Hebei Metals & Minerals Import & Export Corp v. United States

OPINION Introduction Restani, Chief Judge: Under 19 U.S.C. § 1677b(c)(l)(B) (2000), an antidumping duty imposed on a product exported from a nonmarket economy (“NME”) country is calculated using surrogate values from an appropriate market economy country or countries. Plaintiffs Hebei Metals & Minerals Import & Export Corporation and Hebei Wuxin Metals & Minerals Trading Co., Ltd. (referred to collectively hereinafter as “Hebei”) move for judgment on the agency record that the United States Department of Commerce (“Commerce”) improperly calculated the antidumping duty imposed on its lawn and garden steel fence posts from the People’s Republic of *1186 China (“PRC”), an NME country. Three aspects of Commerce’s duty calculation are at issue: (1) the use of an Indian import price rather than an Indian domestic price for the surrogate coal value; (2) the refusal to exclude as aberrational a Swedish import value from the surrogate value for steel pallet packing materials; and (3) the removal of internal consumption from raw material expenditures in the calculation of surrogate ratios for general expenses and profit. These decisions cannot be sustained. First, Commerce used the Indian import price for the surrogate coal value, but failed to provide substantial evidence demonstrating why imported coal yielded a more accurate surrogate value than domestic coal. Second, Commerce chose not to exclude the Swedish import value from the surrogate value for steel pallets, but the Government fails to provide a reasonable explanation why this uniquely high-priced/ low-volume import value was not aberrational. Third, Commerce removed internal raw material consumption from its surrogate ratio calculations, even though it lacked substantial evidence to demonstrate internal consumption’s significance and how its removal from the denominator would increase the accuracy of the ratios. Accordingly, the case is remanded for reconsideration and action consistent with this opinion. Background Commerce’s antidumping investigation determined that Hebei and other manufacturers of lawn and garden steel fence posts from the PRC had sold their products at less than fair value, based on a normal value calculation using surrogate values from India. Final Determination of Sales at Less Than Fair Value: Lawn and Garden Fence Posts from the People’s Republic of China, 68 Fed. Reg. 20,373 (Dep’t Commerce April 25, 2003) [hereinafter Notice of Final Deter mination]. Commerce explained its conclusions in Decision Memorandum for the Final Determination of the Antidumping Duty Investigation of Lawn and Garden Steel Fence Posts from the People’s Republic of China (Dep’t Commerce April 18, 2003), P.R. 158, Pis.’ App., Ex. 2 thereinafter Decision Memorandum or Decision Mem.]. At issue here are Commerce’s final decisions regarding the calculation of the surrogate coal value, the surrogate steel pallets value, and the surrogate ratios. These final decisions were the culmination of an antidumping duty investigation initiated by a petition from Steel City Corporation. Petition (May 1, 2002), P.R. Doc. 1, Def.’s App., Tab 3. I. The Antidumping Duty Investigation The period of investigation (“POI”) extended from October 1, 2001, through March 31, 2002. Notice of Initiation of Antidumping Duty *1187 Investigation: Lawn & Garden Steel Fence Posts From the People’s Republic of China, 67 Fed. Reg. 37,388, 37,389 (Dep’t Commerce May 29, 2002). The investigation sought to determine whether the subject merchandise was sold at less than fair value, based on a comparison between the export price and normal value of the merchandise. See id. at 37,390. Because Commerce considers the PRC to be a nonmarket economy country (“NME”), normal value was derived from factors of production as valued in India, a market economy country used as a surrogate for the PRC. Id. at 37,390. At the invitation of Commerce, Hebei provided surrogate value information for the various factors used in the production of the subject fence posts. Letter from Grunfeld Desiderio to Commerce (Sep. 18, 2002), P.R. Doc. 67, Def.’s App., Tab 7 [hereinafter Hebei First Surrogate Data Submission]. This included a surrogate coal value derived from Indian domestic prices for “steam coal” published in the Tata Energy Research Institute’s (“TERI”) Energy Data Directory & Yearbook for 2000/2001. Id., Ex. 9, at 44, Def.’s App., Tab 7. For steel pallets used as packing materials, Hebei provided the import prices for scrap steel (HTS 7204.29.09) published in the 2000-2001 Monthly Statistics of Foreign Trade of India Volume II (“MSFTI” or “Indian Import Statistics”). Id., Ex. 17, Def.’s App., Tab 7. Hebei also submitted a copy of the 2001 Annual Report of Surya Roshni Ltd., an Indian steel tube manufacturer, for use in calculating the surrogate ratios for selling, general and administrative expenses (SG&A), factory overhead, and profit. Id., Ex. 19, Def.’s App., Tab 7. The Surya Roshni Annual Report included a profit and loss statement (“Surya Roshni P&L Statement”) listing the income and expenditures for the year ended March 31, 2001. Id. II. The Preliminary Determination and Comment Period After reviewing the parties’ submissions, Commerce issued its preliminary determination on December 4, 2002. Preliminary Determination of Sales at Less Than Fair Value and Postponement of Final Determination: Lawn and Garden Steel Fence Posts from the People’s Republic of China, 67 Fed. Reg. 72,141 (Dep’t Commerce Dec. 4, 2002) [hereinafter Preliminary Determination], amended by Correction: Preliminary Determination of Sales at Less Than Fair Value and Postponement of Final Determination: Lawn and Garden Steel Fence Posts from the People’s Republic of China, 68 Fed. Reg. 8,737 (Dep’t Commerce Feb. 25, 2003) (correcting the scope of the investigation to correspond with the International Trade Commission’s preliminary determination). Commerce valued coal using import prices for an “others” basket of coal (HTS 2701.1909) as published in the 2001-2002 Indian Import Statistics. Memorandum Regarding Factors of Production Valuation for the Preliminary Results (Dep’t Commerce Nov. 27, 2002), at *1188 5-6 and Ex. Y, at 113-15, P. R. Doc. 104, Def.’s App., Tab 12 [hereinafter Preliminary FOP Mem.]. The surrogate value of steel pallets was determined based on the MSFTI data for steel bars (HTS 7213) and seamless tubes/pipes (HTS 7304.9000) during April 2001 through December 2001. Id., at 3 n.5 and Ex. U, Def.’s App., Tab 12. The overall average price for seamless steel tubes/pipes was 70 rupees per kilogram (“Rs/Kg”) based on imports from 25 countries, among which imports from Sweden had the highest price of 629 Rs/Kg. Id. In calculating the surrogate ratios for SG&A, factory overhead, and profit — where the formula calls for direct manufacturing expenses or material costs — Commerce used the figure from the “Raw Material Consumed” line-item in the Surya Roshni P&L Statement, which included costs attributed to internal consumption. Id., at 6 and Ex. Z, Def.’s App., Tab 12. In the subsequent comment period, Hebei challenged several aspects of the Preliminary Determination, including Commerce’s use of the Indian import price for imported coal. Brief from Grunfeld, Desiderio to Commerce (Mar. 13, 2003), at 9—11, P. R. Doc. 147 [hereinafter Hebei Case Br. to Commerce]. As for steel pallets, Hebei did not object to Commerce’s inclusion of imports from Sweden in the calculation of surrogate value, but submitted a new set of data for Indian imports of seamless tubes/pipes published in the World Trade Atlas, which was more contemporaneous with the POI than the data used in the Preliminary Determination. Letter from Grunfeld, Desiderio to Commerce (Jan. 21, 2003), at 2 and Ex. 12, P.R. Doc. 132, Def.’s App., Tab. 14. According to the new data, the average price for imports from 21 countries was 82 Rs/Kg, among which imports from Sweden had the highest value: 706 Rs/Kg. Id. Petitioner Steel City — in its comments regarding the calculation of surrogate ratios for SG&A, overhead, and profit — argued that Commerce should deduct from Surya’s raw material expenditures the amount shown for internal consumption on the Surya Roshni P&L Statement. Brief from Baker Hostetler to Commerce (Mar. 18, 2003), at 7-8, P. R. Doc. 150, Def.’s App., Tab 15 [hereinafter Pet’s Case Br. to Commerce]. III. The Final Determination and Hebets Ministerial Error Allegations In the Final Determination, Commerce continued to value coal using the Indian import prices. Decision Mem., at cmt. 4 at 10-11, Pis.’ App., Ex. 2. Commerce rejected the TERI domestic coal prices for steam coal on the grounds that (1) there was no record showing that “steam coal, which is suitable for use in boiler generating steam and most often used for electricity generation, was used in the production process;” and (2) Hebei “did not demonstrate the ‘useful heat value’ (UHV) of the coal used in the production.” Id. *1189 Commerce recalculated the steel pallets value using the new data submitted by Hebei because the new data was more contemporaneous with the POI than the data utilized in the Preliminary Determination. Id., at cmt. 3 at 9-10, Pls.’App., Ex. 2. Commerce “examined the contemporaneous data and found that the values were based on a significant volume of imports from various market economy countries, and did not appear aberrational.” Id. In evaluating the surrogate ratio calculations, Commerce recalculated the overhead and SG&A ratios by removing internal consumption from Surya’s raw material costs. Id., at cmt. 8 at 15-16, Pis.’ App., Ex. 2. Commerce reasoned that: “Internal consumption,” in so far as it represents the use of raw materials to produce internal assets rather than finished products for sale, should not be applied to the cost of goods sold. Only those materials consumed in the production of finished goods should be included in the cost of goods sold. Likewise, if the material costs were increased to include internal transfers between factories or cost centers, only the net material cost figure would avoid double-counting material costs in the denominator of the financial ratios. Id. In response to the Final Determination, Hebei first alleged that Commerce committed a ministerial error by including the Swedish import value in its valuation of steel pallets. Letter from Grunfeld, Desiderio Regarding Clerical Errors in the Final Determination (April 28, 2003), at 5-6, P. R. Doc. 171, Pls.’App., Tab 4. Specifically, Hebei argued that the steel tube imports from Sweden must be considered aberrational given Commerce’s other finding in the Final Determination that the Indian domestic price for powder coating submitted by Hebei as surrogate value was “aberrational” because it was 43% lower than the import price. Id. Hebei claimed that, if a 43% difference constitutes the standard for an “aberrational” price, the Swedish steel tube import value should be disregarded as aberrational because it is 1,134% greater than the average of other countries’ import values. Id. Commerce found no ministerial error, claiming that it intended to exclude from the calculation only those Indian imports sourced from NME countries and countries maintaining non-industry specific export subsidies, which might distort export prices. ITA Memo Re: Ministerial Error Allegations (May 12, 2003), at 4, P. R. Doc. 173, Pis.’ App., Ex. 5. Because the Swedish steel tube value did not fall into either category, Commerce found no reason to exclude it. Id. Hebei then initiated a challenge to the Final Determination before the court. Hebei filed the instant motion for judgment on the agency record on November 21, 2003. *1190 IV. Jurisdiction and Standard of Review The court has jurisdiction over this case pursuant to 28 U.S.C. § 1581(c) and 19 U.S.C. § 1516a(a)(2)(B)(i), but the Government— citing the exhaustion doctrine — challenges the court’s power to review Commerce’s inclusion of a Swedish import price in the calculation of the surrogate steel pallets value. This issue is discussed infra at Part III(A). Commerce’s antidumping duty calculation shall be sustained if it is supported by substantial evidence and is otherwise in accordance with law. See 19 U.S.C. § 1516a(b)(l)(B) (1988). Discussion I. The Statutory Framework For NME Antidumping Duty CalCULATIONS The antidumping duty represents the amount by which the “normal value” of the subject merchandise exceeds its “export price;” i.e., the price at which the merchandise was sold, or was threatened to be sold, in the United States. 19 U.S.C. § 1673. Where the exporting country has a nonmarket economy (“NME”) and where Commerce determines that the available information does not permit a standard normal value calculation, Commerce must determine normal value on the basis of surrogate values for “the factors of production utilized in producing the merchandise” plus “an amount for general expenses and profit plus the cost of containers, coverings, and other expenses.” 19 U.S.C. § 1677b(c)(l)(B). 1 The surrogate value of the factors of production “shall be based on the best available information regarding the values of such factors in a market economy country or countries considered to be appropriate by [Commerce].” Id. Hebei does not contest the designation of India as the surrogate economy, but argues that Commerce failed to use the best available information from the Indian surrogate data in its normal value calculations. Because the statute provides little guidance as to what constitutes the “best available information” (“BAI”), Commerce is accorded “wide discretion in the valuation of factors of production in the application of those guidelines.” Nation Ford Chem. Co. v. United States, 166 F.3d 1373, 1377 (Fed. Cir. 1999) (citing Lasko Metal Prods., Inc. v. United States, 43 F.3d 1442, 1446 (Fed. Cir. 1994)) (internal citations omitted). Thus, “Commerce need not prove that its methodology was the only way or even the best way to calculate surrogate values for factors of production as long as it was reasonable.” Shandong Huarong Gen. Corp. v. United States, 159 F. Supp. 2d 714, 721 (Ct. Int’l Trade 2001). *1191 Despite the broad latitude afforded Commerce, its discretion is not unlimited, but must be exercised “in a manner consistent with underlying objective of [the statute] — to obtain the most accurate dumping margins possible.” Shandong Huarong, 159 F. Supp. 2d at 719; see also Shakeproof Assembly Components, Div. of Ill. Tool Works, Inc. v. United States, 268 F.3d 1376, 1382 (Fed. Cir. 2001) (“In determining the valuation of the factors of production, the critical question is whether the methodology used by Commerce is based on the best available information and establishes antidumping margins as accurately as possible.”). To achieve the statutory purpose of accuracy, Commerce’s choice of what constitutes BAI must evidence a rational and reasonable relationship to the factor of production it represents. See Shandong Huarong, 159 F. Supp. 2d at 719. In the context of products from the PRC, the requirement of BAI entails use of “the price that results in the most accurate calculation of what the cost of production would be in the PRC if the PRC were a market-economy country.” Rhodia v. United States, 185 F. Supp. 2d 1343, 1353 (Ct. Int’l Trade 2001) (“Rhodia I"); see also Baoding Yude Chem. Indus. Co. v. United States, 170 F. Supp. 2d 1335, 1345 (Ct. Int’l Trade 2001). But see Tianjin Mach. Import & Export Corp. v. United States, 16 CIT 931, 938, 806 F. Supp. 1008, 1016 (1992). In pursuing the most accurate calculation, Commerce’s practice has been to prefer “surrogate price data which is: (1) an average non-export value; (2) representative of a range of prices within the POR if submitted by an interested party, or most contemporaneous with the POR; (3) product-specific; and (4) tax-exclusive.” Taiyuan Heavy Mach. Imp. & Exp. Corp. v. United States, 23 CIT 701, 706 (1999) (internal citations omitted). II. Commerce’s Use of an Import Price for the Surrogate Coal Value In calculating the surrogate value for coal, Commerce rejected the Indian domestic coal price from the TERI Energy Data Directory in favor of the Indian import values for an unspecified “others” basket category of coal products taken from the MSFTI. Decision Mem., at cmt. 4, Pls.’App., Ex. 2. Commerce gave two reasons why it rejected the TERI data. First, Hebei “did not put any information on the record to indicate specifically that steam coal, which is suitable for use in boiler generating steam and most often used for electricity generation, was used