Opinion
Pogue, Judge:
This case is before the Court on a motion for judgment upon the agency record pursuant to USCIT R. 56.2. Plaintiffs, Magnesium Corporation of America, International Union of Operating Engineers, Local 564, and United Steelworkers of America, Local 8319 (“Plaintiffs”) bring this action under section 516A of the Tariff Act of 1930 for review of the final affirmative determination of the International Trade Administration of the United States Department of Commerce (“Commerce”) that imports of pure magnesium from Russia are sold at less than fair value (“LTFV”). Notice of Final Determinations of Sales at Less Than Fair Value: Pure Magnesium and Alloy Magnesium From the Russian Federation, 60 Fed. Reg. 16,440 (March 30, 1995) (“Final Determination”).
Background
On March 31, 1994, plaintiffs filed an antidumping petition alleging material injury by reason of LTFV imports of pure and alloy magnesium [1093]*1093from China, Russia and Ukraine.1 Thereafter, Commerce initiated anti-dumping duty investigations. In June 1994 Commerce sent the anti-dumping questionnaire (“questionnaire”) to Berezniki Titanium-Magnesium Works (“Avisma”) and Solikamsk Magnesium Works (“Soli-kamsk”) in Russia. Commerce subsequently requested information from 56 potential exporters of Russian magnesium.
On October 27, 1994, Commerce issued preliminary determination that imports of magnesium from Russia were being sold at less than fair value2 within the meaning of section 733(b) of the Tariff Act of 1930, as amended, 19 U.S.C. § 1673b(b) (1988).3
On March 30, 1995, Commerce published its final determination of LTFV sales for imports of magnesium from Russia.4 Commerce reaffirmed its conclusion in the preliminary determination that Russia is a nonmarket economy country (“NME”). Commerce determined that Avisma did not make direct exports of magnesium to the United States during the period of investigation (“POI”), and that Solikamsk did make direct exports and qualified for a separate rate, which Commerce set at zero. Certain reseller/exporters, i.e. AIOC Corporation (“AIOC”), Gerald Metals, Inc. (“Gerald Metals”), Greenwich Metals (“Greenwich”), Hunter Douglas Metals (“HDM”), Hochschild Partners (“Hochschild”), Interlink Metals and Chemicals, S.A. (“Interlink”), MG Metals (“MG”), and Razno-Alloys, Ltd. (“Razno”), received zero or de minimis dumping margins. Commerce also established a 100.25 percent “All Others” rate based on best information available (“BIA”), which applied to all exporters not assigned an individual rate. This rate would also apply to reseller/exporters which received an individual rate, if they were to sell magnesium produced by a Russian producer different from the producer from which they exported magnesium to the United States during the POI.
Commerce issued its Antidumping Duty Order, together with an Amended Final Determination on May 8, 1995.5
In the related proceeding before the International Trade Commission (“Commission”), the Commission determined that the domestic indus[1094]*1094try was materially injured by reason of imports of pure magnesum from China, Russia and Ukraine.6
This action presents the following issues:
1. Whether Commerce’s use of Brazilian raw dolomite to establish a surrogate value for concentrated carnallite rather than for raw carnallite is supported by substantial evidence, and otherwise in accordance with law?
2. (a) Whether Commerce’s calculation of the surrogate value for electricity used in the production of magnesium is supported by substantial evidence, and otherwise in accordance with law?
(b) Whether Commerce violated plantiffs’ due process rights when it relied on an outside expert’s consultation which plaintiffs could not review?
3. Whether Commerce’s determination to value factory overhead costs using a Brazilian silicomanganese producer’s factory overhead as a surrogate value, and without including an adjustment reflecting the electrolytic cell rebuild costs incurred by petitioner MagCorp, is based on substantial evidence, and otherwise in accordance with law?
4. Whether Commerce properly used a Brazilian surrogate value to calculate the Russian producers’ selling, general and administrative expenses (“SG&A”)?
5. Whether Commerce’s calculation of by-product credits to the Russian producers’ cost of manufacturing magnesium without reducing the value of the by-product credits by the after separation processing costs is supported by substantial evidence, and otherwise in accordance with law?
6. Whether Commerce’s reliance on sales information reported by Razno, Interlink and AIOC immediately prior to verification is consistent with existing Commerce precedent, supported by substantial evidence on the record, and otherwise in accordance with law?
7. Whether Commerce’s exclusion of resellers’ SG&A from the calculation of foreign market value (“FMV”) is consistent with existing Commerce precedent, supported by substantial evidence, and otherwise in accordance with law?
8. (a) Whether Commerce’s refusal to deduct from USP the export taxes paid by Russian producers to the Russian government is in accordance with law?
(b) Whether the exchange rate balancing requirement, to which Russian producers were subject, represents an implied export tax that should be deducted from USP?
Standard of Review
The Court will uphold a Commission determination in an antidump-ing investigation 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)(i) (1994). Substantial evidence is “such relevant evidence as a reasonable mind might accept as adequate to support a con[1095]*1095clusion.” Consolidated Edison Co. v. NLRB, 305 U.S. 197, 229 (1938); Universal Camera Corp. v. NLRB, 340 U.S. 474, 477 (1951); Matsushita Elec. Indus. Co., Ltd. v. United States, 750 F.2d 927, 933 (Fed. Cir. 1984). The court may not substitute its judgment for that of the agency. See Matsushita, 750 F.2d at 936. “[T]he possibility of drawing two inconsistent conclusions from the evidence does not prevent an administrative agency’s finding from being supported by substantial evidence.” Consolo v. Federal Maritime Comm’n, 383 U.S. 607, 620 (1966) (citations omitted).
