QWEST CORPORATION, Appellant, v. Anne C. BOYLE, in her official capacity as Commissioner of the Nebraska Public Service Commission; Gerald L. Vap, in his official capacity as Commissioner of the Nebraska Public Service Commission, Rod Johnson, in his official capacity as Commissioner of the Nebraska Public Service Commission; Frank E. Landis, Jr., in his official capacity as Commissioner of the Nebraska Public Service Commission; Timothy Schram, in his official capacity as Commissioner of the Nebraska Public Service Commission; Nebraska Public Service Commission, Appellees, 589 F.3d 985

GRUENDER, Circuit Judge. Qwest Corporation challenges an order issued by the Nebraska Public Service Commission setting the rates that competitors must pay to lease elements of Qwest’s local telephone network in Nebraska. The district court, applying a deferential standard of review, affirmed the Commission’s order. For the following reasons, we affirm in part and remand to the district court with instructions to remand to the Commission for further proceedings concerning how Qwest will implement the order. 1.

BACKGROUND The Telecommunications Act of 1996 requires established local telephone companies such as Qwest to lease elements of their networks to rival companies seeking to enter a local market.

The established companies are conventionally called “ILECs,” which stands for “incumbent local exchange carriers.” The ILECs’ rivals are called “CLECs,” which stands for “competitive local exchange carriers.” And the network elements that ILECs are required to lease to CLECs are called “UNEs, which stands for ‘unbundled network elements’.” The Act allows ILECs to negotiate and contract with CLECs regarding the rates for leasing UNEs. See 47 U.S.C. § 252(a). In the event that carriers cannot reach an agreement, the Act authorizes state public utilities commissions to set rates.

See id. § 252(b); see also Implementation of the Local Competition Provisions in the Telecommunications Act of 1996, 11 F.C.C.R. 15499, ¶¶ 620, 693 (Aug. 8, 1996). State commissions must, in turn, follow implementing regulations issued by the Federal Communications Commission. See MPower Commc’ns Corp. v. Ill. Bell Tel. Co., 457 F.3d 625, 627 (7th Cir.2006). Two FCC regulations are particularly relevant here. The first regulation provides that rates for UNEs must be based on a standard known as “total element long-run incremental cost,” or “TELRIC.” 47 C.F.R. § 51.505(a); see also Verizon Commc’ns Inc. v.

FCC, 535 U.S. 467, 122 S.Ct. 1646, 152 L.Ed.2d 701 (2002) (holding that the FCC has authority under the Act to require state commissions to set rates based on TELRIC). Under TEL-RIC, rates “are calculated according to what it would cost today to build and operate an efficient network that can provide the same services as the ILEC’s existing network.” Qwest Corp. v. Koppendrayer, 436 F.3d 859, 863 (8th Cir.2006).

In other words, “TELRIC obliges ... state regulators to set prices based on the long-run costs that would be incurred to produce the services in question using the most-efficient telecommunications technology now available, and the most efficient network configuration.” AT & T Commc’ns of Ill., Inc. v. Ill. Bell Tel. Co., 349 F.3d 402, 405 (7th Cir.2003). TEL-RIC thus differs from “old-style rate regulation,” in which a commission simply determined “how much capital a utility ha[d] reasonably invested in its plant and then set[ ] the reasonable rate of return on that investment.” MPower Commc’ns, 457 F.3d at 629.

As other courts have noted, TELRIC is a very flexible standard. See, e.g., id. at 630 (“Because the endeavor is hypothetical and prospective, it is impossible to find ‘right’ answers; there are only better and worse estimates.”); AT & T Commc’ns of Ill., 349 F.3d at 405 (“TELRIC is a framework rather than a formula; there is considerable play in the joints.”); AT & T Corp. v. FCC, 220 F.3d 607, 616 (D.C.Cir.2000) (“[Ejnormous flexibility is built into TELRIC.”).

The second regulation requires state commissions to “establish different rates for elements in at least three defined geographic areas within the state to reflect geographic cost differences.” 47 C.F.R. § 51.507(f). The process of establishing different rates for UNEs in different areas is called “geographic deaveraging.” According to the FCC, rates must be deaver-aged to “more closely reflect the actual costs of providing ... unbundled elements.” Implementation of the Local Competition Provisions, 11 F.C.C.R. 15499, ¶ 764.

In 2002, the Nebraska Public Service Commission set the leasing rates for Qwest’s “local loops,” a term of art referring to the “last mile” of copper wire or fiber-optic cable that connects customers to the local network. In Docket C-2516, the Commission used three economic cost studies to determine TELRIC for Qwest’s loops. The Commission then used a statistical technique called cluster analysis to place “wire centers” (the connecting point for local loops and the carrier’s central office) with similar cost characteristics into three geographically-deaveraged zones.

