OPINION and ORDER
This action is again before the court following a second redetermination and a voluntary partial third redetermination. In the third redetermination, the Depart.ment of Commerce (“Commerce”) reaffirmed the second redetermination of the final results of the antidumping (“AD”) duty investigation of multilayered wood flooring from the People’s Republic of China (“PRC” or'“China”).1
*1380Still at issue are the AD duty rates assigned to eight separate rate respondents—the Plaintiffs and Plaintiff-Interve-nors here (collectively, “Plaintiffs”)2—for the underlying AD duty investigation. Specifically, Plaintiffs challenge Commerce’s decision to assign seven of them an unspecified, non-de minimis AD duty rate for the investigation, to provide for liquidation of their entries at the rates established for them in the first administrative review3 (as limited by the provisional measures deposit cap), and to initiate a full investigation of the remaining eighth Plaintiff, Changzhou Hawd Flooring Co. (“Changzhou Hawd”), as it has certified no shipment of subject merchandise in the first administrative review and therefore otherwise lacks any relevant calculated rate. The court has jurisdiction pursuant to § 516A(a)(2)(B)(i) of the Tariff Act of 1980, as amended, 19 U.S.C. § 1516a(a)(2)(B)(i) and 28 U.S.C. § 1581(c) (2012).4
As explained below, Commerce’s determination regarding the group of seven Plaintiffs is based on a reasonable reading of the law and record evidence. However, the agency’s decision to conduct, at this late date, a full investigation of Changzhou Hawd is arbitrary and capricious. Therefore, the court remands again for further consideration in accordance with this opinion.
BACKGROUND
Litigation of the separate rate5 has so far produced two court opinions,6 two corresponding ' redeterminations by Com-*1381meree,7 and, most recently, a voluntary remand and redetermination by Commerce.8
In each successive determination, Commerce has established the separate rate in a different way. In the Final Determination, having individually investigated three fully cooperative mandatory respondents,9 Commerce loosely followed 19 U.S.C. § 1673d(c)(5)(A) and took a simple average10 of the two non-de minimis mandatory respondent rates (resulting in a separate rate of 3.31 percent). Final Determination, 76 Fed. Reg. at 64,321-22. Plaintiffs challenged the determination. Compl., ECF No. 9 at ¶ 3. It was ultimately remanded on other grounds. Baroque Timber, — CIT -, 925 F.Supp.2d 1332.
In the First Redetermination, changes to the underlying surrogate values and calculation methodology resulted in all three mandatory respondents receiving AD duty rates of zero. First Redetermination, Consol. Ct. No. 12-00007, ECF No. 132, at 2, 52. Because of this, Commerce recalculated the separate rate under 19 U.S.C. § 1673d(c)(5)(B), and decided that “any reasonable method” included a simple average of the three zero mandatory rates and a rate based on adverse facts available (“AFA”).11 This resulted in a higher separate rate of 6.41 percent. First Redetermination, Consol. Ct. No. *138212-00007, ECF No. 132, at 27. The court found that this method, while not per se unreasonable, was unsupported by substantial evidence, because Commerce had failed to articulate a rational connection between Plaintiffs’ economic reality and the use of the AFA rate in the calculation of their rate. Baroque Timber, — CIT -, 971 F.Supp.2d at 1344-45. The court accordingly remanded to Commerce for a redetermination of the separate rate. Id. at 1346.