When Commerce’s interpretation of the antidumping statute is challenged, this court applies the two step analysis set forth in Chevron, U.S.A., Inc. v. Natural Resources Defense Council, Inc., 467 U.S. 837, 842-43 (1984): Using the traditional tools of statutory construction the court ascertains whether congressional intent on the disputed issue is clear, and, if clear, the court applies the statute in the manner Congress intended, regardless of the agency’s position.7 If the statute is ambiguous, the court, rather than interpreting the statute anew and rendering its own interpretation, must defer to an administrative agency’s “permissible construction of the statute,”8 whether that construction manifests itself in the application of the statute, see, e.g., Daewoo Electronics Co., Ltd. v. International Union of Electronic Elec., Technical, Salaried and Mach. Workers, 6 F.3d 1511, 1516 (Fed. Cir. 1993), cert. denied, 114 S.Ct. 2672 (1994), or in the promulgation of a regulation, see, e.g., Smith Corona Group v. United States, 1 Fed. Cir. (T) 130, 136, 713 F.2d 1568, 1575 (1983), cert. denied, 465 U.S. 102 (1984). This deference is at its greatest level on issues involving Commerce’s technical expertise in interpreting the antidumping statute it administers. Fujitsu General Ltd. v. United States, No. 95-1343 at 8 (Fed. Cir. July 3, 1996).
Discussion
This case involves the review of Commerce’s final determination in the investigation of exports of pure magnesium from the Russian Federation (“Russia”) to the United States.9 Commerce determined that Russia is a nonmarket economy (“NME”).10 The prices of the goods produced in an NME are subject to discrepancies which distort their value. Consequently, the antidumping statute provides that, in such cases, Commerce calculates the foreign market value (“FMV”) of the merchandise “on the basis of the value of the factors of production utilized in [1096]*1096producing the merchandise * * *.” 19 U.S.C. § 1677b(c) (1988).11 Commerce utilizes, “to the extent possible, the prices or costs of factors of production in one or more market economy countries that are — (A) at a level of economic development comparable to that of the nonmarket economy country, and (B) significant producers of comparable merchandise.” 19 U.S.C. § 1677b(c)(4).
In their petition, plaintiffs proposed Brazil as the appropriate surrogate market economy for Russia.12 Commerce selected Brazil because it “is the only country that produces magnesium and is comparable in terms of per capita GNP to Russia.”13 To determine the FMV of magnesium, Commerce calculated surrogate values for the factor-of-production inputs (including raw materials, labor, and energy) using the Brazilian magnesium industry. Commerce then multiplied these values by the quantities reported by the Russian producers and exporters.14 A factory overhead figure was also included in the FMV calculation, and certain by-product offsets were granted against the cost of manufacturing. Finally, Commerce added an amount for general expenses and profit, the cost of containers and coverings, and other expenses incidental to the shipment of the merchandise to the United States.15
1. Carnallite v. Dolomite:
Avisma and Solikamsk use concentrated carnallite as the basic feedstock in the production of magnesium.16 In their antidumping petition, plaintiffs were unable to determine a value for carnallite in Brazil, because the Brazilian magnesium industry utilizes dolomite instead of carnallite.17 Plaintiff therefore proposed a price of raw dolomite as a good surrogate value for raw carnallite. In its final determination, Commerce used the price of raw dolomite in Brazil18 as the surrogate value for concentrated carnallite, because it had no surrogate value information for concentrated carnallite, and because concentrated carnallite and raw dolomite have a comparable content of magnesium.19
Plaintiffs criticize Commerce’s use of a raw material, raw dolomite, to value a processed material, concentrated carnallite. Plaintiffs do not challenge the use of raw dolomite as a surrogate value. However, plaintiffs argue that raw dolomite is similar to raw carnallite, whereas the value relevant to the factor-of-production calculation is concentrated [1097]*1097carnallite. Therefore, plaintiffs contend, Commerce should have adjusted the value of raw dolomite by adding the processing costs incurred to convert raw carnallite into concentrated carnallite.20 Plaintiffs claim that Commerce has usually allowed adjustments to surrogate country values when necessary to reach an accurate determination.21
Despite plaintiffs’ claim, the administrative record does not indicate a surrogate value for concentrated carnallite, because this material is not used in Brazil, the surrogate country. Commerce’s use of raw dolomite was reasonable because its magnesium content is comparable to that of concentrated carnallite. The statute permits Commerce the use of “comparable merchandise” for valuing a raw material input as a basis for foreign market value.22 Moreover, it is reasonable to conclude that the cost associated with processing raw dolomite into a suitable feedstock is included in factory overhead for the Brazilian industry.23 Commerce’s determination is therefore supported by substantial evidence on the record, and otherwise in accordance with law.