The Commission’s application of this methodology resulted in the following monthly leasing rates for individual loops in each zone: Zone 1, $15.14; Zone 2, $35.05; and Zone 3, $77.92. Qwest later moved to reduce these rates by 20 percent. The Commission granted Qwest’s motion, resulting in monthly rates of $12.14, $28.11, and $62.49. The parties agree that these rates complied with the TELRIC standard.

In addition to its rate-setting responsibilities, the Commission is responsible for administering Nebraska’s “universal service fund,” which is intended to “ensure[ ] that all Nebraskans, without regard to their location, have comparable accessibility to telecommunications services at affordable prices.” Neb.Rev.Stat. § 86-317. The Commission supports universal access by allocating subsidies to service providers. In 2004, the Commission adopted a new method of allocating these subsidies.

The so-called “long-term universal service funding mechanism” that the Commission adopted in Docket NUSF-26 was designed to target subsidies to rural, out-of-town areas where the costs of providing service are highest. In particular, the Commission developed a model using econometric regression (another form of statistical analysis) to calculate subsidies based on the relationship between household density and “forward-looking costs.” The Commission also ordered certain subsidies for leased UNEs to be “ported” (ie., transferred) from the ILEC to a CLEC.

The Commission adjusted the size of portable subsidies linked to loops providing service to residential customers and eliminated portable subsidies linked to many loops providing service to business customers. While this proceeding did not affect the leasing rates for Qwest’s loops, the new method of allocating subsidies led to a competitive imbalance in Zone 2 and Zone 3. In 2006, the Commission opened an investigation into whether the three geo-graphieally-deaveraged zones established in Docket C-2516 remained appropriate in light of the funding mechanism announced in Docket NUSF-26.

In Docket C-3554, the Commission’s staff proposed adding in-town and out-of-town designations to the existing zones. According to the staff, further deaveraging leasing rates would account for geographic cost differences “on a more targeted basis.” Furthermore, the staff suggested that adding in-town and out-of-town designations would restore competitive neutrality by harmonizing leasing rates with portable subsidies. The Commission adopted the staffs proposal to deaverage leasing rates into four zones: Zone 1 (out-of-town); Zone 2 (out-of-town); Zone 3 (out-of-town); and a zone encompassing in-town areas.

The Commission defined in-town areas as “cities, villages or unincorporated areas with 20 or more households and densities greater than 42 households per square mile” and defined out-of-town areas as the “remaining areas” in each zone “that have not been assigned to a town.” The Commission also adopted the staffs proposed deaveraging method, which had four basic steps. First, the existing leasing rates were multiplied by the number of loops in each zone to determine the “total cost” in each zone.

Second, the percentage of “expected cost,” calculated using the econometric regression model from Docket NUSF-26, was allocated to in-town and out-of-town areas. Third, the total cost in each zone determined in step one was allocated to in-town and out-of-town areas according to the percentages determined in step two. Fourth, the aggregate costs for in-town and out-of-town areas were divided by the number of in-town and out-of-town loops to determine per loop costs for in-town and out-of-town areas in each zone.

The Commission’s application of this methodology resulted in the following monthly leasing rates for loops in each zone: In-town, $10.76; Zone 1 (out-of-town), $37.04; Zone 2 (out-of-town), $95.32; and Zone 3 (out-of-town), $210.90. The Commission ordered Qwest to implement its order by identifying in-town and out-of-town customers (as proxies for in-town and out-of-town loops) using the distribution model developed in Docket NUSF-26.

Qwest moved for reconsideration and rehearing, asserting that the distribution model did not contain the data needed to classify many of its customers’ addresses as in-town or out-of-town. Qwest requested permission to use municipal tax records, which were already integrated into Qwest’s billing system, instead of the distribution model. Qwest described this alternative method of implementing the Commission’s order as “a reasonable approximation” of the Commission’s definition of in-town and out-of-town areas. The Commission denied Qwest’s motion because it was untimely.

Qwest then moved for clarification, modification, or waiver of the Commission’s order, renewing its request for permission to use municipal tax records instead of the distribution model. The Commission granted Qwest’s request for clarification, stating “that Qwest may use the municipal tax information in its existing billing systems in the implementation of the four zone methodology.” Qwest filed this action under 47 U.S.C. § 252(e)(6), which authorizes any party aggrieved by the rates set by a state commission to seek judicial review in federal court.

The district court found that Qwest “[made] a good case for setting aside the rate[s]” set by the Commission in Docket C-3554. Qwest Commc’ns Corp. v. Vap, No. 8:07-cv-430, 2008 WL 4866351, at *1 (D.Neb. Nov.7, 2008) (unpublished). Specifically, the court noted that it was “doubtful” whether the Commission’s deaveraging method resulted in TELRIC-based rates. Id. at *9. Nevertheless, the court construed Qwest’s challenge as raising a mixed question of law and fact, subject to review under the arbitrary and capricious standard. Id.

