This appeal is from the final judgment of the Court of International Trade in Olympic Adhesives, Inc. v. United States, 708 F. Supp. 344 (Ct.Int'l Trade 1989) (Musgrave, J.). Animal glue imported from Sweden has been subject to an antidumping order imposed by the Treasury Department on August 3, 1977.1 The original investigation found that the dumping margin for Swedish glues was 92.72%. In the first review of the amount of duties following the order which covered the period August 3, 1977, through May 31, 1978, Treasury found that there were no sales at less than fair value and, thus, no dumping duties were due. This appeal concerns the next review period for which the International Trade Administration (ITA), now responsible for conducting such reviews, reimposed the original 92.72% duty rate. The ITA did not base this rate on sales data during the review period. Rather, the ITA found that the Swedish manufacturer, Extraco Geltec AB, failed to supply certain requested information. In reliance on that finding, the ITA resorted to its authority, under 19 U.S.C. Sec. 1677e(b) (1982) (redesignated in 1988 as Sec. 1677e(c)2) as implemented by 19 C.F.R. 353.51(b) (1988), to use the "best information available" to determine the amount of the duty. In the ITA's judgment, the "best information" on the dumping margin for the review period was the 92.72% margin found in the original investigation. The United States importer of Swedish glue, who must pay the duty, is appellant Olympic Adhesives, Inc. Olympic brought suit in the Court of International Trade challenging the ITA's resort to the "best information" rule on the record of this case, and asserting that, in any event, the ITA did not use the "best information available." Rejecting both of these positions, the Court of International Trade affirmed the ITA's determination of the duty rate. Because we agree with the first of Olympic's arguments, we reverse and remand with instructions.

* Background

The 1977 antidumping duty investigation into animal glue and inedible gelatin imported from Sweden involved only one Swedish producer, Extraco Geltec AB, and only one U.S. importer, Olympic Adhesives. At that time, Extraco was selling animal glue to Olympic primarily in grades 12, 16, and 20, which grades Extraco also sold to unrelated purchasers in Sweden. A comparison of U.S. and Swedish prices on these grades yielded the initial dumping margin of 92.72%. 42 Fed.Reg. 39288-89 (1977).

The first administrative review was conducted by the Treasury Department and covered the period from August 3, 1977 to May 31, 1978. For that review period, Extraco reported no sales to Olympic of animal glue grades 12, 16, and 20. It did report sales to Olympic and to one Swedish customer of grades 14, 18, and 22. The comparison of the respective prices for these grades resulted in a determination that grades 14, 18, and 22 glues were not being sold at less than fair value (LTFV) during this review period. Olympic Adhesives, 708 F. Supp. at 346.

Before the second review was completed by Treasury, responsibility for conducting the review was transferred to the ITA. Treasury had been reviewing the period June 1, 1978 to December 31, 1978. The ITA expanded this review through November 30, 1980, but treated it as two periods divided at January 1, 1979. In response to the ITA's request for information regarding Extraco's sales subject to the dumping order for the extended period, Extraco again supplied information on sales of 14, 18, and 22 grade glues to Olympic in the United States and to the same single company in Sweden. On May 19, 1981, the ITA verified Extraco's responses by an investigation in Sweden. Based on the information supplied by Extraco and verified by the ITA, the ITA published its preliminary determination that the dumping margin on sales to Olympic during the period from June 1, 1978 through December 31, 1978, was 8.95%, and that the margin on sales during the period January 1, 1979 through November 30, 1980, was 1.93%. See 46 Fed.Reg. 41540-41 (August 17, 1981).

Following the preliminary determinations of the above rates, the U.S. glue industry objected to the ITA, asserting that the data supplied by Extraco was questionable and might not reflect sales made in the ordinary course of trade, as required for the ITA's determination of the amount of the duty. It suggested that Extraco was in fact selling grades 12, 16, and 20 but designating them as "new grades" 14, 18, and 22; that these "new grades" were not known in Europe; and that because there was only one home market purchaser of the "new grades", the Swedish sales could have been contrived. It also noted the great drop in volume of Extraco's reported total sales compared to its capacity and surmised that all sales had not been reported.

The above assertions by the U.S. glue industry concerning the alleged identity of the "old" and "new" grades and the possibility of contrived sales were not supported by any direct evidence either before or at a hearing held by the ITA to consider the objections. Counsel for the U.S. glue industry simply posited the theory that there could have been an agreement or arrangement between Extraco and Olympic to denominate the grades of the initial investigation (12, 16 and 20) as different grades in the annual review (14, 18 and 22) and between Extraco and its single Swedish customer to make enough purchases of these "new" grades to establish a home market price. Counsel pointed to the tremendous reduction in the dumping margin as, in itself, raising a suspicion of falsification.

If the domestic industry's theory were true, the reported Swedish sales could not be used to determine the foreign market value under 19 U.S.C. Sec. 1677b(a) (1982) which then read:3

(a) Determination; fictitious market; sales agencies

For the purposes of this subtitle--

(1) In general

The foreign market value of imported merchandise shall be the price, at the time of exportation of such merchandise to the United States--

(A) at which such or similar merchandise is sold or, in the absence of sales, offered for sale in the principal markets of the country from which exported, in the usual wholesale quantities and in the ordinary course of trade for home consumption, ...

. . . . .

In the ascertainment of foreign market value for the purposes of this subtitle no pretended sale or offer for sale, and no sale or offer for sale intended to establish a fictitious market, shall be taken into account. [Emphasis added.]

The applicable implementing regulation, 19 C.F.R. Sec. 353.18 (1988) similarly provides:

In determining foreign market value, no pretended sale or offer for sale, and no sale or offer for sale intended to establish a fictitious market, will be taken into account.

In response to the objections, the ITA initially defended its preliminary determinations of 8.95% for the 1978 period and 1.93% for 1979-1980. It found nothing suspect in the drop in margin inasmuch as the imposition of antidumping duties is intended to achieve precisely that effect. However, it reopened its investigation to look into the specific assertions of the U.S. glue industry.

The assertion that the glue grades purchased by Olympic had merely been redesignated turned out to be baseless. Extraco had, in fact, sold two other grades of glue (17 and 25-S) to Olympic even during the original investigation period, albeit in small amounts. Further, information supplied by Olympic established, without contradiction, that the three suspect grades (14, 18 and 22) were standard grades of glue which Olympic had been purchasing from other sources prior to the publication of the initial antidumping duty finding. Finally, test reports proved that the glues were in fact 14, 18 and 22 grades, not 12, 16 and 20.4 That asserted basis for the U.S. industry's theory of fictitious sales was wholly unsupported.

Extraco's drop in overall production and decreased sales to the United States also were proved to be genuine. The record showed that because of a decline in the production of English bone china to which Extraco sold processed bones, Extraco began to limit its production of glue and, before the renewed investigation was over, Extraco had gone out of the glue business entirely.

The remaining allegation of the U.S. industry was that the information on Swedish sales was falsified or incomplete. As indicated, the reported transactions had been investigated prior to the preliminary determinations and were found to be in order. Even minor discrepancies in Extraco's records, e.g., a failure to deduct a 2% discount on a few sales, were resolved after the ITA's investigative trip.

The ITA pursued the matter of Swedish sales by first contacting Olympic to obtain more information from Extraco with respect to Extraco's sales of grades 14, 18 and 22. On November 11, 1981, Olympic forwarded telexes from Extraco which provided the answers to all of these questions. Specifically, Extraco answered that (1) it had continuously offered grades 14, 18 and 22 to Swedish customers since January 1978; (2) its single Swedish customer manufactured water-remoistenable sealing tape; and (3) it switched to grades 14, 18 and 22 at Olympic's request, because it had the capability to do so, and it was able to sell these grades in Sweden also.

By letter dated November 19, 1981, the ITA submitted a questionnaire directly to Extraco advising that counsel for the domestic industries raised issues concerning whether the reported sales were made in the ordinary course of trade which "the Department believes Extraco can clarify by answering the enclosed questions. Failure to respond to this request may result in [use of the original 92.72% margin on sales in the relevant period]." Questionnaires were also sent to Extraco on November 25, 1981 (apparently the same as above), and on December 21, 1981 (contents unknown, but which may have asked, in addition, about sales to the United States via London brokers).

Extraco replied on January 8, 1982, advising that it did not control European brokers and could not answer questions concerning their sales to the United States. It further advised that it had reduced its production from 4,000 tons in 1978, to 1,500 tons in 1981; that it had stopped all sales to the United States in July 1981; that it enclosed its answer to the November '81 questionnaire; but that because of a decrease in staff, "we have small possibilities to continue answering further questionaires [sic], and considering the fact that we no longer sell any glue to the United States we presume that it would not be needed either."

In May 1982, the ITA sent another questionnaire to Extraco by telex raising a different objection to the information on Swedish sales of grades 14, 18, and 22 which had been supplied. This questionnaire stated that for the periods, January 1, 1979 through November 30, 1980, and for December 1, 1980 through November 30, 1981, the Swedish sales "were not close in time to U.S. sales." Nothing was said about the June-December 1978 review period. By this time, it is apparent that the ITA had already begun its review for the next year inasmuch as the telex refers to Extraco's responses omitting some sales for the period December 1980 through November 1981, apparently on still another questionnaire. Extraco's sales manager immediately responded to the telex by telephone that the company would not or could not answer. However, it appears that some matters were resolved as shown by the ITA's determination of a duty of 3.01% for the December 1980 to November 1981 period. See 49 Fed.Reg. 39883 (October 11, 1984).

On January 24, 1983, Extraco sent a letter to the ITA in response to the ITA's letter of December 30, 1982 (contents unknown), which referred to yet another request for information covering the review period December 1, 1981--November 30, 1982. In its January 1983 letter, Extraco stated it was no longer producing any animal glue; that its sales to the United States had stopped in July 1981; and that it had no people who could answer the ITA's questions. No further contacts apparently were made between the ITA and Extraco.

On September 9, 1983, the ITA published the final results of its review for the period June 1, 1978, through November 30, 1980, which imposed the 92.72% duty. See 48 Fed.Reg. 40769-70 (1983). The entirety of its explanation for overturning its preliminary determinations of 8.95% and 1.93% duty rates for the review periods is the following: "Based on Extraco's failure to provide home market sales data for similar merchandise, we cannot resolve the petitioners' [i.e., the domestic industry's] allegation that home market sales of the three grades were not in the ordinary course of trade."5 Therefore, the ITA determined it could use "best information available" and found that to be the original 92.72% margin.

Olympic timely contested the final results of the administrative review on the grounds that, inter alia, the ITA was not entitled to resort to the best information rule as authorized by 19 U.S.C. Sec. 1677e(b) and implemented by 19 C.F.R. Sec. 353.51(b). Olympic asserted that the ITA did not comply with the regulatory requirements for notification that Extraco's answers were deficient; that Extraco, in any event, supplied all requested information; and that any inability of the ITA to make a determination regarding whether the home market sales of glue were not in the ordinary course of trade resulted from the ITA's failure to conduct a proper investigation rather than any deficiency in the information supplied by Extraco. Rejecting Olympic's arguments, the trial court found that the ITA had no affirmative legal duty to supplement the administrative record and that the ITA had on "several occasions" requested that Extraco supply it with certain information relevant to the ordinary course of trade determination. The trial court further found that Extraco failed to adequately supply the requested information, either because it was no longer able to do so or because it simply refused. The trial court concluded that, after the ITA received Extraco's January 24, 1983 letter, the ITA was not "entirely unjustified" in resorting to the best information available. Olympic Adhesives, 708 F. Supp. at 350. On appeal, Olympic argues that the trial court decision sustaining the ITA's determination under the "best information available" rule was contrary to law and not supported by substantial evidence.

This court has jurisdiction over this appeal under 28 U.S.C. Sec. 1295(a)(5) (1982). Olympic brought this action pursuant to 19 U.S.C. Sec. 1516a(a)(2)(B)(iii) (1988) which permits any interested party to commence an action in the Court of International Trade to review, inter alia, "[a] final determination, ... by the administering authority [the ITA] ... under section 1675 of this title."6 The applicable standard for judicial review of this ITA determination is prescribed by 19 U.S.C. Sec. 1516a(b)(1) (1988):

The court shall hold unlawful any determination, finding, or conclusion found--

. . . . .

(B) in an action brought under paragraph (2) of subsection (a) of this section, to be unsupported by substantial evidence on the record, or otherwise not in accordance with law.

In the appeal before us, we are presented with a situation involving two-tiered appellate review of agency action7 which is to be reviewed on the administrative record under the above standard. In reviewing the trial court's affirmance of ITA's dumping duty assessment, we must decide whether the trial court correctly concluded that ITA's determination is in accordance with the law and supported by substantial evidence. As aptly stated in Atlantic Sugar, Ltd. v. United States, 744 F.2d 1556, 1559 n. 10 (Fed.Cir.1984), with regard to appeals under this statute, "[w]e review that court's review of an [agency] determination by applying anew the statute's express judicial review standard."

The question raised in this appeal is whether the ITA's action in imposing the 92.72% duty on animal glues from Sweden during the years in question is authorized under 19 U.S.C. Sec. 1677e(b) (1982) which provides:

(b) Determinations to be made on best information available

In making their determinations under this subtitle, the administering authority and the Commission 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 [emphasis added];

and in accordance with the applicable implementing regulation, 19 C.F.R. Sec. 353.51(b) (1988), which provides in pertinent part:

(b) Whenever information cannot be satisfactorily verified, or is not submitted in a timely fashion or in the form required, the submitter of the information will be notified ald [sic] the affected determination will be made on the basis of the best information then otherwise available which may include the information submitted in support of the petition. An opportunity to correct inadequate submissions will be provided if the corrected submission is received in time to permit proper analysis and verification of the information concerned; otherwise no corrected submission will be taken into account.

Assuming for purposes of analysis that the 92.72% duty rate would be based on "the best information otherwise available," we cannot agree that the facts here, which are essentially undisputed, justify the ITA's invocation of this authority.

The basic error we perceive arises from the ITA's overly sweeping view of the authority it is granted under section 1677e(b). In essence, the ITA interprets the phrase "whenever a party ... refuses or is unable to produce information requested" to cover, in the ITA's discretion, any inadequacy or insufficiency of a reply to a request for any type of information. Indeed, even where a reply is complete, the ITA may, as it did in this case, conclude that the information does not answer a question it wishes to resolve, and for that reason the party is deemed to "refuse" or "be unable to supply" information within the meaning of the statute. We cannot agree that the ITA's authorization under section 1677e(b) extends so far.

In Atlantic Sugar, 744 F.2d at 1560, this court recognized that one may view section 1677e(b) as giving an agency (there, the ITC) some leverage against recalcitrant or noncooperative parties because the agency is required to "arrive at some determination." We agree that the ITA cannot be left merely to the largesse of the parties at their discretion to supply the ITA with information. This is particularly the case when the ITA is attempting to obtain information to conduct statutorily mandated administrative reviews because unlike ITC, the ITA has no subpoena power. See, e.g., Pistachio Group of the Ass'n of Food Indus. v. United States, 671 F. Supp. 31, 40 (Ct.Int'l Trade 1987). Thus, if the responses provided to an information request are only partially complete in that not all questions requiring a response are answered or answers to questions do not fully or accurately supply the information requested, partial completeness under section 1677e(b) may justify resort to the best information rule. See, e.g., Chinsung Indus. Co. v. United States, 705 F. Supp. 598, 600-01 (Ct.Int'l Trade 1989); Ceramica Regiomontana, S.A. v. United States, 636 F. Supp. 961, 966-67 (Ct. Int'l Trade 1986) (resort to best information available justified in countervailing duty determination where requested information as supplied was inaccurate in significant and material respects); Ansaldo Componenti, S.p.A. v. United States, 628 F. Supp. 198, 205 (Ct.Int'l Trade 1986) (resort to best information available justified where submissions to requests for information consistently partially complete); Tai Yang Metal Indus. Co. v. United States, 712 F. Supp. 973, 977 (Ct.Int'l Trade 1989) (resort to best information available justified where party served with questionnaire does not submit any answer even amidst assertion that party lacked financial capacity to assemble requested information). Otherwise, alleged unfair traders would be able to control the amount of antidumping duties by selectively providing the ITA with information. See, e.g., Pistachio Group, 671 F. Supp. at 40; Chinsung Indus., 705 F. Supp. at 601.

On the other hand, the ITA has not been given power that can be "wielded" arbitrarily as an "informal club." Atlantic Sugar, 744 F.2d at 1560. For example, the ITA may not properly invoke section 1677e(b) by making repeated requests for information which a party has already submitted until the party becomes frustrated and refuses to comply. Nor may it characterize a party's failure to list and give details of sales as a "refusal" or "inability" to give an answer where, in fact, there are no sales.

In the notice of final determination in this case, the ITA states that the reason for resort to the best information rule is that Extraco failed "to provide market sales data for similar merchandise." In its brief, the ITA relies solely on Extraco's alleged failure to supply data relating to grades 14, 18, and 22, and the parties' arguments as well are limited to those products. An agency may not, of course, shift the grounds it relied upon for taking action. As stated in SEC v. Chenery Corp., 318 U.S. 80, 87, 63 S. Ct. 454, 459, 87 L. Ed. 626 (1943), "[t]he grounds upon which an administrative order must be judged are those upon which the record discloses that its action was based." An initial question, therefore, is whether grades 14, 18, and 22, which are identical to the U.S. grades, are "similar" goods within the meaning of the above notice of final determination. We conclude that they are. In the May '82 telex, grades 14, 18 and 22 are denominated similar goods. Moreover, under 19 U.S.C. Sec. 1677(16) (1988), which defines "such or similar merchandise," the ITA would have to make its calculations of duty on grades 14, 18, 22 if it has valid data respecting those grades.8 Thus, we conclude that the published notice is based on Extraco's alleged failure to supply requested data on grades 14, 18, and 22 which caused the fictitious market issue to be resolved against Extraco. Accordingly, the propriety of the ITA's invocation of the "best information" rule depends on (1) whether Extraco's responses to information requests relating to grades 14, 18, and 22 were deficient in connection with the fictitious market issue, and (2) whether the ITA gave Extraco warning and an opportunity to correct any such deficiencies.

The ITA correctly notes that its letter accompanying the November '81 questionnaire contained an appropriate advance warning that complete answers were required to avoid the ITA's resort to the best information available. This multipage questionnaire was answered in January 1982, and it is the allegedly inadequate answers to part of this questionnaire on which the ITA seeks to lay the foundation for invoking its Sec. 1677e(b) authority. In one section of the questionnaire, the ITA asked a series of questions about Extraco's sales of grades 14, 18, and 22 to other countries prior to January 1978. Extraco had responded that it began producing grades 14, 18, and 22 in the fall of 1977, and that no such sales were made prior to January 1, 1978. The ITA maintains that Extraco's answers to these questions were deficient because Extraco listed no sales to other countries. In another section of the questionnaire, the ITA had asked, "Has any other customer in Sweden [other than the one customer for which data was given] ever purchased collex 14, 18, or 22?" to which Extraco replied, "No." The ITA maintains that this answer is deficient because Extraco "listed only one purchaser of Collexes [glues] 14, 18 and 22 in Sweden."

The ITA characterizes Extraco's conduct as a "refus[al] to provide the requested data." Contrary to the ITA's view, a "No" answer is not a refusal to provide data. If there is no data, "No" is a complete answer. Moreover, Extraco stated that it did not begin making grades 14, 18, 22 until the fall of 1977, and that the Swedish customer first ordered the glues on December 9, 1977. These answers are entirely consistent with its answers that it made no sales of these grades to Europe or third countries prior to January 1, 1978.9 The ITA points to the statutory language that it may use "best information available" when a submitter is "unable to produce information requested." The ITA is not so bold as to say explicitly this means it may resort to the "best information" rule where a submitter cannot produce data because such data never existed. In essence, however, that is its interpretation of its section 1677e(b) authority here, and we reject it.

But assuming the ITA did have reason to pursue the fictitious market allegations further, the ITA had to comply with 19 C.F.R. Sec. 353.51(b) requiring it to give notice to Extraco of perceived inadequacies of the responses to the November '81 questionnaire. The required notice, per the ITA, was provided in the May '82 telex. We disagree.

In the November '81 questionnaire, as indicated, the ITA was pursuing an inquiry into whether Extraco's sales to its U.S. and Swedish customer were in the ordinary course of trade or were fictitious sales. The ITA stated expressly in the accompanying letter that the answers to the questions in that questionnaire would likely resolve that issue. The ITA's telex, on which the ITA relies as notice, did not tell Extraco to amplify its responses to the November '81 questionnaire concerning the fictitious sales issue. On the contrary, the ITA raised a new objection, namely, that some of the Swedish sales of grades 14, 18, and 22, during part of the investigation period, were not "close in time" to the U.S. sales.10 There is no mention of their being fictitious. As far as this record shows, the inquiry into whether or not the reported sales were made in the ordinary course of trade was never pursued after Extraco's response to the November '81 questionnaire. Thus, the May '82 telex did not give notice of deficiencies in answers to the November '81 questionnaire. Moreover, the May '82 telex simply asked for data on additional sales of grades 14, 18, and 22 as before, albeit for a different reason, sales which Extraco had previously reported were nonexistent.11

Nothing in the record suggests that Extraco did, in fact, withhold any information on sales of grades 14, 18, and 22. To show that Extraco's responses were incomplete, the ITA would have to put forth at least some evidence tending to suggest that the answers provided by Extraco were inaccurate (e.g., that sales to other purchasers during the period were not listed). There is none. Moreover, most of the same information had been supplied with respect to sales of grades 14, 18, and 22 prior to the preliminary determination, and the ITA had already verified the existence of the reported sales. The exasperation of Extraco's officials in being asked to supply the information for a third time is understandable.

It may well be that the information provided by Extraco in response to the November '81 questionnaire was "incomplete" or "deficient" in the sense that the ITA did not obtain information which unequivocally negated or established the allegation that the home market sales were fictitious. But we agree with Olympic that this "incompleteness" or "deficiency" is not attributable to any deficiency in the answers provided by Extraco.12 The ITA may not properly conclude that resort to the best information rule is justified in circumstances where a questionnaire is sent and completely answered, just because the ITA concludes that the answers do not definitely resolve the overall issue presented. Although the ITA may properly request additional supplemental information, if needed, to fully resolve the issue, section 1677e(b) clearly requires noncompliance with an information request before resort to the best information rule is justified, whether due to refusal or mere inability. See, e.g., Daewoo Elec. Co. v. United States, 712 F. Supp. 931, 944 (Ct.Int'l Trade 1989). The ITA suggests that Extraco should have made efforts apart from the questionnaire to rebut the allegations of fictitious sales. However, section 1677e(b) does not place that obligation on the submitter.13 To avoid the threat of Sec. 1677e(b), a submitter need only provide complete answers to the questions presented in an information request.

In sum, because Extraco did not fail to provide the information requested in the November '81 questionnaire with respect to sales of grades 14, 18, and 22, this questionnaire provides no substantial evidence indicating that Extraco "refused" or "was unable" to supply the ITA with information within the meaning of those terms in the statute. Moreover, the May '82 telex did not give notice that Extraco's answers were deficient with respect to the allegation that the reported sales of grades 14, 18, and 22 were fictitious. The ITA could not, therefore, justifiably resort to the "best information" rule.

Because the ITA's decision to use the best information rule did not comply with the statutory requirements of 19 U.S.C. Sec. 1677e(b) and the regulatory requirements of 19 C.F.R. Sec. 353.51(b), the judgment of the Court of International Trade must be reversed. Further, due to the passage of time, Extraco's 1981 exit from the production of bone glues,14 the lack of any evidence supporting the fictitious home market sales allegation, and in the interests of judicial economy, we remand this cause to the Court of International Trade with instructions to direct the ITA to assess antidumping duties in accord with the margins determined in the preliminary investigation for the years in question.

Costs are awarded to Olympic Adhesives, Inc.