The regional Bell operating companies ("BOCs"), split off from AT&T in the 1982 antitrust settlement, provide most local telephone service. Section 271 of the Telecommunications Act of 1996 (the "Act"), 47 U.S.C. § 271, offers the BOCs a deal: such a company can enter the long distance business in a state within its service area if it takes specified steps to open the local-service market to competition. SBC Communications, a provider of local service in Kansas and Oklahoma, applied to the Federal Communications Commission for authorization to enter the long distance market in those states. Various long distance providers, including the five appellants before us, objected. The FCC granted the authorization, see Joint Application by SBC Communications, Inc. et al. for Provision of In-Region InterLATA Services in Kansas and Oklahoma, 16 F.C.C.R. 6237 (2001) (the "Order"), and this appeal followed. See 47 U.S.C. § 402(b)(6) & (9) (giving this court exclusive jurisdiction over challenges to the Commission's § 271 orders).

The full regulatory context of a § 271 application is set forth comprehensively in our decision affirming the FCC's approval of Bell Atlantic's § 271 application for New York. AT&T v. FCC, 220 F.3d 607, 610-15 (D.C. Cir. 2000), affirming Bell Atlantic New York Order, 15 F.C.C.R. 3953 (1999) ("New York Order"). Here we state only the bare bones. The Act entitles each of the BOCs to begin offering long distance service originating outside their local-service areas immediately. See 47 U.S.C. § 271(b)(2). But for authority to offer "in region" long distance service (i.e., service originating in a state where it provided local service), the Act requires a BOC to apply for Commission approval. Id. § 271(b)(1). The Commission then has 90 days to decide whether the BOC has shown that it is in compliance with the statutory prerequisites. Id. § 271(d)(3).

First, the BOC must satisfy either "Track A" or "Track B"--names derived from subparagraphs of § 271(c)(1). For Track A it must show that it provides network access to "one or more unaffiliated competing providers of telephone exchange service ... to residential and business subscribers." Id. § 271(c)(1)(A). If no competing provider has requested such access, the BOC may invoke Track B, showing that it is ready and willing to provide its competitors with network access and interconnectivity under terms "approved ... by the State commission." Id. § 271(c)(1)(B).

Besides prevailing on Track A or B, the BOC must establish that its offering of interconnection and access to competitive local exchange companies ("CLECs") meets the fourteen requirements of a "competitive checklist" contained in § 271(c)(2)(B). Many of these requirements are simply incorporations by reference of obligations independently imposed on the BOCs by §§ 251-52 of the Act, id. § 271(c)(2)(B)(i) & (ii), and enforced by state regulatory commissions pursuant to § 252. The required interconnection and access must be available on non-discriminatory terms and at cost-based rates. See, e.g., id. §§ 251(c)(2) & (3), 252(d)(1). Finally, the BOC must convince the FCC that "the requested authorization is consistent with the public interest, convenience, and necessity." Id. § 271(d)(3)(C).

SBC filed an application for long distance service authorization in both Kansas and Oklahoma on October 26, 2000. Its petition relied on the network element rates that were set by the Kansas Corporation Commission and the Oklahoma Corporation Commission in § 252 proceedings implementing SBC's duties under § 251. See Order at p p 12-16, 22-23. Numerous parties--including the Department of Justice, the Kansas Commission and the Oklahoma Commission, and the five appellants--filed comments. The state commissions weighed in on the side of SBC. DOJ's recommendations endorsed neither denial nor approval of the applications, but instead urged the FCC to scrutinize particular aspects of the petition, including SBC's prices for network elements.

On January 22, 2001--ninety days after the application was filed, and for the first time in a situation involving predominantly rural states--the FCC released its order granting SBC the authorizations. See Order p 1.

The Commission found first that SBC had satisfied Track A in both Kansas and Oklahoma because the company was supplying network access to one or more unaffiliated competitors providing residential and business customers with "facilities-based service." Order p p 40-44; see also § 271(c)(1)(A).

Next the Commission concluded that SBC had fully met the requirements of the competitive checklist in both states. Evidence in the record was virtually uncontested with respect to eleven of the fourteen checklist items. Order p p 39, 24155. As for the remaining three items, the Commission considered and rejected commenters' contentions that SBC failed to provide network elements and interconnection to CLECs at cost-based rates. See Order p p 45-240.

