Full opinion text
OPINION RESTANI, Judge. This matter is before the court on cross Motions for Judgment on the Agency Record, pursuant to USCIT Rule 56.2, by Borden, Inc., Gooch Foods, Inc., and Hershey Foods Corp. (collectively “Borden” or “the domestic industry”) and Delverde, SrL and Delverde USA, Inc. (collectively “Delverde”). This matter is, also before the court on a Motion for Judgment on .the Agency Record by F.lli De Ceceo di Filippo Fara San Marti-no S.p.A. (“De Ceceo”). The International Trade Administration, U.S. Department of Commerce’s (“Commerce” or “the agency”) determinations under review are Certain Pasta from Italy, 61 Fed.Reg. 30,326 (Dep’t Commerce 1996) (final determination) ^‘Final Determination”] and 61 Fed.Reg. 38,547 (Dep’t Commerce 1996) (amended final determination and antidumping order). Borden asks the court to find that Commerce erred in failing to calculate dumping margins for Delverde using transactionspe-cific export prices, rather than weighted-average prices, pursuant to 19 U.S.C. § 1677f-1(d)(1) (1994), the “targeted dumping” provision. Borden also challenges Commerce’s commission offset methodology. Delverde argues that Commerce, during Delverde’s level of trade inquiry, unlawfully denied its request for a constructed export price (“CEP”) offset, an adjustment to normal value which Delverde claims would have led to a de minimis dumping margin. The defendant agrees in part and requests a remand to correct analytical errors. Defendant-intervenor Borden opposes this request. Delverde also argues that Commerce erred in rejecting the capital asset depreciation expense component of the cost of production data submitted by Delverde affiliate Tamma Industrie Alimentan, SrL (“Tamma”). De Ceceo asks the court to review the 46.67% antidumping margin assigned to it by Commerce based on an “adverse facts available” analysis. The court considers issues raised by Borden, Delverde, and De Ceceo separately, in that order. The facts relating to each issue will be stated separately. JURISDICTION The court has jurisdiction pursuant to 28 U.S.C. § 1581(c) (1994). STANDARD OF REVIEW The court must uphold Commerce’s Final Determination unless it is “unsupported by substantial evidence on the record, or otherwise not in accordance with law.” 19 U.S.C. § 1516a(b)(l)(B) (1994). I. Borden A. Targeted Dumping Investigation Background On May 12, 1995, Borden filed a petition with Commerce seeking the imposition of antidumping and countervailing duties against certain pasta from Italy, pursuant to 19 U.S.C. § 1677f-l(d)(l)(B). Pertain Pasta from Italy, 60 Fed.Reg." 30,268, -30,268 (Dep’t Commerce 1995) (notice of initiation of investigation). Borden used average weekly retail prices to demonstrate that, for all Italian exporters, certain regions of the United States experienced significantly different pricing than others, submitting that this suggested targeting and justified the use of individual (ie. transaction-specific) export prices rather than weighted-averages to detect dumping without masking targeting. Letter from Borden to Commerce (Oct. 20, 1995), P.R. 355, PI. Borden’s App., Tab 4. Commerce initiated a sales at less than fair value (“LTFV”) investigation on June 1, 1995. Certain Pasta From Italy, 60 Fed. Reg. at 30,268. Commerce found Borden’s attempts to demonstrate targeting insufficient reason to depart from its normal methodology and employ instead a comparison with transaction-specific prices. Final Determination, 61 Fed.Reg. at 30,329. Commerce did not perform an independent targeted dumping analysis of the. data. Commerce did find a 2.8% dumping margin using the weighted-average to weighted-average comparison. Id: at 30,365. On February 14, 1997, Borden moved for Judgment on the Agency Record, alleging that Commerce erred in refusing to conduct a transaction-specific comparison and that such a comparison would have produced a dumping margin of 6.14%. PI. Borden’s Br. at 6. Discussion The issue before the court is whether Commerce erred in rejecting Borden’s targeted dumping allegation due to the inadequacy of Borden’s pricing pattern evidence. The court concludes that Commerce erred by failing to articulate the methodology or standards by which a targeted dumping allegation would be judged and failed to clearly allocate burdens of production and analysis in the targeting context. 1. Transaction-Specific Price Comparisons, Not Weighted-Average Price Comparisons, are the Exception Requiring Justification Section 1677f — 1(d)(1) of Title 19 of the United States Code describes the LTFV price comparison analysis. The first part of the section describes the normal methodology- (A) In general [T]he administering authority shall determine whether the subject merchandise is being sold in the United States at less than fair value- (i) by comparing the weighted average of the normal values to the weighted average of the export prices (and constructed export prices) for comparable merchandise, or (ii) by comparing the normal values of individual transactions to the export prices (or constructed export prices) of individual transactions for comparable merchandise. 19 U.S.C. § 1677f-l(d)(l)(A). The statute continues with a description of conditions under which Commerce might deviate from this method. 19 U.S.C. § 1677f-l(d)(l)(B). (B) Exception The administering authority may determine whether the subject merchandise is being sold in the United States at less than fair value by comparing the weighted average of the normal values to the export prices (or constructed export prices) of individual transactions for comparable merchandise, if - (i) there is a pattern of export prices (or constructed export prices) for comparable merchandise that differ significantly among purchasers, regions, or periods of time, and (ii) the administering authority explains why such differences cannot be taken into account using a method described in paragraph (l)(A)(i) or (ii). Id. Commerce reasonably interprets the statute to mean that the weighted-average to weighted-average comparison is the normal method required by law, but that it may, though is not required to, deviate from this requirement under certain conditions and then only upon explicit justification of its decision. Borden’s opposite interpretation, that Commerce must default to the transaction-specific methodology and justify its use of weighted-average price comparisons, is not borne out by the statutory language. 