Between the second remand and the corresponding redetermination, Commerce issued the final determination in the first administrative review following the investigation at issue here. Final Review, 79 Fed. Reg. 26,712. Because of this, in the Second Redetermination, rather than recalculate the separate rate for all separate rate respondents, Commerce inferred that, because there were 110 non-cooperative respondents in the investigation, see Part IIA, infra, the appropriate separate rate for the investigation was more than de minimis. It then assigned seven of the Plaintiffs12 the rate calculated for them in the first administrative review (as limited by the provisional measures deposit cap). Second Redetermination, ECF No. 52, at 6-8. The remaining eighth Plaintiff, Changzhou Hawd, having certified no shipments, did not have a calculated rate for the first review. Commerce concluded that it did not have enough data on the record to calculate a rate reflective of that company’s economic reality and initiated an individual investigation of this eighth respondent. Id. at 8-9.13
The Second Redetermination was challenged in extensive briefing before the court,14 and, at the court’s suggestion, see Telephone Conf., ECF No. 79, Commerce requested a partial voluntary remand “to determine whether it should conduct a limited investigation of the eight separate rate [PJlaintiffs,” rather than a full investi*1383gation of just Changzhou Hawd. Mot. for Voluntary Remand, ECF No. 92 at 1 (quotation marks omitted). The court granted the voluntary remand. Changzhou Hawd, — CIT at-, 6 F.Supp.3d at 1362. It was ultimately a futile exercise. Commerce essentially decided that it was impossible to take an approach that was both measured and fact-based, and reaffirmed its results and reasoning in the Second Redetermination. See Third Redetermi-nation, ECF No. 107, at 17.
STANDARD OF REVIEW
The court will sustain Commerce’s determinations unless they are “unsupported by substantial evidence on the record, or otherwise not in accordance with law.” 19 U.S.C. § 1516a(b)(l)(B)(i). The court will set aside agency actions found to be arbitrary and capricious. Changzhou Wujin Fine Chem. Factory Co., Ltd. v. United States, 701 F.3d 1367, 1377 (Fed. Cir.2012) (citing Bowman Transp., Inc. v. Arkansas-Best Freight Sys., Inc., 419 U.S. 281, 284, 95 S.Ct. 438, 42 L.Ed.2d 447 (1974)).
DISCUSSION
I. Commerce’s Methodology
Commerce generally follows 19 U.S.C. § 1673d(c)(5) to establish the separate rate. Second Redetermination, ECF No. 52, at 3; Yangzhou Bestpak Gifts & Crafts Co. v. United States, 716 F.3d 1370, 1374 (Fed.Cir.2013). Thereunder, the general rule sets the separate rate as 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 [based entirely on facts otherwise available].” 19 U.S.C. § 1673d(c)(5)(A). The exception to this rule, which applies only when all individually investigated rates are zero, de minimis, or based entirely on facts otherwise available, allows Commerce to use “any reasonable method to establish the estimated [separate rate] for exporters and producers not individually investigated.” Id. at § 1673d(c)(5)(B). “Any reasonable method” is expected to mean the average of the rates calculated for individually investigated respondents. 19 U.S.C. § 1673d(c)(5)(B); Uruguay Round Agreements Act, Statement of Administrative Action (“SAA”), HR. Doc. No. 103-316 (1994) at 873, reprinted in 1994 U.S.C.C.A.N. 4040, 4201.15 However, “if [the expected] 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 873, reprinted in 1994 U.S.C.C.A.N. at 4201.
Here, all three individually investigated respondents had AD duty rates of zero. Second Redetermination, ECF No.- 52, at 3. Commerce accordingly established the separate rate under the exception—using “any reasonable method”—rather than the rule. Id.; 19 U.S.C. § 1673d(c)(5)(B). Commerce elected not to use the expected method, or even to calculate a specific separate rate for the investigation. Rather, the agency went no further than inferring that the separate rate, on the record evidence, must be more than de minimis. Second Redetermination, ECF No. 52, at 4-7. Plaintiffs argue that- this is not in accordance with law, contending Commerce must calculate a de minimis separate rate for the investigation.16
*1384The AD statute does not speak directly to the question at issue;17 it only requires “any reasonable method to establish” the separate rate. See 19 U.S.C. § 1673d(c)(5)(B). “Any reasonable method” is a “lenient standard,” Bestpak, 716 F.3d at 1378, and “establish” is a broader term than “calculate.” The court must “leave the discretion provided by the ambiguities of [the AD] statute with the implementing agency,” Fine Furniture (Shanghai) Ltd. v. United States, 748 F.3d 1365, 1369 (Fed.Cir.2014) (quoting United States v. Eurodif S.A., 555 U.S. 305, 316, 129 S.Ct. 878, 172 L.Ed.2d 679 (2009)), even where “the court might have preferred” a different interpretation, Koyo Seiko, 36 F.3d at 1570 (citation omitted). The broad language of the statute allows Commerce to tailor its method to the record evidence before it. See Eurodif, 555 U.S. at 317-18, 129 S.Ct. 878 (“[In reading regulatory statutes] form should be disregarded for substance and the emphasis should be on economic reality.” (quotation marks and citation omitted)).
