Opinion for the Court filed by Circuit Judge DYK. Dissenting opinion filed by Circuit Judge REYNA.
[1370]*1370DYK, Circuit Judge.
Jiangsu Jianghai Chemical Group, Ltd. (“Jiangsu Jianghai”) appeals a decision of the United States Court of International Trade that the Department of Commerce (“Commerce”) did not exceed the scope of a remand order when it recalculated the U.S. price and that the explanation given by Commerce for its calculation of the separate rate on remand was not unreasonable. We affirm in part, and reverse and remand in part.
Background
When merchandise is sold in the United States at less than fair value, Commerce is authorized by statute to impose antidumping duties. These duties are equal to the amount by which the price of the merchandise in the exporting country (“normal value”) exceeds the price of the merchandise in the United States (“export price” or “U.S. price.”). See 19 U.S.C. §§ 1673e(a)(l), 1677b(a)(l), 1677a(a). The statute provides that Commerce will generally determine individual dumping margins for each known exporter or producer, but if that is not practicable, Commerce may individually investigate a reasonable number of respondents. 19 U.S.C. § 1677f-l(c)(l)-(2). Those not individually investigated are assigned a separate rate, sometimes referred to as the “all-others” rate. 19 U.S.C. § 1673d(e)(l)(B)(ii). In proceedings involving non-market economy countries, including China, Commerce presumes that exporters and producers are state-controlled, and assigns them a single state-wide rate. See 19 C.F.R. § 351.107. This presumption is rebuttable; a company that demonstrates sufficient independence from state control may apply to Commerce for a separate rate. Here, Commerce concluded that the appellant was entitled to a separate rate. The central issue concerns the calculation of that separate rate.
This underlying proceeding involves an investigation of imports of 1-hydroxyethy-lidene-1, 1-diphosphonic acid (“HEDP”) from the People’s Republic of China, initiated in response to a petition filed by Compass Chemical International, LLC (“Compass Chemical”). Commerce sent initial quantity and value (“Q & V”) questionnaires to ten known HEDP exporters and producers identified in the petition. 1 Hydroxyethylidene 1, 1 Diphosphonic Acid from the Republic of India and the People’s Republic of China, 73 Fed.Reg. 62,470 n. 1 (Dep’t of Commerce Oct. 21, 2008) (“Preliminary Determination”). Five companies responded: Changzhou Wujin Fine Chemical Factory Co., Ltd. (“Wujin Fine Chemical”); Changzhou Kewei Fine Chemical Factory (“Kewei”); BWA Water Additives U.S. LLC (“BWA”); Nanjing University of Chemical Technology Changzhou Wujin Water Quality Stabilizer Factory Ltd. (“Wujin Water”); and Jiangsu Jianghai. Because Jiangsu Jianghai was not one of the ten companies identified in the petition, Commerce did not send it an initial Q & V questionnaire; thus, its response was voluntary.
The questionnaire responses indicated that BWA was one of the largest exporters of HEDP from China to the U.S. during the period of investigation. However, Commerce did not select BWA as a mandatory respondent because BWA refused to permit public disclosure of its supplier, and was therefore ineligible to be individually investigated. Commerce instead selected Wujin Water and Kewei as the mandatory respondents. Kewei subsequently notified Commerce that it would no longer participate in the investigation, leaving Wujin Water as the only cooperating mandatory respondent. Wujin Water, Wujin Fine Chemical, and Jiangsu Jianghai all [1371]*1371responded to Commerce’s supplemental questionnaires, and Jiangsu Jianghai and Wujin Fine Chemical requested that Commerce assign them separate, company-specific rates. Because it refused to supply the necessary information, BWA was ineligible to apply for a separate, company-specific rate.