The court held that the Commission’s order did not violate the Act or applicable FCC regulations and was not arbitrary and capricious. Id. at *9-11. Accordingly, the court concluded that “[e]ven though it is ... a close question, I will defer to the [Commission’s] judgment, confident that it will take prompt corrective action in the event the marketplace discloses that there is a[n] imbalance between rates and costs in any zone.” Id. at *9. Qwest appeals. II. DISCUSSION Qwest makes five principal arguments for overturning the Commission’s order.

Although there is some overlap between these arguments, we find it convenient to address them in turn. At the outset, however, we will recite the general principles governing our standard of review. We review the district court’s decision de novo, “applying the same standards as the district court.” Ace Tel. Ass’n v. Koppendrayer, 432 F.3d 876, 878 (8th Cir.2005). Thus, we review a state public utilities commission’s

interpretation of federal law de novo. Id. (citing Qwest Corp. v. Minn. Pub. Utils. Comm’n, 427 F.3d 1061, 1064 (8th Cir.2005)). Recognizing the state commission’s “superior technical expertise,” however, “we review its factual determinations under the arbitrary and capricious standard.” Id. (citing Qwest Corp. v. Minn. Pub. Utils. Comm’n, 427 F.3d at 1064). The scope of review under the arbitrary and capricious standard is “narrow”; the standard does not permit “a court ... to substitute its judgment for that of the agency.” Connect Commc’ns Corp. v. Sw.

Bell Tel., L.P., 467 F.3d 703, 711 (8th Cir.2006) (quoting In re Core Commc’ns, Inc., 455 F.3d 267, 277 (D.C.Cir.2006)). We have applied the arbitrary and capricious standard in reviewing commissions’ decisions concerning “rate-setting issues.” See, e.g., WWC License, L.L.C. v. Boyle, 459 F.3d 880, 894 (8th Cir.2006) (citing Qwest Corp. v. Koppendrayer, 436 F.3d 859, 863 (8th Cir.2006)). For instance, the court in WWC License applied the arbitrary and capricious standard in reviewing a challenge to a commission’s cost estimates. Id. Likewise, the court in Qwest Corp. v.

Koppendrayer applied the arbitrary and capricious standard in reviewing challenges to a commission’s “adoption of ... cost models.” 436 F.3d at 871. These cases appropriately focused on concrete rate-setting issues rather than abstract methodological principles. See AT & T Commc’ns of Ill., Inc. v. Ill. Bell Tel.

Co., 349 F.3d 402, 408 (7th Cir.2003) (“Congress provided for federal judicial review of rates set by state commissions; it did not provide for review of individual factors that influence those rates.”)• Given the fact-intensive nature of rate-setting, applying a deferential standard of review makes good sense. See WWC License, 459 F.3d at 894. On the other hand, we have said that we review a commission’s decision “for compliance with federal law de novo.” See, e.g., Connect Commc’ns, 467 F.3d at 708 (citing Qwest Corp. v. Minn. Pub. Utils. Comm’n, 427 F.3d at 1064).

The meaning of “compliance” in this context is not entirely clear. Presumably, our statements concerning review for compliance with federal law refer to purely legal questions rather than the mixed questions that we have elsewhere described as “rate-setting issues.” For instance, the court in WWC License applied the de novo standard in reviewing a dispute about an ILEC’s statutory duties, noting that “resolution of the ... issue requires only

interpretation of the Act.” 459 F.3d at 890. As we mentioned above, the court went on to apply the arbitrary and capricious standard in reviewing a challenge to the commission’s cost estimates! Id. at 894. The challenged cost estimates in WWC License were based on the commission’s application of federal pricing standards. Id. at 893-94. Thus, we can state with confidence that not every challenge to a commission’s application of federal law is reviewed de novo.

We do not want to bog down in a “standard of review quagmire,” Connect Commc’ns, 467 F.3d at 709, so we will not attempt a more comprehensive exegesis of these general principles.

Turning to the merits, Qwest first argues that the Commission’s order violates federal law because it sets rates based on a “mathematical equation” rather than a “TELRIC cost study.” Qwest relies almost exclusively on 47 C.F.R. § 51.505(e), which provides in pertinent part that “[a]n incumbent LEC must prove to the state commission that the rates for each element it offers do not exceed the forward-looking economic cost per unit of providing the element, using a cost study that complies with the methodology set forth in this section.” The parties disagree about whether this challenge to the Commission’s deaveraging method raises a pure question of law or a mixed question of law; and fact.

