DYK, Circuit Judge.
The Department of Commerce (“Commerce”) selected two exporters, Jacobi Carbons AB (“Jacobi”) and Calgon Carbon (Tianjin) Co., Ltd. (“CCT”), to individually examine in the third administrative review of an antidumping order. Commerce assigned both exporters de minimis dumping margins. Rather than using the “expected method” of averaging those de minimis margins to' calculate a separate rate for non-examined exporters Ningxia Guan-ghua Cherishmet' Activated Carbon Co., Ltd. (“Cherishmet”),
' Shanxi DMD Corp. (“Shanxi”), and' Ningxia Huahui Activated Carbon Company Ltd. (“Huahui”),
Commerce continued to apply the rates it had assigned those exporters in the second, immediately preceding administrative review. The Court of International Trade (“CIT”) held that Commerce’s use of prior dumping margins was impermissible with respect to Cherishmet and Shanxi but permissible with respect to Huahui. We affirm with respect to Cherishmet and Shanxi and. reverse arid remand with respect to Huahui. In each case, Commerce has failed to justify using the rate from the prior administrative review.
Background
■ I .
When merchandise is sold in the United States at less than fair value, Commerce is authorized by statute to impose antidump-ing duties. '
See
19 U.S.C. § 1673. These duties are equal to the dumping margin, 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 id.
§§ 1673e(a)(l), 1677b(a)(l), 1677a(a);
Changzhou Wujin Fine Chem. Factory Co.,
v.
United States,
701 F.3d 1367, 1370 (Fed.Cir.2012). Under the statute, Commerce is generally charged with determining individual dumping margins for each known exporter. 19 U.S.C. § 1677f-l(c)(l). When it is “not practicable” to determine individual margins for each exporter, the statute provides that Commerce may limit its examination to a “reasonable number of exporters” that either constitute a statistically representative sample, of all known exporters or account for the largest volume ,of the subject merchandise from the exporting country.
Id.
§ 1677f-l(c)(2).
In proceedings, involving ..non-market economy countries, including China, Commerce presumes that exporters , are state-controlled, and assigns them a single statewide dumping rate.
Changzhou,
701 F.3d at 1370; 19 C.F.R. § 351.107(d). t This presumption is rebuttable;. an exporter that demonstrates sufficient independence from state control may apply to Commerce for ,a different rate.
Changzhou,
701 F.3d at 1370. A separate rate, sometimes referred to as the “all-others” rate, is assigned to all non-individually examined exporters (“separate respondents”) when Commerce limits its examination to fewer than all known exporters. 19 U.S.C. § 1673d(c)(l)(B)(i)(II);
Changzhou,
701 F.3d at 1370.
Typically,-as discussed;'below,'this separate 'Or all-others rate is calculated by" averaging the rates of the individually examined exporters. Non-selected parties can request individual examination pursuant to 19 U.S.C. § 1677m(a), but Commerce is not obligated to grant such requests. Statement of Administrative Action accompanying the Uruguay Round Agreements Act, H.R. Doc. No. 103-316 (1994) [“SAA”],
reprinted in
1994 U.S.C.C.A.N. 4040, 4201 (“Commerce may decline to analyze voluntary responses because it would be unduly burdensome.”);
Yangzhou Bestpak Gifts & Crafts Co. v. United States
(“Bestpak”), 716 F.3d 1370, 1373 (Fed.Cir.2013). Here, Commerce determined that all of the parties to this appeal were entitled to separate rates. The central issue concerns the calculation of those separate rates for Cherishmet, Shanxi, and Huahui. As noted earlier, rather than using the average ,.of; the rates calculated for the individually examined exporters during the third " administrative review, Commerce used the rates applied to Cherishmet, Shanxi, and Huahui in the previous administrative review.
II
The underlying proceedings involve an initial investigation followed by three administrative reviews.of imports of “certain activated carbon*’ from the People’s Republic of China, which encompasses “powdered, granular, or pelletized carbon products obtained by ‘activating’ with heat and steam various materials, containing carbon.”
