Full opinion text
OPINION RIDGWAY, Judge. Pending before the Court are the Final Results of Redetermination Pursuant to Court Remand, filed by the U.S. Department of Commerce pursuant to the decision in Habas. See generally Final Results of Redetennination Pursuant to Court Remand (“Remand Determination”); Habas Sinai Ve Tibbi Gazlar Istihsal Endustrisi A.S. v. United States, 31 CIT -, 2007 WL 3378201 (2007) (“Habas ”). Habas remanded to Commerce two issues concerning the agency’s analyses in the seventh administrative review of the antidumping duty order on Certain Steel Concrete Reinforcing Bars From Turkey: (1) Commerce’s use of annual Period of Review (“POR”) average costs (rather than Habas’ quarterly costs) in the agency’s “sales-below-cost” analysis, and (2) Commerce’s use of invoice date (rather than contract date) as the date of sale for Habas’ U.S. sales in the agency’s anti-dumping duty margin calculations. See Habas, 31 CIT at -, -, 2007 WL 3378201 *5, 8. In its Remand Determination, Commerce reaffirmed its earlier decision to use POR average costs, rather than quarterly costs. However, Commerce reversed its prior determination on the date of sale issue, concluding that contract date is the appropriate date of sale. See generally Remand Determination. In its comments on the Remand Determination, Habas requests that the quarterly costing issue be remanded once again, but argues that the Remand Determination on the date of sale issue should be sustained. See generally Brief of Plaintiff Habas Sinai ve Tibbi Gazlar Istihsal Endustrisi A.S. Concerning Remand Final Determination of Department of Commerce (“Pl.’s Brief’); Reply Brief of Plaintiff Habas Sinai ve Tibbi Gazlar Istihsal Endustrisi A.S. Concerning Remand Final Determination of Department of Commerce (“Pl.’s Reply Brief’). The Domestic Producers — DefendantIntervenors Nucor Corporation, Gerdau Ameristeel Corporation, and Commercial Metals Company — oppose Habas on both counts. According to the Domestic Producers, the Remand Determination should be sustained as to the quarterly costing issue, while the date of sale issue should be remanded to the agency once more. See generally DefendanNIntervenors’ Comments on the Remand Results (“Def.-Ints.’ Brief’); DefendanNIntervenors’ Supplemental Brief (“Def.-Ints.’ Reply Brief’). The Government maintains that Commerce has complied fully with the Court’s instructions in Habas, and that the Remand Determination is supported by substantial evidence and is otherwise in accordance with the law. The Government therefore contends that the Remand Determination should be sustained in its entirety. See Defendant’s Response to Comments Regarding Remand Redetermination (“Def.’s Response Brief’) at 4. Jurisdiction lies under 28 U.S.C. § 1581(c) (2000). For the reasons set forth below, the Remand Determination is sustained as to Commerce’s determination on the use of contract date as the date of sale. However, as to the issue of quarterly costs versus POR-average costs, this matter must be remanded to the agency yet again. I. Background This action arises out of the seventh administrative review of the antidumping duty order on imports of steel concrete reinforcing bar (“rebar”) from Turkey. In the Preliminary Results of the administrative review, Commerce made a preliminary determination that the dumping margin for Habas was 26.07%. See generally Certain Steel Concrete Reinforcing Bars from Turkey; Preliminary Results and Partial Rescission of Antidumping Duty Administrative Review and Notice of Intent to Revoke in Part, 70 Fed.Reg. 23,990 (May 6, 2005) (“Preliminary Results”); see also Habas, 31 CIT at -, 2007 WL 3378201 *2. Following publication of the Preliminary Results, Habas’ advocacy before Commerce focused principally on the two issues in dispute in this action — whether Commerce erred in using annual POR-average costs (rather than Habas’ quarterly costs) in the agency’s sales-below-cost analysis, and whether Commerce erred in using invoice date (rather than contract date) as the date of sale for Habas’ U.S. sales in the agency’s antidumping duty margin calculations. In the Final Results, Commerce rejected Habas’ arguments on both issues, and left Habas’ dumping margin unchanged at 26.07%. See generally Certain Steel Concrete Reinforcing Bars From Turkey: Final Results, Rescission of Antidumping Duty Administrative Review in Part, and Determination to Revoke in Part, 70 Fed.Reg. 67,665 (Nov. 8, 2005) (“Final Results”); see also Issues and Decision Memorandum for the Antidumping Duty Administrative Review on Certain Steel Concrete Reinforcing Bars from Turkey — April 1, 2003, through March 31, 2004, 2005 WL 3054566 (Nov. 8, 2005) (Pub.Doc. No. 256) (“Decision Memorandum”). This action followed, contesting Commerce’s determination in the Final Results. See generally Habas, 31 CIT -, 2007 WL 3378201. In Habas, the Court granted in part Habas’ Motion for Judgment Upon the Agency Record, remanding to Commerce for further consideration the issues of quarterly costs and date of sale. See generally Habas, 31 CIT at -, -, 2007 WL 3378201 *5, 8. In its Remand Determination, Commerce reaffirmed the agency’s earlier decision to use annual POR average costs— rather than quarterly costs — in its sales-below-cost analysis. See generally Remand Determination at 1-19, 21-40, 49. However, Commerce reversed its earlier determination on the date of sale issue, concluding that contract date is the appropriate date of sale for use in the agency’s antidumping duty margin calculations. See generally Remand Determination at 1-2, 19-21, 40-^49. Commerce therefore recalculated Habas’ dumping margin, which now stands at 22.53%. See Remand Determination at 1-2, 21, 49. II. Standard of Review In reviewing a challenge to Commerce’s final determination in an antidumping case, the agency’s determination must 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)(l)(B)(I); see Elkem Metals Co. v. United States, 468 F.3d 795, 800 (Fed.Cir.2006). “[S]ubstantial evidence is more than a mere scintilla. It means such relevant evidence as a reasonable mind might accept as adequate to support a conclusion.” Universal Camera Corp. v. NLRB, 340 U.S. 474, 477, 71 S.Ct. 456, 95 L.Ed. 456 (1951) (quoting Consol. Edison Co. v. NLRB, 305 U.S. 197, 229, 59 S.Ct. 206, 83 L.Ed. 126 (1938)). Moreover, any determination as to the substantiality of evidence “must take into account whatever in the record fairly detracts from its weight,” including “contradictory evidence or evidence from which conflicting inferences could be drawn.” Suramerica de Aleaciones Laminadas, C.A. v. United States, 44 F.3d 978, 985 (Fed.Cir.1994) (quoting Universal Camera, 340 U.S. at 487-88, 71 S.Ct. 456). On the other hand, the mere fact that “it [may be] possible to draw two inconsistent conclusions from evidence in the record ... does not prevent Commerce’s determination from being supported by substantial evidence.” Am. Silicon Techs. v. United States, 261 F.3d 1371, 1376 (Fed.Cir.2001); see also Conso lo v. Federal Maritime Commission, 383 U.S. 607, 620, 86 S.Ct. 1018, 16 L.Ed.2d 131 (1966) (same). III. Analysis Habas and the Domestic Producers each challenge one aspect of Commerce’s Remand Determination. Specifically, Habas takes issue with Commerce’s continued adherence to the use of a single cost-averaging period, contemporaneous with the period of review (POR), in the agency’s sales-below-cost analysis, while the