Plaintiffs argue at length that they placed evidence on the record describing the Soviet method for producing artificial carnallite from raw carnallite.24 Based upon this information, plaintiffs claim that they
were able to estimate that the surrogate value for concentrated car-nallite was $15.32 per MT for Avisma and $17.84 per MT for [Soli-kamsk], The reasonableness of these calculations is shown by a commercial proposal by Silvinit, a Russian metals trade from whom [1098]*1098[Solikamsk] purchased concentrated carnallite * * * [which] states * * * an export price of 22.59 per MT.25
The Court recognizes that in applying the factors of production methodology to a nonmarket economy country Commerce has discretion to take a combined approach and to consider actual costs paid by the NME producer for each factor of production, if Commerce finds that such actual costs represent the best information available. Lasko Metal Products, Inc. v. United States, 43 F.3d 1442, 1445-46 (Fed. Cir. 1994). In the case of the Russian magnesium producers, the combined approach would allow Commerce to use the market cost of concentrated carnal-lite — if available and if verified as accurate. Commerce’s regulations provide that, if such or similar merchandise is not produced in the surrogate country, Commerce “may calculate the foreign market value using constructed value based on factors of production incurred in the home market country * * * if the Secretary obtains and verifies such information from the producer of the merchandise in the home market country. ” 19 CFR § 353.52(c) (1995).
The producers of concentrated carnallite were not under investigation here because Russian magnesium producers purchase their concentrated carnallite from unrelated Russian suppliers.26 Therefore, neither calculation nor verification was at issue in the proceeding.
Plaintiffs are not claiming that Commerce should account for the actual cost paid by the Russian producers to purchase concentrated car-nallite. Rather, plaintiffs seek to calculate the value for concentrated carnallite partly based on the surrogate value of raw dolomite, and partly based on unverified information about the Soviet method of production. Such a methodology would calculate the surrogate value of concentrated carnallite by comingling the surrogate value for raw carnallite (the value of Brazilian raw dolomite) with estimated Russian costs related to converting raw carnallite into concentrated carnallite. In following this methodology, plaintiffs point to their own estimated Russian costs and claim that Commerce should make the appropriate adjustments which reflect such costs.
Plaintiffs’ methodology is not codified anywhere in the statute, and would be an aberration within the factors-of-production methodology for an NME country.
2. Electricity:
The magnesium production process is electricity intensive.27 Commerce stated that it would prefer to use publicly available, published information (“PAPI”) from Brazil in order to determine which surro[1099]*1099gate value to use for electricity,28 but such information was unavailable for purposes of its preliminary determination. Initiálly, therefore, Commerce used information gained from a staffperson at the U.S. Consulate in Belo Horizonte, Brazil.29
The surrogate value for electricity was calculated to be at $0.055 per kWh, a simple average of the reported $/kWh rates for industrial use in Brazil.30 Commerce then requested that the parties provide factor price data.31 The Russian respondents provided PAPI for Brazilian large industrial users, such as magnesium or aluminum producers.32 Plaintiffs did not submit any PAPI data for electricity, but commented on Russian respondents’ submissions.33
In its Final Determination, Commerce decided that the Brazilian “large industry user” rate should apply and that the value for electricity should therefore be $0.0235/kWh.34 Commerce noted that the electricity consumption of Avisma and Solikamsk “was significantly higher than the minimum necessary to receive the lowest rate in Brazil.”35 Commerce also requested advice from a private firm, CSA Energy Consulting (“CSA”), on the relationship between rates and electricity consumption, and between electricity consumption and line tension.36
Upon learning that Commerce had consulted CSA, plaintiffs also contacted CSA and solicited its advice, providing it with a copy of the Brazilian tariff schedules.37 Then, based on the results of their own consultation contained in a letter sent by another CSA official, Mr. Palermo, plaintiffs requested correction of ministerial errors allegedly con-[1100]*1100tamed in the final determination.38 Commerce rejected plaintiffs’ allegations.39
Plaintiffs now contend that the Brazilian tariff schedules used by Commerce were not the tariffs of CEMIG (which supplies electricity to Brasmag, the local magnesium producer), which has only two rates, domestic and industrial. The tariffs used by Commerce, instead, reported customer rates varying according to the tension of the line, where the lowest rate of 2.3 cents per kWh corresponds to a tension line of 230 kV A line of230kY plaintiffs argue, is far more than Russian producers would need.40 Plaintiffs refer to the consultation that they received from CSA, where Mr. Palermo indicated that “[a]n industrial customer with a peak load of about 20,000 kw would take service” on A3a or A4 rates, which are more expensive than the A1 rate.41 All these errors were brought to Commerce’s attention in the ministerial error submission.
Plaintiffs’ alleged errors, however, are immaterial because Commerce did not rely on the CSA statement in making its final determination. Instead, Commerce relied on PAPI provided by the Russian respondents on Brazilian electricity rates;42 this is Commerce’s preferred methodology for valuing factors of production.43 The record shows that the production of aluminum is electricity intensive, and that in this regard the aluminum industry is comparable to the magnesium industry.44 Based on the tariff schedules and an article from the “American Metal Market” which reported the electricity rates charged to the largest industrial producers of aluminum, Commerce averaged the A1 rate for each month during the POI, and determined the surrogate value for electricity to be $0.0234 per kWh.
The critical issue, however, is whether the magnesium industry requires enough electricity to qualify for the lowest rate, Al, as applied [1101]*1101by Commerce. The record indicates that it does. Commerce asserts that it had verified that both Avisma and Solikamsk used very large amounts of electricity,45 and that the electricity rate charged to Brazilian large users is the A1 rate. Electricity constitutes a large portion of the costs incurred in the production of magnesium. Based on the evidence on the record, it is reasonable to conclude that magnesium producers use electricity at the lowest rate available. Moreover, record evidence also shows that a planned magnesium investment in Brazil would have an energy line of 230kV46 Finally, no evidence indicates that the lowest rate was normally granted only upon “special deals,” as plaintiffs object.47 Consequently, Commerce’s determination is based on substantial evidence on the record.
