Opinion
Restani, Judge:This matter is before the court on cross motions for judgment upon the agency record, pursuant to USCIT Rule 56.2, brought by Rautaruukki Oy (“Rautaruukki” or “Importer”) and by Bethlehem Steel Corporation and U.S. Steel Corporation, aunit of USX Corporation (collectively “Domestic Producers”). The International Trade Administration, U.S. Department of Commerce’s (“Commerce”) determination under review is Certain Cut-to-Length Carbon Steel Plate from Finland, 62 Fed. Reg. 18,468 (Dep’t Commerce 1997) (final results of second antidumping duty administrative review) [hereinafter “Final Results IF’], pursuant to 19U.S.C. § 1516a(a)(2)(B) (1994). The court addresses issues raised by Rautaruukki and the Domestic Producers in their respective motions separately, in that order.
Jurisdiction and Standard of Review
This court has jurisdiction pursuant to 28 U.S.C. § 1581(c) (1994). Commerce’s final determination must be upheld unless it is “unsup*787ported by substantial evidence on the record, or otherwise not in accordance with law.” 19U.S.C. § 1516a(b)(l)(B) (1994).
I. Grade “A” Steel Plate Specifications for Matching Purposes
Facts
In Rautaruukki’s first administrative review, Commerce treated all steel plate certified as grade “A” by any national classification society as identical merchandise. See Certain Cut-To-Length Carbon Steel Plate from Finland, 61 Fed. Reg. 2,792, 2,797 (Dep’t Commerce 1996) (final results of first antidumping administrative review) [hereinafter “Final Results I”]. Commerce concluded that Rautaruukki had not sold subject merchandise at less than fair value (“LTFV”) during the review period and gave Rautaruukki a zero dumping margin. Id. at 2,797.
During the second administrative review, Commerce issued a questionnaire to Rautaruukki which described the eight product characteristics comprising the model match criteria Commerce would use to compare merchandise when calculatingthe dumping margin. Questionnaire (Sep. 14,1995), at V-14-V-16, ER. Doc. 346, Def.’s App., Tab 1, at 2-4. Commerce stated that when the third characteristic asks for plate “specification and/or grade,” id. at V-14, Def’s App., Tab 1, at 2, “specification” refers to “broad-based characteristics of a particular type of plate,” which includes the “method of manufacture, testing requirements, heat treatment, chemical requirements, etc.” and “grade” is a subcategory of specification, which “indicates the specific mechanical and chemical characteristics of aproductconformingto a specification,” Questionnaire, at V-17, Def.’s App., Tab 1, at 5.
The explanatory notes to the model match list described the information Commerce required Rautaruukki to provide with regard to “specification and/or grade.” Commerce requested “all of the relevant mechanical and chemical characteristics which define all specifications and grades sold.” Questionnaire, at V-18, Def.’s App., Tab 1, at 6. Commerce provided a template from which Rautaruukki was to create a specification chart. Questionnaire, at V-19, Def.’s App., Tab 1, at 7. Commerce explained that, “[s]ince there are different specifications for identical steel plate sold around the world, we need some basis of comparison between U.S. standards and those of other countries.” Id. All the information and explanations Commerce provided and the requests it made were identical to those of the first administrative review. Compare Questionnaire, at V-18, Def.’s App., Tab 1, at 6, with Questionnaire, at 8, PI. Rautaruukki.’s Reply App., Tab B, at 4.
The chart Rautaruukki provided Commerce in response reported information identical to that submitted in response to Commerce’s identical request during the first administrative review. Compare Rautaruukki’s Response (Nov. 13,1995), Ex. B-5, ER. Doc. 364, Def.’s App., Tab 2, at 3, with Rautaruukki’s Response (Feb. 15,1995), PI. Rau-taruukki’s Reply App., Tab A, at 1. The chart described the chemistry and strength requirements for the specification of plate sold in the *788United States (hereinafter “AB A”), as well as the requirements for those specifications of plate sold in the home market. Rautaruukki’s Response, Ex. B-5, Def.’s App., Tab 2, at 3. Rautaruukki identified the specifications of plate sold in the home market as “equivalent specifications” to the U.S. AB A specification, including with the AB A specification BV A, GL A, LR A, NV A, and PC A1 specifications. Id. The chart described chemical and strength requirements of the combined group of home market “equivalent specifications, ” which varied at certain points from those for the AB A specification. Id.
