Opinion
I. Introduction
Barzilay,
Judge:
The court reviews the Department of Commerce’s (“Commerce” or “Department”)
Remand Determination in Stainless Steel Sheet and Strip in Coils from Germany (“Remand Determination”).
This case originated pursuant to Plaintiffs’ USCIT R. 56.2 Motion for Judgment Upon the Agency Record. In
Krupp Thyssen Nirosta GmbH and Krupp Hoesch Steel Products, Inc. v. United States,
Slip Op. 00-89, 24 CIT 666 (2000), 2000 WL 1118114
(“Krupp I”),
the court examined Plaintiffs’ challenge to certain aspects of the
Final Determination
of the Department of Commerce’s International Trade Administration in the antidumping investigation of stainless steel sheet and strip from Germany.
See Final Determination of Sales at Less Than Fair Value; Stainless Steel Sheet and Strip in Coils From Germany,
64 Fed. Reg. 30710 (June 8,1999),
as amended,
64 Fed. Reg. 40557 (July 27, 1999)
(“Final Determination”).
The court remanded the case back to
Commerce for reconsideration and redetermination pursuant to the court’s instructions in
Krupp I.
The court exercises jurisdiction pursuant to 28 U.S.C. § 1581(c) (1994) which provides for judicial review of a final determination by an administering authority or commission in accordance with the provisions of 19 U.S.C. § 1516a(a)(2)(B)(i) (1994).
II. Background
Familiarity with the facts presented in
Krupp I
is presumed; however, a brief summation is necessary to properly delineate the pending issues. Krupp Thyssen Nirosta GmbH (“KTN”) is a German producer of stainless steel and Krupp Hoesch Steel Products, Inc. (“KHSP”) is KTN’s United States sales affiliate. KTN was a subsidiary of Krupp Thyssen Stainless (“KTS”). KTS, in turn, was a joint venture holding company owned by two German steel manufacturers, Krupp AG Hoesch-Krupp (“Krupp”) and Thyssen Stahl Ag (“Thyssen”). Kxupp owned sixty percent of KTS and Thyssen owned forty percent. Thyssen also owned resellers in Germany (“German Resellers”) and in the United States (“USR”). KTN sold the subject merchandise in Germany directly from its factory and inventory through a KTN-affiliated service center specializing in resale of second quality stainless products and through the German Resellers. KTN sold the subject merchandise in the United States through KHSR USR and two other Thyssen-owned importers. Commerce requested that KTN answer its standard dumping questionnaire, which asked,
inter alia,
that KTN report all sales (“downstream sales”) by affiliated customers to the first unaffiliated customer in both the United States and Germany. KTN supplied information for USR but failed to supply downstream sales data for the German Resellers. Commerce found that KTN failed to act to the best of its ability and applied partial adverse facts available as a surrogate for the missing sales data so that it could compute the dumping margin.
Final Determination
at 30726-27. Additionally, in
the Final Determination,
Commerce rejected the data KTN provided for USR. Commerce determined that KTN failed to cooperate to the best of its ability because it failed to ensure the accuracy of USR’s sales data prior to verification.
See id.
at 30739. Therefore, Commerce rejected USR’s sales data entirely and applied total adverse facts available as a surrogate for USR’s sales data.
This court ordered on remand that Commerce (1) explain why the adverse facts used as surrogate sales data for Plaintiffs’ affiliated German resellers were rationally related to KTN’s sales and were not unduly harsh or punitive, (2) demonstrate why its rejection of USR’s cost and sales data and use of adverse facts available was supported by substantial evidence, (3) point to substantial evidence demonstrating that KTN had the ability to run checks to discover and correct the errors in its cost and sales data prior to verification, (4) explain why the allocation methodology for USR’s sales of unknown origin was not unduly harsh or punitive, (5) explain its decision not to deduct movement and selling expenses from USR’s gross selling price prior to applying adverse facts,
and (6) exclude USR’s sales of non-subject merchandise from the margin calculation as it had agreed to do after publication of the final administrative determination.
III. Standard of Review
The court must evaluate whether the remand findings are supported by substantial evidence on the record or otherwise in accordance with law. See 19 U.S.C. § 1516a(b)(l)(B). “Substantial evidence is more than a mere scintilla;” it is “such relevant evidence as a reasonable mind might accept as adequate to support a conclusion.”
Consolidated Edison Co. of New York v. NLRB,
305 U.S. 197, 229 (1938);
Matsushita Elec. Indus. Co., Ltd. v. United States,
750 F.2d 927, 933 (Fed. Cir. 1984). This court noted, “[i]n applying this standard, the court affirms [the agency’s] factual determinations so long as they are reasonable and supported by the record as a whole, even if there is some evidence that detracts from the agency’s conclusions.”
