In these consolidated antidumping duty cases, a Chinese exporter and five U.S. importers of iron eastings from the People’s Republic of China (PRC) challenge several related orders of the Court of International Trade. Through those orders, the court upheld the Commerce Department’s final determinations in a series of administrative reviews of a 1986 antidumping duty order directed at iron construction castings from the PRC. We affirm-in-part, reverse-in-part, and remand.
I
In May 1986, the Department of Commerce issued an antidumping duty order directed to iron construction castings imported into the United States from the PRC. As part of the order, Commerce directed that antidumping duties be collected in an amount “equal to the amount by which the foreign market value of the merchandise exceeds the United States price for all entries of castings from the PRC.” 51 Fed.Reg. 17,222, 17,223 (1986). Commerce subsequently conducted a series of administrative reviews of the anti-dumping duty order for the period between May 1, 1987, and April 30, 1991. The first two annual reviews were consolidated into a single review covering May 1, 1987, through April 30,1989.
Each of the reviews resulted in a final determination that was appealed to the Court of International Trade. That court directed Commerce to conduct remand proceedings with respect to several issues, and following the remand proceedings the court affirmed the final remand determinations in each case. Appeals were then taken to this court. With respect to the 1990-91 review, we have separately disposed of the appeal taken by the U.S. importers. See D & L Supply Co. v. United States, 113 F.3d 1220 (Fed.Cir.1997). In this case, we consider appeals from Commerce’s final remand determinations for the 1987-89 and 1989-90 reviews. Rather than describe in detail the complex procedural background of the various final determinations and the proceedings with respect to each final determination in the Court of International Trade and in the Department of Commerce on remand, we turn directly to the issues raised on appeal. The procedural background will be described, as necessary, in connection with the discussion of each appealed issue.
II
For the 1987-89 review, Commerce concluded that the PRC was a “nonmarket economy country,” see 19 U.S.C. § 1677(18) (1988), and it derived the dumping margin by using the “constructed value” method, one of the methods provided for calculating dumping rates for nonmarket economy countries. See 19 U.S.C. § 1677b(c), (e) (1988). Under the version of the antidumping statute that was applicable at the time, “constructed value” was calculated “on the basis of the value of the factors of production utilized in producing the merchandise,” to which was added “an amount for general expenses and profit plus the cost of containers, coverings, and other expenses.” 19 U.S.C. § 1677b(c)(l) (1988). The statute directed the factors of production to be valued based on “the best available information regarding the values of such factors in a market economy country or countries” that Commerce considered appropriate. Id. The constructed value, or foreign market value, was then compared with- the *1405United States price. See 19 U.S.C. § 1677a (1988). If the United States price was lower than the foreign market value, an antidump-ing duty was imposed in the amount by which the foreign market value exceeded the United States price. See 19 U.S.C. § 1673 (1988).
In its preliminary determination for the 1987-89 period, Commerce assigned the Chinese manufacturers separate dumping margins, but in the final determination Commerce assigned the manufacturers a single country-wide rate. Commerce noted that because the PRC is a nonmarket economy, all commercial entities in the country are presumed to export under the control of the state, and that no manufacturer would receive a separate antidumping duty rate unless it could demonstrate that it enjoyed both de jure and de facto independence from the central government. Finding that the evidence was unclear on the issue, Commerce determined that the presumption of state control had not been rebutted and therefore adopted a single country-wide margin.
D & L and other U.S. importers of iron construction eastings challenged the final determination for 1987-89 in the Court of International Trade, asserting that Guangdong was independent of the national Chinese corporation, China National Metals and Minerals Import and Export Corporation (“China National”), and that Commerce had therefore erred in applying a country-wide antidump-ing margin to Guangdong. The Court of International Trade found procedural error in Commerce’s change of position on that issue between the preliminary and final determinations, and it remanded for Commerce to provide Guangdong an opportunity to demonstrate its de jure and defacto independence during the two years covered by the 1987-89 review.
On remand, Guangdong submitted a document that included copies of Chinese laws and a certification from China National stating that China National did not exercise control over Guangdong during the pertinent time periods. Commerce, however, determined that the evidence failed to justify giving Guangdong a separate dumping margin. The evidence, Commerce concluded, failed to show de jure independence for 1987-88. With respect to 1988-89, Commerce found that the evidence established Guangdong’s de jure independence but failed to show defacto absence of state control over Guangdong’s activities.
