Full opinion text
Opinion CARMAN, Chief Judge. This matter comes before the Court on motion for judgment upon the agency record filed by Plaintiffs Hynix Semiconductor, Inc. and Hynix Semiconductor America, Inc. (“Hynix”). Plaintiffs challenge the United States Department of Commerce’s (“Commerce”) final antidumping determination in the Seventh Administrative Review of certain dynamic random access memory semiconductors (“DRAMs”) of one megabit or above from the Republic of Korea for the period of May 1, 1999 to December 81, 1999. See Dynamic Random Access Memory Semiconductors of One Megabit or Above From the Republic of Korea: Final Results of Antidumping Administrative Review, 66 Fed.Reg. 52,097 (Oct. 12, 2001) (“Final Results”). The Court has jurisdiction over this ease pursuant to 28 U.S.C. § 1581(c) (2000). Background On May 10, 1993, Commerce published the antidumping duty order on DRAMs from Korea. See Dynamic Random Access Memory Semiconductors of One Megabit or Above from the Republic of Korea, 58 Fed.Reg. 27,520 (May 10, 1993). Commerce published a notice of opportunity to request an administrative review of the antidumping duty order on DRAMs from Korea on May 16, 2000. Antidumping or Countervailing Duty Order, Finding, or Suspended Investigation; Opportunity To Request Administrative Review, 65 Fed. Reg. 31,141 (May 16, 2000). Micron Technology, Inc. (“Micron”), the Defendant-Intervenor in this case, requested an administrative review of Plaintiffs (two Korean DRAMs manufacturers) and six Korean resellers of DRAMs for the period of May 1, 1999 to April 30, 2000. Micron also requested a cost of production (“COP”) investigation of Plaintiffs pursuant to 19 U.S.C. § 1677b(b) for the period of review (“POR”). See Dynamic Random Access Memory Semiconductors of One Megabit or Above From the Republic of Korea: Preliminary Results of Antidumping Duty Administrative Review, 66 Fed.Reg. 30,688, 30,689 (June 7, 2001) (“Preliminary Results”). On May 31, 2000, Plaintiffs responded to Commerce’s notice and requested that Commerce review their exports entered into the United States during the stated POR. Commerce initiated the Seventh Administrative Review of the antidumping order on DRAMs from Korea on July 7, 2000. See Initiation of Antidumping and Countervailing Duty Administrative Reviews and Requests for Revocations in Part, 65 Fed.Reg. 41,942 (July 7, 2000). While the Seventh Administrative Review was pending, Commerce published the final results of a separate sunset review of the subject DRAMs from Korea. See Dynamic Random Access Memory Semiconductors (“DRAMs’’) of One Megabit and Above From the Republic of Korea; Final Results of Full Sunset Review and Revocation of Order, 65 Fed. Reg. 59,391 (Oct. 5, 2000) (“Final Sunset Results”). Commerce concluded that it would be unlikely that dumping of DRAMs from Korea would continue or resume. Id. at 59,391-92. Thus, the antidumping duty order was revoked effective December 31, 1999. Id. at 59,392. The revocation of the anti-dumping duty order shortened the POR of the Seventh Administrative Review to eight months, covering May 1, 1999 to December 31, 1999, rather than a full twelve month review period. See Preliminary Results, 66 Fed.Reg. at 30,689. DRAMs entering on or after January 1, 2000, were to be liquidated without regard to antidumping duties. Final Sunset Results, 65 Fed.Reg. at 59,392. Commerce published the Preliminary Results of the Seventh Administrative Review on June 7, 2001. See Preliminary Results, 66 Fed.Reg. 30,688. In the Preliminary Results, Commerce calculated the dumping margin by using the sales plus POR-entries approach suggested by Plaintiffs. Id. at 30,692. The sales plus POR-entries approach included the constructed export price sales of products completed during the POR and entries of the subject merchandise that entered within the POR with sales that concluded after the POR. Id. Using this hybrid approach, Commerce arrived at a 3.01 percent dumping margin. Id. at 30,694. Commerce published the Final Results, accompanied by the Final Decision Memorandum of the Seventh Administrative Review on October 12, 2001. See Final Results, 66 Fed.Reg. 52,097; Issues and Decision Memorandum for the Administrative Review of Dynamic Random Access Memory Semiconductors from Korea — 5/1/1999 through 12/31/1999; Final Results (Oct. 5, 2001) (Pub.Doc. No. 72) (Def.’s Pub.App. Ex. 7) (“Final Decision Memorandum”). The Final Results included four determinations that Plaintiffs challenge in the instant case. First, Commerce used a different methodology to calculate the dumping margin in the Final Results than it had used in the Preliminary Results to arrive at a 2.92 percent dumping margin in the Final Results. Final Results, 66 Fed.Reg. at 52,099; Final Decision Memorandum at cmt. 7 (Pub.Doc. No. 72) (Def.’s Pub.App. Ex. 7). Instead of using the sales plus POR-entries approach as the basis for calculating the dumping margin, Commerce only used sales completed during the POR. Id. Second, Commerce recalculated Plaintiffs’ reported research and development (“R & D”) costs. Final Decision Memorandum at cmts. 2 and 3. Third, Commerce rejected Plaintiffs’ accounting adjustments for the average useful lives (“AULs”) of Plaintiffs’ semiconductor equipment. Id. at cmt. 5. Fourth, Commerce rejected Plaintiffs’ use of offsets to foreign currency exchange losses for the revaluation of Plaintiffs’ fixed assets. Id. at cmt. 1. Plaintiffs commenced this action on December 10, 2001, to challenge the Final Results. Standard of Review The Court will sustain Commerce’s 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 “such relevant evidence as a reasonable mind might accept as adequate to support a conclusion.” Consolidated Edison Co. v. NLRB, 305 U.S. 197, 229, 59 S.Ct. 206, 83 L.Ed. 126 (1938); Micron Tech., Inc. v. United States, 117 F.3d 1386, 1393 (Fed.Cir.1997). “The specific determination we make is whether the evidence and reasonable inferences from the record support” Commerce’s findings. Daewoo Elecs. v. Int’l Union, 6 F.3d 1511, 1520 (Fed.Cir.1993) (quoting Matsushita Elec. Indus. Co. v. United States, 750 F.2d 927, 933 (Fed.Cir.1984)). In determining whether Commerce’s interpretation and application of the dumping duty statute is in accordance with law, the Court must consider whether “Congress has directly spoken to the precise question at issue,” and if not, whether the agency’s interpretation of the statute is reasonable. Chevron U.S.A. Inc. v. Natural Res. Def. Council, Inc., 467 U.S. 837, 843-44, 104 S.Ct. 2778, 81 L.Ed.2d 694 (1984). “[A] court must defer to an agency’s reasonable interpretation of a statute even if the court might have preferred another.” Koyo Seiko Co. v. United States, 36 F.3d 1565, 1570 (Fed.Cir.1994). Additionally, the Court recognizes that “Commerce’s special expertise in administrating the antidumping law entitles its decisions to deference from the courts.” Ta Chen Stainless Steel Pipe, Inc. v. United States, 298 F.3d 1330, 1335 (Fed.Cir.2002). Discussion I. Commerce’s Decision to Calculate the Dumping Margin Based Solely on Sales Completed During the Period of Review is Supported by Substantial Evidence or Otherwise in Accordance with Law. A. Plaintiffs’ Contentions Plaintiffs believe that Commerce erred in calculating the “ultimate antidumping duty liability” because Commerce only used constructed export price sales completed during the POR to arrive at the dumping duty published in the Final Results. (Plaintiffs’ Motion for Judgment Upon the Agency Record (“Pis.’ Br.”) at 6.) Plaintiffs contend that Commerce’s approach resulted in an inaccurate and incomplete “universe of U.S. sales” upon which Commerce based the dumping calculation. (Id.) According to Plaintiffs, the basis for the dumping calculation was incomplete because Commerce did not include sales of the subject merchandise that entered the United States during the POR, but which were not shipped to the customer prior to the end of the POR. (Id. at 7.) Thus, Plaintiffs explain, these entries did not have a “sale date” during the POR and were not used by Commerce in determining the final dumping margin. (Id.) Plaintiffs argue that the dumping margin calculation was inaccurate because all POR-entries of the subject merchandise were assessed a duty, but not all POR-entries were included in calculating that duty. (Id. at 13-14.) Plaintiffs submit that Commerce should have calculated the dumping margin using Plaintiffs’ suggested sales plus POR-entries approach. (Id. at 10-12.) Plaintiffs’ suggested approach is not a “pure entry-based approach” seen in previous administrative reviews. (Reply Brief of Plaintiffs Hynix Semiconductor, Inc. and Hynix Semiconductor America, Inc. (“Pis.’ Reply Br.”) at 8.) Rather, Plaintiffs’ sales plus POR-entries approach uses “all sales during the POR plus sales that occurred after the POR but were entered before termination of the [antidumping] order” to calculate the dumping margin. (Pis.’ Br. at 14.) Plaintiffs note that Commerce used their suggested sales plus POR-entries approach to determine the dumping margin in the Preliminary Results. (Id. at 7 n. 15.) However, Commerce changed to a sales-based approach in the Final Results. (Id. at 7.) Plaintiffs maintain that this change “cannot be sustained.” (Id.) Plaintiffs argue that Commerce’s decision to rely on a sales-based approach rather than Plaintiffs’ sales plus POR-entries approach in determining the dumping margin is contrary to the statutory intent of determining accurate antidumping duty liability. (Id. at 8.) Plaintiffs advance that Commerce failed to take into account the “unique circumstances” of this case: (1) the POR is truncated to eight months, rather than the typical twelve months; and (2) this is the last administrative review of the subject merchandise. (Id. at 12-15.) Plaintiffs contend that because of these unique circumstances, the opportunity to correct inaccuracies in a subsequent review is no longer available. (Id. at 12; Pis.’ Reply Br. at 6-7.) Therefore, Plaintiffs assert, the facts warrant the use of the sales plus POR-entries approach. B. Defendant’s Contentions Defendant maintains that Commerce’s methodology used in the Final Results to calculate the dumping duty is in accordance with law. (Defendant’s Opposition to Plaintiffs’ Motion for Judgment Upon the Agency Record (“Def.’s Br.”) at 8.) Defendant contends that 19 U.S.C. § 1675(a)(2), the statute addressing anti-dumping duty determinations, does not prescribe the methodology that Commerce must employ in determining the dumping duty. (Id.) Defendant asserts that the Court should defer to Commerce’s calculation methodology, if it is reasonable. (Id. at 9.) According to Defendant, the reasonableness of the decision to use a sales-based approach to calculate the dumping margin is supported by Commerce’s explanation in the Final Results and the fact that this Court has upheld this methodology in other administrative reviews. (Def.’s Br. at 9 (citing Ad Hoc Comm. of S. Cal. Producers of Gray Portland Cement v. United States, 914 F.Supp. 535 (CIT 1995))); Final Decision Memorandum at 22-23 (Pub.Doc. No. 72) (Def.’s Pub.App. Ex. 7). Defendant argues that Commerce’s use of a sales-based approach in this case is in accordance with Commerce’s normal practice of basing the dumping margin calculation on sales during the POR, especially in order to ensure consistency with prior reviews and to take into account the difficulties of accurately connecting entries to sales. (Def.’s Br. at 9-10.) Defendant cites the rationale outlined in the promulgating regulations pursuant to the Uruguay Round Agreements Act to support Commerce choice of methodology in the present case. (Id. (citing Antidumping Duties; Countervailing Duties, 62 Fed.Reg. 27,296, 27,314 (May 19, 1997)).) Defendant maintains that Commerce’s decision to base the dumping margin calculation on sales of the subject merchandise during this POR was reasonable because that approach has been used in all six previous administrative reviews involving the subject merchandise and Plaintiffs are admittedly unable to tie all POR entries to post-POR sales. (Id. at 11-13.) Defendant refutes Plaintiffs’ contention that Commerce should have exercised its discretion to use post-POR sales of subject merchandise that entered during the POR. (Id. at 12.) Defendant concedes that there is no “rule” that requires a sales-based approach be used in calculating all dumping margins. (Id. at 12-13.) However, Defendant points to the preamble of Commerce’s regulations which states that Commerce will use either entries or sales for calculating a dumping margin in administrative reviews and that “consistency across administrative reviews is the essential factor when Commerce considers whether to use entries instead of sales for its dumping analysis.” (Id. at 12.) Defendant continues that nowhere in Commerce’s regulations or in the agency’s practice has a hybrid sales plus POR-entries approach been used to calculate dumping duties. (Id.) Contrary to Plaintiffs’ claim of increased accuracy, Defendant asserts that it is not clear how the use of a hybrid approach in this case would yield more accurate results, especially given the fact that Plaintiffs could not tie all POR-entries to post-POR sales. (Id. at 14.) Defendant argues that utilizing the hybrid sales plus POR-entries approach proposed by Plaintiffs will lead to distortions in the dumping calculation because such an approach was not used in the prior administrative reviews involving Plaintiffs. (Id. at 13.) To further support its position, Defendant states that this Court has upheld Commerce’s use of a sales-based methodology even in cases where the information necessary to use an entry-based methodology was available. (Id. at 14 (citing Gray Portland Cement, 914 F.Supp. at 544 n. 7).) Therefore, Defendant asserts, Commerce’s use of a sales-based approach for the Final Results was reasonable and in accordance with Commerce’s regulations, Commerce’s prior practice, and the anti-dumping statute. (Id. at 14-15.) C. Defendant-Intervenor’s Contentions Because the Court finds Defendanh-In-tervenor’s (“Micron”) arguments on this issue are substantially similar to those presented by Defendant, the Court will not recount them in this opinion, although they have been duly considered. D. Analysis For the reasons set forth below, the Court holds that Commerce’s decision to use a sales-based approach in the Final Results to calculate duty margins is supported by substantial evidence or is otherwise in accordance with law pursuant to 19 U.S.C. § 1516a(b)(1)(B). Plaintiffs are incorrect in their assertion that Commerce’s change of calculation methodology from a sales plus POR-entries approach in the Preliminary Results to the sales-based approach in the Final Results cannot be sustained. This Court notes that Commerce adopted Plaintiffs’ calculation methodology in the Preliminary Results without explanation. See Preliminary Results, 66 Fed.Reg. at 30,689, 30,692. However, in the Final Results, Commerce explained the rationale behind returning to the sales-based approach. See Final Decision Memorandum at 22-23 (Pub.Doc. No. 72) (Def.’s Pub.App. Ex. 7). Generally, “an agency must either conform itself to its prior decisions or explain the reasons for its departure.’ ” Hussey Copper, Ltd. v. United States, 834 F.Supp. 413, 418 (CIT 1993) (quoting Citrosuco Paulista, S.A. v. United States, 704 F.Supp. 1075, 1088 (CIT 1988)). The Court will reject an agency’s determination to depart from past practice if there is an insufficient explanation. Am. Silicon Techs. v. United States, 19 F.Supp.2d 1121, 1123 (CIT 1998). Commerce’s practice during the prior administrative reviews of the subject merchandise has been to base the dumping margin calculations on sales made during a given POR. (Def.’s Br. at 11.) Commerce departed from that practice in the Preliminary Results of this administrative review, but returned to that method for the Final Results. As detailed below, the Court finds that Commerce supplied a reasonable explanation to support its decision to apply the sales-based approach in this final administrative review. Commerce’s application of 19 U.S.C. § 1675(a)(2)(A)(ii) in this case is supported by substantial evidence and is otherwise in accordance with law because the statute does not limit Commerce to using only entries to calculate the dumping margin. Subsection 1675(a)(2)(A)(ii) states that Commerce “shall determine the dumping margin for each ... entry,” and the dumping duty determinations shall be the basis of assessment of duties on the entries of merchandise during the POR. However, this Court has held that “Congress could not have intended to limit Commerce’s margin calculations solely to entries.” Am. Silicon Techs. v. United States, 23 Ct. Int’l Trade 237, Slip Op. 99-34 at 7 (Apr. 9, 1999). Because the statute does not specify the basis upon which Commerce must calculate the dumping margin, “the Court must defer to Commerce’s interpretation of the statute as long as the interpretation is reasonable.” Helmerich & Payne, Inc. v. United States, 24 F.Supp.2d 304, 309-10 (CIT 1998) (citing Koyo Seiko Co., 36 F.3d at 1573). Commerce’s regulations accord it the flexibility to rely upon sales occurring during the POR to make margin calculations where appropriate. Specifically, 19 C.F.R. § 351.213 allows Commerce to base its margin calculation on “entries, exports, or sales.” 19 C.F.R. § 351.213(e)(1) (1997). Although Commerce applied Plaintiffs’ suggested sales plus POR-entries approach in the Preliminary Results, Commerce returned to the sales-based method used throughout the prior administrative reviews in this case for the Final Results in order to be consistent. The importance of consistency across reviews can be inferred from the rationale outlined in the promulgating regulations pursuant to the Uruguay Round Agreements Act and supports Commerce’s choice of margin calculation methodology in the present case. See Antidumping Duties; Countervailing Duties, 62 Fed.Reg. at 27,314. Although some limited circumstances may exist to afford the use of an entry-based methodology, in this case, Commerce declined to exercise its discretion to include post-POR sales in the administrative review because there was no compelling evidence to cause Commerce to deviate from using completed sales as it had done in the prior administrative reviews. Final Decision Memorandum at 20-23 (Pub.Doc. No. 72) (Def.’s Pub.App. Ex. 7). Moreover, nothing in Commerce’s regulations supports the use of a hybrid sales plus POR-entries approach for calculating dumping margins. This Court has upheld Commerce’s decision to use sales during the POR rather than entries during the POR for the purposes of calculating antidumping margins. See Gray Portland Cement, 914 F.Supp. 535, 544. Commerce’s approach is not made unreasonable by the fact that POR-sales used to calculate the dumping margin are not necessarily tied to the POR-entries that will be assessed that dumping margin or by the fact that Plaintiffs could provide Commerce with the information to calculate the dumping margin based on entries during the POR. See id.; see also NSK Ltd. v. United States, 825 F.Supp. 315, 320 (CIT 1993). This Court has acknowledged that the use of sales rather than entries in calculating the dumping margin “may result in the ... failure to consider entries made, but not sold, during the period of review.” Gray Portland Cement, 914 F.Supp. at 544. Nevertheless, “such consequences are permissible so long as dumping duties are assessed only upon entries made during the period of review.” Id. at 544-45. Plaintiffs do not demonstrate that the circumstances of the present case are so unique that Commerce was unreasonable in its decision to use the sales-based approach that had been used in the previous six administrative reviews. For the foregoing reasons, the Court holds that Commerce’s decision to use POR-sales to calculate the dumping margin for the Seventh Administrative Review is supported by substantial evidence and is otherwise in accordance with law. II. Commerce’s Decision to Reject Plaintiffs’ Method of Accounting for Research and Development Expenses is Remanded. Plaintiffs challenge two aspects of Commerce’s treatment of Plaintiffs’ research and development (“R & D”) costs used in constructing the cost of production (“COP”) for the subject merchandise based on Defendant-Intervenor’s request for the COP calculation under 19 U.S.C. § 1677b(f)(1)(A). See Preliminary Results, 66 Fed.Reg. at 30,689. Plaintiffs question: (1) whether Commerce properly rejected their amortized R&D costs; and (2) the appropriateness of Commerce’s reallocation of total semiconductor R&D costs over all semiconductor production based on its application of its cross-fertilization of R & D theory. (Pis.’ Br. at 18, 26.) These two issues are addressed separately in the following discussion. For the reasons stated below, this Court finds that Commerce’s determination regarding Plaintiffs’ reported amortized R & D costs and the appropriateness of Commerce’s reallocation of total R&D costs over all semiconductor production by its application of its cross-fertilization theory are unsupported by substantial evidence, or otherwise not in accordance with law. Accordingly, these issues are remanded to Commerce for further consideration and explanation. 1. Commerce’s Rejection of Plaintiffs’ Reported R&D Costs A. Plaintiffs’ Contentions Plaintiffs provided Commerce with amortized R&D costs for 1999 to be used as one of the components in calculating the COP for the subject merchandise. (Id. at 17; see also “Memorandum to File From Paige Rivas Regarding Verification of the Response of Hyundai Electronics Industries Co., Ltd. and Hyundai Electronics America, for the 1999 Administrative Review of Dynamic Random Access Memory Semiconductors of One Megabit and Above from Korea” (May 31, 2001) (Pis.’ Conf.App. Ex. 3) (“Sales Verification Report”).) Commerce verified the reported R & D costs, noting that Plaintiffs maintained costs by project, and found no discrepancies between the reported costs and Plaintiffs’ financial statements. (Sales Verification Report at 14-15 (Pis.’ Conf. App. Ex. 3).) In the Final Results, Commerce rejected Plaintiffs’ amortized R & D for two reasons: (1) alleged distortions to the COP resulting from changes in Plaintiffs’ accounting practices during the Fifth Administrative Review; and (2) Plaintiffs’ accounting practice of deferring certain development expenses. See Final Decision Memorandum at 8-11 (Pub.Doc. No. 72) (Def.’s Pub.App. Ex. 7). Plaintiffs argue that Commerce’s decision to reject the reported amortized R & D costs in the Final Results is without merit and is not supported by substantial evidence on the record. (Pis.’ Br. at 18.) Plaintiffs deny that distortions arose such that the reported R & D costs did not accurately reflect the COP simply because Plaintiffs changed R & D accounting practices in the past. (Id. at 20-21.) First, Plaintiffs contend that Commerce overstates the frequency with which they have changed R & D accounting methods. (Id. at 21.) Plaintiffs note that prior to the initial investigation in this case, while operating as Hyundai Industries Co., Ltd. (“Hyundai”), they historically amortized R & D costs. (Id.) In 1996, Hyundai switched to expensing R & D costs and then in 1997, returned to amortization. (Id.) Plaintiffs explain that LG Semicon, the company that Plaintiffs acquired in 1999, expensed its R & D costs from 1993 to 1997. (Id. at 21 n. 61.) Plaintiffs state that their business decision in 1997 to return to their previously-used accounting method of amortizing R & D costs does not establish a practice of “continuously” changing accounting methodologies from one review to the next. (Id. at 22.) Plaintiffs contend that a change in their R & D accounting method prior to and during the Fifth Administrative Review does not continue to have distortive effects on the COP in this review. (Id.) Plaintiffs challenge the validity of Commerce’s assertion that changing accounting methods fails to capture all R & D costs because the new method does not include R & D costs that would have otherwise been amortized in prior years. (Id.) Plaintiffs maintain that Commerce overlooked the fact that “by expensing R & D costs in 1996 ... all the amounts that would have been amortized [in 1996] were taken into account in the prior year’s costs, in previous administrative reviews.” (Id.) Plaintiffs assert that the information provided to Commerce demonstrates that R & D costs incurred in the subsequent years are then carried forward over the life of the R & D asset. (Id. at 22.) Plaintiffs claim that they provided Commerce with evidence to support this contention in then-response to Commerce’s request for information during the administrative review. (Id. (citing “Response of Hyundai Electronics Indus. Co., Ltd. and Hyundai Electronics America to the Department’s Supplemental Request for Information,” vol. 2 at DS-2 (Dec. 28, 2000) (BPI Doc. No. 4) (Pub.Doc. No. 31) (Pis.’ Conf.App. Ex. 4)).) Plaintiffs note that Commerce has previously accepted amortized R & D costs for other products and even rejected a respondent’s attempt to expense current R & D costs where these costs were amortized in the respondent’s financial statement. (Id. at 23-24 (citing Final Determination of Sales at Less Than Fair Value: Polyethylene Terephthalate Film, Sheet, and Strip From the Republic of Korea, 56 Fed.Reg. 16,305, 16,312 (Apr. 22, 1991); Final Determination of Sales at Less Than Fair Value: Certain Welded Stainless Steel Pipe From the Republic of Korea, 57 Fed.Reg. 53,693 (Nov. 12, 1992)).) Plaintiffs acknowledge that Commerce has, however, rejected amortized R & D costs in favor of expensing R & D costs in administrative reviews involving DRAMs and SRAMs from Korea and Taiwan. (Id. at 24 (citing Final Decision Memorandum at 10-11 (Pub.Doc. No. 72) (Def.’s Pub.App. Ex. 7)).) Nevertheless, Plaintiffs assert that the use of amortized R & D costs for COP calculations in administrative reviews of other products is equally appropriate for the subject DRAMs in this administrative review. (Id. at 25.) Plaintiffs cite to Micron v. United States, where “this Court expressly recognized in a semiconductor context that amortizing R & D costs ‘reasonably reflects’ the cost of production.” (I. (citing Micron Tech., Inc. v. United States, 893 F.Supp. 21, 28 (1995) (“Micron I”)).) Commerce also rejected Plaintiffs’ accounting practice of indefinitely deferring certain R & D costs for merchandise not produced in that year because, according to Commerce, the practice did not reasonably reflect the COP of the subject merchandise. Final Decision Memorandum at 9 (Pub.Doc. No. 72) (Def.’s Pub.App. Ex. 7). Plaintiffs claim that the indefinite deferral of certain R & D costs until the project is commercialized is permissible because such deferral is allowed under Korean GAAP and International Accounting Standard No. 9. (Id. at 25-26.) Further, Plaintiffs argue that indefinite deferral of R & D costs more accurately reflects the actual costs of producing DRAMs because such deferral matches R & D costs with the period of time in which that research and development generates revenue. (Id. at 26.) Plaintiffs rely on International Accounting Standard No. 9 to support their position that deferring R & D costs does not distort the COP. (Id. at 23, 25-26.) Plaintiffs argue that “[International Accounting Standard No.] 9 states that deferral of R & D costs is required in order to match the expenses with the benefits derived from the R & D expenditures, and therefore ‘[a]mortization commences when the product or process is available for sale or use.’ ” (Id. at 23 (quoting International Accounting Standard No. 9 at para. 21).) Plaintiffs assert that there is no distortion of COP because deferring R & D costs until benefits are realized more accurately matches future revenue to expenses. (Id. at 24.) Plaintiffs add that expensing costs when incurred creates the very distortions that Commerce is concerned about by “artificially depressing] profits prior to the introduction of a new product and artificially inflating] profits in production years.” (Id.) Thus, Plaintiffs argue, Commerce is distorting R & D costs for this POR by not permitting Plaintiffs to defer these costs. (Id. at 22-23.) B. Defendant’s Contentions Defendant argues that Plaintiffs’ use of different accounting methods and their practice of indefinitely deferring certain R & D costs created distortions in their reported R & D costs for the POR. (Def.’s Br. at 15.) Therefore, Defendant asserts that Commerce properly rejected Plaintiffs’ reported R & D costs and properly substituted R & D costs actually incurred during the POR for the COP calculation. (Id.) According to Defendant, the principle reason that Commerce rejected Plaintiffs’ amortized R & D costs is the numerous changes in accounting methods that Plaintiffs employed during the administrative reviews of the subject merchandise. (Id.) Defendant contends that Plaintiffs’ use of different accounting methods from year to year distorted their reported R & D costs for dumping purposes. (Id. at 15-20.) Defendant contends that Plaintiffs’ reliance on the fact that Plaintiffs’ accounting practices are in accordance with prior rulings of this Court and the Korean GAAP is misplaced. (Id. at 15.) Defendant states that § 1677b(f)(1)(A) directs Commerce to rely on the records of the exporter or the producer of the subject merchandise “if such records are kept in accordance with the generally accepted accounting principles of the exporting country (or the producing country where appropriate) and