Notice of Antidumping Order: Certain Activated Carbon from the People’s Republic of China,
72 Fed.Reg. 20,-988, 20,988 (Apr. 27, 2007). In the initial investigation in 2007, Commerce individually investigated only the two largest volume exporters, Jacobi and CCT. Commerce determined that appellant Huahui and appellees Cherishmet and Shanxi were entitled to separate rates.
After receiving several requests for review of the initial antidumping order, Com
merce conducted a series of three administrative reviews. In the first review in 2009, Jacobi and CCT were selected as individual respondents, and Commerce granted Cherishmet’s request to be individually examined as a voluntary respondent.
Certain Activated .Carbon from, the People’s Republic of China: Notice of Preliminary Results of the Antidumping Duty Administrative Review and Extension of Time Limits for the Fiual Results,
74 Fed.Reg. 21,317, 21,318 .(May 7, 2009). In. the second review in 2010, Jacobi and Huahui were individually examined, while CCT, Cherishmet, and Shanxi, were given separate rates.
Certain Activated Carbon from the People’s Republic of China: Final Results and Partial Rescission of Second Antidumping
Duty,
Administrative Review,
75 Fed.Reg. 70,208, 70,208, 70,-209-10 (Noy. 17, 2010). Commerce assigned Jacobi a rate of $0.11/kg and Hua-hui a rate of $0.44/kg. It calculated the separate rate by averaging those individual margins, resulting in a separate rate of $0.28/kg, which was applied to Cherishmet and Shanxi.
In the third -and final review in 2011, which is the subject of this appeal, Commerce individually examined. Jacobi and CCT.
Certain Activated Carbon from the People’s Republic of China: Final Results and Partial Rescission of the Third Administrative Antidumping Duty Administrative Review [“Final Results
”], 76 Fed. Reg. 67,142, 67,142 (Oct. 31, 2011). Cher-ishmet, Shanxi, and Huahui were held entitled to a separate rate. Huahui submitted a request to be individually examined as a voluntary respondent, but Commerce denied its request. In the
Final Results,
Commerce determined thát Jacobi and CCT, the individually examined respondents, were not dumping, and assigned them de minimis margins.
The question was whether these de minimis rates should be ■ averaged and applied to the separate respondents.
As discussed in detail below, 19 U.S.C. § 1673d(c)(5) governs Commerce’s calculation of separate rates. Under the statute, when' all - individually examined exportérs are assigned de minimis margins, the “expected method” is for Commerce to calcü-late the separate rate by taking the average of the de minimis margins assigned to the individually examined respondents.
See
SAA, at 4201. If Commerce determines that following the expected method would not be feasible or would result in margins that would “not be reasonably reflective of potential dumping margins” for the separate respondents, Commerce may use “other reasonable methods.”
Id.
Rather than following the “expected method” of averaging the de minimis margins calculated for the individually examined respondents here, Commerce calculated separate rates for Huahui, Cherishmet, and Shanxi by continuing to apply the margins it had assigned them -during the previous period of review — Huahui was given $0,44/kg, the same rate it was assigned when individually examined during the second review, and Cherishmet and Shanxi were given $0.28/kg, the same separate rate that was calculated during the second review by averaging the rates for the individually examined respondents. Commerce does not contend that employing the expected method would be unfeasible. Instead, Commerce determined that the expected method would result in margins that would, not be reasonably reflective of the separate respondents’ actual dumping, explaining primarily that its policy was to exclude de minimis margins from
all separate rate calculations: “We agree with Petitioners that [Commerce] should not diverge from the practice of excluding zero and
de minimis
margins when calculating the separate rate .margin.!’ ’Memo-rándum to the File, through Ronald K. Lorentzen from Christian Marsh re: Issues and Dec. Mem. for the Final Results of the Third Antidumping Duty Administrative Review, dated Oct. 24, 2011, at 5
[“Memorandum
”],
available at.
J.A. 100992.
III
Huahui, Cherishmet, and Shanxi challenged Commerce’s separate rate calculations in the CIT.
