US Telecom Assn v. FCC

Notice: This opinion is subject to formal revision before publication in the Federal Reporter or U.S.App.D.C. Reports. Users are requested to notify the Clerk of any formal errors in order that corrections may be made before the bound volumes go to press. United States Court of Appeals FOR THE DISTRICT OF COLUMBIA CIRCUIT Argued January 28, 2004 Decided March 2, 2004 No. 00-1012 UNITED STATES TELECOM ASSOCIATION, PETITIONER v. FEDERAL COMMUNICATIONS COMMISSION AND UNITED STATES OF AMERICA, RESPONDENTS BELL ATLANTIC TELEPHONE COMPANIES, ET AL., INTERVENORS Consolidated with 00–1015, 00–1025, 01–1075, 01–1102, 01–1103, 03–1310, 03–1311, 03–1312, 03–1313, 03–1314, 03–1315, 03–1316, 03–1317, 03–1318, 03–1319, 03–1320, 03–1324, 03–1325, 03–1326, 03–1327, 03–1328, 03–1329, 03–1330, 03–1331, 03–1338, 03–1339, 03–1342, 03–1347, 03–1348, 03–1360, 03–1372, 03–1373, 03–1385, 03–1391, 03–1393, 03–1394, 03–1395, 03–1400, 03–1401, 03–1424, 03–1442 –———— Bills of costs must be filed within 14 days after entry of judgment. The court looks with disfavor upon motions to file bills of costs out of time. 2 On Petitions for Writ of Mandamus and for Review of an Order of the Federal Communications Commission Michael K. Kellogg argued the cause for ILEC petitioners. With him on the briefs were Mark L. Evans, Sean A. Lev, Colin S. Stretch, Michael T. McMenamin, James D. Ellis, Paul K. Mancini, Joseph E. Cosgrove, Jr., Gary L. Phillips, James P. Lamoureux, Robert B. McKenna, Charles R. Mor- gan, James G. Harralson, William P. Barr, Michael E. Glover, and Edward Shakin. Donna M. Epps, Daniel L. Poole, John H. Harwood II, William R. Richardson, Jr., and Matthew R. Sutherland entered appearances. Donald B. Verrilli, Jr. and Christopher J. Wright argued the cause for CLEC petitioners. With them on the briefs were Mark D. Schneider, Marc A. Goldman, Michael B. DeSanctis, William Single IV, Jeffrey A. Rackow, David W. Carpenter, David L. Lawson, C. Frederick Beckner III, An- drew D. Lipman, Russell M. Blau, Richard M. Rindler, Patrick J. Donovan, Harisha J. Bastiampillai, Dennis D. Ahlers, Steven A. Augustino, Albert H. Kramer, Jonathan E. Canis, Robert J. Aamoth, Carl S. Nadler, Adelia S. Borrasca, Jason D. Oxman, Timothy J. Simeone, Charles C. Hunter, Catherine M. Hannan, Genevieve Morelli, Glenn B. Manish- in, Jonathan E. Canis, Teresa K. Gaugler, Jonathan Jacob Nadler, and Jonathan D. Lee. Jennifer M. Kashatus, Paul J. Rebey, Eric J. Branfman, Joshua M. Bobeck, and Angela M. Simpson entered appearances. James Bradford Ramsay argued the cause for State peti- tioners. With him on the briefs were Grace Delos Reyes, Jonathan Feinberg, John L. Favreau, John C. Graham, Helen M. Mickiewicz, Gretchen T. Dumas, Maryanne Reyn- olds Martin, Christopher C. Kempley, Maureen A. Scott, Michael A. Cox, Attorney General, Attorney General’s Office of the State of Michigan, Thomas L. Casey, Solicitor General, 3 and David A. Voges and Michael Nickerson, Assistant Attor- ney Generals. David C. Bergmann, Irwin A. Popowsky, Philip F. McClel- land, Patricia A. Smith, Billy Jack Gregg, and F. Anne Ross were on the briefs for petitioner National Association of State Utility Consumer Advocates. John E. Ingle, Deputy Associate General Counsel, Federal Communications Commission, and James M. Carr, Counsel, argued the cause for respondents. With them on the brief were R. Hewitt Pate, Assistant Attorney General, U.S. De- partment of Justice, Catherine G. O’Sullivan and Nancy C. Garrison, Attorneys, John A. Rogovin, General Counsel, Federal Communications Commission, and Laurence N. Bourne, Joel Marcus and Christopher L. Killion, Counsel. Andrea Limmer, Attorney, U.S. Department of Justice, and Lisa S. Gelb, Counsel, Federal Communications Commission, entered appearances. Michael K. Kellogg argued the cause for ILEC intervenors and Catena Networks, Inc. in support of respondents. With him on the brief were Mark L. Evans, Aaron M. Panner, Michael T. McMenamin, James D. Ellis, Paul K. Mancini, Joseph E. Cosgrove, Jr., Gary L. Phillips, James P. Lamour- eux, Robert B. McKenna, Charles R. Morgan, James G. Harralson, William P. Barr, Michael E. Glover, Edward Shakin, and Stephen L. Goodman. Alfred G. Richter, Hope E. Thurrott, Lawrence E. Sarjeant, and Jonathan E. Canis entered appearances. David W. Carpenter argued the cause for CLEC interve- nors in support of respondents. With him on the brief were Donald B. Verilli, Jr., Mark D. Schneider, Marc A. Gold- man, Michael B. DeSanctis, William Single IV, Jeffrey A. Rackow, David L. Lawson, C. Frederick Beckner III, Teresa K. Gaugler, Charles C. Hunter, Catherine M. Hannan, An- drew D. Lipman, Russell M. Blau, Richard M. Rindler, Patrick J. Donovan, Harisha J. Bastiampillai, Albert H. Kramer, Jonathan D. Lee, Carl S. Nadler, Adelia S. Borras- ca, Janson D. Oxman, Robert J. Aamoth, Genevieve Morelli, John T. Nakahata, Sara F. Leibman, John J. Heitmann, 4 Jennifer M. Kashatus, Christopher J. Wright, and Timothy J. Simeone. Roy E. Hoffinger, Charles J. Cooper, Hamish P. Hume, and Richard J. Metzger entered appearances. Jonathan Feinberg, John L. Favreau, John C. Graham, Helen M. Mickiewicz, Gretchen T. Dumas, Maryanne Reyn- olds Martin, Christopher C. Kempley, Maureen A. Scott, Michael A. Cox, Attorney General, Attorney General’s Office of the State of Michigan, Thomas L. Casey, Solicitor General, David A. Voges and Michael Nickerson, Assistant Attorney Generals, James Bradford Ramsay, and Grace Delos Reyes were on the brief for State intervenors in support of respon- dents. Laura H. Philips, Douglas G. Bonner, Michael F. McBride, Thomas J. Sugrue, Howard J. Symons, Sara F. Leibman, and Douglas I. Brandon were on the brief of Wireless intervenors in support of respondent. Brian A. Coleman entered an appearance. Before: EDWARDS and RANDOLPH, Circuit Judges, and WILLIAMS, Senior Circuit Judge. Opinion for the Court filed by Senior Circuit Judge WILLIAMS. Table of Contents I. Legal Background TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTT 6 II. ILEC ObjectionsTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTT 11 A. Unbundling of Mass Market SwitchesTTTTTTTTTT 11 1. Subdelegation of § 251(d)(2) impairment determinations to state commissions TTTTTT 12 2. Impairment in provision of mass market switchingTTTTTTTTTTTTTTTTTTTTTTTTTTTTTT 18 3. The Commission’s definition of ‘‘impair- ment’’TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTT 22 B. Unbundling of High–Capacity Dedicated Transport Facilities TTTTTTTTTTTTTTTTTTTTTTTT 26 1. Unlawfulness of the delegation to the states and the national impairment finding TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTT 26 2. Remaining dedicated transport issues TTTTTT 28 5 a. Route-specific analysis of dedicated transport TTTTTTTTTTTTTTTTTTTTTTTTTT 28 b. Wireless providers’ access to unbun- dled dedicated transport TTTTTTTTTTTTT 29 C. Network Modification RequirementsTTTTTTTTTTT 33 III. CLEC Objections TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTT 34 A. Unbundling of Broadband Loops TTTTTTTTTTTTTT 34 1. Hybrid loops TTTTTTTTTTTTTTTTTTTTTTTTTTTT 35 2. Fiber-to-the-home (‘‘FTTH’’) loops TTTTTTTTT 42 3. Line sharing TTTTTTTTTTTTTTTTTTTTTTTTTTTT 44 B. Exclusion of ‘‘Entrance Facilities’’ TTTTTTTTTTTTT 46 C. Unbundling of Enterprise Switches TTTTTTTTTTTT 47 D. Unbundling of Call–Related Databases and Signaling Systems TTTTTTTTTTTTTTTTTTTTTTTTT 49 E. Unbundling of Shared Transport Facilities TTTT 50 F. Section 271 Pricing and Combination RulesTTTT 51 IV. Unbundling of Enhanced Extended Links (‘‘EELs’’)TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTT 54 A. The Qualifying Service/Non–Qualifying Ser- vice DistinctionTTTTTTTTTTTTTTTTTTTTTTTTTTTT 56 B. The EEL Eligibility Criteria TTTTTTTTTTTTTTTTT 58 V. Miscellaneous TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTT 59 A. NASUCA’s Standing TTTTTTTTTTTTTTTTTTTTTTTT 59 B. Ripeness of the State Preemption ClaimsTTTTTTT 60 VI. Conclusion TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTT 61 WILLIAMS, Senior Circuit Judge: The Telecommunications Act of 1996, Pub. L. 104–104, 110 Stat. 56, codified at 47 U.S.C. § 151 et seq. (the ‘‘Act’’), sought to foster a competi- tive market in telecommunications. To enable new firms to enter the field despite the advantages of the incumbent local exchange carriers (‘‘ILECs’’), the Act gave the Federal Com- munications Commission broad powers to require ILECs to make ‘‘network elements’’ available to other telecommunica- tions carriers, id. §§ 251(c)(3),(d), most importantly the com- petitive local exchange carriers (‘‘CLECs’’). The most obvi- ous candidates for such obligatory provision were the copper wire loops historically used to carry telephone service over the ‘‘last mile’’ into users’ homes. But Congress left to the 6 Commission the choice of elements to be ‘‘unbundled,’’ speci- fying that in doing so it was to consider, at a minimum, whether TTT the failure to provide access to such network elements would impair the ability of the telecommunications carrier seeking access to provide the services that it seeks to offer. Id. § 251(d)(2) (emphasis added). The Act became effective on February 8, 1996, a little more than eight years ago. Twice since then the courts have faulted the Commission’s efforts to identify the elements to be unbundled. The Supreme Court invalidated the first effort in AT&T Corp. v. Iowa Utilities Board, 525 U.S. 366, 389–90 (1999) (‘‘AT&T’’). We invalidated much of the second effort (including separately adopted ‘‘line-sharing’’ rules) in United States Telecom Association v. FCC, 290 F.3d 415 (D.C. Cir. 2002) (‘‘USTA I’’). The Commission consolidated our remand in that case with its ‘‘triennial review’’ of the scope of obligatory unbundling and issued the Order on review here. See Report and Order and Order on Remand and Further Notice of Proposed Rulemaking, Review of the Section 251 Unbundling Obligations of Incumbent Local Exchange Carriers, CC Docket Nos. 01–338 et al., FCC 03– 36, 18 FCC Rcd 16978 (Aug. 21, 2003) (‘‘Order’’); Errata, Review of the Section 251 Unbundling Obligations of Incum- bent Local Exchange Carriers, CC Docket Nos. 01–338 et al., FCC 03–227, 18 FCC Rcd 19020 (Sep. 17, 2003). Again, regrettably, much of the resulting work is unlawful. After a brief summary of the legal background, we address first the ILECs’ claims, then the CLECs’ claims, then the ILEC and CLEC claims relating to a special area, enhanced extended links (‘‘EELs’’), and finally a couple of miscellane- ous claims. I. Legal Background Section 251(c)(3) of the Act imposes on each ILEC the duty to provide any requesting telecommunications carrier with 7 access to network elements on an unbundled basis at any technically feasible point on rates, terms, and conditions that are just, reasonable, and nondiscriminatory in accor- dance with TTT the requirements of this section and section 252 of this title. 47 U.S.C. § 251(c)(3). The statute says that the ILECs may charge a ‘‘just and reasonable rate’’ for these unbundled network elements (‘‘UNEs’’), see id. § 252(d)(1), and the Commission adopted as its standard ‘‘total element long-run incremental cost,’’ or ‘‘TELRIC.’’ Under this criterion UNE prices are to be ‘‘based on the use of the most efficient telecommunications technology currently available and the lowest cost network configuration, given the existing location of the incumbent LEC’s wire centers.’’ 47 CFR § 51.505(b)(1). In litigation over this pricing rule, which the Supreme Court upheld in Verizon Communications v. FCC, 535 U.S. 467 (2002) (‘‘Veri- zon’’), it appears to have been common ground that, because of ongoing technological improvement (among other things), prices so determined would fall well below the costs the ILECs had actually historically incurred in constructing the elements. Id. at 503–04, 508–09. Certainly the ardent pref- erences of the parties as to the scope of the Act’s unbundling requirements—the ILECs seeking a narrow reading, the CLECs seeking a broad one—suggest such a relationship. In its first effort to interpret the ‘‘impairment’’ standard of § 251(d)(2), the Commission held that lack of unbundled access to an element would ‘‘impair’’ a CLEC’s ability to provide telecommunications service ‘‘if the quality of the service the entrant can offer, absent access to the requested element, declines and/or the cost of providing the service rises.’’ Implementation of the Local Competition Provisions in the Telecommunications Act of 1996, First Report and Order, CC Docket No. 96–98, 11 FCC Rcd 15499, 15643 (1996) (‘‘First Report and Order’’), ¶ 285. The Supreme Court found this reading of ‘‘impair’’ unrea- sonable in two respects. First, the Commission had irration- ally refused to consider whether a CLEC could self-provision 8 or acquire the requested element from a third party. AT&T, 525 U.S. at 389. Second, the Commission had considered any increase in cost or decrease in quality, no matter how small, sufficient to establish impairment—a result the Court con- cluded could not be squared with the ‘‘ordinary and fair meaning’’ of the word ‘‘impair.’’ Id. at 389–90 & n.11. The Court admonished the FCC that in assessing which cost differentials would ‘‘impair’’ a new entrant’s competition with- in the meaning of the statute, it must ‘‘apply some limiting standard, rationally related to the goals of the Act.’’ Id. at 388. Responding to the AT&T decision, the Commission adopted a new interpretation under which a would-be entrant is ‘‘impaired’’ if, ‘‘taking into consideration the availability of alternative elements outside the incumbent’s network, includ- ing self-provisioning by a requesting carrier or acquiring an alternative from a third-party supplier, lack of access to that element materially diminishes a requesting carrier’s ability to provide the services it seeks to offer.’’ Implementation of the Local Competition Provisions of the Telecommunications Act of 1996, Third Report and Order and Fourth Further Notice of Proposed Rulemaking, 15 FCC Rcd 3696, 3725 (1999) (‘‘Third Report and Order’’), ¶ 51 (emphasis added). But in USTA I we held that this new interpretation of ‘‘impairment,’’ while an improvement, was still unreasonable in light of the Act’s underlying purposes. The fundamental problem, we held, was that the Commis- sion did not differentiate between those cost disparities that a new entrant in any market would be likely to face and those that arise from market characteristics ‘‘linked (in some de- gree) to natural monopoly TTT that would make genuinely competitive provision of an element’s function wasteful.’’ USTA I, 290 F.3d at 427. This distinction between different kinds of incumbent/entrant cost differentials is qualitative, not merely quantitative, which is why the Commission’s addition of a requirement that the cost disparity be ‘‘material’’ was inadequate. Id. at 427–28. 9 We also made clear that the Commission’s broad and analytically insubstantial concept of impairment failed to pur- sue the ‘‘balance’’ between the advantages of unbundling (in terms of fostering competition by different firms, even if they use the very same facilities) and its costs (in terms both of ‘‘spreading the disincentive to invest in innovation and creat- ing complex issues of managing shared facilities,’’ id. at 427), a balance that we found implicit in the AT&T Court’s insis- tence on an unbundling standard ‘‘rationally related to the goals of the Act,’’ id. at 428 (quoting AT&T). We also objected to the Commission’s decision to issue, with respect to most elements, broad unbundling require- ments that would apply ‘‘in every geographic market and customer class, without regard to the state of competitive impairment in any particular market.’’ USTA I, 290 F.3d at 422. Though the Act does not necessarily require the Com- mission to determine ‘‘on a localized state-by-state or market- by-market basis which unbundled elements are to be made available,’’ id. at 425 (quoting Third Report and Order, 15 FCC Rcd at 3753, ¶ 122), it does require ‘‘a more nuanced concept of impairment than is reflected in findings TTT de- tached from any specific markets or market categories.’’ USTA I, 290 F.3d at 426. Thus, the Commission is obligated to establish unbundling criteria that are at least aimed at tracking relevant market characteristics and capturing signifi- cant variation. Finally, we vacated the Commission’s decision to require ILECs to unbundle the high-frequency portion of their cop- per loops to requesting CLECs—a practice known as ‘‘line sharing’’ and used by CLECs to provide broadband DSL service—because the Commission had failed to consider ade- quately whether intermodal competition from cable providers tilted the balance against this form of unbundling in the broadband market. In response to USTA I the Commission again revised its definition of impairment. This time around, the Commission determined that a CLEC would ‘‘be impaired when lack of access to an incumbent LEC network element poses a barrier 10 or barriers to entry, including operational and economic barri- ers, that are likely to make entry into a market uneconomic. That is, we ask whether all potential revenues from entering a market exceed the costs of entry, taking into consideration any countervailing advantages that a new entrant may have.’’ Order ¶ 84 (emphasis added). The Commission clarified that the impairment assessment would take intermodal competi- tion into account. Id. ¶ ¶ 97–98. The Commission responded to our demand for a more ‘‘nuanced’’ application of the impairment standard by purport- ing to adopt a ‘‘granular’’ approach that would consider ‘‘such factors as specific services, specific geographic locations, the different types and capacities of facilities, and customer and business considerations.’’ Id. ¶ 118. Where the Commission believed that the record could not support an absolute nation- al impairment finding but at the same time contained too little information to make ‘‘granular’’ determinations, it adopted a provisional nationwide rule, subject to the possibili- ty of specific exclusions, to be created by state regulatory commissions under a purported delegation of the Commis- sion’s own authority. The Commission also resolved to use the ‘‘at a minimum’’ language in § 251(d)(2) to ‘‘inform [its] consideration of un- bundling in contexts where some level of impairment may exist, but unbundling appeared likely to undermine important goals of the 1996 Act.’’ Id. ¶ 173. Specifically, in connection with two broadband elements, ‘‘fiber-to-the-home’’ (‘‘FTTH’’) and hybrid loops (see below), it brought into the balance the risk that an unbundling order might deter investment in such facilities—contrary, as it saw the matter, to the statutory goal of encouraging prompt deployment of ‘‘advanced telecommu- nications capability.’’ Id. ¶ ¶ 172–73 (quoting § 706 of the Act). Additional issues also emerged in the rulemaking and will be addressed below. The ILECs filed two mandamus petitions with this Court, arguing that the Order violated our decision in USTA I, and in addition filed a petition for review here. Various CLECs, state commissions, and an association of state utility consum- 11 er advocates filed petitions for review in several other cir- cuits; these petitions were transferred to the Eighth Circuit under the random lottery procedure established in 28 U.S.C. § 2112(a)(3), and then transferred to this court by the Eighth Circuit under 28 U.S.C. § 2112(a)(5). We consolidated the petitions for review with the mandamus petitions. II. ILEC Objections A. Unbundling of Mass Market Switches The Commission made a nationwide finding that CLECs are impaired without unbundled access to ILEC switches for the ‘‘mass market,’’ consisting of residential and relatively small business users. This finding was based primarily on the costs associated with ‘‘hot cuts’’ (discussed below), which must be performed when a CLEC provides its own switch. Order ¶ ¶ 464–75. But the Commission, apparently con- cerned that a blanket nationwide impairment determination might be unlawfully overbroad in light of the record evidence of substantial market-by-market variation in hot cut costs, delegated authority to state commissions to make more ‘‘nu- anced’’ and ‘‘granular’’ impairment determinations. First, the Commission directed the state commissions to eliminate unbundling if a market contained at least three competitors in addition to the ILEC, id. ¶ ¶ 498–503, or at least two non-ILEC third parties that offered access to their own switches on a wholesale basis, id. ¶ ¶ 504–05. For pur- poses of this exercise the Commission gave the states virtual- ly unlimited discretion over the definition of the relevant market. Id. ¶ ¶ 495–97. Second, where these ‘‘competitive triggers’’ are not met, the Commission instructed the states to consider whether, despite the many economic and opera- tional entry barriers deemed relevant by the Commission, competitive supply of mass market switching was neverthe- less feasible. Id. ¶ ¶ 494, 506–20. The Commission also instructed the states to explore specific mechanisms to ame- liorate or eliminate the costs of the ‘‘hot cut’’ process. Id. ¶ ¶ 486–90. The Commission mentioned, for example, the possible use of ‘‘rolling’’ hot cuts, a process in which CLECs 12 could use ILEC switches for some time after a customer selected the CLEC as its provider, and after an accumulation of such customer changes, the ILEC would make all the necessary hot cuts in one fell swoop. Id. ¶ ¶ 463, 521–24. If a state failed to perform the requisite analysis within nine months, the Commission would step into the position of the state commission and do the analysis itself. Id. ¶ 190. Final- ly, the Order provided that a party ‘‘aggrieved’’ by a state commission decision could seek a declaratory ruling from the Commission, though with no assurance when, or even wheth- er, the Commission might respond. Id. ¶ 426; see also 47 CFR § 1.2. We consider first whether the Commission’s subdelegation of authority to the state commissions is lawful. We conclude that it is not. We then consider whether the Commission’s nationwide impairment determination can nevertheless sur- vive, even without the safety valve provided by subdelegation to the states. We conclude that it cannot. We therefore vacate the Commission’s decision to order unbundling of mass market switches, subject to the stay discussed in Part VI. 1. Subdelegation of § 251(d)(2) impairment determina- tions to state commissions The FCC acknowledges that § 251(d)(2) instructs ‘‘the Commission’’ to ‘‘determine[ ]’’ which network elements shall be made available to CLECs on an unbundled basis. But it claims that agencies have the presumptive power to subdele- gate to state commissions, so long as the statute authorizing agency action refrains from foreclosing such a power. Given the absence of any express foreclosure, the Commission ar- gues that its interpretation of the statute on the matter of subdelegation is entitled to deference under Chevron U.S.A. v. Natural Resources Defense Council, 467 U.S. 837 (1984). And it claims that its interpretation is reasonable given the state commissions’ independent jurisdiction over the general subject matter, the magnitude of the regulatory task, and the need for close cooperation between state and federal regu- lators in this area. 13 The Commission’s position is based on a fundamental mis- reading of the relevant case law. When a statute delegates authority to a federal officer or agency, subdelegation to a subordinate federal officer or agency is presumptively per- missible absent affirmative evidence of a contrary congres- sional intent. See United States v. Giordano, 416 U.S. 505, 512–13 (1974); Fleming v. Mohawk Wrecking & Lumber Co., 331 U.S. 111, 121–22 (1947); Halverson v. Slater, 129 F.3d 180, 185–86 (D.C. Cir. 1997); United States v. Mango, 199 F.3d 85, 90–91 (2d Cir. 1999); Inland Empire Pub. Lands Council v. Glickman, 88 F.3d 697, 702 (9th Cir. 1996); United States v. Widdowson, 916 F.2d 587, 592 (10th Cir. 1990), vacated on other grounds, 502 U.S. 801 (1991). But the cases recognize an important distinction between subdelegation to a subordinate and subdelegation to an outside party. The presumption that subdelegations are valid absent a showing of contrary congressional intent applies only to the former. There is no such presumption covering subdelegations to outside parties. Indeed, if anything, the case law strongly suggests that subdelegations to outside parties are assumed to be improper absent an affirmative showing of congression- al authorization. See Shook v. District of Columbia Fin. Responsibility & Mgmt Assistance Auth., 132 F.3d 775, 783– 84 & n.6 (D.C. Cir. 1998). See also Nat’l Ass’n of Reg. Util. Comm’rs (‘‘NARUC’’) v. FCC, 737 F.2d 1095, 1143–44 & n.41 (D.C. Cir. 1984); Nat’l Park and Conservation Ass’n v. Stanton, 54 F. Supp. 2d 7, 18–20 (D.D.C. 1999). (We discuss below some cases that might, mistakenly, be thought to support a contrary view.) This distinction is entirely sensible. When an agency delegates authority to its subordinate, responsibility—and thus accountability—clearly remain with the federal agency. But when an agency delegates power to outside parties, lines of accountability may blur, undermining an important demo- cratic check on government decision-making. See NARUC, 737 F.2d at 1143 n.41; cf. Printz v. United States, 521 U.S. 898, 922–23 (1997). Also, delegation to outside entities in- creases the risk that these parties will not share the agency’s ‘‘national vision and perspective,’’ Stanton, 54 F. Supp. 2d at 14 20, and thus may pursue goals inconsistent with those of the agency and the underlying statutory scheme. In short, sub- delegation to outside entities aggravates the risk of policy drift inherent in any principal-agent relationship. The fact that the subdelegation in this case is to state commissions rather than private organizations does not alter the analysis. Although United States v. Mazurie, 419 U.S. 544 (1975), noted that ‘‘limits on the authority of Congress to delegate its legislative power TTT are [ ] less stringent in cases where the entity exercising the delegated authority itself possesses independent authority over the subject mat- ter,’’ id. at 556–57 (emphasis added), that decision has no application here: it involved a constitutional challenge to an express congressional delegation, rather than an administra- tive subdelegation, and the point of the discussion was to distinguish the still somewhat suspect case of congressional delegation to purely private organizations. Two Ninth Circuit cases have invoked Mazurie to suggest that limitations on an administrative agency’s power to sub- delegate might be less stringent if the delegee is a sovereign entity rather than a private group. See Assiniboine & Sioux Tribes v. Bd. of Oil and Gas, 792 F.2d 782, 795 (9th Cir. 1986); Southern Pacific Transp. Co. v. Watt, 700 F.2d 550, 556 (9th Cir. 1983). But in neither of these cases was this principle necessary to the outcome, and in neither did the court seek to justify the extension of Mazurie from its context—the validity of an express delegation of Congress’s powers. We therefore hold that, while federal agency officials may subdelegate their decision-making authority to subordinates absent evidence of contrary congressional intent, they may not subdelegate to outside entities—private or sovereign— absent affirmative evidence of authority to do so. The Commission’s plea for Chevron deference is unavailing. A general delegation of decision-making authority to a federal administrative agency does not, in the ordinary course of things, include the power to subdelegate that authority be- yond federal subordinates. It is clear here that Congress has 15 not delegated to the FCC the authority to subdelegate to outside parties. The statutory ‘‘silence’’ simply leaves that lack of authority untouched. In other words, the failure of Congress to use ‘‘Thou Shalt Not’’ language doesn’t create a statutory ambiguity of the sort that triggers Chevron defer- ence. See Ry. Labor Exec. Ass’n v. Nat. Mediation Bd., 29 F.3d 655, 671 (D.C. Cir. 1994) (‘‘Were courts to presume a delegation of power absent an express withholding of such power, agencies would enjoy virtually limitless hegemony, a result plainly out of keeping with Chevron and quite likely with the Constitution as well.’’); see also Aid Ass’n for Lutherans v. U.S. Postal Service, 321 F.3d 1166, 1174–75 (D.C. Cir. 2003); Motion Picture Ass’n of Am. v. FCC, 309 F.3d 796, 801 (D.C. Cir. 2002); Ethyl Corp. v. EPA, 51 F.3d 1053, 1060 (D.C. Cir. 1995). The FCC invokes a number of other cases in support of its idea of a presumptive authority to subdelegate to entities other than subordinates. These are inapposite because they do not involve subdelegation of decision-making authority. They merely recognize three specific types of legitimate outside party input into agency decision-making processes: (1) establishing a reasonable condition for granting federal approval; (2) fact gathering; and (3) advice giving. The scheme established in the Order fits none of these models. First, a federal agency entrusted with broad discretion to permit or forbid certain activities may condition its grant of permission on the decision of another entity, such as a state, local, or tribal government, so long as there is a reasonable connection between the outside entity’s decision and the federal agency’s determination. Thus in United States v. Matherson, 367 F. Supp. 779, 782–83 (E.D.N.Y. 1973), aff’d 493 F.2d 1339 (2d Cir. 1974), the court upheld the decision of the Fire Island National Seashore Superintendent to condi- tion issuance of federal seashore motor vehicle permits on the applicant’s acquisition of an analogous permit from an adja- cent town. And Southern Pacific, 700 F.2d at 556, citing Matherson, sustained the Secretary of Interior’s conditioning of right-of-way permits across tribal lands on the tribal government’s approval. In contrast to these cases, where an 16 agency with broad permitting authority had adopted an obvi- ously relevant local concern as an element of its decision process, the Commission here has delegated to another actor almost the entire determination of whether a specific statuto- ry requirement—impairment—has been satisfied. Second, there is some authority for the view that a federal agency may use an outside entity, such as a state agency or a private contractor, to provide the agency with factual infor- mation. While Assiniboine & Sioux Tribes found that a delegation of decision-making power to a state board would be unlawful, it left open whether reliance by the federal agency on the state board for ‘‘nondiscretionary activities such as compiling, hearing, and transmitting technical infor- mation might not be permissible and desirable.’’ 792 F.2d at 795. And National Association of Psychiatric Treatment v. Mendez, 857 F. Supp. 85, 91 (D.D.C. 1994), upheld a federal certifying agency’s decision to hire a private contractor to conduct surveys of residential treatment centers and pass its results on to the agency, which retained final certification authority. While the FCC has sought to characterize the state commissions’ role here as fact finding, see Order ¶ ¶ 186, 493, in fact the Order lets the states make crucial decisions regarding market definition and application of the FCC’s general impairment standard to the specific circumstances of those markets, with FCC oversight neither timely nor as- sured. The Commission’s attempted punt does not remotely resemble nondiscretionary information gathering. Our own decision in Tabor v. Joint Board for Enrollment of Actuaries, 566 F.2d 705, 708 n.5 (D.C. Cir. 1977), seems to straddle the two above variants of permissible relationships. There the federal Joint Board for Enrollment of Actuaries, exercising its broad discretion to set conditions for certifying actuaries to administer ERISA pension plans, required appli- cants either to pass a Board exam or to pass an exam administered by one of the recognized private national actuar- ial societies. 