19 U.S.C. § 1677f-l(d)(1)(B)(ii). The practice Borden recommends is contrary to law as it now exists; it reflects the law as it was prior to the changes enacted by the Uruguay Round Agreements Act (“URAA”). Pub.L. No. 103—465, 108 Stat. 4809 (1994). Statement of Administrative Action accompanying the URAA, at 842, H.R. 5110, H.R. Doc. No. 316, Vol. 1, 103d Cong., 2d Sess. (1994) (“SAA”). Borden quotes the SAA selectively: Although current U.S. law permits the use of averages on both sides of the dumping equation, Commerce’s preferred practice has been to compare an average normal value to individual export prices in investigations and reviews. In part, the reluctance to use an average-to-average methodology has been based on a concern that such a methodology could conceal “targeted dumping.” In such situations,, an exporter may sell at a dumped price to particular customers or regions, while selling at higher prices to other customers or regions., SAA at 842 (emphasis supplied). The passage continues, however, to describe as normal the methodology set forth in 19 U.S.C. § 1677f—1(d)(1)(A). Consistent with the Agreement, new [19 U.S.C. § 1677f-l(d)(l)(A) ] provides that in an investigation, Commerce normally will establish and measure dumping margins on the basis of a comparison of a weighted-average of normal values with a weighted-average of export prices or constructed export prices. SAA at 842. The SAA explains that the exception to the normal methodology is an antidote to targeted dumping: New [19 U.S.C. § 1677f-l(d)(l)(B) ] provides for a comparison of average normal values to individual export prices or constructed export prices in situations where an average-to-average or transaction-to-transaction methodology cannot account for a pattern of prices that differ significantly among purchasers, regions, or time periods, i.e., where targeted dumping may be occurring. Before relying on this methodology, however, Commerce must establish and provide an explanation why it cannot account for such differences through the use of an average-to-average or transaction-to-transaction comparison. In addition, the Administration intends that in determining whether a pattern of significant price differences exist, Commerce will proceed on a case-by-ease basis, because small differences may be significant for one industry or one type of product, but not for' another. Id. at 843. Averages allow higher prices to cancel out some amount of dumping, see Potassium Chloride from Israel, 50 Fed.Reg. 4,560, 4,562 (Dep’t Commerce 1985) (final determination), but transaction-specific. price comparisons are statistically biased toward a dumping finding, as high priced sales are disregarded, and only the lower priced ones are included in the calculation. Congressional Budget'Office, How the GATT Affects U.S. Antidumping and Countervailing Duty Policy, 33-35, 66 (1994). [The transaction-specific methodology] contains a statistical error that will yield a positive average margin of dumping in every real-world situation except the one that would take place if every single transaction (including both the home market and U.S. sales) during the ... period of investigation occurred at the same price.... If the true margin of dumping is quite high, say, 20 percent, this statistical bias will not be very large.... However, if the estimated margin of dumping is small, the bias could be several times the magnitude of the true margin of dumping; in such cases it is more likely that no dumping exists. Tracy Murray, The Administration of the Antidumping Duty Law by the Department of Commerce, in Down in the Dumps 23, 36-37 (Richard Boltuck & Robert E. Litan eds., 1991). Economists and politicians argued masking versus amplifying during the Uruguay Round. See Congressional Budget Office, at 33-35, 66; Murray at 36-37. Borden’s argument revisits that debate. At the Uruguay Round, the U.S. argued, consistent with its past administrative practice,, that comparisons using weighted-average prices mask dumping. Trade Agreements Resulting from the Uruguay Round of Multilateral Trade Negotiations: Hearings before the Committee on Ways and Means and its Subcommittee on Trade, 103d Cong. 73, at 401 (1994) (statement of Jeffrey E. Gartner, Under Secretary of Commerce for International Trade) [“House Hearings ”]. The prevailing view, however, which was made part of U.S. domestic law through the URAA, was that transaction-specific comparisons ran the different risk of amplifying dumping margins. See URAA, Pub.L. 103-465, § 219; 108 Stat. 4855-57 (1994). During the-Uruguay Round meetings, the U.S., out of concern that targeted dumping might be masked by weighted-average to weighted-average comparisons, insisted upon a provision in the international agreement on antidumping permitting the use of individual export prices in investigations when targeting is occurring. House Hearings, at 401. Despite the inclusion of this provision, the fact remains that antidumping law under the URAA instructs Commerce to be more concerned about amplification of dumping margins through the use of transaction-specific prices than about masking them with averaging. Serampore Industries Pvt, Ltd. v. United States, 11 CIT 866, 874, 675 F.Supp. 1354, 1360-61 (1987) (agency practice of excluding non-dumped sales is to prevent a foreign producer from masking dumping) and Drycleaning Machinery from Germany, 56 Fed.Reg. 66,838, 66,840 (Dep’t Commerce 1991) (final results of antidumping duty administrative review) (averaging rejected because of masking of dumping), upon which Borden relies, no longer represent normal practice. 2. Commerce Rejection of Borden’s Targeting Allegations Each of three times Borden submitted a petition alleging targeting, Commerce articulated various reasons for rejecting it. Commerce advised Borden, after its first submission, that targeted dumping analyses should be company-specific, based on product-, not brand-specific calculations, and should use exporter, not retail prices. Memorandum from Team to Barbara R. Stafford (Nov. 8, 1995), at 2, P.R. 440, Pl. Borden’s App., Tab 5 at 2. Borden’s first revision was criticized for its weak statistical analysis that demonstrated price differences but not a pattern. Memorandum from Team to Barbara R. Stafford (Dec. 8, 1995), at 2, P.R. 507; Pl. Borden’s App., Tab 8, at 3. Commerce faulted Borden’s second revision for failing to address the statutory criteria of 19 U.S.C. § 1677f-l(d)(l)(B)(ii). Final Determination, 61 Fed.Reg. at 30,329. Specifically, Commerce noted that Borden failed to demonstrate a pattern of significant price differences, predetermined the results through customer groupings, failed to provide benchmark prices, and failed to explain why the demonstrated price differences could not be accounted for by weighted-average comparisons. Id. Commerce thus found the pattern of demonstrated price differences to be “predetermined” by the initial composition of the customer groups, that the domestic industry failed to supply any relevant “benchmark” prices that would demonstrate the “significance” of price variations, and that the analysis did not address the statutory criteria of 19 U.S.C. § 1677f-l(d)(l)(B)(ii). Final Determination, 61 Fed.Reg. at 30,329. Commerce concluded that statistical flaws in Borden’s analysis precluded a finding of the requisite pattern of pricing differences indicative of targeting. See id. The record shows Borden failed to remove outliers, Letter from Delverde to Commerce (Dec. 4, 1995), at 6, C.R. Doc. 179; Def.