Albemarle Corp. v. United States (“Albemarle I”),
931 F.Supp.2d 1280 (Ct. Int’l Trade 2013). The exporters argued that Commerce’s calculations were unreasonable because they were based exclusively on non-contemporaneous data, and asserted that Commerce failed to adequately explain its departure from the “well-established premise that the' Final Results of a proceeding should be based solely on the facts on'the record
in that proceeding.” Id.
at 1290. The CIT remanded and ordered Commerce to reconsider the margin it assigned to Cherishmet and Shanxi, explaining'that the
$0.28/kg margin was not based on data pertaining to any pricing behavior that occurred in the third [period of review]. Nor was it based on any data pertaining to these respondents; instead, Commerce reverted to a margin it determined in
another
review for
other
respondents! This margin does not reflect commercial reality with respect to [Shanxi or'Cherishmet], and is, in that sense, arbitrary.
Id.
at 1291. With! respect to Huahui, the CIT “reserve[d] any decision on whether the .margin assigned [] was permissible. Commerce may or. may not decide to assign Huahui a different margin based on other decisions it makes upon remand.”
Id.
at 1293.
On remand, cross-appellant Calgon Carbon Corporation (“CCC”), the parent company of CCT and a United States competitor of Cherishmet, Shanxi, and Huahui, petitioned Commerce to reopen the administrative record and collect additional data from the separate respondents to support its position that those companies should be assessed dumping duties.
Final Results of Redetermination Pursuant to Court Remand, Albemarle Corp. v. U.S.,
Consol. Ct. No. 11-00451 at 21 (Jan. 9, 2014)
[“Final Remand Redeterminatión”'],'available at
J.A. 101177. Commerce declined, stating, “we do not have the resources to individually review more than two respondents.”
Id.
Commerce recalculated the separate rates for Cherishmet and Shanxi; “fol-lowfing] the [CIT’s] logic, under protest, to its natural conclusión,” and averaging the de minimis margins assigned to the individually examined'respondents in the third review.
Id.
at 13.' This resulted in de minimis margins for Cherishmet and Shanxi. With respect to Huahui,' Commerce “decline[d] to reconsider Huahui’s dumping margin” and continued to assign the previous rate of $0¡44/kg.
Id.
at 22.
Following the Final. Remand Redetermi-nation, the CIT, having retained jurisdiction, affirméd Commerce’s' redetermina-tions and the dumping margins assigned to all exporters.
Albemarle Corp, Ningxia Huahui Activated Carbon Co. v. United States (“Albemarle II”),
27 F.Supp.3d 1336, 1352 (Ct. Int’l Trade 2014). While following the CIT’s instructions on remand, Commerce continued to view its original determinations as being correct. With respect to Cherishmet and Shanxi, the CIT reiterated that Commerce’s initial decision to carry forward the prior dumping margins was impermissible. It explained that Commerce’s assertion that the previous margins were contemporaneous
and reasonably reflective of actual margins was “factually incorrect ■ when viewed in the context of the record evidence of the third review.”
Id.
at 1344. The CIT explained that “no data on the record demonstrated that the pricing behavior of [Cher-ishmet and Shanxi] matched the pricing behaviors of the mandatory respondents in the previous review.”
Id.
Instead,. using contemporaneous data from the third. review .for the individually examined respondents resulted in a “reasonable reflection of the potential dumping margin” that Cherishmet and Shanxi “would have been assigned in the third review, had they been examined.”
Id.
at 1344-45 (internal quotation marks, citations, and alterations omitted).
With respect to Huahui, the CIT affirmed Commerce’s decision to not recalculate the $0.44/kg margin previously assigned, holding that Commerce’s method was reasonable because it “rélie[d] on data that were specific to Huahui’s sales and factors of production,” unlike the rate that had been carried over for Cherishmet and Shanxi.
Id.
at 1348. The CIT recognized that the data were derived from the previous period of review but concluded that Commerce’s, decision to choose “specificity to Huahui over contemporaneity” was reasonable.
Id.
Huahui appeals. The government and CCC cross-appeal. We have jurisdiction pursuant to 28 U.S.C. § 1295(a)(5).
Discussion
I
Wé review decisions of thé CIT concerning Commerce’s antidumping determinations by applying the same standard of review used by the CIT.
Bestpak,
716 F.3d at 1377;
Changzhou,
701 F.3d at 1374. Commerce’s determination will be set aside if it is arbitrary and capricious or not supported by substantial .evidence.