Here, Commerce’s decision to infer a more' than de minimis but otherwise unspecified separate rate for the investigation, using instead the cash deposit rates from the first administrative review, as limited by the provisional measures deposit cap, is within a reasonable construction of the statute.18 That “any reasonable method” is available to Commerce, not just the expected method, indicates the statute contemplates the possibility of a more than de minimis separate rate even where, as here, all individually investigated rates are zero. See 19 U.S.C. § 1673d(c)(5)(B). Further, while in most circumstances Commerce would need a specific separate rate for the investigation, so that an AD duty can be assessed (or not) with publication of an AD duty order, see 19 U.S.C. § 1673e, that is not the case here. Because Commerce has already calculated rates for some Plaintiffs from the first administrative review (based on their actual sales experience, not the assortment other companies’ de minimis and AFA rates otherwise available in this investigation), and these rates will apply to the period at issue regardless, see 19 U.S.C. § 1675(a)(2)(C), Commerce has established rates “reasonably reflective of potential dumping margins” for the separate rate respondents. See SAA at 873, reprinted in 1994 U.S.C.C.A.N. at 4201. According*1385ly, Commerce’s method is not per se unreasonable.
II. Commerce’s Methodology in the Context of the Record
A. Commerce’s Inference that the Separate Rate is More Than De Min-imis
In the investigation, 110 companies did not respond to Commerce’s Q & V questionnaire. Second Redetermination, ECF No. 52, at 4. Commerce assumes that, when a company so completely fails to participate, it has made “a knowing and rational decision” not to respond “based on which choice will result in the lower rate.” Id. at 5 (citations omitted). Commerce is permitted to make this assumption, see Ta Chen Stainless Steel Pipe, Inc. v. United States, 298 F.3d 1330, 1339 (Fed.Cir.2002) (citing Rhone Poulenc, Inc. v. United States, 899 F.2d 1185, 1190 (Fed.Cir. 1990)),19 and Plaintiffs have not offered evidence sufficient to suggest that Commerce is wrong in doing so here.20
This rational actor assumption is the core of the well-worn presumption that allows Commerce to use AFA against non-cooperating respondents, see 19 U.S.C. § 1677e(b), thereby shifting the burden of production21 and incentivizing future cooperation.22 Similar but distinct, this same rational actor assumption allows Commerce to infer23 from companies’ non-co*1386operation that “[their] dumping margins during the period of investigation were not zero or de minimis, and that, if [Commerce] had received complete information, [it] may have chosen one of these companies as a mandatory respondent.” Second Redetermination, ECF No. 52, at 4 (footnote omitted). That is, the 110 non-cooperating respondents would have participated if their rates were zero or de minimis, and the gap in the evidentiary record their non-cooperation creates reflects on the separate rate respondents only insofar as it conceals data that would have applied in the calculation of the separate rate. Where, as here, all individually investigated respondents have received a zero rate (or de minimis rate, or AFA rate), this gap is effectively dispositive: “if the 110 companies had chosen to cooperate,” and one had been selected as a mandatory respondent, “the examined company’s rate would have been above de minimis” but below AFA, and, pursuant to 19 U.S.C. § 1673d(c)(5)(A), “would have been assigned to the separate rate plaintiffs as [the] separate rate in the Final Determination.” See Second Redetermination, ECF No. 52, at 6.