On October 21, 2008, Commerce published a preliminary determination that HEDP from China was, or was likely to be, sold in the United States at less than fair value, and assigned antidumping rates to Chinese producers of HEDP.1 Preliminary Determination 1. Cooperating mandatory respondent Wujin Water received a preliminary rate of 24.30%, based on the data it had submitted to Commerce. Id. When Commerce determines that a respondent has not cooperated to the best of its ability in supplying Commerce with requested information, Commerce may subject that respondent to “adverse facts available” (“AFA”) and employ an inference adverse to the respondent’s interests when selecting among the facts available to determine its rate. See 19 U.S.C. § 1677e(b). Applying AFA, Commerce assigned a preliminary rate of 72.42% to non-cooperating mandatory respondent Kewei. Kewei’s AFA rate reflected the amount by which the petition’s alleged normal value exceeded the petition’s alleged export price. As required by section 1677e(c), Commerce corroborated Kewei’s AFA rate using data submitted by Wujin Water.2 Because Kewei had not demonstrated its independence from state control, Commerce treated Kewei as part of a “China-wide entity”; therefore, all Chinese exporters and producers that did not qualify for separate rates were likewise assigned the 72.42% rate.
Turning to Jiangsu Jianghai and Wujin Fine Chemical, Commerce determined that they had provided sufficient evidence of their independence from state control and qualified for a separate, company-specific rate. For preliminary purposes, Commerce assigned them “a weighted-average margin based on the experience of mandatory respondents and excluding any de minimis or zero rates or rates based on [AFA]”. 73 Fed.Reg. at 62,473; see also 19 U.S.C. § 1673d(c)(5)(A). Because the AFA rate assigned to mandatory respondent Kewei was excluded from this calculation, Jiangsu Jianghai and Wujin Fine Chemical were assigned cooperating mandatory respondent Wujin Water’s preliminary rate of 24.30%. Id.
Wujin Water provided Commerce with its final data submission in December of 2008. As a result, Commerce reduced Wujin Water’s margin from 24.30% to de min[1372]*1372imis. 1-Hydroxyethylidene-l, 1-Diphos-phonic Acid from the People’s Republic of China: Final Determination of Sales at Less Than Fair Value, 74 Fed.Reg. 10,545 (Dep’t of Commerce Mar. 11, 2009). Commerce continued to apply the single anti-dumping rate of 72.42% to all other Chinese exporters and producers, with the exception of Wujin Fine Chemical and Jiangsu Jianghai, which continued to qualify for a separate rate.
Under the trade statute, the separate rate for separate rate respondents like Wujin Fine Chemical and Jiangsu Jianghai is normally “an amount equal to the weighted average of the estimated weighted average dumping margins established for exporters and producers individually investigated, excluding any zero and de minimis margins, and any margins determined entirely [on the basis of AFA].” 19 U.S.C. § 1673d(c)(5)(A). Commerce determined that it was no longer appropriate to assign Wujin Water’s rate to the separate rate applicants, because that rate had been reduced to de minimis. Commerce looked to section 1673d(c)(5)(B), which provides that when the margins established for all individually investigated respondents are zero, de minimis, or based entirely on AFA, Commerce “may use any reasonable method to establish the estimated all-others rate for exporters and producers not individually investigated, including averaging the estimated weighted average dumping margins determined for the exporters and producers individually investigated.” Id. Commerce explained that
because there are no rates other than de minimis [i.e., Wujin Water’s] or those based on AFA [i.e., Kewei’s], we have determined to take a simple average of the AFA and the de minimis rate calculated for Wujin Water as a reasonable method for purposes of determining the rate assigned to Wujin Fine Chemical and Jiangsu Jianghai.
74 Fed.Reg. at 10,546. Consequently, Wujin Fine Chemical and Jiangsu Jianghai were assigned a rate of 36.21%, representing a simple average of Kewei’s 72.42% rate and Wujin Water’s de minimis rate. Id. at 10,547.