Changzhou,
701 F.3d at 1374;
see also SKF U.S.A., 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.”). The question, of whether Commerce’s statutory interpretation accords with law is guided by the two-part test articulated in
Chevron, U.S.A., Inc. v. Natural Resources Defense Council, Inc.,
467 U.S. 837, 842-43, 104 S.Ct. 2778, 81 L.Ed.2d 694 (1984).
See Bestpak,
716 F.3d at 1377;
Nan Ya Plastics Corp. v. United States,
810 F.3d 1333, 1341 (Fed.Cir.2016). We have acknowledged “Commerce’s special expertise” in antidumping cases and have “accorded substantial deference to its construction of pertinent statutes.”
SKF,
254 F.3d at 1028 (internal quotation marks, citations, and alterations omitted),
Under the statute, Commerce normally calculates the separate rate by averaging the “dumping margins established for exporters and producers individually investigated, excluding any zero and de minimis margins.” 19 U.S.C. § ■ 1673d(c)(5)(A);
Changzhou,
701 F.3d at 1372. The statute provides an exception, however, for situations, like the present one where all individually examined respondents receive de minimis margins. ‘In that case, 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)(5)(B). The Statement of Administrative Action, legislative history that is “recognized by Congress as an authoritative expression concerning the interpretation and application of the Tariff Act under 19 U.S.C. § 3512(d),”
Bestpak,
716 F.3d at 1373, explains that when all individually examined respondents are . assigned de minimis margins,
The expected method in such cases will be tó weight-average the zero and de minimis margins
... provided that volume data is available.
However, if this method is not feasible, or if it results in an average that would not be reasonably reflective ' of potential dumping margins
for non-investigated exporters 'or producers,
Commerce may use other reasonable methods;
SAA, at 4201 (underscoring added).
The SAA thus makes clear that under the statute, when all individually examined respom dents are assigned de minimis margins, Commerce is expected to calculate the separate rate by taking the average of those margins. Commerce may use “other reasonable methods,” but only if Commerce reasonably concludes that the expected method is “not feasible” or “would not be reasonably reflective of potential dumping margins.”
Id.
It is true, as the government points out, that 19 U.S.C. § 1673d applies on its face only to investigations, not periodic adrhin-istrative reviews.
See Amanda Foods (Vietnam) Ltd. v. United States,
714 F.Supp.2d 1282, 1290 (Ct. Int’l Trade 2010). But the statutory framework contemplates that Commerce will employ the same methods for calculating a separate rate in periodic administrative reviews as it does in initial investigations.
See
19 U.S.C. § 1675(a)' (amended 2016) (In conducting periodic administrative reviews, Commerce is required to “determine the dumping margin” to calculate “the amount of- any antidumping duty,” just as it must do -in initial, investigations).
.Indeed, for
this reason Commerce itself has found the statute’s methodology applicable in periodic administrative reviews as well as initial investigations, as it did in this case.
See Memorandum,
at, 4.,
The “expected method”, under the statute makes sense in light of the general assumption underlying the statutory framework. Here, the individually examined respondents account for a majority of the market during the relevant period, and are representative at the very least in terms of aggregate volume. The government argues that “the possibility exists that the pricing behavior of the largest exporters [selected for individual examination] may not reflect the pricing behavior of smaller exporters,” Br. of United States at 23-24, but there is no evidence hete that the largest exporters are not representative. The very fact that the- statute contemplates using data from the largest volume exporters suggests an assumption that those data can be viewed as representative of all exporters. The statute assumes that, absent such evidence, reviewing only a limited number of exporters will enable Commerce to reasonably approximate the margins of all known exporters. As the CIT has explained, “[t]he representativeness of the investigated exporters "is the essential characteristic that justifies an ‘all1 others-’ rate based on a weighted average for such" respondents.”
Nat'l Knitwear & Sportswear Ass’n v. United States,
779 F.Supp. 1364, 1373-74, 15 C.I.T. 548, 559 (1991). Thus the government’s repeated argument that “no record evidence demonstrated that the separate rate respondents engaged in pricing behavior similar to Jacobi or CCT,” Br. of United States at 24, is backwards. . The burden is not on the separate respondents to show that their: dumping is the same as that of the individually examined respondents. Rather, Commerce must find based on substantial evidence that there is a reasonable basis for concluding that the separate respondents’ dumping is different.