-Int. Del-verde’s App., at AII-13, presumed the existence of a pattern and manipulated the data to show it, Letter from Borden to Commerce (Feb. 13, 1996), Pub. A.R. 680, Pl. Borden’s App., Tab 9, at 7, and failed to standardize the data, for example, by running regressions, to account for or rule out differences that could be explained by factors other than targeted dumping. Borden did not control for volume or customer status. Borden also failed to prove the null hypothesis, which would have shown the intrinsic probability of finding a pattern even where none existed. Instead, Borden simply grouped customers by price and performed univariate and linear modeling procedures. See id. Borden complains that Commerce rejected its revised petitions, despite Borden having addressed all the concerns Commerce raised regarding Borden’s first petition. In raising those concerns, Commerce did not guarantee a particular finding, even upon Borden’s strict conformity of its revised petition with Commerce’s response. At best, Commerce’s criticism of the first petition amounted to an incomplete expression of minimum conditions for an effective petition, not a roadmap toward an affirmative finding. ' Commerce found the results in Borden’s third petition predetermined because Borden grouped the customers by price and then used statistical methods to show differences in price between the groups. Final Determination, 61 Fed.Reg. at 30,329. Borden understood Commerce as having required such grouping when Commerce instructed Borden not to examine each of the multitude of transactions individually. Pl. Borden’s Br. at 26. What Commerce did say was that there was a statutory preference for weighted-average price comparisons and that simple averages might distort where certain individual customers had a great number of transactions. Memorandum from Team to Barbara R. Stafford (Dec. 8,1995), at 3, P.R. 507, Pl. Borden’s App., Tab 8, at 3. Grouping the data by price, Borden’s results better reflected its method than the data. Borden’s statistical analysis demonstrated the existence of price variation, showing a distribution but not a pattern. See Final Determination, 61 Fed.Reg. at 30,329. Even if Borden had found a pattern, its analytical technique would have been too simple to convey the statistical significance of its findings. Commerce concluded that Borden’s petitions failed to show targeting, because they failed to employ analytical techniques which could have shown it. Borden did not understand the targeted dumping provision in the statute, in part for the failure to understand what targeting is. Borden considers targeted dumping “the practice of selling to selected customers or regions at different and preferential prices as compared to the prices charged to other customers or regions.” PI. Borden’s Br. at 7. Thus, Borden searched for statistical evidence of price variance. By Borden’s definition, however, most pricing would constitute targeted dumping, in that there is price variance along multiple dimensions in many markets. Certainly, not all price variation, not even all statistically significant variation, results from targeted dumping. The concept of targeted dumping is that a company might not be able to, or might choose not to, use a dumping strategy toward a whole market but might strategically focus on a more narrowly defined market. Michael Coursey, Comment, in Down in the Dumps, at 240. To ferret out this more complicated dumping, the statute instructs Commerce to look not only at the magnitude of price variance but also for a pattern of significant price differences. 19 U.S.C. § 1677f-l(d)(l)(B)(i). Nonetheless, Commerce failed from the outset to convey standards for satisfying the statute. Borden also maintains that, contrary to Commerce’s view, the weighted-average price Borden submitted was adequate to establish a benchmark price. Commerce appears to have wanted a U.S. price for comparison. Final Determination at 30,329. Using the average for comparison shows only divergence from the average price, highlighting the magnitude of price variation, but not demonstrating a pattern. Commerce complained that Borden failed to demonstrate that price differences could not be taken into account using weighted-average prices. Borden intended its presentation of the masking effect of averaging to satisfy this requirement. Because Commerce did not understand Borden to have provided evidence of targeting at all, Commerce was not satisfied that Borden had made an adequate showing that the targeting would be masked by averaging. Even if Commerce accepts Borden’s targeted dumping petition, however, the agency would not necessarily err in rejecting transaction-specific methodology. 19 U.S.C. §§ 1677f-l(d)(l)(A)-(B). Commerce “shall [compare] ... the weighted average of the normal values to the weighted average,” 19 U.S.C. § 1677f-l(d)(l)(A)(i) (emphasis supplied), but Commerce “may [compare] ... the weighted average of the normal values to the export prices ... of individual transactions for comparable merchandise,” 19 U.S.C. § 1677f-l(d)(l)(B) (emphasis supplied). Under the appropriate circumstances Commerce has the discretion to not apply the targeted dumping exception to its normal methodology, even upon a finding of targeted dumping. 3. Commerce Has Not Fulfilled Its Duty to Articulate Standards Whether Commerce was reasonable or not in thrice rejecting Borden’s allegation, the court finds that in rejecting the targeting petition on methodological grounds, Commerce never reached the question of whether the data submitted by Delverde revealed a pattern of price differences adequate to trigger transaction-specific price comparisons. Commerce concluded merely that methodological inadequacies in Borden’s three petitions precluded a finding of targeting based on that methodology. Under the URAA, however, Commerce may not abandon the targeting inquiry simply because of a petitioner’s lack of statistical sophistication. See 19 U.S.C. § 1677f-l(d)(l)(B). Despite pointed questioning at oral argument, defendant could not describe in a clear way how targeting could be demonstrated. Commerce has only said what it does not want; it has not made clear what it requires. Thus, the court is not convinced that Commerce is fulfilling its duty to rationally decide whether it should perform a transaction-specific investigation when the statutory prerequisites are met. Indeed, Commerce has not answered several critical questions: What methods could show targeting? By what standards does Commerce evaluate correctly analyzed data? What does Commerce deem a pattern significant enough to constitute evidence of targeting? Would acceptable methods, applied to the Delverde data, reveal targeting? Commerce failed to articulate the standards by which it would determine that a “pattern of export prices” that “differ significantly” did or did not exist. Id. Commerce was obliged to articulate the standards on which it based its decision. National Steel Corp. v. United States, 18 CIT 1126, 1132-33, 870 F.Supp. 1130, 1136-37 (1994) (remand to Commerce to articulate standards for determining “non-aberrant” margins where agency failed to provide explanation of term). While the negative rationale Commerce gives for rejecting Borden’s final petition may be reasonable when viewed in isolation, the demand by the domestic industry for standards is no less reasonable. The court finds that Commerce erred in failing to articulate the standards by which it will evaluate the targeted dumping petitions. To facilitate future inquiries, Commerce will need at some point to explain what targeted dumping is, what methods will identify or rule out the pricing patterns referred to by the statute, what degree of significance in those patterns will trigger Commerce to exercise its discretion to make a case-by-case determination to depart from its normal methodology, and on what basis it will make that decision. For example, regarding the significance of the pattern, bearing in mind that Congress intended a case-by-case analysis with reference to variations in price sensitivity by industry, SAA at 843, Commerce might suggest a calculus which relates pricing patterns with price elasticity. In May 1997, Commerce independently announced its intention to issue policy bulletins setting forth specific criteria for targeted dumping “as the Department