Commerce then requested that Rautaruukki “provide requested specification and grade information for each model separately as specified in the questionnaire,” Supplemental Questionnaire (Jan. 29,1996), at 4, ER. Doc. 396, Def.’s App., Tab 3, at 2, and “explain how identical models AB A in the U.S. market and AB A in the home market have different carbon levels.” Id.
In response, Rautaruukki offered a chart which reflects “the target or ‘aim’ chemistry for each grade with identical and similar specifications,” Rautaruukki’s Supplemental Questionnaire Response (Feb. 20, 1996), at SUPP-14, C.R. Doc. 411, Def.’s App., Tab 4, at 5, entitled “List of Identical and Most Similar Specifications and Grades,” Rautaruuk-ki’s Supplemental Questionnaire Response, at SUPP-18, Def’s App., Tab 4, at 6, and a compilation of mill certificates, id., at SUPP-18, Tab 4, at 7-13. The mill certificates show some small differences in the chemical characteristics for AB A, Rautaruukki’s Supplemental Questionnaire Response, Def.’s App., Tab 4, at 9 {e.g., carbon levels 0.11 to 0.18 percent), LR A, Rautaruukki’s Supplemental Questionnaire Response, Def.’s App., Tab 4, at 11 (e.g., carbon levels 0.13 to 0.15 percent), and GL A, Rautaruukki’s Supplemental Questionnaire Response, Def.’s App., Tab 4, at 13 {e.g., carbon levels 0.13 to 0.18 percent).
Commerce diverged from its position on plate specifications in the first review, in which it had assigned one value to all plate specifications. See Final Results I, 61 Fed. Reg. at 2797. In the second review Commerce assigned one value to U.S. AB A and a different value to grade “A” plate classified by all other national societies. Final Results II, 62 Fed. Reg. at 18470-71.
Rautaruukki first learned of Commerce’s decision upon publication of the Preliminary Results. Certain Cut-to-Length Carbon Steel Plate from Finland, 61 Fed. Reg. 51901,51902 (Dep’t Commerce 1996). In its response, Rautaruukki maintained that Commerce erred in treating subject merchandise produced to different specifications as non-identical merchandise, given that Rautaruukki had submitted what it had deemed to be evidence of the identicalness of “A” grade plate, as defined by each of the various national classification societies. Rautaruukki’s *789Response to Preliminary Results (Nov. 18,1996), at 3-6, P.R. Doc. 494, Def.’s App., Tab 10, 2-5.
Commerce explained in Final Results II, 62 Fed. Reg. at 18,470-71, that it decided to treat plate classified “A” by different societies differently, notwithstanding its recognition that Rautaruukki used identical plate to meet orders governed by different classification societies. Id. Commerce noted the change from its practice in the first review, explaining that it had not become aware of the need for the change until the second review. Id. Commerce stated that it decided to make the change based upon Rautaruukki’s failure to prove the specifications of the different classification societies to be identical. Id. The second administrative review resulted in a dumping margin of 24.95 percent for Rautaruukki. Id. at 18,475.
Discussion
Rautaruukki challenges Commerce’s decision to treat U.S. grade “AB A” steel plate and all other grade “A” plate as identical merchandise, Final Results II, 62 Fed. Reg. at 18,469, contending that this decision was unsupported by substantial evidence and that this change2 was procedurally unfair, as Rautaruukki had insufficient notice and lacked adequate opportunity to submit a factual response. The court agrees.
In successive reviews, Commerce is not absolutely bound by its prior decisions. Commerce may change its position, provided that “it explains the basis for its change and providing that the explanation is in accordance with law and supported by substantial evidence.” Cultivos Miramonte S.A. v. United States, 980 F. Supp. 1268, 1274 (Ct. Int’l Trade 1997) (footnotes omitted). Thus, Commerce is not compelled to adhere to prior decisions if new arguments or facts emerge that support a different conclusion. Citrosuco Paulista, S.A. v. United States, 12 CIT 1196, 1209, 704 F. Supp. 1075, 1088 (1988). Commerce learns through experience and is allowed to refine and adjust its stance. It is still, however, required to explain its actions. Cultivos, 980 F. Supp. at 1274.