Commerce corroborates its inference of a more than de minimis separate rate for the investigation with citation to the results of the first administrative review. Second Redetermination, ECF No. 52, at 7, 30. There, Commerce individually investigated three respondents, including Plaintiffs Fine Furniture and Armstrong, and ultimately found a more than de minimis rate for Fine Furniture and a zero rate for Armstrong.24 Commerce views this as confirmation that dumping occurred during the period of investigation: if dumping occurred during the review, under the discipline of an AD order, it is likely to have also occurred here, without the discipline of an AD order to disincentivize such pricing behavior.25 While it is true that “each administrative review is a separate segment of proceedings with its own unique facts,” Peer Bearing Co.-Changshan v. United States, 32 *1387CIT 1307, 1310, 587 F.Supp.2d 1319, 1325 (2008)(quotation marks and citation omitted), and that Commerce cannot consider AD duty rates from other reviews when those rates bear “no rational relationship to any pricing behavior during the [period of review] or to the likely pricing behavior of the recipients of the margin,” Albemarle Corp. v. United States, — CIT-, 931 F.Supp.2d 1280, 1292 (2013),26 this does not undermine Commerce’s determination here. Commerce references the first review results as corroboration, not for calculation. The first review serves to confirm that the separate rate respondents’ economic reality is more varied and complicated than the mandatory respondent de minimis rates here suggest. It confirms that the separate rate respondents merit the closer consideration that keeping them subject to the order affords, some receiving de minimis rates and others not. The individually investigated rates for two Plaintiffs, one of which, as the only non-de minimis rate, defines the separate rate for five other of the Plaintiffs, bear a rational relationship to the pricing behavior of the recipients of the margin. As the rates at which the entries at issue will be liquidated (as limited by the provisional measures deposit cap), they are also reasonably related to the time period at issue.
Because “the question here is whether the evidence and reasonable inferences from the record support [Commerce’s] finding,” Matsushita Electric Indus. Co. v. United States, 750 F.2d 927, 933 (Fed.Cir. 1984), “not whether some other inference could reasonably have been drawn,” Daewoo Elecs. Co. v. Int’l Union of Elec., Elec., Technical, Salaried & Mach. Workers, AFL-CIO, 6 F.3d 1511, 1520 (Fed.Cir. 1993), Commerce’s determination holds. Commerce’s conclusion that—based on the silence of 110 respondents, the resultant gap in the record, and the mixed results of the first administrative review—the separate rate (and thus Plaintiffs’ rate) in this investigation is somewhat more than de minimis and less than AFA, while not the only possible inference, is a reasonable inference from the record, and therefore supported by substantial evidence. See Consolo v. Fed. Mar. Comm’n, 383 U.S. 607, 620, 86 S.Ct. 1018, 16 L.Ed.2d 131 (1966).
B. Commerce’s Refusal to Calculate a Specific Separate Rate
Having reasonably inferred that the separate rate for the period of investigation is more than de minimis, Commerce declined to calculate a specific (higher than de minimis) rate for seven of the eight Plaintiffs. Second Redetermination, ECF No. 52, at 7. The agency concluded that “[w]hile it is normally necessary to assign a specific rate to separate rate respondents ... in this instance, it would be an unnecessary use of administrative and judicial resources” because specific rates would be without consequence and without use. Id. at 7-8.
Commerce is correct that further precision would be without consequence. In an AD investigation, Commerce calculates dumping margins for respondents and imposes an AD order based on those margins. Union Steel v. United States, 713 F.3d 1101, 1103 (Fed.Cir.2013) (citing 19 U.S.C. §§ 1673a, 1673b(b), 1673b(d), 1673d(a), 1673d(c)). Respondents with de minimis or zero margins are excluded from the order (and therefore subsequent *1388administrative reviews). See 19 U.S.C. §§ 1673b(b)(3), 1673d(a)(4). This exclusion is not premised on a specific rate, but rather whether the rate is de minimis or not. Having reasonably inferred that the separate rate is more than de minimis, Commerce has made the determination necessary to impose the AD order on Plaintiffs.
Commerce is also correct that a specific rate for the seven Plaintiffs would be without use. This is because “the rate[s] determined in the first administrative review supersede) ] the cash deposit rate established in the final determination of the investigation.” Second Redetermination, ECF No. 52, at 7 (citing 19 U.S.C. § 1675(a)(2)(C)). Plaintiffs’ entries would have been and will be liquidated at the rates established in the first administrative review (as limited by the provisional measures deposit cap)27 regardless of whatever non-de minimis rate might be assigned to them in the investigation.