On June 25, 2009, Jiangsu Jianghai and Wujin Fine Chemical filed an action in the Court of International Trade, in which they challenged Commerce’s corroboration of the AFA rate used to calculate their separate rate, as well as the financial ratios and surrogate values used to calculate the AFA rate. They argued, inter alia, that using a flawed AFA rate in calculating a separate rate “would, in essence, punish a fully cooperative respondent and apply adverse inferences in the absence of any finding of a lack of cooperation.” Mot. J. Agency R. at 16, Changzhou Wujin Fine Chem. Factory Co. v. United States, No. 09-CV-0216 (Ct. Int’l Trade Sept. 21, 2009), ECF No. 25. The court granted the government’s request for a partial voluntary remand, and issued an order authorizing Commerce
to reconsider the separate rate amount after examining the following issues []: (i) whether Commerce corroborated the [AFA] rate upon which it relied in calculating the separate rate and (ii) whether Commerce used the proper financial ratios and the proper surrogate value for Phosphorus Trichloride and steam in calculating the [AFA] rate Commerce used as a basis for the separate rate.
Changzhou Wujin Fine Chem. Factory Co. v. United States, No. 09-CV-0216, 2010 Ct. Intl. Trade LEXIS 108, at *1 (Feb. 8, 2010) (“Remand Order”). The court retained jurisdiction pending the remand. Id. at *2.
On remand, Commerce abandoned the AFA rate based on the petition, because it [1373]*1373found that Wujin Water’s data could no longer corroborate that rate.3 See Final Results of Redetermination Pursuant to Court Order at 5, Changzhou Wujin Fine Chem. Factory Co. v. United States, No. 09-CV-0216 (Ct. Int’l Trade May 3, 2010), ECF No. 44 (“Final Remand Redetermination”). As a preliminary step in recalculating the separate rate, Commerce calculated a new, hypothetical AFA rate of 30.94%, which it based on Wujin Water’s verified normal value data and unverified U.S. price data obtained from the non-cooperating respondent BWA. Commerce did not assign this hypothetical AFA rate to any party; rather, it was used solely as the predicate for calculating the new separate rate for Wujin Fine Chemical and Jiangsu Jianghai. Commerce determined that because the hypothetical AFA rate was based on information obtained during the course of the investigation, corroboration was unnecessary. Averaging the hypothetical AFA rate of 30.94% with Wujin Water’s de minimis rate, Commerce obtained a separate rate of 15.47%, which it assigned to Wujin Fine Chemical and Jiangsu Jianghai.4
In their comments to Commerce on its draft remand results, Wujin Fine Chemical and Jiangsu Jianghai argued that Commerce should not base the hypothetical AFA rate on BWA’s data. See Comments on Draft Remand Determination by Wujin Fine Chemical & Jiangsu Jianghai at 6-8, Changzhou Wujin Fine Chem. Factory Co. v. United States, No. 09-CV-0216 (Ct. Int’l Trade Mar. 31, 2010), ECF No. 47-5. In the Final Remand Redetermination, Commerce stated that any other data — including the mandatory respondents’ data, or an average of BWA’s data with those of any other respondent — would produce a hypothetical AFA rate of zero. According to Commerce, this would be unacceptable, because
[a]n AFA rate of zero would not be sufficiently adverse as to effectuate the purpose of the facts available rule to induce respondents to provide the Department with complete and accurate information in a timely manner and would not ensure that uncooperative parties do not obtain a more favorable result by failing to cooperate than if they had cooperated fully.
Final Remand Redetermination 16.
Following the Final Remand Redetermination, the Court of International Trade considered the parties’ comments on the remand results. The appellant objected that Commerce’s deterrence rationale was inappropriate under the circumstances:
The ultimate objective in this case is to calculate a margin for the Plaintiffs using a reasonable method. The most appropriate, if not the only appropriate method, is to use the Q & V [data] of the [cooperating] plaintiffs.... The Department rejected the use of this data alleging that it would result in an AFA rate which would be zero and that a rate of zero would not act as a deterrent. Plaintiffs note that the new ‘AFA’ rate would not be used for any purpose other than assigning a rate to [cooperating separate rate applicants].... [A]s discussed above, deterrence is not the only factor involved in calculating an AFA rate, and deterrence is never a factor in calculating a non-AFA rate.
[1374]*1374Pis.’ Comments on Remand Determination 12-13, Changzhou Wujin Fine Chem. Factory Co. v. United States, No. 09-CV-0216 (Ct. Int’l Trade June 4, 2010), ECF No. 48 (“Pis.’ Comments on Remand”).