Qwest contends that the Commission’s compliance with the Act and applicable FCC regulations is a question of law and that “[n]o deference is owed to the cost methodologies a state commission adopts to set rates.” Appellant’s Br. at 39. The Commission contends that the question whether its deaveraging method complies with the Act and applicable regulations depends in part on its underlying factual findings and in turn on “whether the record contains sufficient evidence to support those factual findings.” Appellee’s Br. at 27-28.

This challenge is difficult to classify because it assumes a crucial fact: that the Commission’s deaveraging method nullified the results of the three cost studies that the Commission used in Docket C-2516. That assumption fails to account for the limited scope of the current rate-setting proceeding before the Commission. Recall that the Commission’s avowed purpose in Docket C-3554 was to deaverage the leasing rates for Qwest’s loops by adding in-town and out-of-town designations to the existing geographic zones. The Commission’s deaveraging method started with the existing leasing rates in each zone.

Those rates satisfied the requirements set out in 47 C.F.R. § 51.505(

e) because they were based on the averaged results of three cost studies. The Commission used those cost studies to determine TELRIC for Qwest’s loops. As the district court noted, the Commission’s “division of the [existing] zones into ‘in-town’ and ‘out-of-town areas’ necessitated a change in rates, [but] the loop costs that were used in the rate calculation remained unchanged.” Qwest Commc’ns Corp. v. Vap, No. 8:07-cv-430, 2008 WL 4866351, at *9 (D.Neb. Nov.7, 2008) (unpublished). Hence, a new study of loop costs would have been superfluous.

Qwest suggests that there is a hole in this logic because the deaveraged rates “bear no resemblance to the original TEL-RIC-compliant rates.” Reply Br. at 9. In particular, Qwest asserts that “[a]ll rates for out-of-town loops have increased between 205% and 239%, and rates for in-town loops in zones 2 and 3 have decreased between 61% and 83%.” Reply Br. at 9. That is quite obviously an invalid comparison; one should expect the bottom-line rates to differ, perhaps by large amounts, based on the number of geographic zones (four versus three) and the use (or not) of in-town and out-of-town designations.

Notwithstanding those differences, the key input — estimated loop costs — remained the same for both sets of rates. The whole point of further deaveraging the existing rates was to account for geographic cost differences “on a more targeted basis.” The magnitude of the .increases and decreases in bottom-line rates that Qwest identifies only goes to show that the three-zone model was obscuring large differences in loop costs in in-town versus out-of-town areas.

Once the current rate-setting proceeding is placed in proper context, the relevant question of law becomes whether a new cost study is required every time a state commission wants to further deaverage existing rates that comply with the TELRIC standard.

Section 51.505(

e) does not mention geographic deaveraging, so it is inap-posite on this point. Qwest has not identified any other authority for the proposition that deaveraging existing rates that comply with the TELRIC standard requires a new cost study, and we have found none. Because the Commission’s deaveraging method did not nullify the results of the three cost studies that the Commission used in Docket C-2516, a new cost study was not required. Accordingly, we reject Qwest’s first argument for overturning the Commission’s order.

Qwest next argues that the Commission’s order violates federal law because the rates it sets are inaccurate and not “cost-based.” Qwest relies almost exclusively on the following sentence in the FCC’s 1996 order on proposed rulemaking: “Where costs differ, rate differences that accurately reflect those differences are not discriminatory.” Implementation of the Local Competition Provisions in the Telecommunications Act of 1996, 11 F.C.C.R. 15499, ¶ 860 (Aug. 8, 1996).

The relevance of this sentence is not immediately apparent, for Qwest does not contend that the rates violate the Act’s anti-discrimination provision. See 47 U.S.C. § 252(d)(1)(A)(ii) (providing that “[t]he just and reasonable rate for network elements ... shall be— nondiscriminatory”).

Instead, the crux of Qwest’s argument is that the rates violate the Act’s requirement “that prices must be based on estimates of forward-looking costs that are accurate.” Appellant’s Br. at 51; see 47 U.S.C. § 252(d)(1)(A)(i) (providing that “[t]he just and reasonable rate for network elements ... shall be — based on the cost (determined without reference to a rate-of-return or other rate-based proceeding) of providing the ... network element”). On this view, inaccurate rates are, ipso facto, not cost-based. Qwest frames the issue as a pure question of law.

Specifically, Qwest asserts that it is undisputed that the rates “are at least 22% inaccurate and are more likely 35% inaccurate.” Reply Br. at 7. According to Qwest, these inaccuracies are an unavoidable consequence of the regression model’s use of household density to allocate costs between in-town and out-of-town areas. Reply Br. at 14-15. The question therefore becomes whether rates with that “level of error meet the Act’s requirement of accurate, cost-based rates.” Reply Br. at 13. We reject the premise that the rates suffer from an “undisputed level of error.” Appellant’s Br. at 52.