Domestic Producers dispute Commerce’s decision to reverse its earlier determination and use contract date (rather than invoice date) as the date of sale in its antidumping duty calculations. Both issues are discussed in turn below. For the reasons detailed there, Habas’ challenge to the Remand Determination is sustained, and this matter is remanded to Commerce for a second time, for further consideration of the issue of the use of POR-average costs versus quarterly costs. On the other hand, the Domestic Producers’ challenge to Commerce’s decision to use contract date (rather than invoice date) as the date of sale is rejected, and the Remand Determination on that issue is sustained. A. Commerce’s Determination on Use of Qucurterly Costs versus POR-Average Costs In order to make fair comparisons between U.S. sales and normal value, and between home market sales and costs, Commerce must determine the appropriate time period(s) for its weighted-average cost calculations. In the instant case, Habas claims that Commerce’s use of POR-average costs created a mismatch between sales and costs which distorted the comparisons between U.S. price and normal value. See Habas, 31 CIT at -, 2007 WL 3378201 *3. According to Habas, Commerce’s use of POR average costs “causes a 14% increase in normal value ... which, in turn, inflates the dumping margins by 20%.” See Pl.’s Brief at 4. The Remand Determination at issue here is, in part, the result of the Government’s request for a voluntary remand on the issue of POR-average costs versus quarterly costs. Habas vigorously opposed the request for a voluntary remand, claiming that Commerce “simply want[ed] another chance to come up with a rationale to support its previous decision,” and cautioning against giving Commerce “another chance to find a theory that will support [its] predetermined result,” by allowing the agency yet “another bite at this apple.” See Habas, 31 CIT at -, -, 2007 WL 3378201 *4 (internal quotation marks and citations omitted). The source of frustration with the Government’s request for a voluntary remand, explained Habas, was that: During the administrative proceeding, Commerce issued a preliminary result based on a particular rationale. Habas’ case brief addressed Commerce’s rationale. Commerce then chose to keep the same result, but to formulate a new rationale [in its Final Results].... In its principal brief [filed with the Court], Habas exposed the errors of Commerce’s rationale. Now, having read Habas’ principal brief, the government would like another chance before this court to formulate a more persuasive rationale. Habas, 31 CIT at -, 2007 WL 3378201 *4 (internal quotation marks and citations omitted). Habas instead sought a directed remand, urging the Court to require Commerce to recalculate Habas’ dumping margin using Habas’ quarterly costs (rather than POR-average costs). See Habas, 31 CIT at -, 2007 WL 3378201 *4-6. Notwithstanding Habas’ request, the Court granted the Government’s request for a voluntary remand in Habas, noting that — under SKF — an agency is generally entitled to a voluntary remand to reconsider its position if the agency’s concern is substantial and legitimate. See Habas, 31 CIT at -, 2007 WL 3378201 *4 (citing SKF USA, Inc. v. United States, 254 F.3d 1022, 1028-29 (Fed.Cir.2001)). Habas also took note of the Government’s assurances that Commerce intended to “take a fresh look” at the issue on remand. See Habas, 31 CIT at -, 2007 WL 3378201 *5; see also Pl.’s Brief at 5-6. Habas now asserts that — in its Remand Determination — Commerce has once more “shown itself unwilling to consider this issue without prejudgment,” and has “made no attempt to ‘take a fresh look’ at quarterly cost, nor did it ‘consider anew1 its methodologies.” See PL’s Brief at 6. Habas essentially claims that the Remand Determination has yet again “moved the goalpost” on the issue of the use of multiple cost-averaging periods, and that Commerce “continue[s] to rely on flawed tests, illogical propositions, and selective statistics.” See PL’s Brief at 39. Although they are strong, there is at least some truth to Habas’ charges. As a threshold matter, four interrelated overarching points bear note. First, throughout its briefs, the Government repeatedly alludes to the general “virtue” of Commerce’s standard practice of using annual (POR) average costs in its sales-below-cost analysis — i.e., that the use of annual POR-average costs tends to smooth out swings in production costs that respondents may experience over shorter periods of time. See, e.g., Def.’s Response Brief at 13 (noting that “the use of annualized costs ‘normally evens out swings in production costs’ ” that may occur over shorter periods) {quoting Remand Determination at 12). At the same time, the Government strives to depict Habas as seeking to carve out for itself some novel, aberrant, extraordinary “special exception” to Commerce’s standard practice — going so far as to characterize the relief that Habas seeks as “a dramatic change from Commerce’s normal practice.” See Def.’s Response Brief at 7, 8 (emphasis added). As the Government acknowledges, however, Habas is not challenging in principle Commerce’s standard practice of using annual POR average costs. See Def.’s Response Brief at 7 (noting that “Habas does not challenge the reasonableness of Commerce’s general practice”), 13 (noting that Habas does not claim that “Commerce’s annual-based methodology is impermissible”). Nor — contrary to the Government’s implication- — is Habas attempting to fashion and then exploit some creative, brand new, unprecedented, one-off “loophole” or caveat to Commerce’s standard practice. Instead, Habas has sought merely to demonstrate that it falls within the preexisting, longstanding, and well-established exception to the Government’s standard practice — an exception that provides for the use of multiple, shorter cost-averaging periods by respondents that meet certain heretofore relatively clear-cut criteria, in situations where the fundamental underlying “virtue” of Commerce’s standard practice does not hold true (ie., where the use of annual POR-average costs does not serve to “smooth out” swings in production costs, but — rather— has a distortive effect). See Pl.’s Reply Brief at 2 (observing that “Habas is simply asking Commerce to apply a well-established test to Habas’ facts”), 3 (same). The second threshold observation is related to the first. As noted immediately above, Habas has here sought simply to avail itself of the longstanding, well-recognized exception to Commerce’s standard practice — an exception permitting the use of multiple, shorter cost-averaging periods under certain specific circumstances. As Habas quite properly complains, however, Commerce’s test for the use of multiple, shorter cost-averaging periods has (to say the least) been a constantly moving target in the administrative review at issue here. See generally section III.A.1, infra; see, e.g., PL’s Brief at 2 (summarizing evolution of Commerce’s approach over course of this proceeding, and emphasizing that agency’s “approach [in this case] is completely different in kind from that of all previous cases”), 6-10 (reviewing evolution of Commerce’s approach in this case). The third threshold observation is related to the second. The Government’s brief is peppered with various references casting aspersions on Habas’ motives in taking the positions that Habas has taken in this