' There Is ho contention 'here that the expected method is not feasible. The questions therefore are (1) whether Commerce properly determined that the expected method (utilizing the average of the margins calculated for the examined respondents) “would not be reasonably reflective of potential dumping margins” for Cherishmet, Shanxi, and Huahui; and (2) if so, whether Commerce’s chosen method of carrying forward their margins from the previous review was “reasonable.” .
Ill'
Commerce primarily seeks to justify its approach on the ground that there is a policy against using de minimis margins to calculate, separate rates.
See Memorandum,
at 4 (“We agree with. Petitioners that [Commerce] should not diverge from the practice of excluding zero and
de minimist
margins when calculating the separate rate margin.”) (emphasis in original). The government contends that this ,is reasonable because the statute disfavors rising zero or de minimis rates as a general matter. We disagree. ,
It is true that when there are non-de minimis margins assigned to individually examined respondents, the statute instructs ' Commerce to calculate the separate rate by averaging the margins assigned to the individually examined
respondents, “excluding any zero and de minimis margins.” 19 U.S.C. § 1673d(c)(5)(A). But it is equally clear that when all individually examined respondents are assigned de minimis margins, Commerce has no similar mandate to routinely exclude zero or de minimis margins. Congress has spoken directly to this precise situation in §. 1673d(c)(5)(B), and the SAA unambiguously provides that the expected method to calculate the. separate rate in such circumstances is to average the individually examined respondents’ de minimis margins.
See
SAA, at 4201.
The government cannot contend that methodology employing de minimis margins is disfavored when Congress has unmistakably explained that it is, in fact, preferred.
The government’s policy simply cannot be distilled from the statute in this context, and Commerce’s insistence on using its hostility to de minimis rates as the driving force behind its methodology is on its face arbitrary and capricious. Indeed, counsel for the government'admitted at oral argument that Commerce would have used $0.05/kg, which CCT was assigned in the preliminary results of the third review, as the separate rate if one of the two individually examined respondents .had been assigned that rate in the final review.
See
Oral Arg. at 44:30^0. This demonstrates the arbitrariness of Commerce’s approach.
. Commerce also seeks to justify its approach on the ground that Commerce has a legitimate interest in allocating its own limited resources.
While this interest is certainly relevant, it alone is not sufficient to render an otherwise unreasonable methodology reasonable.
Finally, as our cases have explained, accuracy and fairness must be Commerce’s primary objectives in calculating a separate rate for cooperating' exporters.
See Bestpak,
716 F.3d at 1379 (“An overriding purpose of Commerce’s administration of antidumping laws is to calculate dumping margins as accurately as possible.”);
Gallant Ocean (Thailand) Co., v. United States,
602 F.3d 1319, 1323 (Fed.Cir.2010) (a rate must be a “reasonably accurate estimate of the respondent’s actual rate”) (internal quotation marks and citations omitted);
SNR Roulements v. United States,
402 F.3d 1358, 1363 (Fed.Cir.2005) (“Antidumping laws intend to calculate antidumping duties on a fair and equitable basis.”);
Rhone Poulenc, Inc. v. United States,
899 F.2d 1185, 1191 (Fed.Cir.1990), (the “basic purpose of the statute” is to “determin[e] current margins as accurately as possible”);
see also Nan Ya,
810 F.3d at 1344-45 (accuracy represents a “reliable guidepost[] for Commerce’s determinations,” and a determination is “accurate” if it is “supported by substantial evidence”).
IV
We therefore turn to Commerce’s other justifications for its approach. We first address Commerce’s methodology with respect to Cherishmet and Shanxi. Cher-ishmet and Shanxi were not individually examined .in either the second or third reviews. In the second review, they were assigned a separate rate based on the simple average of the dumping margins calculated for the individually examined respondents, Jacobi and Huahui, which resulted in a dumping margin of $0.28/kg. In the third review, Commerce individually examined Jacobi and CCT, ultimately assigning them both de minimis margins. But Commerce continued to apply the $0.28/kg margin to Cherishmet and Shanxi as their separate rate in the third review.