Qwest apparently defines “error” as any deviation from the “estimates of true loop costs” derived from its preferred cost study, the “Benchmark Cost Proxy Model,” or “BCPM.” Reply Br. at 12-14. But we see no reason to treat the BCPM as the gold standard; the Commission certainly was not required to do so. The real benchmark here is TELRIC, and TELRIC is a flexible standard. We will not proceed as if there is a single “right” answer when it comes to estimating loop costs. See MPower Commc’ns, 457 F.3d at 630.

Thus, Qwest’s assertion that the rates are at least 22 percent inaccurate amounts to an exercise in semantics with little or no practical significance. Moreover, since Qwest has failed to provide any objective criteria for evaluating the accuracy of the regression model or the rates that it produced, its second argument turns out to be an ipse dixit. The Commission’s expert, Dr.

David Rosen-baum, testified that the regression model explained 95 percent of the variation between household density and the natural logarithm of loop costs and 78 percent of the variation between household density and loop cost estimates under the BCPM. Qwest asserts that an “error rate” of more than 20 percent is too high. But the fact that a regression model does not explain a certain amount of variation between two variables does not necessarily mean that the model is “inaccurate” in any legally meaningful sense.

In this instance, the amount of unexplained variation in the relationship between household density and loop cost estimates {viz., 22 percent) does not necessarily mean that the resulting rates deviate from TELRIC, much less that the rates deviate from TELRIC by at least 22 percent. Qwest also points to testimony from its own expert, Dr. William Fitzsimmons, who testified that the results obtained using the regression model deviate from loop cost estimates under the BCPM by 35 percent.

The Commission found that Qwest “did not provide any alternative statistical criteria for evaluating whether or not the fit of the regression model was poor or the variation substantial.” Put differently, Qwest’s expert failed to provide adequate context for the Commission to assess the significance of his findings. As a result, the Commission determined that the regression model was “sufficiently accurate.” In our view, the dispositive question is whether the Commission’s adoption of the deaveraging method proposed by its staff was arbitrary and capricious. Accord MCI Worldcom Commc’ns, Inc. v.

BellSouth Telecommc’ns, Inc., 446 F.3d 1164, 1177 (11th Cir.2006) (reviewing a state commission’s decision to adopt a certain method of geographically deaveraging rates under the arbitrary and capricious standard).

The applicable FCC regulation is 47 C.F.R. § 51.507(f), which provides that “[s]tate commissions shall establish different rates for elements in at least three defined geographic areas within the state to reflect geographic cost differences.” As the Eleventh Circuit has recognized, “section 51.507 does little to cabin the discretion of a state commission when devising a geographic cost-based deaveraging method.” MCI Worldcom Commc’ns, 446 F.3d at 1177. Qwest has not shown that the Commission’s deaveraging method violated the general requirements set out in

section 51.507(f). Amd we are not convinced that the Commission’s adoption of that method was arbitrary and capricious. On the contrary, Dr. Rosenbaum’s testimony leaves little doubt that household density was “the single most significant cost variable.” Qwest Commc’ns Corp. v. Vap, 2008 WL 4866351, at *9. Qwest has not proven that the rates are unreasonable, unsupported by substantial evidence, or otherwise inconsistent with the TELRIC standard. See MPower Commc’ns, 457 F.3d at 629-30. Nor has Qwest offered a viable alternative method of further deaveraging the existing rates. Cf. Qwest Corp. v.

Koppendrayer, 436 F.3d at 868 (upholding the commission’s adoption of a potentially-flawed cost model, in part because the ILEC failed to provide a “satisfactory alternative”). Qwest’s conclusory assertion that the deaveraged rates are inaccurate and not cost-based is not enough to justify setting aside the Commission’s order. Accordingly, we reject Qwest’s second argument. Qwest’s third argument is that the Commission’s order violates federal law because the rates it sets will undermine “facilities-based competition” in Nebraska.

Qwest variously describes facilities-based competition as “[t]he Act’s ultimate objective,” Reply Br. at 24, and, more modestly, as “one of the Act’s central purposes,” Appellant’s Br. at 58. Qwest asserts that the new leasing rate for loops in in-town areas underestimates loop costs. The in-town rate will discourage facilities-based competition, the argument runs, because CLECs would rather lease Qwest’s loops at an “artificially low” price than invest in building a parallel network. Reply Br. at 24-25.

Qwest claims that the Commission’s order “creates the opposite problem” in out-of-town areas because “it is simply uneconomic for CLECs to compete at the out-of-town rates.” Appellant’s Br. at 61. We will review this challenge under the arbitrary and capricious standard, keeping in mind that if the Commission’s order contravenes federal policy then the order must give way. See AT & T Corp. v. Iowa Utils. Bd., 525 U.S. 366, 378 n. 6, 119 S.Ct. 721, 142 L.Ed.2d 835 (1999). Initially, we reject the proposition that promoting facilities-based competition is the sine qua non of valid rate-setting under the Act.