case. Ordinarily, it would suffice simply to dismiss such finger-pointing with the general observation that presumably all litigants take their positions with an eye toward promoting their own self-interests. See Pl.’s Reply Brief at 7-8. In this ease, however, it is not much of a stretch to view the Government’s efforts to impugn Habas as a classic case of “the pot calling the kettle black.” As noted above (and discussed in greater detail below), Commerce’s test for multiple, shorter cost-averaging periods has been such a moving target that one might be forgiven for wondering whether, in fact, in this case it is the Government (specifically, Commerce) that is pursuing a (questionably) single-minded agenda — in other words, whether it is actually Commerce that has a “result in search of a rationale.” The fourth, and final, threshold observation is also related to the second' — i.e., the apparently still-evolving nature of Commerce’s test for the use of multiple, shorter cost-averaging periods. Commerce has (to put it most charitably) sought to reformulate and refine its test for the use of multiple, shorter cost-averaging periods over the course of these proceedings; but — at the same time — Commerce and the Government (and, to a somewhat lesser extent, the Domestic Producers) seek to confíne Habas to the administrative record developed before Commerce fully defined and articulated the (still, frankly, rather amorphous and unclear) criteria that the agency now seeks to apply. See, e.g., Remand Determination at 26-27 (noting Domestic Producers’ arguments that agency should refuse to consider various authorities cited by Habas to demonstrate that 5% to 10% difference in COM is significant), 30-31 (stating that “Habas’ reliance on information which is not on the administrative record before the agency is inappropriate and [Commerce] will not address this line of argument further”). There is thus an obvious issue of fundamental fairness, to which Commerce and the Government (and, to a somewhat lesser extent, the Domestic Producers) seem to turn a blind eye. 1. Commerce’s Evolving Test for Use of Multiple Cost-Averaging Periods When the underlying administrative review began, Commerce’s then well-established test for the use of multiple cost-averaging periods in cases involving ordinary industrial products (such as the rebar at issue here) “focused on the behavior and economics of the respondent,” inquiring: (1) whether the product at issue had a “single primary input,” and (2) if so, whether the cost of that single primary input experienced a change during the POR that was of a quality and magnitude to warrant the use of multiple, shorter cost-averaging periods (ie., less than a full POR). If those two criteria were satisfied, and if the respondent’s changes in prices tracked changes in costs, Commerce used multiple cost-averaging periods. See generally Pl.’s Brief at 6; see also id. at 2. As Habas puts it, “[t]he core issue was whether costs had increased or decreased markedly across the POR.” Id. at 6. The Preliminary Results in the case at bar reflect Commerce’s then well-established test, as it was being applied by the agency at that time. Specifically, Commerce stated in the Preliminary Results: The Department has used monthly or quarterly costs in non-inflationary cases only where [1] there was a single primary input and [2] that input experiences a significant and consistent decline or rise in its cost during the reporting period. Preliminary Results, 70 Fed.Reg. at 23,993 (emphases added). Applying that then well-established test, Commerce concluded in the Preliminary Results that Habas did not qualify for the use of multiple cost-averaging periods: In this case, because we do not find that the price of scrap [the single primary input in rebar] experienced a significant and consistent increase during the POR, we have continued to follow [Commerce’s] normal practice of using weighted-average POR costs for all respondents. Preliminary Results, 70 Fed.Reg. at 23,993 (emphasis added). In other words, Commerce implicitly found in the Preliminary Results that Habas satisfied the first criterion — i.e., Commerce found that rebar is a “single primary input” product. But Commerce also found that Habas had not demonstrated that it satisfied the second criterion — i.e., Commerce found that Habas had not shown that the cost of the single primary input (scrap) had “experienced a significant and consistent increase during the POR.” Habas briefed the issue extensively before the agency, seeking to demonstrate that it satisfied Commerce’s then well-established test for the use of multiple cost-averaging periods. Specifically, Habas argued that rebar is a single primary input product, that it had experienced a 28% increase in material cost between the first and last quarters of the POR, and that such a change was of the type and magnitude to qualify Habas for the use of quarterly costing. See generally Pl.’s Brief at 7 (and sources cited there). Habas also sought to explain that Commerce’s use of POR-average costing resulted in mismatches in the normal value calculation by driving below cost many sales that were actually well above cost at the time that they were made. Id. In the Final Results, Commerce apparently abandoned the well-established test that it had applied in the Preliminary Results in this case, and in other, prior cases. No longer was Commerce focused on the existence and cost of a single primary input. Nevertheless, relying on a seemingly brand new test (and on a rather different rationale as well), Commerce once again concluded that Habas did not qualify for the use of multiple cost-averaging periods: [Commerce] analyzed the significance of the change in the COM [cost of manufacturing], whether the change in cost occurred consistently and significantly throughout the POR, and whether the direct material inputs causing the cost fluctuation can be directly tied to the related sales transactions. In this case, the COM experienced by the respondents both decreased and increased during the first three quarters of the POR. It was not until the third and fourth quarters of the POR that the COM increased steadily. Because of this end of POR increase, the respondents claim that the COM for the first two quarters of the POR become inflated when using an annual average method, as compared to a quarterly average method. While we agree with the respondents that the annual average COM is higher than the quarterly average COM for the first two quarters of the POR, we disagree that the difference is significant. In analyzing this point, we first identified the 5 highest volume home market control numbers and examined the impact of using annual average costs of manufacturing versus quarterly average costs of manufacturing. We computed the difference in the cost of the input raw materials for the first two quarters of the POR using quarterly average cost data versus annual average cost data, and noted that in both instances, the difference ranged from approximately 5 to 10% of the COM.... In the past, [Commerce] has not considered one to 10% increases significant. See Pasta from Italy 1998-1999 Reviews. Therefore, we find the respondents’ reliance on Thai Pineapple 1 and Thai Pineapple 2 irrelevant given that, in the instant case, we have found no significant change in the cost of scrap during the POR. Decision Memorandum at 11-12. As the excerpt above reflects, in