Plaintiffs also claim that Commerce’s use of CSA’s consultation violated their due process rights, because they were not given any notice nor opportunity to comment.48
Commerce did not inform plaintiffs of its request to CSA and did not give plaintiffs any opportunity to review CSA’s findings prior to the final determination. It is evident, though, that plaintiffs had been aware of the question submitted to CSA since January 13,1995, when the Russian respondents submitted the PAPI of electricity rates in Brazil, and that plaintiffs had full opportunity to respond and contest the data.49 Moreover, even though Commerce may seek corroboration from an outside expert on information already on the record, CSA’s opinion was not material to the final decision. In fact, Commerce issued its final determination before receiving any response from CSA.
Thus, Commerce did not rely on CSA’s consultation in its determination. Commerce, in fact, reported CSA’s statement using a patently incorrect electrical terminology.50 The Court is persuaded that such a statement could not possibly have been the basis of Commerce’s decision without creating dramatic abnormalities within the decision itself. The decision contains no such abnormalities.
The cases cited by plaintiffs are inapposite. In PPG Industries v. United States, 13 CIT 183 (1989), in particular, this court reviewed Commerce’s attempt to include in the record a document containing [1102]*1102information on which Commerce had relied and which formed the basis for the final decision. In the present case Commerce sought confirmation from an outside expert with regard to information already on the record. CSA’s consultation, therefore, was immaterial to Commerce’s determination. Commerce relied on its own expertise and reached the same conclusion based on evidence on the record.
The fact pattern of the present case is similar to that in Timken v. United States, 12 CIT 955, 699 F. Supp. 300 (1988), aff'd, 894 F.2d 385 (Fed. Cir. 1990). In Timken, plaintiff claimed denial of due process alleging that it had not been provided with an opportunity to comment on a telex Commerce received eight days before publication of its final determination. The court rejected plaintiffs claim because record evidence showed plaintiffs full involvement in the investigation, where it “received and submitted data on contested matters, participated in public hearings, and filed briefs.” Id. at 309.51 Similarly, this Court rejects plaintiffs’ claim of violation of due process rights.
3. Factory Overhead:
Factory overhead represents the indirect manufacturing costs that a company incurs. Included in factory overhead are costs such as depreciation, insurance, taxes, repairs and maintenance, supervisory salaries, manufacturing supplies, power, and indirect labor.52 The value for factory overhead is generally calculated as a percentage of manufacturing costs.53 For its calculation of factory overhead, Commerce relied on data taken from another antidumping investigation on a Brazilian silico-manganese producer, and calculated the surrogate value for factory overhead at a 22 percent rate.54
Plaintiffs indicate that the magnesium industry uses electrolytic cells which must be rebuilt every 450-600 days, whereas silicomanganese furnaces last five to seven years. Furthermore, plaintiffs assert that the electrolytic cells used to produce magnesium require continuous maintenance because of the nature of the materials employed, whereas silico-manganese furnaces require minimal repair costs.55 Thus, plaintiffs contend that Commerce should have adjusted the factory overhead costs of silicomanganese producers upward in order to account for the higher costs in the production of magnesium. According to plaintiffs, absent [1103]*1103any appropriate surrogate information on factory overhead in the Brazilian magnesium industry, Commerce should have used petitioners’ experience to value the cell rebuild overhead costs, as it has done in similar circumstances in the past.56 By adding MagCorp’s electrolytic cell rebuild expense rate, plaintiffs claim, Commerce should have reached a total factory overhead rate that is far higher than 22 percent.
Commerce counters that the factors-of-production analysis is complex, especially when examining a particularly complicated production process like the one used in the magnesium industry.57 Consistent with its usual practice,58 Commerce rejected an “item-by-item evaluation of overhead components,” and instead calculated factory overhead as a percentage of the total cost of manufacturing. Commerce maintains that no data were available on magnesium production in Brazil, so it resorted to the best information publicly available on silicomanganese production in Brazil as surrogate value.59
Quite correctly, Commerce explained that “[fjactory overhead is a combination of elements, some of which may be more or less expensive depending on the product or even the company.”60 Factory overhead reported by MagCorp is composed of cell rebuilds, maintenance materials, plant supplies, depreciation, and other fixed overhead.61 Factory overhead reported by the Brazilian silicomanganese producer, conversely, is composed of costs for labor, contractors, indirect materials, maintenance, rents, others, amortizations, insurance, depreciation, power — demand, and power — auxiliaries.62 The discrepancy stems from different cost accounting methods rather than different overall factory overhead costs. Cost accounting is characterized by its multiple methodologies, which although different may all be appropriate.
The discrepancy in cost accounting methodologies for factory overhead determines the differences at issue here. Even if electrolytic cell rebuild costs in the magnesium industry are higher than the correspondent costs in the silicomanganese industry, there may be other costs in the silicomanganese industry factory overhead which are higher than in [1104]*1104the magnesium industry.63 This might result from the adoption of different, even though equally acceptable, accounting methodologies, which are inconsistent with one another (i.e., in the depreciation of fixed costs). There is no evidence on the record that MagCorp’s method of allocating inventory better reflects the subject NME country realities or the experience of any Brazilian magnesium producer.