The Court of International Trade affirmed Commerce’s remand results. Changzhou Wujin Fine Chem. Factory Co. v. United States, No. 09-CV-0216, 2010 Ct. Intl. Trade LEXIS 96, at *15 (Aug. 5, 2010). The court rejected the appellant’s contention that Commerce exceeded the scope of the remand order by recalculating the U.S. price, concluding that Commerce properly used other sources of U.S. price data when it determined that it could not corroborate its initial AFA rate on remand. The court found that the separate rate recalculation was supported by substantial evidence and was reasonable in view of Commerce’s well-established methodologies. It rejected the contention that BWA’s data failed to reflect commercial reality as required by Gallant Ocean (Thail.) Co. v. United States, 602 F.3d 1319, 1323 (Fed.Cir.2010), observing that because BWA had a large market share, its data substantially reflected the HEDP market. The court likewise considered and rejected the appellant’s argument that Commerce “erroneously chose BWA’s data solely to obtain a positive separate rate and, instead, should have used data from another uncooperative respondent.” Changzhou, 2010 Ct. Int’l. Trade LEXIS 96, at *11.
After the Court of International Trade denied Jiangsu Jianghai and Wujin Fine Chemical’s motions for reconsideration, Changzhou Wujin Fine Chem. Factory Co. v. United States, No. 09-CV-0216, 2010 Ct. Intl. Trade LEXIS 106, at *4 (Sept. 13, 2010), this appeal followed.5 We have jurisdiction pursuant to 28 U.S.C. § 1295(a)(5).
Discussion
This appeal presents the following issues: (1) whether the Court of International Trade has jurisdiction to review Commerce’s post-remand modification of U.S. price; and (2) whether Commerce acted arbitrarily when recalculating the separate rate assigned to appellant Jiangsu Jianghai, a cooperative respondent.
This court reviews decisions of the Court of International Trade concerning Commerce’s antidumping determinations by applying the same standard of review used by the trade court. Gallant Ocean, 602 F.3d at 1323 (citing Tung Mung Dev. Co. v. United States, 354 F.3d 1371, 1378 (Fed.Cir.2004)). In essence, Commerce’s determination will be sustained unless it is “unsupported by substantial evidence on the record, or otherwise not in accordance with law.” 19 U.S.C. § 1516a(b)(l)(B)(i); see F.lli De Cecco Di Filippo Fara S. Martino S.p.A. v. United States, 216 F.3d 1027, 1031 (Fed.Cir.2000). Commerce’s decision will be set aside if it is arbitrary and capricious. See, e.g., SKF USA, Inc. v. United States, 254 F.3d 1022, 1028 (Fed.Cir.2001) (“We review Commerce’s decision under the Administrative Procedure Act and any other applicable law.”).
I
We must first decide whether Commerce’s recalculation of U.S. price was within the scope of the Court of International Trade’s remand order.
We generally disfavor limited remands that restrict Commerce’s ability to collect and fully analyze data on a contested issue. See Am. Silicon Techs. v. United States, 334 F.3d 1033, 1038-39 (Fed.Cir.2003) (“By sharply limiting Commerce’s [1375]*1375inquiry, the trial court’s remand actually prevented Commerce from undertaking a fully balanced examination that might have produced more accurate results.”). So too, an appellant “faces a very high hurdle when it tries to convince us that, despite the remanding Court’s satisfaction, we must conclude that the ITC on remand acted outside the scope of the remand directions.” Trent Tube Div., Crucible Materials Corp. v. Avesta Sandvik Tube AB, 975 F.2d 807, 814 (Fed.Cir.1992).
The appellant argues that the voluntary remand order was limited in scope, only permitting Commerce to reconsider the normal value and the impact the surrogate values and financial ratios had on the AFA rate. According to the appellant, the remand order placed only two issues before Commerce: “whether Commerce corroborated the adverse facts available rate upon which it relied in calculating the separate rate,” and whether Commerce used the proper financial ratios and surrogate values. Remand Order 1-2. The appellant argues that “[t]he first of these issues is clearly inapplicable to the U.S. price as it merely goes to the test of the selected final rate,” Appellant’s Br. 17, while the second issue relates to calculation of the normal value, not U.S. price. Because the remand order required only corroboration of the normal value and no party challenged the U.S. price, the appellant concludes, Commerce had no authority to recalculate U.S. price.