Further, contrary to Plaintiffs’ arguments,28 any rate calculated pursuant to this litigation would not affect the provisional measures deposit cap. The provisional measures deposit cap ensures that, for the interstitial period of the investigation—after the preliminary determination but prior to the issuance of an AD order— importers are not liable for more than the rate set for them at the time of entry. 19 U.S.C. § 1673f(a); 19 C.F.R. § 351.212(d) (2014). If the AD duty rate set in the first administrative review (or subsequent litigation) is less, the difference between it and the cash deposit, bond, or other security provided at entry, is refunded. If the AD duty rate is ultimately more, then the difference is not owed. Thai Pineapple Canning Indus. Corp. v. United States, 273 F.3d 1077, 1086 (Fed.Cir.2001). Because the cap “limits the rate based on the deposited amount, not an amount that a final determination indicates should have been deposited,” Universal Polybag Co. v. United States, 32 CIT 904, 925, 577 F.Supp.2d 1284, 1303 (2008), this action and its results do not affect the cap. Rather, the cap is set by the amount collected, “[not] the amount that, should have been collected.” Id., 32 CIT at 925, 577 F.Supp.2d at 1303-04; accord Yantai Oriental Juice Co. v. United States, 26 CIT 605, 623, 2002 WL 1347018 (2002) (not reported in the Federal Supplement) (“[The cap] merely directs how the deposit rate should be used, not how it should be calculated.”).
Accordingly, as a specific rate for the seven Plaintiffs would be without use and without effect, in the interest of administrative and judicial economy,29 it was reasonable for Commerce to decline to calculate a more specific rate for this investigation.
C. Commerce’s Decision to Individually Investigate Changzhou Hawd
Having inferred that the separate rate for the investigation is more than de min-imis, but declining to calculate a specific *1389separate rate in favor of rates from the first administrative review, Commerce determined it was necessary to conduct an individual investigation of the one Plaintiff that did not receive a rate in the first administrative review, Changzhou Hawd. Changzhou Hawd has certified no shipments of subject merchandise for the period of the first administrative review, and therefore has no calculated rate for that period. Final Review, 79 Fed. Reg. at 26,713. Commerce believes that “with the very limited information currently on the record, [it] is unable to calculate a dumping rate based on Changzhou Hawd’s own economic reality” without a full individual investigation. Second Redetermination, ECF No. 52, at 8-9; see also Third Rede-termination, ECF No. 107, at 17 (concluding that anything short of a full investigation would not be practically or legally feasible). Plaintiffs challenge this determination as arbitrary and capricious.30
While the decision to reopen the record is generally within the agency’s discretion, see Essar Steel, 678 F.3d at 1277-78, that discretion cannot be exercised in a manner that is arbitrary and capricious. See Changzhou Wujin Fine Chem., 701 F.3d at 1377 (citing Bowman Transp., 419 U.S. at 284, 95 S.Ct. 438). Arbitrary and capricious is a “narrow” standard of review, but still “searching and careful.” Citizens to Pres. Overton Park, Inc. v. Volpe, 401 U.S. 402, 416, 91 S.Ct. 814, 28 L.Ed.2d 136 (1971). Commerce must “articulate a rational connection between the facts found and the choice made.” Bowman, 419 U.S. at 285, 95 S.Ct. 438 (quotation marks and citation omitted). The agency’s decision cannot have “relied on factors [that] Congress has not intended it to consider, entirely failed to consider an important aspect of the problem, offered an explanation for its decision that runs counter to the evidence before [it], or [be] so implausible that it could not be ascribed to a difference in view or the product of agency expertise.” Motor Vehicle Mfrs. Ass’n of U.S., Inc. v. State Farm Mut. Auto. Ins. Co., 463 U.S. 29, 43, 103 S.Ct. 2856, 77 L.Ed.2d 443 (1983).