There is no evidence that supports Commerce’s determination that averaging the de minimis margins assigned to the individually examined respondents in the third review would have resulted in margins for Cherishmet and Shanxi that would not have been reflective of their actual dumping margins. Commerce had no data specific to Cherishmet or Shanxi from either the second or third review. And Commerce in fact assumed that the examined respondents were- reasonably representative of Cherishmet and Shanxi during the second review, as Commerce calculated the separate rate by averaging the margins of the individually examined respondents from that review.
Having assumed that the individually examined respondents were reasonably representative' of Cherishmet and Shanxi in the second, review, Commerce lacked any basis to reverse course and conclude that Cherishmet and Shanxi were somehow different in the third review. Commerce did not collect any additional information regarding Cherishmet or Shanxi, nor was there any evidence that would indicate different exporting behavior. In fact, one of the two individually examined exporters, Jacobi, was examined in both the second and third reviews. And in both reviews Commerce selected the individually examined respondents pursuant to the same method of examining the largest volume exporters. In the second review, Jacobi’s rate was used to calculate the rates for Cherishmet and Shanxi, but was found non-representative in the third review. The government offers no explanation as to why Cherishmet and Shanxi were no longer reasonably represented by the individually examined respondents in the third review, while they had been in both reviews prior. Commerce’s conclusion was arbitrary and unsupported by substantial evidence. Accordingly, we affirm the CIT with respect to Cherishmet and Shanxi.
V
We next consider Huahui. Unlike Cherishmet and Shanxi, Commerce did have information specific to Huahui because Huahui was one of the individually examined respondents in the second review. Huahui was assigned an individual dumping margin of $0.44/kg in the second review, significantly higher than the $0.1i/kg margin assigned to Jacobi, the other individually examined respondent during that review, and the $0.00/kg rates assigned to the individually examined respondents in the third review. For this reason, Commerce did have substantial evidence to support its conclusion that simply averaging the de minimis rates assigned to Jacobi and CCT in the third review might not reasonably reflect Hua-hui’s potential dumping margin during the third period; Commerce was therefore entitled to use “other reasonable methods” under the statute. SAA, at 4201; 19 U.S.C. § 1673d(c)(5)(B). The question here is whether Commerce’s- chosen meth
od of carrying forward Huahui’s data from the second period of review to the third was reasonable. We conclude that it was not.
In assessing the reasonableness of Commerce’s methodology, our analysis is guided by the statute’s manifest preference for contemporaneity in periodic administrative reviews. Under the statute, Commerce is obligated to review an anti-dumping duty order if requested “[a]t least once during each 12-month period beginning'on the anniversary of the date of publication” of the order. 19 U.S.C. § 1675(a)(1) (1999) (amended 2016). Commerce must commence a review within six months of any request. 19' U.S.C. § 1675(a)(2)(B)(ii) (1999) (amended 2016). The purpose of periodic administrative reviews is to reassess dumping margins previously calculated in light of data made available during the intervening period since the antidumping order was issued.
See Union Steel v. United States,
713 F.3d 1101, 1103, 1108 (Fed.Cir.2013);
Allegheny Ludlum Corp. v. United States,
346 F.3d 1368, 1373 (Fed.Cir.2003).
Unlike investigations, which consider “overall pricing behavior” to “determine the appropriateness of imposing an anti-dumping duty order” in the first place, administrative reviews’begin with “an anti-dumping duty order already in place,” and typically employ methodology that “permits greater specificity” to “further[]" the transactional accuracy interests” at the core of the review process.
Union Steel,
713 F.3d at 1108. Thus, when it comes to administrative reviews, “it is reasonable for the agency to look for more accuracy [than in the initial investigation], which it achieves in some measure through monthly averaging.”
Id.
There is no basis to simply assume that the underlying facts or calculated dumping margins remain the same from period, to period. “[I]f the facts remained the same from period to ’ period, there would be no need for administrative reviews.”
Shandong Huarong Mach. Co. v. United States,
29 C.I.T. 484, 490-91 (2005). Thus Commerce itself has explained that “it is well established and upheld practice that the Department must base its decisions on the record of the administrative proceeding before it in each review.”