We agree with Commerce that recalculating the U.S. price was within the scope of the voluntary Remand Order. Nothing in the remand order limited Commerce to recalculating the normal value when “examining ... whether [it] corroborated the adverse facts available rate upon which it relied in calculating the separate rate.” See Remand Order 1-2. Commerce specifically requested the partial voluntary remand “to examine the issues related to the adverse facts available rate, to make any necessary recalculations based on its examination, and to obtain comments from the parties.” Def s. Resp. to Mot. for J. on the Admin. R. at 2, 23, Changzhou Wujin Fine Chem. Factory Co. v. United States, No. 09-CV-0216 (Ct. Int’l Trade Jan. 26, 2010), ECF No. 36.
Commerce’s reexamination of the AFA rate used in calculating the appellant’s separate rate necessarily involved reconsideration of the underlying variables, i.e., normal value and U.S. price. See 19 U.S.C. § 1673e(a)(l). Once Commerce determined that it could not corroborate Kewei’s AFA rate as required by statute, see 19 U.S.C. § 1677e(c), Commerce reasonably chose to abandon the secondary information it originally used for normal value and U.S. price. We reject the appellant’s argument that the remand order’s second instruction to Commerce, regarding the proper financial ratios and surrogate values, limited Commerce to reconsideration of the normal value only. We conclude that Commerce acted within the remand order’s broad scope when it decided to select a new U.S. price for purposes of calculating the appellant’s new separate rate.
II
The second question is whether Commerce acted arbitrarily in recalculating the appellant’s separate rate, when Commerce constructed a hypothetical AFA rate based on U.S. price data obtained from a non-cooperating respondent, averaged that hypothetical AFA rate with the de minimis rate for Wujin Water, and then assigned the resulting separate rate to the appellant. The hypothetical AFA rate Commerce generated on remand was an “AFA rate” in name only: its sole purpose was to serve as a value in the [1376]*1376recalculation of the separate rate. The hypothetical AFA rate was not assigned to any individually investigated entity, and only affected cooperating respondents.6
The applicable statute contemplates that the separate rate will generally be calculated by taking a weighted average of dumping margins assigned to individually investigated entities, excluding those margins that are zero, de minimis, or determined entirely on the basis of AFA. 19 U.S.C. § 1673d(c)(5)(A). Where, as here, the dumping margins for all individually investigated entities are zero, de minimis, or determined entirely on the basis of AFA, Commerce “may use any reasonable method to establish the estimated all-others rate for exporters and producers not individually investigated, including averaging the estimated weighted average dumping margins determined for the exporters and producers individually investigated.” 19 U.S.C. § 1673d(c)(6)(B) (emphases added). We have recognized that in some circumstances, antidumping investigations require calculations that are “by necessity imperfect,” and we consider the circumstances when reviewing Commerce’s application of the statute. Thai I-Mei Frozen Foods Co. v. United States, 616 F.3d 1300, 1309 (Fed.Cir.2010).
The appellant has not argued that the simple average methodology suggested by the statute is unreasonable in all circumstances, nor even that it is necessarily unreasonable when applied to cooperating parties.7 Rather, the appellant argued to Commerce and before the Court of International Trade that Commerce’s stated reasons for its approach were arbitrary, because deterrence is not a sufficient justification when calculating a rate that solely affects cooperating respondents.8 The [1377]*1377parties continued to dispute this point in their briefs on appeal. The appellant contended that
[t]he most appropriate, if not the only-appropriate method [of calculating the appellant’s separate rate], is to use the Q & V [data] of the appellant, and another cooperative party which actively participated in the investigation and provided multiple responses to Department questionnaires____The Department rejected the use of this data alleging that it would result in an AFA rate which would be zero and that a rate of zero would not act as a deterrent. Appellant notes that the new “AFA” rate would not be used for any purpose other than assigning a rate to Appellant, a cooperative [separate rate applicant] respondent. ... [Deterrence is not the only factor involved in calculating an AFA rate, and deterrence is never a factor in calculating a non-AFA rate.