in the production process.” Decision Mem., at cmt. 4, Pis.’App., Ex. 2. Second, Hebei “did not demonstrate the ‘useful heat value’ (UHV) of the coal used in the production process.” Id. Instead of the TERI steam coal data, Commerce used import values for an “others” basket category of coal products taken from the MSFTI in making the Preliminary Determination and Final Determination. Id. This “others” category corresponds to HTSUS category 2701.1909, which encompasses “other coal;” i.e., coal that is not an *1192 thracite, bituminous, or bituminous-metallurgical. See Preliminary FOP Valuation Mem., at 5, Def.’s App., Tab 12. Hebei alleges three flaws in Commerce’s use of the Indian import price for the surrogate coal value: (1) Commerce misread the term “non-coking steam coal” and erred in discarding the domestic coal data published in TERI Energy Data Directory & Yearbook; (2) Commerce failed to explain how Indian imported coal price represented the cost incurred by Indian fence post manufacturers; and (3) Commerce failed to follow its established practice of using the domestic coal price. Hebei Op. Br. at 5-10. While Hebei observes correctly that Commerce failed to base its decision on substantial evidence, Hebei’s Indian domestic data appears to be an inadequate alternative basis for a surrogate coal value. A. The Inadequacy of Hebei’s Surrogate Coal Data Coal is used in the production of the subject fence posts to generate heat that aids in the drying of coating materials. Decision Mem., at cmt. 4, Pls.’App., Ex. 2, at 11. Commerce, conceiving of steam coal as that which is used for steam and electricity generation, did not find record evidence to indicate that steam coal was used in the production process. Id. According to Hebei, however, the meaning of “steam coal” is much broader, representing not a specific type of coal used to generate steam but rather all coal not used for metallurgical purposes. Pls.’ Op. Br., at 5-7. In advancing this argument, Hebei relies primarily on an attachment to its opening brief to this court. See Pis.’ Op. Br. at 6. 2 The Government correctly characterizes this as information outside the record that cannot form the basis for Commerce’s decision, see NEC Corp. v. United States, 151 F.3d 1361, 1373 (Fed. Cir. 1998), and the court declines to take judicial notice of the information contained therein. The Government, however, failed to address Hebei’s somewhat more persuasive secondary argument: that the broad scope of the term “steam coal” is evident from Hebei’s submissions to the record. See Pls.’ Op. Br. at 6-7. In the main text of the Hebei First Surrogate Data Submission, the brief discussion of coal refers initially to “steam coal” and then to “non-coking steam coal”: Steam Coal should be valued using data from the Teri Energy Data Directory & Yearbook for 2000/2001. The value is derived from price for non-coking steam coal as of April 20, 2000. These steam coal prices are based on grades for non-coking coal that *1193 are determined by coals UHV (“Useful Heat Value”), the UHV is measured by a range of kcal/kg. The average values for non-coking steam coal are as follows: GRADE A (UHV over 6200 kcal/kg.) 1109.26 RS/MT GRADE B (UHV 5600-6200 kcal/kg.) 1017.89 RS/MT GRADE C (UHV 4940-5600 kcal/kg.) 870.42 RS/MT GRADE D (UHV 4200-4940 kcal/kg.) 742.95 RS/MT Source documents for these surrogate values have been provided in Exhibit 9. Hebei First Surrogate Data Submission at 6, Pls.’ App., Ex. 1. Exhibit 9 to the Hebei First Surrogate Data Submission provides pages from the TERI Energy Data Directory & Yearbook for 2000/20001. Table 1.15 of the TERI Energy Data Directory makes no reference to “steam coal” but does refer to “non-coking coal,” which it defines as “coals other than coking, semi-coking, or weakly coking coals.” Id. at Ex. 9, p. 44 n.3, Pls.’ App., Ex. 1. Table 1.28 provides the selling prices of non-coking coal on April 20, 2000, with the prices broken down by producer, characteristics, grade, and classification. Id. at Ex. 9, p. 52-54, Pls.’App., Ex. 1. The table provides three classifications: “steam coal and rubble,” “slack coal and washery middlings,” and “run-of-mine coal.” The logical inference to be drawn from Table 1.28 is that steam coal is a type of non-coking coal, not that steam coal is synonymous with non-coking coal or comparable to non-metallurgical coal or a basket of “others” coal products. Indeed, nothing in Table 1.28 suggests that steam coal is a more appropriate surrogate data source than the other two categories of non-coking coal provided by the table. Without additional evidence, it is a matter of speculation whether steam coal is used in the production of the subject fence posts. Thus, Hebei failed to provide record evidence to show that its steam coal data pertained to a sufficiently broad category of coal. Hebei’s data is further flawed because it lacks contemporaneity with the POI. See Union Camp Corp. v. United States, 20 CIT 931, 939, 941 F. Supp. 108, 116 (1996) (noting that, where possible, Commerce will select a publicly available published value which is, among other things, “representative of a range of prices within the POI”). B. The Lack of Substantial Evidence for Commerce’s Use of Indian Import Values The shortcomings of Hebei’s data do not, however, resolve the issue of whether Commerce properly selected Indian coal import values over Indian coal domestic values. 3 This decision must still be *1194 supported by substantial evidence showing that the use of Indian import values is consistent with Commerce’s duty to calculate normal value as accurately as possible on the basis of the best information available. “The decision on which price to use-domestic or import-should be based on which value will result in a more accurate normal value.” Rhodia I, 185 F. Supp. 2d at 1352. Commerce’s use of import prices has been upheld on several occasions. See Nation Ford, 166 F.3d at 1378; however, Commerce here did not explain why an Indian manufacturer would pay for imported coal. Commerce defended its use of the Indian Import Statistics only on the grounds that the data was contemporaneous and “free of taxes and duties.” Decision Mem., at cmt. 4, Pls.’App., Ex. 2. This justification compares unfavorably with what was required in Yantai Oriental Juice Co. v. United States, Slip Op. 02-56 at 34 (Ct. Int’l Trade June 18, 2002). In Yantai, the court concluded that Commerce’s rejection of the Indian domestic coal price in favor of the Indian import coal price was not supported by substantial evidence because (1) there was no indication that the domestic Indian coal market was distorted, e.g. by a high tariff that inflated the domestic price; and (2) there was no indication that the use of imported coal values best approximated the cost encountered by Indian apple juice producers. Id. at 22-23. On remand, Commerce was ordered to either recalculate normal value using domestic coal data or provide an explanation of why the use of domestic coal data would not more accurately approximate the costs experience of Indian apple juice production. Id. at 24. In the instant case, Commerce similarly failed to identify a distortion in the Indian domestic coal market or explain how import coal values best approximate the cost incurred by Indian fence post production. Other cases have affirmed Commerce’s choice between import and *1195 domestic values where Commerce’s decision demonstrated a reasonable basis for the superior accuracy of one over the other. In Nation Ford, the Federal Circuit upheld Commerce’s use of an import price for aniline because the records showed that India protected its domestic industry with a high import tariff, which inflated the domestic price. Nation Ford, 166 F.3d at 1375-78 (Fed. Cir. 1999). The tariff, however, was not paid by Indian producers if they used the material to produce for export. Id. at 1376. Therefore,. Commerce’s use of the import price was justified upon findings that Indian producers who exported their product bought imported material instead of domestic material because it was less expensive. See id.; Rhodia I 185 F. Supp. 2d at 1351-52 (finding a similar rationale sufficient to support the use of import values for phenol). In Shangdong made in the instant case. 159 F. Supp. 2d at 722-24. The use of the Indian domestic price for aniline was sustained in Baoding Yude, where Commerce concluded that the reduction of the high tariff rate effectively removed the previous distortions in the domestic price and Indian manufacturer no longer depended on imported aniline. 