Commerce asked Rautaruukki to report its U.S. and home market sales in accordance with Commerce’s eight model match product characteristics. Questionnaire, at 1, Def.’s App., Tab 1, at 1. Based on the results of the first review, Rautaruukki treated all plate with shipbuilding grade “A” as identical merchandise. See, e.g., Rautaruukki’s Response, at B-5, Def.’s App., Tab 2, at 3. Commerce found Rautaruukki’s reporting unsatisfactory and ambiguous and requested additional information and clarification. Supplemental Questionnaire, at 4, Def’s App., Tab 3, at 2. Commerce, however, did not make it clear that the identicalness of all grade “A” plate was at issue.
Apparently minor chemical composition variations shown in the mill certificates Rautaruukki submitted, see Rautaruukki’s Supplemental *790Questionnaire Response, Def.’s App., Tab 4, at 9, 11, 13, are the only record evidence upon which Commerce might find that the specifications for the various grades differed. Commerce did not request, and respondent did not furnish independently, any nonsubjective evidence from which Commerce could determine the significance of those differences. Rautaruukki provided only its own assurance that the difference was not commercially significant. While this is evidence, Commerce is not required to accept it. But neither must the court regard as substantial evidence seemingly nominal differences in chemical composition, the significance of which Commerce has not explained.
At oral argument, defendant stated that Commerce had based its decision to consider plate classified “A” by different classification societies as non-identical, not upon the mill certificates, but upon Rautaruukki’s failure to provide evidence of identicalness. Apparently Commerce had expected Rautaruukki to respond to its questionnaires by submitting complete specifications of each classification society. The court finds that Commerce did not specifically request this information, and that Rautaruukki reasonably attempted to comply with Commerce’s requests, as it understood them. Rautaruukki likely concluded, based on the past history of the antidumpingorder, that Commerce’s concerns in its two questionnaires and other correspondence related only to coding and did not require a further showing by Rautaruukki of the identicalness of the classification societies’ specifications. See Questionnaire, at V-14, Def.’s App., Tab 1, at 2; see also Supplementary Questionnaire, at 4, Def.’s App., Tab 3, at 2; and see Commerce’s Sales Verification Report (Sep 5, 1996), at 4, C.R. Doc. 452, Def.’s App., Tab 7, at 3.
Apparently, Commerce made its decision based on Rautaruukki’s failure to meet an evidentiary burden the existence of which Commerce had not made clear. To that extent, the manner in which Commerce made its decision to treat U.S. AB A plate as not identical to plate meeting other classification societies’ “A” specification was lacking in procedural fairness. While Commerce may correct what it discovers to be a prior flaw, see Cultivos, 980 F. Supp. at 1273-74, that right is balanced against respondent’s right to procedural fairness, see Citrusuco, 12 CIT at 1209, 704 F. Supp. at 1088 (agency must explain departures from prior decisions to insure consistent application of statute); see also Budd Co. v. United States, 14 CIT 595, 602, 746 F. Supp. 1093, 1099 (1990) (fairness central to antidumping law enforcement by Commerce).
Because Commerce has indicated it lacked sufficient information to make a factually supportable determination on this issue, remand is ordered to provide respondent an opportunity to submit the necessary information. If Commerce relies on nominal differences in chemical composition, it must explain why they are significant.
*791II. Wide Flats and Normal Cut-to-Length Comparison
Facts
■ Commerce compared normal cut-to-length carbon steel plate sold to the U.S. market with wide flats sold in the home market, having concluded that the distinction was not and should not have been one of the model match criteria. Final Results II, 62 Fed. Reg. at 18471. Rauta-ruukki maintained in its communication with Commerce that cut-to-length carbon steel plates and wide flats are distinct, supplying a products list, Rautaruukki’s Products List, C.R. Doc. 454, Pl. Rautaruukki’s App., Tab 15, at 1, and a “report [that] lists the two first characters of product control numbers and shows how they are created,” See Report (Nov. 7,1995), PI. Rautaruukki’s App., Tab 16, at 1. Rautaruukki wrote to Commerce to argue the importance of the demonstrated differences, Letter from Rautaruukki to Commerce (Aug. 2, 1996), at 3, C.R. Doc. 443, PI. Rautaruukki’sApp., Tab 17, at 3, and submitted documentation indicating that the raw material for wide flats is basic cut-to-length plate, Hot-Rolled-Steel Products — Production Programme (Jan. 1995), PR. Doc. 356, PI. Rautaruukki’sApp., Tab 2; Rautaruukki Steel Guide Hot-rolled Plate and Sheet, PR. Doc. 356, PI. Rautaruukki’sApp., Tab 19, at 1. Rautaruukki attempted to show that, although wide flats may have some of the same physical characteristics as basic cut-to-length, they are manufactured by different processes and have different uses.
