Aimcor v. United States

OPINION WALLACH, Judge. I INTRODUCTION This case comes before the Court on Plaintiffs’ Motion For Judgment Upon The Agency Record (“Plaintiffs’ Motion”). Plaintiffs challenge the decision of the International Trade Administration of the U.S. Department of Commerce (“Commerce” or “the Department”) not to apply its methodology for hyperinflationary economies or otherwise adjust the direct cost-of-materials reported by Defendants Intervenor, Companhia de Ferro Ligas da Bahia (“Ferbasa”), to account for hyperinflation. For the reasons below, the Court finds Commerce’s decision not to be supported by substantial record evidence, since (a) Commerce failed to explain why, in contrast to previous examinations, it would not examine whether hyperinflation distorted costs incurred before the relevant reporting period; (b) Commerce appears to have erred in its assumptions concerning hyperinflation in Brazil and Ferbasa’s inventory turnover period; and (c) Commerce failed to address certain record evidence which indicates that Fer-basa’s costs were distorted by hyperinflation. Accordingly, a further remand is necessary to allow Commerce to remedy its apparent errors or, where appropriate, provide further explanations. *1250 II BACKGROUND Underlying Plaintiffs’ Motion is Commerce’s administrative review in Ferrosili-con From Brazil; Final Results of Anti-dumping Duty Administrative Review (“Final Results ”), 61 Fed.Reg. 59,407 (1996), which covered exports of ferrosili-con by Ferbasa for the period August 16, 1993, through February 28, 1995. In this review, AIMCOR and SKW Metals & Alloys, Inc. requested that Commerce use its hyperinflationary methodology in measuring the cost-of-production of Ferbasa’s home-market sales. Application of this methodology would have entailed using the replacement cost of Ferbasa’s raw materials in inventory, as opposed to the historic or “normal inventory value” of such materials, in valuing Ferbasa’s cost-of-production. Final Results, 61 Fed.Reg. at 59,-408. According to Plaintiffs, such action was necessary to keep Ferbasa’s reported cost-of-materials from being distorted by hyperinflation, since the materials in Fer-basa’s inventory were valued at the time of purchase, but not used in production until a later time. Id. at 59,407. In the Final Results, Commerce rejected Plaintiffs’ suggestion, finding that, although Brazil experienced hyperinflation during the review period as a whole, it did not experience hyperinflation during the six months of the period of review (September 1, 1994, through February 28, 1995) in which the home-market sales at issue took place. Id. at 59,408. Further, Commerce found Plaintiffs’ argument that Ferbasa’s material costs were distorted by inventory bought during a period of hyperinflation to be “speculative and not supported by facts on the record.” Id. Commerce noted that because the home market sales in question occurred fully two months after the period of hyperinflation ended (in June 1994), and because Ferbasa had an inventory turnover rate of approximately one month, it could reasonably conclude that Ferbasa’s “costs were not distorted by inflation.” Id. Despite this initial rejection of Plaintiffs’ requested adjustments, on June 18, 1998, Commerce asked that the Court remand this action so that it could reconsider using its hyperinflationary methodology. See Defendant’s Motion For A Remand of June 18, 1998. Commerce requested this remand because, upon review of the record, it found that Brazil’s economy had experienced hyperinflation through at least the beginning of August 1994, and not, as it initially found in the Final Results, through June 1994. Id. at 5. Separately, Commerce also requested a remand so that it could correct and review its treatment of certain kinds of interest income. Id. at 6. By Order dated June 24, 1998, the Court granted Defendant’s Motion and remanded this case for further consideration of these issues. On December 7, 1998, Commerce issued its Final Results Of Redetermination Pursuant To Court Remand (“Redetermi-nation”). In the Redetermination, Commerce again found that, although the Brazilian economy (on average) experienced hyperinflation over the period of review, application of its hyperinflationary methodology was inappropriate for valuing Fer-basa’s cost-of-production. Specifically, Commerce stated as follows: During the administrative proceeding, Ferbasa notified the Department at an early stage of this review that it made only one U.S. sale of ferrosilicon during the POR [period of review] (i.e., that sale took place in December 1994). Given this single sale, Ferbasa requested the Department’s permission to report cost and sales information for only those home market sales of subject merchandise that were made contemporaneously with the reported U.S. sale. Because under our matching methodology we would analyze only home market sales made during the window of contemporaneity [sales within 90 days before and 60 days after the month of the U.S. sale], we allowed Ferbasa to report only six months worth of cost and home market sales data. *1251 We agree with AIMCOR that the annual inflation rate in Brazil during the entire 18-month POR was hyperinfla-tionary. However, for purposes of our analysis, we must consider the particular circumstances of this case, namely, that we decided to allow Ferbasa to report home market sales and cost data for a limited six-month period covering September 1994 through February 1995. Thus, our analysis must focus on this limited period, as opposed to the entire POR. The record evidence in this case demonstrates, and AIMCOR does not dispute this fact, that the inflation rate for the six-month period for which Fer-basa reported its costs is only 9.9 percent, or 21 percent per annum. We thus determine that the six-month reporting period in this case is not hyperinflation-ary. Redetermination at 4-5 (citations and footnotes omitted). Because Commerce found that the six-month reporting period was not hyperinflationary (and, accordingly, that application of its hyperinflationary methodology was not necessary), it continued to rely upon Ferbasa’s historical cost data in calculating its cost-of-production. Id. at 19. In the Redetermination, Commerce also rejected Plaintiffs’ argument that, notwithstanding the lack of hyperinflation during the six-month reporting period, Ferbasa’s reported costs should be adjusted to account for hyperinflation prior to the reporting period. First, Commerce noted that its practice is to look to the relevant reporting period to determine whether an economy is hyperinflationary, and not to shift the reporting period to adjust for inflation in a prior period. Id. at 19. Even if Commerce had accepted Plaintiffs’ argument that it should take into account Ferbasa’s 37-day inventory turnover period, however, Commerce noted that “the rate of inflation during this period would not exceed 50 percent, and therefore, would not be considered hyperinflation-ary.” Id. at 20. In this regard, Commerce also observed that “[although the actual inventory turnover period is 37 days, the month of August covers 31 of the 37 days. It is therefore reasonable to state that the September costs are based on August prices.” Id. at 20 n. 5. Second, Commerce disagreed with the conclusions Plaintiffs drew from the evidence at hand. Examining Plaintiffs’ argument that Ferbasa’s input purchase prices for the six-month reporting period greatly exceeded the cost of its material inputs consumed during this period, Commerce stated: We do not agree with AIMCOR’s implied conclusion since it is already clear from the record that inflation in Brazil during the six-month reporting period was not hyperinflationary. Moreover, the values cited by AIMCOR appear to reflect total consumption value and total purchases value without regard to quantities. Therefore, these figures are irrelevant to any analysis of distortion of cost data resorting from inflation. Id. at 20 (citation omitted). In . addition to its findings concerning hyperinflation, and in accordance with the Court’s Order of June 24, 1998, in the Redetermination Commerce deducted from Ferbasa’s claimed offset to financial expenses both the interest income it collected for late payments on outstanding accounts receivables and the income Fer-basa derived from currency fluctuations that positively affected its accounts receivables. Id. at 9-11, 15. 