Issues and Decision Memorandum, for the Final Results in the Second Antidumping Duty Order Administrative Review of Diamond Sawblades and Parts Thereof from the Republic of Korea,
78 Fed.Reg. 36,524, cmt. 4 (June 18, 2013).
In short, as we have previously recognized, “there is a clear congressional intent” that administrative reviews.,“.be as accurate
and current
as possible.”
Allegheny,
346 F.3d at 1373 (emphasis, added). The legislative history “emphasized the importance of using current information with respect to making determinations. ‘The Committee
intends that the' Authority and the ITC should
always
use the most up-to-date information available.’ ”
Freeport Minerals Co. (Freeport McMoran, Inc.) v. United States,
776 F.2d 1029, 1032 (Fed.Cir.1985) (quoting H.R.Rep. No. 96-317, at 77 (1979)).
In light of this established'doctrine, it is not open to Commerce to argue that prior review data is reliable simply because it is “temporally proximate,” 1 Br, of United States at 13, 25. The government’s rationale contravenes this fundamental premise of periodic administrative reviews that each “administrative review is a separate exercise of Commerce’s -authority that allows for different conclusions based on different facts in the record.”
Qingdao,
766 F.3d at 1387. That the prior rates were near in time cannot in and of itself justify their use in a subsequent review.
To be sure, there are at least two circumstances where use of data from a prior period may be reasonable. But neither prevails here. First, there are situations where there is evidence that the overall market and the dumping margins have not changed from period to period. Thus in
Atar S.R.L. v. United States,
we upheld Commerce’s use of surrogate data from a prior review period in calculating an exporter’s constructed value profit cap because the record demonstrated that the market had not meaningfully changed between periods. 730 F.3d 1320, 1327 (Fed.Cir.2013). But that is not the case here. This is not a situation in which there was any consistency with respect to the dumping margins of the individually examined respondents throughout the reviews: For both parties that were individually examined in more than one review here, the numbers demonstrate a- significant decline in dumping margins. For example, Jacobi’s margin decreased from 61.95% in the initial investigation to 18.19% in the first review, 'and then again from $0.11/kg in the second review to $0.00/kg in the third. CCT’s margin also significantly decreased from 69.54% in the initial investigation to 14.51% in the first’review, and then again to a, de minimis margin in the third review. There was also a significant decrease in the financial ratios used in calculating the normal values between the second and third reviews, which, all other things being equal, suggests a decline in overall dumping. Thus if anything, the history here is one of generally declining dumping margins, and Commerce had no reason to believe that'Huahui’s margin would hot have similarly declined.
Second, as the government points out, in the Adverse Facts Available (“AFA”) context, where Commerce is allowed to consider deterrence .as a factor,
we have upheld Commerce’s use of .data from a previous administrative review. We have explained that when an exporter is not cooperating, “Commerce is permitted to use a ‘common sense inference that- the highest prior [dumping] margin is . the most probative evidence of current margins.’”
KYD, Inc. v. United States,
607 F.3d 760, 766 (Fed.Cir.2010) (quoting
Rhone Poulenc,
899 F.2d at 1190). In other words, Commerce may presume that
“a prior dumping margin imposed against an exporter in an earlier administrative review continues to be valid if the exporter fails to cooperate in a subsequent administrative review’.”
Id.
at 767;
see also Ta Chen
Stainless
Steel Pipe, Inc. v. United States,
298 F.3d 1330, 1339-40 (Fed.Cir.2002). We have upheld this presumption because “if it were not so, the [exporter], knowing of the rule, would have produced
current
information showing the margins to be less.”
KYD,
607 F.3d at 766 (emphasis in original). Commerce is thus permitted to infer from the exporter’s lack of cooperation that its dumping has not decreased since the previous review. But the current situation is quite different.
Huahui is not a non-cooperating party. To the contrary, Huahui specifically requested leave to be individually examined as, a voluntary respondent under 19 U.S.C. § 1677m(a), or alternatively to submit additional supplementary data, but. Commerce denied both requests. “A presumption used to encourage some companies to submit more accurate information1 may not reasonably be transposed onto companies which are expressly prevented from submitting more accurate information.”