develops its practice in this area,” Antidumping Duties; Countervailing Duties, 62 Fed.Reg. 27,296, 27,374 (Dep’t Commerce 1997) (final rule) [“Final Rule ”]. Recognizing that Commerce as yet may not be prepared-to articulate a methodology to be employed by domestic industries alleging targeting generally, and in light of-its expressed intention to issue new regulations, id., the court leaves it to Commerce to decide, on remand, whether to articulate the standards by which it evaluates a domestic industry’s targeted dumping petitions, in general or for only this case, or to conduct its own analysis to determine whether there is targeted dumping based on the data submitted by respondent. The court has considered Borden’s assertion that Commerce, not the domestic industry, was obliged by statute to assess whether targeted dumping had occurred and that Commerce therefore improperly shifted its own burden to Borden by denying Borden’s petition solely on the inadequacies of Borden’s analysis, without itself examining the data to which it had equal access. The government rejoins that Commerce was reasonable in requiring a minimally sufficient allegation of targeted dumping from the domestic industry prior to proceeding with its own inquiry into targeted dumping. The SAA refers to the collection of transaction-specific data “so Commerce may determine ... whether the exception for targeted dumping is applicable.” SAA at 843. This does not clarify whether Commerce would make its determination on the basis of its own or the domestic industry’s analysis. The statute is silent in that regard. 19 U.S.C. § 1677f-1(d)(1)(B). The court will not dictate the division of labor between Commerce and the domestic industry; Commerce is better positioned to decide that issue. In the absence of statutory direction as to the allocation of responsibility between Commerce and the domestic industry in a targeted dumping inquiry, the matter is left to Commerce’s discretion to develop some reasonable system. Chevron USA Inc. v. Natural Resources Defense Council, 467 U.S. 837, 843, 104 S.Ct. 2778, 81 L.Ed.2d 694 (1984). Thus far, Commerce has not clearly explained its allocation of responsibility for performing the targeting analysis. Commerce has recently spoken somewhat mysteriously of the “assistance of the domestic industry.” Final Rule, 62 Fed. Reg. at 27,374. Unless Commerce determines that it will perform its own analysis based on the data on hand, Commerce shall specify what burden it assigns petitioners. 4. Pricing Pattern Standards In a sub-argument relevant to its targeting challenge Borden asserts that Commerce’s application of different standards as to targeted dumping and the level of trade (“LOT”) adjustment is in error. The court is not convinced that the “rigid” standard employed by Commerce in analyzing the domestic industry’s allegations of a pattern of price differences for purposes of targeted dumping and the allegedly inconsistent “liberal” standard applied when analyzing a pattern of export prices in the LOT context taken together constitute error. See PI. Borden’s Br. at 9. Because the domestic industry’s allegation of a pattern of price differences to show targeted dumping allegedly “met and surpassed” the standards Commerce accepted as showing a pattern of price differences in its LOT analysis, Borden calls Commerce’s decision to accept the pricing patterns for LOT but to reject them for targeted dumping arbitrary and capricious. PI. Borden’s Br. at 10. Commerce is instructed that a LOT adjustment may be made if the existence of different selling functions “is demonstrated to affect price comparability, based on a 'pattern of consistent price differences ” relating to different selling practices at different levels of trade. 19 U.S.C. § 1677b(a)(7)(A)(ii) (1994) (emphasis supplied). Individual export prices may be used for comparison under the targeted dumping provision if Commerce finds “there is a pattern of export prices (or constructed export prices) for comparable merchandise that differ significantly among purchasers, regions, or periods of time.” 19 U.S.C. § 1677f-l(d)(l)(B)(i). Borden refers to the language of these two statutory directives as “almost identical” and their application “very different and inconsistent.” Pl. Borden’s Br. at 9. In an elaborate, if contradictory, textual analysis, Borden argues that identical words used in different parts of the same statute are intended to have the same meaning. Borden contends that the difference between the “almost identical” provisions is the use of the word “consistent” in 19 U.S.C. § 1677b(a)(7)(ii) and “significantly” in 19 U.S.C. § 1677f-1(d)(1)(B). Relying on the dictionary, Borden argues that the word “consistent” should connote a higher, more rigorous standard than “significant.” Significant merely means “a noticeably or measurably large amount” or “caused by something other than mere chance;” whereas consistent means “showing steady conformity.” Webster’s Ninth New Collegiate Dictionary, at 1096, 280 (1985). Price differences can be “consistent” without being “significant,” as when prices to all end-users exceed all prices to wholesalers by a very small percentage. Likewise, prices can differ “significantly” without doing so “consistently,” as in the case of a customer whose price exceeds that of another by 20% one week but is 25% lower the next. It is correct that where the identical word or phrase is used more than once in the same act, there is a presumption that they have the same meaning. See Sutherland Stat. Const. § 46.06 (5th Ed. 1992) at 119, 120; Cemex, S.A. v. United States, 16 CIT 251, 254, 790 F.Supp. 290, 294 (1992), aff'd, 989 F.2d 1202, 1993 WL 26767 (Fed.Cir.1993) (court may apply parallel constructions where phrase appears more than once in same statutory provision); Champagne v. United States, 35 Fed. Cl. 198, 210 (1996) (“common sense requires that the same words used twice in the same act should have the same meaning.”). This rule is tempered, however, by the canon that the same word or phrase can possess different meanings in different contexts. See Cemex, 16 CIT at 254, 790 F.Supp. at 294 (“[w]here, however, the contexts and purposes of the provisions differ, parallel constructions are inappropriate”); New England Telephone and Telegraph Co. v. Public Utilities Commission of Maine, 742 F.2d 1, 7 (1st Cir.1984) (eonstru-ing the identical word differently because there existed a “sufficient difference in the functions of the two sections to justify assigning a different scope to the same word.”), cert. denied, 476 U.S. 1174, 106 S.Ct. 2902, 90 L.Ed.2d 988 (1986). Commerce concedes that different methodologies were used in the LOT adjustment to normal value reported by Commerce in the Final Determination, 61 Fed.Reg. at 30,331, and Commerce’s denial of Borden’s targeting petition, PI. Borden’s Br. at 30, but that difference was justified and permitted, because the two provisions differ in ways that render the different standards appropriate. The way Commerce treats patterns of price differences for purposes of LOT adjustments is largely irrelevant to a targeted dumping allegation. Moreover, the LOT adjustment inquiry is mandatory and standard, while the targeted dumping provision is guiding and an exception. The standards are understandably different because they were designed for different circumstances, one of which is the norm and the other an exception. The targeted dumping statute, which is labeled “exception,” includes the additional requirement that Commerce show why there should be a deviation from the normal procedure. 