1 Also per the *1252 Court’s instruction, Commerce reexamined the other categories of interest income reported by Ferbasa and found that, in all instances, Ferbasa failed to adequately demonstrate that the income at issue was short-term interest income that should be used to offset its financial expenses in the calculation of cost-of-production and constructed value. Id. at 11-15, 24-32. In so finding, Commerce reversed the position it had taken in the Final Results. See id. at 8 (citing Final Results, 61 Fed.Reg. at 59,412-13). Because Plaintiffs claim that Commerce has now “correctly made no offset to Fer-basa’s financial expenses,” and no longer challenges this aspect of the Final Results, the Court does not address this issue in the analysis below. See Supplemental Brief In Support Of Plaintiffs’ Motion For Judgment Upon The Agency Record (“Supplemental Brief’) at 16. Ill DISCUSSION A STANDARD OF REVIEW In reviewing Plaintiffs’ challenge, the Court “shall hold unlawful any determination, finding, or conclusion found ... to be unsupported by substantial evidence on the record, or otherwise not in accordance with law.” 19 U.S.C. § 1516a(b)(1)(B) (1994). Substantial evidence is something more than a “mere scintilla,” and must be enough evidence to reasonably support a conclusion. Primary Steel, Inc. v. United States, 17 CIT 1080, 1085, 834 F.Supp. 1374, 1380 (1993); Ceramica Regiomontana, S. A. v. United States, 10 CIT 399, 405, 636 F.Supp. 961, 966 (1986), aff'd, 810 F.2d 1137 (Fed.Cir. 1987). “As long as the agency’s methodology and procedures are reasonable means of effectuating the statutory purpose, and there is substantial evidence in the record supporting the agency’s conclusions, the court will not impose its own views as to the sufficiency of the agency’s investigation or question the agency’s methodology.” Ce ramica Regiomontana, S.A., 10 CIT at 404-5, 636 F.Supp. at 966. Further, should the Court be called upon to review Commerce’s construction of a statute which it administers, the Court’s initial inquiry will be to determine “whether Congress has directly spoken to the precise question at issue.” Chevron, U.S.A, Inc. v. Natural Resources Defense Council, Inc., 467 U.S. 837, 842, 104 S.Ct. 2778, 81 L.Ed.2d 694 (1984). “If Congress has explicitly left a gap for the agency to fill, there is an express delegation of authority to the agency to elucidate a specific provision of the statute by regulation.” Id. at 843-44,104 S.Ct. 2778. Consequently, “[t]he court will defer to the agency’s construction of the statute as a permissible construction if it ‘reflects a plausible construction of the plain language of the statute[s] and does not otherwise conflict with Congress’ express intent.’ ” Torrington Co. v. United States, 82 F.3d 1039, 1044 (Fed.Cir.1996) (citing Rust v. Sullivan, 500 U.S. 173, 184, 111 S.Ct. 1759, 114 L.Ed.2d 233 (1991)). It is under these standards that the Court considers Plaintiffs’ claims. B Commerce’s Decision Not to Examine the Entire Period of Review in Deciding Whether to Employ its Hy-perinflationary Methodology Is Supported by Substantial Record Evidence. Plaintiffs first challenge the Rede-termination on the grounds that Commerce did not adequately explain why it departed from its “long-established prac *1253 tice” of examining the entire period of review (i.e., August 16, 1993 through February 28, 1995) in deciding whether to employ its hyperinflationary methodology. Supplemental Brief at 7. According to Plaintiffs, Commerce’s current practice is to recognize an economy as hyperinflation-ary if its annual inflation rate during the period of review exceeds 50%. Id. at 7-8. Given this standard, Plaintiffs argue, Commerce should have recognized Brazil as hyperinflationary, since Brazil’s 2,618% annual inflation rate during the eighteen-month period of review greatly exceeds this 50% threshold. Id. at 8. At its heart, Plaintiffs’ argument is centered on the premise that Commerce’s “longstanding practice” is to use a replacement-cost methodology when an economy experiences hyperinflation over the entire period of review. Id. at 7. A review of the parties’ submissions and the antidumping investigations cited therein, however, indicates that Plaintiffs’ underlying premise is incorrect. For example, as an illustration of Commerce’s “practice,” Plaintiffs cite, inter alia, Commerce’s statement in Carbon Steel Pipe From Turkey that “we routinely examine the entire review period to determine whether high inflation exists.” Notice of Final Results of Anti-dumping Duty Administrative Review: Certain Welded Carbon Steel Pipe and Tube From Turkey (“Carbon Steel Pipe From Turkey”), 62 Fed.Reg. 51,629, 51,-630 (1997). A review of that investigation, however, shows that the quoted statement is qualified by the clause “although not dispositive of the issue.” Id. (“[Although not dispositive of the issue, we routinely examine.... ”). Read as a whole, this statement indicates that, while it may be Commerce’s practice to look at inflation during the entire period of review in deciding whether it should use a hyperinflation methodology, Commerce has not always viewed this factor as being determinative, as Plaintiffs seem to contend. Similarly undermining Plaintiffs’ premise is Defendant’s argument that, rather than routinely focusing on the entire review period, Commerce’s practice actually focuses on the period in which a respondent reports its cost and price information. See Defendant’s Response To Plaintiffs’ Supplemental Brief In Support Of Plaintiffs’ Motion For Judgment Upon The Agency Record (“Defendant’s Response”) at 11. According to Defendant, “[i]n some cases, the period of reporting this [price and cost] information coincides with the POR. Other times, Commerce may limit its reporting period to a 90/60 day window of contemporaneity.” Id. at 11-12 (citations omitted). To illustrate this methodology, Defendant cites the approach Commerce took in Silicomanganese from Brazil; Final Results of Antidumping Duty Administrative Review {“Silicomanganese from Brazil; Final Results ”), 62 Fed.Reg. 37,-869, 37,876 (1997). Although the period of review in this investigation covered fifteen months, the respondent was allowed to submit home-market sales and cost-of-production data for only the six months surrounding its sole U.S. sale. See Silico-manganese From Brazil: Preliminary Results of Antidumping Administrative Review, 62 Fed.Reg. 1,320, 1,322-23 (1997). As in the case at bar, because Commerce only examined cost and price data from a six-month period, it looked only to this same six-month period, and not the entire period of review, in deciding whether the use of a replacement cost (i.e., hyperinflationary) methodology was necessary. Silicomanganese from Brazil: Final ResultSj 62 Fed.Reg. at 37,876. Based on the foregoing, as well as other authority, 2 the Court doubts whether the *1254 “longstanding practice” identified by Plaintiffs is as uniform as Plaintiffs suggest. While it is clear that Commerce has repeatedly looked at inflation over an entire period of review as an important guide in deciding whether it should use a respondent’s replacement costs, it is not clear that is Commerce’s exclusive “practice.” Instead, it appears Commerce treats each investigation on a case-by-case basis, and decides whether to employ its hyperinflation methodology based on the facts before it, rather than any standard time measurement. Regardless of how one chooses to characterize Commerce’s “practice,” however, the Court finds that the explanation Commerce set out in the Redetermination provides substantial record evidence in support of its actions. Assuming, as Plaintiffs allege, that Commerce’s standard “practice” is to use a respondent’s replacement costs whenever the period of review, when viewed as a whole, is hyperinflationary, Commerce may still depart from this practice so long as (a) it articulates its reason for doing so, and (b) its explanation is supported by substantial record evidence and is otherwise in accordance with law. Queen’s Flowers de Colombia v. United States, 981 F.Supp. 617, 626 (CIT 1997). 