Amanda Foods,
714 F.Supp.2d at 1294.
VI
Nonetheless, the government argues that Commerce’s reliance on’data from-the previous period was reasonable as to Hua-hui for two additional reasons.
First, the government argues that Commerce’s method was reasonable in light of Huahui’s “history of dumping” over the course of the administrative proceedings here. Br. of United States at 24. While evidence of dumping from previous administrative reviews is relevant and may inform Commerce’s methodology, in itself it is not sufficient to demonstrate that Hua-hui’s dumping continued, let alone that it continued at the same rate.
Second, the government argues that it was reasonable to carry over Huahui’s old rate because there were no data on record specific to Huahui in the third review. As we described earlier, Commerce has significant authority to, make administrative decisions regarding the allocation of its own limited resources.
See, e.g., Torrington Co. v. United States,
68 F.3d 1347, 1351 (Fed.Cir.1995). But as we have explained, Commerce may not justify “the absence of evidence by invoking, procedural difficulties that were, at least in part a creature of its own making.”,
Bestpak,
716 F.3d at 1378. It was unreasonable, in this case for Commerce to choose .to limit its review to the two largest volume exporters, refuse to collect additional data from Huahui, and then draw inferences adverse to Huahui based on the lack of data available in the record.
See Albemarle I,
931 F.Supp.2d at 1293 (“The available data pertaining to the [period of review] for the third review were limited by [Commerce’s] decision to individually examine only two mandatory respondents.... Commerce made this determination despite its general statutory obligation to examine all respondents for which a review was requested.”).
While it is clear that 19 U.S.C. § 1677f-1 does not require Commerce to calculate individual margins for every known exporter in all instances, it is equally clear that Commerce has at its disposal broad authority to gather information,
see
19 U.S.C; §§ 1677f-l(a), (b); 1675(a), including from separate respondents. On at least one prior occasion, Commerce has reopened the1 administrative record and collected additional information from separate respondents when all individually examined respondents were assigned de minimis margins:
See Amanda Foods (Vietnam) Ltd. v. United States,
774 F.Supp.2d 1286, 1289-90 (Ct. Int’l Trade 2011). When it did, fit found that the separate respondents, like the individually
examined respondents, were not dumping.
See id.
The availability' of updated information from Huahui and its request to submit such information here suggest that simply applying the old rate in the third, period of review was not reasonable. Commerce had available additional quantity and value data, which would not have required elaborate antidumping calculations, but rather could have served as a basis for Commerce to make approximate comparisons of Hua-hui’s export price.
See, e.g., Amanda Foods,
774 F.Supp.2d at 1289-90, n. 8. What is more, as Commerce acknowledged at argument, Commerce already had at least partial data specific to Huahui’s factors of production on record because Hua-hui was a supplier to Jacobi, and Huahui’s information had already been collected as part of Jacobi’s individual examinations.
See
Oral Arg. at 12:24-12:52. Commerce had at least partial information regarding Huahui’s contemporaneous normal value, and the ability to gather information as to U.S. sales price. . >
To be clear, we are not suggesting that Commerce was required to either individually examine Huahui or assign it a de minimis margin. Rather, what was necessary was some evidence, for example, by a sampling process, that Huahui’s. earlier data continued to be reasonably reflective of its current practice. Far from suggesting that Huahui’s old data would continue to be reasonably reflective, here, the. current record suggests that normal values decreased and dumping margins declined between the second and third reviews generally. It was unreasonable Under these circumstances for Commerce to reassign Huahui in the third review its rate calculated during the second review rather, than take the average of the de minimis margins assigned to Jacobi and CCT.
VII
Accordingly, we áffírm the decision of the CIT' with respect to Cherishmet and Shanxi, and reverse with respect to Hua-hui. We hold that Commerce could not on this record utilize data from the previous review. Rather, Commerce, having declined to collect additional information, was required to follow the “expected method” of utilizing the de minimis margins of the individually examined respondents from the contemporaneous period. The case is remanded to the CIT so that it may issue appropriate instructions to Commerce.
AFFIRMED-IN-PART, REVERSED-IN-PART, AND REMANDED-IN-PART
Costs
Costs to Cherishmet, Beijing Pacific Activated Carbon Products, Shanxi, Albe-marle, and Huahui.