19 U.S.C. § 1677f-l(d)(l)(B). Commerce “may” deviate from the normal methodology defined in 19 U.S.C. § 1677f-1(d)(1)(A) if the conditions for the exception are met and the administering authority explains why such differences cannot be taken into account using [the normal, weighted-average to weighted-average or transaction-specific to transaction-specific comparison] a method described in paragraph (l)(A)(i) or (ii). 19 U.S.C. § 1677f-l(d)(l)(B)(ii). This itself justifies application of a stricter standard in that inquiry. B. Challenge to Offset of the Difference in Commissions in the United States and Foreign Markets. Pursuant to 19 U.S.C. § 1677b(a)(6)(C)(iii) (1994) and 19 C.F.R. § 353.56(a)(2) (Apr. 1, 1996), Commerce adjusted normal value for differences in the circumstances of sale between sales of pasta made to the United States and sales made in Italy. See Final Determination Calculation Memorandum (June B, 1996), P.R. Doe. 848, Def.-Int. Borden’s App., Tab 54; Final Margin Computer Printout, at lines 292-95, C.R. Doc. 407, PI. Borden’s App., Tab. 10. Where there were commissions in both markets, the agency did not simply adjust for the difference in commission expenses incurred by adding U.S. direct expenses to and deducting foreign market direct expenses from normal value, but instead added home market indirect selling expenses to home market commissions to offset the additional commissions incurred in selling to the United States. See id. Commerce’s methodology was the same in both the Preliminary and Final Determinations. See Certain Pasta from Italy, 61 Fed. Reg. 1,334, 1,349 (Dep’t Commerce 1996) (preliminary determination); Preliminary Determination Calculation Memorandum (Dec. 14, 1995), at 7, Def.-Int. Delverde’s App., at A-62. Although Borden noted, as a clerical error after the Final Determination, PI. Borden’s Reply Br. at 14, that the calculations did not match the Final Determination language, this does not amount to a timely challenge to the actual methodology used. If the language used in the Final Determination is in error, it may be corrected, but the methodology actually employed was set early on and could not be challenged for the first time after the Final Determination. If the challenge had been timely raised, Commerce could have addressed it before the Final Determination. Accordingly, this challenge is rejected for failure to exhaust administrative remedies. 28 U.S.C. § 2637(d) (1994). II. Delverde A. Depreciation Recalculation Facts Commerce examined Delverde affiliate Tamma’s data in the course of evaluating Delverde’s costs of production. In 1994, Tamma revised its method of calculating its depreciation to double the useful lives of the company’s assets. Tamma Verification Report for Cost of Production and Constructed Value Data (Apr. 19, 1996), at 27-28, Pl. Delverde’s App., at AII-1-2. These changes were made for income tax purposes. Final Determination, 61 Fed.Reg. at 30,354. The relevant effect of changing the rate of depreciation of Tamma’s productive capital assets is to reflect lower marginal production costs, permitting comparison with lower export or constructed export prices without a finding of dumping. Commerce rejected the new calculation as not standard for antidumping inquiries and substituted a depreciation calculation consistent with its normal calculation of depreciation in antidumping investigations. Final Determination, 61 Fed.Reg. at 30,354-55. Calculations based on Tamma’s calculation allegedly would result in a de minimis dumping margin. Discussion The basic rules for the calculation of cost of production, 19 U.S.C. § 1677b(b)(3) (1994), are supplemented by 19 U.S.C. § 1677b(f)(l)(A) (1994), referring specifically to depreciation in the antidumping context. Costs shall normally be calculated based on the records of the exporter or producer of the merchandise, if such records are kept in accordance with the generally accepted accounting principles of the exporting country (or the producing country, where appropriate) and reasonably reflect the costs associated with the production and sale of the merchandise. The administering authority shall consider all available evidence on the proper allocation of costs, including that which is made available by the. exporter or producer on a timely basis, if such allocations have been historically used by the exporter or producer, in particular for establishing appropriate amortization and depreciation periods, and allowances for capital expenditures and other development costs. 19 U.S.C. § 1677b(f)(l)(A). This approach to accounting standards is consistent with pre-URAA practice. SAA at 834. The SAA provides further detail regarding when it may be appropriate for Commerce to reject a reported depreciation, even though the reported depreciation meets the first statutory condition of consistency with home country accounting standards. Id.; see also 19 U.S.C. § 1677b(f)(l)(A). In determining whether a company’s records reasonably reflect costs, Commerce will consider U.S. generally accepted accounting principles employed by the industry in question. For example, a company’s records might not fairly allocate the cost of an asset if a firm’s financial statements reflect an extremely large amount of depreciation for the first year of an asset’s life, or if there is no depreciation expense reflected for assets that have been idle. In such a situation, it would be appropriate for Commerce to adjust depreciation expenses. Costs shall be allocated using a method that reasonably reflects and accurately captures all of the actual costs incurred in producing and selling the product under investigation or review. SAA at 834-35; see also Certain Cold-Rolled and Corrosion-Resistant Carbon Steel Flat Products from Korea, 62 Fed.Reg. 18,404, 18,443 (Dep’t Commerce 1997) (final results of antidumping duty administrative review) (allocation based on home country accounting practice rejected for not accurately capturing costs). In other words, the SAA counsels Commerce to refer to U.S. GAAP, as employed by the importer’s industry generally, as a guide to the reasonableness of the importer’s allocations based on home country accounting principles. SAA at 834-35. The SAA further contemplates that Commerce will make adjustments when it concludes that the importer’s calculations do not reasonably reflect costs. Id. In the original investigation, Commerce considered the depreciation methods presented in Tamma’s cost of production response. At verification, the Department found that Tamma had replaced its former depreciation schedule, based on industry-specific asset lives authorized under the Italian Civil Code and GAAP, with a new, more tax-beneficial, while still Italian GAAP consistent, depreciation schedule. Final Determination, 61 Fed. Reg. at 30,354. Commerce rejected Tamma’s reported depreciation method, as “contrary to sound accounting principles and the Department’s practice.” Id. at 30,355. Commerce reasoned that the methodological change “was not the result of new events, changing conditions, experience, or additional information.” Id. at 30,354. Instead, Commerce concluded that the change distorted costs for the purposes of an antidumping analysis, and implied that the change was inconsistent with the Italian Civil Code standards for