Commerce did not change its model match program but did address Rautaruukki’s arguments, reiterating that “[wjhether or not subject merchandise is beveled or wide flat is not a model match criterion.” Final Results II, 62 Fed. Reg. at 18471. Commerce added that Rautaruuk-ki had sua sponte deviated from Commerce’s model match program in its response, without suggesting to Commerce that it change the program, giving proper explanations, or submitting information on the record to demonstrate the relevance of beveling or wide flats as a model match criterion. Final Results II, 62 Fed. Reg. at 18471.
Discussion
Rautaruukki submits that Commerce erred in comparing basic cut-to-length carbon steel plate with wide-flats by failing to distinguish them in its model match program. The court finds Commerce’s determination that differences between wide flats and normal cut-to-length plate do not necessitate a distinct model match criterion to be supported by substantial evidence in the record.
Commerce has the discretion “to choose the manner in which ‘such or similar’ merchandise shall be selected,” Koyo Seiko Co., Ltd. v. United States, 66 F.3d 1204, 1209 (Fed. Cir. 1995), provided its actions are reasoned and supported by substantial evidence on the record, see Rhone-Poulenc, Inc. v. United States, 927 F. Supp. 451, 454 (Ct. Int’l Trade 1996). Respondent bears the burden of demonstrating the necessity for *792a change in Commerce’s model match program. Koyo Seiko Co., Ltd. v. United States, 19 CIT 1272, 1277, 905 F. Supp. 1112, 1117 (Ct. Int’l Trade 1995) (citing Timken Co. v. United States, 11 CIT 786, 804, 673 F. Supp. 495, 513 (1987)). Respondent must also make timely explanations to ensure that Commerce understands and correctly uses the data it submits. Zenith Elecs. Corp. v. United States, 18 CIT 870, 880 (1994).
Commerce did not find the evidence Rautaruukki provided that wide flats are physically distinct from cut-to-length plate sufficient to alter its program. Having failed to persuade Commerce of the relevance of its evidence, Rautaruukki could not then unilaterally change the model match methodology, as it appears to have tried to do through its data coding. It was incumbent upon Rautaruukki instead to demonstrate that a change in Commerce’s model match was required. See Koyo Seiko, 905 F. Supp. at 1117. Rautaruukki did not clearly indicate to Commerce the need to amend the model match program. Instead, Rautaruukki failed to comply with Commerce’s model match methodology and explained its own reasons for deviating from Commerce’s model. Letter from Rautaruukki to Commerce, at 3, PI. Rautaruukki’s App., Tab 17, at 3.
Rautaruukki showed that the products were different, but did not explain the relevance of those differences to demonstrate why Commerce needed to alter its model to avoid comparing them. Rautaruukki simply assigned different code numbers to wide flats and only later explained this coding deviation in a letter to Commerce in response to Domestic Producers’ comments on alleged deficiencies in Rautaruukki’s questionnaire response. Letter from Rautaruukki to Commerce, at 3, PI. Rau-taruukki’s App., Tab 17, at 3. Respondent argued that wide flats “require separate processing and handling on a different product line than basic cut-to-length plate. The raw material to wide flats and beveled plates is cut-to-length plate.” Id. This explained why Rautaruukki considered wide flats different from normal cut-to-length plate but did not show how the differences would distort the model match or required Commerce to change it. The court therefore finds that Rautaruukki did not satisfy its burden and sustains Commerce’s comparison of wide flats and normal cut-to-length plate.