The Court of International Trade upheld Commerce’s rulings with respect to both periods. On appeal to this court, D & L asserts that Commerce’s findings with respect to the de jure and de facto control issues are contrary to the evidence offered to prove that Guangdong was not controlled by China National.
Consistent with Commerce’s approach to this issue, the Court of International Trade has ruled that an exporter in a nonmarket economy country must “affirmatively demonstrate” its entitlement to a separate, company-specific margin by showing “an absence of central government control, both in law and in fact, with respect to exports.” Tianjin Mach. Import & Export Corp. v. United States, 806 F.Supp. 1008, 1013-14 (Ct. Int’l Trade 1992). Absence of de jure government control can be demonstrated by reference to legislation and other governmental measures that decentralize control. Id. at 1014. Absence of de facto government control can be established by evidence that each exporter sets its prices independently of the government and of other exporters, and that each exporter keeps the proceeds of its sales. Id.
D & L questions the applicability of the presumption of state control as applied to the PRC, and particularly to a company operating in Guangdong province in the late 1980s. Commerce, however, has broad authority to interpret the antidumping statute and devise procedures to carry out the statutory mandate. See Torrington Co. v. United States, 68 F.3d 1347, 1351 (Fed.Cir.1995). We agree with the government that it was within Commerce’s authority to employ a presumption of state control for exporters in a nonmarket economy, and to place the burden on the exporters to demonstrate an absence of central government control. The antidumping statute recognizes a close correlation between a nonmarket economy and *1406government control of prices, output decisions, and the allocation of resources. See 19 U.S.C. § 1677(18)(B)(iv), (v). Moreover, because exporters have the best access to information pertinent to the “state control” issue, Commerce is justified in placing on them the burden of showing a lack of state control. See Zenith Electronics Corp. v. United States, 988 F.2d 1573, 1583 (Fed.Cir.1993) (“The burden of production should belong to the party in possession of the necessary information.”).
The only law to which D & L refers to establish de jure independence for the 1987-88 period is the 1986 General Code of the Civil Law of the PRC. Commerce concluded that the 1986 Code, which simply provided that “[enterprises ... shall be qualified as a legal person upon approval of registration by the competent authority,” did not confer independence on the branches of state-owned enterprises. The language of the 1986 Code contrasts sharply with the language of the 1988 Law of the PRC on Industrial Enterprises Owned by the Whole People, which was the basis of Commerce’s finding of de jure independence for Guangdong during the 1988-89 period. The 1988 law provides that “[enterprises may, under the guidance of state planning, arrange on [their] own the production or provision of products or services required by the society,” language that goes well beyond the right to qualification “as a legal person” granted by the 1986 Code.
With respect to the issue of de jure independence, Commerce concluded that “absent some objective indication from the government that branch corporations such [as] Gu-angdong Minmetals were legally independent prior to or concurrent with the time of the 1987-1988 review period, the record does not support a finding that Guangdong Minmetals was legally independent from its parent corporation during the 1987-1988 review period.” D & L has not persuaded us that Commerce’s analysis of the pertinent legislative materials is wrong; we therefore sustain Commerce’s determination that Guangdong failed to satisfy its burden of proving de jure independence from the central government during the 1987-88 review period.
With respect to the 1988-89 period of review, Commerce found that-Guangdong established de jure independence from China National, but that it failed to satisfy its burden of showing defacto independence. Commerce noted that Guangdong “provided almost no evidence on the record establishing that it negotiated prices or sales with its customers, or that it was actually responsible for its profits or losses.” After listing the types of evidence that might show such de facto independence, Commerce noted that Guangdong provided only a certification from its parent corporation attesting that Guangdong acted independently during the 1987-89 period of review. Commerce concluded that such a certification, “standing alone without any supporting documentation does not establish the absence of defacto control.”
Commerce noted that Guangdong had submitted copies of its 1986 and 1988 business licenses in an effort to demonstrate its independence. The 1988 license supports the claim that Guangdong was independent from China National, as it states that Guangdong was in the business of “import and export (at its own account)” and was “[i]ndependently responsible” for its accounting. As Commerce concluded, however, the 1988 business license does not “establish clear independence from the parent corporation,” as it “describes the scope of Guangdong’s business activities without identifying the government’s role in those activities,” and fails to prove the absence of de facto government control with respect to export sales and pricing decisions.