3 As explained by the Supreme Court: [T]he agency, to engage in informal rule-making, must consider varying interpretations and the wisdom of its policy on a continuing basis. An agency is not required to establish rules of conduct to last forever, but rather must be given ample latitude to adapt its rules and policies to the demands of changing eir-cumstances. Rust v. Sullivan, 500 U.S. at 186, 111 S.Ct. 1759 (citations and internal quotations omitted). See also Hussey Copper, Ltd. v. United States, 17 CIT 993, 997, 834 F.Supp. 413, 418 (1993) (“This rule [against creating conflicting precedents] is not designed to restrict an agency’s consideration of the facts from one case to the next, but rather it is to insure consistency in an agency’s administration of the statute.”) (internal quotations omitted). In this case, Commerce adequately explained why it did not look to inflation over the entire period of review in deciding whether to apply its hyperinflationary methodology. As Commerce stated in the Redetermination, this case presented a unique situation, since Ferbasa made only one U.S. sale of ferrosilicon during the eighteen-month period of review. See Re-determination at 4. Given this fact, Commerce decided that it did not need to examine price and cost data from across the entire period of review. Id. at 5. Rath *1255 er, and in line with its long-standing practice of using a six-month window for comparing home market and export sales, 4 Commerce limited its period of review of home-market sales to the six-month period between September 1994 and February 1995. Id. Because Commerce only needed to examine price and cost data from a six-month section of the entire period of review, Commerce decided that it did not need to look at the entire eighteen months in deciding whether to use a hyperinfla-tionary methodology. Rather, to determine whether an adjustment to Ferbasa’s reported costs was appropriate, Commerce examined the inflation rate both during, and slightly before, the six-month reporting period. Id. at 5, 18-20. 5 It then reviewed the (historic) prices Ferbasa reported for its material inputs over the six-month period, id. at 6 (“[Tjhere is no pattern of direct material price increases during the six-month period.”), and it considered Plaintiffs’ argument that increasing replacement costs during the six-month reporting period were evidence of price distortions, id. at 20 (“[T]he values cited by AIMCOR appear to reflect total consumption value and total purchases value without regard to quantities. Therefore, these figures are irrelevant to any analysis of distortion of cost data resulting from inflation.”). 6 In so doing, Commerce concluded that hyperinflation had not distorted the costs reported by Ferbasa for the six-month reporting period, and that these figures could properly be used for the Department’s cost-of-production determination. See id. at 7,18-19. In light of this explanation, the Court finds that substantial record evidence supports Commerce’s alleged departure from its “practice” of using a respondent’s replacement costs whenever the period of review, when viewed as a whole, is hyper-inflationary. As Commerce noted in the Redetermination, the purpose of substituting replacement costs for the historic costs reported by a respondent is to avoid “distortions in dumping margin calculations [that] can occur when costs incurred in one *1256 period are compared to sales prices from a later period when nominal currency values have fallen significantly.” Id. at 3. In this case, Commerce essentially did no more than examine those factors which would most likely indicate whether the historic material costs reported by Ferbasa were understated due to hyperinflation and, finding that these costs likely were not so distorted, proceeded to use these figures. Given the fact that Commerce only had to compare home-market prices and costs over a six-month period, such an approach appears well-reasoned and more appropriate than looking at inflation over the entire period of review in deciding whether Fer-basa’s reported costs were understated. The Court also notes that Plaintiffs have failed to indicate how hyperinflation from an earlier part of the review period (e.g., late 1993) would have distorted the costs Ferbasa reported for the last six months of this period. In both the Final Results and the Redetermination, Commerce attached importance to the fact that Ferba-sa’s inventory turnover period was only 37 days. See id. at 20 n. 5 (noting that [i]t is therefore reasonable to state that the September costs are based on August prices.); Final Results, 61 Fed.Reg. at 59,408 (“[Biased upon the company’s inventory turnover rate of approximately one month, Ferbasa produced ferrosilicon for these sales at most approximately one month earlier (i.e., at a time when the Brazilian economy was not hyperinflationary”).). Although, as will be discussed below, Commerce appears to have erred in its assumption that Ferbasa’s inventory of material inputs, as opposed to its inventory of finished products, was actually turned over every 37 days, this record evidence nevertheless indicates that it was reasonable for Commerce to conclude that Ferbasa used its material inputs within a few months of purchase; a fact which strongly suggests that hyperinflation in late 1993 or early 1994 would have had no effect on the material costs Ferbasa reported for the period September 1994 through February 1995. Finally, Plaintiffs argue that Commerce’s decision in this case will create a “conflicting precedent” with Carbon Steel Pipe From Turkey, an investigation in which Commerce expressly rejected a respondent’s claim that it should break the period of review into discrete periods in deciding whether to apply its hyperinfla-tionary methodology. Id. at 8-9. Carbon Steel Pipe From Turkey involved an administrative review in which the respondent reported price and cost data for sales over the entire, twelvemonth period of review; a situation significantly different from the facts at bar, where Ferbasa reported prices and costs for only a six-month, non-inflationary, section of the period of review. More importantly, in Carbon Steel Pine From Turkey Commerce found that, even if it were to split the period of review into the two segments requested by the respondent, inflation during both periods would exceed the Department’s 50% hyperinflationary threshold. See Carbon Steel Pipe From Turkey, 62 Fed.Reg. at 51,630 (“[EJven if we were to split the POR into 1993 and 1994 segments as requested by Borusan, we would And high inflation to exist for the entire period, since the inflation rate was greater than 50 percent during both 1993 and 1994.”). Thus, Commerce’s decision in Carbon Steel Pipe From Turkey not to break the period of review into discrete segments was not based on an institutional practice against doing so, but was based on the particular facts of that investigation. Since those facts are significantly different from the facts in this case (in which there was a significant, non-hyperinflationary period), Commerce’s actions in Carbon Steel Pipe From Turkey do not conflict with its actions here. 7 *1257 In short, the Court finds that, even if it were to agree with Plaintiffs’ characterization of Commerce’s “long-standing practice” (which it does not), substantial record evidence supports Commerce’s decision not to follow this “practice” in the Redeter-mination. Contrary to Plaintiffs’ claims, in the Redetermination Commerce made clear that, in determining whether to make adjustments to Ferbasa’s reported costs, the question of whether there was hyperinflation over the entire period of review was largely irrelevant to the facts at hand. Rather, because this review involved prices and costs reported for only a six-month period of time, Commerce reasonably found that a narrower examination of the effects of hyperinflation was more appropriate. Accordingly, because the Court agrees that this was a reasonable and adequate explanation for Commerce’s decision, and as Commerce’s action in this instance was otherwise in accordance with law, 8 the Court rejects this aspect of Plaintiffs’ challenge. C Commerce’s Decision Not to Adjust Fer-basa’s Costs to Account for Hyperinflation Is Not Supported by Substantial Record Evidence. Plaintiffs’ second basis for challenging the Redetermination is that, regardless of whether Commerce looked to the correct period of time for determining whether to employ its hyperinflationary methodology, record evidence shows that hyperinflation actually did cause distortions in Ferbasa’s reported cost-of-materials which require adjustment. In contrast to Plaintiffs’ first