Appellant’s Br. 34-35. Commerce responded that “a zero percent adverse facts available rate would not be sufficiently adverse to effectuate the purpose of the adverse facts available rule,” Appellee’s Br. 32, and therefore
when Commerce determines an [AFA] rate, it selects a rate that is sufficiently adverse and selects the highest rate in the record of the proceeding to ensure the party does not obtain a more favorable result by failing to cooperate than if it had cooperated fully. Selecting Kewei’s data, which would have yielded a zero percent facts available rate, would not satisfy this standard.
Id. at 36 (citation omitted). In its reply brief, the appellant continued to argue that “[t]he United States[’] entire basis for the selection of a new U.S. price in this matter is that of ‘deterrence.’ This results oriented method is not a sufficient basis for the selection of an AFA rate,” Reply Br. 9, to be used solely “as a substitute in the calculation” of a separate rate for cooperating respondents. Id. at 9 n. 1.
The question is whether Commerce’s method was arbitrary and capricious. The dissent urges that the arbitrary and capricious standard is inapplicable, and that we should instead apply a substantial evidence standard. What the dissent fails to appreciate, however, is that a reviewing court must apply both standards, and that “an agency’s finding may be supported by substantial evidence,” yet “nonetheless reflect arbitrary and capricious action.” Bowman Transp., Inc. v. Arkansas-Best Freight Sys., Inc., 419 U.S. 281, 284, 95 S.Ct. 438, 42 L.Ed.2d 447 (1974). The dissent correctly points out that the substantial evidence standard applies to review of factual determinations. But here we are evaluating the agency’s reasoning, which is reviewed under the arbitrary and capricious (or contrary to law) standard. See Motor Vehicle Mfrs. Ass’n v. State Farm Mut. Auto. Ins. Co., 463 U.S. 29, 48-49, 103 S.Ct. 2856, 77 L.Ed.2d 443 (1983). Therefore, we first ask whether Commerce articulated an adequate reason for calculating the appellant’s separate rate in the manner chosen on remand.
“The grounds upon which an administrative order must be judged are those upon which the record discloses that its action was based.” SEC v. Chenery Corp., 318 U.S. 80, 87, 63 S.Ct. 454, 87 L.Ed. 626 (1943). Tellingly, Commerce has not con[1378]*1378tended that BWA’s data are more reliable or accurate than data from the other respondents.9 Rather, the Final Remand Redetermination explained that Commerce used BWA’s data to obtain an adverse margin. Commerce determined that any other respondent’s U.S. price data (or any combination of BWA’s data with another respondent’s data) would have produced a de minimis hypothetical AFA rate, a result Commerce found unacceptable:
[a]n AFA rate of zero [or de minimis ] would not be sufficiently adverse as to effectuate the purpose of the facts available rule to induce respondents to provide the Department with complete and accurate information in a timely manner and would not ensure that uncooperative parties do not obtain a more favorable result by failing to cooperate than if they had had cooperated fully.
Final Remand Redetermination 16 (emphasis added). In other words, Commerce chose to use BWA’s data in the course of recalculating the appellant’s separate rate because those data gave the most adverse margin possible, and an adverse margin is necessary to deter non-cooperating respondents. See also de Cecco, 216 F.3d at 1032.
Deterrence is not relevant here, where the “AFA rate” only impacts cooperating respondents. We find no support in our caselaw or the statute’s plain text for the proposition that deterrence, rather than fairness or accuracy, is the “overriding purpose” of the antidumping statute when calculating a rate for a cooperating party. See, e.g., Parkdale Int’l v. United States, 475 F.3d 1375, 1380 (Fed.Cir.2007) (“[A]n overriding purpose of Commerce’s administration of antidumping laws is to calculate dumping margins as accurately as possible ... ”) (citing Rhone Poulenc, Inc. v. United States, 899 F.2d 1185, 1191 (Fed.Cir.1990)). Quite the contrary: applying an adverse rate to cooperating respondents undercuts the cooperation-promoting goal of the AFA statute. Commerce misses the point when it argues that the appellant cannot complain because it does not bear an AFA rate directly, but only a separate rate derived from the AFA rate, which is only half as adverse. Although the hypothetical AFA rate was not directly applied to a cooperating respondent, cooperating respondents were the only entities impacted by the recalculated rate.