It was proper for Commerce to require D & L to do more to establish Guangdong’s independence of the central government. Moreover, we are unpersuaded by D & L’s arguments that Commerce should have found de jure and de facto independence based on (1) the fact that the domestic producers did not object to Commerce’s preliminary determination to give Guangdong a separate rate; (2) the fact that the Court of International Trade ordered Commerce to assess duties against another Chinese exporter at the deposit rate of 11.66 percent; and (3) the assertion that at the time Guangdong province, where Guangdong Minmetals was located, *1407was “far ahead of the rest of China in moving toward market economics.” In view of the presumption that an exporter in a nonmarket economy country is part of the national corporation and Commerce’s reasonable conclusion that D & L’s evidentiary showing was insufficient to overcome that presumption, we affirm the order of the Court of International Trade upholding Commerce’s conclusion on the independence issue.
Ill
The appellants next raise a series of challenges to Commerce’s calculation of various components of the constructed foreign market value for Guangdong’s iron castings. They contend that Commerce committed error in calculating the freight costs that were used in determining the foreign market value of Guangdong’s products; that Commerce improperly added a sum for warehousing expenses to Guangdong’s foreign market value; and that Commerce’s calculation of Gu-angdong’s overhead was not supported by substantial evidence in the record.
A
In deriving foreign market value for the 1987-89 review, Commerce used the Philippines as a surrogate market economy country for purposes of valuing the factors of production for the Chinese eastings exporters. In the final results for that period, Commerce agreed with the Chinese exporters and the U.S. importers that the import price of pig iron in the Philippines included foreign inland and ocean freight (ie., the freight necessary to transport the pig iron from the pig iron mill in a third country to the port in the Philippines), and that it was therefore not appropriate to increase the constructed foreign market value “to account for the freight expense of transporting materials from the point of purchase to the foundry.” The domestic producers appealed that ruling to the Court of International Trade, arguing that, in calculating the freight component of foreign market value, Commerce should have used the distances from the pig iron mills to the foundries reported by the Chinese exporters. The court agreed and remanded, directing Commerce to recalculate the freight costs using the information submitted by the Chinese exporters.
On remand, Commerce completely altered its method of calculating freight. As before, it started with the import price of pig iron in the Philippines, ie., the price of pig iron delivered to port in the Philippines, with foreign inland and ocean freight expenses already included. Commerce then ascertained the distance from the pig iron mill in China to the foundry and added a constructed freight cost for that distance to the Philippine import price.
In deriving the foreign market value of Guangdong’s products for the 1989-90 review, Commerce followed a protocol similar to the one it followed on remand in the 1987-89 review, but instead of the Philippines, it used India as the surrogate market economy country. Commerce started with the import price of pig iron in India, ie., the price of pig iron delivered to port in India, with foreign inland and ocean freight expenses included. It then added a constructed price to account for the transportation from the Chinese pig iron mill to Guangdong’s foundries. The Court of International Trade affirmed Commerce’s disposition of the freight issue with respect to both the 1987-89 review and the 1989-90 review.
The appellants argue that in both instances Commerce’s method of calculating the constructed freight costs resulted in double counting of freight expenses and therefore unreasonably increased the foreign market value of Guangdong’s products. We agree that Commerce’s methodology overstated the value of the freight component of foreign market value for both the 1987-89 and 1989-90 reviews, and that the resulting antidump-ing duty rates for those review periods cannot stand.
Simply put, the import prices in the Philippines and India already included ocean freight and foreign inland freight, a substantial portion of the total cost of transporting imported pig iron from the pig iron mill to the foundry. By adding a constructive freight charge for the entire trip from the mill to the foundry in China on top of the import prices in the Philippines or India, *1408Commerce’s methodology double-counted a substantial component of the total freight expense. While the process of constructing foreign market value for a producer in a nonmarket economy country is difficult and necessarily imprecise, the method that Commerce chose in this case falls outside the limits of permissible approximation.
The appellees’ answer is that the import prices in the Philippines and in India are presumed to be the same as the market price for domestically produced pig iron in each country, which does not include any element of freight. What Commerce has done, the appellees explain, is to add to the presumed domestic prices in the Philippines and India a constructed value for transporting pig iron from the Chinese pig iron mill to the Chinese foundry.