170 F. Supp. 2d at 1342-44. Each of the above cases required Commerce to demonstrate that either the import value or the domestic value was more accurate than the other. Commerce’s observation in the instant case — that the Indian import coal value was free of taxes and duties — does not meet this standard because it does not address whether taxes and duties had a distortive effect on the Indian domestic coal market. Accordingly, Commerce’s selection of an Indian import value for coal was not based on substantial evidence. On remand, Commerce must either provide further explanation based on record evidence or conduct further investigations to determine whether Indian import or domestic data provides a value that more accurately reflects the coal consumption patterns of producers in the relevant industry. III. Commerce’s Use of a High-Price/Low-Volume Swedish Value in the Calculation of a Surrogate Value for Steel Pallets A. Exhaustion of Administrative Remedies According to the Government, the doctrine of exhaustion precludes judicial review of Commerce’s inclusion of a Swedish import value in the calculation of the surrogate value for steel pallets. Def.’s Br. at 26-29. As a general matter, “[t]he exhaustion doctrine requires a party to present its claims to the relevant administrative agency for the agency’s consideration before raising these claims to the Court.” Timken Co. v. United States, 201 F. Supp. 2d 1316, 1340 (Ct. Int’l Trade 2002). The doctrine furthers two main purposes: (1) allowing the administrative agency to perform the functions within its area of special competence; and (2) promoting judicial efficiency by affording the agency the opportunity to correct its mistakes so as to resolve *1196 the controversy without judicial intervention. See Parisi v. Davidson, 405 U.S. 34, 37 (1972); Sandvik Steel Co. v. United States, 164 F.3d 596, 600 (Fed. Cir. 1998). “There is, however, no absolute requirement of exhaustion in the Court of International Trade in non-classification cases.” China Steel Corp. v. United States, 306 F. Supp. 2d 1291, 1310 (Ct. Int’l Trade 2004) (quoting Consol. Bearings Co. v. United States, 25 CIT_, _, 166 F. Supp. 2d 580, 586 (2001) (citations omitted), rev’d on other grounds by 348 F.3d 997 (Fed. Cir. 2003)). Congress has directed that “the Court of International Trade shall, where appropriate, require the exhaustion of administrative remedies.” 28 U.S.C. § 2637(d). By its use of the phrase “where appropriate,” Congress granted “discretion to determine the circumstances under which it is appropriate to require the exhaustion of administrative remedies.” China Steel, 306 F. Supp. 2d at 1310; see also, Cemex, S.A. v. United States, 133 F.3d 897, 905 (Fed. Cir. 1998). Employing this discretion, the court has recognized certain exceptions to the requirement of exhaustion. One exception applies where the respondent did not have the opportunity to raise the relevant issue at the administrative level. Philipp Bros., Inc. v. United States, 10 CIT 76, 83-84, 630 F. Supp. 1317, 1324 (1986); see also Al Tech Specialty Steel Corp. v. United States, 11 CIT 372, 377, 661 F. Supp. 1206, 1210 (1987) (noting that, in determining whether a question is precluded from judicial review, “the Court will assess the practical ability of a party to have its arguments considered by the administrative body”). In Philipp Bros., the plaintiff was not afforded an opportunity to raise its objections at the administrative level because Commerce did not address the issue until the final determination. 10 CIT at 83-84, 630 F. Supp. at 1324. With regard to the methodology Commerce uses to resolve an issue, the exhaustion doctrine is inapplicable where a respondent did not have the opportunity to challenge the methodology because Commerce failed to articulate the methodology it would use until the final determination. See LTV Steel Co. v. United States, 21 CIT 838, 869, 985 F. Supp. 95, 120 (1997); see also SKF USA, Inc. v. U.S. Dep’t of Commerce, 15 CIT 152, 159 n.6, 762 F. Supp. 344, 350 n.6 (1991) (declining to apply the exhaustion doctrine where a respondent did not have a chance to contest Commerce’s recalculation of the foreign market value because the agency did not reveal the result of the recalculation until the final determination). Application of the exhaustion doctrine is inappropriate here because (1) while the elimination of the Swedish value at the time of the Preliminary Determination would have reduced the average steel pallet price by 11%, the elimination of the Swedish value at the time of the Final Determination would have reduced the surrogate steel *1197 pallet price by 24%, a significantly greater figure; 4 and (2) only with the Final Determination did Commerce offer a benchmark for aberrational values when it excluded a Indian domestic value for powder coating that was 43% lower than the Indian import value and 34% lower than the Indonesian import value. Decision Mem., at cmt. 4, Pls.’App., Ex. 2. Thus the potentially significant aberrational nature of the Swedish steel tube value only became apparent upon the issuance of the Final Determination, after the conclusion of the administrative investigation. 5 Application of the exhaustion doctrine here, while possible, would be overly technical and unfair to Hebei. Accordingly, review of this issue is appropriate. B. Commerce’s Unreasonable Decision to Include the Swedish Import Value Commerce’s inclusion of the Swedish import value is an unreasonable departure from the agency’s approach to aberrational data. “Consistent with the statutory requirement to use the best available information, Commerce must evaluate all data in the record to determine their reliability.” Shanghai Foreign Trade Enters. Co. v. United States, No. 03-00218, Slip Op. 04-33 at 22 (Ct. Int’l Trade April 9, 2004). Commerce’s general practice in this regard is “to value inputs using surrogate values derived from the import statistics of the surrogate country. Further, [Commerce] has excluded— where appropriate — aberrational data that appear to distort the overall value for a specific import category.” Issues and Decision Memorandum to Final Determinations of Sales at Less Than Fair Value: Steel Wire Rope from India and People’s Republic of China, at cmt. 1, 66 Fed. Reg. 12,759 (Dep’t Commerce Feb. 28, 2001) [hereinafter Steel Wire Rope from India and the PRC]. 1. The Aberrational Nature of the Swedish Value, Based on Price Variation Alone In the Final Determination, Commerce obtained a surrogate value for steel pallets using the prices for Indian imports of steel tubes from 21 countries between October 2001 and March 2002. See Factors of Production Valuation of the Final Determination (Dep’t Com *1198 merce April 18, 2003), at Ex. L, P.R. Doc. 159 at 49, Def.’s App., Tab 17 [hereinafter Final FOP Valuation Mem.]. In this period, the total quantity of steel tubes imported was 5,253,028 Kg, and the total import value was 435,494,000 Rs. Id. Ninety-seven percent of these steel tube imports originated in 15 countries, with prices ranging between 14 Rs/Kg and 105 Rs/Kg. Id. 6 Notably, over 60% of the total import quantity concentrated in three countries: Germany (1,697,422 Kg, 32% of total import quantity), Japan (926,149 Kg, 17.7%), and Argentina (583,347 Kg, 11%). Id. Their respective prices were Germany (52 Rs/Kg), Japan (69 Rs/Kg), and Argentina ( 49 Rs/ Kg). Id. In contrast, only 168,424 Kg were imported from Sweden and at much higher unit price: 706 Rs/Kg. Id. 7 The overall average import price for steel tubes in the Final Determination — including the Swedish value — was 82 Rs/Kg. Id. Exclusion of the Swedish value reveals its aberrational nature: it would lower the overall average to 62 Rs/Kg. Id. This large drop in the average price reflects the fact that the Swedish value is 1,134% higher than the average of import values from all other countries and, considering the relatively low volume of Swedish imports, had a singularly disproportionate impact on the overall average value. A 1,134% price variation appears aberrational on its face, and this variation is more striking when compared to a value for Indian domestic powder coating that Commerce treated as aberrational in this case. Commerce rejected this powder coating value on the ground that it was 43% lower than the Indian import price and 34% lower than the Indonesian import price. If such variations from the average price are a benchmark for aberrational values, the Swedish value is clearly aberrational. 