reasonably reflect the costs associated with production and sale of the merchandise.” (Id. at 15-16 (citing 19 U.S.C. § 1677b(f)(1)(A)).) Defendant also cites to the Statement of Administrative Action accompanying the Uruguay Round Agreements Act (“SAA”), which emphasizes that “[t]he exporter or producer will be expected to demonstrate that it has historically utilized such allocations, particularly with regard to the establishment of appropriate amortization and depreciation periods and allowances for capital expenditures and other development costs.” (Id. at 16 (citing SAA, Pub.L. Doc. 103-465 at 834 (1994), reprinted in 1994 U.S.C.C.A. 4040, 4172).) Defendant concedes that Commerce has accepted a respondent’s accounting methodology where that methodology is in accordance with the home country’s GAAP. (Id.) Defendant also concedes that Plaintiffs’ method of amortizing and deferring R & D costs is permissible under Korean GAAP, as is their previous method of ex-pensing all R & D costs in the year incurred. (Id. (citing Final Decision Memorandum at 8 (Pub.Doc. No. 72) (Def.’s Pub.App. Ex. 7)).) However, as indicated in the Final Decision Memorandum, Defendant contends that Commerce found Plaintiffs’ use of different accounting methods created distortions in the dumping calculations. (Id. at 16-20 (citing Final Decision Memorandum at 8-9 (Pub. Doc. No. 72) (Def.’s Pub-App. Ex. 7)).) Because of these distortions, Defendant asserts that the R & D costs reported by Plaintiffs do not reasonably reflect the cost of producing the subject merchandise. (Id. (citing Final Decision Memorandum at 8-11 (Pub.Doc. No. 72) (Def.’s Pub.App. Ex. 7)).) Defendant disagrees with Plaintiffs’ argument that Commerce’s treatment of R & D costs is contrary to the Court’s decision in Micron I. (Id. at 18.) Defendant argues that Micron I is not on point. (Id.) According to Defendant, Micron I dealt with Commerce recalculating R & D costs despite the fact that the accounting methods were consistent with Korean GAAP. (Id.) Here, Commerce rejected the reported amortized R & D costs because Plaintiffs changed accounting methods between review periods, not because of the accounting method selected by Plaintiffs. (Id.) Defendant argues that Plaintiffs’ “business decision” justification for changing accounting methods is irrelevant. (Id. at 19.) Defendant contends that the reasons behind the changes are not at issue; rather, the significance of the changes in accounting is that they have the effect of “distorting] the magnitude of [Plaintiffs’] R & D expenses, and thus ... artificially lowering Plaintiffs] costs [of production].” (Id.) Defendant cites to the second Micron v. United States case to support the proposition that a change from expensing R & D costs to amortizing R & D costs frustrates “the object of the cost of production exercise ... to capture ... those expenses that reasonably and accurately reflect the respondent’s actual production costs for a period of review.” (Id. (citing Micron Tech., Inc. v. United States, 23 Ct. Int’l Trade 380, 382 (1999) (“Micron II”)).) Defendant finds Plaintiffs’ accounting practice of indefinitely deferring certain R & D costs to be problematic. (Id.) Defendant notes that prior to the Fifth Administrative Review, Plaintiffs recognized R & D costs in the year incurred; however, they now defer these R & D costs indefinitely. (Id. at 20.) Defendant argues that this change is inconsistent with previous administrative reviews and is contrary to the accounting principle of conservativism: “where an expense is recognized when incurred if the probability of associated revenue is remote or uncertain.” (Id. (quoting Final Decision Memorandum at 8 (Pub. Doc. No. 72) (Def.’s Pub. PubApp. Ex. 7)).) Defendant contends that substantial evidence demonstrates that Plaintiffs’ reported amortized R & D costs and their accounting practice of indefinitely deferring certain R & D costs do not accurately reflect the COP of the subject merchandise; therefore, Commerce properly rejected Plaintiffs’ reported R & D costs and properly used actual R & D costs in the COP calculations. (Id. at 19-20.) C. Defendantr-Intervenor’s Contentions The Court finds that the majority of Defendant-Intervenor’s arguments on this issue are substantially similar to those presented by the Defendant. Therefore, the Court will not recount the arguments in their entirety, although they have been duly considered. The Court will, nevertheless, note some of the arguments that were addressed in greater detail by Micron. Micron provides a more detailed time line of Plaintiffs’ changes in accounting methods. Micron notes that Plaintiffs, while operating as Hyundai and LG Semi-con, amortized R & D costs in 1992, during the initial antidumping investigation. (Brief of Defendant-Intervenor Micron Technology, Inc. in Opposition to Plaintiffs’ Rule 56.2 Motion for Judgment Upon the Agency Record (“Def.-Int.’s Br.”) at 7 (referencing “Issues and Decision Memorandum from Holly A. Kuga, Acting Deputy Assistant Secretary, Import Administration, to Troy H. Cribb, Assistant Secretary for Import Administration,” (Nov. 3, 2000) adopted by Dynamic Random Access Memory Semiconductors of One Megabit or Above From the Republic of Korea: Final Results of Antidumping Duty Administrative Review, 65 Fed.Reg. 68,976, 68,978 (Nov. 15, 2000)).) In 1998 and 1996, LG Semicon and Hyundai, respectively, changed accounting methodologies and began expensing all R & D costs in the year incurred. (Id. at 7-8.) In 1997, Hyundai and LG .Semicon returned to amortizing R & D costs. (Id. at 8.) Micron explains that in switching “from amortizing to expensing of R & D, the companies included in their transition year R & D costs all of the accumulated R & D expenses that had been incurred in prior years but which had been capitalized and were scheduled to be amortized in subsequent periods. This transition resulted in the companies recording unusually high R & D costs in the transition year (all R & D incurred in that year plus previously un-amortized R & D from prior years).” (Id. at 7-8.) Micron observes that in the First Administrative Review, this Court agreed with LG Semicon’s argument that the transition from amortizing to expensing “caused the R & D expenses as recorded in LG [Semiconj’s accounts for 1993 to be distortive for antidumping purposes.” (Id. (discussing Micron II, 23 Ct. Int’l Trade at 380-81).) Micron disagrees with Plaintiffs’ assertion that Commerce’s decision to reject Plaintiffs’ amortized R & D costs is contrary to Commerce’s practice of accepting a respondent’s reported R & D costs. (Id. at 31.) Micron asserts that Commerce’s decision to reject amortized R & D costs is “in accord with its established practice” in semiconductor cases. (Id. at 31-32.) Micron notes that in the Final Decision Memorandum, Commerce observed that “for at least the last six years and throughout the course of this proceeding, [Commerce] has constantly required that respondents recognize R & D expenses in the year incurred” even though amortization of R & D expenses has been permitted in administrative reviews of other products. (Id. at 32 (citing Final Decision Memorandum at 10 (Pub.Doc. No. 72) (Def.’s Pub.App. Ex. 7)).) Micron also points to an administrative review in which a respondent changed accounting practices during the course of the administrative review and offered arguments similar to those offered by the Plaintiffs in this case. (Id. at 32-33 (citing Structural Steel Beams from Korea, Issues and Decision Memorandum (July 5, 2000)).) Micron notes that in that administrative review, Commerce similarly rejected respondent’s reported amortized R & D costs and recognized actual R & D costs for the given year. (Id. at 33.) Micron argues that Commerce properly rejected Plaintiffs’ practice of indefinitely deferring R & D costs for certain projects as distorting the COP of the subject merchandise. {Id. at 29-30.) Micron offers additional clarification of International Accounting Standard No. 9. {Id. at 30.) Micron interprets International Accounting Standard No. , 9 to distinguish between research costs and development costs for accounting purposes. {Id.) “[C]osts identified as research [must] be charged to current expenses, and costs identified as development are to be capitalized, but only if specified criteria are met, including that the ‘costs attributable to the product can be separately identified and measured reliably.’ ” {Id. at 30-31) (citing The IASC-U.S. CompaRison Project: A Report ON the Similarities and Differences between IASC Standards and U.S. GAAP 174 (Financial Accounting Standards Bd., 1996) (emphasis omitted).) Micron explains that if the criteria for costs identified as “development” are not met, International Accounting Standard No. 9 requires that those costs be expensed in the year incurred. {Id. at 31.) Micron notes that Plaintiffs did not demonstrate that their R & D costs in 1999 could be tied to future revenues with any reasonable degree of certainty. {Id.) Thus, according to Micron, even under International Accounting Standard No. 9, Commerce properly rejected Plaintiffs’ practice of indefinite deferral of certain R & D costs. {Id.) D. Analysis This Court holds that Commerce’s rejection of Plaintiffs’ amortized R & D costs and Commerce’s rejection of Plaintiffs’ indefinite deferral of certain R & D expenses are unsupported by substantial evidence and are otherwise not in accordance with law. The following analysis will first address Commerce’s treatment of Plaintiffs’ amortized R & D costs and will then discuss Commerce’s refusal to allow Plaintiffs’ deferral of certain R & D expenses. First, this Court finds that Commerce’s decision to expense R & D costs in the COP calculation is not supported by substantial evidence or otherwise in accordance with law. Accordingly, this issue is remanded to Commerce to reconsider and further explain why the use of Plaintiffs’ amortized R & D costs would not reasonably and accurately reflect Plaintiffs’ actual R & D expenses for this POR, and to identify what distortions, if any, would arise in the COP calculation if amortized R & D costs were used. This Court finds that Commerce faces similar problems in this case as it did in Micron I. In Micron I, the Court held that Commerce failed to provide an adequate explanation for its decision to reject the respondent’s amortized R & D costs that were consistent with Korean GAAP. Micron I, 898 F.Supp. at 28. The Court, in Micron I, noted the established principle that “Commerce will apply the GAAP of the home country if it is satisfied that those principles reasonably reflect all of the costs associated with production of the subject merchandise.” Id. (citing Camargo Correa Metais, S.A. v. United States, 17 CIT —, 1993 WL 366964, *2 (Aug. 13, 1993) and providing numerous administrative reviews exemplifying Commerce’s practice of accepting reported amortized costs). Here, as in Micron I, the Court finds that Commerce has failed to articulate sufficient reasoned analysis to justify Commerce’s departure from its established practice of amortizing R & D costs when such amortization is supported by the GAAP of the home country. See id. at 29. Again, Commerce rejected Plaintiffs’ reported R & D costs despite Commerce’s own acknowledgment that Plaintiffs applied Korean GAAP-consistent accounting practices and the reported costs were verified. This Court finds that Commerce failed to provide sufficient explanation or point to evidence in the record to support its decision to reject Plaintiffs’ reported amortized R & D costs. Defendant and Micron argue that Commerce has an established practice to expense R & D costs in the year incurred in administrative reviews dealing with the semiconductor industry. Final Decision Memorandum, at 10-11 (Pub.Doc. No. 72) (Def.’s Pub.App. Ex. 7); (Def.-Int.’s Br. at 32 n. 66.) Defendant’s and Defendant-Intervenor’s authorities do not support their assertion. The administrative reviews cited by Defendant to demonstrate this established practice do not provide any reasoned support for Commerce’s position. In those administrative reviews, Commerce expensed R & D costs, but never discussed whether or not the respondents in those reviews amortized the costs in their financial statements. See, e.g., SRAMs from Taiwan, 63 Fed.Reg. at 8,914, 8,925; SRAMs from Korea, 63 Fed Reg. at 8,939-40; DRAMs from Taiwan, 64 Fed.Reg. at 56,319. Commerce relies heavily on a contention that changing accounting methods across periods of review “distorts the cost calculation in an antidumping analysis.” Final Decision Memorandum at 8 (Pub.Doc. No. 72) (Def.’s Pub.App. Ex. 7). It is undisputed that Plaintiffs changed R & D accounting methods in the past, switching from amortizing to expensing and back to amortizing during the Fifth Administrative Review of the subject merchandise. Plaintiffs stopped expensing R & D costs in the year incurred in 1997, and since then have utilized amortization as their R & D accounting method. Id at 8-11; (Pis.’ Br. at 21.) As this Court stated in Micron II, “the object of the cost of production exercise is not to capture all past expenses, but rather those expenses that reasonably and accurately reflect a respondent’s actual production costs for a period of review.” Micron II, 23 Ct. Int’l Trade at 382 (emphasis added). The distortive impact of Plaintiffs’ accounting changes prior to the Fifth Administrative Review on the actual COP of the subject merchandise for this POR does not seem to be as apparent as Defendant and Micron advance. In the instant case, Commerce does not identify evidence in the record and does not show how Plaintiffs’ practice of amortizing R & D costs, after previously ex-pensing R & D costs in their accounting, distorts the reported R & D costs such that Plaintiffs’ reported R & D costs do not reasonably reflect the COP for the current POR. In the Final Decision Memorandum, Commerce stated that changes in accounting methods “can lead to distortions for antidumping purposes because the fluctuating costs tend to overstate per unit amounts in one period and understate these amounts in other periods.” Final Decision Memorandum at 9 (Pub.Doc. No. 72) (Def.’s Pub.App. Ex. 7) (emphasis added). This statement appears to be a conclusion and does not clearly indicate how Plaintiffs’ changes lead to such results. This Court finds that Micron II does not support Commerce’s position, as the facts of that case are distinguishable. In Micron II, the POR that the Court examined included the transition period when the change in accounting methods from amortization to expensing R & D costs occurred. Micron II, 23 Ct. Int’l Trade at 381. Under those circumstances, the Court held that it was distortive for Commerce to include in the COP calculations both the current year’s R & D costs and the unamortized amount of the prior