Opinion
Restani, Judge:This case is before the court after a second remand determination concerning the use of best information available (“BIA”) in selecting an antidumping duty rate to be applied to merchandise of plaintiff Pulton Chain Co., Inc. (“Pulton”).1 The issues presented are whether BIA should have been applied to Pulton and whether the rate chosen by the United States Department of Commerce, International Trade Administration (“ITA”) was proper.
Facts
Pulton made several shipments of roller chain, other than for bicycles, from Japan during the period of April 1, 1981 through March 31, 1983. Antidumping duties were originally assessed by the Department of the Treasury in 1973. See Roller Chain, Other than Bicycle, from Japan, 46 Fed. Reg. 44,488, 44,488 (Dep’t Comm. 1981) (final admin, review). The law in effect from 1979 to 1984 required annual administrative reviews. Trade Agreements Act of 1979, Pub. L. No. 96-39, sec. 101, § 751, 93 Stat. 144, 175 (1979). ITA began the administrative review of the 1981-82 shipments sometime after March 31, 1982. Pulton submitted a 280-page questionnaire response on June 24, 1982 and a supplemental response on February 24, 1983. ITA requested an update to the data previously submitted for 1981-82 by mailing out questionnaires on April 1, 1983. On April 27, Pulton requested that the requirement for submitting data on computer tapes be waived because it did not keep records in that fashion. The waiver was granted and Pulton submitted a written 250-page response containing data for 1982-83 on June 17, 1983 and a supplemental response on December 16, 1983.
Although ITA used data submitted by Pulton to prepare a draft preliminary results notice, the notice was not published because statutory provisions eliminating the necessity for mandatory annual reviews were soon to be passed by Congress. When the domestic industry later requested a periodic review under the new law, ITA reinitiated a review *1137on July 9, 1986. ITA did not request a new questionnaire response until almost four years later, on December 18, 1990. Pulton replied on January 17, 1991 by means of a letter, stating that it was “puzzled” by ITA’s questions because an unpublished, draft preliminary notice had calculated the dumping margin to be 0%. Plaintiffs Appendix A, Pub. Doc. 20, at 1. The letter also noted that many of the relevant documents had been discarded when Pulton moved into different offices and asked ITA to “reconsider” its request for information. Id.
On January 25, 1991, ITA again urged Pulton to submit the necessary data. Pulton claims that an official in the Tokyo Import Attaché’s office told it that a submission of no information was preferable to a partial submission. Therefore, on January 29, 1991, Pulton sent a second letter to ITA informing it that the documents were lost and no data would be forthcoming. Id., Pub. Doc. 22, at 1.
ITA published preliminary results of the 1981-82 and 1982-83 annual reviews on March 7, 1991. Roller Chain From Japan, 56 Fed. Reg. 9674 (Dep’t Comm. 1991) (prelim, admin, review). The results stated that Pulton “provided an inadequate questionnaire response originally, and then declined to respond to the supplemental questionnaire/deficiency letter or to participate further in the reviews.” Id. at 9675. ITA proclaimed that it would resort to BIA, using “the highest rate for another manufacturer under review in these periods, 15.92 percent.” Id. The 15.92% figure derived from the final results of the antidumping duty review of 18 other companies, published on May 13, 1987. Roller Chain, Other than Bicycle, From Japan, 52 Fed. Reg. 18,004, 18,005 (Dep’t Comm. 1987) (final admin, review). The results assigned a 15.92% rate for several companies exporting between 1981 and 1983. One company that exported in 1979 and 1980 was assigned a rate of 43.29%. Id.
After ITA published the preliminary results of the 1981-82 and 1982-83 reviews, Pulton objected that ITA already had most of the information requested by ITA in the 1990 and 1991 questionnaires. ITA took the position that Pulton’s prior submissions included
no information on the value of sales which is critical in determining home market viability, inconsistent units of sale (feet in some cases, links in others), dates of sale missing on approximately 20 percent of both U.S. and home market sales, charges reported in various units with no explanation of how they relate to units of sale, * * * [and] no computer tapes.
Roller Chain from Japan, 56 Fed. Reg. 32,175, 32,177 (Dep’t Comm. 1991) (final admin, review) (‘Final Results”). It therefore applied the 15.92% BIA rate in its publication of final results, citing its policy to apply the higher of “(a) the highest rate for a responding firm with shipments during the period or (b) that firm’s own last rate.” Id. at 32,176 (“highest rate analysis”).
Pulton challenged the final results in a suit brought before this court. ITA admitted that its choice of 15.92% as the BIA rate for Pulton under *1138the highest rate analysis was in error and requested a remand. Defendant’s Memorandum in Opposition to Plaintiffs Motion for Judgment upon the Administrative Record, at 39-40. The first remand determination resorted to a different analysis for the calculation of a BIA rate than the original determination. According to ITA, before the original results were published, it had already begun implementation of a new BIA policy, known as the two-tier policy. The two-tier policy provided that if a company refused to cooperate, ITA would apply the higher of (a) the highest rate for any firm in the original less than fair value (“LTFV”) investigation, or (b) the highest rate for any firm in the subject review. If a company cooperated substantially, ITA would apply the higher of (a) the firm’s LTFV rate, or (b) the highest calculated rate in the subject review. Final Results of Redetermination Pursuant to Court Remand, at 1-3 (April 5, 1993) {“First Remand Results”); see Antifriction Bearings (Other than Tapered Roller Bearings) and Parts Thereof from the Federal Republic of Germany, 56 Fed. Reg. 31,692, 31,705 (Dep’t Comm. 1991) (final admin, review) (“AFBs 1”).2
Subsequent to the issuance of the original results in the case at bar, ITA revised the policy articulated in AFBs 1. The new policy provided that for either cooperative or uncooperative respondents, ITA could choose to apply the highest rate in the LTFV determination or any prior administrative review. Antifriction Bearings (Other than Tapered Roller Bearings) and Parts Thereof from France, 57 Fed. Reg. 28,360, 28,379 (Dep’t Comm. 1992) (“AFBs 2”). Contending that the revision made in AFBs 2 was merely a clarification, ITA applied the amended two-tier policy in the first remand results and selected a BIA rate of 43.29%, which was assigned to another company in a prior administrative review. First Remand Results, at 4-6.
Pulton objected to the results of the first remand on the ground that ITA could not reasonably rely upon the revised two-tier policy, initiated after the final results in the Roller Chain determination had been reached. Pulton also argued that the 43.29% figure was never finalized and was therefore an improper BIA rate. ITA agreed that the revised two-tier policy did not apply, and consented to a second remand. Final Results of Redetermination Pursuant to Court Remand, at 5 (June 22, 1993) (“Second Remand Results”).
In the second remand determination, ITA noted that the two-tier policy in place at the time of the final results in Roller Chain restricted applicable BIA rates to either the original LTFV rate or the highest rate of any firm in the subject review. Id. at 3-4. Because the original LTFV determination was made by the Department of the Treasury, under the old statutory scheme, ITA had no documentation concerning the original LTFV rate and asserted that it could not rely on such a rate. Id. at 7. It nevertheless contended that the highest rate of any firm during the sub*1139ject review was too low to “provide the incentive necessary to ensure compliance by Pulton in future reviews.” Id. at 7-8. Clinging to the notion that BIA is determined on a case-by-case basis, ITA continued to find 43.29% to be the proper rate. Pulton protests strenuously.
Discussion
I. Applicability of BIA to Pulton:
Congress has mandated that ITA “shall, whenever a party or any other person refuses or is unable to produce information requested in a timely manner and in the form required, or otherwise significantly impedes an investigation, use the best information otherwise available.” 19 U.S.C. § 1677e(c) (1988). Similarly, the regulations provide for the use of BIA when ITA “(1) Does not receive a complete, accurate, and timely response to * * * [a] request for factual information; or (2) Is unable to verify * * * the accuracy and completeness of the factual information submitted.” 19 C.F.R. § 353.37(a) (1991). ITA may take the cooperativeness of the party into account in determining what constitutes BIA. Id. § 353.37(b).
In the absence of a subpoena power, ITA must often resort to BIA in order to ensure compliance with its requests for information. See Rhone Poulenc, Inc. v. United States, 899 F.2d 1185, 1191 (Fed. Cir. 1990). BIA has thus been characterized as “an investigative tool, which th[e] agency may wield as an informal club over recalcitrant parties.” Id. [quoting Atlantic Sugar, Ltd. v. United States, 744 F.2d 1556, 1560 (Fed. Cir. 1984)). Nonetheless, the ultimate purpose of BIA is not to punish non-responding parties, but to “determin[e] current margins as accurately as possible.” Id.
Pulton claims that resort to BIA was improper in this case for four reasons. First, ITA had already compiled sufficient accurate information to prepare draft preliminary results by late 1984 and there was no subsequent change in circumstances. Second, most of the information requested by ITA was already present in its files in paper form and Pul-ton was not required to submit computer tapes of identical data. Third, ITA requested irrelevant information. Fourth, Pulton’s refusal to provide information resulted from the alleged advice of a Department of Commerce official that no information was better than partial information.
This court cannot accept the argument that ITA’s final decision should conform to the draft preliminary results prepared in 1984. In a similar case to this one, ITA had informed the plaintiffs that a forthcoming preliminary determination would calculate the antidumping margin to be de minimis. Koyo Seiko Co. v. United States, 796 F. Supp. 517, 520 (Ct. Int’l Trade 1992). Believing that the assessment of antidump-ing duties would soon be revoked, plaintiffs destroyed certain important documents. Id. at 525. The preliminary results never appeared and ITA continued its investigation, relying on BIA instead of data from the destroyed documents. Id. at 525-26. This court stated that while plaintiffs *1140“may have been justified in expecting a revocation, it was never confirmed. Under these circumstances, [plaintiffs] should have retained their records.” Id. at 525. ITA’s preparation of draft results in this case cannot justify Pulton’s lack of care, which led to the loss of relevant documentation during a move to different offices. Like the plaintiffs in Koyo Seiko, Pulton is not entitled to rely on its expectation of future agency action.
Furthermore, ITA need not reach the same result in a final determination as it did in preliminary results. The purpose of publishing preliminary results is to discover inaccuracies and correct them before coming to a final decision. See Tehnoimportexport v. United States, 15 CIT 250, 254, 766 F. Supp. 1169, 1175 (1991). Therefore, even if ITA had published the preliminary results as expected, Pulton could not assume that the final results would arrive at the same conclusion.
Pulton’s second argument, that it submitted complete documentation in paper form and ITA could not request computerized information, must also fail. ITA concluded that there was insufficient information regarding value of sales, date of sales, inconsistent units of quantity, and no explanation of various charges such as freight charges, commissions, discounts and rebates. Final Results, 56 Fed. Reg. at 32,177. Although Pulton is correct that certain data were previously submitted, substantial information was lacking from the administrative record.
Pulton did provide ITA with value of sales information for the rele-vantperiod. Plaintiffs Appendix B, Conf. Doc. 1, at 2; id., Conf. Doc. 3, at 2. Likewise, all reported sales to the United States or the home market seem to include dates of sale. Id., Conf. Doc. 1, at 002-106; id., Conf. Doc. 3, at S83,001-76, S83,137-71. On the other hand, itemized freight costs are missing from several sales reports. Id., Conf. Doc. 1, at 002-106. In briefs submitted after oral argument, Pulton finally explained how to use tables attached to the sales data to determine freight and packing costs. The tables of freight charges are customer specific and tables for only the main customers are included. Id. at 132-35. The data is not in easily useable form. Although Pulton argues that ITA was able to calculate dumping margins for succeeding periods using Pulton’s questionnaire responses, it admits that in this respect the 1981-82 response was structured differently.3 Plaintiffs Reply to Defendant’s July 28, 1993 Comments, at 6.
ITA is correct in noting that whereas most of Pulton’s sales records express quantity in terms of feet, many of them use measurements such as links or pieces. Plaintiffs Appendix B, Conf. Doc. 1, at 049-069 (expressed in terms of links); id., Conf. Doc. 3, at S83,144, S83,164-66, S83,171 (expressed in terms of pieces). Pulton contends that ITA could have converted the sales data to comparable units of quantity by using *1141conversion ratios that are “common knowledge.” Memorandum in Support of Plaintiff s Motion for Judgment on the Agency Record, at 30. Although Pulton demonstrates that the conversion ratios were made known to ITA by other companies in other administrative reviews, there is no evidence that ITA was informed about the conversion ratios in the course of the contested review.
It is incumbent upon the person receiving the questionnaire to provide information “in the form requested." N.A.R., S.p.A. v. United States, 14 CIT 409, 415, 741 F. Supp. 936, 941 (1990). In N.A.R., the plaintiffs internal accounting procedures organized cost data regarding pressure sensitive plastic tape by classes of colors rather than by the length of rolls. When ITA requested calculations according to length, plaintiff refused on the ground that it had already provided cost information. Id. at 414, 741 F. Supp. at 941. This court upheld the use of BIA in that situation, saying “it is not sufficient that NAR provided cost information according to its internal procedures if those procedures did not produce what the ITA specifically requested. ”4 Id. at 415, 741 F. Supp. at 941-42. The same situation exists in the present case.
ITA also requested computer tapes, but Pulton refused to submit them on the ground that ITA had no authority to make such a request. The legislative history states that ITA “may require the exporter to furnish information on computer tape if the exporter maintains records in that form, but it may not require the exporter to purchase a computer or computerize information.” H.R. Rep. No. 317, 96th Cong., 1st Sess., pt. V, at 69 (1979).
It is undisputed that Pulton did not have computerized records in 1983, when the investigation began. Nevertheless, in 1990, Pulton submitted computer disks containing information for the 1983 to 1985 period as part of a separate annual review proceeding. Memorandum in Support of Plaintiff s Motion for Judgment on the Agency Record, at 25 n.52; Plaintiffs Reply Brief, at 17-18. ITA, alert to the possibility that Pulton had computerized its old data base, requested computer tapes for the 1981 to 1983 period. Pulton responded that ITA had no authority to request computer tapes. Memorandum in Support of Plaintiff s Motion for Judgment on the Agency Record, at 25. ITA is correct in characterizing this behavior as recalcitrant.5 Ansaldo Componenti, S.p.A. v. United States, 10 CIT 28, 37, 628 F. Supp. 198, 205 (1986) (respondent *1142should not be allowed to withhold information based on unilateral legal conclusions).
The argument that ITA requested irrelevant information from Pul-ton in its second set of questionnaires has no merit. Quite simply, “[i]t is Commerce, not the respondent, that determines what information is to be provided for an administrative review.” Id. ITA may “conduct its an-tidumping investigations the way it sees fit, not the way an interested party seeks to have it conducted.” N.A.R., 14 CIT at 415, 741 F. Supp. at 941. If a party resists ITA’s control and attempts to direct the investigation itself, ITA is justified in using BIA. Id. By making its own determination as to the relevancy of the requested information, Pulton was improperly attempting to direct the investigation.
The final argument to be addressed is whether Pulton properly withheld information based on the advice of the Tokyo Import Attaché, who allegedly said that no information was preferable to partial information. As ITA points out, there is no actual evidence of what the Import Attaché said to Pulton’s representatives. Assuming the truth of Pul-ton’s allegation, Pulton had no right to rely on this advice in failing to respond to ITA’s questionnaires. First, communications from ITA clearly stated that if Pulton did not provide the requested information, BIA would be applied. Plaintiff s Appendix A, Pub. Doc. 21, at 1. Second, regardless of the Attaché’s advice, Pulton could not have submitted data that it claims to have lost in the relocation of its offices.
In conclusion, ITA correctly resorted to BIA under these circumstances. Pulton had no right to expect a final decision commensurate with the low dumping margins established in the draft preliminary results, which never appeared in the Federal Register. ITA properly found Pulton’s questionnaire data to be insufficient and not in easily useable form. Pulton’s obligation was to provide the data it could, explain why data not provided was unobtainable, and to explain how the data previously provided answered any new questions. It did not do this until this litigation. Thus, its response was deficient. Pulton’s arguments that ITA requested irrelevant information and that a government representative advised it not to respond fully to the questionnaires do not change this conclusion.
II. Use of43.29% as the BIA Rate:
In the course of several successive remands, ITA has narrowed the disputed issue of law to a very fine point. In the final determination, ITA applied the “highest rate” analysis, which used the higher of the highest rate for any responding firm with shipments during the period or the individual firm’s own last rate. In asking for the first remand, ITA conceded that the “highest rate” analysis was inapplicable. It recalculated BIA according to the revised two-tier analysis, which gave ITA the option of applying a rate from the subject review, a prior administrative review, or the original LTFV investigation. ITA then agreed to a second remand, conceding that the revised two-tier analysis had developed af*1143ter the final determination in this proceeding had been issued. The original version of the two-tier analysis is thus the one that pertains to this case.
As noted in the background section, the original two-tier policy provides that if a company refuses to cooperate, ITA will apply the higher of the highest rate for any firm in the original LTFV investigation or the highest rate for any firm in the subject review. If a company cooperates substantially, ITA will apply the higher of the firm’s LTFV rate or the highest calculated rate in the subject review. See AFBs 1, 56 Fed. Reg. at 31,705.
Pulton’s one-page response to ITA’s supplemental questionnaire cannot be characterized as cooperative. Plaintiffs Appendix A, Pub. Doc. 20, at 1. The letter states several unacceptable reasons for Pulton’s failure to provide more data, and nowhere does it indicate a willingness to cooperate in ITA’s review. Id. As an uncooperative respondent, Pulton could receive, as BIA, the highest rate for any firm in the original LTFV investigation. ITA has admitted, however, that it has no documentation from the original LTFV investigation, which was conducted by the Department of the Treasury under the old law. Second Remand Results, at 7. The parties agree that the applicable rate would be the highest calculated rate in the subject review.
In the subject review, ITA applied a BIA rate of 15.92% to several participating firms. Final Results, 56 Fed. Reg. at 32,177. ITA concluded, however, that 15.92% did not constitute the highest rate of any firm in the subject review, because it was not a contemporaneously calculated rate. Pulton agrees6 and contends that it is entitled to a calculated rate of its own, which it believes to be close to zero. The court declines to require ITA to apply such a rate, because the result would “in effect reward [Pulton] for its failure to cooperate with Commerce in the investigation.” Krupp Stahl A.G. v. United States, 822 F. Supp. 789, 793 (Ct. Int’l Trade 1993).
Pulton argues in the alternative for BIA rates of 5.22% for 1981-82 and 5.45% for 1982-83, which represent the highest rates for an exporter of the same products from the same country during the relevant time period. Roller Chain Other than Bicycle, from Japan, 57 Fed. Reg. 43,697, 43,703 (Dep’t Comm. 1992).7 Pulton’s past actual dumpingmar-gins have varied between zero and 5%. In 1979-80, Pulton’s margin was zero. 46 Fed. Reg. at 44,491. In 1980-81, the numbers ranged from zero to 5%. 48 Fed. Reg. 51,801, 51,807 (Dep’t Comm. 1983). The most recent final non-BIA rates calculated for Pulton were zero and 0.11%. 57 Fed. *1144Reg. 56,319, 56,321 (Dep’t Comm. 1992) (final admin, review for 1990-91).
ITA has not suggested a number that would apply on the facts of this case and under the policy applicable at the time of the final determination herein. As far as the court can determine the only BIA numbers available on the unique facts of this case, where the original LTFV margin is unusable, are the 5.22% and 5.45% figures. ITA’s attempt to assert a 43.29% rate is arbitrary and capricious and has no basis in law or fact. Such a rate cannot be justified on the “case-by-case nature of BIA determinations.” There is nothing in this case to indicate that 43.29% is in any way relevant, or which suggests that the 5-6% rate does not provide an appropriate incentive to comply with ITA’s information requests. Finally, as is agreed by ITA, the policy in effect at the time of the original determination did not allow, as a general rule, use of another firm’s rate from a prior administrative review as a BIA rate. Therefore this case should be remanded again to give ITA the opportunity to apply a rate based on the calculated rates for other exporters for the period at issue. ITA has the option of calculating actual antidumping duty rates based on Pulton’s recent explanation of its data, if ITA believes it can do so with reasonable accuracy. Because of Pulton’s previous lack of full cooperation, the choice is ITA’s to make, based on what it considers to be administratively sound.
The two remands were agreed to by the parties.
The Federal Circuit has upheld ITA’s use of the two-tier policy in Allied-Signal Aerospace Co. v. United States, 996 F.2d 1185, 1192 (Fed. Cir. 1993).
An additional discrepancy in Pulton’s data is that records of commissions appear in supplemental tables although Pulton’s introductory explanation states that no commissions were paid. Plaintiffs Appendix B, Conf. Doc. 3, at 3, S83,111-12. Because this discrepancy also appeared in Pulton’s 1983-84 response from which ITA calculated actual dumping margins, the court declines to consider it. See Plaintiffs Appendix D, Attachment 1, at 3, 84,015.
Cases cited by Pulton that decline to uphold the use of BIA are distinguishable. This court has found the use of BIA to be inappropriate where the party submitted too much information pursuant to unclear instructions by ITA, because ITA could analyze and reprocess the data. Daewoo Electronics Co. v. United States, 13 CIT 253, 266, 712 F. Supp. 931, 945 (1989). Pulton, rather than providing too much information, submitted not enough and not the right kind. In another case cited by Pulton, ITA had requested information about additional buyers in the domestic market. Olympic Adhesives, Inc. v. United States, 899 F.2d 1565, 1573 (Fed. Cir. 1990). When the company responded that there were no additional domestic buyers, ITA applied BIA. Id. ITA’s decision was overturned on the ground that “a ‘No’ answer is not a refusal to provide data. If there is no data, ‘No’ is a complete answer.” Id. The facts in Olympic Adhesives are clearly distinguishable from the present case.
Not until oral argument did Pulton state that it had not compiled computerized data for the relevant period. This explanation should have been made to ITA.
As the parties agree that 15.92% is inapplicable, the court will not address situations in which a non-contemporaneously calculated rate may be used as the “highest rate” in the subject review.
ITA complains that the 5.22% and 5.45% rates appeared in a Federal Register notice published after the date of the final determination at issue. But ITA has consented to two subsequent remands and it cannot show that it is prejudiced by the application of these rates, because the only calculated rates for the subject review in existence at the time of the determination were de minimis. See 52 Fed. Reg. 18,004, 18,005 (Dep’t Comm. 1987) (final admin. review).