Commerce “utilize[d], to the extent possible, the prices or costs of factors of production in one * * * market economy countr[y] * * * at a level of economic development comparable to that of the nonmarket economy country. ”64 Adhering to the statutory mandate, Commerce used factory overhead data taken from an investigation of a Brazilian silicomanga-nese producer,65 because the Brazilian data were deemed to reflect an experience close to that of magnesium producers in the subject NME countries.66 In addition, the Brazilian data offered the advantage of referring to the same period of investigation.
It is Commerce’s standard practice to disregard petitioners’ costs because they are not “an appropriate benchmark by which to test the accuracy of surrogate country values.”67 While in an extraordinary circumstance — such as an absolute lack of information — it would be appropriate for Commerce to refer to petitioners’ data, plaintiffs have not demonstrated that such a circumstance exists in this case. Commerce may use considerable discretion in the evaluation of factors of production. See Technoimportexport v. United States, 16 CIT 13, 14, 783 F. Supp. 1401, 1404 (1992). Commerce’s decision to rely on the factory overhead ratio of a Brazilian silicomanganese producer is therefore appropriate.
4. SG&A:
Selling, General & Administrative Expenses (“SG&A”) is a ratio of general expenses to the cost of manufacturing. Commerce used the information submitted in the antidumping investigation of Silicoman-ganese from Brazil to calculate a representative ratio of SG&A for a [1105]*1105Brazilian silicomanganese producer.68 From the silicomanganese producer’s data, Commerce took the values within the field for general selling expenses (“GNAU”), and calculated an SG&A of 10 percent.
Plaintiffs contest the calculations because the line entitled “GNAU” includes general and administrative expenses but does not include selling expenses. Plaintiffs indicate that the field “TOTGENU,” on the other hand, is more appropriate because it represents the total SG&A percentage (approx. 18 percent), which includes selling expenses.
Commerce acknowledges that the field GNAU used to calculate the surrogate value for SG&A does not include selling expenses. Commerce therefore requests a remand to correct its valuation of SG&A expenses for the Russian producers, and indicates that on remand it will use the field “TOTGENU.”
Defendant-intervenors Avisma and Solikamsk argue that if the court concludes that Commerce’s calculations are not supported by substantial evidence then the data they placed on the record offer the best alternative. These data show SG&A rates between 11.8 and 13.6 percent.69 Defendant-intervenors also object that plaintiffs’ claim is raised here for the first time, and should therefore not be entertained.
Defendant-intervenors’ objection has merit. Plaintiffs had full opportunity to point out the alleged errors contained in Commerce’s preliminary determination, but they failed to do so. Since the preliminary investigation, Commerce has calculated the SG&A ratio using the financial data from the GNAU field rather than from the TOTGENU field. Normally, a plaintiff that did not exhaust its administrative remedies would be estopped from raising the claim before this court.70 The-Court, however, recognizes that Commerce has requested remand to correct its calculations of SG&A expenses.
Time and effort will be saved if the calculations challenged before the Court are what Commerce considers to be its true and accurate final results. Therefore, the Court remands to Commerce to recalculate the surrogate value for Russian producers’ SG&A expenses because Commerce has admitted that its valuation is not supported by substantial evidence on the record. On remand, all the parties will be afforded the opportunity to review Commerce’s proposed corrections, and to make comments on them. Therefore, at this time, the Court does not address the issue of SG&A.
5. By-Product Credits:
Chlorine and potassium chloride are by-products of the magnesium production process.71 Since Avisma and Solikamsk either use or sell part [1106]*1106of these by-products, Commerce computed by-product credits in the calculation of the FMV In the preliminary determination, Commerce accounted for the by-products by subtracting their value from the surrogate value of the cost of materials. In its final determination, accepting plaintiffs’ argument, Commerce “subtracted the total by-product cost [sic] from the cost of manufacture (sum of materials, labor, energy, and imputed factory overhead, or “COM”).”72 Deducting the by-product credits from the cost of manufacture rather than from the cost of materials increases the factory overhead amount,73 because the factory overhead amount is calculated by multiplying a percentage by an amount which has not been reduced by the amount of the by-product credits. Commerce asserts that the increase in the amount of factory overhead reflects the by-product processing costs, thereby “eliminating the need for valuing any additional processing-related elements.”74
Plaintiffs disagree. Under generally accepted accounting principles (“GAAP”), plaintiffs explain, the by-product processing costs should have been deducted from the by-product revenues. Plaintiffs demonstrate that domestic magnesium producers incur significant processing costs in order to render the by-products suitable for use or sale.75 Plaintiffs argue that Commerce should have resorted to this information because respondents did not cooperate in furnishing their questionnaire answers. Plaintiffs also assert that Avisma, which uses some of its chlorine by-product to produce titanium sponge, had already claimed the same by-product credit for the same period in another antidumping investigation, Titanium Sponge from the Soviet Union.