Olympia Indus., Inc. v. United States,
22 CIT 387, 389, 7 F. Supp. 2d 997, 1000 (1998)
(citing Atlantic Sugar, Ltd. v. United States,
744 F. 2d 1556, 1563 (Fed. Cir. 1984)). The court may not re-weigh the evidence or substitute its own judgment for that of the agency.
See Granges Metallverken AB v. United States,
13 CIT 471, 474, 716 F. Supp. 17, 21 (1989).
TV.
Discussion
A. German Resellers
As explained in
Krupp I,
“Commerce [is] required to compare the U.S. prices of the subject merchandise to the prices (‘normal value’) for the same or similar merchandise in the home market.”
Mannesmannrohren-Werke AG & Mannesmann Pipe & Steel Corp. v. United States,
23 CIT 826, 827, 77 F. Supp. 2d 1302, 1304 (1999) (citations omitted)
(“Mannesmann”).
To facilitate this process, Commerce requests information from the parties, in particular sales price data, to make the required comparison. If a party fails to cooperate with the requests, Commerce may use adverse facts otherwise available on the record as a surrogate for the missing information.
See
19 U.S.C. § 1677e(b) (1994).
In this instance, KNT did not provide price data for the downstream sales to its affiliated German resellers. Commerce, therefore, used “the ‘highest NV [normal value] reported by control number’ for ‘KTN’s and NSC’s sales to its affiliates for which KTN did not report home market
downstream sales.’”
Krupp I
at *4 (citing
Final Determination,
64 Fed. Reg. at 30728). As determined by this Court in
Krupp I,
the use of adverse facts for the missing German reseller sale data was supported by substantial evidence, but Commerce was asked to show that the adverse facts used were rationally related to sales, indicative of customary selling practices, and not unduly harsh or punitive.
Plaintiffs contend that Commerce has not offered any new evidence in the
Remand Determination
to support its decision to use the highest normal value (“NV”) reported by the control group. Commerce argues that its choice to use the respondent’s own sales best reflects the company’s normal sales practices. Additionally, Commerce offers that it applied the adverse facts based on (1) above-cost sales made at arm’s-length prices, (2) in the ordinary course of trade, (3) in the home market, and (4) on a model-specific basis. Commerce argues that the adverse facts it selected best reflect the customary selling practices for each specific product. In evaluating the parties’ competing claims the court is guided by the language of the Federal Circuit’s holding in
F. LII De Cecco di Filippo Fara S. Martino S.p.A v. United States,
216 F.3d 1027 (Fed. Cir. 2000). In
De Cecco,
the Court of Appeals interpreted Commerce’s discretion in using adverse facts against uncooperative parties.
In the case of uncooperative respondents, the discretion granted by the statute appears to be particularly great, allowing Commerce to select among an enumeration of secondary sources as a basis for its adverse factual inferences.
See
19 U.S.C. § 1666e(b) * * *.
Particularly in the case of an uncooperative respondent, Commerce is in the best position, based on its expert knowledge of the market and the individual respondent, to select, adverse facts that will create the proper deterrent to non-cooperation with its investigations and assure a reasonable margin. Commerce’s discretion in these matters, however, is not unbounded.
De Cecco,
216 F.3d at 1032.
The court holds that here Commerce has shown the prices it selected as adverse facts were rationally related to sales and indicative of customary selling practices. Commerce selected KTN’s own sales data as a surrogate for the missing data of the downstream German resellers. This information is rationally related to sales because it is derived from KTN’s actual sales figures. Commerce used the NV figures that KTN provided for its above-cost sales made at arm’s-length. These sales represented common products sold during the period of investigation by companies similarly situated to the German resellers and accurately re-
fleet the prices KTN was able to charge within its home market. Additionally, Commerce applied the adverse facts on a model-specific basis which ensured the selling price was customary for each specific product. Commerce used its discretion appropriately in choosing these elements of KTN’s own sales data to substitute for the missing information from its German resellers.
Commerce used the highest NV per control number (which related to a specific grade and size of coil) in KTN’s or NSC’s databases. Plaintiffs argue that the use of this adverse sales data dramatically impacted the dumping margin. The missing information represented only [ ] of total sales, but increased the overall weighted margin by [ ]. Therefore, Plaintiffs claim that a relatively small proportion of sales drastically increased the total weighted margin.
Although the adverse facts selected must be rationally related to sales, indicative of customary selling practices, and not unduly harsh or punitive, the court acknowledges that the selection of adverse facts against non-cooperative parties must also serve as a deterrent for withholding data in future proceedings.