The problem with that rationale is that it assumes a castings producer in the surrogate country would choose to pay the highest combination of prices for pig iron plus freight, something that no rational producer in a market economy would do. For example, if a producer in a surrogate country had a foundry next to a port (and thus had negligible freight expenses from the port to the foundry), it would purchase its pig iron at the import price, rather than purchasing equivalently priced domestic pig iron that had to be shipped at significant expense from a domestic pig iron mill. By the same token, if the surrogate country producer had a foundry next to the domestic pig iron mill and far from the nearest port, it would purchase its pig iron from the domestic mill and thereby avoid the inland freight charge on equivalently priced imported pig iron. Thus, if the import price in a surrogate country is to be used as the basis for the pig iron price in the PRC, Commerce cannot assume that a Chinese iron castings manufacturer would purchase domestic pig iron at the import price, rather than imported pig iron at the import price, regardless of the respective freight costs for inland transportation of the domestic and imported pig iron. Realistically, such a manufacturer would minimize its material and freight costs by purchasing imported pig iron if the cost of transportation from the port to the foundry were less than the cost of transportation from the domestic pig iron mill to the foundry.
We recognize the difficulty, in a nonmarket economy case, of selecting a methodology that produces reasonably accurate estimates of the true value of the factors of production, as the statute directs. See 19 U.S.C. § l677b(c)(l), (e) (1988). Accordingly, we do not dictate the particular methodology that Commerce must use to determine the freight component in this case, but leave that decision to the discretion of Commerce. Even acknowledging the difficulties of calculating the value of the factors of production by using economic data from a surrogate market economy country, however, we conclude that the methodology chosen by Commerce in this case has led to a substantial overvaluation of the total freight expense, and that the constructed foreign market value must therefore be recalculated.
B
We next turn to the question whether Commerce erred in constructing foreign market value for the 1988-89 period by adding to the constructed value a cost for Guangdong’s after-sale warehousing of iron castings prior to their shipment to customers. In its final results, Commerce explained the adjustment as follows:
[A]n adjustment should be made to account for after-sale warehousing expenses. Guangdong Minmetals reported that it stored' finished merchandise in a warehouse for an average of one week after the time of sale and before the time of shipment. Therefore, because we are calculating U.S. price on a purchase-price basis, we made an adjustment to constructed value, using, as the best information otherwise available, petitioners’ estimate of the average U.S. costs for storing finished castings in rented warehouses during 1988.
56 Fed.Reg. 2742, 2745 (1991). The Court of International Trade upheld Commerce’s decision on that issue.
D & L challenges Commerce’s addition of constructed warehousing expenses to the constructed foreign market value of Guang-dong’s products, asserting that because Gu-*1409angdong owned the warehouse where the castings were stored before their shipment, the cost of warehousing should have been treated as an indirect cost, a form of general overhead, and should not have been separately added to foreign market value.
We disagree. Because the warehousing was an expense “incidental to placing the merchandise under consideration in condition, packed ready for shipment to the United States,” 19 U.S.C. § 1677b(e)(l)(C) (1988), and because Guangdong’s warehousing expenses were not otherwise accounted for in Commerce’s calculation of the constructed foreign market value, it was legitimate for Commerce to adjust the constructed value to account for those expenses. Moreover, even though Guangdong owned the warehouse at which the iron castings were stored prior to shipment to the United States, the warehousing necessarily involved some costs to Guangdong, whether those costs be regarded as fixed or variable. In light of the fact that Guangdong provided Commerce with little information on which to base its calculation of the warehousing costs, we hold that it was not improper for Commerce to calculate those costs by reference to warehouse rents in the United States. We therefore sustain the decision of the Court of International Trade on the warehouse issue.
C
We next consider Guangdong’s challenge to Commerce’s use of a Pakistani foundry as a surrogate for Guangdong’s foundries for the purpose of calculating the factory overhead component of the constructed foreign market value. Commerce’s finding is based on a single cable from the United States embassy in Pakistan reporting on overhead rates at Pakistani castings foundries. Guangdong argues that the cable does not support Commerce’s finding.
Paragraph one of the cable provides a general description of the Pakistani castings industry:
Manhole coverings and other products from east iron are fabricated at some two thousand mini-foundries, located throughout the country. Pakistan’s six foundries with monthly capacities in excess of 1,000 metric tons and high overheads cannot compete with these, cottage industry competitors. Most of the mini-foundries are located in the industrial cities of Lahore, Karachi, Gujranwala and Gujrat, not in industrial sites, but in houses or small towns. Such small informal enterprises normally do not document their activities, both to reduce costs and to avoid taxation. Paragraph 16 of the cable explicitly deals
with factory overhead:
Factory Overhead: Overheads at small units inclusive of profit and tax fall at 20 to 30 percent. At large foundries however, factory overheads are 40 to 50 percent.