stark contrast to the test articulated in the Preliminary Results in this case (and also applied in other, prior cases), Commerce in the Final Results here focused not on the existence and cost of a single primary input (i.e., scrap), but, rather, on the total cost of manufacturing (“COM”). Moreover, rather than comparing the cost at the beginning of the POR to the cost at the end of the POR (as Commerce had done in the past, both in the Preliminary Results in this case and in other, prior cases), Commerce instead compared Habas’ actual quarterly COM to Commerce’s calculated POR-average COM. In essence, rather than comparing costs at two different points in time, Commerce instead compared the results of two different methodologies. As Habas observes, the effect of this change was to repudiate the agency’s historic focus on “the behavior and economics of the respondent” in favor of “an examination of the difference between [two] competing methodologies” — an approach “completely different in kind from that of all previous cases.” Finding that the difference between the results of the two methodologies ranged from 5% to 10%, Commerce then concluded — ostensibly relying on Pasta from Italy as precedent — that a difference of 10% was not “significant.” See Decision Memorandum at 12; see generally PL’s Brief at 2, 7-8. On the strength of that analysis, Commerce determined that the use of multiple cost-averaging periods was not warranted. Habas challenged Commerce’s Final Results in this action. In its initial briefs (filed pre-remand), Habas once again addressed the quarterly costing issue in detail, focusing primarily on the new test that Commerce had articulated and applied in the Final Results. Habas argued that the surge in scrap prices in the fourth quarter of the POR increased Habas’ cost of manufacturing (“COM”) by 21% across the POR. Habas further argued that— when quarterly costs were averaged across the POR — the surge in fourth-quarter COM created “fictitious profits and losses on home market sales,” violating the rule of Brass Sheet and Strip From the Netherlands, and artificially increasing normal value by 14.5%. And, significantly, Habas criticized Commerce’s reliance on Pasta from Italy, explaining that Pasta from Italy simply does not stand for the proposition for which the agency cited it in the Final Results. See generally PL’s Brief at 8-9 (citing Notice of Final Results of Anti-dumping Duty Administrative Review and Determination Not to Revoke the Anti-dumping Duty Order: Brass Sheet and Strip From the Netherlands, 65 Fed.Reg. 742, 747 (Jan. 6, 2000) (“Brass Sheet and Strip from the Netherlands”); RE: Certain Pasta from Italy (Period of Review: July 1, 1988 through June 30, 1999), Subject: Issues and Decision Memorandum for the Third Antidumping Duty Administrative Review; Final Results of Review, 2000 WL 1880666 (Dec. 13, 2000) (“Pasta from Italy”), at comment 18). In particular, Habas explained that the gravamen of Pasta from Italy is that an increase in raw material cost of 10% to 12% between the beginning and the end of the POR is not significant, when costs increased for half of POR and decreased for the other half of the POR. But, notably, in Pasta from Italy, Commerce was not comparing the results of two different costing methodologies (as the agency did in the Final Results here). Thus, contrary to Commerce’s implication in the Final Results in this case, Pasta from Italy does not stand for the proposition that a 10% difference between two methodologies is “not significant.” See generally PL’s Brief at 9. In the course of the remand proceedings, Commerce reaffirmed that its new approach to deciding whether to apply multiple cost-averaging periods is to compare, during particular quarters, the quarterly cost of manufacturing (COM) with the POR-average COM. According to Commerce, if the difference between the results of these two methodologies is not “significant,” the agency will not use multiple cost-averaging periods. In the Draft Remand Results, Commerce summarized its new analysis: [W]e reviewed the precise impact of using Habas’s quarterly-average COM approach versus [Commerce’s] preferred annual-average COM method, and found that the difference of approximately 5 to 10% was not significant. See Final Results at Comment 1. Accordingly, Commerce found that using annual-average costs was consistent with [the agency’s] practice, more predictable, and reasonable. Draft Results of Redetermination Pursuant to Court Remand (Pub.Doc. No. 3) (“Draft Remand Results”), at 7. Habas points out that the language of the Draft Remand Results thus largely tracked the language of the (pre-remand) Final Results — with one significant difference: The Draft Remand Results omitted the Final Results’ reference to Pasta from Italy. As Habas notes, that omission reflects Commerce’s tacit admission that the agency’s earlier reliance on Pasta from Italy was misplaced. See Pl.’s Brief at 11. The relevant language of the final Remand Determination parrots the language of the Draft Remand Results (quoted above), word-for-word. See Remand Determination at 8. As noted above, in its briefs now before the Court, Habas complains that — notwithstanding the assurances that the Government gave in seeking a voluntary remand — Commerce failed to take a “fresh look” at this issue during the course of the remand proceeding. See Habas, 31 CIT at -, 2007 WL 3378201 *5; see also Pl.’s Brief at 6, 10, 12. Habas asserts that, on remand, “Commerce simply restated the test for which it had previously been unable to articulate a rationale,” and then— for good measure — sought to buttress that test with an additional “two evidentiary tests that purport! ] to support the conclusion that Habas did not experience a significant increase in cost in the POR and that Habas’ prices were not correlated closely with its [cost of manufacturing].” See PL’s Brief at 10. According to Habas, “Commerce’s criteria for whether to apply multiple cost-averaging periods, as developed in this case, remain selective, ad hoc, and unprincipled, unsupported by law or fact.” See PL’s Brief at 10. And, as to the two so-called “evidentiary tests Commerce formulated in the remand,” Habas contends that “the first actually supports quarterly costing, while the second has no substantive bearing on the issue at hand.” Id. 2. The First Prong of Commerce’s Test for Multiple Cosh-Averaging Periods As summarized above, Commerce’s principal criteria for the use of multiple, shorter cost-averaging periods historically have been (1) whether the product was a “single primary input product,” and (2) if so, whether the cost of that single primary input increased or decreased significantly across the period of review (“POR”). If a respondent satisfied those two criteria (which in the past together comprised the first, and the main, prong of Commerce’s test for shorter cost-averaging periods), and if the respondent’s changes in prices tracked its changes in costs (the second prong of the test as applied in the past), then Commerce used multiple, shorter cost-averaging periods, because use of the agency’s standard annual POR-average costs would be distortive. But Commerce approached this case quite differently. As the Remand Determination explains, as to the first prong of the test, Commerce conducted two different analyses. First, Commerce analyzed the difference between calculating Habas’ costs on a quarterly