National Steele Corp. v. United, States,
20 CIT 100, 103, 913 F. Supp. 593, 596 (1996) (citing
Rhone Poulenc, Inc. v. United States,
899 F.2d 1185, 1191 (Fed. Cir. 1990)). In
De Ceceo,
the court stated:
[i]t is clear from Congress’s imposition of the corroboration requirement in 19 U.S.C. § 1677e(c) that it intended for an adverse facts available rate to be a reasonably accurate estimate of the respondent’s actual rate, albeit with some built-in increase intended as a deterrent to non-compliance. * * * Obviously a higher adverse margin creates a stronger deterrent, but Congress tempered deterrent value with the corroboration requirement. It could have only 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.
De Cecco,
216 F.3d at 1032. Commerce explains that it could have used the highest absolute price or third party data as a surrogate for the missing German Reseller information. However, Commerce chose Plaintiffs’ own data instead.
[T]he Department did not use a more adverse facts available approach which would have assigned the
absolute
highest home market sale price of any product to all other varying product types. Neither did the Department rely upon third party data, such as the petition, as a source of facts available. Instead, the Department has applied to unreported sales only prices which the respondent itself has applied to its sales under customary selling practices.
Remand Determination
at 7. Although Commerce cannot prove that the adverse facts it selected are not unduly harsh or punitive simply by stating that they could have chosen to use data more adverse to Plaintiffs, the use of the highest NV figures illustrates Commerce’s attempt to en
sure that the surrogate data was most closely related to the missing information while simultaneously attempting to deter non-compliance. In striking this balance, Commerce applied adverse facts based on Plaintiffs’ own reported sales figures and on a model match test basis. Although Commerce used the highest NV for the control group, the values used were product-specific and had to meet the match'test criteria. Commerce explains the methodology it used and illustrated the impact that the adverse facts selected had on the dumping margin.
Of the total number of reported home market sales ([ ]), a total of [ ] sales passed the Department’s model match test (this means the remaining [ ] were therefore not used in our calculations). These [ ] “matching home market sales” comprise the [ ] matching sales to which facts available were applied and the remaining [ ] matching sales which did not receive any facts available. The Department’s application of adverse facts available to the sub-group off] sales raised the normal value of the collective [ ] matching sales by [ ] percent.
Remand Determination
at 9 n. 11.
Commerce applied adverse facts only to a small subgroup of the sales used to calculate the dumping margin. The net effect of Commerce’s application of adverse facts available was to raise the average sales price by [ ]. However, Commerce’s use of the adverse facts resulted only in an additional [ ] increase in the dumping margin. This slight increase and the close relationship between the adverse facts used and Plaintiffs’ customary sales establishes that Commerce achieved the balance sought by section 19 U.S.C. § 1677e(b). Therefore, the court finds that the adverse facts selected and used by Commerce were not unduly harsh or punitive.
B. USR
On remand, the court asked Commerce to point to substantial evidence to sustain (1) its rejection of the USR’s entire database, (2) its finding that KTN had the ability, prior to verification, to discover the computer errors that caused the mistakes in the further manufacturing data base, (3) its adverse inference concluding that KTN could have checked the further manufacturing cost data, and (4) the use of adverse facts available for USR’s cost and sales data. Additionally, Commerce was asked to explain its allocation methodology for the sales of unidentified origin and its decision not to deduct movement and selling expenses from gross unit prices.
1.
Rejection of USR’s Entire Data Base
During an antidumping investigation, failure of a party to respond completely or comply with a request for information may result in Com
merce’s rejecting the party’s supplied information.
Commerce requested that Plaintiffs provide cost and sales data information that was vital to the antidumping investigation because the information was to be used to compare home market prices to export prices. In determining the export price or constructed export price Commerce must make adjustments in accordance with 19 U.S.C. § 1677a(d).
Plaintiffs argue that Commerce has not complied with the court’s Remand Order. Plaintiffs contend “Commerce merely restated the same reasoning it articulated in the
Final Determination.
”
Pis. ’ Comments on the Results of Redetermination Pursuant to Remand and Br. in Supp. of Mot. for J. Upon the Agency R.
at 9
(“Pis. ’ Br. ”).
Plaintiffs further argue that the errors were minor and that most of the cost data was verified or verifiable. Furthermore, Plaintiffs believe that Commerce has failed to substantiate that the minor errors in the cost data contaminated USR’s sales data.
Commerce contends that the cost and sales information that Plaintiffs supplied was not capable of being verified. Commerce argues that a logic flaw in the computer program Plaintiffs used to produce its cost and sales data caused the errors found during verification. Commerce asserts that Plaintiffs’ reliance on its “integrated” computer system produced erroneous cost data that was then merged directly into the sales data. Additionally, because Plaintiffs did not “provide the linking variable that would tie each further manufactured product in the U.S. to the product(s) as imported. * * * Commerce, was obligated to rely upon the integrated cost and sales database” and could not “correct” the errors in the computer program.