The Supreme Court has noted that the Act authorizes CLECs to pursue three “strategies” for competing in a local market: First, a competitor entering the market ... may decide to engage in pure facilities-based competition, that is, to build its own network to replace or supplement the network of the incumbent.... At the other end of the spectrum, the statute permits an entrant to skip construction and instead simply to buy and resell “telecommunications service,” which the incumbent has a duty to sell at wholesale....

Between these extremes, an entering competitor may choose to lease certain of an incumbent’s “network elements,” which the incumbent has a duty to provide “on an unbundled basis” at terms that are “just, reasonable, and nondiseriminatory.” Verizon Commc’ns Inc. v. FCC, 535 U.S. 467, 491-92, 122 S.Ct. 1646, 152 L.Ed.2d 701 (2002) (citing and quoting 47 U.S.C. § 251(a)-(c)).

The Act thus recognizes the “practical difficulties” of promoting competition, id. at 491, 122 S.Ct. 1646, and “allows for an entrant that may have to lease some ‘unnecessarily expensive’ elements in conjunction with building its own elements to provide a telecommunications service to consumers,” id. at 510 n. 27, 122 S.Ct. 1646 (quoting id. at 546, 122 S.Ct. 1646 (Breyer, J., concurring in part and dissenting in part)).

Indeed, the FCC has acknowledged that “entrants may need to share some facilities that are very expensive to duplicate (say, loop elements) in order to be able to compete in other, more sensibly duplicable elements (say, digital switches or signal-multiplexing technology).” Id. at 510 n. 27, 122 S.Ct. 1646 (emphasis added) (recounting the FCC’s position on facilities-based competition).

Assuming, then, that the Commission’s order will limit facilities-based competition in in-town areas, that result is not necessarily at odds with federal policy, for the Act does not require CLECs to construct parallel networks of loops if leasing the ILEC’s existing loops would be equally or more efficient. To the extent the Act does embody a federal policy of promoting facilities-based competition wherever possible, Qwest has not proven that the Commission’s order will not do so.

Qwest insists that the new rate for loops in in-town areas will discourage facilities-based competition because the rate underestimates loop costs. Here again, Qwest compares the new rate for loops in in-town areas to the rates for those loops under the three-zone model. As we have said, that comparison is invalid since the point of further deaveraging the existing rates was to account for geographic cost differences (ie., the differences in loop costs in in-town versus out-of-town areas) that the three-zone model obscured. Qwest also points to testimony from Dr.

Fitzsimmons indicating that the results obtained using the Commission’s regression model deviate from loop cost estimates under the BCPM. But as we have said, the Commission was not required to treat the BCPM as the gold standard. The Commission found that the new rate for loops in in-town areas complies with the TEL-RIC standard, and that is what matters here. See Qwest Corp. v. Koppendrayer, 436 F.3d at 863 (noting that rates based on TELRIC “can provide accurate market signals to inform CLECs’ decisions about whether to invest in their own facilities or lease the ILEC’s facilities”).

Qwest’s claim that the Commission’s order will discourage all competition in out-of-town areas also lacks merit. Once more, Qwest compares the new rate for loops in out-of-town areas to the rates for those loops under the three-zone model. That comparison is invalid, for the reasons we have already discussed. In addition, Qwest posits that the “dramatic disparities” in rates for in-town and out-of-town areas “will stifle or altogether eliminate competitive alternatives offered by CLECs in rural areas.” Appellant’s Br. at 61.

In support of this point, Qwest notes that a CLEC would be required to pay more than $200 to lease a loop in Zone 3, yet the rate it could charge the customer would be sharply constrained under state law. Qwest reasons that “[a]ny rational CLEC would ... choose not to provide service to the customer.” Appellant’s Br. at 61. We are not persuaded. This line of argument overlooks the existence of portable subsidies, which the Commission has targeted to rural, out-of-town areas where the costs of providing service are highest.

Returning to Qwest’s example, a CLEC that leases an out-of-town residential loop in Zone 3 must pay $210.90 to Qwest, but receives up to $186.76 from the universal service fund. Taking portable subsidies into account, we cannot say that CLECs will be unable to compete with Qwest in out-of-town areas. In any event, the Commission found that the new rates for loops in out-of-town areas comply with the TEL-RIC standard. We may not overturn the Commission’s order based on mere speculation about how CLECs will react to the “market signals” sent by those rates. See Qwest Corp. v. Koppendrayer, 436 F.3d at 863.

We conclude that the Commission’s order does not contravene federal policy. In particular, we are not convinced that the new leasing rates will have anticompetitive effects in any of the four geographically-deaveraged zones. Accordingly, we reject Qwest’s third argument. Qwest’s fourth argument is that the Commission’s order is based on arbitrary and capricious decision making. This challenge rehashes Qwest’s previous arguments regarding the Commission’s deaver-aging method and the accuracy of the new leasing rates. Qwest also alleges that the Commission ignored testimony from Dr. Fitzsimmons and another expert.