The parties do not contest Commerce’s finding that chlorine and potassium chloride are by-products of magnesium production. A byproduct is qualified by its “overall * * * insignificant [value] when compared to the relative value of the * * * subject merchandise.”76 When accounting for by-products, Commerce usually subtracts the sales revenue (or the value of by-products recovered) from the production costs of the product under investigation.77
[1107]*1107The antidumping statute provides that the foreign market value is determined on the basis of the factors-of-production methodology. 19 U.S.C. § 1677b(c)(1). The factors of production include, but are not limited to, labor, raw materials, energy and other utilities consumed, and capital cost. 19 U.S.C. § 1677b(c)(3). The statute also directs Commerce to add an amount for general expenses to the value of the factors of production. 19 U.S.C. § 1677b(c)(1). The cost of general expenses encompasses factory overhead. IPSCO, Inc. v. United States, 965 F.2d 1056, 1059 (Fed. Cir. 1992) (discussing constructed value).78
Commerce allocated by-products processing costs to factory overhead.79 Commerce determined that the increase in factory overhead was representative of the costs of processing the by-products,80 and thus eliminated the need for calculating costs related to the by-product processing.81 In addition, Commerce “adjusted the total by-product cost to reflect the actual purity concentration of the by-product in the production process.”82
Commerce’s methodology reflects a reasonable interpretation of the provisions for nonmarket economy countries, codified under 19 U.S.C. § 1677b(c), because it accounts for both the value of the by-products and the costs related to by-product processing. This methodology is reasonably based on substantial evidence in the administrative record. Commerce verified the indirect costs, including by-product processing costs, at the respondents’ facilities in Russia.83 The incidental nature of byproducts and their insignificant value does not warrant costly accounting procedures.84 This Court cannot substitute its own construction for the reasonable interpretation made by Commerce. See Chevron U.S.A. v. Natural Resources Defense Council, 467 U.S. 837, 844 (1984); Fujitsu [1108]*1108General Ltd. v. United States, No. 95-1343 at 8 (Fed. Cir. July 3, 1996). Thus, Commerce’s determination is sustained.
Finally, plaintiffs’ contention that Avisma allegedly duplicated its byproduct credits is immaterial to this action, because it is uncontroverted that Commerce has not granted any offset claim in the titanium sponge case. That case is not before the Court.
6. U.S. Sales:
Razno-Alloys, Ltd. (“Razno”), AIOC Corporation (“AIOC”), and Interlink Metals and Chemicals, S.A. (“Interlink”) are three of the trading companies that reported some of their U.S. sales of pure magnesium just prior to, or during, verification.85
At the start of the investigation, Razno reported one sale of magnesium directed to the United States, and declared that the major volume of sales during the POI was directed to ‘“w/house [warehouse] Kotka and/or Loviisa and/or Rotterdam’ and ‘FOB Kotka’.”86 Subsequently, Razno submitted information on ten additional sales of stockpile magnesium.87 Commerce verified Razno’s sales documentation in Zurich, Switzerland, where the accounts and sales records were maintained,88 and found that Razno had reported all of its sales made during the POI. Razno justified its initial omission of the ten sales by declaring that it was unaware that they were destined for the U.S. market.89 Commerce verified Razno’s claim and found no discrepancies.90 However, Commerce’s examination of “Razno’s sales ledgers and contract files revealed two POI U.S. sales* * * that had not been reported as POI sales to the United States,”91 apparently because the final destination was not indicated in the invoice.92
Plaintiffs claim that the information which Razno failed to report was so significant that Commerce should have questioned the veracity of Razno’s entire submission, and, in accordance with Commerce’s common practice, resorted to total or partial best information available [1109]*1109(“BIA”) to calculate dumping margins.93 Instead, plaintiffs argue, Commerce departed from its standard practice without any explanation.94
Plaintiffs also censure Commerce’s treatment of late submissions from two other trade companies, AIOC and Interlink. During its verification, AIOC reported the existence of a purchase contract which had not been included in its response to Commerce’s questionnaire.95 Twenty-one sales were connected to this purchase contract, and were subsequently verified by Commerce.96 Similarly, during verification Interlink provided Commerce with information about a number of sales not included in its questionnaire response, thereby increasing the total quantity of magnesium related to its sales.97 Commerce examined logbooks and all sales contracts filed within the POI, noting no evidence of unreported sales to the U.S.98
Plaintiffs claim that Commerce should have rejected AIOC and Interlink’s sales submissions as untimely, and used BIA instead, or explained its diversion from its standard practice.
The antidumping statute requires Commerce to use BIA whenever Commerce is “unable to verify the accuracy of the information submitted.”99 Verification includes “access to all files, records, and personnel of the producers, resellers, importers, or unrelated purchasers which the Secretary considers relevant to factual information submitted.”100 If the agency is unable to verify within a specific time the accuracy and completeness of the information submitted, it will use BIA.101 Similarly, Commerce will use BIA if a party refuses or is unable to provide the information requested, or if the party significantly impedes an investigation.102 This is consistent with “the basic purpose of the statute: determining current margins as accurately as possible.” Rhone Poulenc, Inc. v. United States, 899 F.2d 1185, 1191 (1990). Never[1110]*1110theless, the statute does not define explicitly what constitutes best information available.103
The absence of a statutory definition of BIA vests the administering agency with broad discretion. The agency’s construction of BIA, therefore, must be accorded considerable deference. Chevron U.S.A., Inc. v. Natural Resources Defense Council, Inc., 467 U.S. 837, 843-44 (1984); Allied-Signal Aerospace Co. v. United States, 996 F.2d 1185 (Fed. Cir. 1993). In deciding whether to resort to any BIA, the agency must consider the cooperation shown by the parties to the investigation “to gather the data needed for an accurate determination.” Atlantic Sugar, Ltd. v. United States, 744 F.2d 1556, 1560 (Fed. Cir. 1984). Noncooperation may warrant the agency’s “use of whatever other best information it may have available.” Id.104 “Commerce’s discretion to determine whether a respondent has complied with an information request” is consistent with “the discretion afforded Commerce in its administration of the antidumping laws.” Daido Corp. v. United States, 893 F. Supp. 43, 49 (CIT 1995) (citing Smith-Corona Group v. United States, 713 F.2d 1568, 1571 (1983)). The same discretion allows Commerce to request additional information from respondents before resorting to BIA.105
In this case, the information necessary to reach an accurate dumping determination was made available by the respondents and was verified as accurate and reliable by Commerce. Commerce considered the initial inaccuracies in the data as inadvertent errors. In the end, the errors were corrected and the data were verified.106 The low number of U.S. sales of magnesium allowed Commerce to conduct thorough and independent verifications, during which Commerce examined all the POI data, matching the data received from the producers with those received from the resellers/exporters.107 Plaintiffs correctly point out that Commerce usually rejects untimely submissions and uses BIA instead, because “the late submission of this information [leaves] no opportunity to analyze'the sales reporting and provide deficiency questions, and no opportunity for Petitioners to analyze and comment on these sales.”108 In this case, however, Commerce rendered a thorough verification, and [1111]*1111plaintiffs were offered full opportunity to comment on the verification report.109 Therefore, Commerce did not depart from its usual practice.
