, the parties shall be permitted to file comments with Commerce. Commerce shall then have another 60 days to complete a final redetermination which will be filed with the court no later than Monday, December 16, 2013. The parties shall have thirty days thereafter to file comments on the remand in this court.

SO ORDERED.