As we have said, review under the arbitrary and capricious standard does not permit “a court ... to substitute its judgment for that of the agency.” Connect Commc’ns, 467 F.3d at 711 (quoting In re Core Commc’ns, 455 F.3d at 277). But we will not blindly defer to an agency decision that is uninformed or unexplained. The agency must “examine the relevant data and articulate a satisfactory explanation for its action including a rational connection between the facts found and the choice made.” Id. (quoting In re Core Commc’ns, 455 F.3d at 277); see also Motor Vehicle Mfrs. Ass’n v. State Farm Mut. Auto. Ins.

Co., 463 U.S. 29, 43-44, 103 S.Ct. 2856, 77 L.Ed.2d 443 (1983). We conclude that the Commission considered the relevant evidence and gave an adequate explanation for its decision. In particular, we find that the Commission gave a reasoned explanation for its decision to adopt the deaveraging method proposed by its staff, which we have already discussed in connection with Qwest’s first argument. Likewise, we find that the Commission gave a reasoned explanation for its determination that the regression model was “sufficiently accurate,” which we have already discussed in connection with Qwest’s second argument.

Finally, we find that Qwest’s allegation that the Commission ignored relevant testimony is not supported by the record. The fact that the Commission “chose the ... methodology of one expert over another” does not mean that it ignored the testimony of Qwest’s experts. See Qwest v. Koppendrayer, 436 F.3d at 871. Although the Commission’s order “could have provided more detail,” see id., the order shows that the Commission engaged in rational decision making that was not arbitrary and capricious. Accordingly, we reject Qwest’s fourth argument.

Qwest’s fifth argument is that it cannot implement the Commission’s order because there is no reliable method of delineating the boundaries between in-town and out-of-town areas. Specifically, Qwest claims that using municipal tax records is not a workable solution because the records are incomplete. Qwest asserts that implementing the Commission’s order would require it “to arbitrarily designate thousands of Nebraska customers as in-town or out-of-town and, in turn, arbitrarily determine whether the prices for their loops would be $10.76 or between SA and 19.6 times that rate.” Appellant’s Br. at 56.

We are troubled by the fact that Qwest proposed the solution that it now claims is woefully inadequate. Before the Commission, Qwest described its alternative method of implementing the Commission’s order as “a reasonable approximation” of the Commission’s definition of in-town and out-of-town areas. Qwest now says that method is “imperfect,” and that it only proposed using municipal tax records as a stopgap measure. Appellant’s Br. at 55. In our view, Qwest has not shown that delineating the relevant boundaries is impossible, only that doing so would be costly and inconvenient.

Nevertheless, at oral argument, the Commission invited us to remand for further proceedings concerning how Qwest will implement its order. We agree that a remand is appropriate, but only for the limited purpose of settling on a workable method of implementing the order. To be clear, we reject Qwest’s argument that there is no reliable method of delineating the boundaries between in-town and out-of-town areas. We express no opinion, however, on whether using municipal tax records is a viable long-term solution. III.

CONCLUSION For the foregoing reasons, we affirm the judgment of the district court insofar as it upheld the leasing rates set by the Commission in Docket C-3554. We remand to the district court with instructions to remand to the Commission for further proceedings concerning how Qwest will implement its order. HANSEN, Circuit Judge, concurs in the judgment. . The Honorable Richard G. Kopf, United States District Judge for the District of Nebraska. . The ILECs were spun off from the old AT & T in 1982 as part of the antitrust settlement that ended the national monopoly in telephone service.

See AT & T Commc’ns of Ill., Inc. v. Ill. Bell Tel. Co., 349 F.3d 402, 404 (7th Cir.2003). . Qwest sought to reduce its leasing rates in connection with an application to provide long-distance phone service in Nebraska and other states, which was then pending before the FCC. See 47 U.S.C. § 271. . This is another term that the parties use to refer to hypothetical estimates of "long-run costs” for Qwest’s loops (as opposed to the actual “historical costs” that Qwest has invested in its plant). See generally Verizon Commc’ns, 535 U.S. at 495-501, 122 S.Ct. 1646 (2002); AT & T Corp. v. Iowa Utils.

Bd., 525 U.S. 366, 424-25, 119 S.Ct. 721, 142 L.Ed.2d 835 (1999) (Breyer, J., concurring in part and dissenting in part). .Specifically, the Commission authorized continuing portable subsidies to CLECs that had existing leases on loops providing service to business customers but eliminated portable subsidies linked to business loops that were not already leased by a CLEC. . For instance, the portable subsidy linked to loops providing service to residential customers in Zone 3 exceeded the leasing rate, so a CLEC profited from leasing such loops even without charging the customer. .