basis versus calculating Habas’ costs on an annual POR-average basis — in essence, seeking to determine the “significance” of the difference between using Commerce’s standard POR-average costing methodology versus using a quarterly costing methodology (as Habas has proposed). See generally Remand Determination at 7-8, 28, 32. Commerce’s analysis in this case thus represented a sea change from what the agency has done in such cases in the past. As noted above, Commerce historically has evaluated the “significance” of a respondent’s actual changes in cost over the POR — not the “significanee” of the difference between the results of two different costing methodologies (as Commerce did in the Remand Determination in this case). In effect, Commerce here fundamentally alters the nature of the first prong of its two-prong test for the use of multiple cost-averaging periods. Second, in addition to its comparative analysis of the difference between the annual POR-average cost methodology and the quarterly-average cost methodology, Commerce also conducted a “price volatility” analysis. Specifically, Commerce analyzed Habas’ home market rebar prices within each quarter of the POR as compared to fluctuations in Habas’ costs, to attempt to gauge the extent of normal cost and price fluctuations in Habas’ home market over a short period of time. See generally Remand Determination at 10-11, 29-30, 32. Based on its two analyses, Commerce concluded in the Remand Determination that the use of annual POR-average costs was not distortive. See generally Remand Determination at 8, 10-11, 19, 28-30, 32-33. In particular, Commerce found that the difference between costs calculated on a quarterly basis versus on an annual POR-average basis was approximately 5% to 10%, which Commerce concluded was not a sufficiently significant difference between the two methodologies to warrant the use of quarterly costs. See Remand Determination at 8, 19, 28, 32. In addition, Commerce found, as a result of its price volatility analysis, that — in light of the magnitude of the fluctuation in Habas’ home market sales prices within a given quarter — -Habas’ cost fluctuations over the course of the POR were “not unusual or significant.” See Remand Determination at 10-11, 29-30, 32. Habas attacks both of Commerce’s analyses. Habas first argues that Commerce’s analysis of “significance” in effect measures the wrong thing, by evaluating the significance of the difference between the end results of two competing costing methodologies, rather than the significance of the actual increase in Habas’ costs over the course of the POR. See generally Pl.’s Brief at 2-3, 15, 23, 26; PL’s Reply Brief at 4, 9-11; Remand Determination at 22. But see Remand Determination at 25-26, 28-29; Def.’s Response Brief at 16-17; Def.-Ints.’ Brief at 13-15, 21; Def.-Ints.’ Reply Brief at 1. Further, Habas faults the Remand Determination’s comparative analysis of the “significance” of the difference between the annual POR-average cost methodology and the quarterly-average cost methodology, on the ground that the analysis was limited to only two quarters of the POR. In addition, Habas dismisses Commerce’s price volatility analysis as “fatally flawed.” PL’s Brief at 20; see also id. at 2-5, 12-17, 19-24; PL’s Reply Brief at 12-13; Remand Determination at 23. But see Remand Determination at 7 & n. 1, 8, 10-11, 29-30, 32; Def.’s Response Brief at 15-17, 21-23; Def.-Ints.’ Reply Brief at 2-6. a. Commerce’s Comparison of Two Competing Methodologies Habas initially takes aim at Commerce’s decision to evaluate “significance” in this case by analyzing the “significance” of the difference between the use of two different costing methodologies. Habas contends that Commerce instead should have evaluated the “significance” of the increase in Habas’ costs between the beginning and the end of the POR, as Commerce has done in every other such case in the past. See, e.g., PL’s Reply Brief at 10-11 (noting that change in costs over POR “was specifically articulated as the test for a shorter cost-averaging period. in Pasta from Italy”; that “[i]n all of the precedents Commerce compared the costs across the POR; it is the core of a long-standing and clearly-articulated test”; and that “if this test is met, then the precedents require that costing be done on a shorter cost-averaging period precisely because POR-average costing is distortive in an environment where cost is rising rapidly”). Habas allows that “Commerce’s decision to analyze the difference in COM [cost of manufacturing] rather than in raw material cost [e.g., a single primary input, as the agency has done in the past] does not seem unreasonable.” Pl.’s Brief at 3. But, Habas maintains, Commerce should have evaluated “whether the COM increased (or decreased) significantly across the POR”— not whether there was a “significant” difference between the use of the two competing costing methodologies. Id.; see also id. at 15 (observing that “in all of the previous cases, Commerce analyzed the movement of cost across the period of review, whether the cost of a single input or the total COM”). In the Remand Determination, Commerce gave Habas’ point short shrift: Habas continues to argue that the change in costs from the beginning of the POR to the end is the proper method for analyzing this issue---- Other than the change in cost resulting in a larger figure, this approach provides little use in assessing the issue at hand. To simply look at costs at two points in time fails to recognize all the production activity throughout the year. That is, it simply represents the cost at two specific points in the POR, and calculates the percentage difference between those two points. By calculating the percent difference in the COM based on two specific points of time, we would be ignoring both the volume and COP lie., cost of production] occurring during the remaining time in the POR. Thus, ... Habas’ proffered analysis is [not] appropriate. Remand Determination at 28-29; see also id. at 25-26; Def.’s Response Brief at 16-17; Def.-Ints.’ Brief at 13-15, 21. Commerce’s treatment of Habas’ argument is far too dismissive. Although the parties differ as to the “significance” of the increase, no party appears to seriously dispute that Habas’ costs escalated over the course of the POR (driven largely by a 28% increase in the cost of scrap). Further, neither Commerce nor the Government nor the Domestic Producers contests Habas’ assertion that, in all previous cases, the first prong of Commerce’s analysis focused on the significance of the change in a respondent’s costs over the course of the POR. And, as Habas aptly observes, “Commerce never explains why a test that was appropriate for 20 years’ of precedent is suddenly irrelevant” (much less, as Commerce apparently contends, affirmatively misleading). See Pl.’s Brief at 15. Although Commerce refers to it in the Remand Determination (in the excerpt quoted above) as “Hotels’ proffered analysis” (emphasis added), analyzing the change in cost across the POR in fact historically has been Commerce’s approach to the analysis. To be sure, it is settled black letter law that an agency generally has the right to change its practices and methodologies. But it is equally well-established that the agency is obligated to fully explain and adequately justify any such changes. See, e.g., NSK Ltd. v. United States, 510 F.3d 1375, 1381 (Fed.Cir. 2007). To the extent that