566 F.2d at 708 n.5. The court found that the process was ‘‘superintended by the Board in every respect,’’ and that the Board had not abdicated its decision-making authority but merely created a reasonable ‘‘short-cut,’’ contin- 17 gent on the approval of certain private organizations, to satisfy one of the Board’s own regulatory requirements. Id. The opinions in both Southern Pacific (from our first catego- ry) and Mendez (from our second) invoke Tabor. Neither Tabor nor its progeny relied on any principle that subdelegations to outside parties were presumptively valid, since the result in each of these cases was supportable on the theory that no subdelegation of decision-making authority had actually taken place. To the extent that Tabor’s citation of United States v. Giordano, 416 U.S. 505, 512–13 (1974), might be thought to suggest that external delegations enjoy the same favorable presumption as internal ones, that suggestion was clearly rejected by our decision in Shook, 132 F.3d at 783–84 & n.6. Third, a federal agency may turn to an outside entity for advice and policy recommendations, provided the agency makes the final decisions itself. Thus in Shook, 132 F.3d at 784, we disapproved the D.C. Control Board’s delegation of governance powers over D.C. schools to a private Board of Trustees, but we suggested that the Control Board could use an entity of that sort ‘‘as an advisory board charged with recommending certain actions and policies to the Control Board.’’ See also Stanton, 54 F. Supp. 2d at 19–20 & n.6; Mendez, 857 F. Supp. at 91. An agency may not, however, merely ‘‘rubber-stamp’’ decisions made by others under the guise of seeking their ‘‘advice,’’ see Assiniboine & Sioux Tribes, 792 F.2d at 795, nor will vague or inadequate asser- tions of final reviewing authority save an unlawful subdelega- tion, see Stanton, 54 F. Supp. 2d at 19, 20–21. Finally, the Commission’s claim that Diamond Internation- al Corp. v. FCC, 627 F.2d 489, 492–93 (D.C. Cir. 1980), and New York Telephone Co. v. FCC, 631 F.2d 1059, 1065 (2d Cir. 1980), uphold ‘‘virtually indistinguishable’’ FCC subdel- egations to state commissions, FCC Br. at 25, is (or should be) embarrassing. These cases involved a wholly unrelated issue: whether the FCC properly interpreted the Communi- cations Act when it decided to permit carriers to file state tariffs for local services used in connection with interstate 18 services. The issue was not delegation of federal authority but rather the scope of federal authority to preempt state authority. We note that the ILEC petitioners invoke standard expres- sio unius reasoning to attack the delegation. They point out that other provisions of the Act—e.g., the procedures for arbitration and approval of agreements under § 252—ex- pressly specify a state role, and urge us to infer congressional preclusion of such a role under § 251(d)(2). We do not rely on this theory. Our conclusion would be unchanged if no provision of the Act mentioned any role for the state commis- sions, because the general conferral of regulatory authority does not empower an agency to subdelegate to outside par- ties. That said, the fact that other provisions of the statute carefully delineate a particular role for the state commissions, but § 251(d)(2) does not, reassures us that the our result is consistent with congressional intent. We therefore vacate, as an unlawful subdelegation of the Commission’s § 251(d)(2) responsibilities, those portions of the Order that delegate to state commissions the authority to determine whether CLECs are impaired without access to network elements, and in particular we vacate the Commis- sion’s scheme for subdelegating mass market switching deter- minations. (This holding also requires that we vacate the Commission’s subdelegation scheme with respect to dedicated transport elements, discussed below.) We now turn to whether, without that safety valve, the FCC’s national impair- ment findings for mass market switches can be reconciled with USTA I. 2. Impairment in provision of mass market switching Without the (unlawful) innovation of transforming a nation- al impairment finding into a provisional national impairment finding from which state commissions could deviate if they found no impairment under local market conditions, the FCC’s Order on mass market switches must stand or fall as a nationwide determination that CLECs are impaired in the mass market without unbundled access to ILEC switches. After reviewing the record, we conclude that we must vacate 19 the (no longer provisional) national impairment finding as inconsistent with our conclusion in USTA I that the Commis- sion may not ‘‘loftily abstract[ ] away from all specific mar- kets,’’ 290 F.3d at 423, but must instead implement a ‘‘more nuanced concept of impairment,’’ id. at 426. The Commission’s national finding of impairment for mass market switches is based on entry barriers related to the need for ILECs to perform ‘‘hot cuts’’ (manual connections) for CLECs if the latter choose to self-provision mass market switches. See Order ¶ ¶ 459, 464–76. A ‘‘hot cut’’ requires an ILEC technician to physically disconnect a customer loop from the ILEC switch (to which the loop was hard-wired) and re-wire the loop to the CLEC switch, while simultaneously reassigning the customer’s phone number from the ILEC switch to the CLEC switch. Order ¶ 465 n.1409. A hot cut must be performed every time a CLEC seeks to connect a new customer. In contrast, ILEC connection of a customer generally only requires a software change (unless the custom- er had already switched to a CLEC switch, in which case the hot cut must be undone via the same physical re-connection). Order ¶ 465. The Commission explains that, according to evidence in the record, the need to perform hot cuts can delay a CLEC in providing service with its own switch and can cause service disruptions, and that these delays and disrup- tions, even if minor, can damage customer perceptions of CLEC service and impede the CLECs’ ability to compete. Order ¶ ¶ 466–67. Though the Commission in its brief alludes to ‘‘other opera- tional and economic factors’’ that might create barriers to competition in mass market switching, FCC Br. at 36, the Order makes clear that the national impairment finding was based solely on hot cuts. Order ¶ ¶ 459 n.1405 & 476. (The other factors were to be considered by state commissions in the exercise of the unlawfully delegated authority.) There appears to be no suggestion that mass market switches exhibit declining average costs in the relevant markets, or even that switches entail large sunk costs. The Commission nonetheless concluded that hot cut costs are not the sort of cost disparity that a new entrant into any market might face, 20 since they arise due to the fact that ‘‘incumbent LECs’ networks were designed for use in a single carrier, non- competitive environment,’’ which means that CLECs face operational costs that the ILECs do not. Order ¶ 465. Though certain sections of the Order suggest that impair- ment due to hot cut costs might be sufficiently widespread to support a general national impairment finding even in the absence of more ‘‘nuanced’’ determinations to be made by the state commissions, Order ¶ ¶ 459, 470, 473, the Commission at other points concludes that a national finding, without the possibility of market-specific exceptions authorized by state commissions, would be inconsistent with USTA I. See Order ¶ ¶ 186–88, 196, 425, 485, 493. At the very least, these latter passages demonstrate that the Commission’s own conclusions do not clearly support a non-provisional national impairment finding for mass market switches, and thus require us to vacate and remand. Moreover, we doubt that the record supports a national impairment finding for mass market switches. In another context the Commission has already addressed a kindred issue. Under § 271 of the Act, the subset of ILECs that used to be operating companies of AT&T before its break-up (the Bell Operating Companies, or ‘‘BOCs’’) can enter the interLATA market (the market for calls between different local access and transport areas) only by showing, among other things, that they are providing CLECs adequate un- bundled access to various network elements, including local loops. See Act § 271(c)(2)(B)(iv). The Commission acknowl- edges that in that context it has in fact found that the BOCs were doing so ‘‘in the quantities that competitors demand and at an acceptable level of quality,’’ see, e.g., Memorandum Opinion and Order, Application by SBC Communications, Inc., et al., Pursuant to Section 271 of the Telecommunica- tions Act of 1996 To Provide In–Region, InterLATA Services in Texas, 15 FCC Rcd 18354, 18480 (2000), ¶ 247; Memoran- dum Opinion and Order, Application of Ameritech Michigan Pursuant to Section 271 of the Communications Act of 1934, as Amended, To Provide In–Region, InterLATA Services in Michigan, 12 FCC Rcd 20543, 20601–02 (1997), ¶ 110. In 21 none of those proceedings did the Commission find the hot cut process inadequate to meet this standard. See Separate Statement of Chairman Michael K. Powell Approving in Part and Dissenting in Part, FCC 03–36 (‘‘Powell Statement’’) at 4. But it distinguished those cases on the ground of uncertainty about whether ILECs would be able to handle the increases in hot cut demand that would flow from denying CLECs access to switches as UNEs. Order ¶ 469 & n.1435. The ILECs contend that in fact hot cut processes are ‘‘scalable,’’ so that existing sufficiency can be projected onto larger-scale usage. See ILEC Br. at 16 (citing Powell Statement at 5; Memorandum Opinion and Order, Application by Bell Atlan- tic New York for Authorization Under Section 271 of the Communications Act to Provide In–Region, InterLATA Ser- vice in the State of New York, 15 FCC Rcd 3953, 4114 (1999), ¶ 308). The record on the matter is mixed, perhaps sufficiently so that the Commission’s ‘‘provisional’’ assumption to the con- trary might be sustainable as an absolute finding, given the deference we would owe the Commission’s predictive judg- ment and the inevitability of some over- and under- inclusiveness in the Commission’s unbundling rules. But the Commission implicitly conceded that hot cut difficulties could not support an undifferentiated nationwide impairment find- ing. Order ¶ ¶ 425, 485, 493. Moreover, we made clear in USTA I that the Commission cannot proceed by very broad national categories where there is evidence that markets vary decisively (by reference to its impairment criteria), at least not without exploring the possibility of more nuanced alterna- tives and reasonably rejecting them. 290 F.3d at 425–26. One can imagine the Commission successfully identifying criteria based, for example, on an ILEC’s track record for speed and volume in a market, integrated with some projec- tion of the demand increase that would result from withhold- ing of switches as UNEs. The Commission, however, has made no visible effort to explore such possibilities. Additionally, the ILEC petitioners suggested several more narrowly-tailored alternatives to a blanket requirement that mass market switches be made available as UNEs. Consid- ering such narrower alternatives is essential in light of our admonition in USTA I that the Commission must balance the 22 costs and benefits of unbundling. 290 F.3d at 429. ‘‘Rolling’’ hot cuts are one such proffered alternative. Under that concept the Commission could require unbundled access to ILEC switching on new lines for 90 days (or some other period of time) in order to give the ILEC time to perform the accumulated backlog of hot cuts simultaneously, Order ¶ ¶ 463, 521–24, or the Commission could require the ILEC to provide unbundled access to its switch only until it was able to perform the hot cut. The FCC’s only real answer to these proposed alternatives, at least the only answer that appears in the Order or the FCC’s brief, is that the Commission directed the state commissions to consider these alternatives and to implement them if they would remedy impairment. See FCC Br. at 38–39; Order ¶ ¶ 463, 521–24. But since we have held such subdelegation unlawful, that response is un- available. Moreover, even if the FCC had adopted some lawful mech- anism for making exemptions from its general national rule, it could not necessarily rely on the existence of that mechanism as the sole justification for not adopting a more narrowly tailored rule. While a rational rule that would otherwise be impermissibly broad can be saved by ‘‘safety valve’’ waiver or exception procedures, the mere existence of a safety valve does not cure an irrational rule. See ICORE, Inc. v. FCC, 985 F.2d 1075, 1080 (D.C. Cir. 1993); Alltel Corp. v. FCC, 838 F.2d 551, 561–62 (D.C. Cir. 1988). And a rule is irrational in this context if a party has presented to the agency a narrower alternative that has all the same advantages and fewer disad- vantages, and the agency has not articulated any reasonable explanation for rejecting the proposed alternative. We therefore vacate the FCC’s determination that ILECs must make mass market switches available to CLECs as UNEs, subject to the stay discussed in Part VI below, and remand to the Commission for a re-examination of the issue. 3. The Commission’s definition of ‘‘impairment’’ The Commission claims that no party in this litigation has challenged the concept embodied in its new interpretation of 23 ‘‘impairment.’’ All the disputes, it says, are about the proper implementation of that standard. FCC Br. at 18. Not exactly. For example, although the ILEC petitioners’ objec- tions to the Commission’s mass market switching provisions are all within the framework of the Commission’s subdelega- tion scheme, a number of them clearly go to the character of the impairment standard embodied in that scheme. As a general matter the ILECs argue the Commission’s impairment standard is so open-ended that it imposes no meaningful constraints on unbundling, and would be unlawful even if applied by the FCC itself. ILEC Br. at 28; see also Separate Statement of Commissioner Kathleen Q. Abernathy Approving in Part and Dissenting in Part, FCC 03–36 at 6–7 & n. 16 (claiming that the Commission’s multifactor test is no different from the totality-of-the-circumstances approach struck down in USTA I). More specifically, the ILECs claim that the Commission’s unbundling test unlawfully permits states to consider as a potential source of impairment retail rates that are held below cost by state regulation against the ILECs’ will, and unlawfully precludes consideration of inter- modal competition when determining whether a market is suitable for competitive supply. On the general point about the open-endedness of the Commission’s standard, we observe that the Order’s interpre- tation of impairment is an improvement over the Commis- sion’s past efforts in that, for the most part, the Commission explicitly and plausibly connects factors to consider in the impairment inquiry to natural monopoly characteristics (de- clining average costs throughout the range of the relevant market), see Order ¶ ¶ 75–76 & nn.245, 256, 258–59, ¶ 87 & n.283, or at least connects them (in logic that the ILECs do not seem to contest) to other structural impediments to competitive supply. These barriers include sunk costs (Order ¶ 75 & n.244, ¶ ¶ 76, 80, 86, 88), ILEC absolute cost advan- tages (Order ¶ 75 & n.247, ¶ 90 & n.302), first-mover advan- tages (Order ¶ 75 & n.249, ¶ 89), and operational barriers to entry within the sole or primary control of the ILEC (Order ¶ 91). In contrast to the First Report and Order and the Third Report and Order, the Commission has clarified that 24 only costs related to structural impediments to competition are relevant to the impairment analysis. In light of our remand, this is not the occasion for any review of the Commission’s impairment standard as a general matter; it finds concrete meaning only in its application, and only in that context is it readily justiciable. A few general observations are pertinent, however. Relation of ‘‘impairment’’ to the ‘‘at a minimum’’ clause. We note that there are at least two ways in which the Commission could have accommodated our ruling in USTA I that its impairment rule take into account not only the benefits but also the costs of unbundling (such as discourage- ment of investment in innovation), in order that its standard be ‘‘rationally related to the goals of the Act.’’ See USTA I, 290 F.3d at 428. One way would be to craft a standard of impairment that built in such a balance, as for example by hewing rather closely to natural monopoly features. The other is to use a looser concept of impairment, with the costs of unbundling brought into the analysis under § 251(d)(2)’s ‘‘at a minimum’’ language. The Commission has chosen the latter, and we cannot fault it for doing so. This is especially true as the statutory structure suggests that ‘‘impair’’ must reach a bit beyond natural monopoly. While for ‘‘proprie- tary’’ network elements the statute mandates a decision whether they are ‘‘necessary,’’ § 251(d)(1)(A), for non- proprietary ones it requires a decision whether their absence would ‘‘impair’’ the requester’s provision of telecommunica- tions service, § 251(d)(1)(B). Thus, in principle, there is no statutory offense in the Commission’s decision to adopt a standard that treats impairment as a continuous rather than as a dichotomous variable, and potentially reaches beyond natural monopoly, but then to examine the full context before ordering unbundling. That said, we do note that in at least one important respect the Commission’s definition of impairment is vague almost to the point of being empty. The touchstone of the Commis- sion’s impairment analysis is whether the enumerated opera- tional and entry barriers ‘‘make entry into a market uneco- 25 nomic.’’ Order ¶ 84. Uneconomic by whom? By any CLEC, no matter how inefficient? By an ‘‘average’’ or ‘‘representa- tive’’ CLEC? By the most efficient existing CLEC? By a hypothetical CLEC that used ‘‘the most efficient telecommu- nications technology currently available,’’ the standard that is built into TELRIC? Compare 47 CFR § 51.505(b)(1). We need not resolve the significance of this uncertainty, but we highlight it because we suspect that the issue of whether the standard is too open-ended is likely to arise again. Intermodal alternatives. As for the ILECs’ claim that the Commission’s impairment standard unlawfully excludes con- sideration of intermodal alternatives, we observe that the Commission expressly stated that such alternatives are to be considered when evaluating impairment. Order ¶ ¶ 97–98, 443. Whether the weight the FCC assigns to this factor is reasonable in a given context is an question that we need not decide, except insofar as we reaffirm USTA I’s holding that the Commission cannot ignore intermodal alternatives. 290 F.3d at 429. Impairment in markets where state regulation holds rates below historic costs. In the name of ‘‘universal service,’’ state regulators have commonly employed cross-subsidies, tilting rate ceilings so that revenues from business and urban cus- tomers subsidize residential and rural ones. USTA I, 290 F.3d at 422. On remand from our decision in USTA I, the Commission decided to consider regulated below-cost retail rates as a factor that may ‘‘impair’’ CLECs in competing for mass market customers. See Order ¶ 518. The ILECs object strenuously, and it appears virtually certain that the issue will recur on remand. The Commission’s brief treatment of the issue makes no attempt to connect this ‘‘barrier’’ to entry either with struc- tural features that would make competitive supply wasteful or with any other purposes of the Act (other than, implicitly, the purpose of generating ‘‘competition,’’ no matter how synthet- ic). The Commission rightly says that if prevailing rates are too low to elicit CLEC entry even with the benefit of UNEs, the unbundling mandate will have no consequences. True 26 enough. But it is no defense of a rule to say that it is harmless in those cases where it has no effect at all; that presumably is true even of the most absurd rule. The interesting case is the one where TELRIC rates are so low that unbundling does elicit CLEC entry, enabling CLECs to cut further into ILEC revenues in areas where the ILECs’ service is mandated by state law—and mandated to be of- fered at artificially low rates funded by ILECs’ supracompeti- tive profits in other areas. If the scheme of the Act is successful, of course, the very premise of these below-cost rate ceilings will be undermined, as those supracompetitive profits will be eroded by Act-induced competition. In com- petitive markets, an ILEC can’t be used as a pi˜ata. The n Commission has said nothing to address these obvious impli- cations, or otherwise to locate its treatment of the issue in any purposeful reading of the Act. We recognize, of course, that the historic accounting costs relied upon by state regulators are, like TELRIC itself, an artificial construct that may not closely track true economic cost. But that is no justification for the Commission’s refusal to evaluate the probable consequences of its approach, and to adopt, in the light of those estimations, a policy that it can reasonably say advances the goals of the Act. B. Unbundling of High–Capacity Dedicated Transport Fa- cilities 1. Unlawfulness of the delegation to the states and the national impairment finding The Commission has made multiple impairment findings with respect to dedicated transport elements (transmission facilities dedicated to a single customer or carrier), varying the findings by capacity level. First, it found that competing providers are not impaired without unbundled access to ‘‘OCn’’ transport facilities (very high-capacity transport facili- ties or bandwiths within such facilities), Order ¶ ¶ 359, 372, and all petitioners appear to accept that finding. Second, the Commission found that competitors are impaired without unbundled access to DS1 transport, DS3 transport, and dark 27 fiber transport, but made this nationwide impairment finding subject to variation by state commissions applying specific ‘‘competitive triggers.’’ Id. ¶ 359; see also id. ¶ ¶ 381–93. Explaining this latter decision, the Commission observed that its nationwide impairment findings for DS1, DS3, and dark fiber were based on ‘‘aggregated data’’ and frankly acknowl- edged that competitive alternatives are available ‘‘in some locations.’’ Id. ¶ 398. The Commission declared that it did not need to resolve ‘‘the factual identification of where alter- native facilities existTTTT [B]ecause we recognize that the record is insufficiently detailed to make more precise findings regarding impairment, we delegate to the states, subject to appeal back to this Commission if a state fails to act, a fact- finding role to determine on a route-specific basis where alternatives to the incumbent LECs’ networks exist such that competing carriers are no longer impaired.’’ Id. ¶ 398. Specifically, the Commission instructed states to apply two competitive triggers on a route-by-route basis. Id. ¶ ¶ 399– 401. First, the ‘‘self-provisioning’’ trigger required states to find no impairment if three or more competitors had deployed non-ILEC transport facilities along a specific route. Id. ¶ ¶ 400, 405–09. Second, the ‘‘wholesale facilities’’ trigger required states to find no impairment if two or more compet- ing carriers were immediately able and willing to sell trans- port along a given route at wholesale rates. Id. ¶ ¶ 400, 412– 16. Even where the triggers were not satisfied, the FCC allowed a finding of non-impairment if a state, applying seven criteria (all quite fluid and none quantified), determined that the route was suitable for multiple competitive supply. Id. ¶ 410. If a state believed that there was impairment on a specific route despite facial satisfaction of the self- provisioning trigger, it could petition the Commission for a waiver. Id. ¶ 411. As we explained in the mass market switching context, the Commission may not subdelegate its § 251(d) authority to state commissions. Although the Commission characterizes the states’ role as ‘‘fact-finding,’’ Order ¶ 394, the character- ization is fictitious. It is the states, not the FCC, that determine whether the competitive triggers, or the Commis- 28 sion’s numerous and largely unquantified alternative criteria, are satisfied; it is the states that issue binding orders, subject only to the Commission’s discretionary review. And, as with mass market switching, the Order itself suggests that the Commission doubts a national impairment finding is justified on this record. Id. ¶ ¶ 360, 394, 398. We therefore vacate the national impairment findings with respect to DS1, DS3, and dark fiber and remand to the Commission to implement a lawful scheme. 2. Remaining dedicated transport issues The ILECs have raised two additional issues about the Commission’s treatment of dedicated transport, and the CLECs yet another. We address the ILECs’ objections here, and that of the CLECs (which relates to so-called ‘‘entrance facilities’’) below in the portion of the opinion devoted to their claims. a. Route-specific analysis of dedicated transport In USTA I we expressed skepticism regarding whether there could be impairment in markets ‘‘where the element in question—though not literally ubiquitous—is significantly de- ployed on a competitive basis,’’ giving as a specific example interoffice dedicated transport. 290 F.3d at 422. We also instructed the Commission, as noted above, to apply a ‘‘nu- anced’’ concept of impairment connected to ‘‘specific markets or market categories.’’ Id. at 426. Any process of inferring impairment (or its absence) from levels of deployment de- pends on a sensible definition of the markets in which deploy- ment is counted. For dedicated transport elements the Commission decided that the appropriate market was not a geographic market (e.g., a Metropolitan Statistical Area (‘‘MSA’’), as the ILECs urged, or general customer class), but rather a specific point- to-point route. Thus, for example, the fact that dedicated transport facilities are widely deployed within one MSA does not, in the Commission’s view, necessarily preclude a finding of impairment between two specific points within that MSA, if 29 deployment has not satisfied the Commission’s competitive ‘‘triggers’’ on that route. We do not see how the Commission can simply ignore facilities deployment along similar routes when assessing impairment. Suppose points A, B, and C are all in the same geographic market and are similarly situated with regard to the ‘‘barriers to entry’’ that the Commission says are control- ling. See Order ¶ ¶ 84 et seq. Suppose further that multiple competitors supply DS1 transport between points A and B, but only the ILEC and one other CLEC have deployed DS1 transport between A and C. The Commission cannot ignore the A–B facilities deployment when deciding whether CLECs are impaired with respect to A–C deployment without a good reason. The Commission does explain why competition on the A–B route should not be sufficient to establish competi- tion is possible on the A–C route, Order ¶ 401, but this cannot explain the Commission’s implicit decision to treat competi- tion on one route as irrelevant to the existence of impairment on the other. Nor does the Commission explain whether, and why, the error costs (both false positives and false negatives) associated with a route-by-route market definition are likely to be lower than the error costs associated with alternative market definitions. While it may be infeasible to define the barriers to entry in a manageable form, i.e., in such a way that they may usefully be applied to MSAs (or other plausible markets) as a whole, the Commission nowhere suggests that it explored such alternatives, much less found them defective. b. Wireless providers’ access to unbundled dedicated transport In addition to their general challenge to the FCC’s provi- sional national finding that competitors are impaired without access to dedicated transport facilities, the ILEC petitioners also attack the Commission’s conclusion that providers of wireless service (also known as commercial mobile radio services, or ‘‘CMRS’’) qualify for unbundled access to these facilities. According to the ILECs, the Commission not only failed to conduct the requisite impairment analysis for wire- less providers, but in fact found that wireless growth has 30 been ‘‘remarkable’’: 90% of the U.S. population lives in areas served by at least three wireless providers, 40% of Americans and 61% of American households own a wireless phone, wireless prices have been steadily declining, and 3–5% of wireless customers use wireless as their only phone, treating it as a full substitute for traditional land line service. Order ¶ 53. Although the ILECs implicitly concede that wireless providers would be impaired if they were denied any access to ILEC dedicated interoffice transport facilities, they point out that wireless providers have traditionally purchased such access from ILECs at wholesale rates (a transaction classi- fied, since adoption of the Act, under § 251(c)(4)). And the data above clearly show that wireless carriers’ reliance on special access has not posed a barrier that makes entry uneconomic. Indeed, the multi-million dollar sums that the Commission regularly collects in its auctions of such spec- trum, see, e.g., Annual Report and Analysis of Competitive Market Conditions With Respect to Commercial Mobile Ser- vices, Seventh Report, FCC 02–179 (July 3, 2002), Table 1B, and that firms pay to buy already-issued licenses, see, e.g., Annual Report and Analysis of Competitive Market Condi- tions With Respect to Commercial Mobile Services, Eighth Report, FCC 03–150 (July 14, 2003), ¶ ¶ 42–44, seem to indicate that wireless firms currently expect that net reve- nues will, by a large margin, more than recover all their non- spectrum costs (including return on capital). The FCC and the wireless intervenors do not challenge the assertion that the current regime has witnessed a rapidly expanding and prosperous market for wireless service. Rath- er, they rely on the principle that ‘‘evidence that requesting carriers are using incumbent LEC tariffed services’’ is not ‘‘relevant to [the] unbundling determination.’’ Order ¶ 102. The Commission offers several justifications for its decision to treat special access availability as irrelevant to the impair- ment analysis. None withstands scrutiny. First, the Com- mission suggests that it would be inconsistent with the Act if we permitted the incumbent LEC to avoid all unbundling merely by providing resold 31 or tariffed services as an alternative. Such an approach would give the incumbent LECs unilateral power to avoid unbundling at TELRIC rates simply by voluntarily making elements available at some higher price. Order ¶ 102 (footnote omitted). While the possibility to which the Commission points is undeniable, its implications for the Act’s implementation aren’t as horrifying as the Commission seems to think. After all, the purpose of the Act is not to provide the widest possible unbundling, or to guarantee com- petitors access to ILEC network elements at the lowest price that government may lawfully mandate. Rather, its purpose is to stimulate competition—preferably genuine, facilities- based competition. Where competitors have access to neces- sary inputs at rates that allow competition not only to survive but to flourish, it is hard to see any need for the Commission to impose the costs of mandatory unbundling. We recognize that, given the ILECs’ incentive to set the tariff price as high as possible and the vagaries of determin- ing when that price gets so high that the ‘‘impairment’’ threshold has been crossed, a rule that allowed ILECs to avoid unbundling requirements simply by offering a function at lower-than-TELRIC rates might raise real administrability issues. Those complications might in principle support a blanket rule treating the availability of ILEC tariffed service as irrelevant to impairment. But the FCC hasn’t defended its decision in those terms or even tried to explicate these complications. Moreover, where (as here) market evidence already demonstrates that existing rates outside the compul- sion of § 251(c)(3) don’t impede competition, and where (as here) there is no claim that ILECs would be able drastically to hike those rates, those possible complications recede even farther in the background. The FCC also suggests that the ILECs’ view would effec- tively read unbundled access out of the Act. Both the Commission and the wireless intervenors argue that this conclusion finds support in Iowa Utilities I, which held that ILECs could not avoid unbundling requirements by classify- ing certain features as ‘‘services’’ rather than ‘‘network ele- 32 ments.’’ 