Here, Commerce has decided to conduct an individual investigation of a single separate rate respondent in the third iteration of a much-contested AD determination. Second Redetermination, ECF No. 52, at 8-9, 36-37. This, despite Commerce’s emphatic claims of limited administrative resources. Final Determination I & D Mem., cmt. 43 at 110 (“[T]he Department lack[s] the resources required to examine more than three respondents in this investigation.”); Second Redetermination, ECF No. 52, at 7-8 (declining to calculate a specific separate rate because of “limited administrative resources”). Moreover, Commerce has repeatedly declined to conduct an individual investigation of another Plaintiff in this investigation, would-be voluntary respondent Fine Furniture, citing lack of resources.31 Final Determination I & D Mem., cmt. 43 at 110-112; First Redetermination, Consol. Ct. No. 12-00007, ECF No. 132, at 49; Second Redetermination, ECF No. 52, at 3740; Third Redetermination, ECF No. 107, at 10-11.
*1390Commerce cannot have it both ways. It is well-established that “[a]n agency action is arbitrary when the agency offers insufficient reasons for treating similar situations differently.” SKF USA Inc. v. United States, 263 F.3d 1369, 1382 (Fed. Cir.2001) (alteration, quotation marks and citation omitted). An agency “must cogently explain why it has exercised its discretion in a given manner.” State Farm, 463 U.S. at 48, 103 S.Ct. 2856 (citations omitted). Internal inconsistency and self-contradiction do not satisfy this requirement.
Commerce asserts that because the current record has only “very limited information” on Changzhou Hawd (specifically, only “aggregate [Q & V] data and Changzhou Hawd’s separate rate application”), Commerce “is unable to calculate a dumping rate based on Changzhou Hawd’s own economic reality” without a full investigation. Second Redetermination, ECF No. 52, at 9. While Commerce is correct that a separate rate respondent’s AD duty rate must be reasonably related to its economic reality, Bestpak, 716 F.3d at 1380, that cannot reasonably be said to necessitate a full individual investigation in every instance. If Commerce can, indeed must, tie an AFA rate to the recipient’s actual dumping margin,32 where, by definition, Commerce cannot conduct a meaningful, let alone full, investigation to establish a rate,33 it cannot be impossible to do the same for a fully cooperative separate rate respondent with the record evidence present here and Commerce’s continued ability to reasonably reopen the record.34 Cf. Amanda Foods (Vietnam) Ltd. v. United States, — CIT-, 774 F.Supp.2d 1286 (2011).
Commerce also believes that it is statutorily impossible for it to do anything less than a full investigation. Third Redeter-mination, ECF No. 107, at 8-9, 17. But this does not comport with the plain language of the applicable statute. Commerce is only obliged to use “any reasonable method” to calculate a separate rate. 19 U.S.C. § 1673d(c)(5)(B). Commerce’s internally inconsistent rationalization is “so implausible that it could not be ascribed to a difference in view or the product of agency expertise.” State Farm, 463 U.S. at 43, 103 S.Ct. 2856.
Commerce now has both an investigation and first administrative review, each with three fully cooperative individually investigated respondents. Second Rede-termination, ECF No. 52, at 3-4; Final Review, 79 Fed. Reg. at 26,713. It has denied multiple voluntary respondent applications, Preliminary Determination, 76 Fed. Reg. at 30,658; Final Determination I & D Mem., cmt. 43 at 110, but still has an evidentiary record much more robust than would be available in a typical investigation. In this context, while Commerce retains the discretion to reasonably reopen the record, its decision to conduct a full individual investigation of Changzhou Hawd at such a late date is arbitrary and capricious.
*1391CONCLUSION
While it is reasonable on this record for Commerce to infer that the separate rate is more than de minimis, and to decline to calculate a specific rate in favor of those already calculated for the first administrative review, it is arbitrary and capricious for Commerce to now launch an individual investigation of Changzhou Hawd.
Accordingly, this matter is affirmed in part and remanded in part to Commerce for further consideration in accordance with this opinion. Commerce shall have until March 24, 2015 to complete and file its remand redetermination. Plaintiffs shall have until April 7, 2015 to file comments. Defendant and Defendant-Inter-venor shall have until April 17, 2015 to file any reply.
IT IS SO ORDERED.