argument, which concerned an alleged practice, or methodology, that Commerce used in other investigations, this argument is based on the specific facts at issue in the Redetermination. Before examining the merits of Plaintiffs’ factual challenge, the Court notes that the adjustments requested by Plaintiffs are not without precedent in Commerce’s antidumping investigations. Effectively, Plaintiffs request that Commerce adjust Ferbasa’s direct material costs to reflect the fact that, although the six-month reporting period itself was not hy-perinflationary, some of Ferbasa’s material inputs were purchased during a period of hyperinflation. In at least two other investigations, Commerce has made such an adjustment. First, in Silicon Metal From Brazil; Final Results of Antidumping Duty Administrative Review and Determination Not To Revoke in Part (“Silicon Metal From Brazil ”), 62 Fed.Reg. 1,970 (1997), Commerce found that the respondents’ cost-of-materials for the first several months of the period of review showed significant fluctuations, and that “[tjhese fluctuations occurred because these respondents consumed inventory which they had purchased during a period of hyperinflation.” Silicon Metal From Brazil, 62 Fed.Reg. at 1,972. As Commerce explained, “[bjecause the [respondents’] reported costs of materials included the cost of the beginning inventory based on historic costs, these amounts were understated by the rates of inflation that occurred from the date of purchase until June 30, 1994 [the day before the period of review began].” Id. Accordingly, Commerce re *1258 quested information on the “purchase dates, quantities, and amounts recorded in [the respondents’] July 1, 1994 beginning inventory,” so that it could revalue this inventory to account for hyperinflation. Id. Second, in Final Determination of Sales at Not Less Than Fair Value; Paint Filters and Strainers From Brazil (“Paint Filters and Strainers From Brazil”), 52 Fed.Reg. 19,181 (1987), Commerce made adjustments to a respondent’s cost-of-production to account for inflation that occurred before the period of review. In this investigation, although the Brazilian inflation rate over the six-month period of review (February 1, 1996, through July 31, 1996) was clearly not hyperinflationary, Brazil did experience hyperinflation up until that time. See Paint Filters and Strainers From Brazil, 52 Fed.Reg. at 19,184 (“The Cruzado plan was introduced as of February, 1986, and marked the end of the hyperinflationary period.”). In light of this fact, Commerce adjusted “[t]he cost of paper produced in the period prior to February, 1986 ... to reflect the replacement cost when used in a later month’s production.” Id. In so doing, Commerce explicitly rejected the respondent’s argument that it was Commerce’s policy to only use replacement costs when the inflation rate exceeds 50% during the period of review. Id. Notwithstanding these precedents, in this investigation Commerce decided not to make a similar adjustment to Ferbasa’s reported costs. As discussed in the “Background” section above, Commerce explained its decision on both institutional and factual grounds, essentially stating that neither its institutional practice, nor the record evidence before it, justified replacing Ferbasa’s reported costs with replacement costs. Upon close review, however, the Court does not find Commerce’s explanation to be supported by substantial record evidence. [1] Commerce Has Not Adequately Explained Why it No Longer Accounts for Inventory Which, Though Consumed During a Non-Hyperinflationary Period, Was Actually Purchased During a Period of Hyperinflation. As part of its justification of not substituting Ferbasa’s reported costs with replacement costs, Commerce stated the following: AIMCOR asserts that Ferbasa’s reported cost data was distorted by hyperinflation because materials used in production during the six-month reporting period were actually purchased in a hyperinflationary period prior to the cost reporting period. The Department’s practice is to determine whether an economy is hyperinflationary based on the rate of inflation during the relevant reporting period. Thus, we do not shift the reporting period to account for inflation in a period prior to the reporting period (i.e. to account for purchases consumed during the reporting period that were purchased prior to that peñod). Since the level of inflation for this period is not hyperinfla-tionary, we have rejected AIMCOR’s argument that Ferbasa’s costs are distorted by hyperinflation. Redetermination at 19-20 (emphasis added). On its face, this explanation appears to directly contradict the approach Commerce took in both Silicon Metal From Brazil and Paint Filters and Strainers From Brazil. As noted above, in these investigations, Commerce found it necessary to adjust certain historic costs reported by the respondents to account for the fact that some materials used for final sales during the periods of review were purchased prior to these periods, during times of hyperinflation. In fact, in Paint Filters and Strainers From Brazil, Commerce explicitly rejected the argument that “the Department’s policy [is] to use *1259 replacement cost only when the inflation rate exceeds 50 percent during the period of investigation.” Paint Filters and Strainers From Brazil, 52 Fed.Reg. at 19,184 (emphasis added). Based on above quoted explanation, however, Commerce appears to have departed from this approach in the Redetermination and taken the position that, regardless of the facts before it, Commerce will not consider the effects of hyperinflation on inventory purchased prior to the period under review. 9 As discussed previously, in its administration of the antidumping law, Commerce need not be forever wedded to a policy or practice that it employed in prior investigations. Before departing from a prior practice, however, Commerce must articulate the reasons why a departure is appropriate, and the explanation it provides must be supported by substantial record evidence and otherwise be in accordance with law. In this case, Commerce has not provided such an explanation. Rather, it has simply made conclusory statements such as “we do not shift the reporting period to account for inflation in a period prior to the reporting period” and “[s]ince the level of inflation for this period is not hyperinfla-tionary, we have rejected AIMCOR’s argument that Ferbasa’s costs are distorted by hyperinflation.” Redetermination at ID-EO. Because these general policy statements do not set forth any reasons why Commerce believed it appropriate to depart from its former practice, its allegedly new “practice” of not accounting for purchases consumed during the reporting period that were made prior to that period, without more, cannot serve as a basis for refusing to make the adjustments requested by Plaintiffs (assuming, of course, that such adjustments are otherwise appropriate). In light of the foregoing, the Court remands this case for further consideration by Commerce. Commerce is instructed to either (a) conform its practice in this case to that used in Silicon Metal From Brazil, 62 Fed.Reg. 1,970 (1997), and Paint Filters and Strainers From Brazil, 52 Fed. Reg. 19,181 (1987), for determining whether to make necessary adjustments to account for the effects of hyperinflation on some or all of a respondent’s reported costs, or (b) explain why it was appropriate for Commerce to depart from its former practice in this investigation. In so doing, Commerce may consider those statutory provisions which underlie Commerce’s use of its hyperinflationary methodology, and discuss how Commerce’s practice in this case constitutes a reasonable interpretation of these statutes. 