Def.’s Mem. in Opp’n to Pis.’ Mot. for Summ. J. Upon the Agency R.
at 9-10.
(“Def.’s Br. ”).
Therefore, because the errors affected the sales and cost data, Commerce could neither properly match home market sales to U.S. sales nor accurately construct the export price.
a. Cost Data
The only costs reported by USR were the further manufacturing costs.
See Section E Resp., Conf. Doc. 36
(November 16,1998)
(“Section E Resp. ”).
The amount delineated as total further manufacturing cost (“FURMANU”) is the compilation of many cost fields and sub-cost fields. For instance, it includes further manufacturing costs (“FUR-COM”), further manufacturing general costs (“FURGNA”), and further manufacturing interest (“FURINT”). FURCOM, in turn, is comprised of three sub-costs, two of which, further manufacturing labor (“FUR-LAB”) and further manufacturing overhead (“FURFOH”), could not be verified as — Plaintiffs concede. Thus, FURCOM remains unverified and may be inaccurate. The inaccuracy in the FURCOM field created systemic errors throughout the cost data base because FURCOM is used to derive both FURGNA and FURINT.
As illustrated, the errors in the FURCOM cost data field and its sub-cost data created errors that affected the entire cost data base, even though the actual errors appeared in only a few fields, because several of the conglómerate-cost fields were contingent on erroneous information. Therefore, Commerce’s determination to reject the cost database was supported by substantial evidence.
b. Sales Data
Section 1677a(d)(2) requires that Commerce deduct the cost of any further manufacture from the USR’s sales price to its first unaffiliated customer in the United States to calculate the constructed export price (“CEP”). The CEP is critical in the dumping margin calculation because CEP must be compared to NV in order to calculate an accurate dumping margin. As described above, computer errors corrupted the data reflecting total further manufacturing costs. Clearly, if the total further manufacturing costs are erroneous, then the resulting CEP will also be erroneous. Plaintiffs argue that “the further manufacturing cost data is used as one of several adjustments to CER and its impact on the dumping calculation is relatively minor.”
Pis. ’ Br.
at 13. Additionally, Plaintiffs argue that Commerce’s reliance on a prior case involving steel wire rod from Germany is misplaced and “highlights the extreme position that Commerce has taken in this case.”
Id.
(citing
Notice of Final Determination of Sales at Less Than Fair Market Value: Steel Wire Rod from Germany,
63 Fed. Reg. 8953 (Feb. 23, 1998)
(“Steel Wire Rod from Germany”)).
The court is not persuaded by Plaintiffs’ argument. In
Steel Wire Rod from Germany,
Commerce rejected the respondent’s sales data base because errors in the cost of production database affected the NV calculation. Although in this instance the errors are in the further manufacturing data and affect the CEP calculation, the rationale for rejecting the data remains consistent. Plaintiffs’ attempt to illustrate that the errors are isolated inaccuracies minimizes the widespread effect of the computer logic errors. It is not the number of errors that is the deter
mining factor, but the impact the errors have on the data. In this instance, the computer programming errors were so systemic and pervasive that the inaccuracies affected the cost and sales data and prevented Commerce from calculating an accurate CEE As Commerce states in the
Remand Determination,
[w]hile the further manufacturing cost may only be a single field in the U.S. Reseller’s
sales
data file (TOTFMG), its impact on the calculated CEP and, therefore, the dumping margin, is significant. This is true because the TOTFMG field is a direct reduction from the U.S. Reseller’s sales price to the first unaffiliated customer in the United States in order to arrive at the CEE As a result, an error in this field will corrupt the resulting CEP that is compared to NV to calculate the dumping margin.
Remand Determination
at 20. Plaintiffs’ decision not to provide a linking variable compounded the problems caused by the faulty computer program. Plaintiffs stated:
the Company has provided the further processing costs in an integrated computer file that also incorporates the specific transactions in which the processing costs were incurred. Since the cost information is already merged with the sales data, there is no need to provide a linking variable. Therefore, this field has been omitted.
Section E Resp.
at E-18. In essence, the linking variable requested by Commerce would have allowed for proper matching of KTN’s home sales to identical or similar merchandise sold in the United States.
Since KTN provided integrated sales figures without accompanying linking variables, further processed products were not properly compared to appropriate U.S. sales. Additionally, the errors in the further manufacturing field dramatically impacted the sales data because [ ] percent of KTN’s total USR’s sales underwent further manufacturing. Inaccurate further manufacturing figures would distort the costs and sales data on the
majority
of USR’s sales. The programming errors in the cost data also caused product classification problems.