Finally, paragraph 17 of the cable provides a specific cost breakdown for “a large Lahore-based foundry” and notes that no cost breakdowns are “available for small foundries.” Commerce used the data from that cost breakdown to calculate the surrogate overhead that was then made part of the constructed value of Guangdong’s merchandise.
As a source of information from which to calculate overhead, the cable is seriously deficient. Paragraph one describes two types of foundries operating in Pakistan: “mini-foundries” and “six foundries with monthly capacities in excess of 1,000 metric tons and high overheads.” Paragraph 16 then divides the foundries into “small” foundries with overheads of 20 to 30 percent and “large” foundries with overheads of 40 to 50 percent. The debate among the parties on appeal centers on whether the “large Lahore-based foundry” referred to in paragraph 17 is meant to represent one of the six foundries with “high overheads” identified in paragraph one; one of the “large” foundries with overheads of 40 to 50 percent identified in paragraph 16; or one falling somewhere on a continuum between the “mini-foundries” and the large, high overhead foundries identified in paragraph one.
Guangdong asserts that its four foundries, which have capacities ranging from approximately 90 metric tons per month to approximately 420 metric tons per month, are small in comparison to the 1000-metric-ton-eapaci-ty foundries described in paragraph one of *1410the cable. Guangdong interprets the cost breakdown data in paragraph 17 of the cable to apply to the 1000-metrie-ton-eapacity foundries referred to in paragraph one, and not to foundries of Guangdong’s size. The government; on the other hand, argues that the reference to “large” foundries in paragraph 17 does not apply to what it characterizes as the “mega-foundries” identified in paragraph one. According to the government, the reference to a “large” foundry in paragraph 17 denotes a foundry that is larger than cottage or “mini-foundries,” but smaller than the “mega-foundries,” and is therefore comparable in size to Guangdong’s foundries.
From the skimpy record materials pertaining to this issue, it is impossible to determine whether the “large Lahore-based foundry” referred to in paragraph 17 of the cable is meant to refer to a foundry comparable in size to Guangdong’s foundries. What is clear is that the government’s interpretation of the cable is based on pure conjecture. The government asserts that the large foundry referred to in paragraph 17 is reasonably comparable in size to Guangdong’s foundries and therefore is a proper surrogate, but the government points to nothing in the cable or elsewhere in the record to back up that assertion. So far as the record reflects, the “large” foundry referred to in paragraph 17 of the cable could well refer to the 1000-metrie-ton-capacity foundries referred to in paragraph one of the cable, which the government concedes would be “larger than Gu-angdong’s foundries and, thus, perhaps not comparable.” There is simply no way of knowing; in any event, there is no basis in the record to accept the- government’s assertions on this point.
The domestic producers follow an approach quite different from — and indeed contradictory to — that taken by the government. In their brief, the domestic producers argue that the “large Lahore-based foundry” referred to in paragraph 17 of the cable must have been one of the 1000-metric-ton-eapac-ity foundries referred to in paragraph one— exactly the argument made by Guangdong and disavowed by the government. The domestic producers then argue that the large Lahore-based foundry was a proper surrogate for calculating overhead even though, according to the cable, it was considerably larger than Guangdong’s foundries.
The problem with the domestic producers’ argument is that it assumes, without any support in the record, that the overhead for the 1000-metric-ton~capaeity foundry in Lahore is comparable to the overhead for considerably smaller foundries such as the Guangdong foundries. The report in the embassy cable of a huge disparity between the overhead rates for large and small foundries in Pakistan easts doubt on the assertion that overhead rates for the large Lahore-based foundry and for foundries the size of Guangdong’s would be comparable; the government’s acknowledgment that a 1000-metric-ton-capacity foundry would “perhaps not [be] comparable” to the Guangdong foundries makes the domestic producers’ assertion even more suspect. That assertion therefore cannot serve as a basis for upholding Commerce’s decision on the issue of the valuation of overhead.