calculating the useful life of assets for the pasta industry. Id. Thus, Commerce substituted a depreciation calculation consistent with its normal practice in antidumping investigations. Id. at 30,333. Delverde argues that, in rejecting its revised capital asset depreciation expenses, Commerce disregarded its statutory obligation under 19 U.S.C. § 1677b(f)(l)(A) to use a respondent’s reported costs where such costs are maintained in conformity with home market GAAP, are according to the company’s historical use, and reasonably reflect the cost of producing subject merchandise. Del-verde presents the new calculations as historical, alleging that Tamma changed to the new method before the commencement of the an-tidumping investigation. The home country legality (i.e. conformity with Italian GAAP) of the new calculation is evidenced, according to Delverde, by the approval of Tamma’s local auditors. (Of course, they approved the earlier method, selected by Commerce here, as well.) Finally, Delverde argues that the true cost of production is represented by the new calculation because the new method extends the period of depreciation to more closely approximate the age of certain fixed assets associated with production, namely Tamma’s decades old pasta factory and wheat mill. The court finds the statutory .mandate to Commerce under 19 U.S.C. § 1677b(f)(l)(A) to be relatively flexible. The court reads the statutory instruction to clearly state what factors Commerce shall consider but to leave to Commerce’s judgment what relative weight and meaning to give them. First, the question of any alleged bad faith behind Tamma’s changed calculation is not at issue here, regardless of whether the change occurred prior to or during the pasta investigation. Questions of timing and motivation simply distract from the central concern, which is the accuracy of the calculations. Commerce’s mention of the motivation issue, however, is not reversible error, as that was not the focus of its decision. Commerce could, as it did, find Tamma’s tax-based reason to change from an otherwise acceptable useful life to be some evidence of distortion in the absence of evidence of new events, changing conditions, experience, additional information, or a change in underlying economic assumptions which would relate the change to accurate cost accounting. This is not necessarily an additional affirmative legal burden on respondent, though a respondent such as Del-verde might find it advisable and persuasive to present an explanation for an otherwise seemingly arbitrary change. Second, Commerce properly determined that Tamma’s calculation was not historical. Delverde wanted Commerce to read “historically” to mean “in the normal course of business,” PL Delverde’s Reply Br. at 13, but Commerce was entitled to do otherwise, provided its interpretation was reasonable. The term historical can be variously understood to mean prior to the period of inquiry, not in immediate anticipation of the investigation, or long-standing industry practice. These variations suggest some ambiguity to the term, leaving Commerce free to choose an interpretation. Third, the statutory directive is conditional, requiring Commerce to use the company’s own calculation only if satisfied with the accuracy of the cost representations they render. The government contends that Tamma calculated the useful life of its assets and then altered that calculation, albeit as allowed by Italian GAAP, rendering the calculations inaccurate. Further, Tamma’s auditor’s approval of prior use of the calculation Commerce did apply was part of the evidence upon which Commerce could rely reasonably for an accurate reflection of production costs. After weighing the less than clear evidence, Commerce did not agree that Tamma’s new calculations accurately reflected its costs. Thus, it was not obliged to accept them, and could resort to the prior figures. The court finds no error in Commerce’s rejection of the new calculation. Commerce was entitled to reject the revision on the basis of its factual findings, namely that the new calculation was not in accord with Tam-ma’s historical practice and did not reflect . costs accurately. B. Level of Trade Adjustment and CEP Offset Facts Commerce issued a questionnaire to exporters of pasta from Italy, including Del-verde, requesting information concerning their sales in Italy and the United States from May 1,1994 through April 30,1995, and asking the amount, if any, claimed by each responding exporter for the “level of trade” adjustment to normal value authorized by 19 U.S.C. § 1677b(a)(7)(B), along with a statement justifying the claimed amount. Anti-dumping Questionnaire (July 10, 1995), at B-19, P.R. Doc. 66; Pl. Delverde’s App., at A-44. Commerce also asked respondents to report the amount, if any, claimed for the constructed export price offset (“CEP offset”) adjustment to normal value authorized by 19 U.S.C. § 1677b(a)(7)(B). Id. Delverde responded that it claimed a LOT adjustment in the form of a CEP offset pursuant to 19 U.S.C. § 1677b(a)(7)(B), maintaining that it could not quantify price differences attributable to sales at different levels of trade in the Italian home market. Delverde’s Questionnaire Response (Sept. 18, 1995), at 18-19, 30-31, P.R. Doc. 273, Def.Int. Borden’s App. Tab 14, PI. Delverde’s App., at A-45-48. In a letter to all Italian respondents, Commerce asked four questions regarding the requested LOT adjustments and CEP offset. Supplemental Antidumping Questionnaire (Oct. 23, 1995), at 2, P.R. Doc. 359, Pl. Del-verde’s App., at A-51. Delverde responded only to the fourth question, maintaining that the first three questions pertained only to those requesting LOT adjustments, thus not to Delverde. Letter from Delverde to Commerce (Nov. 6, 1995), at 3, P.R. Doc. 432, Pl. Delverde’s App., at A-53, A-55. Delverde explained its inability to quantify price differences by LOT was due to the very small portion of home market sales made at an ex-factory LOT. Id., at 5-6, Pl. Delverde’s App., at A-57-58. In its Preliminary Determination, Commerce denied Delverde’s requested CEP offset, explaining that “U.S. sales were matched to normal values at the same levels of trade.” 61 Fed.Reg. at 1,347. Nonetheless, Delverde was found to have a de minimis dumping margin of .06%. Id. at 1,351. Upon solicitation of its views, Letter from Commerce to Delverde (Jan. 22, 1996), P.R. Doc. 583, PI. Delverde’s App., at A-65, Delverde notified Commerce of its objection to the methodology used in the Preliminary Determination. Letter from Delverde to Commerce (Feb. 5, 1996), at 2-4, P.R. Doc. 636, PI. Delverde’s App., at A-67, A-68-70. Commerce used a different LOT methodology in the Pinal Determination,' one that considered Delverde’s CEP sales to include some selling functions. Final Determination, 61 Fed.Reg. at 30,331, 30,339. On the basis of this analysis, Commerce still concluded that Delverde’s U.S. sales and home market sales were made at the same LOT and, again did not grant Delverde’s requested CEP offset. Id. at 30,354. Delverde argues that Commerce erred in concluding that Delverde’s CEP sales in the United States were made at the same LOT as sales in Italy, Delverde’s home market. Delverde contends that Commerce erroneously attributed selling expenses to Del-verde’s CEP LOT, expenses which had been excluded previously during the calculation of CEP. The government requests a remand to correct errors in