Appellants raised two rate-related arguments that are novel in § 271 litigation. Pointing to the rather low level of residential service by CLECs, they argued that the unbundled network element ("UNE") rates could not have genuinely conformed to the cost requirement, or else competition would have flourished, or at least not proven so modest. Further, pointing to submissions of evidence that SBC's UNE rates were too high to provide profitable residential service, they argued that SBC was engaged in a "price squeeze" (charging prices for inputs that precluded competition from firms relying on those inputs), and that accordingly the Commission could not find that authorization of its entry into the long distance market was "consistent with the public interest," as required by § 271(d)(3)(C). The Commission rather summarily rejected both claims. Order p p 92, 268.

Appellants here pursue three basic arguments. First, they make the arguments summarized above about the relation between UNE rates and low volumes of residential service. Second, they make a series of detailed attacks on the Commission's findings that the UNE rates were cost-based. Finally, they say that the FCC improperly relied on ex parte communications in its Kansas Track A determination, which was, in any event, independently erroneous. We consider the arguments in that order. We conclude that appellants have made out a case for a remand to the Commission only on the "public interest" aspect of the first issue.

1. Low-volume local competition, possible price squeeze and the public interest.

In contrast to the situation in the other two states where the FCC has previously granted long distance authority to a requesting BOC (New York and Texas), Oklahoma and Kansas have local telephone markets characterized by relatively low volumes of residential competition from non-BOC firms. Order p p 34, 92, 268. Appellants also point to evidence they submitted, evidently uncontradicted in the record, that competition in the residential market that was dependent on UNEs at SBC's prices could not succeed. They argue that the low volumes and the evidence of the impossibility of profitable competition both contradict the Commission's finding of cost-based rates and undermine its conclusion that granting SBC's applications was consistent with the public interest.

The Commission's standard for cost--"TELRIC" (or total long-run incremental cost)--is one that might normally be expected to generate competition. In principle, there is no reason to think the BOC's real costs could be lower. In an otherwise undistorted market, firms capable of efficiently supplying the non-BOC elements should be able to compete. Appellants' proposal that the Commission consider the volume of competition as a cross-check for its cost finding is therefore understandable.

But we can hardly find the Commission's rejection of appellants' proposal unreasonable. The statute imposes no volume requirements for satisfaction of Track A, so that it would be odd for the Commission to use low volume to defeat a finding of TELRIC-compliant rates. And it would be reasonable for the Commission to treat any questions raised by the low volumes, or by the appellants' evidence showing the difficulty of making a profit under SBC's rates, as subsumed within the issue of TELRIC compliance. See AT&T, 220 F.3d at 619. As the appellants concede, the lack of competition they allude to is neither a direct nor a conclusive proof of a checklist violation. Appellants' Br. at 52.1 The Justice Department argued that the low volumes "compela closer look" at whether SBC's rates were "properly costbased," Evaluation of the U.S. Department of Justice ("DOJ Report") at 10, and the FCC did just that, as we shall see.

Appellants repackage the data on volumes and potential profitability as an attack on the Commission's public interest finding. Concededly, Congress did not expressly consider whether such data could prevent a public interest finding or even require further inquiries by the Commission. See Chevron U.S.A. v. Natural Resources Defense Council, 467 U.S. 837, 842-43 (1984). But appellants in substance argue that the only reasonable construction of the requirement calls at least for such an inquiry. They point to "price squeeze" decisions, construing public-interest provisions in statutes that are not explicitly aimed at fostering competition, which reviewed--and found wanting--agencies' asserted grounds for failing to follow up similar claims. See, e.g., Mid-Tex Elec. Coop., Inc. v. FERC, 773 F.2d 327, 351-62 (D.C. Cir. 1985). With a statute that proclaims competition as the congressional purpose, Pub. L. No. 104-104, purpose statement, 110 Stat. 56, 56 (1996), it follows a fortiori, appellants say, that the Commission should pursue their price squeeze claim, or at the very least explain why the public interest does not require it to do so. After all, "the words 'public interest' in a regulatory statute ... take meaning from the purposes of the regulatory legislation." NAACP v. FPC, 425 U.S. 662, 669 (1976).

In fact, the Commission gave appellants' claim rather a brush-off. First, the Commission said that under its reading of the Act, the "profitability" considerations raised by appellants were "irrelevant" because the Act directed it to assure that the rates were cost-based, "not [to determine] whether a competitor can make a profit by entering the market." Order p 92; see also id. p 65. This, of course, is unresponsive. The issue is not guarantees of profitability, but whether the UNE pricing selected here doomed competitors to failure.

Second, the Commission reasoned that consideration of the price-squeeze claims would have exceeded the Commission's authority under the Act, because it would have required the FCC to invade state commissions' exclusive jurisdiction over retail rates. Order p 92 ("Were we to focus on profitability, we would have to consider the level of a state's retail rates, something which is within the state's jurisdictional authority, not the Commission's."). But the Supreme Court rejected precisely this argument in FPC v. Conway Corp., 426 U.S. 271 (1976). There the Federal Power Commission had refused to hear price-squeeze evidence on the theory that it was powerless to eliminate any such squeeze, as retail rates fell in the exclusive jurisdiction of state commissions. Id. at 277. The Supreme Court responded that the remedy, if any, could take the form of the Commission's fixing the wholesale rates, which were under its jurisdiction, at a lower level within "the zone of reasonableness." Id. at 279 (internal quotation marks omitted).

The Commission makes a related argument in its brief, saying that any price-squeeze claim is effectively rebutted by the Commission's finding that UNE rates were cost-based. That is true to the extent that an agency can pinpoint TELRIC rates; as we suggested earlier, in an undistorted market "perfect" TELRIC rates would seem inconsistent with any price squeeze. But to the extent that an agency can confidently identify TELRIC rates only within some band, like those involved under conventional "just and reasonable" regulation, the possibility exists that the agency has chosen too high a point within the band. As the Court said in Conway:

This argument, however, assumes that ratemaking is an exact science and that there is only one level at which a wholesale rate can be said to be just and reasonable.... [H]owever, there is no single cost-recovery rate, but a [wide] zone of reasonableness....

Conway, 426 U.S. at 278.

Finally, the Commission observes, "[f]actors beyond a BOC's control, such as individual CLEC entry strategies for instance, might explain a low residential customer base." Order, p 268. This is, of course, correct in principle, although it may leave the inquisitive wondering why firms would forego what superficially appear to be promising opportunities for profit. But the observation is no basis for rejecting a proffer of evidence that the ceiling level for UNE rates-fixed by the state commissions and approved by the FCC itself--precluded profitable entry.

At oral argument Commission counsel offered another analysis, not explicitly mentioned in the Order. He suggested that state commissions have historically set relatively low residential rates, especially rural ones, allowing the incumbent monopoly to make it up in other aspects of their business. Compare City of Batavia v. FERC, 672 F.2d 64, 90 n.52 (D.C. Cir. 1982) ("Conway [does not] require the Commission to set a wholesale rate so that wholesale customers are guaranteed the ability to compete in the retail market. It may be, for instance, that the state commission is not allowing an adequate rate of return. If so, the Commission is not obliged to follow, and indeed may not follow suit, for the rates it sets must fall within the zone of reasonableness."). This was in part contested by appellants' counsel, who suggested that even with state commission regulation it would be possible to offer certain enhanced services profitably--if only UNE rates were capped at correct TELRIC levels, or, in the alternative formulation, at lower levels within the correct TELRIC range.

In any event Commission counsel's observation is not a ground for rejecting appellants' claim; we can affirm only on the Commission's reasoning. See SEC v. Chenery Corp., 318 U.S. 80, 88 (1943). We therefore remand the case to the Commission for reconsideration of this issue. Of course Conway itself "merely holds that the Commission must weigh anticompetitive effects along with other factors in setting rates." Kansas Cities v. FERC, 723 F.2d 82, 94 (D.C. Cir. 1983) (citing Conway, 426 U.S. at 278-79). Here, as the Act aims directly at stimulating competition, the public interest criterion may weigh more heavily towards addressing potential "price squeeze." But the Commission's reasoning, when it is proffered, will presumably help establish the reasonable range for interpretations of the statutory criterion.

In closing we note two points, without evaluation. First, the potential scale of a serious price squeeze inquiry may present a problem; we have already recognized that Congress's 90-day limit constrains the scope of the Commission's inquiries. AT&T, 220 F.3d at 631. Second, if the Commission is correct in reading Track A to require only a minimal volume of competition to be present, see Order p 42, and that reading is not challenged here (though its application is), it may reflect a recognition that the residential market may not be attractive to competitors even if UNE costs are at the lower end of TELRIC (assuming it to have a material range). See City of Batavia, 672 F.2d at 90. While we remand for consideration of this issue, we do not vacate the Order. See, e.g., Allied-Signal, Inc. v. NRC, 88 F.2d 146, 150-51 (D.C. Cir. 1993) ("The decision whether to vacate depends on 'the seriousness of the order's deficiencies (and thus the extent of doubt whether the agency chose correctly) and the disruptive consequences of an interim change that may itself be changed.' ").

2. Specific attacks on the Commission's UNE rate findings.

General background. There is no dispute here on the basic principle governing the rates at which a BOC must offer unbundled network elements. The 1996 Act gives the Commission general authority to promulgate rules necessary to implement the statute, while entrusting the individual state commissions with the application of those principles in the process of approving interconnection agreements and rates pursuant to § 252. See AT&T Corp. v. Iowa Utils. Bd., 525 U.S. 366, 385 (1999); AT&T, 220 F.3d at 615. In August 1996 the FCC adopted general rules, including the TELRIC standard. See First Report and Order, Implementation of Local Competition Provisions in the Telecommunications Act of 1996, 11 F.C.C.R. 15499, 15844 p 672 (1996).

A forward-looking rather than a historical measure, TELRIC bases rates on "the cost of operating a hypothetical network built with the most efficient technology available." Iowa Utils. Bd., 525 U.S. at 374 n.3. "TELRIC is not a specific formula," but rather a collection of "methodological principles." AT&T, 220 F.3d at 615 (internal quotation marks omitted). As it allows the states wide latitude to account for "local technological, environmental, regulatory, and economic conditions," its application "may result in different rates in different states." Id. (internal quotation marks omitted). When the Commission adjudicates § 271 applications, it does not--and cannot--conduct de novo review of state ratesetting determinations. Instead, it makes a general assessment of compliance with TELRIC principles. Id.

Specifically, the Commission has said that it "will not reject an application ... [unless] basic TELRIC principles are violated or the state commission makes clear errors in factual findings on matters so substantial that the end result falls outside the range that the reasonable application of TELRIC principles would produce." Order p 59 (quoting New York Order, 15 F.C.C.R. at 4084, p 244).

We review the Commission's decision on TELRIC compliance under the arbitrary and capricious standard. 5 U.S.C. § 706(2)(A). In AT&T we identified three reasons why "special deference" was in order for our review of a § 271 finding by the Commission: that the issues at stake were ones involving a high level of technical expertise in an area of rapidly changing technological and competitive circumstances; that the Commission itself was reviewing a state agency with considerable expertise; and that "enormous flexibility built into TELRIC." 220 F.3d at 616. These reasons of course continue to apply, and so shall the deference.

Thus, a challenger can prevail here by making one of two showings. First, he may demonstrate that the FCC acted arbitrarily and capriciously in finding that the state commission followed basic TELRIC principles. Alternatively, he may point to specific factual errors made by the state commission, and demonstrate either that the FCC failed to consider these errors or that it arbitrarily determined that the rates were nevertheless within the range acceptable under TELRIC.

Kansas non-recurring charges. A CLEC leasing a network element from the BOC incurs two types of charges. Recurring charges are assessed for the use of the network element in question over some time period. Non-recurring charges are for one-time activity of processing orders for UNEs and for physically providing them initially. Under the FCC's rules, TELRIC applies to both. See 47 C.F.R. 51.507(e). Our inquiry in this section focuses on SBC's non-recurring charges in Kansas.

The key dispute revolves around the so-called "fall-out factor." Even with the best technology available, each stage in the various automatic processes may fail, requiring more expensive manual intervention. When this occurs, a service order is said to "fall out." Kansas Commission Nov. 3, 2000 Non-Recurring Charges Order p 35;2 see also Order p 57 (comparing a $2.35 non-recurring service charge for electronically-processed orders with a $12.35 charge associated with manual processing).

At the outset of its § 252 proceeding, the Kansas Commission found itself faced with what it deemed unrealistic cost proposals submitted by both parties (SBC and the long distance providers). Over a four-year period it conducted "extensive workshops, hearings, and other types of discovery" aimed at obtaining an accurate estimate of SBC's UNE costs. Order p 49. Finally, it ordered both sides to rerun their TELRIC recurring and non-recurring cost studies "using certain prescribed inputs and cost assumptions" that the commission thought would better approximate efficient network design and operation. See id.

For non-recurring charges this attempted solution proved defective. Making an estimate that fall-out probably would occur about 5% of the time in a well-functioning system, the commission directed the two sides to re-run their studies on this aspect yet again, using this figure. Kansas Commission Sept. 17, 1999 Reconsideration Order p 70; Kansas Commission Nov. 3, 2000 Non-Recurring Charges Order p 35. Both sides, however, applied the 5% factor creatively so as to inflate or understate the net rate of manual processing, as would suit their respective cases. Kansas Commission Nov. 3, 2000 Non-Recurring Charges Order p p 13(C), 35, 38 (noting that in the case of one particular network element, SBC's unfounded assumptions ballooned the cumulative fall-out factor to 59.3%). Although the exact nature of AT&T's fudging is obscure, see id. p 13(C), AT&T here makes no claim to have complied precisely with the commission order.

Concerned with preventing further delay in the already lengthy proceedings, and frustrated by the carriers' failure to follow its directions, the commission decided to set nonrecurring rates based on "information previously received in this matter" and "its best judgment." Kansas Commission Nov. 3, 2000 Non-Recurring Charges Order p 4. To fix the majority of non-recurring rates it split the baby, adopting a weighted average of the AT&T (2/3) and SBC (1/3) proposals. Kansas Commission Nov. 3, 2000 Non-Recurring Charges Order p 32; Revised Attachment B at 10 n.2. In weighting the studies "so that the final price fell toward the low end of the range of possible prices," id. p 32, the commission aimed to avoid "reward[ing] an inefficient service provider," id. It set the remaining rates by adopting either SBC or AT&T cost proposals, or by relying on similar rates from comparable jurisdictions.

SBC's initial § 271 application to the FCC relied on the non-recurring charges calculated as described above. Responding to comments filed by the Department of Justice and private commenters, which noted that many of the charges were a good deal higher than the corresponding rates in Texas, SBC proposed to lower its non-recurring rates in Kansas by 25%. SBC December 28, 2000 Ex Parte Letter at 2, Joint Appendix ("J.A.") 1447; see also Order p 56. The Kansas Commission adopted these cuts on January 5, 2001. Order p 52. In light of "these additional voluntary reductions," the Commission reasoned that it "need not reach a conclusion as to whether the carriers' failure to follow the Kansas Commission's directions" in running cost studies resulted in non-recurring rates that violate TELRIC. Order p 60. In its opinion, the cuts eliminated "any remaining concerns" about the non-recurring rates in Kansas.

Appellants claim the FCC could not properly find these rates TELRIC-compliant, as they are the product of a crude "settlement" method, trimmed by an arbitrary 25% "haircut." Accusing the Kansas Commission of acting out of selfimposed pressure to support SBC's § 271 application, the appellants say that the Kansas Commission should have required SBC to continue re-running its studies until they reflected "accurate and Commission-approved cost data." Appellants' Br. at 39 (quoting Kansas Commission Nov. 3, 2000 Non-Recurring Charges Order p 30 (internal quotation marks omitted)). (Appellants often cite to pages in the Kansas Commission's rulings where the commission finds fault with SBC's studies, never mentioning that the same pages allocate seemingly equal fault to AT&T's.)

Appellants' complaints overlook the procedural context of the Kansas proceeding. The commission stated that the burden of producing cost studies "squarely fell upon AT&T and SWBT," Kansas Commission Nov. 3, 2000 Non Recurring Charges Order p 28 (J.A. 800-01); appellants have not contested this (or mentioned it!). Given that calculation of forward-looking costs for a hypothetical network requires far more pervasive use of predictive judgments than does standard cost-of-service ratemaking, some such allocation was likely essential. Especially in the rural states, the burden on regulatory resources might otherwise have proven overwhelming. Cf. Order p 2. Perhaps AT&T might have legitimately protested the assignment of any burden to the protesters, but it appears not to have done so.

In fact the Kansas Commission diligently labored for several years scrutinizing AT&T's and SBC's submissions; to the extent that it believed those studies contained errors, it made adjustments. As the appellants concede, the Kansas Commission "identified the many cost-inflating flaws in SBC's cost studies." Appellants' Repl. Br. at 17. In such instances it would then use "inputs, in some instances those proposed by AT&T, that it found more reasonable," accordingly reducing SBC's non-recurring charges for service orders and several kinds of loops. Order p 63.

As we have seen, however, sound application of a fall-out rate proved comparatively intractable. Appellants argue that the weighted (2/3)/(1/3) formula, even when coupled with the 25% cut, was "mathematically incapable" of bringing SBC's exaggerated fall-out ratios down to the correct 5% level. But in fact appellants waited until their reply brief before deigning to offer any detailed numbers in support of their complaint. See id. at 20-21 n.5. Further, we note that if the calculations were so simple, surely AT&T itself could have offered the Kansas Commission some easy way of bringing SBC's submissions--or its own--into line with the 5% criterion. In AT&T, rejecting AT&T's insistence that switching costs should have been adjusted to reflect newly discovered information, we endorsed the New York agency's observation that its computations were "the result of a complex analysis that does not lend itself to simple arithmetic correction through the adjustment of a single input." 220 F.3d at 617 (internal quotes omitted). Here, too, we cannot fault the FCC for approving the Kansas Commission's compromise resolution of an issue that the parties' behavior had left a muddle.

Nor are we persuaded by appellants' suggestion that because some of SBC's non-recurring charges (even after its 25% discount) remain significantly higher in Kansas than in Texas, such charges are not TELRIC-compliant. First, even if the appellants' calculations were properly done, only two of the five non-recurring UNE rates on which appellants rely in this appeal--loop and cross-connect charges--are higher in Kansas; the other three--service order charge, analog line switch port, and central office access--are identical in both states. AT&T Clarke Supp. Decl. Ex. 1, Cols. F & J (J.A. 1526-27). Second, SBC submitted evidence indicating that if the relevant non-recurring charges (pre-discount) were amortized over a reasonable period of time, the total monthly UNE costs for CLECs serving comparable exchanges would be lower in Kansas than in Texas. Order p 61 n.171. Although appellants dispute the latter conclusion, saying that it was generated by selectively comparing charges in urban areas of Kansas to those in more costly suburban Texas, Appellants' Reply Br. 21, their own citations (see J.A. 151920, 1563-64) do not clearly establish a fallacy in SBC's contention.

Even if the two states' rates in fact differ, the FCC identified a likely cause. Whereas the Texas Commission had explicitly rejected SBC's claimed entitlement to a "trip charge" for various installation and maintenance activities, a comparable charge was evidently allowed in Kansas. Order p 61. Because the "trip charge" is not a directly observable quantity, but like many other inputs to the TELRIC model is a prediction about future costs faced by a not-yet existent efficient provider, a state-to-state difference in prediction scarcely proves that either resulting rate is non-TELRIC.

Appellants argue that the Kansas Commission never explicitly approved a trip charge. But in fact the Kansas Commission cited a statement by AT&T that "20 percent of service orders will require sending a truck (and technician) to work the order." Kansas Commission Nov. 3, 2000 NonRecurring Charges Order p 24; see also id. p 48-49. (Presumably this 20% is applied to a set different from and smaller than the set for which the 5% fall-out rate was considered appropriate.) While there may be no explicit Kansas finding of a trip charge, the Kansas Commission's express recognition of AT&T's own suggestion makes its exclusion by Kansas seem implausible.

Accordingly, we reject appellants' claim that the Commission was arbitrary or capricious in finding no violation by the Kansas Commission of basic TELRIC principles and no "clear errors in factual findings on matters so substantial that the end result falls outside the range that the reasonable application of TELRIC would produce." Order p 59.

Oklahoma rates. Oklahoma's recurring and non-recurring UNE rates were set by the Oklahoma Commission's administrative law judge on June 30, 1998, and approved by the Oklahoma Commission without modification on July 17, 1998. Although SBC's initial recurring and non-recurring cost proposals were founded on cost studies similar to those it had used in Kansas, the Oklahoma ALJ did not--unlike its Kansas counterpart--order a re-run of estimates. Declaration of Baranowski & Flappan, November 15, 2000, at 7. Instead, he recommended adoption of rates that SBC and Cox Oklahoma, a cable competitor, had agreed to in a § 252 arbitration, convinced that they appropriately reflected forward-looking costs. Order p 69.

Oklahoma non-recurring charges. Appellants now challenge the FCC's conclusion that Oklahoma's non-recurring rates, after a 25% discount paralleling that in Kansas, satisfied the § 271 checklist. Pointing out that Cox uses its own network to provide telephone service, has no need for most UNEs that other CLECs must rely on, and thus would benefit--or, at least, be indifferent--if SBC overcharges other local competitors for these elements, AT&T insists that the Oklahoma non-recurring rates are not supported by any cost evidence. Thus, it reasons, the FCC should have independently determined compliance with TELRIC.

We are unconvinced. First, the fact that the non-recurring charges recommended by the ALJ (before the 25% haircut) were 33% below those originally proposed by SBC, Order p 100, tends to rebut the claim that Cox didn't really object to high rates for SBC's UNEs. Second, that the ALJ's ultimate selection was of stipulated rates did not in itself show that they were not backed by any cost studies. The ALJ had SBC's and AT&T's cost data before him, and found that the stipulated rates fell within the range of cost-based rates proposed by the parties. Order p p 69, 90; see also Oklahoma ALJ Report at 159. The FCC ultimately concluded that the Oklahoma ALJ had "carefully analyzed the various cost studies submitted for nonrecurring charges, and was committed to TELRIC principles in making his evaluations." Order p 98. While appellants point to passages in the ALJ's report that may be ambiguous on the articulation of TELRIC, Appellants' Repl. Br. at 24 (citing Oklahoma ALJ Report at 165-66), even these snippets are not clearly inconsistent with a sound understanding of TELRIC.

Appellants correctly note that the ALJ mistakenly assumed manual processing for all service order charges. But the FCC saw that error too, and concluded that it was neutralized by SBC's subsequent implementation of a reduced charge for electronic orders, so that the higher charge originally adopted by the ALJ was in fact applied only to manually processed orders. Order p 99. Appellants question that conclusion, and, again challenging the ALJ's grasp of TELRIC, point to a section of the record where the ALJ rejected the contention--just as the Kansas Commission had--of an AT&T witness who claimed that almost all of the actual provisioning of UNEs could be done electronically. Order p 100 & n.286; see also Oklahoma ALJ Report at 166. The passage in fact doesn't show the ALJ to be the hopeless rube appellants paint him as.

Oklahoma recurring charges. Following the same line of reasoning it had applied when considering Oklahoma nonrecurring charges, the FCC also determined that SBC's Oklahoma recurring rates, with one exception, complied with TELRIC. Order p p 75, 90, 91. To the extent that appellants' challenge to the recurring rates in Oklahoma tracks their arguments raised against the non-recurring charges, we reject it for the reasons just stated.

The FCC's only serious doubt concerned the ALJ's choice of what the Commission viewed as an unreasonably low loop "fill factor" for the computation of recurring loop rates. "A fill factor is the estimate of the proportion of a facility that will be used. In other words, the per-unit cost associated with a particular element ... [is] the total cost associated with the element divided by a reasonable projection of the actual total usage of the element." Order p 78. Thus, the lower the fill factor, the higher the rate ceiling. The ALJ had erred, the Commission thought, by basing the fill factor on its current value (30%) instead of "consider[ing] the forwardlooking fill or assum[ing] that the fill factor would increase over time." Order p 80. AT&T was urging 50%. Id.

The FCC resolved this issue by (1) taking into account SBC's proposal of a 25% reduction in its rates and (2) comparing the resulting loop charge with that approved by the Commission for Texas. Order p p 82-87. To do so, it took into account estimates from its "USF" (universal service fund) cost model, which it regards as generally valid for comparisons between states, indicating that loop costs were 23% higher in Oklahoma than in Texas. Id. p 84. As the weighted average of the various Oklahoma discounted loop rates was only 11% higher than the equivalent for Texas, the FCC concluded that the Oklahoma rates, though not calculated by TELRIC means, nonetheless fell "within the range that TELRIC would produce." Order p 86. To create a distinction between properly derived cost-based rates and rates that were equal to them, the Commission said, "would promote form over substance, which, given the necessarily imprecise nature of setting TELRIC-based pricing, is wholly unnecessary." Order p 87.

AT&T's main complaint here is that the Commission was wrong to accept the Texas rates as a standard, as "the reasonableness of Texas network element rates was not litigated" in the Texas § 271 proceedings. But all of the appellants in this case participated in the Texas administrative proceedings and were free to litigate the matter if they thought litigation promising. As the stakes for appellants must have been high, and as they have not generally been shy about pressing claims in this area, we see no abuse of discretion in the Commission's reliance on the Texas rates, with suitable adjustments, to verify TELRIC compatibility for the Oklahoma recurring loop rates.

Against the Texas comparison AT&T also urges that new evidence, coming to light after the Texas rates had been set, shows that CLECs have abandoned their efforts to provide residential service in the state. But nothing that AT&T cites ties the alleged decline to excessive UNE rates.

Finally, AT&T says that as the FCC used some Texas recurring rates for comparison, it should have also compared all the Oklahoma recurring rates to Texas's. This issue was not raised before the FCC and is therefore waived. See 47 U.S.C. § 405; Coalition for Noncommercial Media v. FCC, 249 F.3d 1005, 1008-09 (D.C. Cir. 2001).

3. Track A. Under the FCC's interpretation of § 271, a BOC cannot qualify under Track A unless it shows that at least one competing provider, with which it has entered into an interconnection agreement, serves "more than a de minimis" number of residential and business customers using either exclusively or predominantly that CLEC's own telephone exchange facilities (including therein UNEs leased from the BOC, Ameritech Michigan Order, 12 F.C.C.R. 20543, 20588 p 101 (1997)) and thus constitutes "an actual commercial alternative to the BOC." SBC Oklahoma Order, 12 F.C.C.R. 8685, 8695 p 14 (1997); see also Order p 42. Appellants contend that SBC had failed to make that showing with respect to Kansas residential subscribers.

The weakness of appellants' position is most easily viewed in relation to the evidence for residential service offered by a CLEC named Ionex. SBC was not privy to Ionex's own figures on the subject. But because Ionex was offering service via UNEs leased from SBC, SBC was in a position to estimate the sum of Ionex's aggregate service volumes, and, by further extrapolation, to estimate the residential service. In its Reply Comments before the Commission, it produced an analysis suggesting that Ionex's residential service via UNE was sufficient to meet the Commission's standard-"more than a de minimis number." See SBC Reply Comments at 73, J.A. 1240; Affidavit of J. Gary Smith dated December 11, 2000 at 6, J.A. 1269.

Later SBC submitted an "ex parte" letter dated December 20, 2000, with actual numbers for Ionex and an explanation of its methodology. J.A. 1370-71, 1373. SBC made the filing of the letter public by an open filing on January 12, 2001, and appellants secured a copy from SBC on January 17, just two days before the Commission decided SBC's application. Appellants cry foul.

But Ionex was itself a party to the proceeding, sturdily resisting SBC's application and presumably fully aware of its residential services. The public SBC Reply put it on notice that SBC was using Ionex's service to satisfy Track A. Ionex uttered not a peep in protest, correction or qualification. So the FCC's observation that it used the ex parte letter only as "additional support" for the Track A finding, Order p 43, was entirely legitimate.

Appellants also say the Commission contradicted its own prior decision when it counted Birch as a competitive alternative, even though Birch was "not actively marketing" residential service in Oklahoma. They point to the Commission's 1997 decision on SBC's first Oklahoma § 271 application, where it found Brooks Fiber's residential service inadequate: the only "customers" were four Brooks employees receiving free service as a test, and Brooks was refusing to accept any new business. See SBC Oklahoma Order, 12 F.C.C.R. 8685, 8698 p 20 (1997). The Commission's finding no commercial alternative there hardly contradicts a positive finding here, where the Commission found that more than a de minimis number of real customers were actually being served. Order p 42.

* * *

Because the Commission has offered an inadequate justification for why it thought that evidence of a "price squeeze" precluding profitable CLEC competition was irrelevant to its public interest analysis, we remand the case for reconsideration of that issue. We reject all other claims.

So ordered.