Finally, Commerce’s verification does not contain any methodological problems or mathematical errors. Indeed, plaintiffs do not claim that the calculation of respondents’ margins is incorrect. Plaintiffs’ claim, rather, proposes a blind, punitive use of BIA, which is clearly disfavored. See Rhone Poulenc, Inc. v. United States, 899 F.2d 1185, 1190 (Fed. Cir. 1990). Commerce determined that the errors contained in respondents’ answers to Commerce’s questionnaire “were inadvertent and were, in the end, verified;” that the margins imputed to those companies were correct; that the record was “ complete and accurate. ”110 Upon review of the administrative record, the Court concludes that Commerce’s determination to rely on the reported information and not to resort to any BIA was neither arbitrary nor capricious, is supported by substantial evidence, and is otherwise in accordance with law.
7. Third-country Resellerlexporters’ SG&A Expenses and Profits:
Most of the magnesium sold to the United States was exported by trading companies located in third-country market economies. Commerce did not include these companies’ SG&A expenses and profits in the calculation of the FMV of Russian magnesium.111
Plaintiffs claim that Commerce’s exclusion of reseller/exporters’ SG&A and profit amounts to a failure to include the costs of the exporter together with those of the producer, “to the extent necessary to accurately calculate the total amount incurred and realized for costs, expenses, and profits in connection with production and sale of that merchandise.”112 By doing so, plantiffs aver, Commerce did not compare United States price (“USP”) and FMV on a fair basis, i.e., at a common point in the chain of commerce. Plaintiffs argue that USP was based on resellers’ U.S. selling prices, whereas reseller/exporters’ constructed value (“CV”) included only the SG&A expenses and profits of the producers but not of the intermediate resellers and the final reseller/ exporter.113
The Court observes that for nonmarket economy countries the statute mandates that the FMV of the merchandise be determined “on the [1112]*1112basis of the value of the factors of production utilized in producing the merchandise and to which shall be added an amount for general expenses and profit plus the cost of containers, coverings, and other expenses, as required by subsection (e) of this section.” 19 U.S.C. § 1677b(c)(l) (1988). The amount for general expenses and profit is “equal to that usually reflected in sales of merchandise * * * which are made by producers in the country of exportation * * 19 U.S.C. § 1677b(e)(l)(B) (1988) (emphasis provided).
Thus, the statutory amount for general expenses and profit is limited to sales made by producers in the country of exportation. The statute does not mention resellers’ sales. Therefore, the only general expenses and profit to be accounted for are those of the producers and not the resellers. The language of the statute provides an unambiguous directive for the agency and for the court, to which the agency and the court must give effect. This result accords with the recent decision of this Court in Rhone-Poulenc, Inc., v. United States, slip op. 96-74 at 20-21 (Ct. Int’l Trade, May 9, 1996).
The statute requires a comparison between USP and FMV based on the factors of production to which general expenses, profit and the cost of containers are added. 19 U.S.C. § 1677b(c).114 General expenses are based on the sales made by producers in the country of exportation. This is in accord with Commerce’s precedent115 and with Commerce’s determination in this case.
Plaintiffs have acknowledged that a circumstance of sale adjustment “may be an appropriate mechanism for adjusting FMV to account for additional reseller costs. ” (Plaintiffs’ Reply Brief at 45.) However, plaintiffs make no such claim. Rather, they focus on Commerce’s alleged unexplained deviation from its practice by excluding resellers’ SG&A from FMV
In asserting that Commerce has an established practice to include reseller expenses in CV¡ plaintiffs rely on Fresh and Chilled Atlantic Salmon from Norway,116 in which Commerce included the exporters’ expenses in the calculation of FMV Salmon from Norway, however, is distinguishable from the present case on at least two grounds: First, the producers had no selling expenses because most of their selling activities were handled by the exporters;117 Second, the FMV was based on [1113]*1113third country sales.118 To accurately calculate the cost of production of salmon sold by the exporters, Commerce added the exporters’ SG&A to the salmon farmers’ cost of production.119 Thus, in Salmon from Norway, Commerce had to calculate a third-country sale value which necessarily included the exporters’ SG&A. This situation, however, is different from the one that plaintiffs presented in their petition.