its attribution of selling functions to Delverde’s CEP sales for LOT analysis purposes and to reconsider whether Delverde’s CEP sales can be matched to home market sales at the same LOT. It also seeks to reconsider certain adjustments to CEP itself. The parties dispute neither the de minimis volume of ex-factory sales in the home market, nor the difficulty of determining the price effects of different levels of trade in the home market due to the lack of ex-factory sales. Delverde supports remand without a CEP recalculation. Borden opposes remand. Discussion 1. Level of Trade Statute Not all differences between export prices and home market prices are the result of dumping. Therefore, Commerce adjusts the prices it compares to tease out other factors, in part through the level of trade adjustment inquiry. See 19 U.S.C. § 1677b(a)(7)(A). A “level of trade” adjustment to normal value is appropriate when U.S. prices and normal value are compared at different levels of trade, if the different levels of trade involve different selling functions, and price differences attributable to the different levels of trade can be documented. 19 U.S.C. § 1677b(a)(7)(A). In the case of CEP sales, when normal value is determined to be at a more advanced LOT (by stage of distribution) than CEP, but the price effect of the differences in LOT in the home market cannot be determined, a “CEP offset” adjustment is authorized in lieu of a LOT adjustment. 19 U.S.C. § 1677b(a)(7)(B). 2. Commerce’s Level of Trade Methodology Commerce makes certain additions and deductions in calculating normal value, 19 U.S.C. § 1677b(a)(6), and export price or constructed export price, 19 U.S.C. §§ 1677a(e)-(d) (1994). These adjustments are divided into those common to export price and CEP, at 19 U.S.C. § 1677a(c), and those specific to CEP only, at 19 U.S.C. § 1677a(d). The 19 U.S.C. § 1677a(d) deductions are those associated with the transaction between the affiliated importer and the first unaffiliated purchaser. See 19 U.S.C. §§ 1677a(b), (d). In the LTFV analysis, Commerce must compare prices that are at the same LOT by adjusting normal value accordingly where sales are at different levels of trade. 19 U.S.C. § 1677b(a)(7). Thus, where Commerce deems sales in the foreign market and sales in the U.S. to have been made at different levels of trade, the statute provides for an additional, LOT adjustment to normal value, 19 U.S.C. § 1677b(a)(7)(A), or for a CEP offset, in the case of a comparison involving CEP sales where normal value is at a more advanced LOT but data about the relationship between LOT and price is inadequate to determine the size of a LOT adjustment, 19 U.S.C. § 1677b(a)(7)(B). Commerce interprets the mandate generally to mean it must examine.levels of trade, determining where an adjustment is appropriate through a comparison of the selling functions attributable to the respective sales, the expenses for which are reflected in the starting prices. New 19 C.F.R. § 351.412 provides that where the LOT comparison is between normal value sales and export price sales, ie. where the first sale in the United States is to an unaffiliated party, for its LOT inquiry, Commerce compares the normal value starting price (ie. unadjusted price) with the starting export price. Final Rule, 62 Fed. Reg. at 27,414. In cases involving CEP sales, Commerce tries to approximate the LOT comparison made in export price cases. In cases involving CEP sales, Commerce has, thus, developed the following level of trade methodology. See Def.’s Br. at 69-72. First, beginning with the U.S. starting price used for the constructed export price calculation, Commerce adjusts that price to account for selling expenses in the U.S., making the deductions provided for in 19 U.S.C. § 1677a(d). Id. at 69. Commerce then examines the result to determine its LOT, based on the selling expenses remaining therein. Id. at 70-71. Next, Commerce determines whether the LOT of the CEP sale is matched by sales at a comparable LOT in the home market. To that end, Commerce examines unadjusted home market (normal value) sales to determine levels of trade. Id. at 71. Because the 19 U.S.C. § 1677a(c) adjustments are of the type that normally apply to all sales, and are not made prior to comparison in the export price cases, Commerce does not make “(c)” adjustments to either price before the final LOT comparison. Because the § 1677a(d) adjustments allegedly will be made to the constructed export price only, 19 U.S.C. § 1677a(d), Commerce does make the “(d)” adjustments prior to the comparison, in an effort to approximate an export price starting price. If there are no sales in the home market at the same LOT as the “(d)” adjusted CEP sales, Commerce considers making a LOT adjustment, which is an adjustment to normal value intended to offset the price effect of the difference in LOT in the two markets. Def.’s Br. at 71. Commerce determines that levels of trade differ only when sales involve different selling functions and are associated with a consistent pattern of price differences. Id. If home market price data is available but reveals no consistent pattern of price differences, Commerce makes no LOT adjustment. Id. at 72. Where the data is inadequate to make a determination regarding the pattern of prices, Commerce makes a CEP offset, pursuant to 19 U.S.C. § 1677b(a)(7)(B). Id. Commerce applied the latter methodology-in this case. Final Determination, 61 Fed. Reg. at 30,339 (“[F]or CEP sales, we considered the selling functions reflected in the price after the deduction of expenses ... under [19 U.S.C. § 1677a(d) ]”)- Making § 1677a(d) deductions prior to the level of trade comparison is Commerce’s established methodology. Antifriction Bearings (Other than Tapered Roller Bearings) and Parts thereof from France, Germany, Italy, Japan, Singapore and the United Kingdom, 62 Fed. Reg. 2,081, 2,107 (Dep’t Commerce 1997). Commerce’s new regulations also embody this practice. New § 351.412(c)(l)(ii) provides that in the case of CEP sales, Commerce will identify levels of trade based on “the starting price, as adjusted under [19 U.S.C. § 1677a(d) ].” Final Rule, 62 Fed. Reg. at 27,414. 3. Delverde’s Position In its Motion, Delverde argued that Commerce erroneously concluded that Delverde’s CEP was not a price devoid of selling functions, but rather retained the following three selling functions: price lists and/or contracts; inventory services, and freight and/or delivery. At oral argument, defendant conceded that expenses for the three selling functions were deducted from the CEP starting price under 19 U.S.C. § 1677a(d), and were not 19 U.S.C. § 1677a(c) expenses, as Commerce had mistakenly presumed when assessing CEP level of trade. Delverde argues that CEP is defined as a “price” at a specific LOT (ex-factory), devoid of all selling expenses, U.S. or otherwise, because the statute requires the construction of CEP without any movement costs, commissions, direct selling expenses, costs paid by the seller on behalf of the buyer, or any indirect selling expenses (defined as any expenses not already deducted from the U.S. selling price). 