III. Selection of Partial Facts Available for Normal Cut-to-Length Plate
Facts
In response to Commerce’s questionnaire relating to the calculation of cost of production, Questionnaire, at 1, Def.’s App., Tab 1, at 1, Rauta-ruukki indicated that it uses an “actual” cost accounting system, which generates information used to calculate weighted-average production costs, from which base costs are derived. Cost Verification Report (Sep. 11, 1996), at 12-13, C.R. Doc. 454, Def.’s App., Tab 8, at 6-7. Extras costs, accounting for differences in quality and dimensions in various plate products, are then added to base cost to derive the total cost of the product for the specified reported sale. Id.
*793Domestic Producers commented to Commerce that Rautaruukki’s method of calculating manufacturing costs was “wholly inadequate.” Domestic Producers’ Comments on Rautaruukki’s Questionnaire Response (Jan. 11, 1996), at 22, ER. Doc. 393, Def.’s App., Tab 11, at 4. Commerce requested that, at verification, complete supporting documentation be available for some merchandise so the verifiers might create an accurate and thorough trail of the process of cost allocation. Letter from Commerce to Rautaruukki (May 20,1996), at 1, C.R. Doc. 17, PI. Domestic Producers’ App., Tab 3, at 1. At verification, Rautaruukki provided Commerce access to its on-line computer system and a cost extras book, neither of which was specifically for the period of review (“POR”). See Final Results II, 62 Fed. Reg. at 18,472-73. Rautaruukki indicated that it does not maintain a log or any documentation indicating cost changes for extras from one period to another. Commerce Cost Verification Report, at 4, Def.’s App., Tab 8, at 4.
After verification, Commerce indicated that it had “reviewed the base-cost build-up of this material and the weighted average cost for all production of this material during the POR. [It] also reviewed the reported ‘extras costs’ associated with this specific product.” Commerce’s Cost Verification Report, at 4, Def.’s App., Tab 8, at 4. Domestic Producers responded to Commerce’s verification report, assertingthat “Rauta-ruukki was unable to provide the necessary data to support the calculation of product-specific costs.” Domestic Producers’ Comments on Verification Reports (Sep. 17,1996), at 3, C.R. Doc. 448, Def.’s App., Tab 9, at 2. Domestic Producers claimed Rautaruukki did not maintain documentation sufficient to verify the accuracy of the cost-extras adequately, additionally noting allegedly “[cjonsistent and pervasive deficiencies” that precluded verification. Id. Domestic Producers thus urged Commerce to resort to total facts available for all plate products. Id. at 13, Def.’s App., Tab 9, at 5.
Commerce found resort to total facts available unnecessary with respect to normal cut-to-length steel plate products, because “the Department was able to tie Rautaruukki’s base costs to appropriate financial and accounting documentation” and had found these base costs represented “the largest portion of Rautaruukki’s total cost.” Final Results II, 62 Fed. Reg. at 18472-73. Commerce acknowledged that the documentation Rautaruukki did provide for cost extras was not for the period of review. Id. at 18473. Commerce further noted that, at verification, of the 48 different reported cost extras it compared to the costs listed in the cost extra book, 16 had shown slight inconsistencies of one or two Finnish Marka (“FIM”). Id. at 18473. As a result of these small discrepancies, which Commerce considered negligible, and because Commerce was unable to verify cost extras for the exact period in question, it used partial facts available for the normal cut-to-length cost extras. Id. Commerce used as partial facts available the highest reported cost extras for products that were not beveled or wide flat and the highest reported cost extras for painted and non-painted plate. Id.
*794Discussion
Domestic Producers challenge Commerce’s resort to partial rather than total facts available to calculate Rautaruukki’s dumping margin in the second review. Commerce may resort to “facts otherwise available in reaching the applicable determination,” 19 U.S.C. § 1677e(a) (1994), if an antidumpingrespondent provides information which cannot be verified, 19 U.S.C. § 1677e(a)(2)(D). There is no clear statutory guidance on when Commerce should fill in with partial facts available information and when it should use a substitute margin. Thus, the court looks for reasoned decision making supported by evidence.