Plaintiffs also rely on Tubeless Steel Disc Wheels From Brazil,120 which involved merchandise produced in a market economy and sold only to the United States. Since the merchandise was not sold in the home country nor to third countries, Commerce had to calculate CV¡ and used U.S. selling expenses as a surrogate.121
The Court finds that Salmon from Norway and Tubeless Steel Disc Wheels From Brazil are inapposite, and that Commerce did not depart from its practice.
Finally, no evidence points to mistakes committed by Commerce in the construction of FMV Commerce included all costs related to the Brazilian production and sale of merchandise to the first unrelated seller in the home market.122 Commerce’s action was in accordance with the the statute, which grants Commerce discretion to select the appropriate surrogate value for SG&A expenses. In the case of an NME country, Commerce usually refers to surrogate values because of the distortions to which NME costs and prices are subject. Nevertheless, Commerce may use prices paid by the NME country to market-economy suppliers in combination with surrogate values, when this serves the statutory purpose of “determining current margins as accurately as possible,” Rhone-Poulenc, Inc. v. United States, 899 F.2d 1185, 1191 (Fed. Cir. 1990), and on “fair and equitable basis,” Koyo Seiko Co. v. United States, 36 F.3d 1565, 1573 (Fed. Cir. 1994).123 Commerce did not deem it appropriate to deviate from the surrogate values calculated on the basis of the Brazilian information and to follow a different methodology. The Court finds that Commerce’s determination is in accordance with law.
8. Export Taxes and Mandatory Monetary Conversion:
Avisma and Solikamsk reported that, based on the quantity exported, their magnesium sales were subject to Russian export duties and administrative procedure fees.124 As the exporting companies, Solikamsk and Avisma were responsible for paying these duties and fees, either in [1114]*1114rubles or in dollars.125 Commerce’s final determination stated that “ [w]e made no deduction from USP to account for exporter-incurred selling expenses, nor did we deduct export taxes paid by Russian companies to the Russian government because the actual amounts paid are an internal expense within an NME country.”126
Commerce has requested that the Court remand Commerce’s determination on Russian export taxes. Commerce has declared that the agency “is currently reevaluating its interpretation of section 772 of the Act as applied to nonmarket economies, ” and “the legal and policy bases for its treatment of expenses related to the USE”127
A remand is warranted to ensure a consistent approach in this and other continuing nonmarket economy proceedings. Consequently, the Court does not address the issue of the deductibility of export taxes from USl^ and remands to Commerce on the question of whether the export taxes paid by an NME producer can be deducted from USE
Plaintiffs’ second claim is that Avisma and Solikamsk’s requirement to sell half of their hard currency revenue for rubles to a Russian bank within fourteen days of receipt128 amounts to “an implied export tax.”129 This exchange rate balancing requirement, according to plaintiffs, represents a direct cost to producers because it involves the application of a less favorable government exchange rate. Plaintiffs acknowledge that they are raising this claim for the first time here. They point out, however, that Commerce refused to make adjustments for export taxes, and that therefore it would have been futile for plaintiffs to raise this other export tax issue, the exchange rate balancing requirement.130 Consequently, plaintiffs contend that it is appropriate to address their second claim now, together with the export taxes claim.
Defendant-intervenors insist that plaintiffs are barred from raising this claim here. The fact that plaintiffs mentioned the exchange rate balancing requirement once in a submission discussing whether or not Russia could be considered a market economy country131 is insufficient to include this question within the broad issue of the export taxes. See Timken, 16 CIT 429, 436, 795 F. Supp. at 442-44 (citing Rhone Poulenc, Inc. v. United States, 899 F.2d 1185, 1191 (Fed. Cir. 1990)) (holding that it is insufficient to argue that raising broad issues implicitly raises specific issues).
Certainly, plaintiffs could have raised this claim before the agency. Plaintiffs fully discussed the export tax issue during the proceeding, but [1115]*1115never raised or even addressed the question of the exchange rate balancing requirement. The two claims are essentially different. One constitutes an export tax, whereas the other resembles an involuntary loan made in currency. Plaintiffs have not demonstrated that the mandatory conversion requirement is included in the export taxes issue, or that the two issues are related in any way.
A plaintiff that did not exhaust its administrative remedies is estopped from raising the claim before the court. 28 U.S.C. § 2637(d) (1994). This Court would usurp the agency’s function if it examined Commerce’s determination upon a ground not presented before Commerce, and for which Commerce did not have an opportunity to consider the matter and to state the reason for its determination. Unemployment Compensation Comm’n of Alaska v. Aragon, 329 U.S. 143, 155 (1946). See also Rhone Poulenc, Inc. v. United States, 899 F.2d 1185, 1191 (Fed. Cir. 1990). Accordingly, the Court concludes that plaintiffs’ failure to raise the issue of the exchange rate balancing requirement before Commerce bars plaintiffs from raising that issue before the Court now for the first time.
Conclusion
In accordance with the foregoing opinion, this case is remanded to Commerce to recalculate the surrogate value for SG&A for the Russian producers, and to consider whether the export taxes paid by an NME producer should be deducted from USD Remand results are due within sixty days of the date this opinion is entered. Comments and responses are due within thirty days thereafter. Any rebuttal comments are due within fifteen days after the date comments or responses are due. Plaintiffs’ motion upon the agency record is denied in all other respects.