19 U.S.C. §§ 1677a(b)-(d). Delverde therefore believes that Commerce, prior to making the LOT comparison, should make both § 1677a(c) and § 1677a(d) adjustments. According to Delverde, Commerce would then view the LOT. of CEP as ex-factory. The fact that only de minimis sales without any selling expenses exist in the home market makes it self-evident to Del-verde that its CEP sales had no LOT counterparts, no LOT adjustment could have been quantified, and the LOT of its normal value sales was at a more advanced stage of distribution than that of its CEP sales.Thus, it ostensibly meets the three requirements for granting a CEP offset adjustment, pursuant to 19 U.S.C. § 1677b(a)(7)(B). Commerce allegedly erred in failing to recognize this, attributed selling expenses to Del-verde’s CEP, and engaged in an “apples to oranges” comparison, resulting in a positive, significant dumping margin.- Accordingly, Delverde maintains that Commerce’s failure to grant its CEP offset request contravened the “fairness” provision of the statute, 19 U.S.C. § 1677b(a), and the GATT, Art. 2.4, Agreement on Implementation of Article VI of the General Agreement on Tariffs and Trade, in Final Text of the GATT Uruguay Round Agreements (Apr. 15, 1994) [“Anti-dumping Code”]. 4. Borden’s Position Borden opposes defendant’s remand request, arguing that Commerce seeks a remand to perpetuate an unfair, skewed comparison between an adjusted CEP and an unadjusted normal value. Borden denies the import of the distinction Commerce draws between 19 U.S.C. § 1677a(c) and (d) deductions. Borden contends that most selling expenses excluded from CEP before the LOT comparison will subsequently be deducted from normal value. Thus, Borden argues that the comparison is faulty if made after deducting the expenses from one side but before the deduction from the other. Borden maintains that Delverde has failed to distinguish selling expenses from selling functions, that Delverde’s contention is incorrect that all CEP sales have an ex-factory LOT, even though all selling expenses are deducted from CEP during the price calculation. In Borden’s view, the inquiry does not ask whether the price to be compared “contains” selling functions but rather whether the sales are at the same or different levels of trade, as defined by the selling functions they embody. CEP, by definition, does not embody any selling expenses, but that implies nothing about the LOT of the underlying sales. Adjustments to the CEP starting price might render a price with no selling expenses, for example, but that may be an adjusted wholesale price or an adjusted retail price. The LOT inquiry thus concerns the LOT of the sales, not of the price. Thus, in Borden’s view, the LOT of the CEP sales does not exclude selling expenses, even though selling expenses are deducted in calculating CEP. Borden therefore contends that, while CEP itself excludes - selling expenses, in determining the propriety of LOT adjustments or CEP offsets, Commerce correctly examined the selling functions of Del-verde’s CEP sales. 5. Defendant’s Position After oral argument and rebriefing, Commerce reiterated its request for a remand to revise its LOT analysis by correcting its previous miscategorization of selling expenses deducted under 19 U.S.C. § 1677a(d) during the constructed export price calculation as having been deducted under subsection 1677a(c). Commerce would then reevaluate its LOT adjustment and CEP offset decisions in light of any remaining selling expenses. As indicated, Commerce bases its decision to adjust the CEP starting price in two statutory provisions: 19 ' U.S.C. § 1677b(a)(l)(B)(i), which instructs Commerce to calculate normal value “to the extent practicable, at the same level of trade as the ... constructed export price,” and 19 U.S.C. § 1677a(b), which defines “constructed export price” as the affiliate’s price in the United States “as adjusted under subsections (c) and (d).” Commerce finds further justification in the SAA for beginning the LOT inquiry with a partially adjusted CEP. SAA at 823. The SAA describes the purpose of the deductions made in the CEP price calculation as converting the affiliate’s price into, “as closely as possible, a price corresponding to an export price between non-affiliated exporters and importers.” Id. Defendant rejects Borden’s suggestion that Commerce compare levels of trade by evaluating the selling functions in both unadjusted CEP and unadjusted normal value sales. Defendant argues that to do so would mean that the CEP deductions, cannot create a difference in level of trade between the CEP sales and the originally comparable home market sales. And, without a difference in level of trade, there can be no level of trade deduction and no CEP offset from normal value to account for the disparity created by the CEP deductions. Def.’s Post-Hearing Reply, at 4. As defendant understands the inquiry, Commerce’s methodology, rather than creating an automatic adjustment, preserves the 'possibility of any adjustment at all. As interpreted by Borden, the new statutory framework ... precludes any adjustment to normal value that will compensate for the difference in level of trade created by the CEP deductions. Id. In other words, defendant asserts that deductions specific to CEP starting price intended to correct for the sale to an affiliate, because they are deductions of selling expenses, will change the level of trade of CEP. Moreover, if these are made after the LOT inquiry, there will not be a corresponding LOT adjustment to normal value to counterbalance that deduction. Commerce is concerned that the § 1677a(d) deductions will shift only one side of the balance, requiring a counterbalancing adjustment, and that the LOT adjustment is meant to address that, but cannot if the inquiry is completed before the § 1677a(d) deductions. Thus, to Commerce, failure to make the § 1677a(d) adjustments creates the distortion, not the reverse. 6. The Level of Trade Provision is a Conditional Mandate The methodology applied in this case, Final Determination, 61 Fed.Reg. at 30,339, and recently promulgated in its rules, Final Rule, 62 Fed.Reg. at 27,414, derives from Commerce’s interpretation of 19 U.S.C. § 1677b(a)(7). The Supreme Court in Chevron outlined a two-step inquiry to determine when a court should defer to an agency’s statutory interpretation. 467 U.S. at 842, 104 S.Ct. 2778. First the court determines whether the statute is silent or ambiguous. Id. at 842-43, 104 S.Ct. 2778. If the statute is clear, the court does not defer to the agency’s interpretation; only if the statute is silent or not clear as to the issue at bar does the court proceed to the second step and ask if the agency’s interpretation is reasonable. Id. at 843, 104 S.Ct. 2778. The court does not find statutory silence or ambiguity as to the fundamental matters at issue in this case. Whatever are Commerce’s concerns about a balanced final price comparison, Congress has defined how this particular adjustment will be made. The statute clearly provides for a conditional level of trade adjustment, instructing Commerce to make the adjustment to normal value if various conditions obtains. 19 U.S.C. § 1677b(a)(7). By contrast, the methodology employed by Commerce amounts to an unconditional adjustment in every CEP case. This is clear both in Commerce’s actual practice and in defendant’s explanation and defense of that practice, as set forth supra. Commerce’s argument is built both on a misreading of the statute and on the notion that normal value is ultimately unadjusted for “(d)” type expenses. That is, of course, incorrect. Normal value is adjusted for differences in circumstances of sales .(“COS”), covering various selling expenses. 19 U.S.C. § 1677b(a)(6)(C)(iii). Under former prac