Contrary to Commerce’s argument, the AFA statute does not compel this result. Commerce maintains that despite its broad discretion in making antidumping determinations, its hands are tied by section 1673d(c)(5)(B), which requires that the appellant’s separate rate be based on a traditional AFA rate. But the statute does not say that. Section 1673d(e)(5)(B) states that Commerce is to select “any reasonable method to establish the estimated all-others rate for exporters and producers not individually investigated, including averaging the estimated weighted average dumping mar[1379]*1379gins determined for the exporters and producers individually investigated.” 19 U.S.C. § 1673d(c)(5)(B) (emphases added). The statute’s use of the word “including” contemplates the possibility of other valid approaches, and the requirement that the method be “reasonable” imposes a duty on Commerce to select a method appropriate for the circumstances. Furthermore, AFA rates are disfavored by the statute: section 1673d(c)(5)(B) is an exception to the preferred method in section 1673d(c)(5)(A), which instructs that AFA rates, like de minimis and zero rates, should be excluded. Indeed, the only AFA rates contemplated under section 1673d(c)(5)(B)’s simple average methodology are those determined for “individually investigated” parties. Such rates are generally based on verified data, see 19 U.S.C. § 1677m(i)(l), and bear a reasonable relationship to the party’s actual business practices. See Gallant Ocean, 602 F.3d at 1323 (even “[a]n AFA rate must be a reasonably accurate estimate of the respondent’s actual rate, albeit with some built-in increase intended as a deterrent” (internal quotation marks omitted)). Substitution of a hypothetical “AFA rate” inapplicable to any individually investigated party is neither suggested nor compelled by the statute, particularly where the sole impact of that substitution will be on cooperating parties.
Finally, while administrative convenience might support averaging previously-determined, previously-corroborated rates assigned to mandatory respondents, including AFA respondents if there are no alternatives, see 19 U.S.C. § 1673d(c)(5)(B), such a justification can hardly support Commerce’s choice to calculate a hypothetical “AFA rate” for use solely as a “substitute” rate that will not be assigned to any mandatory respondent.
In summary, the issue before the court is not whether Commerce properly considers the policy goals of the anti-dumping statute in conducting its investigations (it does), nor whether Commerce may draw adverse inferences when assigning AFA rates to non-cooperating respondents (it may). Rather, we address the narrower question of whether Commerce acted reasonably when — by its own account — it cherry-picked the single data point that would have the most adverse effect possible on cooperating voluntary respondents, in a situation where there was no need or justification for deterrence. “[R]eview of an administrative decision must be made on the grounds relied on by the agency. ‘If those grounds are inadequate or improper, the court is powerless to affirm the administrative action by substituting what it considers to be a more adequate or proper basis.’ ” In re Lee, 277 F.3d 1338, 1345-46 (Fed.Cir.2002) (quoting Securities and Exchange Commission v. Chenery, 332 U.S. 194, 196, 67 S.Ct. 1575, 91 L.Ed. 1995 (1947)). In this case, we think it clear that Commerce acted in an arbitrary and capricious manner.
For the foregoing reasons, we affirm the decision of the Court of International Trade in part, reverse in part, and remand to Commerce to once again reconsider its approach to calculating the appellant’s separate rate. In doing so, Commerce must act non-arbitrarily and must explain why its approach is a “reasonable method” of calculating a separate rate, in light of the alternatives available, and with recognition of the fact that the remand calculation will affect only cooperating respondents.
AFFIRMED-IN-PART, REVERSED-IN-PART, AND REMANDED.
Costs
Each party shall bear its own costs.