The extent of the errors found in the cost database, particularly classifying a product as further processed when, in fact, it received no further processing, or assigning no further processing costs to a further manufactured product, undermines the credibility of the data in their entirety both sales and cost. Consequently, neither the further manufacturing data nor the non-further manufactured data can be deemed reliable for calculating the margin, since both groups of sales are tainted by the programming errors.
Remand Determination
at 21 (footnotes omitted). Furthermore, Commerce supports its decision to reject the sales database because Plaintiffs failed to: (1) substantiate a sale trace at verification, (2) properly classify non-prime merchandise, (3) identify supplier sales and supply
model match characteristics, and (4) account for early payment discounts.
The computer program failures combined with the additional errors outlined above provide substantial evidence to support Commerce’s rejection of the entire database. As stated in
Consolo v. Federal Maritime Commission,
383 U.S. 607 (1966):
[w]e have defined “substantial evidence” as “such relevant evidence as a reasonable mind might accept as adequate to support a conclusion.” * * * This is something less than the weight of the evidence, and 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,
383 U.S. at 619-20 (citations omitted). Commerce has proved that the cost and sales databases contained significant errors and inaccuracies. The source and cause of the errors was the integrated computer system Plaintiffs relied on to produce their cost and sales information. Plaintiffs could have chosen to comply with Commerce’s request and supplied the requisite linking variables which may have alleviated the effects of the errors. Instead, Plaintiffs relied on the integrated system that it certified could create an integrated sales response for both further manufactured and non-further manufactured sales. In
Krupp I,
this court stated:
Commerce may find on remand that it is appropriate to apply partial facts available to fill any gaps in the sales data it could not successfully verify, but it may not disregard the sales data absent evidence in the record that the sales data was fatally tainted by the errors in the computer program.
Krupp
/at *6 (citations omitted). The court affirms Commerce’s conclusion that there were systemic errors that fatally tainted the entire database. The errors in the cost data were magnified when they were integrated with the sales data and, thus, tainted both the cost and sales data. Furthermore, because no linking variables were provided, the errors could not be isolated, adjusted, or corrected in the final sales data. This chain of events, however unfortunate, was caused by KTN’s reliance on the computer program and its failure to supply Commerce with the linking variables.
2.
Use of adverse facts available for the USR’s cost and sales data
In
Krupp I,
the court ordered that Commerce point to substantial evidence demonstrating that KTN had the ability to run checks to discover and correct the errors prior to verification and that Commerce point to additional evidence in the record, other than errors in the computer program, before drawing an adverse inference. In the event Commence could not point to substantial evidence to demonstrate that KTN could have discovered and corrected the errors in the computer program prior to verification, Commerce was to refrain from drawing an adverse inference in its use of facts otherwise available.
These remand instructions were necessary because “‘[u]nder the URAA, Commerce is now required to make more subtle judgments than under the previous best information available (‘BIA’) standard.’”
Krupp I
at *7 (quoting
Ferro Union,, Inc. v. United States,
23 CIT 178, 198, 44 F. Supp. 2d 1310, 1329 (1999)). Since the enactment of the URAA, Commerce must first find “an interested party has failed to cooperate by not acting to the best of its ability to comply with a request for information” before making an adverse inference. 19 U.S.C. § 1677e(b);
See also Ferro Union,
44 F. Supp. 2d at 1330-31. As explained in
Mannesmann:
[s]ection 1677e(a) provides, however, that the use of facts available shall be subject to the limitations set forth in 19 U.S.C. § 1677m(d) (1994). Section 1677m, which were enacted as part of the Uruguay Round Agreement Act (“URAA”), Pub. L. 103-465 § 231, is “designed to prevent the unrestrained use of facts available as to a firm which makes its best effort to cooperate with [Commerce].”
Mannesmann,
77 F. Supp. 2d at 1312 (citations omitted).
Commerce responds that KTN could have run checks to discover the errors prior to verification because KTN recognized some errors in the cost data prior to verification. In fact, KTN submitted a revised
Section E
response after running checks, correcting errors and resubmitting the response to Commerce. Additionally, Commerce asserts that in preparation for verification KTN was able to isolate certain errors in the cost allocation program and was able to fix the errors prior to verification.
See Remand Determination
at 31.
Plaintiffs argue that the errors in the computer program could not have been discovered prior to verification. Plaintiffs assert that Commerce “has failed to identify” evidence that proves that the errors could have been detected through rudimentary checks and simply “repeat[s] (albeit in different words) the
same
inadequate grounds cited in the
Final Determination.