Commerce is now repudiating its past practice, Commerce must expressly acknowledge that it is doing so, and provide a full explanation and justification for the change and for any new approach that the agency is taking. It has not yet done so here. Commerce’s justification as set forth in the Remand Determination is much too truncated for the sweeping, fundamental changes that Commerce purports to make to the criteria for the use of multiple cost-averaging periods. The Remand Determination barely acknowledges the longstanding agency criteria that Commerce seeks to supplant, and does not address even the most obvious questions about the changes to those criteria. For example, the Remand Determination fails to explain the rationale behind Commerce’s traditional criteria for the first prong of the test for the use of multiple cost-averaging periods — the “single primary input” criterion, and the criterion of “a significant and consistent decline or rise in ... cost.” See Preliminary Results, 70 Fed.Reg. 23,993. Disavowing the “single primary input” criterion, the Remand Determination states that “it is the total COM [cost of manufacturing] that matters, not simply one component of the total manufacturing cost, since it is the COP [cost of production] that is used in the sales-below-cost test,” and that “[w]hile one input ... may represent a significant portion of the COM ..., the other costs incurred to manufacture the finished product are also important in analyzing the significance of cost fluctuations throughout the POR.” See Remand Determination at 8-9. Similarly, disavowing the criterion of “a significant and consistent decline or rise in ... cost” between the beginning and the end of the POR, the Remand Determination states that such an approach “simply represents the cost at two specific points in the POR.” See Remand Determination at 29. Each of these propositions is self-evident, however, and could not possibly have only recently occurred to Commerce — thus calling into question the rationale behind the agency’s original criteria, as well as the rationale for and the timing of the changes at issue here, and raising the spectre of unprincipled, ad hoc, result-oriented decisionmaking. See, e.g., Pl.’s Reply Brief at 15 (arguing that Commerce has “demonstrated its unyielding commitment to its result regardless of the evidence,” and predicting that agency “will simply continue to create new tests in an effort to support its foreordained conclusion”). Accordingly, this matter must be remanded to Commerce, to permit the agency to reconsider once again whether in this case it should evaluate the significance of the difference between the use of the two competing costing methodologies, or the significance of the increase in Habas’ costs between the beginning and the end of the POR (as Habas urges). In addition, Commerce shall detail the rationale behind the agency’s original criteria (discussed above), as well as the rationale for and the timing of any changes applicable in this case. b. Commerce’s Limitation of Its Analysis to Two Quarters Even assuming that Commerce was not required to apply the first prong of the test for multiple cost-averaging periods as the agency has applied it in past cases (i.e., even if Commerce was not required to base its determination on the significance of the increase in Habas’ costs over the course of the POR), it does not necessarily follow that Commerce’s analysis in this case must be sustained. Habas also argues, in the alternative, that — in evaluating the “significance” of the difference between the two competing costing methodologies — Commerce erred by confining its analysis to only two quarters (ie., the first and second quarters of the POR, when Habas made its U.S. sales). See Pl.’s Brief at 3-5, 12-17; see also Pl.’s Reply Brief at 11. But see Remand Determination at 7 & n. 1; Def.’s Response Brief at 15-17; Def.-Ints.’ Reply Brief at 3-4 & n. 2. As a threshold matter, Commerce’s analysis in the Remand Determination at least appears to be inconsistent with the agency’s asserted rationale for declining to compare “the change in costs from the beginning of the POR to the end” (as Habas urges). As quoted above, the Remand Determination essentially disavowed Commerce’s longstanding practice of comparing “the change in costs from the beginning of the POR to the end” on the grounds that — according to Commerce— by “looking] at costs at two points in time,” such an approach “fails to recognize all production activity throughout the year.” See Remand Determination at 28-29. However, Commerce here compared quarterly cost to POR average cost in only two quarters of the POR. The Remand Determination fails to explain how an analysis that is limited to only two quarters “recognize[s] all production activity throughout the year.” See generally PL’s Brief at 3,12-16. Further pressing its challenge to Commerce’s logic, Habas argues that — if Commerce’s intent is (as stated) to “recognize all the production activity throughout the year” — Commerce must extend its analysis beyond the first two quarters, to include all four quarters of the POR. See generally PL’s Brief at 3, 12-16. Habas protests that “Commerce’s examination of only the two quarters in which Habas had U.S. sales underscores the ad hoc nature of [the agency’s] exercise.” PL’s Brief at 14. And Habas asserts that the difference of approximately 5 to 10% which Commerce finds “not significant” in the Remand Determination “arises solely because Commerce chooses not to run its analysis across the entire POR.” Id.; Remand Determination at 8, 28. According to Habas, extending Commerce’s analysis to all four quarters of the POR demonstrates that “the full spread of the difference between POR-average cost and quarterly cost is 20%” and that “the increase in quarterly cost across the POR is 21%.” See PL’s Brief at 13-14; see also id. at 3. Habas thus concludes that, assuming arguendo the validity of Commerce’s approach of evaluating the significance of the difference between the results of two different methodologies, and extending that approach to the full year of the POR (in accordance with Commerce’s stated goal of “recognizing] all the production activity throughout the year”), Commerce’s own methodology would show that the difference between quarterly cost and average cost, measured across the entire POR, is actually 20% — not a “difference of approximately five to ten percent,” as the Remand Determination indicates. See Pl.’s Brief at 8, 13-15; Remand Determination at 8, 28. No party directly addresses Habas’ point. See Def.’s Response Brief at 16 (acknowledging, but not responding to, Habas’ argument); see also Def.-Ints.’ Reply Brief at 3 n. 2. In the Remand Determination, Commerce states that the agency limited its analysis to the two quarters in which Habas had U.S. sales “simply because those are the only quarters where contemporaneous comparison market sales would be used in the dumping margin calculation.” Remand Determination at 7 n. 1; see also Def.-Ints.’ Reply Brief at 3-4. However, Commerce nowhere explains how an analysis which was limited to the first and second quarters of the POR took into account “the surge in [Habas’] cost in the fourth quarter ” of the POR. See Pl.’s Brief at 4 (emphasis added). According to Habas, “[examination of the quarters when U.S. sales were made, in isolation from the rest of the POR, conceals the full impact of POR average cost versus quarterly cost.” Id. Habas asserts that, by comparing quarterly cost to POR average cost in only two quarters of the POR, Commerce fundamentally loses sight of “the central proposition” underlying the use of multiple cost-averaging periods in appropriate cases. Pl.’s Brief at 15-16. As Habas notes, the raison d’etre for multiple cost-averaging periods “is the concern about a mismatch between sales and cost,” such as the mismatch that Habas alleges here' — -a mismatch between sales in one quarter of the POR and the cost of production much later in the POR. Id. at 16. In the Remand Determination, Commerce acknowledges that its test for the use of multiple cost-averaging periods should be designed to “determin[e] whether there is a temporal mismatch between sales and costs.” Remand Determination at 29. But Commerce accuses Habas of “oversimplifying] the issue,” asserting that “[t]he difficulty in this case is to determine at what point the fluctuation in costs is significant enough to depart from [the agency’s] normal annual average method.” Id. The Remand Determination asks rhetorically: “Is a 10-percent difference in costs between an annual average method and a quarterly average method the tipping point? Is it 15 percent?” Id. As Habas notes, Commerce seems to intimate that, because of the difficulty in identifying a precise “tipping point,” multiple cost-averaging periods can never be justified. See Pl.’s Brief at 16. Commerce stops short of throwing the baby out with the bath water, however. Habas’ point is nevertheless well-taken. Commerce here makes no effort to define the “tipping point” which would warrant the use of multiple cost-averaging periods under facts such as those in this case' — -much less, to establish a standard to govern other cases, to ensure that similar cases are treated similarly. Instead, Commerce contents itself with indicating that, whatever may be the agency’s standard (or “tipping point”), it is not met in this case. Habas is not entirely unsympathetic to Commerce’s plight in identifying a precise “tipping point.” But Habas notes that the problem that Commerce faces is essentially one of the agency’s own making: “For nearly 20 years before the Turkish rebar case, Commerce had a consistent and predictable test [for the use of multiple cost-averaging periods]. For normal industrial products, the single-primary input criterion was workable .... [and the criterion of] whether [the] cost of the input had experienced a consistent and significant increase or decrease served well.” See Pl.’s Brief at 16-17 (footnote omitted). Notwithstanding the fact that Commerce apparently now seeks to jettison that longstanding test, Habas does not contend that Commerce must necessarily establish a precise “tipping point” here. According to Habas, “for 20 years, Commerce had no hard numerical test, and it does not need one to administer the statute effectively.” Id. at 17. However, Habas underscores that “[w]hat is required” in this case “is a principled approach,” rather than what seems in critical respects to be “ad hoc and selective decision-making.” Id. (emphasis added). As the discussion above makes clear, the Remand Determination fails to adequately address Habas’ numerous challenges to the logic and substantive validity of Commerce’s analysis and determination as to the first prong of the agency’s test for the use of multiple cost-averaging periods, as applied in this case. Nor does the Remand Determination adequately address the substantive merits of Habas’ other basic claims (taking into consideration matters such as the agency’s past practice, and the fundamental policy underlying the use of multiple cost-averaging periods)— e.g., Habas’ assertions that the use of “POR-average cost masks the 21% cost increase that occurred across the POR, it forces virtually all first- and second-quarter sales to go below cost, it inflates normal value by 14.5%, and it inflates the dumping margin by some 20 percentage points.” See PL’s Brief at 17. But see Remand Determination at 18-19 (concerning 14.5% increase in normal value). Contrary to Commerce’s claims, the Remand Determination does not establish that the use of POR-average cost is “more accurate” than quarterly costs in this case. See Remand Determination at 6. Commerce’s determination thus cannot be sustained on this record. As section III.A.2.a explains, this issue must be remanded to Commerce once again, for further consideration by the agency. On remand, Commerce shall reconsider the substantive merits of the first prong of its current analysis as reflected in the Remand Determination — in particular, the legitimacy of its evaluation of the significance of the difference between two competing costing methodologies (including its limitation of its analysis to only two quarters of the POR) — taking into consideration (and specifically addressing) each of Habas’ claims. Thus, to the extent that Commerce adheres to its current analysis (or to the extent that the matters otherwise remain relevant), Commerce shall explain, inter alia, how its analysis (limited to only the first two quarters of the POR) recognizes all production activity throughout the year, and how the analysis takes into account the surge in Habas’ costs in the fourth quarter of the POR. In addition, Commerce shall specifically address the validity of Habas’ extension of Commerce’s analysis to all four quarters of the POR. Commerce shall ensure that its redetermination on remand sets forth the agency’s rationale in detail and is supported by substantial evidence in the administrative record. 3. The Second Prong of Commerce’s Test for Multiple Cosh-Averaging Periods Commerce fares only slightly better on its analysis of the second prong of the test for shorter cost-averaging periods — ie., the linkage between Habas’ costs and its sales prices. See generally Remand Determination at 6, 11-14, 33^40; see also Pl.’s Brief at 5, 27-38; Pl.’s Reply Brief at 1, 3-9, 11-14; Def.’s Response Brief at 8-15, 21-26; Def.-Ints.’ Brief at 11, 17-21; Def.-Ints.’ Reply Brief at 1, 6-13. In the Remand Determination, Commerce explains that — even if the agency reached an affirmative determination on the first prong of its test for multiple cost-averaging periods (discussed in section III. A.2, above) — the use of quarterly costs nevertheless still would not be warranted absent “evidence of the direct linkage between the resulting quarterly-average costs and sales prices,” because (compared to the agency’s standard use of POR average costs to determine sales below cost) “a more accurate sales-below-cost test only results if the sales during the shorter averaging period can be directly linked with the [cost of production] during the shorter averaging period.” Remand Determination at 11. Commerce notes that “[i]f one’s objective is to determine whether sales within a given quarter were made above the cost to produce those same products in that quarter, production and sale should occur within the same quarter.” Id. The Remand Determination boldly concludes that, here, “there is no evidence ... which supports the proposition that production costs in each quarter were directly related to those sales reported in that same quarter.” Id. (emphasis added). Commerce overstates its case. Habas provided Commerce with an analysis of its costs and prices over the POR, which Habas asserts demonstrates that its home market sales prices “precisely and consistently” tracked its costs in the same quarter, “lockstep.” See Remand Determination at 12 (citation omitted). In the Remand Determination, Commerce faulted Habas’ analysis in two respects, a critique to which Habas does not directly respond. See Remand Determination at 12-13; Pl.’s Reply Brief at 5-6; see also Def.’s Response Brief at 10; Def.