The Commission determined the "expected cost" for a given area based on the observed relationship between forward-looking loop costs and household density. Expected cost thus differs from "total cost,” which is simply the product of existing leasing rates and the number of loops in each zone. .The Commission established the following portable subsidies linked to loops providing service to residential customers: In-town, $0; Zone 1 (out-of-town), $12.90; Zone 2 (out-of-town), $71.18; and Zone 3 (out-of-town), $186.76.

Thus, to take just one example, a CLEC that leases an out-of-town residential loop in Zone 2 must pay $95.32 to Qwest, but receives up to $71.18 from the universal service fund. . In practice, this means that an aggrieved party cannot obtain de novo review merely by framing its challenge to a commission's cost estimates (WWC License) or a commission's adoption of a particular cost model (Qwest Corp. v. Koppendrayer) as a failure to "comply” with federal law. .

Which standard of review we formally apply may often turn out to be academic, for "there is no doubt ... that if the federal courts believe a state commission is not regulating in accordance with federal policy they may bring it to heel.” AT & T Corp. v. Iowa Utils. Bd., 525 U.S. 366, 378 n. 6, 119 S.Ct. 721, 142 L.Ed.2d 835 (1999). It would not be appropriate, however, to defer to a commission's judgment based on the possibility of corrective action.

See AT & T Commc’ns of Ill., 349 F.3d at 411 ("[T]he possibility of repair in the future is no warrant for promulgating today a rate that deviates from the TELRIC standard. Federal law requires any rate for unbundled network elements, adopted by a state commission, comply with TELRIC when adopted.”). . While the rates set in Docket C-2516 were later reduced by 20 percent, Qwest has not shown that the initial rates failed to comply with the TELRIC standard. .

Qwest submitted an updated cost study to the Commission but the Commission did not consider it, noting that “if Qwest would like the Commission to revisit the UNE rates [i.e., the underlying loop costs] established in Docket C-2516, Qwest should file a request for the Commission to open a proceeding to determine whether [those] rates ... should be reviewed in light of alleged rising costs, evolution in technology, increases in competition and other market changes.” As far as we can tell, Qwest never formally requested such a proceeding.

Qwest might have assumed that by opening Docket C-3554 the Commission intended to set new leasing rates from scratch; if Qwest made such an assumption, it was plainly mistaken. . Qwest points to a dictum in which the district court questioned "whether a rate zone that cannot reliably be drawn on a map is a ‘defined geographic area' within the meaning of 47 C.F.R. § 51.507(f).” Qwest Commc’ns Corp. v. Vap, No. 8:07-cv-430, 2008 WL 4866351, at *10 n. 26 (D.Neb. Nov.7, 2008) (unpublished).

The record shows that Qwest did not raise this issue before the district court, and Qwest has not offered any persuasive reason to consider the issue on appeal. See, e.g., Beavers v. Lockhart, 755 F.2d 657, 662 (8th Cir.1985) ("The general rule, of course, is that federal appellate courts will not consider issues not raised in the district court.” (citing Hormel v. Helvering, 312 U.S. 552, 556, 61 S.Ct. 719, 85 L.Ed. 1037 (1941))). In any event, the dictum begs the question by assuming that the deaveraged zones cannot be reliably drawn on a map.

Qwest asserts that delineating the relevant boundaries is impossible, but it has not convincingly established that premise. To be sure, the zones might not be contiguous or compact, but we are not aware of any prohibition on “gerrymandering” in these circumstances. . Even if we treated this challenge as a pure question of law, it would still fail. Qwest has offered no plausible basis for setting the dividing line between "accurate” and "inaccurate” rates; merely asserting that the new leasing rates are inaccurate does not make it so.

In fact, Qwest has not identified any provision in the Act or the FCC’s implementing regulations that uses the term "accuracy,” much less a provision defining the meaning of the term in this context. . We will also keep in mind that the FCC has made clear that state commissions have substantial discretion under the Act to implement a “pro-competitive regime” in their local markets. Implementation of the Local Competition Provisions, 11 F.C.C.R. 15499, ¶ 22. .

Qwest has not identified any evidence in the record about CLECs’ past, current, or projected future investment in building their own loops, in Nebraska or elsewhere. Without reliable information about baseline investment levels, we are left to speculate about the magnitude of the purported reduction in future investment. . Qwest argues that the four-zone model violates 47 C.F.R. § 51.507(

f) because the in-town and out-of-town zones cannot reliably be drawn on a map (and thus do not qualify as "defined geographic areas”). We reject this argument for the reasons discussed in note 13, supra. We also reject any suggestion that the Commission must find a "perfect” solution. Thus, Qwest should not construe this limited remand as an invitation to take a second bite at challenging the validity of the Commission’s order.