”
Pis. ’ Br.
at 15. Plaintiffs contend that the programming error was undetectable prior to verification and it was only after a manual trace,
during
verification, that the error was discovered. Additionally, Plaintiffs argue that the programming error was inadvertent. KTN claims that it cooperated to the best of its ability because the administrative record and Commerce’s own comments reflect that KTN was conducting checks on the submitted data, submitted several revised versions of its further manufacturing data, and manually reviewed invoices. Furthermore, Plaintiffs contend that the factual circumstances in this case are similar to
Nippon Steel Corp. v. United States,
24 CIT 1158, 118 F. Supp. 2d 1366 (2000), where the court held that inadvertence is not sufficient grounds for the application of an adverse inference.
See Pis. ’ Br.
at 19.
The court is not convinced by the record in this case that there is substantial evidence that Plaintiffs’ conduct here demonstrates that they did not act to the best of their ability with regard to discovering programming errors in the cost data prior to verification. Therefore, Com
merce may not, under the new statutory scheme, draw an adverse-inference in its use of facts otherwise available for USR’s cost and sales data.
The court agrees with Plaintiffs that the specific mistakes Commerce identified in the computer program were inadvertent and were-not a result of Plaintiffs’ failure to act to the best of their abilities considering the totality of Plaintiffs’ actions and all the information submitted. Where Plaintiffs were able to correct identified errors they did so. For instance, Commerce itself agrees that Plaintiffs attempted to check the accuracy of the data prior to verification. “In checking the data and in preparing for verification the U.S. Reseller found several items that were incorrectly calculated and presented these corrections at the. beginning of verification.”
Remand Determination
at 30 (citations omitted). Additionally, Commerce acknowledged that Plaintiffs:
submitted no fewer than three section E further manufacturing cost databases during the course of the proceeding. Each cost database was submitted with the required certification as to the accuracy of the information. Because the U.S. Reseller submitted various versions of the cost information, it is evident that the U.S. Reseller was conducting checks on the submitted data and finding areas that needed to be corrected.
Remand Determination
at 28-29. Commerce’s findings support the position that the computer logic flaws that caused the erroneous data were inadvertent and could not have been discovered through rudimentary checks. Commerce agrees that Plaintiffs checked their information prior to verification and submitted amended responses to correct the errors. Howevér, the computer logic flaws were identified at verification and only after a manual trace of the underlying documents was performed were the errors discovered.
See Pis’ Br.
at 17. Therefore, the record shows that the computer errors corrupting USR’s sales data were inadvertently included in the information Plaintiffs provided to Com
merce.
In
Nippon Steel Corp. v. United States,
25 CIT_,_, 2001 WL 406192, *4 (2001), the court held that Commerce’s conclusion that respondent had failed “to act to the best of its ability” pursuant to 19 U.S.C. § 1677e(b) (1994), was unsupported by substantial evidence. The court analyzed the instances where respondents failed to comply with Commerce’s information request and differentiated the factual circumstances that allowed Commerce to draw an adverse inference, and therefore, use adverse facts as surrogate information.
In cases where a respondent claims an
inability
to comply with the agency’s requests for information, the Department may permissibly draw an adverse inference upon a reasonable showing that respondent, in fact, could have complied, (citations omitted) Here, however, where the respondent acknowledges the ability to comply, but claims it did not do so because of simple inadvertence, the Department must show more. As the court observed in its opinion or
dering remand, those cases that do not suggest willfulness on the part of the respondent pose particular challenges for the Department to draw appropriate lines in order to determine whether to draw an adverse inference.
Nippon Steel Corp,
2001 WL 406192 at *4 (citation omitted). Here, Commerce has failed to point to substantial evidence in the record that supports a finding that Plaintiffs willfully failed to comply with Commerce’s information request. In fact, Plaintiffs specifically instituted the use of the computer program in an attempt to facilitate compliance with Commerce’s information demands.
See Krupp I
at *4. The computer program failure was a byproduct of inadvertent computer logic errors. In both its
Final Determination
and
Remand Determination,
Commerce has failed to support by substantial evidence that these inadvertent computer logic errors show that Plaintiffs failed to act to the best of their ability. “While the parties must exercise care in their submissions, it is unreasonable to require perfection.”
NTN Bearing Corp. v. United States,
74 F.3d 1204, 1208 (Fed. Cir. 1995). The statute at issue here has consistently been interpreted by this court to require more than inadvertent error on the part of a respondent to permit the agency to use adverse facts in calculating dumping margins.
See e.g., Nippon Steel Corp.,
118 F. Supp. 2d 1366;
Ferro Union,
44 F. Supp. 2d 1310;
Mannesmann,
77 F. Supp. 2d 1302. Therefore, the court remands this case, yet again, and pursuant to 19 U.S.C. § 1677e(a)(2), instructs Commerce to select neutral facts available for the purpose of calculating USR’s margin rate and any other calculation predicated on USR’s cost and sales data.
3.
Sales of Unidentified Origin
In
Krupp I,
the court sustained Commerce’s use of an adverse inference with respect to the allocation of sales of unidentified merchandise.
See Krupp I
at *9. The court then ordered on remand that Commerce explain how the chosen adverse facts were not unduly harsh or punitive.
Id.
As cited in
Krupp I,
“the purpose of section 1677e(b) is to provide respondents with-an incentive to cooperate, not impose punitive, aberrational, or uncorroborated margins.”
De Cecco,
216 F.3d at 1032 (citations omitted);
See
19 U.S.C. § 1677e(b).
Plaintiffs argue that Commerce’s allocation of sales of unidentified origin at more than double the amount of KTN’s verified sales to USR is unduly harsh and punitive. Plaintiffs assert that Commerce’s allocation increased the overall dumping margin from [ ] to [ ].
Pis. Br.
at 26-27. Commerce argues that it was inappropriate to base the allocation on KTN’s sales to USR and, instead, Commerce divided the unidentified sales on a “pro rated basis, using the verified percentages of the U.S. Reseller’s merchandise supplied by each respondent mill.”
Remand Determination
at 33. Commerce asserts that the overall effect of this
allocation was to increase the dumping margin by [ ]. To apportion the-amount of the dumping margin attributable to the unidentified sales, Commerce determined that USR should be allocated [ ] million. Commerce then multiplied the sales by the adverse facts available margin rate of [ ]. KTN argues that the margin rate should be multiplied by [ ] million because that is the amount of KTN’s verified sales to USR. The court is not persuaded by KTN’s reasoning.
When KTN claims the dumping margin would be [ ] lower, it assumes that Commerce should use KTN’s verified sales figure of [ ] million. As Commerce states:
[t]he Department cannot reasonably assume that the source of the merchandise of
unidentified
origin is the same as for the group of merchandise of identified origin. Such an inference would have been extremely suspect considering that the U.S. Reseller had the ability to provide relevant information on the origin of many of these sales, but chose not to do so. Had the Department apportioned sales of unidentified origin to KTN in accordance with the percentage of verified KTN sales to the U.S. Reseller ([ ] percent), the Department would not have been applying adverse facts available to such sales but, rather, would have been applying non-adverse, or neutral, facts available to these sales.
Remand Determination
at 34-35. Additionally, the [ ] million KTN represents as its verified sales to USR does not include all resales during the period of investigation.
See
KTN Exh. 6; Def. Conf. Exh. 12.
KTN asserts that “Commerce attributed to KTN USR sales of unidentified origin valued as * * * more than double the value of the KTN material sold to USR.”
Pis. ’ Br.
at 25. However, KTN does not account for sales of merchandise by KTN to USR
prior
to the period of investigation that were then resold
during
the period of investigation.
Commerce must strike a balance when applying adverse facts available. In this instance, KTN reported a large number of U.S. Reseller sales transactions for which the manufacturer was not identified. During verification it became apparent that through rudimentary searches KTN could have identified the suppliers. Since KTN failed to provide the supplier information, it was proper for Commerce to draw an adverse inference with respect to the unidentified supplier sales.
See Krupp I
at *8. As evidenced in the
Remand Determination,
the exact number of unidentified sales that originated from KTN is unknown. However, it is clear from the record that USR could have provided at least some of the supplier information but chose not to look behind the data captured in its inventory system. Commerce has adequately ex
plained its methodology for determining the allocation without using the most adverse inferences possible. As Commerce explained:
[t]he respondent has not afforded the Department any reasonable benchmark for ascertaining what allocation methodology would be more appropriate. * * * It is entirely possible that KTN benefitted from withholding information on the U.S. Reseller’s suppliers that would result in higher margins than those actually calculated
absent
respondent data. * * * It is certainly possible that in allocating all of the sales of unidentified origin to the three respondents in this investigation in accordance with each of their verified percentages of sales to the U.S. Reseller, we may have underestimated the actual volume of unidentified origin merchandise that originated from KTN. * * * Despite this possibility, the Department in applying an adverse inference did not apply all of the sales of unidentified origin to KTN.
Remand Determination
at 35-36. The court finds the application of specific adverse facts chosen and the resulting increase in the dumping margin served as both a “reasonably accurate estimate” of KTN’s actual rate and “as a deterrent to non-compliance.”