2. The Aberrational Nature of the Swedish Value, Based on Price Variation and Low Import Volume If the high-price/low volume Swedish value is somehow not clearly aberrational under the general approach of discarding distortive values, any doubt is removed in light of Commerce’s more specific prac *1199 tice to “disregard small quantity import data when the per-unit value is substantially different from the per-unit values of larger quantity imports of that product from other countries.” Shakeproof Assembly, 23 CIT at 485, 59 F. Supp. 2d at 1359-60; see also Shanghai Foreign Trade Enters., Slip Op. at 26 n.5 (“At oral argument, [the Government’s] counsel mentioned one method of determining whether an Indian Import Statistics price is aberrational: when import statistics include imports from several countries, Commerce will compare the price from countries with small quantity imports against those with large quantity imports, and Commerce will discard small quantity import prices if they are aberrational.”); see also Issues and Decisions Memorandum to Notice of Final Determination of Sales at Less Than Fair Value: Ferrovanadium from the People’s Republic of China, at cmt. 13, 67 Fed. Reg. 71,137 (Dep’t Commerce Nov. 29, 2002) (excluding low-volume import values that were “substantially different” from the values of high-volume imports). While Commerce enjoys discretion to consider new arguments or facts, it “must either conform itself to its prior decisions or explain the reasons for its departure.” Citrosuco Paulista, S.A. v. United States, 12 CIT 1196, 1209, 704 F. Supp. 1075, 1088 (1988). In the instant case, Commerce failed to conform itself to its prior rational decisions. The price for Swedish steel tube imports, 1,134% greater than the average price from all other countries and representing a fraction of the quantity of total imports, increased the overall average value by 24%. 3. The Swedish Value Falls Far Beyond the Range of Variation for the Other Values The Government defends the inclusion of the Swedish values on the ground that it was a reasonable response to the variations among the range of Indian import prices. Def.’s Br. at 30-31. Commerce refers to the prices of Switzerland (490 Rs/Kg), Denmark (453 Rs/Kg), Austria (233 Rs/Kg), Nepal (14 Rs/Kg), Belgium (20 Rs/Kg) and the Czech Republic (28 Rs/Kg) to show that the data fluctuated so greatly that the deviation of Swedish price should not be considered aberrational. Id. Reference merely to these per kilogram prices fails, however, to acknowledge a critical dimension to the variations in the data: the low volume of the higher-priced imports. The data showed five countries in addition to Sweden with import prices at or above 200 Rs/Kg: Switzerland (490 Rs/Kg), Denmark (453 Rs/Kg), Taiwan (250 Rs/Kg), Austria (233 Rs/Kg), and Brazil (200 Rs/Kg). See Final FOP Valuation Mem., at Ex. L, Def.’s App., Tab 17. These prices are much closer to the aberrational end of the spectrum, given their relatively low import volumes — Switzerland (202 Kg), Denmark (130 Kg), Taiwan (100 Kg), Austria (3,190 Kg), and Brazil (20 Kg) — and the significantly greater import volumes for the low-priced imports from Nepal (92,570 Kg), Belgium (60,910 Kg), *1200 and the Czech Republic (90,080 Kg). Id. Commerce’s argument fails to provide a reasonable explanation why high-price/low-volume values should be included in this case, in deviation from Commerce’s past practice. 8 The irrationality of this approach is particularly stark with regard to the Swedish data, and this is the only data Hebei seeks to exclude. Even if the volume of the imports is ignored, the Swedish value still appears aberrational. The Swedish value is 8.5 times higher than the average import value of 83.02 Rs/Kg. See Pls.’ Reply Br. at 8 (citing Final FOP Valuation Mem., at Ex. L, Def.’s App., Tab 17). If the Swedish value is set aside, the remaining highest and lowest import values have a roughly equal variation from the average import value. Id. Switzerland, the highest remaining value, is approximately 5.9 times higher than the average, while Nepal, the lowest remaining value, is approximately 5.7 times lower than the average. Id. Thus, while some values do vary significantly from the average, only the Swedish value varies to a uniquely extreme degree. Commerce should have discarded the Swedish value in conformity with its established practice of excluding aberrational data that distort the overall value for a specific import category. See Steel Wire Rope from India and the PRC, 66 Fed. Reg. 12,759 (Dep’t Commerce Feb. 28, 2001) (stating that it is Commerce’s general practice in NME cases to “exclude [ ] — where appropriate — aberrational data that appear to distort the overall value for a specific import category.). On remand, Commerce shall exclude the Swedish import value from its steel pallet surrogate value calculations. IV. Commerce’s Calculation of Surrogate Ratios Hebei alleges that Commerce erred in its surrogate ratio calculations by removing from the ratios’ denominators the surrogate raw material costs for internal consumption. Commerce uses surrogate ratios to implement the provision in 19 U.S.C. § 1677b(c)(l)(B) which requires that the normal value for products of NMEs include amounts for “general expenses and profit” in addition to the cost of the surrogate FOP values. The amounts for general expenses and profit are typically obtained by applying the following surrogate ratios to the surrogate FOP values: selling, general and administrative expenses (“SG&A”), 9 factory (or manufacturing) overhead, 10 and *1201 profit. Shanghai Foreign Trade Enters., Slip Op. at 5. These three ratios derive from the financial statements of one or more surrogate companies that produce merchandise in the surrogate country that is identical or comparable to the subject merchandise. Id. The ratios are calculated and incorporated into the normal value calculation in the following manner: To calculate the SG&A ratio, the Commerce practice is to divide a surrogate company’s SG&A costs by its total cost of manufacturing. For the manufacturing overhead ratio, Commerce typically divides total manufacturing overhead expenses by total direct manufacturing expenses. Finally, to determine a surrogate ratio for profit, Commerce divides the before-tax profit by the sum of direct expenses, manufacturing overhead and SG&A expenses. These ratios are converted to percentages (“rates”) and multiplied by the surrogate values assigned by Commerce for the direct expenses, manufacturing overhead and SG&A expenses. Id. (citation omitted). A. Surrogate Ratios and Internal Raw Material Consumption Because direct manufacturing expenses are a component in the denominator of each ratio, each ratio requires data for raw material costs. To this end, the Preliminary Determination utilized the “Raw Material Consumed” line-item from the “EXPENDITURES” column in the Surya Roshni P&L Statement for the year ended March 31, 2001. Id., Pls.’App., Ex. 3. 