In sum, Commerce should have sought more information from its representatives in Pakistan with regard to the size of the “large Lahore-based foundry,” and whether the overhead for that foundry is comparable to the overhead that would be experienced by a foundry the size of Guangdong’s foundries, information that Commerce presumably can obtain on remand. On the current state of the record, Commerce’s determination of surrogate factory overhead is unsupported by substantial evidence and must be remanded for further consideration.
IV
The final question we must resolve is whether it was lawful for Commerce to use an invalidated dumping margin as the “best information available” in its determination of an “all others” rate to be charged against the non-responsive exporters (i.e., all of the exporters except Guangdong) in the 1989-90 administrative review. In light of our recent decision in D & L Supply Co. v. United States, 113 F.3d 1220 (Fed.Cir.1997), we hold that it was not.
*1411In the final results of the 1989-90 administrative review, Commerce stated that it was using the antidumping margin assigned to Guangdong as the margin for all other Chinese exporters, who did not respond to Commerce’s questionnaire for that review. See 57 Fed.Reg. 10,644 (1992). That decision was based on Commerce’s long-standing practice of assigning to respondents who fail to cooperate with Commerce’s investigation the highest margin calculated for any party in the less-than-fair-value investigation or in any administrative review. See D & L Supply, 113 F.3d at 1222. The rate that Commerce assigned to Guangdong in the final results for the 1989-90 period was 92.74 percent, which was the highest rate assigned to any party up to that point and thus was selected as the “best information available” rate for the other Chinese exporters.
Along with Guangdong, the U.S. importers other than D & L filed suit in the Court of International Trade asserting various errors in the calculation of foreign market value for the 1989-90 period, including the overstatement of freight and overhead. They alleged that Commerce also “made other errors which substantially increase the margins applied to Plaintiffs’ imported castings.”
The Court of International Trade agreed with Guangdong and the importers in several respects and remanded to Commerce for recalculation of Guangdong’s dumping margin. Commerce then issued a revised determination that adjusted the dumping margin downward from 92.74 percent to 31.05 percent. On appeal from the remand results, the Court of International Trade rejected Guang-dong’s and the importers’ challenges to the revised rate. In addition, the court stated that it agreed with Commerce and the domestic producers that the U.S. importers had not challenged the application of the 92.74 percent dumping margin to exporters other than Guangdong, and that they were therefore foreclosed from challenging that aspect of the remand determination.
It is clear and undisputed that the importers challenged Commerce’s final results, asserting various errors in Commerce’s calculation of Guangdong’s dumping margin. It is also clear to us that the importers’ challenge, which was not expressly limited to an adjustment of the dumping margin assigned to Guangdong, was meant to encompass the margin for the other exporters. All parties to the proceeding understood that Commerce was relying on Guangdong’s rate as the best information available to be applied against all of the other respondents. The importers who had purchased castings from Chinese exporters other than Guang-dong had only one interest in the proceeding before the Court of International Trade: to reduce the rate applied to them respective imports. Because of Commerce’s reliance on Guangdong’s rate as the “best information available,” the other exporters’ rates stood or fell with Guangdong’s rate. To attack the rate applied to the other exporters, it was necessary to attack the rate applied to Gu-angdong’s imports.
It is true that the importers did not object to the use of Guangdong’s rate as the rate for the other exporters. That issue, however, is quite distinct from the question of the correctness of Guangdong’s rate, which dictates the “best information available” rate for the other exporters. As to the latter issue, we are satisfied that the importers’ complaint, which refers to the “all others” rate and to the “margins applied to [the importers’] castings” (ie., both the margin applied to Guang-dong’s castings and the margin applied to all other producers’ castings), made the scope of their claim sufficiently clear to preserve the issue before both the Court of International Trade and this court.
With respect to the merits, our decision in D & L Supply controls. In that case, we held it unlawful for Commerce to use the invalidated 92.74 percent rate as the best information available in the 1990-91 administrative review of the same antidumping duty order that is at issue in this case. We held that “when the dumping margin on which the BIA rate is based is invalidated before the BIA rate has become final, it is irrational to ignore the invalidity of the underlying rate and to uphold the BIA rate as purportedly based on the ‘best information available.’” 113 F.3d at 1224. The principles announced *1412in D & L Supply apply with full force here. Accordingly, we reverse Commerce’s reliance on Guangdong’s invalidated 92.74 percent dumping margin as the dumping margin to be applied to products sold by the other Chinese exporters.
Each party shall bear its own costs for this appeal.
AFFIRMED-IN-PART, REVERSED-IN-PART, AND REMANDED.