year’s R & D costs. Id. The present case is the third administrative review that has taken place after Plaintiffs changed their accounting methods. On remand, the Court instructs Commerce to reconsider Plaintiffs’ assertion, supported by their “Response to the Department’s Supplemental Request for Information,” that all R & D costs that would have been amortized in 1996 and carried forward into this POR were fully taken into account prior to or within the Fifth Administrative Review, when Commerce used expensed R & D costs in the COP calculation. The Court further instructs Commerce to identify the distortions in the COP for this POR that arose as a result of Plaintiffs’ accounting practices and explain how the evidence in the record supports this finding. Second, this Court finds that Commerce failed to articulate a reasoned explanation to support its rejection of Plaintiffs’ practice of indefinitely deferring certain R & D expenses. Even though the Korean GAAP allows such an accounting practice, Defendant argues that deferring R & D costs is contrary to the “principle of conservatism in accounting.” (Def.’s Br. at 20.) Commerce does not, however, address Plaintiffs’ justification that deferring R & D costs most accurately matches future revenue to expenses. Final Decision Memorandum at 8-11 (Pub.Doc. No. 72) (Def.’s Pub.App. Ex. 7). According to the matching principle, “an expense should be recorded in the period in which the product makes its contribution to revenue with the intent to ensure that expenses and revenues are recorded in the proper period.” AK Steel Corp. v. United States, 21 Ct. Int’l Trade 1204, 1214 (1997) (referring to Donald E. Kieso & JERRY J. Weygandt, Intermediate ACCOUNTING 46 (8th ed.1995)). As stated by the Court, “[cjonservatism does not supersede the matching principle, but rather is incorporated into it as a general quality found in all information used in financial statements.” Id. at 1215. In the instant case, Commerce flatly rejected Plaintiffs’ deferral of certain R & D costs under the conservativism principle and because of the possibility that Plaintiffs will never realize revenue. Commerce made no finding that the deferral of R & D costs does not reasonably reflect the R & D costs related to the subject merchandise. For the reasons above, these issues are remanded to Commerce with instructions to offer, if it can, a reasoned explanation supported by evidence in the record of how use of Plaintiffs’ reported amortized R & D costs would not reasonably reflect Plaintiffs’ actual R & D expenses of the subject merchandise for this POR. Additionally, the Court remands to Commerce to reconsider and further explain why deferral of certain R & D costs does not reasonably reflect the R & D costs related to the subject merchandise. 2. Commerce’s Decision to Reject Product-Specific R & D Costs (Cross-fertilization of R & D) Commerce rejected Plaintiffs’ reported product-specific R & D costs, based on Commerce’s application of a theory of cross-fertilization in research and development of semiconductor products. Final Decision Memorandum at 11 (Pub.Doc. No. 72) (Def.’s Pub.App. Ex. 7). Commerce explained that cross-fertilization of research and development occurs “where R & D from one type of semiconductor product influenced another semiconductor product.” Id. at 12. A. Plaintiffs’ Contentions Plaintiffs provided Commerce with extensive records of R & D costs, in which memory and non-memory costs were segregated. (Pis.’ Br. at 29.) Plaintiffs emphasize that these records were kept in accordance with Korean GAAP and contend that the reported costs reasonably reflect the actual cost of production of the subject merchandise. (Id.) Plaintiffs assert that the records provided Commerce with the ability to allocate R & D costs on a product-specific basis; therefore, only memory product R & D costs should have been included in Commerce’s COP calculations for the subject merchandise. (Id.) Plaintiffs argue that Commerce is wrong in its cross-fertilization theory that R & D expenses associated with all semiconductor research and development yield some benefit to production of the DRAMs subject to this review. (Id. at 30.) Plaintiffs argue that Commerce did not comply with 19 U.S.C. § 1677b(e), which “specifically contemplate^] the allocation of only those costs associated with the production of the subject merchandise.” (Id. at 27-28 (citing 19 U.S.C. § 1677b(e)(1), (2)(A)).) According to Plaintiffs, “the statute does not permit [Commerce] to attribute costs incurred in producing non-subject merchandise to the cost of producing subject merchandise.” (Id. at 28.) Plaintiffs note that the SAA also instructs Commerce to focus on ascertaining the costs associated with “producing and selling the product under investigation or review,” indicating that Commerce bears a statutory responsibility to use only R & D costs that are “related directly to the subject merchandise.” (Id. (citing SAA at 835).) Plaintiffs then refer to the instructions in § 1677b(f)(1)(A), which direct Commerce to calculate the COP based on the company’s records, as long as the records are kept in accordance with the home country GAAP and reasonably reflect the actual cost of production. (Id. at 29.) Plaintiffs argue that R & D is critical to their business, and as such, they take great efforts to focus R & D resources and document the spending of R & D resources in detail, on a project-by-project basis. (Id.) Plaintiffs contend that Commerce elected to disregard the realities of the semiconductor industry and the specific actions of Plaintiffs when Commerce allocated a portion of R & D costs for all semiconductor products to the subject merchandise. (Id. at 29-30.) Plaintiffs challenge Commerce’s justification for reallocating R & D costs. (Id. at 30-33.) Commerce has advanced a cross-fertilization theory, supported by Dr. Murzy Jhabvala, a semiconductor device engineer at the National Aeronautics and Space Administration, which states that “the benefits from the results of R & D, even if intended to advance the design or manufacture of a specific product, provided an intrinsic benefit to other semiconductor products.” Notice of Final Determination of Sales at Less Than Fair Value: Static Random Access Memory Semiconductors from the Republic of Korea, 63 Fed.Reg. 8,934, 8,939-40 (Feb. 23, 1998) (“SRAMs from Korea”) (referencing Dr. Jhabvala’s report dated Sept. 8, 1997: “Cross Fertilization of Research and Development of Semiconductor Memory Devices” in support of Commerce’s theory (Def.’s Pub. App. Ex. 8)). Plaintiffs note that Dr. Jhabvala never specifically addressed the subject merchandise or considered the facts on the record in this review. (Id. at 26-27.) Rather, Dr. Jhabvala’s report was created for use in an administrative review of SRAMs. (Id. (citing SRAMs from Korea, 63 Fed.Reg. at 8,938-40).) Plaintiffs challenge the validity of Dr. Jhabvala’s support for the cross-fertilization theory further by citing to Micron I, where the Court rejected a cross-fertilization theory advanced by Commerce. (Id. at 30.) In Micron I, the Court found that “the defendant offers mere speculation, stating that ‘there is an overlap of technology between different semiconductors, the result of which is that R & D performed for one model may provide a benefit to another model.’ ... If R & D cross-fertilization ... is as widespread as defendant claims, one would expect an abundance of corroborating evidence which supports and expands upon the statement offered by the Micron official. Yet there is none.” (Id. (quoting Micron I, 893 F.Supp. at 27).) Plaintiffs argue that Dr. Jhabvala’s report fails to provide “an abundance of corroborating evidence” to support Commerce’s R & D cross-fertilization theory. (Id.) Plaintiffs state that the report was prepared for an administrative review of different subject merchandise, by an expert who never reviewed the “confidential record of the companies he purports to evaluate.” (Id. at 31.) Plaintiffs argue that, even if the report is deemed to be credible, the report addresses only “the potential cross-fertilization among memory products, not cross-fertilization between memory and non-memory products.” (Id) Plaintiffs argue that this report demonstrates the close relationship between two memory products, SRAMs and DRAMs, and acknowledge the fact that these memory semiconductors benefit from R & D conducted specifically for memory semiconductors. (Id at 32.) Plaintiffs, themselves, allocate R & D costs to both types of memory semiconductors. (Id) However, Plaintiffs assert that they make a clear distinction between memory and non-memory research, and R & D costs are accounted for separately. (Id) Plaintiffs argue that R & D conducted for non-memory devices do not benefit memory devices because of “fundamental differences in function, design, production” between the products. (Id) Plaintiffs maintain that Dr. Jhabvala’s report fails to provide Commerce with substantial evidence to support the reallocation of R & D expenses attributable to non-memory semiconductors to the R & D expenses of the subject merchandise of this administrative review. (Id at 33.) B.Defendant’s Contentions Defendant contends that all semiconductor R & D benefits all areas of semiconductor production. (Def.’s Br. at 20.) Thus, according to Defendant, Commerce properly allocated a portion of all semiconductor R & D costs over all semiconductor products, including the subject merchandise. (Id at 21 (citing Final Decision Memorandum at 13-14 (Pub.Doc. No. 72) (Def.’s Pub.App. Ex. 7)).) Defendant relies on the report prepared by Dr. Murzy Jhabvala, whose views have been relied on in other semiconductor proceedings. (Id) Defendant acknowledges that the report was prepared for an administrative review involving SRAMs from Korea. (Id (citing SRAMs from Korea, 63 Fed.Reg. at 8,938-940).) The report specifies the way in which SRAMs can benefit from general R & D activities conducted in the semiconductor industry, and Commerce reasons that such benefits extend to the subject merchandise DRAMs in this case. Final Decision Memorandum at 13-14 (Pub.Doc. No. 72) (Def.’s Pub.App. Ex. 7). In the Final Decision Memorandum, Commerce states that Dr. Jhabvala provided numerous statements to support “the existence of cross-fertilization in semiconductor R & D.” Id Defendant maintains that application of this cross-fertilization theory is not a new policy on the issue, and that Commerce has previously reallocated R & D costs for products that benefit from cross-fertilization. (Def.’s Br. at 22 n. 8.) Defendant points out that Plaintiffs failed to submit any direct evidence contradicting the evidence relied on by Commerce, emphasizing that it is Plaintiffs’ burden to create the record before the agency. (Id at 23 (citing Tianjin Mach. Imp. & Exp. Corp. v. United States, 806 F.Supp. 1008, 1015 (CIT 1992)).) Defendant discounts Plaintiffs’ contentions that R & D conducted for non-memory products does not benefit memory products as being unsupported by evidence in the record. (Id) Defendant maintains that absent contrary evidence in the record, Commerce reasonably relied on Dr. Jhabvala’s report to conclude that Plaintiffs’ R & D costs should be allocated over all semiconductor products produced. (Id) C. Defendant-Intervenor’s Contentions Because the Court finds Micron’s contentions to be analogous to those advanced by Defendant on this issue, the Court will not recount them here, though they have been duly considered. D. Analysis In Micron I, this Court rejected a cross-fertilization of R & D theory advanced by Commerce, holding that “the administrative record ... does not support Commerce’s conclusion that R & D cross-fertilization in the semiconductor industry warrants the use of aggregate data to allocate R & D costs for the subject DRAMS.” Micron I, 893 F.Supp. at 27. Commerce advanced an analogous argument in Micron I to justify using aggregate data rather than product-specific data for R & D costs used to calculate COP. Commerce argued that “because the general underlying technology is the same for all semiconductor products, the benefits from the results of [R & D], even if intended to advance the design or manufacture of a specific product, provide an intrinsic benefit to other semiconductor products.” Id. (citing Final Determination of Sales at Less Than Fair Value: Dynamic Random Access Memory Semiconductors of One Megabit and Above From the Republic of Korea, 58 Fed.Reg. 15,467, 15,472 (Mar. 23, 1993)). The evidence on the record in this case seems to differ from Micron I only with the introduction of the Dr. Jhab-vala report. Commerce relies on this report to provide the substantial evidence that the Court found was lacking in Micron I. This Court has “recognize[d] that R & D cross-fertilization may justify allocating R & D costs on an aggregate, rather than product-specific, basis.” Micron I, 893 F.Supp. at 27 (emphasis added). However, such an allocation is appropriate only “if substantial record evidence supports a determination that the subject merchandise benefits from R & D expenditures earmarked for non-subject merchandise.” Id. The Court, in Micron I, continued that “the factual premise upon which Commerce bases its choice of methodology must be supported by substantial evidence on the record,” regardless of the fact that Commerce’s methodology may have “intuitive appeal.” Id. In this case, the Court finds that Dr. Jhabvala’s report does not provide substantial evidence to support the existence of cross-fertilization of R & D for the subject merchandise. Dr. Jhabvala’s report is based upon different products produced by different parties under conditions that have not been demonstrated to be similar to those of the present case. While Plaintiffs acknowledge that there is a close relationship between memory semiconductor products, that close relationship does not necessarily lead to the conclusion that because memory products by a different company benefitted from some R & D conducted for non-memory products at that company, the subject merchandise in this review must have also benefitted from R & D activities for non-memory products produced by Plaintiffs. Commerce historically has excluded R & D expenses for non-subject merchandise where a respondent maintains product-specific R & D costs and the expense “benefitted only non-subject merchandise.” (Def.’s Br. at 22 n. 8.) It is clear that Plaintiffs have the burden of creating the record. See Tianjin Mach. Imp. & Exp. Corp., 806 F.Supp. at 1015. In building the record for the present case, Plaintiffs provided Commerce with R & D costs accounted for on a product-by-product basis. (Verification of the Response of Hyundai Electronics Industries, Co., Ltd. and Hyundai Electronics America for 1999 Administrative Review of Dynamic Random Access Memory Semiconductors of One Me