11 Below this line-item is an indented line- *1202 item, or contra account, 12 that reads “Less: Internal Consumption,” followed by a line-item for raw material consumption net of internal consumption. Id. The Preliminary Determination used the first, gross raw material consumption line-item, not the net raw material consumption line-item. Petitioner Steel City challenged this decision, arguing that the gross figure improperly included “materials that are internally consumed.” Pet.’s Case Br. at 8, Def.’s App., Tab 15. Steel City explained why such materials should not be included in the calculation of surrogate ratios: [the “Raw Material Consumed”] line item overstates the total cost of materials that are entering the overall production process at Surya Roshni, in the same manner that the line item “Sales” overstates the total sales volume by indicating internal transfers. Surya Roshni recognizes this issue, and provides in the financial statement both the net sales and the net materials consumed. The Department should recognize that the factory overhead expenses and SG&A expenses are incurred as the result of the consolidated production process. As such, including materials internally consumed double-counts the value of internal transfers in the denominator, thereby understating the resulting percentage. The value of these internal transfers is shown both on the revenue side (Rs. 730,575,211) and on the cost side (Rs. 717,871,564) of the financial statement, so that the aggregate profit balances properly. In the Department’s calculation, the value of the internal transfers is included as part of the cost, without recognizing that these generate an offsetting income line item. To correct this error, the Department should either define the direct materials cost as the net cost (Rs. 3,879,219,666), or should reduce the total materials cost used in the current calculation by the revenue generated from the internal transfers (Rs. 730,575,211). Pet.’s Case Br. at 8, Def.’s App., Tab 15 (emphasis added). Nowhere in its briefing does Steel City provide authority to support its assertion as to the source of overhead and SG&A expenses or to the effect of including internal consumption. Commerce agreed with Steel City and reversed its position in the Final Determination. This entailed a recalculation of “SG&A and [factory overhead] surrogate ratios using Surya Roshni’s raw material cost, netting out internal transfers.” Decision Mem., at cmt. 8, Pls.’App., Ex. 2. *1203 Hebei challenges the removal of the internal consumption figure on the grounds that (1) it is based merely on the unwarranted assumption that material expense for internal consumption represents a material cost for the production of internal assets rather than a material cost for the production of goods; (2) there is insufficient evidence to support the conclusion that the inclusion of internal consumption would result in inaccurate ratio calculations; (3) if internal consumption is removed from the raw material costs in the denominator of the ratios, it should also be removed from the numerator; and (4) it conflicts with the prior practice of Commerce. Pis.’ Op. Br. at 11. The Government, in turn, argues that, because Hebei cites no record evidence to support its contentions, the Final Determination’s recalculation should be upheld as a reasonable decision using the best available information. See Def.’s Br. at 32. “Because [19 U.S.C. § 1677b(c)(1)] is ambiguous, we review Commerce’s interpretation to determine whether it is reasonable.” Rhodia, Inc. v. United States, 240 F. Supp. 2d 1247, 1252 (Ct. Int’l Trade 2002) (“Rhodia II’); see also Fuyao Glass Indus. Group Co., Ltd. v. United States, No. 02-00282, Slip Op. 03-169, at p. 33 (Ct. Int’l Trade Dec. 18, 2003). In evaluating the reasonableness of Commerce’s calculation methodology, the court remains mindful that Commerce’s general mandate is to calculate normal value as accurately as possible on the basis of the best available information available. This mandate allows Commerce to draw reasonable inferences from the record, Yantai, Slip Op. at 5 (quoting Daewoo Elecs. Ltd. v. United States, 6 F.3d 1511, 1520 (Fed. Cir. 1993)), but it is not a license to guess. China Nat’l Arts and Crafts Imp. and Exp. Corp. v. United States, 15 CIT 417, 424, 771 F. Supp. 407, 413 (1991) (“Guesswork is no substitute for substantial evidence in justifying decisions”). B. The Ambiguous Meaning of “Internal Consumption” To evaluate whether Commerce improperly removed internal consumption from Surya Roshni’s raw material costs, the meaning of internal consumption must first be examined. Commerce defined the internal consumption line-item as representing “materials consumed outside of the normal production process of the goods sold by a company.” Decision Mem., at 15, Pls.’App., Ex. 2. Commerce went on to assume that, within that broad definition, raw materials could be internally consumed in two ways: (1) through production of internal assets; i.e., such that internal consumption “represents the use of raw materials to produce internal assets rather than finished products for sale;” and (2) through intra-facility transfers; i.e., “if the material costs were increased to include internal transfers between factories or cost centers.” Id. Commerce’s rationale for removing internal consumption differs according to the source of the internal consumption. *1204 If internal consumption reflects the production of internal assets, then the figure should be removed because “[o]nly those materials consumed in the production of finished goods should be included in the cost of goods sold.” Id. If internal consumption reflects intra-facility transfers, Commerce believes that failure to remove the figure would result in “double-counting material costs in the denominator of the financial ratios,” id., though it did not explain how this double-counting would occur. Unfortunately, there is no direct record evidence for these interpretations aside from the line items in the Surya Roshni P&L Statement. The Surya Roshni P&L Statement does not indicate how much, if any, of the internal consumption figure is attributable to either type of transaction. Schedule 13 of Surya Roshni’s financial statements defines the raw material internal consumption contra account as “internal consumption of components.” Final FOP Valuation Mem., at Ex. P, Def.’s App., Doc. 17. The addition of “of components” does little to clarify whether such components were used to produce internal assets or were merely transferred among Surya Roshni’s business units. More helpful is an examination of both the sales and expenditure sections of the Surya Roshni P&L Statement, which provides a “Less: Internal Consumption” contra line item for both “expenditure - raw materials consumed” (Rs. 717,871,564) and “income - sales” (Rs. 730,575,211). The fact that internal consumption is listed as a contra line item suggests that, at least in some sense, Surya Roshni’s raw materials expenditures and sales income are more accurately reflected if internal consumption is removed. A gap exists, however, between this reasonable inference and the two explanations offered by Commerce to justify the removal of internal consumption from raw material costs. Nothing in the Surya Roshni P&L Statement — or anywhere else— supports Commerce’s first explanation; that raw materials were used to produce internal assets. On the other hand, the rough equivalence between the internal consumption figures for sales and expenditures lends some credence to Commerce’s second explanation; that the internal consumption figures represent transfers between units of the consolidated Surya Roshni organization. Such an inference fails to explain, however, the discrepancy between the consumption and sales figures. Furthermore, the speculative nature of the intra-facility transfer interpretation is suggested by the language of the Final Determination: “if the material costs were increased to include internal transfers between factories or cost centers, only the net material cost figure would avoid double-counting material costs in the denominator of the financial ratios.” Decision *1205 Mem.., at cmt. 8, Pls.’ App., Ex. 2 (emphasis added). 13 Surprisingly, the internal-transfer interpretation is somewhat similar to that provided by Hebei: “the most reasonable conclusion is that internal consumption represents goods produced that were then sold as intra-company sales.” Pis.’ Op. Br. at 13. The difference is that, by characterizing the internal transfers as “intra-company sales,” Hebei asserts that Surya Roshni incurred the same costs and expenses as would be incurred from arms-length external sales. Id. (“Regardless of whether product is sold to an unrelated customer or sold in an intra-company transfer, the costs and expenses associated with the product remain the same.”). Hebei not only fails to provide record evidence for this claim but also fails to consider the possibility that some or all of the internal transfers may be little more than artificial transactions recorded for accounting purposes. Commerce is allowed to make reasonable inferences, but it does not cite any record evidence that makes the interpretation of the internal consumption figure more than a speculative enterprise. See Kerr-McGee Chem. Corp. v. United States, 21 CIT 1353, 1361, 985 F. Supp. 1166, 1173 (1997) (“For purposes of judicial review, the evidence before this Court is limited to the evidence contained in the administrative record.” (citations omitted)). As shown above, the parties have invoked contradictory — yet supposedly elementary — accounting principles without providing citations. Even if the court were to find that raw material internal consumption represents mainly intra-company transfers, further review of this issue finds only additional unfounded propositions. C. Commerce’s Unsupported Explanation for Removing Internal Consumption If internal raw material consumption resulted from intra-company transfers, Commerce explained that such data should be excluded because “only the net material cost figure would avoid double-counting material costs in the denominator of the financial ratios.” Decision Mem., at cmt. 8, Pls.’App., Ex. 2. Unfortunately, Commerce did not elaborate as to what constitutes double-counting or the precise manner in which double-counting would distort the surrogate ratios. Perhaps double-counting occurs where, for example, one unit of a consolidated entity purchases $1,000 of raw materials from an external source and later transfers those raw materials to another *1206 unit inside the consolidated entity. If the raw material expenditure account of the consolidated entity reflects a $1,000 expenditure for the external transaction and a $1,000 expenditure for the internal transaction, then the account gives a misleading impression as to how much the consolidated entity is really spending on raw materials. That is, the consolidated entity only paid $1,000 for the raw materials, but the account shows $2,000 worth of expenditures for those materials. Such a conception of double-counting is not, as Hebei claims, “an extremely bizarre assumption,” Pis.’ Reply Br. at 13, yet it finds no affirmative support in the record. It remains speculative and is not “render [ed] evident” by Commerce’s use of the phrase “double-counting” in the Decision Memorandum. See China Nat’l Machinery Imp. & Exp. Corp., 264 F. Supp. 2d at 1242 (quoting H.R. Conf. Rep. No. 103—826(I) at 98, reprinted in 1995 U.S.C.C.A.N. 3773). Assuming internal consumption represents intra-company transfers, and assuming as well the validity of the above example, then Commerce expressed a valid concern that the inclusion of internal consumption would overvalue raw material costs in the surrogate ratios. These assumptions, however, reflect the dearth of record evidence on this issue. If that were not reason enough to reject Commerce’s approach, Commerce also failed to follow the intra-facility transfer rationale in a consistent manner. D. Commerce’s Failure to Consider the Effects of Internal Consumption on SG&A and Factory Overhead Expenses While Commerce’s decision purports to purge the surrogate ratios of the distortive effects of internal consumption, Commerce failed to complete this task. Internal consumption was removed only from Surya Roshni’s raw material costs in the denominator of the surrogate ratios. Commerce did not consider the possibility that internal transfers also generate SG&A and factory overhead expenses that would be reflected in the numerator of the SG&A and factory overhead ratios, a possibility raised by Commerce’s own hypothetical conception of Surya Roshni as a consolidated business entity conducting significant internal transfers. Administrative and judicial precedents underscore the importance of addressing the possibility that internal transfers generate SG&A and factory overhead expenses, but the problem in this case is that there is absolutely no basis for determining what amount of these expenses is attributable to internal consumption. As a result, any attempt to remove internal consumption from the numerator of the ratios would involve the same guesswork as the adjustment to the denominator. In the Issues and Decision Memorandum to the 2000-2001 Administrative Review of Stainless Steel Sheet and Strip in Coils from Mexico, 68 Fed. Reg. 6,889 (Dep’t Commerce Feb. 11, 2003), Commerce had the benefit of financial statements that separated sales to *1207 “Parent and affiliates” from sales to “Third parties,” which supported directly the intra-facility transfer rationale for excluding internal transfers. 68 Fed. Reg. at 6,891. On the basis of this information and a citation to the Financial Accounting Standard Board’s statement of a relevant accounting principle, Commerce agreed with petitioners that a subsidiary’s sales of raw materials to its parent should be removed from the denominator of the ratio for indirect selling expenses because “[the subsidiary’s] sales of raw materials to its parent can be construed as an intracompany transfer of merchandise, as they involve only a routine transfer of merchandise.” 68 Fed. Reg. at 6,891. Commerce then addressed the likelihood that some indirect expenses were generated by affiliated transfers: “while we deem it inappropriate to assign an equal amount of indirect selling expenses to the affiliated transfers of raw materials as compared to sales of finished merchandise, we do consider it appropriate to attribute some expenses to these transfers. Thus, we also have reduced the numerator of the indirect selling expense ratio by an amount attributable to the expenses incurred by [the subsidiary] in selling these raw materials to [the parent].” Id. In the instant case, however, the Surya Roshni financial information does not appear to provide any basis for calculating the amount of internal raw material costs attributable to SG&A and factory overhead. A similar lack of evidence confronted the parties in Fuyao Glass Indus. Group, Slip Op. at 40. In that case, the court identified the need to exclude from the numerator “any amount of selling and administrative costs related to [traded goods]” if traded goods were excluded from the denominator. Id. The court also noted the evidentiary problem: “both Commerce and Fuyao acknowledge that there is insufficient evidence to determine where expenses associated with the purchase of traded goods are accounted for in St. Gobain’s financial statement.” The court responded with the following solution: “On remand, Commerce shall correct the calculation of the SG&A ratio by either (1) eliminating expenses relating to the purchase of traded goods from the numerator, (2) including costs relating to the purchase of traded goods in the denominator, or (3) developing some other reasonable method for taking traded goods into account.” Id. Such an approach is instructive for the instant case. As shown above, Commerce’s exclusion of internal raw material consumption is predicated on a series of conjectures, and “[conjectures are not facts and cannot constitute substantial evidence.” China Nat’l Mach. Imp. & Exp. Corp., 264 F. Supp. 2d at 1240 (citing China Nat’l Arts and Crafts Imp. and Exp. Corp., 15 CIT at 424, 771 F. Supp. at 413). Despite the challenges inherent in constructing normal value for a product from a NME country, Commerce’s method for recalculating the surrogate ratios “falls outside the limits of permissible approximation.” See Sigma Corp. v. United States, 117 F.3d 1401, 1407-08 (Fed. Cir. 1997). *1208 Accordingly, this issue is remanded for further explanation and, if necessary, further investigation. See China Nat’l Mach. Imp. & Exp. Corp., 264 F. Supp. 2d at 1243 (remanding the case to Commerce “to review and augment the administrative record and to explain its determinations adequately”). If Commerce is able to explain adequately the rationale for removing internal raw material consumption from the denominator of the surrogate ratios, then Commerce shall: (1) determine to what extent, if any, SG&A and factory overhead expenses are attributable to internal raw material consumption; and (2) remove appropriate amounts from the numerators of the SG&A and factory overhead surrogate ratios. If Commerce is unable to obtain sufficient evidence for this task, Commerce shall: (a) include internal raw material consumption in the denominator of the SG&A, factory overhead, and profit surrogate ratios; or (b) provide a rational explanation why more accurate surrogate ratios result from the removal of internal raw material consumption from the ratios’ denominators only. Conclusion Commerce lacked substantial evidence for its surrogate coal value and surrogate ratio calculations. Commerce’s decision to include the aberrational Swedish value in its surrogate steel pallet calculation was unreasonable and, therefore, not in accordance with the law. Accordingly, Hebei’s motion for judgment on the agency record is granted in part. The case is remanded for reconsideration and action consistent with this opinion. IT IS SO ORDERED. Footnotes [1] The factors of production utililized in producing the merchandise include, but are not limited to, labor hours, raw materials, energy and other utilities, and representative capital cost, including depreciation. 