-Ints.’ Brief at 18; Def.-Ints.’ Reply Brief at 7. To further evaluate Habas’ claims of a “lockstep” relationship between its costs and its prices, Commerce looked to the price volatility analysis that the agency conducted on remand, discussed briefly in section III. A.2, above. See generally Remand Determination at 13, 33-34, 36-37; see also Def.’s Response Brief at 10-11, 21-23; Def.-Ints.’ Reply Brief at 2 n. 1, 7, 13 n. 10. Commerce reasoned that, “[i]f Habas’ quarterly home market prices and costs did in fact track each other in ‘lockstep’ ..., one would expect [Habas’] quarterly profit percentage on home market sales to be consistent.” Remand Determination at 36. However, the results of Commerce’s analysis indicated that Habas’ profits did not remain constant, as prices did not “increase in relatively the same amount as costs.” Id. Indeed, Commerce found that “prices within a given quarter ... fluctuate^] by more than costs fluctuated] over the entire annual POR.” Id. at 36-37. Habas minimizes Commerce’s price volatility evaluation as a mere “secondary analysis,” with “no bearing on whether Habas’ prices and costs were sufficiently correlated to warrant application of multiple cost-averaging periods.” See PL’s Brief at 37. Even more to the point, Habas asserts that Commerce’s analysis is methodologically flawed, in that it fails to account for the fact that Habas’ sales database is prepared on a daily basis, while its cost database is presented on a quarterly basis. Thus, according to Habas, “[i]f the sales are not evenly distributed within a quarter, then the correlation with price will be diminished and ... profitability will fluctuate.” Id.; see also Remand Determination at 33. Neither the Government nor the Domestic Producers point to anything to refute Habas’ critique. The centerpiece of Habas’ case on the relationship between its costs and its prices, however, is its “correlation coefficient” analysis. Specifically, Habas’ brief uses data from the record — presented in the form of tables and graphs — to depict both the quarterly indices of price and cost for key product models during each quarter of the POR, and the actual quarterly prices and costs across the POR. See generally PL’s Brief at 28-35. Habas contends that “any fair comparison of the price and cost curves ... shows a substantial correlation between and costs across the POR.” PL’s Brief at 30. In fact, according to Habas, “the overall correlation coefficient between price and quarterly [cost of manufacturing] for all [product models] is 0.9928, and for individual [models], the coefficient of correlation is above 0.97 for all but one.” See PL’s Brief at 30; see also id. at 30 n. 8 (explaining that a correlation coefficient of 1 indicates a “perfect linear relationship” between two variables). Habas concludes that “a correlation coefficient of 0.992 definitively shows that price and cost are closely correlated.” See PL’s Brief at 31. In a related set of figures, Habas also depicts Commerce’s POR average cost, graphically illustrating the extent to which the use of POR average cost (as Habas puts it) “introduces significant inaccuracies in every quarter,” particularly in those quarters where the change in cost was most pronounced. See PL’s Brief at 33-35. Based on the information presented in its tables and graphs, Habas questions what Commerce could possibly mean when the agency finds that the use of POR-average cost is “more accurate” than quarterly costs. See PL’s Brief at 35. According to Habas, the evidence depicted in its brief demonstrates overwhelmingly that “[p]rice is correlated with quarterly cost, while POR-average cost significantly distorts the price-cost relationship in every quarter of the POR.” PL’s Brief at 35. Invoking the doctrine of exhaustion of administrative remedies, the Government and the Domestic Producers protest that Habas’ statistical correlation argument should be disregarded on the grounds that it is “entirely new.” See Def.-Ints.’ Reply Brief at 7-9, 13 n. 10; see also Def.’s Response Brief at 25. It is unclear, however, exactly what is assertedly “new”— Habas’ reliance on the statistical concept of a correlation coefficient, Habas’ use of graphic formats (i.e., tables and graphs) to present record evidence, or something else. Certainly neither the Domestic Producers nor the Government claim to be surprised by Habas’ basic contention; as the Domestic Producers candidly acknowledge, Habas has consistently argued that “[its] cost and sales [prices] were sufficiently matched or correlated.” See Def.Ints.’ Reply Brief at 8. Moreover, as Habas points out, Commerce itself “opened the door” to Habas’ correlation coefficient analysis by introducing the concept of “correlation” for the first time in the Remand Determination. See PL’s Reply Brief at 14; Remand Determination at 6 (stating that “there must be a close correlation between the costs to produce the product during the shorter period and the sales price of that same merchandise during the same period”). Further, the application of exhaustion principles in trade cases is a matter of judicial discretion. See Corus Staal BV v. United States, 502 F.3d 1370, 1381 (Fed. Cir.2007) (and cases cited there); see also Def.-Ints.’ Reply Brief at 8 (acknowledging that application of doctrine of exhaustion is committed to court’s discretion). On the merits of Habas’ correlation coefficient analysis, the Government and the Domestic Producers argue that even a perfect one-to-one correlation does not necessarily establish causation, as a matter of logic. See Def.’s Response Brief at 11; Def.-Ints.’ Reply Brief at 9-10. As the Domestic Producers phrase their point: “[W]hile correlation coefficients measure the strength and direction of a relationship between two variables, they do not demonstrate the cause of the relationship, and in particular, cannot suffice to demonstrate that the correlation is caused by direct temporal links between input costs and output prices within a quarter.” See Def.Ints.’ Brief at 10. The Government makes the same argument: “[E]ven if ... costs and prices ... moved in the same direction, and ... appeared to correlate to each other, it does not mean that one caused the other.” See Def.’s Response Brief at 11. Habas argues that there has been no “substantive reply” to its “proof of a strong statistical correlation between its costs and its prices,” and that the absence of a “substantive refutation ... is an admission that Habas is correct.” See Pl.’s Reply Brief at 1, 8,11,14. And, on a more basic level, Habas questions the fundamental fairness of yet another constantly moving target — i.e., a test for cost/price relationship that initially required Habas to demonstrate a “direct relationship” between its production costs and its sales within a quarter, which was then subsequently recast to require proof of a “correlation ” and which is now seemingly being transformed into a requirement that Habas establish actual “causation.” Habas asserts that it is being singled out, and is, in effect, being forced to “satisf[y] a more rigorous test than