120 F.3d at 809. There the ILECs had argued that the legislative history of the Act suggested that functions offered as services were meant to be governed by the resale provisions of § 251(c)(4) rather than the unbundling provi- sions of § 251(c)(3). In rejecting this argument, the Eighth Circuit said that the provision ‘‘for the resale of telecommuni- cations services TTT does not establish resale as the exclusive means through which a competing carrier may gain access to such services. We agree with the FCC that such an interpre- tation would allow the incumbent LECs to evade a substantial portion of their unbundling obligation under subsection 251(c)(3).’’ 120 F.3d at 809. Thus the court found that an ILEC offer of functions for sale as services did not preclude classifying these functions as network elements to be unbun- dled under § 251(c)(3). But that decision in no way supports a claim that the availability of services for sale under § 251(c)(4) is irrelevant to whether there is impairment of the sort that would require unbundling. The Commission next argues that considering special ac- cess availability in the impairment analysis would ‘‘be con- trary to the Act’s requirement that unbundled facilities TTT should be priced at cost-based rates and our determination that TELRIC is the appropriate methodology for determining those ratesTTTT’’ Order ¶ 102. This is circular. The ques- tion is which facilities must be unbundled, or, more specifical- ly, what the relevant benchmark is for assessing whether entry is ‘‘impaired’’ if non-ILECs don’t have access to UNEs (at whatever rate the Commission might choose to prescribe). Finally, the FCC suggests that tariffed services ‘‘present different opportunities and risks for the requesting carrier than the use of UNEs or non-incumbent LEC alternatives.’’ Order ¶ 102. This may well be true in certain cases, and on an appropriate record the Commission might find impairment even when services were available from ILECs outside § 251(c)(3). But this possibility doesn’t give the Commission carte blanche to omit consideration of such alternatives in its impairment analysis. And it clearly cannot justify a finding of impairment with respect to wireless, where these different 33 ‘‘opportunities and risks’’ have obviously not made competi- tive entry uneconomic. We therefore hold that the Commission’s impairment anal- ysis must consider the availability of tariffed ILEC special access services when determining whether would-be entrants are impaired, and vacate ¶ ¶ 102–03 of the Order. This of course still leaves the Commission free to take into account such factors as administrability, risk of ILEC abuse, and the like. What the Commission may not do is compare unbun- dling only to self-provisioning or third-party provisioning, arbitrarily excluding alternatives offered by the ILECs. C. Network Modification Requirements In Iowa Utilities I, the Eighth Circuit struck down an FCC rule that required ILECs to provide interconnection and UNEs superior in quality to those that the ILEC provided for itself. 120 F.3d at 812–13. But the court nonetheless ‘‘endorse[d] the Commission’s statement that ‘the obligations imposed by sections 251(c)(2) and 251(c)(3) include modifica- tions to incumbent LEC facilities to the extent necessary to accommodate interconnection or access to network ele- ments.’ ’’ Id. at 813 n.33. The line between impermissible ‘‘superior quality’’ requirements and permissible ‘‘modifica- tion’’ requirements is not always clear. In the Order under review, the Commission ‘‘require[d] incumbent LECs to make routine network modifications to unbundled transmission facilities used by requesting carriers where the requested transmission facility has already been constructed.’’ Order ¶ 632. The Commission elaborated that ‘‘routine network modifications’’ include ‘‘those activities that incumbent LECs regularly undertake for their own custom- ers,’’ but do not include ‘‘construction of new wires TTT for a requesting carrier.’’ Id. Applying this standard, the Com- mission determined that when ILECs supply high-capacity loops as unbundled elements, they must ‘‘engage in activities necessary to activate loops that are not currently activated in the network.’’ Id. ¶ 633. The FCC gave as examples of such necessary loop modifications: ‘‘rearrangement or splicing of cable; adding a doubler or repeater; adding an equipment 34 case; adding a smart jack; installing a repeater shelf; adding a line card; and deploying a new multiplexer or reconfiguring an existing multiplexer.’’ Id. ¶ 634. The ILECs claim that these passages manifest a resurrec- tion of the unlawful superior quality rules. We disagree. The FCC has established a clear and reasonable limiting principle: the distinction between a ‘‘routine modification’’ and a ‘‘superior quality’’ alteration turns on whether the modification is of the sort that the ILEC routinely performs, on demand, for its own customers. While there may be disputes about the application, the principle itself seems sensible and consistent with the Act as interpreted by the Eighth Circuit. Indeed, the FCC makes a plausible argu- ment that requiring ILECs to provide CLECs with whatever modifications the ILECs would routinely perform for their own customers is not only allowed by the Act, but is affirma- tively demanded by § 251(c)(3)’s requirement that access be ‘‘nondiscriminatory.’’ We needn’t reach that claim, however, since the FCC’s principle is at the very least reasonable and consistent with Iowa Utilities I. The ILECs further object that the Order unlawfully per- mits states to find that ILECs are not entitled to compensa- tion for making the requested modifications. We agree with the FCC that this challenge will not be ripe for judicial review until a state actually decides how much an ILEC may charge for a specific network modification. III. CLEC Objections A. Unbundling of Broadband Loops The Commission declined to require ILECs to provide unbundled access to most of the broadband capabilities of mass market loops. In particular, it decided (subject to certain qualifications) not to require unbundling of the broad- band capabilities of hybrid copper-fiber loops, Order ¶ ¶ 288– 89, or fiber-to-the-home (‘‘FTTH’’) loops, id. ¶ ¶ 273–77, and it also decided not to require ILECs to unbundle the high- frequency portion of copper loops, a practice known as ‘‘line 35 sharing,’’ id. ¶ ¶ 255–63. The Commission did require ILECs to unbundle the narrowband portion of hybrid loops, Order ¶ 296, but it permitted ILECs to use a different type of technology to connect the fiber feeder loop to the copper distribution portion of the loop than the ILEC itself used, in light of technological and engineering considerations, Order ¶ 297. The CLEC petitioners attack these decisions as inconsis- tent with the Act. They argue, first, that CLECs are im- paired without access to the broadband capabilities of loops and, second, that the Commission is obligated to unbundle any elements for which impairment has been shown. We consider these claims with respect to each broadband element in question. We then consider the CLECs’ claim that their access to the narrowband portion of hybrid loops is impaired by the FCC’s decision permitting ILECs to substitute an allegedly inferior connection technology. 1. Hybrid loops The Commission found some degree of impairment from competitors’ lack of unbundled access to hybrid loops, Order ¶ 286, but also found that such impairment ‘‘at least partially diminishes with the increasing deployment of fiber,’’ id., and that unbundled access to copper subloops ‘‘adequately ad- dresses’’ that impairment, id. § 291. Nonetheless, evidently assuming some degree of impairment, it proceeded to invoke the ‘‘at a minimum’’ language of § 251(d)(2) to weigh other statutory goals against that effect. Noting the directive in § 706(a) of the Act that the Commission should pursue ‘‘methods that remove barriers to infrastructure investment,’’ it found that the costs of unbundling hybrid loops—stifling investment by both ILECs and CLECs in advanced telecom- munications infrastructure—outweighed the benefits of re- moving this barrier to competition. Id. ¶ ¶ 286, 288, 290. The CLECs object to this interpretation of the ‘‘at a minimum’’ clause, arguing that the Act prohibits ‘‘ad hoc’’ balancing of the statute’s pro-competition goals with an alleg- edly conflicting goal derived from the uncodified § 706. They interpret the ‘‘at a minimum’’ clause to mean that the FCC 36 may order unbundling even in the absence of an impairment finding if it finds concrete benefits to unbundling that cannot otherwise be achieved, and that it may refuse to order unbundling in the face of impairment findings if unbundling would conflict with some other unambiguous requirement of the Act, such as funding universal service. The CLECs offer two main arguments to support their interpretation of the ‘‘at a minimum’’ clause. First, they claim that the Commission’s interpretation contravenes the Act’s ‘‘stated purpose’’ of promoting competition, CLEC Br. at 18, a goal that is an ‘‘end in itself.’’ Id. (quoting Verizon, 535 U.S. at 476). But in fact the passage from Verizon on which the CLECs rely says that eliminating traditional ILEC monopolies ‘‘was considered both an end in itself and an important step toward the Act’s other goals,’’ including ‘‘boosting competition in broader markets.’’ 535 U.S. at 476 (emphasis added). Section 706(a) identifies one of the Act’s goals beyond fostering competition piggy-backed on ILEC facilities, namely, removing barriers to infrastructure invest- ment. The Commission thus acted reasonably in its interpre- tation of the ‘‘at a minimum’’ clause. Second, the CLECs contend that failing to impose unbun- dling in the face of an impairment finding amounts to an unlawful decision to ‘‘forbear’’ from applying the require- ments of § 251(c). See §§ 160(a),(d). Here they rely on Association of Communications Enterprises (‘‘ASCENT’’) v. FCC, 235 F.3d 662, 665–68 (D.C. Cir. 2001), in which, reject- ing the Commission’s argument that the exclusion of ILEC subsidiaries was a reasonable interpretation of the statutory phrase ‘‘successor or assign’’ in § 251(h)(2)(B)(ii), we held that the FCC couldn’t exempt an ILEC subsidiary from § 251(c)(3) obligations unless it complied with the statutory forbearance requirements of § 160. But § 160, prescribing when the Commission may forbear from applying statutory requirements, obviously comes into play only for requirements that exist; it says nothing as to what the statutory requirements are. Thus ASCENT turned on our finding that, even under Chevron’s forgiving standard, 37 the Commission’s exemption of subsidiaries was inconsistent with the statute. 235 F.3d at 668. As we noted above in Part II.A.3, there are at least two ways in which the Commission could take into account the frustration of some of the Act’s goals—such as encouraging facilities-based competition—that would flow from giving § 251(c)(3) unbundling too broad a scope. It could have built those offsets into its concept of ‘‘impairment’’ by reading that term narrowly, or it could have embraced a relatively broad reading of impairment and then considered, element by ele- ment, how an unbundling order might adversely affect the Act’s other goals. The CLECs rightly point to USTA I’s observation that ‘‘impairment’’ was the ‘‘touchstone,’’ 290 F.3d at 425, but that opinion, far from barring consideration of factors such as an unbundling order’s impact on investment, clearly read the Act, as interpreted by the Supreme Court in AT&T, to mandate exactly such consideration, id. at 427–28. We therefore hold that the Commission reasonably inter- preted § 251(c)(3) to allow it to withhold unbundling orders, even in the face of some impairment, where such unbundling would pose excessive impediments to infrastructure invest- ment. But was the Commission’s decision on hybrid loops, on this record, a legitimate application of that principle? The Com- mission explained that its decision would stimulate the infra- structure investment contemplated by § 706 in two ways. First, limiting access to the fiber portion of the hybrid loops would give ILECs incentives to deploy fiber (both feeder fiber and, eventually, FTTH), along with associated next- generation networking equipment, and to develop new broad- band offerings for mass market consumers. Because unbun- dling orders reduce return on investment, such orders would inhibit ILECs from making risky investments in next- generation technology. Second, denying CLECs access to ILEC broadband capabilities will stimulate them to seek innovative access options for broadband, including self- deployment of new facilities; unbundling, by contrast, would 38 be likely to blunt innovation by locking the CLECs into technological choices made by the ILECs. Order ¶ ¶ 290, 295. The Commission also identified two additional consider- ations that would mitigate any negative impact on local competition in broadband. First, CLECs still have unbun- dled access to other loop alternatives in the ILEC network, including copper subloops, which allow CLECs to compete in the broadband market. Order ¶ 291. Second, intermodal competition in broadband, particularly from cable companies, means that, even if CLECs proved unable to compete with ILECs in the broadband market, there would still be vigorous competition from other sources. Id. ¶ 292. The CLEC petitioners reject all these justifications, and pose a series of objections. First, they argue, the FCC should redress any investment disincentives for ILEC broad- band loop investment not by withholding unbundling, but by modifying the UNE pricing rules. But as we have already held, § 251(d)(2)’s ‘‘at a minimum’’ clause allows the Commis- sion to consider the effect on infrastructure investment when determining what elements must be unbundled. And the fact that the Commission and the Court have deemed TELRIC a reasonable methodology for pricing UNEs doesn’t require the Commission to blind itself to the fact that TELRIC may itself be imperfect and may be implemented still more imperfectly. While the Commission might modify its UNE pricing rules to adequately reduce the negative impacts that it fears, until it has done so it may reasonably consider real-world risks in deciding what elements to unbundle. Second, the CLECs insist that the record demonstrates that there is no need for additional incentives for investment in broadband infrastructure. With respect to broadband customers served by hybrid loops, ILECs have already exten- sively deployed fiber feeder loops, and, the CLECs claim, they would continue to do so even without any incentive from expected broadband revenues, since the narrowband cost savings from fiber feeder deployment alone justify ILEC investment in fiber feeder. Provision of broadband involves 39 additional electronic equipment, but the CLECs assert that the costs involved are negligible compared to the fiber up- grade, and that in fact most of these additional investments have already been made. As for alternative means of provid- ing broadband service, the CLECs characterize the FCC’s assertion that eliminating unbundled access to hybrid loops would stimulate ILEC investment in FTTH loops as pure speculation, inconsistent with record evidence that there is no consumer demand for services requiring such loops. And they say that the Commission may not tolerate an impairment of competition that would benefit consumers of today in order to create incentives for investment in systems for which there is no evidence of demand by consumers of tomorrow. The Commission says little in the Order or in its brief to respond the assertion that ILECs would invest in fiber feeder even without revenue from broadband. Indeed, the Commis- sion appears to concede that ILECs are already investing heavily in fiber feeder loops, Order ¶ ¶ 224, 290, and offers no specific evidence suggesting that unbundling the broadband capabilities of these loops would have a substantial negative impact on this investment. (Nor, to be sure, do the CLECs offer any sort of sophisticated econometric analysis demon- strating the likely marginal impact on investment.) But there are at least three other aspects of the Commis- sion’s investment incentives argument to which the CLEC response is either inadequate or non-existent. First, the Commission suggested that greater incentives may be needed for ILECs to deploy the additional electronic equipment needed to provide broadband access over a hybrid loop. While the CLECs are correct that the Commission concluded that the deployment of this equipment was far less ‘‘costly, complex, and risky’’ than deployment of the fiber feeder, Order ¶ 244, the Commission also noted that this equipment had not been widely deployed, and suggested that ILECs had been deterred by the ‘‘regulatory environment.’’ Order ¶ 290 & n.838. Second, the Commission noted that deployment of feeder fiber is the first step toward FTTH, and that limiting access to ILEC fiber facilities increases incumbents’ incentives to 40 develop and deploy FTTH. Order ¶ ¶ 272, 290. Though the CLECs dismissed this as ‘‘pure speculation,’’ the Commission relied on submissions in the record that the CLECs have not directly impeached. Order ¶ 290 n.837. While the CLECs may be right that the Commission’s judgment entails increas- ing consumer costs today in order to stimulate technological innovations for which there is not yet sufficient consumer demand, there is nothing in the Act barring such trade-offs. Cf. Consumer Electronics Ass’n v. FCC, 347 F.3d 291, 300–03 (D.C. Cir. 2003) (upholding Commission rule that increased television prices in order to stimulate transition to digital TV, for which there is little present demand). Third, the Commission rested its judgment not only on the perceived negative effect of unbundling on ILEC investment incentives but also on a conclusion that unbundling hybrid loops would deter CLECs themselves from investing in de- ploying their own facilities, possibly using different technolo- gy. Order ¶ ¶ 288, 290. Although the CLECs argue that this is inconsistent with the Commission’s finding that for fiber loops, as for copper loops, ‘‘the costs are both fixed and sunk, and TTT deployment is characterized by scale economies,’’ id. ¶ 240, that very paragraph, after weighing the various advan- tages of both ILECs and other entrants, concludes that ‘‘the barriers faced in deploying fiber loops, as opposed to existing copper loops, may be similar for both incumbent LECs and competitive LECs.’’ Thus, while declining to unbundle hy- brid loops might reduce broadband competition, the Commis- sion reasonably concluded that such a decision might be effective in stimulating investment in all-fiber loops. We thus believe that, even if the CLECs are correct that unbundling would have no impact on ILEC investment in the fiber feeder portion of hybrid loops, the other investment disincentives the Commission identified are sufficient for us to uphold the reasonableness of the Commission’s determina- tion. Reading the Order as a whole, we see little sign that the Commission would have come out otherwise if it had given the CLEC arguments as much credit as they deserve. See Indiana Muni. Power Agency v. FERC, 56 F.3d 247, 256 41 (D.C. Cir. 1995); Carnegie Natural Gas Co. v. FERC, 968 F.2d 1291, 1294 (D.C. Cir. 1992). Nor can we say that the Commission was arbitrary or capricious in thinking that any damage to broadband competi- tion from denying unbundled access to the broadband capaci- ties of hybrid loops is likely to be mitigated by the availability of loop alternatives or intermodal competition. With regard to loop alternatives, we agree with the CLECs that these alternatives are not a perfect substitute for the ILECs’ hybrid loops, but we understand the Commission to say only that they are a partial substitute; they will mitigate, not eliminate, CLEC impairment. More important, we agree with the Commission that robust intermodal competition from cable providers—the existence of which is supported by very strong record evidence, including cable’s maintenance of a broadband market share on the order of 60%, see Order ¶ 292—means that even if all CLECs were driven from the broadband market, mass market consumers will still have the benefits of competition between cable providers and ILECs. Although the CLECs point to evidence that CLEC broad- band competition has played a role in constraining ILEC pricing, see Declaration of Robert D. Willig, ¶ ¶ 206–08, Joint Appendix (‘‘J.A.’’) 885–87, the evidence itself is hardly rigor- ous and is offset by conflicting material, see Letter of Sus- anne Guyer, Vice President, Verizon, at 2 (J.A. 2146), itself not rigorous. Thus the Commission’s consideration of past pricing effects was not arbitrary, and in any event, as the discussion above shows, its overall judgment turned on a range of factors. We therefore hold that the Commission’s decision not to order unbundling of the broadband capacity of hybrid loops was based on permissible statutory considerations and sup- ported by substantial evidence. Although the Commission refused to unbundle the broad- band portion of hybrid loops, it required ILECs to unbundle the narrowband portion, Order ¶ 296, and the CLECs raise an issue relating to the details of this unbundling. The Commission said for various technical reasons this would be 42 more difficult for hybrid loops that used integrated digital loop carrier (‘‘IDLC’’) equipment to connect the fiber feeder portion of the loop to the copper distribution portion than it would for those that used universal digital loop carrier equip- ment (‘‘UDLC’’). Order ¶ 297 & n.855. The CLECs protest that the record ‘‘unambiguously estab- lished that UDLC substantially degrades the speed and quali- ty of dial-up Internet access,’’ CLEC Br. at 30, though they fail to point us to the portions of the record that supposedly establish this. The Commission acknowledges that ‘‘UDLC can, in some circumstances, negatively affect data transmis- sion speed,’’ FCC Br. 84 n.37, but it disputes the severity of the impact. Moreover, the Order requires that ILECs ‘‘pres- ent requesting carriers a technically feasible method of un- bundled access.’’ Order ¶ 297. Given the CLEC petitioners’ failure to present or highlight evidence that the impact is severe, or to refute the Commission’s technical analysis, we have no basis for finding the Commission decision on this issue arbitrary or capricious. 2. Fiber-to-the-home (‘‘FTTH’’) loops For FTTH loops, the Commission found relatively little impairment except in a specific, limited domain. Although FTTH deployment showed some characteristics in common with copper loops (the costs being ‘‘both fixed and sunk, and deployment [being] expensive,’’ Order ¶ 274), the Commission believed that the revenue opportunities of FTTH deployment were great enough to ‘‘ameliorate many of the entry barri- ers.’’ Id.; see also id. ¶ 276 (same, with respect to FTTH parallel to or in replacement of existing copper plant). With respect to new or so-called ‘‘greenfield’’ FTTH deployments (as for a new subdivision), it denied unbundling without qualification. Id. ¶ 275. For the ‘‘largely theoretical’’ scenar- io in which an ILEC constructed FTTH parallel to or in replacement of its existing copper plant (‘‘overbuild’’), it de- clined to find impairment as to broadband services, id. ¶ 276, but agreed with the CLECs’ concern that an ILEC might replace and ultimately deny access to the copper loops that CLECs were using to serve mass market customers, id. 43 ¶ 277. In the overbuild situations, then, it ruled that the ILEC must either keep the existing copper loop connected after deploying FTTH, or else provide CLECs with unbun- dled access to the narrowband capabilities of the replacement FTTH loop. Id. ¶ ¶ 277, 281–84. Although not contesting the concept that large expected revenue can offset scale economies, the CLECs do object to the Commission’s decision that CLECs are not impaired by lack of unbundled access to FTTH. They argue that the Commission ignored two critical considerations. First, they point out that the FCC made a national finding that CLECs are impaired without unbundled access to enterprise market high-capacity DS3 loops (which are made from the same fiber as mass market FTTH loops), finding that ‘‘a single DS3 loop, generally, can not provide a sufficient revenue opportunity’’ to overcome the entry barriers to deployment. Order ¶ 320. This, the CLECs say, contradicts the Commission’s conclu- sion that ‘‘the substantial revenue opportunities posed by FTTH deployment help ameliorate many of the entry barri- ers presented by the costs and scale economies.’’ Id. ¶ 274. Second, they argue that ILECs enjoy significant ‘‘first mov- er’’ advantages due to their existing customer base, rights-of- way, and their existing networks’ substantial excess fiber capacity (‘‘dark fiber’’) that ILECs can readily use for net- work extensions. While the CLECs’ objections are convincing in many re- spects, they are ultimately unavailing. Even if the CLECs are impaired with respect to FTTH deployment (a point we do not decide), the § 706 considerations that we upheld as legitimate in the hybrid loop case are enough to justify the Commission’s decision not to unbundle FTTH. Although the Commission based its refusal to unbundle on a finding of no impairment, it made clear that its decision was ‘‘inform[ed]’’ by § 706. Order ¶ 278. In particular, it noted that ‘‘remov- ing incumbent LEC unbundling obligations on FTTH loops will promote their deployment of the network infrastructure necessary to provide broadband services to the mass market.’’ Id. ¶ 278; see also id. ¶ ¶ 272, 290 & n.837. 44 We find that these considerations are sufficient to justify the Commission’s decision not to require FTTH unbundling, even if CLECs are to some extent ‘‘impaired’’ in their ability to enter certain segments of the FTTH broadband market. This conclusion is buttressed by the evidence in the record that FTTH deployment is still very limited, Order ¶ 274, that both the costs and potential benefits of deployment are high, id., and, at least in some contexts, ILECs and CLECs face similar entry barriers, Order ¶ ¶ 240, 275 & n.808, ¶ 276. An unbundling requirement under these circumstances seems likely to delay infrastructure investment, with CLECs tempt- ed to wait for ILECs to deploy FTTH and ILECs fearful that CLEC access would undermine the investments’ potential return. Absence of unbundling, by contrast, will give all parties an incentive to take a shot at this potentially lucrative market. 3. Line sharing In USTA I, 290 F.3d at 428–29, we vacated the Commis- sion’s decision to provide CLECs with unbundled access to the high frequency portion of copper loops to provide broad- band DSL services, primarily because the Commission had failed to consider the relevance of intermodal competition in the broadband market. On remand, the Commission decided to reverse its earlier position and eliminated this unbundling mandate. The Commission explained its change of heart as follows. First, the FCC rejected its prior finding that lack of separate access to the high frequency portion would cause impairment. The earlier impairment finding had been based on a notion that broadband revenues would not justify the cost of the whole loop. But now, applying its new decision to focus on all the potential revenues from the full functionality of a loop (voice, data, video, and other services), the Commis- sion believes that these revenues would offset the costs associated with purchasing the entire loop. Order ¶ 258. Additionally, the Commission reasons that CLECs interested only in broadband could obtain broadband frequencies from other CLECs through line-splitting, in which one CLEC 45 provides voice service on the low frequency portion of the loop and the other provides DSL on the high frequency portion. Thus, after taking both costs and revenues into account, the FCC decided that eliminating mandatory line sharing would not impair CLECs’ ability to provide broad- band service. Id. ¶ 259. The Commission also observed that the difficulties of cost allocation for different portions of a single loop had led most states to price the high frequency portion of the loop at approximately zero. This distorted competitive incentives since CLECs that purchased only the high frequency portion had an irrational cost advantage over both ILECs and CLECs that purchased the whole loop to offer a range of services. Order ¶ 260. The anomalous price differential also skewed CLECs’ incentives toward providing only broadband service instead of bundled voice and DSL, discouraged inno- vative arrangements between voice CLECs and data CLECs, and discouraged product differentiation between ILEC and CLEC offerings. Id. ¶ 261. Thus the FCC found the results of mandatory line sharing to be contrary to the Act’s goal of encouraging vigorous competition in all local telecommunica- tions markets. Id. Finally, following our mandate in USTA I, the Commission noted the substantial intermodal competition from cable com- panies, which provide nearly 60% of all high-speed lines. Order ¶ 262 & nn.777–78. Although noting that intermodal competition was not ‘‘dispositive’’ in the impairment analysis, the Commission found that it lessened any competitive bene- fits associated with line sharing. Id. ¶ 263. Taking this into account, along with the negative impact of unbundling on competitive incentives, it found that ‘‘the costs of unbundling the [high frequency portion of the loop] outweigh the bene- fitsTTTT’’ Id. As with FTTH, we find that even if the CLECs are right that there is some impairment with respect to the elimination of mandatory line sharing, the Commission reasonably found that other considerations outweighed any impairment. And again we note the ambiguous state of the record on the price- constraining effect of CLEC DSL service. We read the 46 Commission as concluding that, at least in the future, line sharing is not essential to maintain robust competition in this market, a conclusion based on permissible considerations and supported by evidence in the record. With respect to the skewed incentives from zero pricing of the high frequency portion, it is of course true that alternative cost allocations could have reduced the skew, but any alternative allocation of costs would itself have had some inescapable degree of arbi- trariness. Summary. We therefore uphold the Commission’s rules concerning hybrid loops, FTTH, and line sharing on the grounds that the decision not to unbundle these elements was reasonable, even in the face of some CLEC impairment, in light of evidence that unbundling would skew investment incentives in undesirable ways and that intermodal competi- tion from cable ensures the persistence of substantial compe- tition in broadband. B. Exclusion of ‘‘Entrance Facilities’’ Entrance facilities are dedicated transmission facilities that connect ILEC and CLEC locations. Before the Order, the Commission had defined ‘‘dedicated transport facilities’’ as including entrance facilities. But in the Order it concluded that this definition was ‘‘overly broad,’’ Order ¶ 365, and found that ‘‘a more reasonable and narrowly-tailored defini- tion of the dedicated transport network element includes only those transmission facilities within an incumbent LEC’s transport network, that is, the transmission facilities between incumbent LEC switches,’’ id. ¶ 366. Thus it held, as a matter of statutory interpretation, that entrance facilities were not ‘‘network elements’’ subject to the statutory unbun- dling requirements of § 251(c)(3), id., and accordingly re- quired no impairment analysis, id. ¶ 367 n.1119. As this is an issue of statutory construction, we review under the Chevron standard. The CLEC petitioners object that the Commission’s inter- pretation is flatly inconsistent with the text of the Act. In particular, the CLECs point out that § 153(29) of the Act defines ‘‘network element’’ as ‘‘a facility of equipment used in 47 the provision of a telecommunications service,’’ and that en- trance facilities clearly fall within that definition. Also, the CLEC petitioners continue, the Commission itself, in this Order, addressed the question whether ‘‘network element’’ included only facilities ‘‘actually used by the incumbent LEC in the provision of a telecommunications service’’ or also included facilities ‘‘capable of being used by a requesting carrier in the provision of a telecommunications service re- gardless of whether the incumbent LEC is actually using the network element to provide a telecommunications service,’’ and expressly adopted the latter definition. Order ¶ 59. While the Commission’s reasoning appears to have little or no footing in the statutory definition, we find the record too obscure to make any final ruling. The CLECs helpfully provide a diagram of various telecommunications network facilities, in which entrance facilities appear as completely stand-alone items linking a CLEC switch with an ILEC office. CLEC Reply Br. at 3. But no party offers an explanation as to why ILECs rather than CLECs construct these facilities. If (as appears) they exist exclusively for the convenience of the CLECs, it seems anomalous that CLECs do not themselves provide them, presumably doing so at the costs associated with ‘‘the most efficient telecommunications technology currently available,’’ 47 CFR § 51.505(b)(1), i.e., the TELRIC standard. The Commission hints at this consid- eration in observing that its ruling encourages CLECs to ‘‘incorporate those costs within their control into their net- work deployment strategies.’’ Order ¶ 367. Thus, although the Commission’s ruling superficially violates the statutory language, we simply remand the matter for further consider- ation. If entrance facilities are correctly classified as ‘‘net- work elements,’’ an analysis of impairment would presumably follow. C. Unbundling of Enterprise Switches The Commission determined, on a nationwide basis, that CLECs are not impaired by lack of unbundled access to switching for the enterprise market at DS1 capacity and above. Order ¶ ¶ 451–53. Though observing that the record 48 showed no impairment on a national basis in the absence of unbundling, id. ¶ 454, and indeed did ‘‘not contain evidence identifying any particular markets where competitive carriers would be impaired,’’ id. ¶ 455, the Commission went on to note that ‘‘a geographically specific analysis could possibly demonstrate that competitive carriers are impaired without access to unbundled incumbent LEC local circuit switching for DS1 enterprise customers in a particular market,’’ id. ¶ 454. It therefore permitted state commissions to petition the Commission to waive the ‘‘no impairment’’ finding in particular markets. Id. ¶ ¶ 455–58. The operative passages direct the state commissions to ‘‘examine’’ certain issues, and ‘‘consider [certain] evidence,’’ and to make ‘‘finding[s].’’ It is obscure what weight the Commission intended to give these findings. CLEC petitioners argue that the 90–day time limit on this petition procedure is arbitrary and capricious, given that in the mass market switching context the Order gave states nine months to collect and analyze market data. In what appears to be a throwaway sentence, the CLECs say the harm inflicted by this supposed error is ‘‘compounded’’ by the fact that the 90–day state proceedings are voluntary rather than mandatory (i.e., at the option of the state commissions), and that the impairment issue cannot be revisited absent changed circumstances. Order ¶ 455. Since we have invalidated the FCC’s subdelegation scheme with respect to mass market switches, a challenge based on the inconsistency between the nine-month period for mass market determinations and the 90–day period for enterprise market determinations is moot as a practical matter (though not in the strict jurisdictional sense). Cf. Belton v. Washing- ton Metro. Area Transit Auth., 20 F.3d 1197, 1203 (D.C. Cir. 1994). And in any event, we agree with the FCC that the market data states are to analyze under the enterprise switching provisions are significantly different from the data they were supposed to evaluate in the mass market switching context. 49 Apart from the argument regarding the inconsistency of time limits, the CLECs’ argument boils down to a claim that the no impairment finding for enterprise switches (1) is overbroad; and (2) lacks sufficient ‘‘safety valve’’ procedures to cure this overbreadth. But the CLECs do not contradict the Commission’s observation about the absence of evidence of impairment either nationwide or in specific markets. Thus, in contrast to the mass market switching context, where the evidence indicated the presence of many markets where CLECs suffered no impairment in the absence of unbundling, here there is no showing of any need for a safety valve, except insofar as one may infer a need from the Commission’s creation of one (which may in fact have been only an excess of caution). The CLECs make a rather underdeveloped argument that the vice of the alleged time-limit anomaly is ‘‘compounded’’ by the state proceedings being ‘‘voluntary rather than mandato- ry,’’ and that enterprise switching cannot be re-instated after the 90–day period without changed circumstances. CLEC Br. at 40 (citing Order ¶ 455). But these claims seem ancil- lary to the now-irrelevant time-limit theory, and without a showing of a need for a safety valve, we see no occasion to reach them. Finally, we note that our holding regarding unlawful sub- delegation of FCC authority to state commissions does not control the limited state commission role contemplated in the portion of the Order dealing with enterprise switching. In this context, state commissions are allowed merely to petition the FCC for a waiver of the unbundling order; the FCC has not granted the states authority to make final decisions on such matters as the existence of impairment. Because no party has challenged the limited state role in the enterprise switching context we have no occasion to rule on whether the role contemplated for the states here is legally problematic. D. Unbundling of Call–Related Databases and Signaling Systems Call-related databases are used in signaling networks for billing or for transmission, routing, and other telecommunica- 50 tions services. These databases include, for example, ones that provide name identification for caller ID service and ones that contain information on calling cards. Order ¶ 549. When CLECs have unbundled access to ILEC mass market switches, they also have access to the databases that the signaling network permits carriers to access. Id. ¶ 551. Where CLECs provide their own switches, however, they don’t automatically have access to the needed databases, and they must either self-provision or purchase databases from the ILEC or a third party. Id. The Commission determined that CLECs are not impaired without unbundled access to ILEC databases (other than the 911 database) because of the abundance of alternative provid- ers. Order ¶ ¶ 551–57. The CLECs object, arguing that the only reason alternatives to ILEC databases exist is that the Commission had previously required ILECs to provide un- bundled access to their databases (removing any competitive incentive for the ILECs to withhold the databases from third parties). But the CLECs point to nothing in the record demonstrating that this is so. Even if they did, we doubt that this alone would support a finding of impairment. As it stands, CLECs evidently have adequate access to call-related databases. If subsequent developments alter this situation, affected parties may petition the Commission to amend its rule. E. Unbundling of Shared Transport Facilities The FCC found CLECs that lease ILEC mass market switches are impaired without unbundled access to so-called ‘‘shared transport’’—transmission facilities shared by more than one carrier, including the ILEC, running between end office switches, between end office switches and tandem switches, and between tandem switches within the ILEC’s network. Order ¶ ¶ 533–34. But the FCC also concluded that, ‘‘because switching and shared transport are inextrica- bly linked, if incumbent LECs are no longer obligated to unbundle switching, they should no longer be obligated to unbundle shared transport.’’ Id. ¶ 534. In effect, it found that CLECs are entitled to unbundled shared transport only 51 in cases where mass market switching has also been unbun- dled. Id. The CLECs object to this condition for unbundled shared transport, saying that they are ‘‘impaired’’ without access to shared transport between local tandem switches when they ‘‘transit’’ traffic—that is, when they transport traffic that originates on their network to other carriers’ networks. The Commission in fact recognized the claim, saying that it proposed to address the issue in a pending rulemaking on intercarrier compensation. Id. ¶ 534 n.1640. Although the FCC failed to resolve an impairment question pressed by the CLECs in this Order, the Commission ‘‘need not address all problems ‘in one fell swoop.’ ’’ U.S. Cellular Corp. v. FCC, 254 F.3d 78, 86 (D.C. Cir. 2001) (quoting Nat’l Ass’n of Broadcasters v. FCC, 740 F.2d 1190, 1207 (D.C. Cir. 1984)). The FCC generally has broad discretion to control the disposition of its caseload, and to defer consideration of particular issues to future proceedings when it thinks that doing so would be conducive to the efficient dispatch of business and the ends of justice. See GTE Service Corp. v. FCC, 782 F.2d 263, 273–74 (D.C. Cir. 1986) (citing Nader v. FCC, 520 F.2d 182, 195 (D.C. Cir. 1975) and Cellular Mobile Sys. of Penn., Inc. v. FCC, 782 F.2d 182, 197 (D.C. Cir. 1985)). So long as the FCC’s decision to postpone consideration of the transiting issue doesn’t result in unreasonable delay or impose substantial hardship on the CLECs—which hasn’t been shown here—the Commission’s choice to organize its rulemaking docket in this way is lawful. F. Section 271 Pricing and Combination Rules Section 271 of the Act sets conditions for Bell operating companies (the ‘‘BOCs’’) to enter the interLATA long dis- tance market. These conditions include a ‘‘competitive check- list,’’ § 271(c)(2)(B), specifying fourteen conditions that a requesting BOC must satisfy before it may provide inter- LATA service. Checklist item two requires BOCs to provide ‘‘[n]ondiscriminatory access to network elements in accor- dance with the requirements of sections 251(c)(3) and 251(d)(1),’’ § 271(c)(2)(B)(ii), while checklist items four, five, six, and ten require the BOC to provide unbundled access to, 52 respectively, local loops, local transport, local switching, and call-related databases, §§ 271(c)(2)(B)(iv)-(vi),(x). The FCC reasonably concluded that checklist items four, five, six and ten imposed unbundling requirements for those elements independent of the unbundling requirements imposed by §§ 251–52. In other words, even in the absence of impair- ment, BOCs must unbundle local loops, local transport, local switching, and call-related databases in order to enter the interLATA market. Order ¶ ¶ 653–55. But the FCC also found that the BOCs’ unbundling obli- gations under the independent checklist items differed in some important respects from those under §§ 251–52. Two such differences are salient here. First, the Commission determined that TELRIC pricing was not appropriate in the absence of impairment; for elements for which unbundling was required only under § 271, the ruling criterion is the §§ 201–02 standard that rates must not be unjust, unreason- able, or unreasonably discriminatory. Order ¶ ¶ 656–64. Second, the Commission decided that, in contrast to ILEC obligations under § 251, the independent § 271 unbundling obligations didn’t include a duty to combine network ele- ments. The CLEC petitioners object to both of these differences, arguing that the independent § 271 unbundling provisions incorporate all the requirements imposed by §§ 251–52, in- cluding pricing and combination. Because this is an issue of statutory construction, we review under Chevron and defer to the Commission unless Congress has spoken to the precise question at issue (Chevron step one) or the Commission’s interpretation is unreasonable (Chevron step two). With regard to pricing, the CLECs have no serious argu- ment that the text of the statute clearly demonstrates that the § 251 pricing rules apply to unbundling pursuant to § 271 checklist items four, five, six, and ten. The CLECs contend that checklist item two specifies that the § 252(d)(1) pricing rules apply to all unbundled ‘‘network elements,’’ but check- list item two says no such thing. Rather, checklist item two by its terms requires only ‘‘[n]ondiscriminatory access to 53 network elements in accordance with the requirements of sections 251(c)(3) and 252(d)(1)’’—it says nothing suggesting that the requirements of those sections also apply to the independent unbundling requirements imposed by the other items on the § 271 checklist. The CLECs also claim that it was unreasonable for the Commission to apply a different pricing standard under § 271, but we see nothing unreason- able in the Commission’s decision to confine TELRIC pricing to instances where it has found impairment. See generally Order ¶ ¶ 657–64. As to combinations, the CLECs argue that the Supreme Court decisions in AT&T and Verizon establish that the nondiscrimination provision in § 251(n)(3), not its reference to ‘‘combin[ation],’’ provides the basis for the rules that ILECs may not separate already-combined network elements before turning them over to competitors, and that ILECs must combine unbundled network elements when requested to do so by CLECs. See CLEC Br. at 42 (citing AT&T, 525 U.S. at 394, and Verizon, 535 U.S. at 537). CLEC reliance on AT&T and Verizon is misplaced for two reasons. First, as we’ve already held with regard to pricing, § 271 checklist items four, five, six, and ten do not incorpo- rate any of the specific requirements of § 251(c)(3), including the nondiscrimination prohibition specific to that section. Second, neither AT&T nor Verizon holds that the § 251(c)(3) nondiscrimination requirement mandates the combination rules the FCC promulgated under that section; rather, those cases found the nondiscrimination language in § 251(c)(3) ambiguous and deferred to the agency’s reading of it. AT&T, 525 U.S. at 394–95; Verizon, 535 U.S. at 531–38. These holdings don’t necessarily establish that a different rule would be unreasonable. Cf. Rust v. Sullivan, 500 U.S. 173, 186–87 (1991). We agree with the Commission that none of the require- ments of § 251(c)(3) applies to items four, five, six and ten on the § 271 competitive checklist. Of course, the independent unbundling under § 271 is presumably governed by the gen- eral nondiscrimination requirement of § 202. But as the only 54 challenge the CLECs have presented to the FCC’s § 271 combination rules is grounded in an erroneous claim of a cross-application of § 251, we do not pass on whether the § 271 combination rules satisfy the § 202 nondiscrimination requirement. IV. Unbundling of Enhanced Extended Links (‘‘EELs’’) Enhanced extended links (‘‘EELs’’) are high-capacity loop/transport combinations that run directly between an end user (usually a large business customer) and an IXC/CLEC office. Supplemental Order Clarification, 15 FCC Rcd 9587, 9593 (2000), ¶ 10 n.36. EELs can be used to provide local exchange services, but they can also be used to originate and terminate long-distance calls. IXC providers have tradition- ally purchased these services from ILECs for long distance purposes as a special access service, i.e., under the ILEC’s tariff rather than at TELRIC rates. In its first Order implementing the 1996 Act, the FCC did not impose any limits on the telecommunications services that a CLEC could provide with the UNEs to which it was entitled access. Order ¶ 134 & n.446 (citing Third Report and Order, 15 FCC Rcd at 3911–12 ¶ 484 and First Report and Order, 11 FCC Rcd at 15671–72 ¶ 356). But in 1999 the FCC modified this principle with respect to EELs, and issued (as an interim measure) a supplemental order that limited access to EELs as UNEs to those CLECs that would use unbundled EELs to provide ‘‘a significant amount of local exchange service.’’ Supplemental Order, 15 FCC Rcd 1760, 1760 ¶ 2. The FCC subsequently clarified and refined this principle, adopting three ‘‘safe harbors’’ that required CLECs to certify sufficient local traffic percentages in order to qualify for unbundled access to EELs, Supplemental Order Clarifica- tion, 15 FCC Rcd 9587, 9598–60 ¶ 22, and restricting ‘‘com- mingling’’ by CLECs of EELs and tariffed special access services used for interoffice transmission, id. at 9602 ¶ 28. We upheld these rules—which the FCC characterized as ‘‘interim restrictions’’—in Competitive Telecommunications Ass’n v. FCC, 309 F.3d 8 (D.C. Cir. 2002) (‘‘CompTel’’). 55 In the Order under review, the Commission revised its approach to EELs. First, the Commission generalized the principle underlying its earlier EELs rulings by interpreting the unbundling obligations of § 251(d)(3) to apply only to ‘‘qualifying services,’’ defined as ‘‘those telecommunications services that competitors provide in direct competition with the incumbent LECs’ core services.’’ Order ¶ 139. The FCC also decided that, once a CLEC obtained access to a UNE for a qualifying service, the CLEC could use that UNE to provide additional non-qualifying services. Order ¶ 143. Un- der these principles, CLECs are entitled to unbundled EELs only if they use these facilities for local exchange service (which counts as a qualifying service), but not for use exclu- sively for non-qualifying long distance service. Order ¶ ¶ 591, 595. The Commission also changed its strategy for enforcing this basic principle and for preventing ‘‘gaming’’ by carriers that, while not bona fide providers of local service, might seek to take advantage of the low (TELRIC) price of unbundled EELs. It abandoned the ‘‘safe harbor’’ approach, agreeing with the CLECs that this regime had proved intrusive, unworkable, and susceptible to abuse by ILECs. Order ¶ 596 & n.1831, ¶ 614. It also lifted the prohibition on ‘‘comming- ling.’’ Id. ¶ ¶ 579–84. In place of the old restrictions, the Commission established new ‘‘eligibility criteria’’ as prerequi- sites for a competitor to enjoy the access entitlement of a bona fide provider of a qualifying service. Id. ¶ ¶ 591–611. Each applicant would have to show, first, that it had a state certification to provide local voice service and, second, that at least one local number was assigned to each circuit to be acquired as a UNE. Id. ¶ ¶ 597, 601–02. In addition, the Commission imposed a variety of technical requirements aimed at preventing firms from gaming the system. Id. ¶ ¶ 597, 603–11. While the Commission admitted that none of the anti- gaming requirements by itself would prevent gaming, it con- cluded that they were ‘‘collectively sufficient to restrict the availability of these UNE combinations to legitimate provid- ers of local voice service.’’ Order ¶ 600 (emphasis in original). 56 It justified this conclusion on the logic that ‘‘the burdens and inefficiencies for a provider to meet these criteria for non- qualifying service would deter a carrier of non-qualifying service from re-designing its operations to subvert our rules.’’ Id. The Commission also allowed CLECs that met the eligibility criteria, but that currently purchased EELs from ILECs as special access services at wholesale rates (i.e., not TELRIC), to ‘‘convert’’ these wholesale services to UNEs. Order ¶ 586. The CLECs object both to the concept of distinguishing between qualifying and non-qualifying service, and to the eligibility criteria used to implement the distinc- tion. A. The Qualifying Service/Non–Qualifying Service Distinc- tion The CLECs object to the FCC’s decision that long distance is not a ‘‘qualifying service,’’ claiming that this conclusion is foreclosed by §§ 251(c)(3) and 251(d)(2)(B) of the Act. Long distance services, including the origination and termination functions performed by EELs, are clearly ‘‘telecommunica- tions services,’’ and § 251(d)(2) directs the Commission to provide unbundled access to elements where the lack of such an element ‘‘would impair the ability of the telecommunica- tions carrier seeking access to provide the services it seeks to offer.’’ (The Commission assumes, as we believe it must, that the reference to ‘‘services’’ in § 251(d)(2) is meant to refer to the ‘‘telecommunications services’’ covered by § 251(c)(3). Order ¶ 138). The CLECs therefore argue that the FCC cannot arbitrarily exclude them from this impairment analy- sis. The Commission asserts that ‘‘section 251(d)(2)’s reference to the ‘services that [the carrier] seeks to offer’ is ambiguous as to the question of which services we should analyze in the context of our impairment analysis.’’ Order ¶ 137 (alteration in original). Having thus ‘‘conclude[d] that the language of section 251(d)(2) is ambiguous concerning the scope of the impairment inquiry,’’ Order ¶ 138, the FCC looked to the history and purposes of the Act and concluded that ‘‘a reason- able interpretation of the statute’’ would restrict the impair- 57 ment inquiry to those services offered in direct competition with ILEC core services such as local voice and data services, id. ¶ 139. In CompTel we agreed with the Commission that § 251(d)(2) was ambiguous on the question whether the FCC could make impairment decisions on a service-by-service ba- sis. 309 F.3d at 12. That is, we considered a situation where an element could be used to provide services A and B, and a carrier requested unbundling for both. We held that the Commission acted reasonably in disaggregating the impair- ment issue, and in ordering unbundling only with respect to the service for which it found impairment. 309 F.3d at 12–13 (service-by-service impairment analysis permissible); 14 (im- pairment finding made by FCC as to local service but not as to long distance). Here the Commission asserts an entirely different sort of statutory ambiguity, namely, whether long distance services are ‘‘services’’ at all and therefore require the Commission, on request, to perform an impairment analysis. We are not persuaded by the Commission’s claim that the ambiguity regarding the permissibility of service-by-service impairment determinations extends to whether long distance services (or other telecommunications services that do not compete direct- ly with ‘‘core’’ ILEC services) are ‘‘services’’ within the meaning of § 251(d)(2) in the first place. Even under the deferential Chevron standard of review, an agency cannot, absent strong structural or contextual evidence, exclude from coverage certain items that clearly fall within the plain mean- ing of a statutory term. The argument that long distance services are not ‘‘telecommunications services’’ has no sup- port. The Commission does suggest that the ‘‘impairment’’ re- quirement is closely linked to natural monopoly conditions that prevail only with respect to the core ILEC services that the Commission defined as ‘‘qualifying services.’’ FCC Br. at 77 (citing USTA I, 290 F.3d at 427). But that argument addresses impairment, not the definition of ‘‘services.’’ We therefore remand those sections of the Order (¶ ¶ 132–53) 58 resting the exclusion of ‘‘non-qualifying’’ services on the Com- mission’s reading of the phrase ‘‘telecommunications services’’ in § 251(d)(2)(B). This does not, of course, necessarily invalidate the Commis- sion’s effort to prevent the use of EELs for long distance service. The CLECs have pointed to no evidence suggesting that they are impaired with respect to the provision of long distance services, and in CompTel we emphatically held that the Act did not bar a service-by-service analysis of impair- ment. 309 F.3d at 12–14. The CLECs do not deny that they have been able to purchase use of EELs as ‘‘special access.’’ As we noted with respect to wireless carriers’ UNE demands, competitors cannot generally be said to be impaired by having to purchase special access services from ILECs, rath- er than leasing the necessary facilities at UNE rates, where robust competition in the relevant markets belies any sugges- tion that the lack of unbundling makes entry uneconomic. On remand, therefore, the Commission will presumably turn to the issue of impairment. Because it may well find none with reference to long distance service, we now turn to the eligibility criteria. B. The EEL Eligibility Criteria Both the CLECs and the ILECs object to the FCC’s eligibility criteria. The CLECs say they are too stringent and are over-inclusive insofar as they preclude access to EELs used to provide services for which CLECs are im- paired. The ILECs claim they are too lax and are under- inclusive insofar as they fail to prevent CLECs from using unbundled EELs exclusively for long distance services. We think that the Commission’s eligibility criteria, though imperfect, reflect a reasonable effort to establish an admin- istrable system that balances two legitimate but conflicting goals: the prevention of ‘‘gaming’’ by CLECs seeking to offer services for which they are not impaired, and the preserva- tion of unbundled access for CLECs seeking to offer services for which they are impaired. We accord considerable defer- ence to such administrative determinations, see WorldCom, 59 Inc. v. FCC, 238 F.3d 449, 459 (D.C. Cir. 2001); Home Box Office, Inc. v. FCC, 567 F.2d 9, 60 (D.C. Cir. 1977), and find that the proxies the FCC used, though imperfect (as the Commission itself candidly admits, Order ¶ 600), are neither inconsistent with the Act nor arbitrary and capricious. The Commission also satisfactorily explained both the problems with the regime previously in place (which the ILECs thought should be retained), Order ¶ 614, and with the CLECs’ proposed alternatives, id. ¶ ¶ 615–19. The ILECs make an independent attack on the Commis- sion’s decision to allow ‘‘conversions’’ of wholesale special access purchases to UNEs. As we discussed in the section on wireless carriers, the presence of robust competition in a market where CLECs use critical ILEC facilities by purchas- ing special access at wholesale rates, i.e., under § 251(c)(4), precludes a finding that the CLECs are ‘‘impaired’’ by lack of access to the element under § 251(c)(3). We realize that this might create anomalies, as CLECs hitherto relying on special access might be barred from access to EELs as unbundled elements, while a similarly situated CLEC that had just entered the market would not be barred. On the other hand, if history showed that lack of access to EELs had not impaired CLECs in the past, that would be evidence that similarly situated firms would be equally unimpaired going forward. Because we have already determined that we must remand to the Commission, given the invalidity of the line it drew between qualifying and non-qualifying services, the Commission can consider and resolve any potential anomaly on remand. V. Miscellaneous There remain two loose ends, attacks on the Order by the National Association of State Utility Consumers Advocates (‘‘NASUCA’’) and by a group of state petitioners. We find that NASUCA lacks standing and that the state petitioners’ claim is unripe. A. NASUCA’s Standing NASUCA is a non-profit association of offices, each of which has been designated by its respective state govern- 60 ments to represent the interests of utility consumers in regulatory and judicial proceedings. We agree with the Commission that NASUCA has failed to establish standing pursuant to the requirements of Sierra Club v. EPA, 292 F.3d 895, 899–901 (D.C. Cir. 2002), though for different reasons than those advanced by the Commission. Under Sierra Club, ‘‘a petitioner whose standing is not self- evident should establish its standing by the submission of its arguments and any affidavits or other evidence appurtenant thereto at the first appropriate point in the review proceed- ing.’’ 292 F.3d at 900. A petitioner’s standing is self-evident only if ‘‘no evidence outside the administrative record is necessary for the court to be sure of it.’’ Id. at 900. Con- trary to the Commission’s assertions, we believe that no evidence outside the administrative record is necessary to explain how (on NASUCA’s view of the merits) the Order injures the consumers that NASUCA claims to represent. See NASUCA ex parte letter (Feb. 13, 2002) at 2–3. On the theories advanced by NASUCA, consumers would enjoy a superior price/quality trade-off in telephone service if the Commission accepted its analysis. But it is not at all self- evident from the record that NASUCA meets the association- al standing criteria established in Hunt v. Washington State Apple Advertising Commission, 432 U.S. 333, 344–45 (1977), for entities that are not voluntary membership organizations. See also Fund Democracy, LLC v. SEC, 278 F.3d 21, 25–26 (D.C. Cir. 2002); Am. Legal Found. v. FCC, 808 F.2d 84, 89– 90 (D.C. Cir. 1987). Although utility consumer interests are clearly affected by the Order, nothing in the administrative record or NASUCA’s opening brief establishes that NASUCA is qualified to represent those interests in federal court. We therefore conclude that NASUCA lacks standing and do not reach the merits of its claims. B. Ripeness of the State Preemption Claims The state petitioners argue that the Order improperly preempts state unbundling regulations that exist independent of the Commission’s federal unbundling regulations enacted pursuant to § 251. Specifically, the state petitioners point to 61 ¶ 195 of the Order, which allows ‘‘[p]arties that believe that a particular state unbundling obligation is inconsistent with the limits of section 251(d)(3)(B) and (C)’’ to seek a declaratory ruling from the Commission, and further predicts that state unbundling requirements for elements that the FCC has determined need not be unbundled under § 251(d)(2) are ‘‘unlikely’’ to be found consistent with the Act. The state petitioners’ challenge to the preemptive scope of the Order is not ripe. The general prediction voiced in ¶ 195 does not constitute final agency action, as the Commission has not taken any view on any attempted state unbundling order. Nor does the states’ claim present a purely legal question, as they acknowledge that Commission regulations will lawfully preempt in some circumstances. See Alascom, Inc. v. FCC, 727 F.2d 1212, 1218–20 (D.C. Cir. 1984); see also Time Warner Entertainment Co. v. FCC, 56 F.3d 151, 193–96 (D.C. Cir. 1995). Besides, the state petitioners have not—and probably could not—identify any substantial hardship that they would suffer by deferring judicial review of the preemp- tion issues until the FCC actually issues a ruling that a specific state unbundling requirement is preempted. We therefore hold the challenge unripe. VI. Conclusion To summarize: We vacate the Commission’s subdelegation to state commissions of decision-making authority over im- pairment determinations, which in the context of this Order applies to the subdelegation scheme established for mass market switching and certain dedicated transport elements (DS1, DS3, and dark fiber). We also vacate and remand the Commission’s nationwide impairment determinations with re- spect to these elements. We vacate the Commission’s decision not to take into account availability of tariffed special access services when conducting the impairment analysis, and we therefore vacate and remand the decision that wireless carriers are impaired without unbundled access to ILEC dedicated transport. 62 We vacate the Commission’s distinction between qualifying and non-qualifying services, and remand (but do not vacate) the decision that competing carriers are not entitled to un- bundled EELs for provision of long distance exchange ser- vice. We remand the Commission’s decision to exclude entrance facilities from the definition of ‘‘network element’’ for further development of the record to allow proper judicial review. The petitions for review are otherwise denied, except for NASUCA’s petition, which is dismissed for want of standing, and the state commissions’ (and that part of the ILEC petitions relating to compensation for modification of ele- ments), which are dismissed as unripe. The ILECs’ manda- mus petitions are dismissed as moot. As to the portions of the Order that we vacate, we tempo- rarily stay the vacatur (i.e., delay issue of the mandate) until no later than the later of (1) the denial of any petition for rehearing or rehearing en banc or (2) 60 days from today’s date. This deadline is appropriate in light of the Commis- sion’s failure, after eight years, to develop lawful unbundling rules, and its apparent unwillingness to adhere to prior judicial rulings. So ordered.