19 U.S.C. § 1677b(e)(3). [2] The attachment contains a printout from the website glossary of the U.S. Department of Energy’s Energy Information Agency (“EIA”), wherein the EIA defines “steam coal” as “all non-metallurgical coal” and “metallurgical coal” as “coking coal and pulverized coal consumed in making steel.” United States Energy Information Administration: Energy Glossary, available at http://www.eia.doe.gov/glossary/glossary_main_page.htm, Pis.’ Op. Br., Attach. 1. [3] Even where a party opposing Commerce’s position has submitted information that ultimately proves inadequate, Commerce is not relieved of the requirement that it support its *1194 antidumping duty calculation with substantial evidence. See 19 U.S.C. § 1516a(b)(l)(B). Section § 1677m(d) of title 19 demonstrates that a flawed data submission does not automatically support the opposing position, as it provides for additional measures where a party submits deficient information: “[Commerce] shall promptly inform the person submitting the response of the nature of the deficiency and shall, to the extent practicable, provide that person with an opportunity to remedy or explain the deficiency in light of the time limits established for the completion of investigations or reviews under this title.” 19 U.S.C. § 1677m(d). Hebei provided surrogate coal value information, the falws of which were only articulated upon the issuance of the Final Determination. At the conclusion of the investigation’s preliminary stage, Commerce stated only that it used the Indian Import Statistics to obtain a surrogate value for coal. Preliminary Determination, 67 Fed. Reg. at 72,145; Preliminary FOP Valuation Mem., at 5-6, Def.’s App., Tab 12. Hebei had little basis from which to understand how to remedy the deficiencies in its surrogate coal data before the conclusion of the investigation, as well as little reason to seek out information in addition to the TERI data. Thus, the emergence of deficiencies in Hebei’s TERI domestic data does not necessarily lead to the conclusion that Commerce’s MSFTI import data is the best available information. [4] In the Preliminary Determination, Commerce calculated the steel pallets value based on imports from 25 countries during the non-contemporaneous period of April 2001 through December 2001. Preliminary FOP Valuation Mem., Ex. U, Def.’s App., Tab 12. In the Final Determination, however, Commerce replaced this data with a set of new data based on imports from 21 countries during the period of October 2001 through March 2002. Factors of Production Valuation of the Final Determination (Dep’t Commerce April 18, 2003), Ex. L, P. R. Doc. 159 at 49, Def.’s App., Tab 17. As a result, the Swedish price increased from 629 Rs/Kg to 706 Rs/Kg; and the overall average price increased from 70 Rs/Kg to 82 Rs/Kg. Id. [5] Hebei challenged the inclusion of the Swedish value as a ministerial error, but this was rejected by Commerce. ITA Memo Re: Ministerial Error Allegations (May 12, 2003), at 4, P. R. Doc. 173, Pls.’App., Ex. 5. [6] The remaining countries are Nepal (14 Rs/Kg; 92,570 Kg), Belgium (20 Rs/Kg; 60,910 Kg), Czech Republic (27 Rs/Kg; 98,080 Kg), UAE (34 Rs/Kg; 113,545 Kg), France (44 Rs/Kg; 260,078 Kg), Argentina (49 Rs/Kg; 583,347 Kg), Germany (52 Rs/Kg; 1,697,422 Kg), Canada (60 Rs/Kg; 96,770 Kg), Japan (69 Rs/Kg; 926,149 Kg), Italy (85 Rs/Kg; 196,437 Kg), Spain (87 Rs/Kg; 76,547 Kg), Netherlands (88 Rs/Kg; 57,512 Kg), United Kingdom (89 Rs/Kg; 377,934 Kg), Singapore (101 Rs/Kg; 194,837 Kg), and the United States (105 Rs/Kg; 220,729 Kg). See Final FOP Valuation Mem.., at Ex. L, Def.’s App., Tab 17 (the first number in the above parentheticals is the individual country’s unit price; the second number is the total import quantity from that country). [7] In addition to Sweden, imports from five countries were valued over 200 Rs/Kg: Switzerland (490 Rs/Kg), Denmark (453 Rs/Kg), Taiwan (250 Rs/Kg), Austria (233 Rs/Kg) and Brazil (200 Rs/Kg). Final FOP Valuation Mem., at Ex. L, Def.’s App., Tab 17. The import quantities from these countries were relatively small: Switzerland - 202 Kg, Denmark - 130 Kg, Taiwan -100 Kg, Austria - 3190 Kg, Brazil - 20 Kg. Id. [8] If the high-cost/low-volume imports were removed from the record, the prices of all remaining countries would fluctuate between 14 Rs/Kg and 105 Rs/Kg. See Final FOP Valuation Mem., at Ex. L, Def.’s App., Tab 17. Moreover, more than 60% of the total import quantity concentrated in a price range between 49 Rs/Kg and 69 Rs/Kg. Id. [9] SG&A reflects the general expenses related to the cost of manufacturing and includes labor, materials, factory overhead, and energy costs. Fuyao Glass Industry Group Co., Ltd. v. United States, No. 02-00282, Slip Op. 03-169, at 38 n.26 (Ct. Int’l Trade Dec. 18, 2003). [10] “As factory overhead is composed of many different elements, the cost for individual *1201 items may depend largely on the accounting method used by the particular factory.” Magnesium Corp. of Am. v. United States, 166 F.3d 1364, 1372 (Fed. Cir. 1999). “The value of factory overhead is calculated as a percentage of manufacturing costs. Commerce calculates a ratio of overhead to material, labor and energy inputs (“MLE”) for producers of comparable merchandise in the surrogate country, India, and then applies this ratio to the NME producer’s MLE.” Rhodia I, 185 F.Supp. 2d at 1346. [11] The Preliminary Determination explained the selection of Surya Roshni as the surrogate for purposes of ratio calculations: To value factory overhead, selling, general and administrative expenses (SG&A) and profit, we used the audited financial statements for the year ended March 31, 2001, from an Indian producer of circular welded steel pipe, Surya Roshni (Surya). See FOP Memo for the calculation of these ratios from Surya’s financial statements. As noted above, section 773(c)(4) of the Act requires that the Department value the NME producer’s factors of production, to the extent possible, based on the prices or costs of factors of production in one or more market economy countries that are significant producers of comparable merchandise. The Department was unable to locate publicly available financial statements for an Indian fence post producer, and therefore, we looked for a producer of comparable merchandise. The production of fence posts and circular welded steel pipe have similar production processes and material inputs, in that the production of these products use steel sheets or strips in coil form as the major input, and the respective products inceptively use the process of roll forming to create the desired shape of the steel. 67 Fed. Reg. 72,141, 72,145 (citations omitted). [12] The court takes judicial notice that a contra account is one that is subtracted from some other account. See Robert N. Anthony & James S. Reece, Accounting Principles 87 (1995). [13] The Government claims that Commerce’s approach is consistent with its past practice. As evidence of past practice, the Government cites only Notice of Preliminary Determination of Sales at Not Less Than Fair Value and Postponement of Final Determination: Structural Steel Beams from Spain, 66 Fed. Reg. 67, 207, 67, 209 (Dec. 28, 2001). In this preliminary determination, Commerce excluded from the home market sales database sales between the mills of a company because these sales were made for internal consumption. Id. Because this decision does not pertain to the calculation of surrogate ratios, it is inappo-site.