*619Opinion
Tsoucalas, Senior Judge:Plaintiffs and defendant-intervenors move this Court for judgment on the agency record pursuant to Rule 56.2 of the Rules of this Court. Plaintiffs and defendant-intervenors challenge the Department of Commerce, International Trade Administration’s (“Commerce”) final results of the fourth administrative review for anti-friction bearings (“AFBs”), entitled Antifriction Bearings (Other Than Tapered Roller Bearings) and Parts Thereof From France, et al.; Final Results of Antidumping Duty Administrative Reviews, Partial Termination of Administrative Reviews, and Revocation in Part of Anti-dumping Duty Orders (“Final Results”), 60 Fed. Reg. 10,900 (Feb. 28, 1995).
Background
The fourth administrative review encompasses imports of AFBs entered during the fourth review period of May 1,1992 through April 30, 1993. See Final Results, 60 Fed. Reg. at 10,900. The present consolidated action concerns imports from Japan.
On February 28,1994, Commerce published the preliminary results of the fourth administrative review. See Antifriction Bearings (Other Than Tapered Roller Bearings) and Parts Thereof From France, Germany, Italy, Japan, Singapore, Sweden, Thailand, and the United Kingdom; Preliminary Results of Antidumping Duty Administrative Reviews, Partial Termination of Administrative Reviews, and Notice of Intent To Revoke Orders (in Part), 59 Fed. Reg. 9,463 (Feb. 28,1994). On February 28,1995, Commerce published the Final Results at issue. See Final Results, 60 Fed. Reg. at 10,900.
NSK Ltd. and NSK Corporation (“NSK”) claims Commerce erred in: (1) failing to apply a tax-neutral methodology in computing the value-added tax (“VAT”) adjustment; (2) treating NSK’s return rebates and post-sale price adjustments as indirect selling expenses; (3) denying NSK a direct adjustment to foreign market value (“FMV”) for home market early payment discounts and distributor incentives; (4) rejecting NSK’s lump sum post-sale price adjustments and stock transfer commissions as indirect expenses; (5) not using NSK’s purchase prices for bearings purchased by NSK from related suppliers; (6) improperly calculating exporter’s sales price (“ESP”) for imported bearing parts further manufactured in the United States; (7) rejecting NSK’s reported interest income offset to interest expense in the calculation of cost of production (“COP”) and constructed value (“CV”); and (8) including zero-priced sample sales in the U.S. database.
Koyo Seiko Co., Ltd. and Koyo Corporation of U.S.A. (“Koyo”) contends Commerce erred in: (1) failing to apply a tax-neutral methodology in computing the VAT adjustment; (2) disallowing certain Koyo home market post-sale price adjustments that were not reported on an invoice- or product-specific basis; (3) investigating the cost of inputs obtained by Koyo from related party suppliers; (4) reclassifying Koyo’s *620non-operating expenses and payments out of retained earnings as production expenses; and (5) committing certain clerical errors.
NTN Bearing Corporation of America, American NTN Bearing Manufacturing Corp., NTN Corporation, NTN Driveshaft, Inc. and NTN-Bower Corporation (“NTN”) argues Commerce erred in: (1) failing to apply a tax-neutral methodology in computing the VAT adjustment; (2) including sample sales in the FMV calculation; (3) crossing levels of trade in comparing U.S. and home market sales; (4) refusing to grant NTN a price-based level of trade adjustment; (5) excluding NTN’s home market sales to related parties in FMV calculation; (6) rejecting NTN’s adjustment for interest on selling expenses; (7) reallocating NTN’s U.S. selling expenses based on the sale price to the first unrelated party; (8) making improper adjustments to NTN’s COP and CV data; and (9) treating home market discounts attributable to sales of subject merchandise as an indirect selling expense.
Nippon Pillow Block Sales Co., Ltd. and FYH Bearing Units USA (“NPB”) asserts Commerce erred in resorting to best information available when NPB failed to report certain negative billing adjustments.
The Torrington Company (“Torrington”) claims Commerce erred in: (1) incorrectly applying the “Roller Chain” and “knowledge” tests to exclude merchandise imported by Honda Motor Co., Ltd., American Honda Motor Co., Inc., Honda of America Mfg., Inc. and Honda Power Equipment Mfg., Inc. (“Honda”); (2) granting billing, post-sale price and warranty credit adjustments that were not linked to specific sales of in-scope merchandise; (3) accepting Koyo’s U.S. freight expenses where air and ocean freight charges were commingled and allocated to all U.S. sales without linkage to specific sales; (4) accepting Koyo’s data regarding “efficiency variances” used in the calculation of cost of production contrary to its own verification; (5) failing to take into account certain related party commissions paid by NTN with respect to purchase price sales; (6) accepting NTN’s designation of certain sales at the “aftermarket” level of trade and NSK’s designation of certain distributor sales as destined to original equipment manufacturers (“OEMs”); and (7) committing certain clerical errors.
Discussion
The Court’s jurisdiction in this action is derived from 19 U.S.C. § 1516a(a)(2) (1994) and 28 U.S.C. § 1581(c) (1994).
The Court must uphold Commerce’s final determination unless it is “unsupported by substantial evidence on the record, or otherwise not in accordance with law.” 19 U.S.C. § 1516a(b)(l)(B). Substantial evidence is “more than a mere scintilla. It means such relevant evidence as a reasonable mind might accept as adequate to support a conclusion.” Universal Camera Corp. v. NLRB, 340 U.S. 474, 477 (1951) (quoting Consolidated Edison Co. v. NLRB, 305 U.S. 197, 229 (1938)). “It is not within the Court’s domain either to weigh the adequate quality or quantity of the evidence for sufficiency or to reject a finding on grounds of a *621differing interpretation of the record.” Timken Co. v. United States, 12 CIT 955,962,699 E Supp. 300,306 (1988), aff’cl, 894 F.2d 385 (Fed. Cir. 1990).
1. VAT Adjustment:
NSK, Koyo and NTN request a remand for Commerce to apply a tax-neutral amount, rather than rate, methodology in computing the VAT adjustment to U.S. price. NSK’s Mem. Supp. Mot. J. Agency R. at 52-53; Koyo’s Mem. Supp. Mot. J. Agency R. at 8-10; NTN’s Mem. Supp. Mot. J. Agency R. at 47-48. Commerce consents to a remand so that it may apply a tax-neutral methodology. Def. ’s Partial Opp’n to Mots. J. Agency R. at 7-10.
In light of Federal-Mogul Corp. v. United States, 63 F.3d 1572, 1580 (Fed. Cir. 1995), requiring Commerce to apply a tax-neutral methodology, the Court agrees that remand is appropriate. This case is therefore remanded to Commerce to apply a tax-neutral methodology in computing the VAT adjustment to U.S. price for NSK, Koyo and NTN.
2. FMV Adjustments for Various Home Market Post-Sale Adjustments:
In calculating FMV and U.S. price, Commerce must determine the price actually charged by a respondent for the merchandise at issue, including discounts, rebates and price adjustments. See 19 U.S.C. §§ 1677a & 1677b (1988). Commerce believes the effect of allowing allocations for price adjustments that are averaged over all sales is to distort the actual price paid for each specific sale. Commerce has therefore primarily required respondents to provide transaction-specific information before allowing direct adjustments to FMV for discounts, rebates and post-sale price adjustments (“PSPAs”).
Upon denying a direct adjustment to FMV for certain selling expenses directly related to underlying sales, Commerce has allowed certain price adjustments to be allocated as indirect selling expenses under the ESP offset. However, this practice was recently rejected by the United States Court of Appeals for the Federal Circuit (“CAFC”). In Torrington Co. v. United States, 82 F.3d 1039 (Fed. Cir. 1996), the CAFC had to determine whether certain PSPAs to FMV were appropriate. The PSPAs at issue involved the correction of invoicing errors and retroactive price changes recorded only on a customer-specific basis. The Court first noted that the determining factor for whether PSPAs qualify for ESP offset treatment is on the PSPA calculation method, not the recordation or allocation method.1 The Court then found the following: .
Both kinds of PSPA [at issue] involve adjustments to the price of a product, or group of products, made in response to billing errors for *622a particular customer or in response to the post-sale raising or lowering of the price for a particular customer.
Torrington, 82 F.3d at 1050. The CAFC went on to conclude that, because these PSPAs represent expenses related to a particular sale or sales, they vary with the quantity of the item sold, and so, constitute direct selling expenses.2 However, the CAFC noted that 19 C.F.R. § 353.56(b)(2) explicitly disallows direct selling expenses from being deducted from FMV as ESP offsets. Torrington, 82 F.3d at 1051 (citing Sharp Corp. v. United States, 63 F.3d 1092,1096 (Fed. Cir. 1995) (stating 19 C.F.R. § 353.56(b) (2) provides for a deduction from FMV for all selling expenses except direct selling expenses)); see also Timken Co. v. United States, 20 CIT 645, 658, 930 F. Supp. 621, 632 (1996).
NSK, NTN, Koyo and Torrington contest Commerce’s treatment of certain post-sale price adjustments in this review.
a. NSK’s Adjustments:
NSK contends Commerce improperly: (1) treated NSK’s return rebate and post-sale price adjustments as indirect selling expenses; (2) denied NSK a direct adjustment to FMV for NSK’s home market early payment discounts and distributor incentives; and (3) rejected NSK’s lump sum post-sale price adjustments and stock transfer commissions as indirect expenses.3 NSK’s Mem. Supp. Mot. J. Agency R. at 21-34.
To retain its Japanese OEM customers that purchased products through distributors yet prevent these distributors from reaping huge profits through sales to aftermarket customers, NSK established a return rebate program. In this program, NSK sells bearings to distributors at a provisional price equal to prices NSK charges to aftermarket customers, but rebates a percentage of this price following proof of the quantities and price charged by the distributor to the OEM. The NSK PSPAs at issue involve the correction of clerical errors, the finalization of a temporary sale price and the price adjustments for changes negotiated subsequent to sale.
*623NSK claims its return rebates and PSPAs should have been treated as direct expenses because they were essentially reported on a transaction-specific basis and, therefore, are directly linked to underlying sales. See NSK’s Mem. Supp. Mot. J. Agency R. at 23-26; NSK’s Reply Mem. Supp. Mot. J. Agency R. at 3-6. While NSK concedes that its business records did not allow it to calculate the return rebates on a transaction-specific basis, it maintains that, because its return rebates did not vary during the review period, its methodology adjusted the selling price of each transaction to reflect the rebate actually paid by NSK to the distributor. NSK’s Mem. Supp. Mot. J. Agency R. at 23.
Commerce maintains it properly dealt with NSK’s return rebates and PSPAs as indirect expenses because they were not reported on a transaction-specific basis. Def.’s Partial Opp’n to Mots. J. Agency R. at 15-18.
Torrington agrees Commerce should not allow NSK a direct adjustment to FMV for return rebates and PSPAs. Torrington’s Opp’n to Mots. J. Agency R. at 10-14. However, Torrington argues Commerce should not allow even an indirect adjustment to FMV for NSK’s PSPAs because: (1) PSPAs are, by nature, price adjustments incurred on, and inseparable from, specific transactions, not selling expenses incurred on all sales; and (2) allowing PSPAs that do not qualify as direct adjustments to price to be treated as indirect selling expenses is erroneous.4 Torrington’s Mem. Supp. Mot. J. Agency R. at 41, 42-45.
As a preliminary matter, the case to which Torrington cites for support, RHP Bearings v. United States, 19 CIT 133, 875 F. Supp. 854 (1995), is distinguishable, as it deals with technical service expenses in the U.S. market, which are treated differently than PSPAs in the home market. See Torrington Co. v. United States, 17 CIT 672, 684, 832 F. Supp. 365, 376 (1993) (stating Commerce.may treat similar expenses differently in the calculation of U.S. price and FMV).
Commerce verified that the return rebate program directly related to sales of in-scope bearings to certain distributors, i.e., NSK kept track of sales to each distributor of standard part numbers eligible for the rebate, less resales by the distributor for which a rebate had already been paid. See NSKHM Sales Verification Report, ER. Doc. No. 342, at 9-10, Def.’s App., Ex. 18, at 9-10 (March 4,1994). Further, as NSK’s PSPAs normally occur after the original sale, Commerce verified that NSK’s records link these expenses on a product- and customer-specific basis, but not to the exact original sale. Id. at 10-11.