At verification, Commerce determined Rautaruukki’s base costs, all of which were verifiable, to comprise the largest portion of its total costs. Final Results II, 62 Fed. Reg. at 18472-73. After verification, with respect to cost extras, Commerce treated wide flats and beveled plate products differently from cut-to-length plate. Id. at 18473. Commerce agreed with Domestic Producers that total facts available should be applied to cost extras for wide flats and beveled plate products, for which no cost extras could be verified. Id. at 18472-73. By contrast, finding only slight discrepancies and having what it considered unalarming problems linking the cost extra data with the period of inquiry, Commerce applied partial facts available for cost extras for normal cut-to-length plate. Id. at 18473. Commerce recognized that its linking problem at verification resulted not from any substantial variation in the magnitude of the cost extras but rather from the particularities of Rautaruukki’s record keeping. Id.
The cost verification report is a detailed account of Commerce’s inquiry into the cost extra data for cut-to-length plate. See Commerce’s Cost Verification Report, at 12-13, Def’s App., Tab 8, at 6-7. Commerce correctly concluded that this case is not akin to Certain Cut-to-Length Carbon Steel Plate From Sweden, 61 Fed. Reg. 51,898 (Dep’t Commerce 1996) (preliminary results) [hereinafter “Sweden”]. There Commerce was unable to verify “numerous and fundamental aspects of the respondents’ responses”. Final Results II, 62 Fed. Reg. at 18473; see also Sweden, 61 Fed. Reg. at 51,899. Here, Commerce’s experience simply does not comport with Domestic Producers’ re-characterization of the experience as wrought with difficulty. For these reasons the court finds Commerce’s decision not to apply total facts available with respect to normal cut-to-length plate to be both reasonable for a generally cooperative respondent and supported by substantial evidence in the record.
IV Revision of Total Facts Available for Wide Flats and Beveled Plate
Facts
During the second administrative review, for facts available for wide flats and beveled plate products, Commerce used 32.80 percent ad valo-rem, id. at 18475, the weighted-average duty rate from the original LTFV investigation, Certain Cut-to-Length Carbon Steel Plate From *795Finland, 58 Fed. Reg. 44165, 44165 (Dep’t Commerce 1993) (order and amendment to final determination) [hereinafter “Final Determination”]. The original LTFV determination was appealed. Subsequent to Commerce’s decision in the second administrative review, this Court affirmed Commerce’s remand determination, which resulted in a final weighted-average dumping margin of40.36 percent ad valorem. Rautaruukki Oy v. United States, Slip Op. 97-56, No. 93-09-00560, 1997 WL 260029 (Ct. Int’l Trade May 13, 1997); Certain Cut-to-Length Carbon Steel Plate from Finland, 62 Fed. Reg. 55782, 55783 (Dep’t Commerce 1997) (amended final determination).
Discussion
Domestic Producers request, and the government consents to, a remand for Commerce to correct its dumpingmargin calculation to reflect a subsequent change in the duty rate it used for total facts available. Domestic Producers request that Commerce re-calculate the weighted-average dumping margin based on the final weighted-average dumping margin of 40.36 percent ad valorem.
Commerce may not base a facts available determination on discredited facts. See D&L Supply Co. v. United States, 113 F.3d 1220, 1222-23 (Fed. Cir. 1997). Accordingly, where total facts available are needed here, the court orders Commerce to apply, in lieu of the original 32.80, the revised weighted-average dumping margin of 40.46 from the LTFV investigation. Final Results II, 62 Fed. Reg. at 18475.
Conclusion
The court remands to Commerce to obtain additional grade “A” plate information from respondent and to reconsider its decision on identicalness. The court sustains Commerce’s decision to compare wide flats with normal cut-to-length plate in its model match program. The court sustains Commerce’s use of partial facts available with regard to normal cut-to-length plate. The court further instructs Commerce to use, for total facts available during its recalculation on remand, the revised weighted-average duty rate of 40.46 ad valorem.
The ABA specification/grade is establishedby the American Bureau of Shipping; GL A by Germanischer Lloyd; LR A by Lloyd’s Register of Shipping; NV A by Det Norske Veritas; BV A by Bureau Veritas; and PC Aby Russian Register of Shipping. Id.
Defendant objects to Rautaruukki’s characterization of Commerce’s decision as a change in methodology. This nomenclature is irrelevant to whether Commerce had evidentiary support for its decision and whether the change was made in a procedurally fair manner: