PUBLIC VERSION
OPINION AND ORDER
MUSGRAVE, Senior Judge: As surety on imports from respondent Xuzhou Jinjiang Foodstuffs Co., Ltd. (“Xuzhou” or “Jinjiang”), the plaintiff Washington International Insurance Company (“WII”) moves for judgment pursuant to USCIT Rule 56.2 on certain decisions of the 2005-2006 administrative review of the antidumping duty order on freshwater crawfish tail meat from the People’s Republic of China (“PRC”), as compiled by the international Trade Administration of the U.S. Department of Commerce (“Commerce”). See Freshwater Crawfish Tail Meat From the People’s Republic of China, 73 Fed. Reg. 20249 (Apr. 15, 2008) (finalresults), Public Document (“PDoc”) 135 (“Final Results”).1 Jurisdiction here is pursuant to 28 U.S.C. § 1581(c). For the following reasons, there is substantial evidence to support the decision to reject Xuzhou’s claim that certain U.S. sales were non-subject merchandise, to resort to facts available, and to apply an adverse inference (“AFA”) therefor; however, the record does not support the decision to use the PRC-wide rate of 223.01% as AFA for Xuzhou, and the matter must therefore be remanded for recalculation of the AFA rate.
Background
Immediately prior to the instant administrative review, Xuzhou obtained its own dumping margin from Commerce through partici*1024pation in new shipper and administrative reviews. See Freshwater Crawfish Tail Meat from the People’s Republic of China, 72 Fed. Reg. 19174 (Apr. 17,2007) (final results of new shipper and administrative reviews); see generally 19 C.F.R. § 351.214. Based upon bona fide sales during the period September 1, 2004 through October 5, 2005, Xuzhou’s margin of dumping was calculated to have been 0.00%. Id. See also Freshwater Crawfish Tail Meat From the People’s Republic of China, 71 Fed. Reg. 59432 (Oct. 10, 2006) (preliminary results of new shipper and administrative reviews).
For the administrative review at bar, Xuzhou reported making [¶] ]] United States sales of subject merchandise between September 1, 2005 and August 31, 2006 (the “POR”).2 Commerce then placed a memorandum on the record, together with copies of entry documentation received from the U.S. Customs and Border Protection (“CBP”), regarding [¶] ]] imports from Xuzhou (the “first disputed entries”) that questioned whether Xuzhou had reported all sales of subject merchandise. Public Record Document (“PDoc”) 61, CDoc 22 (Mar. 30, 2007). The document invited comment from the parties.3
Xuzhou’s response sought to clarify the entry documents with respect to the entry code declarations, count size descriptions, change in product description (purportedly in response to request by CBP), and certain clerical errors. See generally PDoc 65 (Apr. 13, 2007), CDoc 23 (Apr. 12, 2007). By way of broader explanation, Xuxhou alleged that [¶] ]] prepared shipment(s) of subject merchandise involved substitution of non-subject merchandise (i.e. whole crawfish) before leaving port in response to a last-minute request from, and after transmission of the original order documentation to, the customer. Xuzhou alleged that it sent to the customer new set(s) of commercial documents bearing the same invoice number(s) to reflect the change but that apparently the importer or its broker submitted the wrong documentation for customs declaration. This error was further compounded, as told by Xuzhou, because Xuzhou’s sales person did not properly revise the unit price on one of the invoices to reflect non-subject merchandise, although supposedly the proper sales revenue for the invoice was entered on Xuzhou’s sales ledger that had been submitted to Commerce as part of its sales reconciliation. See CDoc 23 at 4-5. Xuzhou also claimed to submit copies of the correct invoices for the [¶] ]] shipment(s) as an attachment to its clarification to Com*1025merce. See id. at Attachment 1.
As of June 4, 2007, Commerce had not found any attempt by Xuzhou to change the allegedly incorrect entry coding of the [¶] ]] shipment(s) declared upon importation to be subject merchandise (apparently pursuant to their Customs Form (“CF”) 7501s). See PDoc 69, CDoc 24 (memorandum to file dated June 6, 2007). By June 7, 2007, for reasons that are unclear, CBP’s database reflected that the remainder of the first disputed entries had been reclassified to subject merchandise. See PDoc 72, CDoc 27 (memorandum to file dated June 18, 2007). Commerce also obtained information from the OASIS database of the U.S. Food and Drug Administration (“FDA”)4 implying [¶] ]] entries of subject merchandise (including [¶] ]] of the first disputed entries) had not been declared as such. See PDoc 70, CDoc 25 (memorandum to file dated June 12, 2007).
In response to those observations, Xuzhou reiterated to Commerce that its earlier comments had “fully explained” each5 of the alleged unreported sales of subject merchandise, and Xuzhou further commented that, as reflected in the OASIS database, the FDA had not conducted actual inspection of any of the [¶] ]] entries in question but had merely re-transcribed the erroneous documentation describing the merchandise as subject merchandise. PDoc 76, CDoc 29 (July 6, 2007). Xuzhou also explained it was not in a position to redress CBP’s reclassifications or address why CBP had reclassified in the first instance, and suggested Commerce seek answers from the customer concerned, but regardless, “[[ ]].” See id. at 2-3. Nonetheless, by September 2007, CBP’s database reflected [¶] ]] entries CBP had reclassified as subject merchandise, and Commerce placed another memorandum to that effect in the record. Cf. PDoc 85, CDoc 35 (Sep. 21, 2007).
Commerce published its preliminary results the following month. Freshwater Crawfish Tail Meat From the People’s Republic of China, 72 Fed. Reg. 57288, 57295 (Oct. 9, 2007) (preliminary results), PDoc 91. They inferred that Xuzhou had failed to report all sales of subject merchandise and that the circumstances of that failure supported application of a total adverse inference and the assessment of anti-dumping duties in the amount of 223.01%. See id.
*1026After gaining admission to the proceeding as an interested party, see PDoc 93 (Oct. 19,2007), PDoc 107 (Nov. 28,2007), WII commented on the preliminary results alongside Xuxhou via case briefing. See PDoc 115, CDoc 43 (Dec. 17, 2007), and PDoc 116, CDoc 44 (Dec. 18, 2007), respectively. The briefing essentially claimed that the evidence of record showed Xuzhou had properly reported all sales of subject merchandise, that total adverse inferences were not appropriate, and that the rate Commerce had selected therefor was inappropriate in any event. See id. Commerce then placed on the record a memorandum summarizing other information obtained from the FDA pertaining to three entries (the “second disputed entries”) that had been reclassified. PDoc 122, CDoc 48 (Feb. 7, 2008). Cf. PDoc 85, CDoc 35. The information included photographs of samples identified as being from two entries, [¶] ]]. The photographs each show a plastic bag labeled “crawfish tails” and the ingredients as [¶] ]] and, for one entry, the count size as [¶] ]].” See PDoc 122, CDoc 48. Xuzhou and WII further responded by explaining that after the purchase order change, Xuzhou insisted the importer accept the whole crawfish in bags labeled as crawfish tail meat because it had already made a substantial investment in the bags, and the importer agreed because it intended to repack the merchandise upon importation anyway. PDocs 126, CDoc 49 (Feb. 22, 2008), and PDoc 128, CDoc 50 (Feb. 25, 2008), respectively. They also argued there was no evidence directly linking these photographs to the entries in question.
Commerce ultimately found this explanation incredible because the “pictures clearly show that the bags contain crawfish tail meat, not whole crawfish,” “tail meat bags are inappropriately small for whole crawfish[,]” Xuzhou “failed to provide documentary evidence that discussed the order change that supposedly led to packing whole crawfish in bags labeled as tail meat (e.g., letters, email correspondence, or cancelled or new purchase orders),” and Xuzhou also was not forthcoming with the fact that “the order change resulted in shipping whole crawfish in bags labeled as crawfish tail meat until Commerce placed the pictures of the bags on the record.” Proprietary Issues and Decision Memorandum (Comment 3), CDoc 52 (Apr. 7, 2008) (“AFA memo”), at 11. For the final review results, Commerce rejected WII’s and Xuzhou’s claims and published the final results in April 2008 reiterating an antidumping duty rate of 223.01% for Xuzhou as a result of applying AFA. 73 Fed. Reg. at 20250; PDoc 135.
Standard of Review
"When reviewing challenges to final results of administrative reviews of antidumping duty orders, this Court will sustain any deter*1027mination, finding or conclusion of Commerce unless it is “unsupported by substantial evidence on the record, or is otherwise not in accordance with law.” 19 U.S.C. § 1516a(b)(l)(B)(i). “Substantial evidence” means “such relevant evidence as a reasonable mind might accept as adequate to support a conclusion.” Consol. Edison Co. v. NLRB, 305 U.S. 197, 229 (1938). It is determined by considering all record evidence, including whatever “fairly detracts from the sub-stantiality of the evidence.” Atlantic Sugar, Ltd. v. United States, 744 F.2d 1556, 1562 (Fed. Cir. 1984). It does not, however, involve reweighing the evidence, as that would usurp the administrative agency’s fact-finding function. See, e.g., Matsushita Elec. Indus. Corp. v. United States, 750 F.2d 927, 936 (1984). Rather, “the possibility of drawing two inconsistent 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 (1966). But still, “the administrative action is to be tested by the basis upon which it purports to rest,” SEC v. Chenery Corp., 332 U.S. 194, 196 (1947), and therefore the agency’s reasoning must be clear, id., although “less than ideal clarity” may be upheld “if the agency’s path may reasonably be discerned.” Bowman Transp., Inc. v. Ark.-Best Freight Sys., Inc., 419 U.S. 281, 286 (1974).
Discussion
For the following reasons, the court concludes, contrary to WII’s arguments, that Commerce’s analysis was not based entirely on “mere” speculation6 and that substantial evidence supports rejecting the claim that Xuzhou had properly reported all of its U.S. market sales of subject merchandise. Substantial evidence also supports finding that Xuzhou failed to act to the best of its ability during the proceeding, which is the necessary predicate for using AFA. The record does not, however, provide substantial evidence to support the use of the PRC-wide rate as AFA, and therefore the matter must be remanded for reconsideration. Further, because the court cannot discern whether Commerce would infer the same results in the absence of speculation, on remand Commerce will consider whether partial rather than total AFA is applicable.
I. The Determination That Xuzhou Failed to Report All POR Sales to the United States
As described above, Commerce relied on four threads of circumstantial evidence to reject Xuzhou’s claim that it had reported all of its *1028POR sales to the United States: (1) entry documentation obtained from CBP, specifically for [¶] ]] entries declared as subject merchandise on their CF 7501s but described as non-subject merchandise in other documents, and for [¶] ]] entries declared as non-subject merchandise on their CF 7501s but described in supporting documentation in ways Commerce interpreted as consistent with subject merchandise, e.g. on [¶] ]] or on the entry’s CF 7501 itself7; (2) CBP reclassification of the latter entries to crawfish tail meat as well as subsequent CBP reclassification of an additional [¶] ]] entries; (3) count sizes listed on the entries’ sales invoices that Commerce interpreted as consistent with entries of subject merchandise; and (4) information obtained from the U.S. Food and Drug Administration (FDA), including OASIS database reports, color photographs by an FDA agent of two of those entries plus a warehouse receiving report for one of those entries, and a summary of an FDA surveillance assignment “identifying” a third entry as crawfish tail meat. See generally CDoc 52.