De Cecco,
216 F.3d at 1032. Therefore, the court upholds Commerce’s allocation method.
4.
Movement and Selling Expenses
On remand, the court ordered that Commerce explain how its decision not to deduct verified moving and selling expenses from USR’s average gross price resales was the proper methodology.
See Krupp I
at *9. In calculating Plaintiffs’ overall dumping margin, Commerce made various margin calculations that were weighted and combined to formulate the overall dumping margin. As Commerce explained:
[b]oth adverse facts available margin percentage rate applied to the U.S. Resellers sales and the calculated weighted-average unit value for the U.S. Reseller sales affect the magnitude of the overall weighted-average margin for all U.S. sales. * * * These estimated dollar margins for the U.S Reseller sales, in turn, are added to the margins calculated for KTN’s properly reported U.S. sales to enable the calculation of a single weighted-average margin percentage rate for all of KTN’s U.S. sales.
Remand Determination
at 37. For USR’s sale prices, Commerce applied an adverse facts available margin to the weighted-average of USR’s sale prices without deducting movement and selling expenses.
See id.
However, the surrogate margin, by Commerce’s own admission, was a “net” margin rate because it “already reflected deductions for movement and selling expenses.”
Id.
Therefore, Commerce was applying a “net” surrogate margin to “gross” sales figures.
Plaintiffs argue that applying the net surrogate margin to USR’s gross sale prices “arbitrarily inflated KTN’s dumping margin because gross prices are higher than net prices.”
Pls.’Br.
at 28. Plaintiffs assert that either (1) USR’s gross sales prices be adjusted to a net sales basis or (2) the surrogate margin percentage be adjusted to a gross margin per
centage. Plaintiffs contend that movement and selling expenses were verified and should be deducted from USR’s gross sales prices to obtain the net weighted-average price of USR’s sales to properly adjust USR’s sales to a net sales basis. In the alternative, Plaintiffs argue that “[t]o properly re-express the surrogate margin on a gross sales basis, all that would he required is to
ignore
these verified adjustments and instead calculate a sales denominator that is based on ‘verified’ gross unit prices for ‘KTN’s properly reported U.S. sales.’”
Pis. ’ Br.
at 33.
Commerce contends that it had to apply the net margin rate to the gross sales figures because the systemic errors in USR’s responses made the cost and sales data unuseable for the purpose of calculating Plaintiffs’ overall weighted-average margin.’
Remand Determination
at 38-39. Commerce explained “[i]n keeping with that practice, when the Department determined that the U.S. Reseller’s sales data were entirely useless for the purpose of calculating KTN’s dumping margin, the response was, so to speak, set aside, and the Department turned elsewhere for surrogate information to serve as the facts otherwise available, in accordance with 776 of the Tariff Act; 19 U.S.C. § 1677e.”
Id.
at 38. In setting aside the entire response, Commerce disregarded movement and selling expenses that had been verified. Commerce asserts it has a long standing practice that prohibits it from “cherry-picking” the verified information from a response that has been completely rejected because “mak[ing] any adjustments to an unverified gross unit price based upon unverified data would render the calculation meaningless.”
Id.
at 38-39.
This statement is contrary to the evidence on the record. The record indicates that Commerce verified forty-seven (47) out of the fifty-two (52) movement and selling expense reported by USR. Additionally, Commerce used facts available for USR sales data and was able to use a verified net margin rate to make the required margin calculation.
Id.
This Circuit has held that the primary purpose of the antidumping laws is to calculate accurate margins.
See Rhone Poulenc,
899 F.2d at 1191 (stating that the basic purpose of the statute is to determine dumping margins as accurately as possible);
Lasko Metal Prods., Inc. v. United States,
43 F.3d 1442, 1446 (Fed. Cir. 1994). “While Congress has left it within Commerce’s discretion to develop methodologies to enforce the antidumping statute, any given methodology must always seek to effectuate the statutory purpose — calculate accurate dumping margins.”
Shakeproof Assembly Components Div. of Ill. Tool Works, Inc. v. United States,
23 CIT 479, 482, 59 F. Supp. 2d 1354, 1358 (1999). In this instance, Commerce should use a methodology that would allow the net margin rate to be applied to USR’s net sales price data. Therefore, on
remand, Commerce is to choose a method that enables the margin rate and sales data to be adjusted to a “net” basis.' Furthermore, if additional information is required, Commerce is to use neutral facts available as surrogate information.
V Conclusion
For the foregoing reasons, the court holds that the Department’s
Results of Redetermination Pursuant to Court Remand; Stainless Steel Sheet and Strip in Coils from Germany
is remanded in part, and sustained in part. Judgment will be entered accordingly.