10 *1260 2 Commerce Erred in its Assumptions Surrounding Hyperinflation in August 1994 and Ferbasa’s 37-Day Inventory Turnover Period. In addition to this new institutional “practice,” Commerce appears 11 to have based its decision not to make hyperinfla-tionary adjustments on record evidence which, it found, did not show that Ferba-sa’s prices had been distorted. Important in this determination, in both the Final Results and the Redetermination, was the fact that Ferbasa had a 37-day finished good inventory period. For instance, in the Final Results, Commerce stated: Petitioners’ arguments that raw materials consumed during the segment of the review period where costs are calculated may have been purchased during a period of hyperinflation is speculative and not supported by facts on the record of this case. The home market sales in question occurred fully two months after the period of hyperinflation ended. We concluded that, based upon the company’s inventory turnover rate of approximately one month, Ferbasa produced ferrosilicon for these sales at most approximately one month earlier (i.e., at a time when the Brazilian economy was not hyperinflationary). Therefore, because the record supports the conclusion that sales in question were produced in a non-hyperinflationary period, we can reasonably conclude, absent evidence to the contrary, that the costs were not distorted by hyperinflation. Final Results, 61 Fed.Reg. at 59,408 (emphasis added). In the Redetemination, Commerce similarly continued to attach significant importance to this measurement, stating that “even if the Department were to include the 37-day finished good inventory period, the rate of inflation during this period would not exceed 50 percent.” Redetermination at 20. Elaborating in a footnote, Commerce remarked that “[a]lthough the actual inventory turnover period is 37 days, the month of August covers 31 of the 37 days. It is therefore reasonable to state that the September costs are based on August prices.” Id. at 20 n. 5. While facially well-reasoned, two of the principal assumptions underlying Commerce’s analysis appear to be incorrect. First, and as Plaintiffs have repeatedly made clear, not only did Brazil experience hyperinflation through July of 1994, but record evidence indicates that inflation continued at a hyperinflationary rate through August of 1994. See, e.g., Brief In Support Of Plaintiffs’ Motion For Judg *1261 ment Upon The Agency Record at 12 & n. 41 (citing Ferbasa’s Cost Questionnaire Response of March 1, 1996, for the assertion that the monthly inflation rate for August 1994 was 7.06%); Reply Brief In Support Of Plaintiffs’ Motion For Judgment Upon The Agency Record (“Plaintiffs’ Reply”) at 6-7 (noting that the adr ministrative record contains two inflation indices which show inflation rates for August 1994 of 7.56% and 7.06%). 12 On remand, however, Commerce only considered this record evidence in passing, 13 and nonetheless concluded that hyperinflation in Brazil “continued through July 1994.” Redetermination at 6. Assuming, as appears to be the case, that Brazil did experience hyperinflation through August 1994, Commerce’s own logic dictates that Ferbasa’s costs for September 1994 were distorted by purchases made in August and late July of that year. As noted above, in the Redetermination, Commerce found that “[although the actual inventory turnover period is 37 days, the month of August covers 31 of the 37 days. It is therefore reasonable to state that the September costs are based on August prices.” Id. at 20 n. 5. Following this logic, the fact that Brazil experienced hyperinflation in both July and August of 1994 would indicate that September costs were, in fact, based on purchases made during a hyperinflationary period (late July and August 1994); a conclusion which undermines Commerce’s finding that no hyperinflationary adjustments are needed. Second, and even more fundamentally, Commerce also appears to have erred in its assumptions concerning Ferbasa’s inventory turnover period. In the Redeter-mination, Commerce appears to have' assumed that the 37-day inventory turnover period meant that Ferbasa’s material inputs were purchased, processed and sold, on average, within a 37-day period — thus meaning that the ferrosilicon Ferbasa sold in September 1994 was produced from inputs purchased in late July or August of 1994. See id. (“It is therefore reasonable to state that the September costs are based on August prices.”). 14 In their Reply, however, Plaintiffs point to two places *1262 in the record where Ferbasa made clear that the 37-day inventory turnover period is the average time during which Ferbasa turned over its inventory of ferrosilicon, the final product, and not the period during which Ferbasa turned over its inventory of materials used in producing ferrosi-licon. See Plaintiffs’ Reply at 7 and n. 22. Specifically, in response to Commerce’s request that it “[d]escribe how the foreign like product is stored prior to sales and provide the average length of time in inventory prior to sale,” Ferbasa stated that 37 days was the “average inventory turnover period for ferrosilicon.” Ferbasa’s Questionnaire Response of Sept. 21, 1995, A.R. Fiche No. 17 at 55. Likewise, and in response to a similar question, Ferbasa stated that “[t]he average period in inventory for ferrosilicon is 37 days.” Id. at 6. Based on the foregoing record evidence, the Court finds Commerce’s conclusion, that Ferbasa’s reported costs for September 1994 were based upon August 1994 input prices, not to be supported by substantial record evidence. Assuming, as seems reasonable, that Ferbasa purchased its material inputs more than a few days before they were processed into ferrosili-con, 15 the record evidence concerning the 37-day average inventory period indicates that the inputs Ferbasa used to produce the ferrosilicon sold in September 1994 were likely purchased in July of 1994, if not earlier. Further, since Commerce itself has found that Brazil suffered hyperinflation during July 1994, see Redetemni-ndtion at 6, it is fair to assume — absent evidence to the contrary- — that at least some of Ferbasa’s reported costs for the six-month reporting period were distorted by hyperinflation. This, of course, is directly contrary to the presumption Commerce appears to have employed in the Final Results and the Redetermination. See, e.g., Final Results, 61 Fed.Reg. at 59,408 (“[Bjecause the record supports the conclusion that sales in question were produced in a non-hyperinflationary period, we can reasonably conclude, absent evidence to the contrary, that the costs were not distorted by hyperinflation.”). Upon remand, Commerce is instructed to reexamine the relevant record evidence concerning Brazilian inflation in August 1994 to determine, based upon the Department’s established standards for measuring hyperinflation, whether this period of time should be considered “hyperinflation-ary.” Similarly, Commerce is to reexamine the relevant record evidence concerning Ferbasa’s 37-day inventory turnover period for ferrosilicon and reconsider, in light of this and other record evidence, 16 whether it still believes that Ferbasa’s reported costs for September 1994 were based upon August 1994 input prices. In light of its analysis of this and other record evidence, Commerce is to reconsider whether some of Ferbasa’s reported costs for the six-month reporting period were distorted by hyperinflation which occurred before September 1994 and, if so, whether all or some of Ferbasa’s costs of production should be adjusted to account for such distortions. *1263 3 A Further Remand Is Necessary So That Commerce May Reconsider All the Record Evidence as to Whether Fer-basa’s Reported Costs Were Distorted by Hyperinflation. In addition to attacking Commerce’s factual assumptions concerning the hyperinf-lationary period in Brazil and Ferbasa’s inventory turnover period, Plaintiffs also argue that “the record in this review demonstrates that the direct materials costs reported by Ferbasa are severely distorted by hyperinflation.” Plaintiffs’ Reply at 8. Through their submissions, Plaintiffs identify the following as ways in which, it claims, “clear and convincing record evidence,” id. at 10, demonstrates that the direct material costs reported by Ferbasa were severely distorted by hyperinflation: • During the September 1994 through February 1995 reporting period, Fer-basa’s monthly, per-unit, direct material costs “steadily increased [confidential information deleted]” Id. at 8. In comparison, Plaintiffs note, the inflation rate for this six-month period was less than 10%. Id. • Over the six-month reporting period, Ferbasa’s monthly direct electricity costs increased [confidential information deleted] for this period. Id. at 8-9. In comparison, Ferbasa’s reported direct material costs [increased at over] the rate of inflation. Id. at 9. • For all six months for which Ferbasa reported costs, the value of its direct materials purchased [was different from] the value of its direct materials consumed. Id. at 9. And, while these figures do not account for quantities purchased, “in order for differences in quantities to explain the [confidential information deleted] differences in these values, Ferbasa would have to have purchased [different] monthly quantities of direct materials for use in ferrosilicon production than Ferbasa consumed in the same months in producing ferrosilicon.” Id. at 9-10. • During the period September through December 1994, “the monthly differences between the value of direct materials purchased and the value of direct materials consumed uniformly are very large [confidential information deleted]” Id. at 10. For the last two months of the review period, however, the monthly differences decreased [confidential information deleted]. Id. According to Plaintiffs, “[t]here decreases in the monthly differences ... show that by the end of the six-month reporting period, Ferbasa had used most of the direct materials purchased during the hyperinflationary period in production.” Id. In the Redetermination Commerce briefly considered, and rejected, Plaintiffs’ arguments concerning the total values of its direct materials purchased and consumed, stating that because “the values cited by AIMCOR appear to reflect total consumption value and total purchases value without regard to quantities. Therefore, these figures are irrelevant to any analysis of distortion of cost data resulting from inflation.” Redetermination at 20 (citation omitted). While the Court is sympathetic to Commerce’s concerns about the weight that should be given this evidence, the Court does not find Commerce’s determination that this data is “irrelevant” to be supported by substantial record evidence. Although, in the Re-determination, Commerce found that Plaintiffs had failed to show that some of Ferbasa’s prices had been distorted by hyperinflation, it did so in light of its presumption that Ferbasa did not purchase any inputs used during the six-month reporting period at a time of hyperinflation. As discussed above, however, not only does this presumption appear to be incorrect, but the record evidence concerning hyperinflation in August 1994 and the 37-day inventory turnover period appear to demonstrate that Ferbasa did purchase at least some inputs during a hyperinflation-ary period. Further, the Court notes that while “the possibility of drawing two in *1264 consistent conclusions from the evidence does not prevent an administrative agency’s finding from being supported by substantial evidence,” Consolo v. Federal Maritime Comm’n, 383 U.S. 607, 620, 86 S.Ct. 1018, 16 L.Ed.2d 131 (1966), in this case Commerce did not articulate an alternative explanation for why the value of Ferbasa’s direct materials purchased might have exceeded the value of its direct materials consumed. Rather, and apparently in light of its incorrect presumptions, Commerce simply found this data to be “irrelevant.” In light of these considerations, the Court finds it possible, if not probable, that had Commerce reviewed Ferbasa’s figures in light of this “correct” presumption, it would have viewed this record evidence as confirming the need to make some hyperinflationary adjustments. Thus, upon remand, Commerce is to reconsider this record evidence in light of what appear to be the “correct” presumptions concerning hyperinflation in August 1994 and Ferbasa’s inventory turnover period for its material inputs. Specifically, Commerce should address the issue of whether the record evidence concerning the total value of Ferbasa’s purchases and cost-of-materials constitutes at least some evidence of price distortions. Should Commerce continue to view these figures as “irrelevant” since they “reflect total consumption value and total purchases value without regard to quantities,” Redeter-mination at 20, it should be specific in stating why it believes this data to be entirely unreliable. Similarly, should Commerce find an alternative explanation for these figures than that provided by Plaintiffs, Commerce should be specific in describing this alternative explanation and the basis for it. As for the record evidence identified by Plaintiffs concerning Ferbasa’s [confidential information deleted] increase in per-unit, direct material costs over the six-month reporting period, the Court notes that, on its face, this data appears to support Plaintiffs’ contention that hyperinflation distorted Ferbasa’s reported costs. The [confidential information deleted] increase in Ferbasa’s per-unit costs over the six-month reporting period appears to provide strong evidence that Ferbasa’s reported costs were distorted by hyperinflation and that appropriate adjustments may be necessary. Commerce, however, did not discuss this record evidence in either the Redetermination or the Final Results. Simply put, such a total failure to consider or discuss data which, on its face, undermines the position Commerce took in the Redetermination is unreasonable. See Olympia Industrial, Inc. v. United States, 1 F.Supp.2d 997, 1002 (CIT 1998) (“Commerce’s decision to reject the data without further investigation or explanation is not reasonable in view of the statute’s mandate to reach the most accurate result.”); see also Usinor Sacilor v. United States, 955 F.Supp. 1481, 1488 (CIT 1997) (“The Court finds that it is unreasonable for Commerce to look only to the intent of the French government and its control over the state-owned firm when there is other economic evidence in the administrative record that Commerce easily could have considered.”), ajfd in part and rev’d in part on other grounds, 17 1999 WL 641231 (Fed.Cir., Aug 24, 1999). Thus, on remand, Commerce is instructed to consider the record evidence concerning Ferbasa’s significant increase *1265 in per-unit, direct material costs over the six-month reporting period and discuss whether, and to what extent, this constitutes evidence of price distortions due to hyperinflation. IV CONCLUSION For the foregoing reasons, Plaintiffs’ Motion For Judgment Upon The Agency Record is partially granted and partially denied. This case is remanded to Commerce so that it may correct its errors and provide further explanations consistent with this opinion. In all other respects, Commerce’s Redetermination is sustained. ORDER This case having come before the Court upon the Plaintiffs’ Motion For Judgment Upon The Agency Record (“Plaintiffs’ Motion”); the Court having reviewed the papers and pleadings on file herein, having heard oral argument by each party, and after due deliberation, having reached a decision herein; now, in conformity with said decision, it is hereby ORDERED ADJUDGED AND DECREED that Plaintiffs’ Motion is DENIED IN PART and GRANTED IN PART; and it is further ORDERED ADJUDGED AND DECREED that the determination of the International Trade Administration of the U.S. Department of Commerce (“Commerce”) in Final Results Of Redetermi-nation Pursuant To Court Remand (“Redetermination ”) is remanded with instructions that Commerce: (1)either (a) conform its practice in this case to that used in Silicon Metal From Brazil, 62 Fed.Reg. 1,970 (1997), and Paint Filters and Strainers From Brazil, 52 Fed.Reg. 19,181 (1987), for determining whether to make necessary adjustments to account for the effects of hyperinflation on some or all of a respondent’s reported costs, or (b) discuss the reasons why it was appropriate for Commerce to depart from its former practice in this investigation. In so doing, Commerce should also consider those statutory provisions which underlie Commerce’s use of its hyperinflationary methodology and discuss how Commerce’s practice in this case constitutes a reasonable interpretation of these statutes. (2) examine the relevant record evidence concerning Brazilian inflation in August 1994 to determine whether this period of time should be considered “hyperinflationary” based upon the Department’s established standards for measuring hyperinflation; (3) examine the relevant record evidence concerning Ferbasa’s inventory turnover of both its finished product, ferrosilicon, and its various material inputs. In particular, Commerce shall reexamine the relevant record evidence concerning Ferba-sa’s 37-day inventory turnover period for ferrosilicon and consider, in light of this and other record evidence, whether it still believes that Ferbasa’s reported costs for September 1994 were based upon August 1994 input prices; (4) reconsider the record evidence concerning the total values of Ferbasa’s direct materials purchased and consumed during the six-month reporting period in light of the record evidence concerning hyperinflation in August 1994 and Ferbasa’s inventory turnover. Specifically, Commerce should address the issue of whether the record evidence concerning the total value of Ferbasa’s purchases and cost-of-materials constitutes at least some evidence of prices distortions; (5) consider, for the first time, the record evidence concerning Ferbasa’s significant increase in per-unit, direct material costs over the six- *1266 month reporting period, and discuss whether this constitutes evidence of price distortions due to hyperinflation; and (6) reconsider, in light of its analyses pursuant to instructions (2) through (5) above and in light of all record evidence in this case, whether part or all of Ferbasa’s costs-of-production should be calculated pursuant to Commerce’s hyperinflationary methodology, or whether other adjustments (to account for hyperinflation) are necessary; and it is further ORDERED ADJUDGED AND DECREED that remand results are due within 90 days of service of this order. Any comments or responses are due within thirty days thereafter. Any rebuttal comments are due within fifteen days after the date comments or responses are due; and it is further ORDERED ADJUDGED AND DECREED that all parties shall review the Court’s Opinion and Order and notify the Court on or before Wednesday, January 12, 2000, whether any information contained in the Opinion and Order is confidential, identify any such information, and request its deletion from the public version of the Opinion and Order to be issued thereafter. If a party determines that no information needs to be deleted, that party shall so notify the Court on or before January 12, 2000. Footnotes [1] As Commerce explained on page 7 of the Redetermination: In calculating cost-of-production (COP) and constructed value (CV), it is the Department's practice to allow a respondent to offset (z.e., reduce) financial expenses with interest income earned from the general operations of the company. See, e.g., Timken v. United States, 852 F.Supp. 1040, 1048 (CIT 1994). In calculating a company’s cost of financing, we recognize that, in order to maintain its operations and business activities, a company must maintain a working capital reserve to meet its daily cash requirements (e.g., payroll, suppliers, etc.). The Department further recognizes that companies normally maintain this working capital reserve in interest-bearing accounts. The Department, therefore, al *1252 lows a company to offset its financial expense with the short-term interest income earned on these working capital accounts. The Department does not, however, allow a company to offset its financial expense with income earned from investing activities {e.g., long-term interest income, capital gains, dividend income) because such activities are not related to the current operations of the company. [2] The citations provided by the parties were inconclusive concerning Commerce's “practice.” One other investigation seemingly supports Plaintiffs’ position (in that Commerce treated the whole POR as hyperinflationary, even though part of it clearly was not), but there Commerce specifically noted that "[t]he hyperinflationary methodology employed by the Department in these preliminary results of review is based on the facts particular to this review.” Preliminary Results of Antidumping Administrative Review Gray Portland Cement and Clinker From Mexico, 61 Fed.Reg. 51,676, 51,681 (1996). In no investigation reviewed by the Court did Commerce expressly discuss *1254 any "practice” or "methodology” of looking to a particular period of time (such as the period of review or even, as Defendant claims, the period for which prices and costs are reported) in determining whether replacement costs should be used. The Court also notes that nothing in Commerce’s Antidumping Manual indicates that Commerce routinely looks to a particular time period in assessing whether hyperinfla-tionary adjustments are necessary. See ITA Antidumping Manual, Chapter 8, pp. 77-81 (discussing high inflation economies). That Manual expressly states that, in calculating cost-of-production and constructed value in high inflation economies, "the decision to use indexation and the selection of an appropriate index/exchange rate system should be made on a case-by-case basis.” Id. at 81. [3] As explained by the Court in Queen's Flowers de Colombia: The court's review of an agency's change of position or practice will typically center on whether the action was arbitrary. A change is arbitrary if the factual findings underlying the reason for change are not supported by substantial evidence. Apart from factual findings, agency arbitrariness may also manifest itself in the particular reasoning offered by the agency; principally, if the reasoning is inconsistent with the statutory mandate, or, to a lesser extent, if the reasoning (or lack thereof) violates general principles of administrative law, or offends standards of procedural fairness implied in the statute. In this context the court reviews an agency change of position or practice to insure it is "in accordance with law.” 981 F.Supp. at 626 n. 7 (citations omitted). [4] See Antifriction Bearings (Other Than Tapered Roller Bearings) and Parts Thereof From France, Germany, Italy, Japan, Singapore, and the United Kingdom; Final Results of Anti-dumping Duty Administrative Reviews, 62 Fed. Reg. 2,081, 2,111 (1997) ("We have a longstanding practice of considering sales within 90 days before and 60 days after the month of the U.S. sale to be acceptable as potential comparators."); Certain Circular Welded Carbon Steel Pipes and Tubes From Thailand; Final Results of Antidumping Duty Administrative Review, 61 1,328, 1,332 (1996) ("The Department has implemented the contemporaneous 90/60 window in order to fulfill the statutory requirements in section 773(a)(1) of the Tariff Act that [fair market value] be based on the price of contemporaneous sales of such or similar merchandise.”). [5] Although Commerce stated that "we do not shift the reporting period to account for inflation in a period prior to the reporting period,” Redetermination at 19, it nevertheless proceeded to examine what the annualized inflation rate would be if it were to include the 37-day inventory period in its calculations. See id. at 20. [6] In the draft results of the Redetermination, Commerce additionally noted that, as Ferbasa had reported replacement costs ("current material prices”), and not historical costs, for its material inputs, Plaintiffs' argument that direct material costs were distorted due to purchases during a hyperinflationary period was "moot.” Redetermination at 6. For the final results, however, Commerce reversed its position on this issue, "agree[ing] with AIMCOR that there is contradictory evidence on the record and the preponderance of the record evidence suggests the conclusion that Ferbasa reported its cost-of-materials based on the weighted-average cost of inventory (z.e., historical costs).” Id. at 18. In all other respects, however, Commerce appears to have adopted the reasoning set out in its draft results. See, e.g., id. at 15 ("Our results have not changed from the draft results of redeter-mination.”), 18 ("[W]e continue to find that the application of our hyperinflationary economy methodology for the calculation of [cost-of-production] and [constructed value] in this case is not warranted.”). Accordingly, the Court reviews the reasons stated in both the draft and final results sections of the Redeter-mination in deciding whether Commerce’s actions are supported by substantial record evidence. [7] Moreover, and as noted earlier, Commerce's approach in this case is consistent with that taken in at least one other administrative review, Silicomanganese from Brazil; Final Results of Antidumping Duty Administrative Review, 62 Fed.Reg. 37,869, 37,876 (1997). There, Commerce similarly broke the period of review into two segments in light of the fact that the respondent only needed to *1257 report price and cost data for a six-month period. [8] As for the question of whether Commerce’s actions were also in accordance with law, the Court notes that "[tjhere is no statutory directive for measuring the effects of inflation in an antidumping analysis.” Asociacion Colombiana de Exportadores de Flores v. United stores, 6 F.Supp.2d 865, 873 (1998). Accordingly, the Court does not find (nor, for that matter, have Plaintiffs alleged) that Commerce has violated any specific statutory mandate in looking at factors besides the inflation rate in Brazil for the entire period of review in. deciding whether to employ its hy-perinflationary methodology. [9] Absent further explanation. Commerce's new practice of not accounting for purchases prior to the reporting period also appears to conflict with Commerce's stated purpose of making hyperinflationary adjustments, which is to avoid "distortions in dumping margin calculations [which] can occur where costs incurred in one period are compared to sales prices from a later period when nominal currency values have fallen significantly.” Rede-termination at 3. [10] Although, as noted previously, "[t]here is no statutory directive for measuring the effects of inflation in an antidumping analysis,” Asociacion Colombiana, 6 F.Supp.2d at 873, Commerce’s practice of making adjustments to account for inflation and hyperinflation appears to be based on its interpretation of various statutes. See, e.g., Redetermination at 21 (citing the requirement, codified in 19 U.S.C. § 1677b(e) (1994), that "the constructed value of imported merchandise shall be an amount equal to the sum of — (1) the cost of materials and fabrication ... during a period which would ordinarily permit the production of the merchandise in the ordinary course of business....”); Final Results, 61 Fed.Reg. at 59,408 (citing the requirement in 19 U.S.C. § 1677b(f)(l)(A) that “[costs shall normally be calculated based on the records of the exporter ... if such records are kept in accordance with the generally accepted accounting principles of the exporting country ... and reasonably reflect the costs associated with the production and sale of the merchandise.]”) See also 19 U.S.C. § 1677b(b)(3)(A) (1994) (stating that the cost-of-production shall equal "the cost of materials ... during a period which would ordinarily permit the production of that foreign like product in the ordinary course of business”); 19 U.S.C. § 1673 (1994) (imposing on Commerce an obligation to impose duties that offset the margin of dumping). *1260 Upon remand, if would be helpful if Commerce would clarify which of these, or other, statutory bases underlie its use of a hyperinf-lationary methodology in appropriate cases, and why Commerce’s decision not to account for purchases prior to the reporting period in this case constitutes a reasonable interpretation of these statutes. [11] The Court uses the word "appears” because, on its face, Commerce’s reasoning in the Redetermination is not entirely clear. On page 19 of the Redetermination, Commerce states what appears to be a clear institutional position: "we do not shift the reporting period to account for hyperinflation in a period prior to the reporting period....” Despite making this categorical statement, however, Commerce appears to have relied upon record evidence concerning actual price distortions and inflation outside of the six-month reporting period in reaching its decision. See Redetermination at 4 (stating, in the draft results, that "we have determined that the application of this hyperinflationary economy methodology is not appropriate in calculating Ferbasa’s COP because the six-month reporting period was not hyperinflationary, and there is no indication that the costs reported by Ferbasa were in any way distorted by hyperinflation.”) (emphasis added); id. at 20 n. 5 (discussing inflation during the 37-day turnover period prior to the reporting period). Such actions indicates to the Court that, despite its categorical statement about not looking outside the reporting period, Commerce’s decision not to make any adjustment was also based upon the facts in this case (which, it found, did not show any price distortions). Thus, the Court finds it appropriate to review Commerce’s treatment of the record evidence, since Commerce’s findings concerning this evidence appear either to have supported, or served as an alternative basis for, its decision. [12] An inflation rate of 7.06% for the month of August translates into a 84.72% annual inflation rate (calculated as 7.06 x 12 months, non-compounded), which is well above Commerce’s 50% hyperinflationary threshold. See Carbon Steel Pipe From Turkey, 62 Fed.Reg. at 51,630 (“[A]lthough Import Administration Policy Bulletin No. 94.5 states that 'an economy is deemed to he hyperinflationary if its monthly or annual inflation rates are greater than 5 percent and 60 percent, respectively,' in recent cases we have considered inflation rates lower than 60 percent to warrant application of our high-inflation methodology to avoid the distortions that may be caused by such inflation.”). [13] On remand, the only place where Commerce considered inflation in August 1994 was in stating that "even if the Department were to include the 37-day finished good inventory period [i.e., August 1994], the rate of inflation during this period [August 1994 through February 1995] would not exceed 50 percent, and therefore, would not be considered hyperinflationary.” Redetermination at 20 (footnote omitted). That inflation in August 1994 would not have caused the seven-month period August 1994 - February 1995 to be hyperinflationary, however, says nothing about whether inflation in this month was hyperinflationary and, more importantly, whether hyperinflation in August 1994 distorted any of Ferbasa’s reported costs. [14] In fact, throughout the Final Results and the Redetermination, Commerce appears to have had various interpretations of what the 37-day inventory turnover period represents. See Final Results 61 Fed.Reg. at 59,408. ("[B]ased upon [Ferbasa’s] inventory turnover rate of approximately one month, Ferbasa produced ferrosilicon for these sales at most approximately, one month earlier....”); Re-determination (draft results) at 5 ("The finished goods inventory period represents the period between the time when raw materials are actually purchased and the time in which those materials enter the production process”). Not only do both these statements appear to conflict with the record evidence concerning the 37-day turnover period, but they also appear to be inconsistent with Commerce’s later statement that (in light of the 37-day turnover period) "[i]t is therefore reasonable to state that the September costs are based on August prices.” Redetermination at 20. [15] At oral argument, the parties represented that the record contains no evidence as to the production processes that were used, whether any intermediate processing steps were required, or how long the production process averaged. [16] To the degree that the record currently contains insufficient evidence concerning the turnover period of Ferbasa’s inventory of raw inputs (i.e., the period between Ferbasa’s purchase of its inputs and the time of their sale as processed ferrosilicon), Commerce may, of course, request such information at its discretion. See 19 C.F.R. § 351.301(c)(2) (1999) (allowing the Secretary of Commerce to request factual information at any time during a proceeding). Similarly, and based on the representation of Defendant’s counsel at oral argument, the Court recognizes that Commerce may have to reopen the administrative record and request replacement cost information from Ferbasa, should it deem it appropriate to make hyperinflationary adjustments to account for the effects of hyperinflation on some or all of respondent’s reported costs. [17] Although the Federal Circuit ultimately reversed the trial court for substituting its judgment for that of Commerce (after the trial court weighed the economic evidence), it did not reverse on the grounds that the trial court erred in ordering Commerce to consider such relevant evidence. Rather, in dicta, the appellate court provides implicit support for the initial decision to remand, noting that the record evidence the trial court instructed Commerce to consider was "clearly relevant” to the question at hand. See Usinor Sacilor, 1999 WL 641231 at *3 n. 3 (stating that "all of the [economic] evidence that [Commerce] considered on remand predated the bestowal of the subsidies and is therefore clearly relevant to an assessment of the likely beneficiaries of the subsidies at the time of their bestowal”).