WII presents a litany of arguments to challenge these findings: Xuzhou had only limited experience with antidumping proceedings, this being only its second antidumping duty proceeding; Xuzhou nonetheless properly reported all of its U.S. sales; innocent inconsistencies and errors are a common occurrence among importers in the preparation of entry documents;8 importers are only required to specify “tail meat” on entry papers but not “crawfish” as “whole” crawfish; the general description of the imported merchandise9 on Xuzhou’s entry papers (even accepting Commerce’s unwarranted interpretation of them) does not support the conclusion that the first disputed entries are in fact entries of tail meat and to infer otherwise is speculative, arbitrary, capricious, and necessarily shows Commerce considered the entry code controlling over general descriptions for some entries but considered the generic description “crawfish” as meaning “tail meat” and controlling over the entry code declaration of non-subject merchandise for the other entries; Commerce improperly relied on the lapse of 200 days, to completion of the Final Results, after the importer had first been alerted to the issue of CBP reclas-*1029sifications, and on the fact that of those only one had been protested, which was denied; notwithstanding CBP-reclassification, all of the affected entries were correctly declared upon entry as whole crawfish, and the importer certified in response to CBP’s request that the entries are all [ [ ] ];10 the fact that CBP may have reclassified non-subject merchandise to subject merchandise is of no evidentiary value because CBP operates under its own standards, regulations, policies and procedures; there is no record evidence on why CBP reclassified certain entries; Commerce cannot properly delegate its administrative mandate to determine whether these entries are, in fact, entries of subject merchandise; Commerce’s action effectively abrogated its long-standing policy of not deferring to CBP on decisions implicating dumping matters; Commerce’s count size theory11 is mere speculation and does nothing to undermine Xuzhou’s differentiation of count sizes for whole crawfish and craw-fish tail meat because when calculating a count for the first disputed entries on a 10-pound basis the count for those entries is “entirely consistent with whole crawfish[;]” the FDA’s OASIS data reports contradict each other over the same entries and are therefore not probative on the issue of the actual merchandise entered; the FDA photographs regarding the second disputed entries amount to a “total failure of proof’ because there is no reliable record evidence directly linking the photographs to them; and Commerce ignored exculpatory documents on the matter purely on procedural grounds. See generally Pl.’s Br.; Pl.’s Reply; see also PDoc 111, CDoc 41 (Dec. 4, 2007) (rejecting submission of WII).
At this stage, the function of the court is not to re-weigh the evidence, only to determine whether the record of it in support of a given conclusion is substantial. The court can agree that mere speculation surrounds much of Commerce’s mustering of it, but the burden was on Xuzhou to create an adequate record.12 WII may be correct that the entry documents for the first disputed entries are inherently *1030contradictory, but when the evidence surrounding an interested party’s claim is inconclusive, Commerce may find that the party has not met its burden of proof and resort to facts available. For example, the entries WII contends were mistakenly declared to CBP to be subject merchandise were, nonetheless, so declared, and a party claiming otherwise must provide sufficient proof of its position beyond asserting that the declarations were mistakes. See 28 U.S.C. § 2639(a)(1).
The record before the court reveals that Commerce attempted to rely on the evidence gathered as a whole, not solely any particular strand, to deny that Xuzhou had properly reported all of its U.S. sales, and that reliance was not entirely speculative. Commerce’s analysis of the first disputed entries emphasizes the absence on the record of accurate [¶] ]] as well as inconsistency in Xuzhou’s explanations, particularly the fact that the declared sales price for a particular entry, alleged to have been mistakenly described on its CF 7501 as crawfish tail meat (entry number [¶] ]]), was nearer the per-pound price of crawfish tail meat than whole crawfish, and the' apparent fact that after mid-July 2006 Xuzhou began to add the word “whole” in response to request from CBP to clarify Xuzhou’s description of “[[ ]],” when “most” of the non-subject merchandise entered after that time were still identified as “[[ ]]” in invoices with no mention of “whole” crawfish, C.Doc 52 at 9. WII’s argumentation does not adequately clarify such inconsistencies, and it therefore cannot be concluded that the weight the agency accorded them was unreasonable.
At the same time, a number of Commerce’s inferences from the record are tenuous. On the subject of CBP reclassifications, for example, although an administrative agency has the power to create a presumption, the presumption “must rest on a sound factual connection between the proved and inferred facts.” British Steel plc v. United States, 20 CIT 663, 698, 929 F. Supp. 426, 454 (1996) (quoting NLRB v. Baptist Hosp., Inc., 442 U.S. 773, 787 (1979)). Presumptions maybe overcome, and they only extend so far.13 Towards that end, WII is correct that the determination of “subject merchandise” is a non-delegable duty that Commerce does not properly discharge to the extent the Final Results imply that reclassification itself is sufficient proof that entries were in fact subject merchandise. If Commerce is to give any weight to CBP reclassification, it must take into account any suspension of liquidation that the importer must abide before protesting, and Commerce does not appear to have done so in this instance. The AFA memo merely notes that an “Exhibit 2” of WII’s administrative case brief lent “credence” to CBP’s reclassification of *1031one entry, CDoc 52 at 10, but that is insufficient to substantiate the reliability of all CBP reclassifications in the non-delegable determination of whether the imported merchandise is in fact whole or tail meat crawfish. t
Regarding Commerce’s rejection of Xuzhou’s count size explanation, WII has a point, but only to ascertain extent: Commerce may have found uniform consistency in per-pound pricing among vendors on the internet, but that does not directly contradict or even undermine Xuzhou’s explanation.14 And yetf-if Xuzhou’s explanation is plausible, Commerce’s theory is also at least as plausible. Commerce observed that there were no other documents of record that describe a 10-pound count, that there was one POR entry of whole crawfish with a per-pound count, and that Xuzhou sold crawfish tail meat with a per-pound count to the same customer to whom Xuzhou claimed to have sold whole crawfish with a jper-10-pound count. The AFA memo summarizes that these “fact[s] raise [ ] questions as to why a customer who wanted whole-crawfish counts to be expressed on a 10-pound basis . . . continued to permit tail meat counts to be on a per-pound basis.” CDoc 52 at 10. Given that the burden was on Xuzhou to provide substantial evidence to support its claim, Commerce’s expressed doubt on Xuzhou’s “proffer” on this point was not unreasonable, but such doubt is to be construed as limited: Commerce’s count theory would apply with respect tó all whole crawfish entries, not merely the first disputed entries, and yet, as Xuzhou pointed out, Commerce does not appear to have éxplicitly voiced any concerns over Xuzhou’s count-size methodology when considering the other entries that Xuzhou had declared as whole crawfish. See, e.g., CDoc 44 at 4.
Regarding the OASIS data, Wil is correct that they are derivative, duplicative, and contradictory, to an extent. Cf. CDoc 25 with CDoc 48. In particular, the reports overlap with respect to three entries, i.e., the second disputed entries. According to the first report, all three entries were released into commerce!, after either [¶] ]], and all three were said to consist of whole crawfish. According to the second FDA report, two of the entries were “sampled and analyzed” while the third entry is summarized as merely having had “an FDA *1032inspector note[ ] that the entry’s package was labeled in part cooked[,] peeled and deveined crawfish tail meat.” This latter observation is plainly not substantial evidence as to the contents of the package. And cf. CDoc 25 (indicating release of entry, of whole crawfish, after [¶] ]]). Evidentiary consideration must necessarily abide its own limits and context, including proffered explanation therefor.
That said, contrary to WII’s argument the court is unable to conclude the extent to which the FDA reports cancel each other out. The most “telling” evidence that persuaded Commerce are two photographs of purported samples from the other two entries in the second FDA report, from which their contents may be deduced, and Commerce deduced that they consisted of tail meat. See also CDoc 52 at 11. Commerce’s AFA memo notes that one of the photographed bag shows a brand name, size, and count that correspond to a warehouse receiving report, which in turn ties to the commercial invoice for the entry (as does the date written on the photograph). CDoc 52 at 11. WII implies there is nothing on the record to prove that the photographs are of actual Xuzhou product, but that is insufficient, without more, to overcome the presumption of administrative regularity that the FDA was entitled to be accorded, at least as to the relevance of the photographs to the purported supporting documentation. The court is thus unable to conclude that the two FDA reports canceled each other out as a matter of law, or that Commerce abused its discretion by according a heavy weight to the photographs in the second FDA report when considering the evidentiary record.
II. The Determination to Resort to “Total” Adverse Facts Available
Taken as a whole, the foregoing amounts to “more than a mere scintilla” to support the agency determination to reject Xuzhou’s assertion that it had properly reported all sales of subject merchandise. It was therefore necessary to resort to “facts otherwise available” in the determination in order to fill a gap in information. See 19 U.S.C. § 1677e(a). Towards that goal, if a respondent has “failed to cooperate by not acting to the best of its ability to comply with a request for information [,]” then Commerce may, but need not, use an adverse inference when resorting to facts otherwise available (AFA). 19 U.S.C. § 1677e(b). If it does, Commerce must “articulate why it concluded that a party failed to act to the best of its ability, and explain why the absence of this information is of significance to the progress of its investigation.” Mannesmannrohren-Werke AG v. United States, 23 CIT 826, 839, 77 F. Supp. 2d 1302, 1313-14 (1999).
According to Nippon Steel Corp. v. United States, 337 F.3d 1373, 1382 (Fed. Cir. 2003), the application of section 1677e(b) is to be *1033evaluated on the bases of objective and subjective showings on the record. The objective showing is that “a reasonable and responsible [respondent] would have known that the requested information was required to be kept and maintained under the applicable statutes, rules, and regulations.” Id. at 1382 (citation omitted). The subjective showing is whether the respondent’s failure to fully respond is the result of a lack of cooperation in either (a) failing to keep and maintain all required records or (b) failing to put forth maximum effort to investigate and obtain requested information from its records. Id. at 1382-83.
The Final Results fail to address Nippon Steel properly; therefore, those showings must be gleaned from the record. Regarding the objective showing, Commerce was asked to consider Xuzhou’s limited antidumping procedure experience (i.e., what is expected of a “reasonable” respondent), but the reason Xuzhou gave for the last-minute order change was [¶] ]]. See CDoc 44 at 12. The record does not reveal whether exigencies at the port of shipment precluded properly revising the [¶] ]] (cf., e.g., CDoc 23 at 4-5; CDoc 44 at 9-11), but whether it was “technically” unnecessary that they be revised for the purpose of antidumping duty administration, Xuzhou’s explanation at least implies awareness that its U.S. sales activities for this review period would be scrutinized.15
Regarding the subjective showing, Commerce ultimately inferred that “the sales in question are subject merchandise sales[,]” that ‘Xuzhou’s failure to report these sales, despite the fact that it possessed the necessary records regarding these sales, indicates a lack of cooperation on its part[,]” and that ‘Xuzhou did not report the sales in question, or indicate that it lacked the records needed to report such sales.”16 CDoc 52 at 15-16. It is unclear what “the sales in question” *1034encompasses, the record lacks evidence sufficient to reject Xuzhou’s non-subject-merchandise claim as to certain sales, and the latter point rather mis-characterizes Xuzhou’s (and WII’s) entire position regarding the disputed sales. See generally CDoc 44. There is no evidence to support the determination that Xuzhou actually “possessed” such documents and failed to cooperate by not providing them; it would have been quite impossible, as WII argues, for Xuzhou to produce requested documents, regarding alleged U.S. sales of subject merchandise, that did not exist. See Pl.’s Reply at 5 (referencing Olympic Adhesives v. United States, 899 F.2d 1565, 1572-73 (Fed. Cir. 1990)). And regardless of whether a failure to make and maintain documentation equates to a failure to act to the “best of one’s ability” in a given instance,17 irresponsible data recording, attributable to Xuzhou, was not clearly and explicitly stated as a reason for applying total AFA in this instance, and it would be presumptuous to so find on judicial review.
Be that as it may, as discussed above there is substantial evidence on the record - the FDA photographs - to support the inference that two entries consisted of crawfish tail meat (entry numbers [¶]
]]), see CDoc 48, that there was therefore irresponsible data reporting as to those two entries, and that Commerce not unreasonably concluded that such failure “indicates a lack of cooperation on [Xuzhou’s] part.” CDoc 52 at 15. As to why that entry situation occurred, the court cannot speculate, but given that Xuzhou engaged in the arduous and expensive process of submitting to a new shipper review and its obvious knowledge from then on that its exports to the U.S. would certainly be subject to strict scrutiny, the circumstance is bizarre, to say the least. At any rate, the question for Commerce at *1035that point was whether that situation justified rejection of all of Xuzhou’s information or whether it had to consider the application of partial adverse facts available.
19 U.S.C. § 1677m(e) requires Commerce not to ignore information that has been submitted and is necessary to the determination, provided, inter alia, the interested party can show “that it acted to the best of its ability to provide the information.”18 The requirement apparently places the burden on the interested party to prove that it acted to the best of its ability to provide the information, cf. NSK Ltd., supra, 481 F.3d at 1360 n.1 (unnecessary to consider whether respondent satisfied best-ability requirement of § 1677m(e) because it failed to satisfy best-ability requirement of § 1677e(b)), but actions speak louder than words, and it may often be discerned from the record whether that was in fact the case.
WII here contends that the sales Commerce considered “unreported” amounted to only [¶] ]]% of the total number of U.S. sales and that such a “relatively small” percentage does not justify resort to total adverse facts available. WII Br. at 22-23. The government, however, points out that the quantity of [¶] ]] “unreported” sales each amounted to approximately [¶] ]]% of the total quantity of reported sales, Def.’s Br. at 20-21 (see CDoc 52 at 15-16 & n.24), and that such a measure cannot be said to be insubstantial. The court is constrained to agree. The better argument for WII is that Commerce did not voice any apparent concern or problem with the majority of Xuzhou’s total reported U.S. sales beyond merely declaring in the AFA memo that “the information . . . provided by *1036Xuzhou cannot serve as a reliable basis for reaching a determination[.]” CDoc 52 at 15. Commerce does not specify why this is so, and the record indicates otherwise, e.g., the only evidence that attempts to cast doubt on the reliability of some of Xuzhou’s other reported U.S. sales is reclassification - which, again, is not self-evidently probative, especially pending liquidation.
Although a decision of less than ideal clarity may be upheld if the reasonableness of an agency’s thought process may be discerned, Bowman, supra, 419 U.S. at 286, the Final Results must stand their own ground. See Chenery, supra, 332 U.S. at 196. Substantial evidence of record supports Commerce’s finding that “Xuzhou made a significant number of sales of subject merchandise that it failed to report [,]” see CDoc 52 at 12, and, accordingly, AFA may be appropriate as to those entries for which substantial evidence of record supports that inference. However, the record does not establish why the remainder of Xuzhou’s information must be rejected, i.e., that all information Xuzhou provided is unreliable, and the court cannot discern whether unwarranted inferences on some of the evidence of record amounted to mere harmless error. On remand, see also infra, Commerce will therefore have a fresh opportunity for reconsideration and elucidation on whether the record supports, in the selection of facts otherwise available, the application of an adverse inference, partial or total, or none at all, as appropriate to the circumstances.
III. The AFA Rate Selected by Commerce
As mentioned, the Final Results increased Xuzhou’s margin from 0% to over 223% based on use of the PRC-wide rate as total adverse facts available. Because that rate is “secondary” information, Commerce was required to corroborate it, to the extent possible, before relying on it. 19 U.S.C. § 1677e(c). But cf. H.R. Rep. 103-826(1) at 105 (1994), reprinted in 1994 U.S.C.C.A.N. 3773, 3877 (“[t]he fact that corroboration may not be practicable in a given circumstance will not prevent the agencies from applying an adverse inference under subsection (b)”). Towards that end, the AFA memo explains
there are no independent sources of information from which [Commerce] can derive calculated dumping margins; the only source [s] for dumping margins [are] administrative determinations. The rate that we are using as AFA is reliable because it was calculated in the 1999-2000 antidumping duty administrative review in this proceeding using respondent data that were accepted by [Commerce] and surrogate values that were selected by [Commerce].... This rate has been used as an AFA rate in every segment of this proceeding since the 1999-2000 anti-*1037dumping duty administrative review and [Commerce] has received no information that warrants revisiting the issue of its reliability.
***
Moreover, we disagree with WII’s claim that this rate is not appropriate for Xuzhou. Comparing the entered value of a large quantity of unreported subject merchandise sales in this review (USD [¶] ]] per kilogram)[FN 30] to the average normal value calculated for Xuzhou in the 2004-2005 new shipper review (USD [¶] ]] per kilogram) results in a[n] estimated dumping margin of [¶] ]]%. . . . The [¶] ]]% is a conservative approximation of Xuzhou’s dumping margin since we did not take into account all of the adjustments normally made in calculating a dumping margin.1 ] Furthermore, we believe that the data underlying the [¶] ]]% rate are more relevant to Xuzhou’s current practices than the rate from the company’s new shipper review because they are based on a large quantity of sales in the current period rather than [¶] ]] sales from the prior new shipper POR. Thus, contrary to WII’s claims, we believe that the AFA rate of 223.01%[ ] is a reasonably accurate, relevant, and reliable estimate of Xuzhou’s experience, and sufficiently adverse as to effectuate the purpose of facts available to induce parties to report complete and accurate information. Use of this highest prior calculated margin reflects the common sense inference that it is probative of Xuzhou’s current margins, “because otherwise {Xuzhou} would have produced current information showing the margin to be less.”[ ]
CDoc 52 at 18-19 (footnotes and citations omitted in part, confidential bracketing moved from “quantity” in footnote 30) (paraphrasing Rhone Poulenc, Inc. v. United States, 899 F.2d 1185, 1190 (Fed. Cir. 1990)).
WII argues the AFA rate is punitive, non-contemporaneous (being over eight years old), and either uncorroborated or “corroborated” via “junk methodology.” Pl.’s Br. at 24-27; Pl.’s Reply at 6-15. The court agrees. While a successful challenge to a margin’s contemporaneity normally ought to require probative evidence of a respondent’s cur*1038rent actual experience,19 in this instance Commerce’s own calculation undercuts its so-called “common sense inference that [223.01%] is probative of Xuzhou’s current margins[.]” CDoc 52 at 19. See Rhone Poulenc, 899 F.2d at 1190. There is no basis on the record or in law for concluding that a [¶] ]] percentage point differential, amounting to [¶] ]]% over the calculated estimate of Xuzhou’s “current market conditions” (ostensibly Xuzhou’s “actual” margin of dumping for [¶] ]] entries) is “corroboration” of the contemporaneity, and therefore applicability, of the 1999-2000 AFA rate, even allowing for a “built-in increase intended as a deterrent to non-compliance.” Fratelli De Cecco di Fillippo Fara San Martino S.p.A. v. United States, 216 F.3d 1027, 1032 (Fed. Cir. 2000).20
Further, the extremity of comparing so-called “unreported subject merchandise sales” of Xuzhou (that were transacted at a U.S. dollar price more obviously approximating that of non-subject merchandise than subject merchandise) to Xuzhou’s new shipper rate may be tempered somewhat by the fact that the former involves a larger quantity of merchandise, as Commerce indicates (CDoc 52 at 19), but Commerce provides no proof to support its implicit inference that the new shipper review did not involve a representative sample population of merchandise, i.e., is unreliable evidence of Xuzhou’s most recent market conditions and experience. See, e.g., Malay Ghosh and Glen Meeden, Bayesian Methods for Finite Population Sampling (Monographs on Statistics and Applied Probability 79) (Chapman & Hall 1997).
Furthermore, based on the foregoing, the methodology behind the [¶] ]] rate currently incorporates the entered values of supposed “unreported” sales of subject merchandise for which the only evidence of record to support that inference is CBP reclassification, which, as mentioned, is insufficient, standing alone, to support that inference.21 And, if partial rather than total AFA is appropriate, then it is obvi*1039ously appropriate to incorporate in such a calculation the entered values of the [¶] ]] declared shipments of subject merchandise and other information for which the record contains no reasonable inference indicating unreliability or incompleteness.
Lastly, the court notes in passing that the matter at bar also bears an uncanny resemblance to Shandong Huarong Gen. Group. Corp. v. United States, 27 CIT 1568 (2003), after remand, 28 CIT 1624 (2004), after re-remand, 29 CIT 1227 (2005), re-re-remand results sustained, Slip Op. 07-04, 31 CIT_(Jan. 9, 2007), which considered a similar extreme divergence in AFA from a respondent’s most recent administratively verified experience to be irrational. 28 CIT at 1634. See also China Kingdom Imp. & Exp. Co., Ltd v. United States, 31 CIT _, 507 F. Supp. 2d 1337, 1361-62 (2007). But then again, it is difficult to discern from the record at bar where responsibility for such irrationality lies.
At any rate, on remand Commerce must use a rate that is a “reasonably accurate estimate” of a respondent’s margin of dumping activity during the period of review. See, e.g., Ta Chen Stainless Steel Pipe, Inc. v. United States, 298 F.3d 1330, 1340 (Fed. Cir. 2002) (use of margin within range of respondent’s actual experience was not unreasonable AFA); PAM, S.p.A. v. United States, 31 CIT_,_, 495 F. Supp. 2d 1360, 1373 (2007) (no finding that rate was “reasonably accurate estimate” of respondent’s activity). Anything beyond that will appear punitive and aberrational, since, as observed many times in the past,
Congress could not have intended for Commerce’s discretion to include the ability to select unreasonably high rates with no relationship to the respondent’s actual dumping margin. Obviously a higher adverse margin creates a stronger deterrent, but Congress tempered deterrent value with the corroboration requirement. It could only have done so to prevent the petition rate (or other adverse inference rate), when unreasonable, from prevailing and to block any temptation by Commerce to overreach reality in seeking to maximize deterrence.
F.lli De Cecco, 216 F.3d at 1032. See, e.g., Tianjin Machinery Import & Export Corp. v. United States, 31 CIT_, Slip Op. 07 — 131 at 35 (2007); Gerber Food (Yunnan) Co., Ltd. v. United States, 29 CIT 753, 770-771, 387 F. Supp. 2d 1270, 1286 (2005); Shanghai Taoen Intern. Co., Ltd. v. United States, 29 CIT 189, 196-97, 360 F. Supp. 2d 1339, 1346 (2005); Shandong Huarong Gen. Group Corp., supra, 28 CIT at *10401631; Allegheny Ludlum Corp. v. United States, 27 CIT 1034, 1050, 276 F Supp. 2d 1344, 1359 (2003); Mannesmannrohren-Werke AG v. United States, 24 CIT 1082, 1097 n.4, 120 F. Supp. 2d 1075, 1087 n.4 (2000); American Silicon Technologies v. United States, 24 CIT 612, 626, 110 F. Supp. 2d 992, 1004 (2000).
Conclusion
For the foregoing reasons, substantial evidence on the record supports resort to facts available and an adverse inference in the selection thereof, but the matter must be, and it hereby is, remanded for redetermination of an adverse facts available (AFA) rate that more closely reflects Xuzhou’s then-current market practices during the POR, “albeit with some built-in increase intended as a deterrent to non-compliance[,]” 216 F.3d at 1032, and also for reconsideration of whether the circumstances warrant partial or total AFA. The results of remand shall be filed within 30 days, comments thereon within 15 days, and rebuttal, if any, within ten days thereafter.
See also Freshwater Crawfish Tail Meat From the People’s Republic of China, 62 Fed. Reg. 48218 (Sep. 15, 1997) (amendment to final less than fair value determination and anti-dumping duty order.
See, e.g., Confidential Record Document (“CDoc”) 7 at Ex. A-1 (Jan. 16, 2007); CDoc 11 at Ex. C-1 (Jan. 31, 2007); CDoc 13 at S-5 (Feb. 14, 2007); CDoc 15 at 2 (Feb. 26, 2007); CDoc 16 (Mar. 2, 2007).
WII does not here directly challenge the sufficiency of “notice” of Xuzhou’s “deficiency(s)” under 19 U.S.C. § 1677m(d).
According to a webpage on the FDA’s website, “OASIS is an automated FDA system for processing and making admissibility determinations for shipments of foreign-origin FDA-regulated products seeking to enter domestic commerce.” http://www.fda.gov/ForIndustry/ ImportProgram/AdmissibilityDeterminationsforShipmentsofForeign-originOASIS/ default.htm (last visited this date).
The court cannot discern that entry number [¶] ]] was ever addressed in Xuzhou’s April 12, 2007 submission; on the other hand, a copy of a commercial invoice referencing that entry number and apparently reflecting a per-pound price perhaps consistent with that of non-subject merchandise is attached to Xuzhou’s July 6,2007 comments. See PDoc 76, CDoc 29.
Speculation is not substantial evidence on the record to support the application of adverse facts available. See, e.g., Asociacion Columbiana Exportadores de Flores v. United States, 23 CIT 148, 153-154, 40 F. Supp. 2d 466, 472 (1999) (speculation does not satisfy the substantial evidence on the record review standard).
According to the government, for [¶] ]]. Def.’s Br. at 11-12, referencing CDoc 52 at 4. For [¶] ]], the accompanying customs entry summary form described the merchandise as [¶] ]]. Id., referencing CDoc 52 at 3, 8-9.
See, e.g. Timken U.S. Corp. v. United States, 434 F.3d 1345, 1353 (Fed. Cir. 2006) (“[w]e suspect that it is not uncommon for an importer to make errors in reporting its sales information to Commerce pursuant to an administrative review”); NTN Bearing Corp. v. United States, 74 F.3d 1204, 1208 (Fed. Cir. 1995) (“[w]hile the parties must exercise care in their submission, it is unreasonable to require perfection”).
E.g. [[]].
See, e.g., entry Nos. [¶] ]].
Commerce also rejected Xuzhou’s explanation that it invoiced customers on the basis of a 10-pound count rather than a per pound count. The AFA memo indicates that crawfish tail meat is generally sold in size-class counts of under 80,80-100,100-150, and 150-200 pieces of tail meat per-pound, whereas whole crawfish is generally sold in size-class counts of under 15,16-20, and 21-80 or more per-pound, and that the counts listed on sales invoices for the first disputed entries were consistent with the size of crawfish tail meat and not whole crawfish. See CDoc 52 at 10-11.
It is the interested party to an administrative review who bears the burden of production on its claim. See, e.g., Tianjin Machinery Imp. & Exp. Corp. v. United States, 16 CIT 931, 936, 806 F. Supp. 1008, 1015 (1992); Chinsung Indus. Co. v. United States, 13 CIT 103, 106, 705 F. Supp. 598, 601 (1989). Once produced, the burden of providing substantial evidence to the contrary for the record shifts. See, e.g., Sichuan Changhong Electric Co., Ltd. v. United States, 30 CIT 1481, 1496, 460 F. Supp. 2d 1338, 1352 (2006); Hangzhou Spring Washer Co., Ltd. v. United States, 29 CIT 657, 670, 387 F. Supp. 2d 1236, 1248 (2005).
E.g., the statutory presumption of correctness on a decision of CBP only extends to a contest of such decision via civil action before this Court. See 28 U.S.C. § 2639(a)(1).
Commerce’s rationale for rejecting WII’s pontention that the count-size stated on the invoice cannot determine whether the product whs invoiced on a per-pound or per-10-pound basis is that knowing the per-unit size can lead to identification of a package of crawfish as either whole or tail meat. However, if it is also true, as Commerce itself observed, that “[t]he size of one whole crawfish is approximately 10 j;imes greater than the size of one piece of crawfish tail meat!,]” CDoc 52 at 10-11, then Commerce’s rationale is a non-sequitur on the weight basis that may be attributable to a given count size, e.g., as discerned from a given invoice.
A reasonable exporter would have insisted on reducing the order-change request to writing, in some fashion, in order to protect itself and/or anticipate request(s) for such information concerning a particular U.S. sale. Even if it “does not require perfection,” the best-of-one’s-ability standard “does not condone inattentiveness, carelessness, or inadequate record keeping.” NSK Ltd. v. United States, 481 F.3d 1355, 1361 (Fed. Cir. 2007) (quoting Nippon Steel, 337 F.3d at 1382). If the shipping documents no longer described the merchandise that was actually shipped, then aside from misstating the shipper’s risk of loss, the risk that the merchandise might be declared subject merchandise at the U.S. border in accordance with such documents - and lead to the type of conundrum Xuzhou faced at the administrative proceeding before Commerce - remained with Xuzhou.
The government further stresses that Commerce also observed that Xuzhou had not been “forthcoming” about the bags allegedly used for the changed order until the pictures had been placed in the record. Def.’s Br. at 14. See CDoc 52 at 11; see also CDoc 51 (memorandum re: color versions of photographs) (Apr. 7, 2008); This was not, however, explicitly stated as a reason for applying an adverse inference. Commerce had not made the packaging an issue until such time, and Xuzhou was to be afforded the opportunity to explain pursuant to 19 U.S.C. § 1677m(d). See, e.g., Ta Chen Stainless Steel Pipe, Ltd. v. United States, 23 CIT 804, 820 (1999) (“it is Commerce, not the respondent, which bears the burden of asking questions!,]” and Commerce must ask “clear” questions “to let the respondent know what *1034information it really wants”); NSK Ltd. v.United States, 19 CIT 1319, 1328, 910 F. Supp. 663, 671 (1995) (“[Respondents should not be required to guess the parameters of Commerce’s interpretation of a phrase in the statute”).
Commerce obviously did not give Xuzhou the benefit of the doubt with respect to the claim of “innocent errors” in entry documentation, although the AFA memo broadly infers that Xuzhou’s explanations were “unsatisfactory” and “inconsistent” with certain record evidence and that requested information had been “withheld” and the proceeding therefore “impeded” within the meaning of 19 U.S.C. § 1677e(a)(2), therefore justifying resort to facts available. CDoc 52 at 15. The AFA memo claims justification for an adverse inference whenever a respondent has not acted to the best of its ability to comply with a request for information, and it also notes that whether a respondent cooperated (by acting to the best of its ability) depends, inter alia, upon the accuracy and completeness of submitted information and whether the respondent has “hindered” the calculation of accurate dumping margins. Id. (referencing Certain Welded Carbon Steel Pipes and Tubes Prom Thailand 62 Fed. Reg. 53808, 53819-20 (Oct. 16, 1997) (administrative review final results)). After proposing that Commerce “must necessarily draw some inferences from a pattern of behavior!,]” id. at 15 (quoting Borden, Inc. v. United States, 22 CIT 1153, 1154 (1998)), the AFA memo then acknowledges Nippon Steel, 337 F.3d at 1383, in noting that Commerce is not required to show intentional conduct on the part of the respondent, only that information was not provided “under circumstances in which it is reasonable to conclude that less than full cooperation has been shown.”
The statute specifically states “the” information, not information generally, and the Statement of Administrative Action likewise specifies that “[w]here a party has not cooperated to the best of its ability, Commerce . . . may employ adverse inferences about the missing information to ensure that the party does not obtain a more favorable result by failing to cooperate than if it had cooperated fully.” H.R. Rep. 103-316 at 870 (1994), reprinted in 1994 U.S.C.C.A.N. 4040, 4199 (as italicized in Gerber Food (Yunnan) Co., Ltd. v. United States, 290 CIT 753, 772, 387 F. Supp. 2d 1270, 1287-88 (2005)). Perhaps therefore, certainly therefor, Commerce’s apparent policy is to resort to “partial” rather than “total” adverse facts available in situations where there is useable information of record but the record is incomplete. See, e.g., Yantai Timken Co., Ltd. v. United States, 31 CIT_,_, 521 F. Supp. 2d 1356, 1364-65 (2007), aff’d 300 Fed. Appx. 934 (Fed. Cir. 2008). Cf. Kawasaki Steel Corp. v. United Steel, 24 CIT 684, 696-697, 110 F. Supp. 2d 1029, 1040-1041 (2000); Ferro Union, Inc. v. United States, 23 CIT 713, 720, 74 F. Supp. 2d 1289, 1296 (1999). Although “total” adverse facts available does not appear in either the governing statute or Commerce’s regulations, administrative usage apparently
referís] to Commerce’s application 'of adverse facts available not only to the facts pertaining to specific sales for which information was not provided, but to the facts respecting all of respondents’ sales encompassed by the relevant antidumping duty order.
Shandong Huarong Mach. Co. v. United States, 30 CIT 1269, 1271 n.2, 435 F. Supp. 2d 1261, 1265 n.2 (2006) (referencing Gerber Food, supra, 29 CIT at 769 n.3, 387 F. Supp. 2d at 1285 n.3).
fn 30] rpj^g -g fljg entered value of [¶] , ]] sales, accounting for [¶] ]] times the quantity of reported sales.
See, e.g., Allied-Signal Aerospace Co. v. United States. 996 F.2d 1185, 1191 (Fed. Cir. 1993); Rhone Poulenc, 899 F.2d at 1190.
Cf., e.g., 19 U.S.C. § 1677b(a)(l)(B)(ii)(II) (5% considered adequate for certain comparison); 19 U.S.C. § 1677i(b)(l) (5% considered adequate for triggering certain agency analysis); 19 C.F.R. § 351.202(b)(7) (2% considered adequate to trigger certain disclosure of information); 19 C.F.R. § 351.206(h)(2) (15% or greater considered “massive”); 19 C.F.R. § 351.208(c) (“substantially all” means exclusion of not more than 15%); 19 C.F.R. § 351.304(c)(1) (up to 10% deviation permitted and at least 1% representative sample required); 19 C.F.R. § 351.403(d) (5% considered adequate for certain comparison); 19 C.F.R. § 351.404(b)(2) (5% considered adequate for certain comparison); 19 C.F.R. § 351.409(b)(1) (20% considered adequate for certain calculation); 19 C.F.R. § 351.413 (1% or less considered “insignificant” for certain adjustments; therefore, greater than 1% considered significant); 19 C.F.R. § 351.523(a)(l)(iii) (1% considered minimum threshold to trigger certain analysis); 19 C.F.R. § 351.524(b)(2) (0.5% or less required for certain allocation).
See CDoc 52 at n.30, supra. Cf. F.lli De Cecco, 216 F.3d at 1032, with, e.g., Polyethylene Retail Carrier Bag Committee v. United States, 232 Fed. Appx. 965, 968, (Fed. Cir. 2007) *1039(“ [b] ecause the attributed electricity cost was greater than the actual cost of electricity used in the production of those bags, the attributed cost was clearly ‘adverse’ to” a particular respondent).