Similar to the PSPAs at issue in Torrington, therefore, NSK’s return rebates and PSPAs represent expenses related to particular sales of in-scope merchandise and vary with the quantity of the particular item *624sold. While Commerce could not allow these expenses a direct adjustment because of NSK’s failure to tie them to specific transactions and ensure they are not attributable to sales of non-scope merchandise, this does not deprive them of their direct relationship to the sales under consideration. See IN A Walzlager Schaeffler KG v. United States, 21 CIT 110,129,957 F. Supp. 251,267 (1997); Torrington, 20 CIT at 640,926 F. Supp. at 1158. Consequently, pursuant to the CAFC’s decision in Tor-rington, Commerce improperly treated the expenses as indirect. This case is remanded to Commerce to deny the adjustment to FMV for NSK’s return rebates and PSPAs.
NSK further disputes Commerce’s treatment of its early payment discounts, where NSK offered a discount to distributors who paid their invoices before the payment due date, and distributor incentives, where NSK offered payments to distributors as incentives for increases in their sales to approved sub-distributors. NSK’s Mem. Supp. Mot. J. Agency R. at 8,9. Similar to NSK’s other PSPAs, NSK was unable to link the discounts to individual transactions. Further, NSK calculated the distributor incentive rebates on the basis of sales to all products to distributors, and not on a specific-model basis. Commerce therefore denied direct adjustment to FMV for the early payment discounts and distributor rebate incentives. Final Results, 60 Fed. Reg. at 10,935.
NSK claims Commerce improperly denied it a direct FMV adjustment for its early payment discounts and distributor rebate incentives because they were directly related to individual transactions. NSK’s Mem. Supp. Mot. J. Agency R. at 27-31; NSK’s Reply Mem. Supp. Mot. J. Agency R. at 6-9. NSK argues that Commerce, at a minimum, should have granted a direct adjustment for NSK’s sales to four distributors granted discounts as a fixed and constant percentage of sales on all transactions for which they were reported. NSK’s Mem. Supp. Mot. J. Agency R. at 28 (citing Final Results, 60 Fed. Reg. at 10,935).
In response, Commerce maintains NSK could not demonstrate that its early payment discounts and distributor rebate incentives did not include discounts or rebates on non-scope merchandise. Hence, they were not entitled to a FMV adjustment. Def.’s Partial Opp’n to Mots. J. Agency R. at 18-20. Torrington agrees generally with Commerce’s rationale in denying the direct adjustment. Torrington’s Opp’n to Mots. J. Agency R. at 15-19.
Upon inspection of the record, the Court agrees with Commerce that NSK cannot tie its early payment discounts and distributor incentive rebates to particular sales, and so, the amount to which they are attributable to non-scope merchandise is unclear. This holds true even for the early payment discounts granted to the four distributors to which NSK refers. While Commerce verified that discounts to the four distributors had remained relatively stable during the period of investigation, see Final Results, 60 Fed. Reg. at 10,935, NSK could not tie these discounts to specific transactions, and so, could not ensure what amount was granted to in-scope merchandise. See Torrington, 20 CIT at 638-40,926 *625F. Supp. at 1157-58. Consequently, Commerce properly disallowed a direct adjustment for these expenses.
NSK’s final claim involves its lump sum PSPAs, granted on a customer-specific basis, and its stock transfer commissions, where NSK would pay a distributor a commission for purchasing from another distributor a particular part number NSK did not have available. NSK’s Mem Supp. Mot. J. Agency R. at 10-12.
NSK claims Commerce should have treated these expenses as indirect expenses because they did not vary with the quantity of merchandise sold. NSK therefore contends they should have been included in the ESP offset without any requirement that they be linked to in-scope sales. Id. at 31-33; NSK’s Reply Mem. Supp. Mot. J. Agency R. at 10-12.
Commerce responds it properly denied the adjustments because they were made on both in-scope and out-of-scope merchandise. Def.’s Partial Opp’n to Mots. J. Agency R. at 20-22. Torrington agrees generally with the position taken by Commerce. Torrington’s Opp’n to Mots. J. Agency R. at 20-24.
Pursuant to the CAFC’s decision in Torrington, the Court concludes NSK’s lump sum PSPAs are direct expenses because they account for price changes calculated with respect to a specific customer for a specific period of time. Hence, NSK’s lump sum PSPAs are direct expenses, which may not be the basis for a deduction from FMV pursuant to the ESP offset. See Timken Co. v. United States, 20 CIT at 659,930 F. Supp. at 633. Further, while the stock transfer commissions paid on all products was a fixed percentage of the sales price, NSK could not isolate the percentage of commissions paid to distributors for stock transfers of the subject merchandise between distributors. See NSK Section C Questionnaire Response, PR. Doc. No. 126, NSK’s App., Ex. 5, at 50-51 (Sept. 29,1993). Consequently, NSK’s stock transfer commissions are also direct expenses, and so, are not entitled to ESP offset treatment.
b. Koyo’s Adjustments:
(1) Koyo’s PSPAs
Koyo originally claimed Commerce improperly disallowed its adjustment to FMV for its PSPAs because Commerce determined they were made or recorded on a lump sum basis and did not distinguish between in-scope and out-of-scope merchandise. Koyo’s Mem. Supp. Mot. J. Agency R. at 10-11 (citing Final Results, 60 Fed. Reg. at 10,931).
In light of the CAFC’s decision in Torrington, Koyo has decided to no longer pursue this claim. Koyo’s Reply Mem. Supp. Mot. J. Agency R. at 3. Consequently, Koyo’s claim regarding its PSPAs is dismissed.
(2) Koyo’s Warranty Expenses
Commerce adjusted for Koyo’s home market warranty expenses as direct expenses pursuant to the COS provision. Final Results, 60 Fed. Reg. at 10,910. Torrington challenges Commerce’s treatment of Koyo’s warranty expense, arguing Koyo did not provide product- or customer-spe*626cific warranty credits. Torrington’s Mem. Supp. Mot. J. Agency R. at 41-42.
Commerce agrees with Torrington and requests a remand to review the record to determine whether it is possible to remove those portions of Koyo’s warranty expenses which relate to non-scope merchandise from the adjustments to FMV or to deny the adjustment if such removal cannot be made. Def.’s Partial Opp’n to Mots. J. Agency R. at 22-23.
Koyo objects to a remand, claiming Commerce’s treatment of its warranty expenses was proper and is distinguishable from the situation involving its PSPAs. Koyo’s Opp’n to Torrington’s Mot. J. Agency R. at 8-17.
Upon inspection of the record, the Court agrees Commerce improperly included Koyo’s warranty expenses relating to non-scope merchandise in adjusting FMV See Koyo Section C Questionnaire Response, PR. Doc. No. 124, Def.’s App., Ex. 6, at 31-33 (Sept. 29, 1993). The Court therefore remands this issue to Commerce to review the record to determine whether it is possible to remove those portions of Koyo’s warranty expenses which relate to non-scope merchandise from the adjustments to FMV or to deny the adjustment if such removal cannot be made.
c. NTN’s Adjustments:
(1) Home Market Discounts
In the Final Results, Commerce treated NTN’s home market discounts as indirect selling expenses. 60 Fed. Reg. at 10,934. NTN asserts Commerce should have treated these discounts as direct selling expenses because it accounted for its discounts using proper product categories, and so, that there was no general pooling of discounts for each customer. NTN’s Mem. Supp. Mot. J. Agency R. at 44-46.
Commerce responds it treated the discounts as indirect selling expenses because they were not reported on a transaction-specific basis and because they were not fixed and constant expenses. Def.’s Partial Opp’n to Mots. J. Agency R. at 24-27. Torrington agrees with Commerce’s position. Torrington’s Opp’n to Mots. J. Agency R. at 86-89.
The Court concludes that Commerce improperly determined that the discounts at issue are indirect selling expenses. Rather, they are direct selling expenses because they represent expenses related to particular sales of in-scope merchandise. While they are not entitled to a direct adjustment because of NSK’s failure to tie them to specific transactions, this does not deprive them of their direct relationship to the sales under consideration. Therefore, pursuant to the CAFC’s decision in Torring-ton, this issue is remanded to Commerce to deny the adjustment to FMV for NTN’s home market discounts at issue.
(2) Billing Adjustments
In the Final Results, Commerce treated certain NTN billing adjustments as direct adjustments to price, stating it had “no reason to believe or suspect that NTN failed to report accurately or completely its [home market] billing adjustments, or that NTN’s method of reporting may *627have included billing adjustments made on sales of non-subject merchandise.” 60 Fed. Reg. at 10,934-35.
NTN claims its total sales value figures were net of billing adjustments, which should have been treated as direct expenses, and asserts Commerce should recalculate NTN’s margins based on sales prices net of billing adjustments. NTN’s Mem. Supp. Mot. J. Agency R. at 46-47.
Torrington alleges Commerce improperly treated certain NTN post-sale billing adjustments that had not been reported on a transaction-specific basis as direct adjustments to FMV Torrington’s Opp’n to Mots. J. Agency R. at 40.
Commerce responds that there was no evidence that NTN’s total sales value figures were reported net of billing adjustments, and so, it properly concluded not to make additional adjustments to the individual sales values for billing adjustments in its formula for calculating expenses for each individual sale. Def.’s Partial Opp’n to Mots. J. Agency R. at 24-26. However, Commerce agrees with Torrington and requests a remand to review the record to determine whether it is possible to distinguish between billing adjustments reported on a transaction-specific basis and those reported on a customer-specific or product-specific basis or, if a distinction cannot be made, to treat all NTN’s billing adjustments as indirect selling expenses. Id. at 27.
NTN maintains Commerce’s direct price adjustments for its billing adjustments was reasonable. NTN’s Opp’n to Torrington’s Mot. J. Agency R. at 11-12.
The Court concludes it is unclear whether NTN’s total sales value figures were reported net of billing adjustments because the evidence on the record does not establish the manner in which NTN calculated the total sales value figure. In essence, there is no specific evidence that all billing adjustments had been accounted for in NTN’s total sales value figure. Hence, Commerce properly declined to make additional adjustments.
With respect to Torrington’s claim, certain NTN billing adjustments were indeed not reported on a transaction-specific basis. See NTN Supplemental Section C Questionnaire Response, C.R. Doc. No. 95, Def.’s App., Ex. 28, at 11-12 (Dec. 8, 1993). Nevertheless, the relevant issue regarding direct adjustments is ultimately not whether they were reported on a transaction-specific basis, but whether a respondent can demonstrate that its adjustments were not made over non-subject merchandise. See Federal-Mogul Corp. v. United States, 20 CIT 1438, 1442-43, 950 F. Supp. 1179, 1184 — 85 (1996) (Commerce properly excluded billing adjustments not recorded on a transaction-specific basis where adjustments on non-scope merchandise could not be removed). Hence, this issue is remanded to Commerce to determine whether the NTN billing adjustments not reported on a transaction-specific basis were made solely over in-scope merchandise and, if so, to allow them a direct adjustment to FMV or, if such a determination cannot conclusively be made, to deny them a direct adjustment to FMV
*6283. COP Data of Inputs Obtained from Related Parties:
a. NSK’s CV Calculation:
During the review, Commerce used the COP data of inputs that NSK obtained from related parties to determine whether the inputs were sold at arm’s-length. In cases where Commerce concluded that they had not been sold at arm’s-length, Commerce used the COP of the inputs as the “best evidence available” for valuing the inputs. See Final Results, 60 Fed. Reg. at 10,924.
NSK contends Commerce is authorized to request COP data for inputs only when it has reason to believe or suspect that the inputs were sold below-COR which was not the case here. NSK further argues that, in this case, Commerce ignored evidence that the prices negotiated between NSK and its related supplier, Supplier X, fairly reflected market value, i.e., that its inputs were sold at arm’s length prices. NSK’s Mem. Supp. Mot. J. Agency R. at 34-43. In addition, NSK claims Commerce illegally coerced NSK to place COP information for finished bearings purchased from related suppliers on the record because it feared another application of best information available (“BIA”). Id. at 43.
Commerce responds that it was authorized to request NSK’s COP data pursuant to 19 U.S.C. § 1677b(e)(2) and the Court’s decision in NSK Ltd. v. United States, 19 CIT 1319,910 F. Supp. 663 (1995). Def.’s Partial Opp’n to Mots. J. Agency R. at 32-35. Commerce further maintains that, even if it is authorized to request COP data of inputs only when it has reason to believe or suspect that the inputs were sold at below-COR it had such grounds in this case due to its conclusions in the third review. Id. at 34-35 (citing Final Results, 60 Fed. Reg. at 10,924). Further, Commerce claims it is not required to consider subjective facts regarding the dealings between NSK and Supplier X before resorting to the COP of an input for that input’s market value. Id. at 37-38.
Commerce was authorized to request cost data to determine whether the transfer prices were at arm’s length pursuant to 19 U.S.C. § 1677b(e)(2). Pursuant to the relevant statutes, Commerce must determine whether transfer prices are at arm’s-length. See id. Commerce does this by comparing the transfer prices to: (1) the prices that the related suppliers charge to unrelated parties; and (2) the prices charged by unrelated suppliers to the respondent. If a transaction is disregardedbe-cause a respondent cannot establish that it was made at arm’s-length, and there are no other such arm’s-length prices to compare with the transfer prices for components, pursuant to 19 U.S.C. § 1677b(e)(2), Commerce is to rely on the best evidence available to determine the value of the element of value. See NSK, 19 CIT at 1325-26,910 F. Supp. at 669-70. In the third review period, where NSK also relied upon transfer prices, this Court concluded the following:
If based on the information considered, Commerce finds that transfer prices do not “fairly reflect the amount usually reflected in sales in the market under consideration, ” then Commerce may disregard these transactions in favor of the best evidence available if “there *629are no other transactions available for consideration.” * * * 19 U.S.C. § 1677b(e)(2) provides a basis for Commerce to request cost data about parts purchased from related suppliers as long as the respondents reported or relied on transfer prices. Thus, 19 U.S.C. § 1677b(e)(2) applies in the case at bar because NSK reported transfer prices in its * * * questionnaire response.
Id. at 669 (citations omitted). In NSK, therefore, the Court found Commerce properly requested COP data pursuant to 19 U.S.C. § 1677b(e)(2) since the respondent reported and relied upon transfer prices. Id. at 669-70. As in NSK, in this case NSK provided no information regarding prices from unrelated parties upon which to determine the market value for inputs it purchased from related suppliers but, rather, reported and relied upon transfer prices. See Final Results, 60 Fed. Reg. at 10,924. Hence, Commerce had the authority to request COP data under 19 U.S.C. § 1677b(e)(2).
It is further clear to this Court that, when there are no purchases of inputs from unrelated suppliers, Commerce may reasonably base the market value of inputs on the COP of inputs. In this case, Commerce used NSK’s COP data without addressing record evidence regarding whether NSK’s purchases from Supplier X were at market value, asserting in its brief that it “appropriately concluded that subjective facts regarding the dealing between NSK and supplier X were not relevant to its determination.” Def.’s Partial Opp’n to Mots. J. Agency R. at 38. Under the statute, Commerce is not required to examine the subjective information relating to the dealings between NSK and its related suppliers over evidence that prices between related parties were not at arm’s-length. Indeed, 19 U.S.C. § 1677b(e)(2) contemplates a comparison of the amount representing an element of value and the amount usually reflected in sales of the merchandise in the market. As Commerce explained in the Final Results, “[flacking information as to what the market value is, we rely on the related supplier’s cost as a measure of the commercial value of that input.” 60 Fed. Reg. at 10,924. The Court finds this process reasonable and in accordance with law.
Finally, NSK’s claim that it placed COP information on the record for finished bearings from related suppliers because it feared another application of BIA, and so, such data was “illegally coerced” is without merit. In NSK, 19 CIT at 1328, 910 F. Supp. at 671, the Court found Commerce improperly resorted to BIA for finished bearings purchased from related suppliers. Unlike the third review, Commerce did not resort to BIA in this case but, rather, used NSK’s COP information. See Final Results, 60 Fed. Reg. at 10,924. Despite NSK’s reasons for including this COP information on the record, Commerce may properly rely on such information once it is on the record.
Consequently, the Court concludes that Commerce properly requested and used COP data for inputs that NSK obtained from related suppliers.
*630b. Koyo’s CV Calculation:
Koyo claims Commerce lacked authority to request COP data of inputs Koyo obtained from related parties, and maintains Commerce may only request such data under § 1677b(e) (3) when Commerce has reason to believe or suspect that the inputs were sold below cost. Pursuant to the above reasoning regarding NSK’s inputs, and in accordance with NSK, 19 CIT at 1325-26, 910 E Supp. at 669-70, the Court concludes Commerce had the authority to request Koyo’s COP data at issue.
4. ESP Calculation for NSK Parts Further Manufactured in the U.S.:
During the review, Commerce calculated ESP pursuant to section 1677a(e)(3)5 for NSK’s AFB parts that were further manufactured in the U.S. prior to their sale to the first unrelated purchaser as completed AFBs. Section 1677a(e)(3) states that if an ESP transaction is involved and the imported merchandise has been further manufactured in the U.S. before the sale to the first unrelated purchaser, then ESP must be reduced by the increased value, including labor and materials, incurred by the additional U.S. manufacturing or assembly. See 19 U.S.C. § 1677a(e)(3).
NSK contends that, according to the legislative history, Section 1677a(e) (3) applies only to products covered by an antidumping duty order that are further manufactured in the U.S. into non-scope merchandise containing a significant amount of the in-scope imported product. NSK therefore maintains that Commerce should have based margins for imported parts on the margins calculated for finished bearings of the same class or kind. NSK’s Mem. Supp. Mot. J. Agency R. at 44-47.
In response, Commerce maintains its use of 19 U.S.C. § 1677a(e)(3) to calculate ESP for NSK’s further manufactured AFB parts is supported by the statutory language and the case law. Def. ’s Partial Opp’n to Mots. J. Agency R. at 41-48.
In support of its position, NSK points to the House of Representatives committee report on the section, which states the following:
This amendment provides that whenever * * * [there is an ESP situation], and the merchandise is changed by further process or manufacture so as to remove it from the class or kind of merchandise involved in the proceeding before it is sold to an unrelated purchaser, such merchandise will not escape the purview of the law * * *.
* * * [T]his amendment shall be applicable only if the manufactured or assembled product that is sold to an unrelated person contains more than an insignificant amount of the imported merchandise.
*631H.R. Rep. No. 571,93d Cong., 1st Sess. 70 (1973) (emphasis added); see also S. Rep. No. 1298, 93d Cong., 2d Sess. 172-73 (1974).6 NSK’s position, while somewhat supported by the language of the legislative history, is incorrect.
First, the Court is to use legislative history to ascertain Congress’s intent where a statute is ambiguous, but not to introduce ambiguity where statutory language is clear. See Terumpo Corp. v. United States, 8 CIT 44,46 (1984); see also Connecticut Nat’l Bank v. Germain, 503 U.S. 249,253-54 (“We have stated time and again that courts must presume that a legislature says in a statute what it means and means in a statute what it says there. When the words of a statute are unambiguous, then, this first canon is also the last: ‘judicial inquiry is complete. ’”) (Citations omitted). In this case, the statute is abundantly clear as to its intent: to reduce the price to the first unrelated purchaser in calculating ESP for an imported product that has been further processed. Hence, it is not necessary for the Court to refer to the legislative history.
Nevertheless, the legislative history does not directly address the situation at issue here, where the imported scope merchandise is further manufactured or assembled into scope merchandise. Rather, it sets forth Congress’ intent that a further-processed scope part used in non-scope merchandise cannot escape the application of the dumping law unless the imported scope part comprises an insignificant amount of the non-scope merchandise. Hence, the legislative history does not preclude the application of the section to situations where the ultimate product sold is in scope. Indeed, the employment of NSK’s interpretation of the statute would permit minor processing to exempt imports from an anti-dumping order, thus thwarting the purpose of the antidumping law.
Further support for the Court’s position is found in NSK’s challenge to the second review, where this Court concluded, and the CAFC recently affirmed, that Section 1677a(e)(3) covers imported parts utilized in the production of AFBs, and so, the parts are subject to a further-manufacturing analysis.7 See NSK, 19 CIT 1013, 1018-1022, 896 F. Supp. 1263, 1268-71 (1995), aff’d in part and rev’d in part, Slip Op. 96-1359, at 25-26 (Fed. Cir. June 10, 1997). See also NSK, 19 CIT at 1337, 910 F. Supp. at 677; Koyo Seiko Co. v. United States, 18 CIT 740, 747, 861 F. Supp. 108, 114-15 (1994) (imported AFB parts further as*632sembled in U.S. qualify as additional material and labor and require an adjustment for further processing under Section 1677a(e)(3)).
The Court therefore concludes that Commerce’s interpretation is reasonable and carries out the congressional intent of reducing the price to the first unrelated purchaser of a further processed import. Consequently, Commerce properly used 19 U.S.C. § 1677a(e)(3) to calculate ESP for NSK’s imported parts that were further-processed into completed AFBs in the U.S.
5. NSK’s Claimed Interest Income Offset to Interest Expense:
In calculating COP and CV during the review, Commerce denied NSK’s claimed interest income offset to interest expense and resorted to BIA. Final Results, 60 Fed. Reg. at 10,928-29. NSK claims that, in doing so, Commerce departed from its standard practice.
Commerce responds it resorted to BIA because NSK failed to explain how its reported interest income offset was related to its bearing manufacturing operations. Def.’s Partial Opp’n to Mots. J. Agency R. at 48-52. Commerce points to the following problems with NSK’s methodology, which it noted in the Final Results: (1) NSK applied a ratio of short-term versus long-term investments to total interest income; and (2) NSK failed to link short-term interest income to business operations. See id. (citing 60 Fed. Reg. at 10,929).
Torrington agrees that Commerce properly denied NSK’s interest income offset because NSK failed to substantiate its alleged short-term interest income. Torrington’s Opp’n to Mots. J. Agency R. at 37-43.
The Court concludes Commerce’s denial of NSK’s reported interest income offset to interest expense is proper. This Court has recognized Commerce’s practice of limiting interest income offset to interest expense for income items shown to have been earned from the general operations of the company in calculating COP and CV See NTN Bearing Corp. v. United States, 19 CIT 1221, 1236-37, 905 F. Supp. 1083, 1096-97 (1995); Timken v. United States, 18 CIT 1,9,852 F. Supp. 1040, 1048 (1994). During verification in this case, NSK provided a non-consolidated “detail of interest income.” See NSK COPICV Verification, C.R. Doc. No. 147, Def.’s App., Ex. 23. Although this document provided a general description of the types of short-term assets that had generated its income, it did not provide a list of the actual accounts or explain the nexus between the interest income and NSK’s bearing manufacturing operations. Consequently, in accordance with the court-approved methodology, Commerce properly denied NSK’s reported interest income offset and resorted to BIA.
6. Inclusion of NSK’s U.S. Zero-Priced Sample Transfers:
Commerce included NSK’s zero-priced sample transfers to U.S. customers in NSK’s U.S. sales database for the margin calculation. Final Results, 60 Fed. Reg. at 10,948. NSK contends the samples in question cannot be considered sales because there was no exchange of monetary consideration. NSK’s Mem. Supp. Mot. J. Agency R. at 50-51. In NSK, *633Ltd. v. United States, 20 CIT 350, 352, Slip Op. 96-50, at 6 (1996), this Court upheld Commerce’s inclusion of NSK’s zero-priced sample transfers because excluding such merchandise “creates a loophole in the anti-dumping law allowing the lowest priced U.S. transactions to escape review and threatens the effectiveness of the law.”
The CAFC recently addressed this issue and reversed the Court’s decision in NSK, concluding that NSK’s transfers of free samples to potential U.S. customers did not amount to sales. NSK, Slip Op. 96-1359, at 17-21. The CAFC reasoned that Congress intended to give the term “sale” its ordinary meaning, hence precluding NSK’s free samples from being treated as sales because they lacked consideration and because there was no evidence that the potential customers had any obligation regarding the samples but, rather, “were free to transact with NSK based solely on their whim.” Id. at 20.
In light of NSK, Slip Op. 96-1359, at 17-21, the Court has no alternative but to find that NSK’s free samples to U.S. customers in this case were not sales. This issue is therefore remanded to Commerce to exclude NSK’s zero-priced sample sales from NSK’s U.S. sales database.
7. Inclusion of NTN’s Home Market Sample and Similar Transfers:
NTN claims Commerce should have excluded NTN’s sample and other similar transfers from its home market database because they were made outside the ordinary course of trade. NTN’s Mem. Supp. Mot. J. Agency R. at 14-16.
Pursuant to the CAFC’s recent decision in NSK, Slip Op. 96-1359, at 17-21, sample transfers do not amount to sales under the dumping statute because there is typically no consideration involved. Similar to NSK’s samples in this review, there is no evidence that NTN’s sample transfers involved consideration in this case. Consequently, the Court concludes Commerce improperly included NTN’s sample and other similar transfers in NTN’s home market database and remands to Commerce to exclude them from the FMV calculation.
8. Koyo’s Non-Operating Expenses:
During the review, Commerce reclassified certain Koyo non-operating expenses and payments out of retained earnings as production expenses. Final Results, 60 Fed. Reg. at 10,926. Koyo claims this reclassification was improper because the expenses at issue were not related to domestic production. Koyo further contends Commerce’s reclassification is contrary to Commerce’s prior practice. Koyo’s Mem. Supp. Mot. J. Agency R. at 17-23.
Commerce agrees a remand is necessary to allow Koyo to submit documentation showing that the nature of the expenses Koyo characterized as non-operating is appropriate. Def.’s Partial Opp’n to Mots. J. Agency R. at 56-57.
Torrington maintains Commerce properly reclassified the expenses and claims Koyo’s management had ample opportunity to explain the *634various categories for expenses at verification. Torrington’s Opp’n to Mots. J. Agency R. at 53-56.
Upon review of the record, the Court agrees with Commerce that the record contains insufficient data upon which to determine the exact nature of the expenses at issue. Consequently, this issue is remanded to Commerce to reopen the record to allow Koyo to submit documentation showing the nature of the expenses Koyo characterized as non-operating expenses.
9. Treatment of NTN’s AFB Sales Sold at Different Levels of Trade:
a. Comparison of Sales Across Different Levels of Trade:
Under 19 U.S.C. § 1677b(a)(4)(B), Commerce must compare a U.S. price and FMV that are free of distortion caused by “other differences in circumstances of sale.” To account for one such distortion, differing levels of trade, Commerce normally calculates FMV and U.S. price based on sales at the same commercial level of trade. See 19 C.F.R. § 353.58.
NTN claims Commerce erred in comparing merchandise across levels of trade because it did not explain its decision to cross levels of trade and did not consider the commercial value of the merchandise sold at various levels of trade. NTN’s Mem. Supp. Mot. J. Agency R. at 16-18.
Commerce responds that it compared sales at different levels of trade because it failed to match “such or similar” merchandise at the same level. Commerce further maintains that it properly considered commercial value by applying its twenty percent cost test. Def. ’s Partial Opp ’n to Mots. J. Agency R. at 60-63.
Commerce may cross levels of trade when sales at the same level of trade are not available. See 19 C.F.R. § 353.58; see also NTN Bearing Corp. v. United States, 17 CIT 1149, 1153-55, 835 F. Supp. 646, 650 (1993). The Court concludes Commerce adequately explained its decision to cross levels of trade, as it used language virtually identical to that upheld in NTN, 19 CIT at 1230-31, 905 F. Supp. at 1092-93. Compare Final Results of Antidumping Duty Administrative Reviews and Revocation in Part of an Antidumping Duty Order, 58 Fed. Reg. 39,729, 39,767-68 (July 26,1993) {“AFBs III”) with Final Results, 60 Fed. Reg. at 10,940. In particular, Commerce noted the following: “[W]hen we were unable to compare NTN’s U.S. sales to [home market] sales at the same level of trade, we attempted to find matches at the next most similar level of trade.” Final Results, 60 Fed. Reg. at 10,940.
The Court further concludes Commerce properly considered commercial value in selecting sales for comparison. In matching U.S. and home market models in this case, Commerce applied its twenty percent cost test, using the CV of the U.S. model as the basis for FMV where “the difference between the variable manufacturing costs of the U.S. and home market models exceeded 20 percent of the total manufacturing cost of the U.S. model.” NTN Preliminary Analysis Memorandum, PR. Doc. No. 309, Fiche 135, Frame 33 (Feb. 24,1994). As this Court determined in NTN, Commerce’s twenty percent cost test appropriately con*635siders commercial value for matching purposes. See NTN, 19 CIT at 1231, 905 F. Supp. at 1092.
b. Denial of Level of Trade Adjustment:
When Commerce compares sales at differing levels of trade, it is authorized to make a COS adjustment to FMV See 19 U.S.C. § 1677b(a)(4)(B). Pursuant to the implementing regulation, 19 C.F.R. § 353.56(a)(1), Commerce may make such an adjustment for a bona fide difference in the circumstances of the sales compared. To ensure respondents do not have an incentive to destroy or fail to produce information supporting or rebutting their own assertions, the party claiming such an adjustment has the burden of substantiating its claim. Timken Co. v. United States, 11 CIT 786, 804, 673 F. Supp. 495, 513 (1987). In this case, NTN produced several documents allegedly supporting its claim for a level of trade adjustment. Nevertheless, Commerce concluded that “NTN * * * failed to demonstrate what portion, if any, of those price differences [between merchandise sold at different levels of trade] is attributable to differences in levels of trade,” as opposed to other factors. Final Results, 60 Fed. Reg. 10,940.
NTN argues Commerce erred in denying NTN a level of trade adjustment to FMV to compensate for the different prices at each level of trade and requests a remand for Commerce to explain its conclusion. NTN’s Mem. Supp. Mot. J. Agency R. at 18-23.
Commerce responds that the information NTN provided failed to adequately quantify the portion of the price differences attributable to differences in the levels of trade, and so, NTN did not demonstrate that it was entitled to a level of trade adjustment. Def.’s Partial Opp’n to Mots. J. Agency R. at 63-69. Torrington agrees generally with the position taken by Commerce. Torrington’s Opp’n to Mots. J. Agency R. at 67-69.
Despite Commerce’s somewhat conclusory statement in the Final Results, upon inspection of the record, it is clear to this Court that NTN failed to substantiate its claim for a level of trade adjustment to FMV As Commerce and Torrington point out in their briefs, NTN could have shown that the price differentials it indicated were due to selling expenses attributable directly to differences in the levels of trade, as opposed to differences in, for instance, quantities sold, arbitrary pricing practices, credit costs, freight, packing or other factors. See Def.’s Partial Opp’n to Mots. J. Agency R. at 66; Torrington’s Opp’n to Mots. J. Agency R. at 68. The information NTN submitted, however, merely indicated variances in prices and selling expenses at the different levels of trade, without illustrating the factors to which they were attributable. See, e.g., NTN’s Section A Questionnaire Response, C.R. Doc. No. 5, Def.’s App., Ex. 21 (Aug. 10,1993) (reporting percentage differences in prices depending on the level of trade at which merchandise was sold).
NTN argues that Commerce has not indicated what proof would suffice for a level of trade adjustment. However, in American Permac, Inc. v. United States, 12 CIT 1134, 1138, 703 F. Supp. 97, 101 (1988), upon *636which NTN relies, the Court held that respondent’s detailed sales information regarding U.S. sales and a third market sale, as well as a detailed accounting study of expenses for home market sales where such information did not exist, was enough to support the respondent’s claim. As indicated above, NTN did not provide such necessary information in this case. See also NTN, 19 CIT at 1231-33,905 F. Supp. at 1093-94 (denying level of trade adjustment claim based on similar information).
Consequently, Commerce’s decision to deny NTN’s claimed level of trade adjustment was supported by evidence on the record and fully in accordance with law.
c. Reallocation of NTN’s Selling Expenses Without Regard to Level of Trade and Denial of a Level of Trade Adjustment Based on Indirect Selling Expenses:
NTN argues Commerce improperly denied NTN’s request for a level of trade adjustment based on differences in selling expenses despite evidence demonstrating that sales made at different levels of trade incurred different selling expenses. NTN acknowledges that the Court decided this same issue in a challenge to the third review, see NTN, 19 CIT at 1234-35,905 F. Supp. at 1094-95, but maintains the facts on this record provide greater support for its position. NTN’s Mem. Supp. Mot. J. Agency R. at 21-23.
After a thorough review of the record, the Court concludes that the facts in this case are substantially the same as those in the third review. The evidence on the record does not establish differences in selling expenses at the three levels of trade because NTN’s allocation methodology again does not reasonably quantify the expenses incurred at each level of trade. Commerce explained the following in the Final Results:
[NTN’s and NTN-Germany’s] expenses are fixed period costs that do not vary according to sales value or the number of employees who allegedly sell each type of merchandise. Further, * * * NTN’s and NTN-Germany’s allocations according to levels of trade [are] misplaced because the types of expenses that they allocated are indirect selling expenses that typically relate to all sales.
60 Fed. Reg. at 10,940. Hence, the Court finds no reason to depart from its decision in NTN, 19 CIT at 1234-35, 905 F. Supp. at 1094-95. Commerce properly denied NTN’s request for a level of trade adjustment based on differences in selling prices at the three levels of trade.
10. Exclusion of Certain NTN Home Market Sales to Related Parties:
Under the relevant statute, Commerce may base FMV on the price paid by a related party. See 19 U.S.C. § 1677b(a)(3). However, Commerce usually excludes related party sales unless a respondent demonstrates to Commerce’s satisfaction that a related party price is an arm’s-length price, i.e., that the related party prices are “comparable” to unrelated party prices. See NTN, 19 CIT at 1239-40,905 F. Supp. at 1099 (citing 19 C.F.R. § 353.45(a)); see also NEC Home Elees., Ltd. v. United States, 18 *637CIT 336, 338-39 (1994), aff’d in part and rev’d in part, 54 F.3d 736, 744 (Fed. Cir. 1995).
In this case, NTN reported both related and unrelated sales, admitting that prices for two classes or kinds of related sales, ball and cylindrical roller bearings, were lower than unrelated party prices while related party prices of spherical roller bearings were actually higher than unrelated party prices. See Final Results, 60 Fed. Reg. at 10,946-47. Commerce developed a weighted-average percent difference in pricing for each class or kind of merchandise and discarded sales to related parties where the related party average price was lower than unrelated party price. See id.
NTN challenges Commerce’s methodology, arguing that Commerce failed to provide an objective basis for determining when prices are comparable and that Commerce should have used other factors in determining whether to use sales to related parties when calculating FMV NTN further contends, with the aid of a hypothetical example, that Commerce should not have used weighted-average prices in determining comparability. NTN’s Mem. Supp. Mot. J. Agency R. at 23-26.
Under the applicable statute, Commerce is allowed considerable discretion in deciding whether to include related party sales when calculating FMV Usinor Sacilor v. United States, 18 CIT 1155, 1158, 872 F. Supp. 1000,1004 (1994). For instance, in NTN, 19 CIT at 1241, 905 F. Supp. at 1100, this Court upheld Commerce’s arms’-length test, emphasizing that respondents failed to present “record evidence tending to show that * * * Commerce’s test was unreasonable.” In this case, NTN has merely restated its position and has not presented any evidence that would require a different conclusion.
This Court has also rejected the contention that Commerce should consider other factors (i.e., factors other than price) in determining comparability and finds no reason to depart from its stated position. See NTN, 19 CIT at 1241-42,905 F. Supp. at 1100. In addition, NTN’s hypothetical example supporting its claim that Commerce should not have used weighted-average prices in determining comparability fails to prove that Commerce’s test is unreasonable, as it does not constitute record evidence demonstrating that its related party prices were comparable to its unrelated party prices.
The Court therefore concludes that Commerce’s methodology, including its decision to use weighted averages, is within its discretion and Commerce need not consider other factors. See id.
11. NTN’s Imputed Interest Expense Allocation:
NTN claims that Commerce improperly denied NTN’s claimed offset for interest expenses incurred in financing cash deposits of estimated antidumping duties. NTN’s Mem. Supp. Mot. J. Agency R. at 26-30. Commerce consents to a remand to conform this issue with the remand redetermination pursuant to Federal-Mogul Corp. v. United States, 20 *638CIT 234, 265, 918 F. Supp. 386, 412-13 (1996). Def.’s Partial Opp’n to Mots. J. Agency R. at 79-80.
Torrington, however, maintains that Commerce properly rejected NTN’s claimed adjustment. Torrington’s Opp’n to Mots. J. Agency R. at 74-77.
In the remand results to the third administrative review, Commerce deterníined, and the Court agreed, that it had properly allowed NTN’s imputed interest expense adjustment because the expense could not be categorized as a selling expense. See Federal-Mogul, 20 CIT at 1440-41, 950 F. Supp. at 1182-83. In light of Federal-Mogul, this issue is remanded to Commerce to allow NTN’s adjustment for interest expenses on an-tidumping duty cash deposits.
12. Reallocation of NTN’s U.S. Selling Expenses Based on the Sale Price to the First Unrelated Purchaser:
In prior reviews, Commerce accepted NTN’s reported U.S. selling expenses allocated on the basis of the transfer prices between NTN-Germany and NTN Bearing Corporation of America. See, e.g., AFBs III, 58 Fed. Reg. at 39,749 (third review final results). Nevertheless, in this review, Commerce chose to reallocate the expenses using the resale prices to the first unrelated purchasers. See Final Results, 60 Fed. Reg. at 10,919.
NTN claims Commerce inexplicably deviated from its previous methodology in reallocating the expenses, contrary to the principles articulated in Shikoku Chems. Corp. v. United States, 16 CIT 382, 795 E Supp. 417 (1992). NTN further alleges Commerce’s reallocation yields less accurate results than the previous methodology. NTN’s Mem. Supp. Mot. J. Agency R. at 30-31.
To sustain Commerce’s choice of methodology, the Court must determine whether the methodology is reasonable, and not whether it is the only reasonable one, or even the one that the Court finds most reasonable. See Cerámica Regiomontana, S.A. v. United States, 10 CIT 399, 404-05, 636 E Supp. 961, 966 (1986), aff’d, 810 F.2d 1137 (Fed. Cir. 1987); see also NTN, 19 CIT at 1241, 905 E Supp. at 1100 (stating the Court will uphold the test that Commerce selects to measure whether sales to related parties were made at arms’-length unless that test is shown to be unreasonable). In this case, Commerce allocated expenses using resale prices to unrelated parties, noting that such a methodology provides a value that is not subject to potential manipulation by respondents. See Final Results, 60 Fed. Reg.' at 10,919. Based on this rationale, and despite the fact that Commerce had previously used a different methodology, the Court concludes that Commerce’s new methodology is reasonable.
First, the present situation is distinguishable from Shikoku. The Shikoku court decided that the methodology Commerce used in the investigation and four administrative reviews had become the law of the proceeding, and so, Commerce could not deviate from that methodology. *63916 CIT at 387, 795 F. Supp. at 422. The court emphasized record evidence that plaintiffs adjusted their prices in accordance with the methodology Commerce had consistently applied in the investigation and previous four reviews. Id. As this Court recently stated in rejecting a similar argument from NTN, “[importantly, in Shikoku, plaintiffs established their reliance on Commerce’s previous methodology consistently applied in several reviews.” NTN, 19 CIT at 1234,905 F. Supp. at 1095; see also Koyo Seiko Co. v. United States, 20 CIT 920, 924, 936 F. Supp. 1040, 1044 (1996).
The Court is disturbed that Commerce would change its consistent methodology in this review, especially after rejecting, as mere speculation, Torrington’s argument that the prices were subject to manipulation. See 60 Fed. Reg. at 10,919. Nevertheless, Commerce’s methodology is reasonable and NTN has not demonstrated any detrimental reliance, such as that in Shikoku, on Commerce’s previous methodology. Further, NTN failed in any way to support its contention that Commerce’s reallocation yields less accurate results than the previous methodology.
Consequently, Commerce’s decision in this case to reallocate NTN’s selling expenses based on the sales price to the first unrelated purchaser is reasonable.
13. Adjustments to NTN’s COP and CV:
During the review, Commerce submitted to NTN a supplemental questionnaire requesting NTN to provide, among other things, a detailed description and an example of the sample calculations presented in NTN’s original questionnaire exhibit for standard costs. At verification, NTN presented a business proprietary worksheet listing actual costs by product line in one of its plants that Commerce subsequently determined was inconsistent with NTN’s prior representations. According to Commerce, the worksheet demonstrated that NTN had overstated the standard costs of non-scope merchandise, and so, had understated the standard costs of scope merchandise. Commerce therefore adjusted NTN’s reported COP and CV See Final Results, 60 Fed. Reg. at 10,928-29.
NTN claims the adjustment was not warranted for several reasons: (1) the basis for the adjustment was a worksheet that was not a regularly maintained business record and that concerned actual costs prior to the review for one plant, with an insignificant portion of the plant’s production forming the basis of the adjustment; (2) Commerce drew erroneous conclusions about revisions in standard costs for subject and non-subject merchandise; and (3) Commerce did not have a legal basis to resort to BIA. See NTN’s Mem. Supp. Mot. J. Agency R. at 32-40. NTN further alleges that, assuming the adjustment was warranted, Commerce’s methodology for calculating the adjustment is less accurate than a methodology that could have been employed. Id. at 42-43.
*640Commerce responds that all of NTN’s claims are without merit. While Commerce admits that the worksheet in question was not a regular business document, it insists that NTN did not offer Commerce such a document that could have been used to test NTN’s standard costs. Further, while the worksheet information was based on costs prior to the period of review, Commerce notes that standard costs are based on historical information that is occasionally updated. In addition, Commerce states that, although it inadvertently typed an incorrect date on the worksheet, it did not misunderstand the temporal origin of the information used to test the standard costs. Commerce further maintains that it did not resort to BIA but, rather, used the information contained in the record to calculate the most accurate dumping margin possible. Finally, Commerce argues that it is not required to address NTN’s proposed methodology, as Commerce has been granted discretion to devise methodology for the administration of the antidumping law. Def.’s Partial Opp’n to Mots. J. Agency R. at 84-94.
Torrington agrees generally with the position taken by Commerce, but contends Commerce properly resorted to BIA because NTN failed to provide accurate cost information. Torrington’s Opp’n to Mots. J. Agency R. at 81-86.
As a preliminary matter, while the worksheet NTN submitted to Commerce was not a regularly maintained business document, NTN did not offer such a document in its response to the supplemental questionnaire. Hence, Commerce properly utilized the document. Further, as verification is a spot check, Commerce was justified in conducting its COP and CV verification at only one of NTN’s plants. See, e.g., Monsanto Co. v. United States, 12 CIT 937, 944, 698 F. Supp. 275, 281 (1988) (“Verification is a spot check and is not intended to be an exhaustive examination of the respondent’s business. ITA has considerable latitude in picking and choosing which items it will examine in detail.”).
To determine whether Commerce acted properly in adjusting NTN’s COP and CV the Court must first decide whether Commerce’s adjustment amounted to the use of BIA and, if so, whether Commerce was justified in resorting to BIA. Using NTN’s verification worksheet, Commerce concluded that NTN’s non-subject standard costs were overstated when compared to actual costs because NTN’s application of a non-product-specific plant-wide variance to all products shifted costs between products. See Final Results, 60 Fed. Reg. at 10,928. Hence, Commerce increased COP and CV (i.e., it made an across the board adjustment to all of NTN’s costs at all plants) to account for the difference it found between the application of the variance to standard costs of a sample non-subject product and the actual costs for the product line.
Commerce claims that it did not resort to BIA in making this adjustment but, rather, utilized information on the record to “correct,” as opposed to reject, NTN’s data. Nevertheless, Commerce made the adjustments only after determining that NTN’s reported information was inaccurate by relying on pre-period of review actual cost informa*641tion from one plant that it believed demonstrated NTN had understated its COP and CV Hence, Commerce’s adjustments were not merely the correction of a computation or other such error but, rather, clearly amounted to the use of the best information Commerce believed it had available to determine the proper dumping margins after disregarding NTN’s reported data.
The relevant statute requires Commerce to resort to BIA when it cannot verify the accuracy of a respondent’s data, when a respondent refuses or is unable to provide information Commerce requests in a timely manner or in the form required, or when a respondent otherwise significantly impedes an investigation. See 19 U.S.C. §§ 1677e(b) & (c) (1994). In this case, NTN was able to, and did, fully comply with Commerce’s requests for information. While Torrington asserts Commerce was unable to verify the accuracy of the information NTN submitted, the record is devoid of any such allegation by Commerce. Rather, it is clear that Commerce based its action on its disagreement with the method NTN used to calculate its standard costs, and not the accuracy of the underlying data NTN had provided.
This situation is similar to that in NTN Bearing Corp. v. United States, 17 CIT 713, 718-20, 826 F. Supp. 1435, 1140-41 (1993), where Commerce made an adjustment based on information from one plant and applied it to all COP and CV for all merchandise. The NTN Court concluded Commerce improperly resorted to BIA, noting Commerce “did not disregard all the information provided by NTN, nor did it question the accuracy of the remaining data. Instead, it questioned the possible distorting effect on overall accuracy the * * * data might have. ” Id. at 718-19, 826 F. Supp. at 1140.
Consequently, in this case, Commerce had no grounds to resort to BIA and Commerce’s attempt to mask its use of BIA by explaining that it used the worksheet information to correct NTN’s reported COP and CV is an improper post hoc rationalization for adjusting the costs. See, e.g., NTN Bearing Corp. v. United States, 74F.3d 1204,1208 (Fed. Cir. 1995) (stating respondent’s litigation-driven argument came too late for serious consideration).8 Because Commerce’s decision to resort to BIA was unwarranted and improper, this issue is remanded for Commerce to recalculate NTN’s COP and CV without resorting to BIA. Having decided that Commerce improperly resorted to BIA, it is not necessary to determine whether Commerce’s methodology was reasonable.
*64214. Resorting to BIA When NPB Failed to Report Negative Billing Adjustments:
In response to Commerce’s questionnaire requests for NPB’s billing adjustments, NPB first stated that it had no such adjustments during the review period, then responded that it did not report its adjustments because of their small magnitude and NPB’s inability to associate them with specific sales transactions, i.e., NPB could not trace the transactions in an invoice-specific manner. See NPB’s Questionnaire Response, PR. Doc. No. 102, Def.’s App., Ex. 2, at 12 (Sept. 21,1993); NPB’s Supplemental Questionnaire Response, PR. Doc. No. 207, Def.’s App., Ex. 10, at 7 (Dec. 1,1993). Upon verification, Commerce discovered numerous errors in NPB’s responses, including NPB’s failure to report home market billing adjustments. See Final Results, 60 Fed. Reg. at 10,908. Commerce therefore corrected those errors for which it had sufficient information and applied BIA to those errors for which it had insufficient or missing information. See id.
With respect to NPB’s unreported billing adjustments, Commerce examined NPB’s home market sales information at verification and found that, when the billing adjustments were included, the weighted-average price of certain products increased for some products and decreased for other products. See NPB Verification Report, PR. Doc. No. 333, Def.’s App., Ex. 17, at 4 (March 1, 1994). The only way to determine the true billing adjustment effect on FMV would be to include them, which was not possible because NPB failed to report them. Hence, because of the potential this result had for masking dumping margins, instead of omitting billing adjustments Commerce resorted to BIA, drawing an inference that the unreported information was adverse to NPB.
NPB claims Commerce improperly changed its position from the preceding review by applying BIA to account for NPB’s unreported negative billing adjustments. NPB further alleges Commerce, in resorting to BIA, treated other respondents with similar responses differently than NPB in this review. Finally, NPB contends that Commerce should have disregarded the negative billing adjustments because they were insignificant. NPB’s Mem. Supp. Mot. J. Agency R. at 4-17.
Commerce responds that its use of BIA was justified because NPB failed to report the billing adjustment information requested in the questionnaires. Further, Commerce maintains there were compelling reasons not to resort to BIA for other respondents. Finally, Commerce notes that, although it may disregard insignificant adjustments, there is no indication what effect NPB’s unreported billing adjustments would have and, further, NPB’s own report at verification indicates the effect on certain AFB models may be above the regulatorily defined level for insignificance. Def. ’s Partial Opp’n to Mots. J. Agency R. at 94-102. Tor-rington agrees generally with Commerce’s position. Torrington’s Opp’n to Mots. J. Agency R. at 91-96.
*643As a preliminary matter, Commerce may alter its policies depending on a change in circumstances in different reviews. See, e.g., Allied Aerospace Co. v. United States, 28 F.3d 1188,1191 (Fed. Cir. 1994), cert, denied, 513 U.S. 1077 (1995) (stating the particular methodology Commerce employed in previous investigations “did not rise to the level of an agency regulation having the force and effect of law”). Although NPB claims Commerce disregarded billing adjustments in a previous review, Commerce found that disregarding the missing billing adjustments in this case might disguise dumping and reward NPB for failing to report the adjustments. Indeed, Commerce did not change its reporting requirements and NPB admits that it “chose not to report negative home market billing adjustments.”9 See NPB’s Mem. Supp. Mot. J. Agency R. at 4 (emphasis added). Commerce was therefore justified in changing its policy and deciding not to disregard NPB’s adjustments.
Further, this case is distinguishable from Shikoku, 16 CIT at 387-88, 795 F. Supp. at 421-22, where Commerce had an established practice over several reviews. In this case, Commerce’s disregard of reported negative billing information was not an established practice NPB could rely upon in deciding not to comply with Commerce’s questionnaire requests. Also, NPB undeniably knew it was requested, if not expected, to report billing adjustments.
Despite NPB’s contention that its billing adjustments would only decrease FMV and, therefore, would be only detrimental to NPB, Commerce’s weighted-average results at verification clearly indicated otherwise. The Court agrees that Commerce’s verification of the respondents that NPB refers to indicated that disregarding the unreported billing adjustments was detrimental to the respondents, as the adjustments solely represented decreases to FMV
In addition, NPB admits it failed to provide the negative billing information Commerce requested in its questionnaire. In circumstances such as this case, where a respondent refuses to produce information requested in a timely manner, Commerce is to resort to BIA. See 19 U.S.C. § 1677e(c).
Finally, under 19 C.F.R. § 353.59(a) (1994), it is within Commerce’s discretion to determine whether an adjustment is insignificant and should, therefore, be disregarded. In this case, the Court agrees that the effect that NPB’s unreported billing adjustments would have is at least unclear, as they were not reported.
Consequently, the Court concludes Commerce properly resorted to BIA when NPB failed to report negative billing adjustments.
*64415. Treatment of Honda’s Imports:
During the period of review, Honda purchased AFBs from various Japanese producers, some of which it imported to two subsidiaries in the U.S. In its Final Results, Commerce classified Honda as a reseller after concluding that Honda’s suppliers were unaware of the ultimate destination of the merchandise they sold to Honda. 60 Fed. Reg. at 10,951-52. The effect of treating Honda as a reseller was that its transactions with home market suppliers were not treated as the purchase price base of U.S. price for the AFBs Honda exported to the U.S. Rather, these transactions were used in the home market sales database and U.S. price was based on the price Honda charged to its purchasers. Commerce further decided to exclude Honda’s AFB imports to its subsidiaries based on the Roller Chain rule. Id. at 10,938; see supra at n.6 (explanation of Roller Chain rule).
Torrington claims Commerce erred in its treatment of AFBs imported by Honda, arguing that Commerce should not have treated Honda as a reseller and should not have applied its Roller Chain rule to Honda’s imports. First, Torrington alleges Commerce improperly required that Honda’s suppliers “had reason to know” whether each transaction of AFB sales to Honda was to be exported to Honda’s U.S. subsidiaries. According to Torrington, it is sufficient under 19 U.S.C. § 1677a(b) if the sale was “for exportation,” whether or not the foreign producer knew or should have known the ultimate destination of the merchandise. However, because of Commerce’s statutorily unjustified high standard of specific knowledge for each transaction, Torrington maintains that Commerce improperly concluded that Honda was a reseller and based purchase price on the prices Honda charged. Torrington’s Mem. Supp. Mot. J. Agency R. at 26-39.
Torrington adds that, based on the information in the record, it is evident that Honda’s suppliers “should have known” that some or all of their sales to Honda were not destined for the home market. In support of this claim, Torrington points to Honda’s substantial manufacturing capacity in the U.S., as reported in its Annual Report, and to the nature of Honda’s suppliers, who had ample knowledge of Honda’s status as a reseller of their parts. Id. at 29-30.
Second, Torrington claims Commerce abused its discretion by applying the Roller Chain rule to exclude sales based on a comparison of entered values with resale prices. In doing so, Torrington argues Commerce relied upon unreliable inter-company prices and compared values at different points in the chain of commerce. Id. at 35-38.
Commerce responds that its knowledge test is fully supported by the legislative history to the statute, as well as case law. Commerce further stands by its decision to exclude Honda’s imports to its subsidiaries under the Roller Chain rule. Def.’s Partial Opp’n to Mots. J. Agency R. at 102-11. Honda and NTN agree generally with the positions taken by Commerce. Honda’s Opp’n to Torrington’s Mot. J. Agency R. at 1-32; *645NTN’s Opp’n to Torrington’s Mot. J. Agency R. at 8-10. Honda emphasizes that the record clearly indicates its suppliers did not know or have reason to know that particular AFB sales to Honda would ultimately end up in the U.S. Honda’s Opp’n to Torrington’s Mot. J. Agency R. at 17-20.
a. Honda as a Reseller:
Under 19 C.F.R. § 353.2(s) (1994), areselleris “anyperson (otherthan the producer) whose sales [Commerce] uses to calculate foreign market value or U.S. price, including the foreign reseller or exporter.” The Court concludes that Commerce’s determination that Honda is a reseller is supported by substantial evidence. Section 1677a(b) states purchase price is “the price at which merchandise is purchased, or agreed to be purchased, prior to the date of importation, from a reseller or the manufacturer or producer of the merchandise for exportation to the United States.” 19 U.S.C. § 1677a(b) (emphasis added). The legislative history to this section clearly demonstrates that Commerce’s knowledge test was anticipated by Congress and is a reasonable interpretation of the statute.
In enacting a new antidumping law as part of the Trade Agreements Act of 1979, Congress modified the definition of purchase price, hence establishing the basis for Commerce’s administrative practice of looking to a producer’s knowledge in determining whether to use the producer’s sales price as the purchase price. Congress stated the following:
If a producer knew that the merchandise was intended for sale to an unrelated purchaser in the United States under terms of sale fixed on or before the date of importation, the producer’s sale price to an unrelated middleman will be used as the purchase price.
S. Rep. No. 249, 96th Cong., 1st Sess. 94 (1979) (emphasis added); see also H.R. No. 153, 96th Cong., 1st Sess. 411 (1979) (“The definition makes clear that if the producer knew or had reason to know the goods were for sale to an unrelated U.S. buyer, * * * the producer’s sales price will be used as ‘purchase price’ * * *.”) (Emphasis added). Further, in 1984 Congress explicitly amended Section 1677a(b) to recognize that a reseller’s price may be used as purchase price. See H.R. Conf. Rep. No. 1156, 98th Cong., 2d Sess. 185 (1984), reprinted in U.S.C.A.A.N. 5220, 5302 (adding “a reseller or” to 19 U.S.C. § 1677a(b)). In addition, this Court has previously upheld Commerce’s application of the knowledge test. See Peer Bearing Co. v. United States, 16 CIT 799, 803-04, 800 F. Supp. 959, 963-64 (1992).
Based on this legislative history and case law precedent, the Court is convinced that Commerce’s decision to apply, and its application of, the knowledge test, is reasonable. The Court recognizes that Commerce’s application of such a high standard is exploitable by the “perfect” scenario, where a reseller ostensibly “hides” the ultimate destination of its purchases from its foreign suppliers by the manipulation of information it chooses to provide the suppliers. Nevertheless, the Court finds that *646such a standard is necessary to fulfill the statutory intent that purchase price be based on sales of goods sold abroad with the intent of being exported to the U.S.
In this case, upon careful examination of the record, the Court agrees with Commerce’s determination that Honda’s suppliers were unaware of the ultimate destination of Honda’s specific AFB purchases. While it is unlikely that Honda’s suppliers had no knowledge that certain AFBs sold to Honda would ultimately be shipped to the U.S., the suppliers must have knowledge that particular sales are destined for import into the U.S. There is no evidence on the record to suggest such knowledge. Indeed, the record indicates the opposite, especially since all parts sold in the U.S. during the period of review were also at least offered for sale in the home market. See Honda HM Verification Report, C.R. Doc. No. 164, Honda’s App., Ex. 10, at 4 (March 4, 1994).
Consequently, the Court concludes that Commerce’s decision that Honda is a reseller is supported by substantial evidence on the record and is fully in accordance with law.
b. Roller Chain Rule Application:
Commerce excluded Honda’s imports to its subsidiaries under the Roller Chain rule. See Final Results, 60 Fed. Reg. at 10,938. Torrington asserts this exclusion was improper because Commerce relied on unreliable inter-company prices. In particular, Torrington claims Commerce compared actual entered values to the resale prices of the finished products. Torrington’s Mem. Supp. Mot. J. Agency R. at 35-38.
Commerce maintains that its use of actual entered values should be upheld because it is even more reasonable than the use of weighted averages, which the Court has found reasonable. Def.’s Partial Opp’n to Mots. J. Agency R. at 107-11.
First, as this Court stated above with respect to the Roller Chain rule, when an imported scope item is further processed and constitutes an insignificant amount of a complete non-scope final product sold to the first unrelated customer in the U.S., Commerce excludes the insignificant import from the coverage of review. See NSK, 19 CIT at 1019-21, 896 F. Supp. at 1269-70. In this case, Torrington does not dispute that the record clearly demonstrates that Honda’s imports at issue fall within the Roller Chain threshold of one percent, but objects to Commerce’s use of actual entered values.
The Court finds Commerce’s use of actual entered values reasonable. First, while Congress expressed its intent to employ the Roller Chain rule, it did not provide guidance as to how “insignificant” merchandise should be measured and, therefore, deference should be given to Commerce’s practice. Further, in Federal-Mogul, 20 CIT at 263,918 F. Supp. at 411, the Court approved the use of weighted-average entered values. Consequently, it follows that the actual entered values may also properly be used to determine whether an imported product satisfied the one percent threshold of the Roller Chain rule.
*647The Court therefore concludes that Commerce’s use of actual entered values to determine whether the imported product satisfied the one percent threshold of the Roller Chain rule, and Commerce’s conclusion that the merchandise at issue satisfied this threshold, are supported by substantial evidence.
16. Koyo’s Allocation of Air Freight Expenses:
Torrington claims Commerce erred in accepting Koyo’s reported air freight expenses because, contrary to Commerce’s policy, they were not reported on a transaction-specific basis and were commingled with ocean freight expenses. Torrington’s Mem. Supp. Mot. J. Agency R. at 45-48.
Commerce agrees that it improperly accepted Koyo’s allocation for air freight expenses and consents to a remand to examine its acceptance of the allocation and, if necessary, to request additional relevant information from Koyo. In particular, Commerce is concerned with the inconsistency between Koyo’s supplemental questionnaire response that air freight charges are not incurred as part of specific sales agreements with particular customers and Commerce’s finding at verification that, at least in one instance, Koyo was able to tie a specific air shipment to a specific customer. Def.’s Partial Opp’n to Mots. J. Agency R. at 111-14 (<comparing Koyo Supplemental Questionnaire Response, ER. Doc. No. 215, Def.’s App., Ex. 11, at 9-11 (Dec. 6,1993) with Koyo Sales Verification Report, PR. Doc. No. 286, Def.’s App., Ex. 14, at 8 (Jan. 13,1994)).
Koyo maintains it was incapable of tracing its air freight expenses on either a sale-specific or customer-specific basis. Koyo’s Opp’n to Mot. J. Agency R. at 17-22.
Upon review of the record, the Court agrees with Commerce that the inconsistency discovered at verification warrants a remand for Commerce to examine its acceptance of Koyo’s allocation of air freight expenses and, if necessary, to request additional relevant information from Koyo.
17. Koyo’s Efficiency Variance:
Commerce’s practice has been to base its cost calculations upon the cost records of the respondent when they are maintained in accordance with the home market’s generally accepted accounting principles (“GAAP”). Commerce will, nevertheless, adjust arespondent’s reported cost data if it finds that it does not reasonably reflect the cost of producing the merchandise. See, e.g., Thai Pineapple Pub. Co. v. United States, 20 CIT 1312,1320-21,946 F. Supp. 11,20 (1996) (“Commerce has never stated that foreign GAAP will suffice when it is unreliable. To the contrary, GAAP consistent methodologies are rejected when they do not reflect actual costs.”).
In the preliminary results, Commerce determined that the variance had a direct effect on the specific product costs and lowered certain factory overhead costs allocated to the subject merchandise. Commerce therefore concluded that, to recapture these costs, an upward adjust*648ment was necessary to Koyo’s fabrication costs. See Koyo COPICVPre-liminary Analysis Memorandum, HR. Doc. No. 322, Def.’s App., Ex. 16, at 1-2 (Feb. 25,1994). In the Final Results, however, Commerce accepted Koyo’s efficiency adjustment, concluding that it more accurately determined the amount of labor costs associated with individual cost centers. See 60 Fed. Reg. at 10,927.
Torrington contends Commerce erred in accepting Koyo’s efficiency variance used in the calculation of COP since, as Commerce concluded at verification, it distorted the estimated cost of the individual bearing models. Torrington therefore requests a remand for Commerce to recalculate Koyo’s margin by eliminating the effect of the questionable variance consistent with Commerce’s preliminary results. Torrington’s Mem. Supp. Mot. J. Agency R. at 49-52.
Commerce stands by its Final Results conclusion on this issue, noting that it did not confirm at verification that the efficiency variance distorted Koyo’s labor costs. Nevertheless, Commerce consents to a remand to explain the basis for its ultimate decision to accept Koyo’s efficiency variance without adjustment. Def.’s Partial Opp’n to Mots. J. Agency R. at 114-19.
Koyo objects to a remand for recalculation of its reported costs, maintaining that Commerce properly accepted its reported COP based on the efficiency variance and emphasizing that Torrington’s contention is based on erroneous assertions by Commerce’s verifier that Commerce rejected in the Final Results. Koyo’s Opp’n to Torrington’s Mot. J. Agency R. at 22-40.
After inspection of the record, the Court agrees that Commerce’s explanation of its change in position regarding Koyo’s efficiency variance was limited, and so, remands this issue to Commerce for a further explanation of its decision.
18. Refusal of COS Adjustment for NTN’s Reimbursement to its U.S. Subsidiary:
During the review, NTN reported certain expenses as commission payments to its U.S. subsidiary for reimbursement of expenses the subsidiary incurred with respect to sales to a specific purchase price customer. Commerce concluded that these expenses were intra-company transfer payments and treated them as indirect selling expenses. In particular, because true commissions had been paid on home market sales, Commerce adjusted FMV for these indirect selling expenses pursuant to the “commission offset,” 19 C.F.R. § 353.56(b)(1). See Final Results, 60 Fed. Reg. at 10,915.
Torrington claims Commerce should have examined NTN’s reported commissions to determine whether they were comparable to an arm’s-length rate and then made a COS adjustment pursuant to 19 U.S.C. § 1677b(a)(4)(B) and 19 C.F.R. § 353.56(a). According to Torrington, pursuant to the CAFC’s decision in LMl-La Metalli Industríale, S.p.A. v. United States, 912 F.2d 455,459 (Fed. Cir. 1990), Commerce adopted a *649practice of allowing a COS adjustment to FMV for commissions paid to related parties when there is adequate evidence supporting a finding that the commissions were at arm’s-length rates and the commissions were directly related to sales. Torrington’s Mem. Supp. Mot. J. Agency R. at 52-56.
Commerce responds that the record clearly demonstrates NTN’s reported commissions were not true commissions, but typical indirect selling expenses. Commerce therefore maintains its decision to treat them as indirect selling expenses, as opposed to making a COS adjustment, was supported by substantial evidence. Def.’s Partial Opp’n to Mots. J. Agency R. at 119-23.
NTN claims this Court recently addressed this issue in Federal-Mogul, 20 CIT at 266-67, 918 F. Supp. at 413-14, where the Court stated that LMI-La Metalli does not impose a duty on Commerce to investigate commissions to determine whether they were based on arms’-length transactions but, rather, focuses on Commerce’s duty not to ignore evidence on the record regarding whether commissions were based on arms’-length transactions. NTN contends that because there was no record evidence that its reported commissions were based on arm’s length transactions, Torrington’s argument must be rejected. NTN’s Opp’n to Torrington’s Mot. J. Agency R. at 12.
Upon careful review of NTN’s questionnaire and supplemental questionnaire responses, the Court agrees with Commerce that the record clearly demonstrates NTN’s reported commissions were not true commissions. In LMI-LaMetalli, 912 F.2d at 459, the expenses at issue were, by agreement with a sales agent, a percentage of the sales by the agent. In contrast, the expenses at issue here were typical indirect selling expenses that do not vary with the quantity of merchandise sold. See NTN Response to Section B Questionnaire, C.R. Doc. No. 16, Def.’s App, Ex. 22, at 19, worksheet #4 (Sept. 21,1993); NTN Response to Supplemental Questionnaire, C.R. Doc. No. 95, Def.’s App., Ex. 28, at 7 (Dec. 8, 1993). Furthermore, deductions for differences in COS have been limited to direct selling expenses, which vary with the quantity sold. See, e.g., Zenith Elees. Corp. v. United States, 77 F.3d 426, 431 (Fed. Cir. 1996). Hence, NTN’s indirect expenses at issue here would not qualify for a COS adjustment, as Torrington suggests.
Consequently, Torrington’s discussion of cases involving true commissions are irrelevant and Commerce properly treated NTN’s reported commissions.
19. Acceptance of NTN’s and NSK’s Levels of Trade:
Torrington claims Commerce erred in improperly shifting the burden of proof to the petitioner to disprove a respondent’s reported levels of trade, hence creating an irrebuttable presumption, and in accepting NTN’s and NSK’s customer classifications without substantial supporting evidence on the record. Torrington further alleges Commerce *650ignored evidence rebutting NTN’s and NSK’s customer classifications. Torrington’s Mem. Supp. Mot. J. Agency R. at 57-64.
Commerce responds that, according to its practice through the functional test, it properly established an economic presumption that NTN’s and NSK’s reported customer classifications existed after verifying the respondents’ information. Commerce therefore claims the burden of proof properly lies with the party attempting to rebut the economic presumption. Commerce further responds it did not ignore record evidence rebutting NTN’s and NSK’s reported classifications but, rather, adhered to its practice of establishing an economic presumption based on verified functional information supplied by the respondents. Def.’s Partial Opp’n to Mots. J. Agency R. at 123-35.
NTN and NSK agree generally with the position taken by Commerce. NTN’s Opp’n to Torrington’s Mot. J. Agency R. at 13-17; NSK’s Opp’n to Torrington’s Mot. J. Agency R. at 4-11.
Under its regulations, Commerce normally calculates FMV and U.S. price based on sales at the same commercial level of trade. See 19 C.F.R. § 353.58. In the Final Results, Commerce accepted NTN’s and NSK’s reported level of trade classifications. 60 Fed. Reg. at 10,940-41. The Court recently addressed and sustained Commerce’s practice of determining a respondent’s levels of trade. In Federal-Mogul, 20 CIT at 1444-45, 950 F. Supp. at 1186, the Court stated that, once Commerce’s functional test is satisfied and an economic presumption is created regarding a respondent’s reported levels of trade, it is proper to require the party opposing a respondent’s classifications to provide a factual basis for rebutting the economic presumption by demonstrating that prices and selling expenses are not correlated to levels of trade. See also Laclede Steel Co. v. United States, 18 CIT at 965, 977 (1994) (discussing the establishment of an economic presumption through Commerce’s functional test and the process of rebutting that presumption). In this case, Torrington has failed to provide such a factual basis.
Torrington claims Commerce improperly accepted NTN’s three reported levels of trade without analyzing NTN’s selling expenses and improperly accepted NTN’s collapse of four categories of customers into two levels of trade. According to its court-approved practice, Commerce performed its functional test on NTN’s reported information and concluded that the evidence supported accepting NTN’s and NSK’s reported levels of trade. See Final Results, 60 Fed. Reg. at 10,940-41 (referring to NTN Section C Questionnaire Response, C.R. Doc. Nos. 25 & 28, Def.’s Exs. 24 & 25, at 5-6,12-13 (Sept. 28,1993)).10 However, Tor-rington failed to rebut this economic presumption because it did not *651present a factual basis demonstrating that there was no correlation between prices, selling expenses and levels of trade.11
Consequently, Commerce properly employed its practice of determining NTN’s and NSK’s levels of trade through the functional test and requiring Torrington to rebut the economic presumption established by the functional test. The Court further agrees Torrington failed to present a factual basis for making such a rebuttal.
20. Clerical Errors:
a. Affecting Koyo:
(1) U.S. Doubtful Debt
During the review, Commerce removed Koyo’s doubtful debt expense from Koyo’s reported home market selling expenses, but not from Koyo’s U.S. selling expenses because it could not locate any record information quantifying the U.S. doubtful debt. Final Results, 60 Fed. Reg. at 10,917. Koyo claims that Commerce’s treatment of this expense was due to an oversight of a relevant Koyo exhibit and requests a remand. Koyo’s Mem. Supp. Mot. J. Agency R. at 24-25.
Commerce agrees that Koyo has noted information that quantifies Koyo’s U.S. doubtful debt and consents to a remand to correct this error. Def.’s Partial Opp’n to Mots. J. Agency R. at 27-28. Torrington, however, contends Commerce’s failure to find the expense amount after examining Koyo’s financial statements demonstrates Koyo did not provide adequate information on the record. Torrington’s Opp’n to Mots. J. Agency R. at 57-59.
Upon inspection of the record, and in particular Koyo’s U.S. Sales Response (see Koyo U.S. Sales Response, PR. Doc. No. 103, Koyo’s App., Ex. 8, at B-15), the Court agrees with Koyo and Commerce that Commerce merely overlooked relevant information on the record. The Court therefore remands this issue for Commerce to remove the U.S. doubtful debt amount from Koyo’s U.S. selling expenses.
(2) Non-Scope Merchandise
Koyo claims the computer program used to calculate Koyo’s final dumping margin contained a typographical error improperly preventing the elimination of a sale of non-scope merchandise from the dumping margin calculation. Koyo’s Mem. Supp. Mot. J. Agency R. at 25-26.
Commerce agrees the computer program used to calculate Koyo’s final dumping margin contained a typographical error and consents to a remand to correct this error. Def.’s Partial Opp’n to Mots. J. Agency R. at 28. Torrington concedes that such a remand may be appropriate. Tor-rington’s Opp’n to Mots. J. Agency R. at 59.
Upon inspection, the Court concludes that a typographical error indeed existed in the computer program used to calculate Koyo’s final *652dumping margin and remands this issue to Commerce to correct this error to allow the elimination of a sale of non-scope merchandise from the dumping margin calculation.
(3) Calculation of Profit in Further Manufacturing
Koyo contends Commerce’s VAT adjustment calculation had the effect of inflating the profit calculated on U.S. sales of further manufactured merchandise and claims this error would be mooted if Commerce were granted the opportunity to recalculate Koyo’s margin using a tax-neutral adjustment methodology. Koyo’s Mem. Supp. Mot. J. Agency it. at 26. Commerce agrees this error exists and would be corrected if Commerce were granted the opportunity to recalculate Koyo’s margin using a tax-neutral adjustment methodology. Def.’s Partial Opp’n to Mots. J. Agency R. at 28-29.
As this Court decided above that Commerce is to recalculate Koyo’s margin using a tax-neutral adjustment methodology upon remand, the error at issue here will be mooted in Commerce’s remand results. Hence, there is no need for a remand to address this specific issue.
(4) Home Market Doubtful Debt
Torrington alleges, and Commerce agrees, the final computer program used to calculate Koyo’s dumping margin did not properly adjust home market indirect selling expenses to eliminate the doubtful debt allowance. Torrington’s Mem. Supp. Mot. J. Agency R. at 64; Def. ’s Partial Opp’n to Mots. J. Agency R. at 29. Upon review, the Court concludes the computer program did not properly adjust home market indirect selling expenses to eliminate the doubtful debt allowance and remands this issue to Commerce for correction of the error.
b. Affecting Ñachi:
Torrington contends, and Commerce agrees, the computer program used to calculate the dumping margin for Ñachi did not properly account for certain U.S. transactions. Torrington’s Mem. Supp. Mot. J. Agency R. at 64; Def.’s Partial Opp’n Mots. J. Agency R. at 29. Upon inspection of the record, the Court concludes Commerce’s program erred in improperly accounting for certain U.S. transactions and remands this issue for Commerce to correct the error.
c. Affecting NTN:
Torrington alleges, and Commerce agrees, the computer program used to calculate the dumping margin for NTN contained a clerical error. Torrington’s Mem. Supp. Mot. J. Agency R. at 65; Def.’s Partial Opp’n to Mots. J. Agency R. at 29. Upon inspection of the record, the Court concludes Commerce’s program indeed contained an error with respect to NTN’s dumping margin and remands this issue for Commerce to correct the error.
d. Affecting NSK:
Torrington claims the computer program used to calculate the dumping margin for NSK contained two clerical errors that resulted in the *653calculation of erroneous CV figures and in the use of an incorrect factor. Torrington’s Mem. Supp. Mot. J. Agency R. at 65-68. Commerce acknowledges Torrington appears to have identified areas of computer program errors and consents to a remand to address these alleged errors. Def.’s Partial Opp’n to Mots. J. Agency R. at 30.
Upon inspection of the record, the Court concludes Commerce’s program may contain errors with respect to NSK’s dumping margin and remands this issue for Commerce to identify specific clerical errors, correct the errors it uncovers and otherwise address these alleged errors.
Conclusion
In accordance with the foregoing opinion, this case is remanded to Commerce to: (1) apply a tax-neutral VAT methodology; (2) deny the adjustment to FMV for NSK’s return rebates and post-sale price adjustments; (3) review the record to (a) determine whether it is possible to remove those portions of Koyo’s warranty expenses which relate to non-scope merchandise from the adjustments to FMV or (b) deny the adjustment if such removal cannot be made; (4) deny the adjustment to FMV for NTN’s home market discounts at issue; (5) determine whether the NTN billing adjustments not reported on a transaction-specific basis were made solely over in-scope merchandise and, if so, to allow them a direct adjustment to FMV or, if such a determination cannot conclusively be made, to deny them an adjustment to FMV; (6) reopen the record to allow Koyo to submit documentation showing the nature of the expenses Koyo characterized as non-operating expenses; (7) exclude NSK’s zero-priced sample transfers from NSK’s U.S. sales database; (8) exclude NTN’s sample and other similar transfers from NTN’s home market sales database; (9) allow NTN’s adjustment for interest expenses on antidumping duty cash deposits; (10) recalculate NTN’s COP and CV without resort to best information available; (11) examine the acceptance of Koyo’s allocation of air freight expenses and, if necessary, request additional information; (12) explain further the basis for accepting Koyo’s efficiency variance without adjustment; and (13) correct clerical errors for Koyo (with respect to U.S. and home market doubtful debt, and failure to eliminate a sale of non-scope merchandise from the dumping margin calculation), Ñachi (with respect to improper accounting for certain U.S. transactions), NTN and NSK. Commerce is sustained as to all other issues.
Commerce claims that, unlike other adjustments, adjustments for PSPAs are not made pursuant to the circumstances-of-sale ("COS”) adjustment contained in 19 U.S.C. § 1677b(a)(4)(B) and 19 C.ER. § 353.56(a). Def.’s Partial Opp’n to Mots. J. Agency R. at 11 n. 3. This is not entirely clear, as the Torrington court noted FMV may be adjusted for PSPAs under the COS provision if Commerce concludes that a respondent’s PSPAs constitute direct selling expenses. See 82 E3d at 1051 nn. 17-18 (referringto Smith-Corona Group u. United States, 713 F.3d 1568,1580 (Fed. Cir. 1983) (COS adjustment allowed for rebates having a reasonably direct relationship to the sales under consideration and where the value determinations are based on proof of actual costs)).
In the Court of International Trade opinion on appeal, the Court decided respondent was not allowed a deduction for its PSPAs because they were calculated using data for both in-scope and out-of-scope merchandise. See Torrington, 17 CIT 199,218,818 E Supp. 1563,1578-79 (1993). As noted above, however, the CAFC came to the same conclusion on different grounds, noting 19 C.F.R. § 353.56(b)(2) explicitly disallows direct selling expenses from being deducted from FMV under the ESP offset. Torrington, 82 F.3d at 1051.
NSK has brought to the Court’s attention that Commerce made direct adjustments to FMV for the identical NSK expenses at issue here (and an indirect adjustment for stock transfer commissions) in the fifth and sixth administrative reviews. Letter from Lipstein, Jaffe & Lawson, L.L.E, attorneys for NSK, to Raymond F. Burghardt, Clerk of the Court (May 30,1997). NSK states Commerce’s acceptance of these expenses in these subsequent reviews illustrates how Commerce believes such expenses should be treated in this case. NSK supports its position by noting that Commerce indicated its new regulations for the allocation of expenses and price adjustments, issued in conformance with the Uruguay Round Agreements Act (“URAA”) and applicable beginning with the sixth review, simply codify its past practice. Id. (citing Antidumping Duties; Countervailing Duties (“New Regulations’’), 62 Fed. Reg. 27,296, 27,346) (May 19, 1997).
As Commerce stated as background for its new regulations, “neither the pre-URAA statute nor the Department’s prior regulations addressed allocation methods [for expenses and price adjustments! * * *. Instead, the Department and the courts resolved these issues on a case-by-case basis.” New Regulations, 62 Fed. Reg. at 27,346. The Court concludes that Commerce’s recent effort to establish more consistent guidelines and decrease litigation over allocation issues has no bearingon litigation over Commerce’sprevious determinations. Inparticular, Commerce’s prior methodology of acceptance of expenses and price adjustments as direct adjustments to FMV primarily pursuant to transaction-specific verification has been approved by this Court and the CAFC and is applicable to this case. See, e.g., Torrington Co. v. United States, 20 CIT 632, 640, 926 F. Supp. 1151,1158 (1996) (stating respondent’s inability to tie PSPAs to particular sales prevented Commerce from allowing a direct adjustment to FMV for PSPAs at issue). This reasoning holds true for Commerce’s pre-URAA fifth review determinations, which must also be resolved on an individual basis.
NSK responds Torrington failed to exhaust its administrative remedies and the NSK adjustments at issue here qualify as selling expenses. NSK’s Opp’n to Torrington’s Mot. J. Agency R. at 11-16. First, the Court concludes Tor-rington sufficiently raised this issue regardingprice adjustments at the administrative level. See Torrington's Administrative Case Brief, ER. Doc. No. 358, Fiche 144, Frames 44,46 (March 1,1994). The Court, nevertheless, concludes the billing adjustments at issue are selling expenses. See Torrington, 82 F.3d at 1050-51 (treating similax* billing adjustments as selling expenses).
The confidential version of the administrative record is designated “C.R.” and the public record is designated “ER.”
Section 1677a(e)(3) states, in relevant part:
(e) Additional adjustments to exporter’s sales price
For .purposes of this section, the exporter’s sales price shall also be adjusted by being reduced by the amount, if any, of-
(3) any increased value, including additional material and labor, resulting from a process of manufacture or assembly performed on the imported merchandise after the importation of the merchandise and before its sale to a person who is not the exporter of the merchandise.
In accordance with this apparent congressional intent, when an imported item constitutes an insignificant amount of a complete final product, Commerce has decided that Section 1677a(e)(3) should not be applied to the insignificant import, which should be excluded from the coverage of review. This principle has come to be known as the “Roller Chain” rule. See Roller Chain, Other Than Bicycle, From Japan; Final Results of Administrative Review of Antidump-ing Finding, 48 Fed. Reg. 51,801,51,804 (Nov. 14,1983) (an imported chain incorporated into a finished motorcycle constituted an insignificant amount of the finished motorcycle).
NSK claims that Commerce has admitted in prior proceedings that it cannot use Section 1677a(e)(3) to reduce ESP for the amount of any value added to bearingparts further manufactured or assembled in the U.S. into the same class or kind of merchandise. NSK’s Mem. Supp. Mot. J. Agency R. at 45-48. Commerce has made no such admission, but has merely employed various simplification techniques that excluded further-processed merchandise from its price comparisons. See, e.g., Antifriction Bearings (Other ThanTapered Roller Bearings) and Parts thereofFrom Japan; Preliminary Results of Antidumping Duty Administrative Reviews and Partial Termination of Antidumping Duty Administrative Reviews, 56 Fed. Reg. 11,186,11,187-88 (March 15,1991); Final Determinations of Sales at Less than Fair Value: Antifriction Bearings (Other Than Tapered Roller Bearings) and Parts Thereof From the Federal Republic of Germany, 54 Fed. Reg. 18,992,19,029 (May 3,1989). Further, in the second review, the Court upheld Commerce’s decision to include information on sales of further-processed merchandise.
In addition to Commerce’s improper use of BIA, it appeal’s that the adjustment was based on several erroneous assumptions regarding the worksheet data. For instance, Commerce originally believed that the actual cost information in the NTN worksheet was within the period of review when, in fact, it was not. See NTN COP/CV Verification. Report, C.R.Doc. No. 157, Def.’s App., Ex. 33,at8n.l3 (Feb. 25,1994). Further, Commerce’s conclusion that the majority of standard costs that remained unchanged were for non-subject merchandise, see Final Results, 60 Fed. Reg. at 10,928, is unsupported by the record.
NPB insists it fully and accurately responded to Commerce’s original questionnaire by responding that it had no negative billing adjustments because the relevant questionnaire instruction required invoice-specific information, which NPB could not trace. NPB’s Reply Mem. Supp. Mot. J. Agency R. at 1-6. Assuming NPB could not trace the necessary information in an invoice-specific manner, the Court is nonetheless disturbed that NPB would respond that it had no such adjustments when, in fact, it had 1,210 billing adjustments that it could report in aproduct-specific manner. See NPB Verification Report, PR. Doc. No. 333, Def.’s App., Ex. 17, at 4.
Torrington claims Commerce’s treatment of NTN was inconsistent with its treatment of Nankai Seiko Co., Ltd. (“Nankai Seiko”). Torrington’s Mem. Supp. Mot. J. Agency R. at 58. However, unlike NTN, Nankai Seiko failed to establish functional differences among the claimed levels of trade. Hence, because Nankai Seiko failed to satisfy the functional test, there was no economic presumption attached to its customer categories.
Torrington’s claim that NSK somehow received a quantity adjustment it was not otherwise entitled to under 19 C.ER. § 353.55, see Torrington’s Mem. Supp. Mot. J. Agency R. at 63-64, is unsupported by the record. Rather, the record indicates quantity was only relevant in Commerce’s level of trade analysis for the correlation test.