BEFORE THE WASHINGTON UTILITIES AND TRANSPORTATION COMMISSION In the Matter of the Pricing Proceeding for Interconnection, Unbundled Elements, Transport and Termination, and Resale NO. UT-960369 In the Matter of the Pricing Proceeding for Interconnection, Unbundled Elements, Transport and Termination, and Resale for US WEST COMMUNICATIONS, INC. NO. UT-960370 In the Matter of the Pricing Proceeding for Interconnection, Unbundled Elements, Transport and Termination, and Resale for GTE NORTHWEST, INC. NO. UT-960371 BRIEF OF COMMISSION STAFF I. LEGAL PARAMETERS AND ISSUES A. Federal Law It is the policy of the United States to promote competition in the telecommunications area. Congress enacted this policy by passing Telecommunications Act of 1996 (“Act”), which favors and promotes local exchange competition. Telecommunications Act of 1996, Pub. L. No. 104 -104, 110 Stat. 56 (1996). The Act is comprised of amendments to the Communications Act of 1934. The Act will be codified in scattered sections of Title 47 of the United States Code. The Act requires incumbent local exchange companies (“LECs”) to unbundle their networks and provide access to unbundled network elements (“UNEs”), interconnect with new entrants, and offer their retail services to resellers at wholesale rates. 47. U.S.C. § 251. More specifically at issue in this case are the costs and prices new entrants must pay for UNEs, interconnection and resold services. With respect to pricing, the Act authorizes state commissions to set rates that conform to the following standards: Interconnection and Network Element Charges - Determinations by a State commission of the just and reasonable rate for the interconnection of facilities and equipment for purposes of subsection (c)(2) of section 251, and the just and reasonable rate for network elements for purposes of subsection (c)(3) of such section-- (A) shall be-- (i) based on the cost (determined without reference to a rate-of-return or other rate-based proceeding) of providing the interconnection or network element (whichever is applicable), and (ii) nondiscriminatory, and (B) may include a reasonable profit. 47 U.S.C. § 252(d)(1), and Wholesale prices for telecommunications services.--For the purposes of section 251(c)(4), a State commission shall determine wholesale rates on the basis of retail rates charged to subscribers for the telecommunications service requested, excluding the portion thereof attributable to any marketing, billing, collection, and other costs that will be avoided by the local exchange carrier. 47 U.S.C. § 252(d)(3). The pricing standards for transport and termination of traffic are set forth in 47 U.S.C. § 252(d)(2). The Act requires incumbent telephone companies to negotiate with new entrants for interconnection or purchase of network elements. See 47 U.S.C. § 251. The negotiation procedures are governed by Section 252 of the Act. If the negotiating companies cannot reach agreement, Section 252 provides that the parties may request the state utilities commission to arbitrate any unresolved issues. The Act preserves the Commission’s “historical role in telecommunications regulation, given the intrastate quality of the local competition provisions of the Act.” Iowa Util. Bd. v. Federal Communications Comm’n, Nos. 96-3321 et al., slip op. at 19 (8th Cir. July 18, 1997). B. Washington Law The Washington Legislature has made competition in the telecommunications markets a matter of state policy. RCW 80.36.300. This policy mandate was reiterated in Electric Lightwave, Inc. v. Utilities and Transp. Comm’n, 123 Wn.2d 530, 869 P.2d 1045 (1994), where the court rejected the contention that exclusive franchise service areas in the telecommunications industry were permissible in Washington. Thus, state law and policy explicitly promotes competition in the local exchange. In keeping with its mandate to promote competition in the local exchange, the Commission three years ago entered into a proceeding to decide the “terms and conditions under which competitors for local exchange service will interconnect their networks so that they can exchange traffic between their customers.” Fourth Supplemental Order Rejecting Tariff Filings and Ordering Refiling, Granting Complaints, In Part, Washington Utilities and Transportation Commission v. US West Communications, Inc., Docket Nos. UT-941464, et al, at 6 (Oct. 31, 1995) (“Interconnection Order”). In that order, the Commission recognized the importance of interconnection in ensuring that competition in the local exchange becomes a reality: Technically and economically efficient interconnection of the incumbent LEC and new entrant ALEC networks is essential to the emergence of a competitive local exchange market. Denial of technically and economically efficient interconnection arrangements creates a barrier to entry. The Commission is persuaded that ALECs should have considerable flexibility to configure their networks in a manner they deem suitable. Id. 45. This case essentially represents a continuation of the unbundling and resale process commenced by the Commission in the local interconnection case. This process, it should be noted, predated the enactment of the federal Telecommunications Act. In the Interconnection Order at 51, the Commission stated: [U]nbundling of the local loop is essential to the rapid geographic dispersion of competitive benefits to consumers and is in the public interest. Unbundling allows customers greater opportunity to choose between a diversity of products, services, and companies. Unbundling also allows for efficient use of the public switched network, reduces the likelihood of inefficient network over-building, and ensures that competition is not held hostage by being bundled with bottleneck functions. The Commission further found that while facility-based competition may be “the preferred future,” retail competition through a strong resale market “may indeed be an important step in the long-term development of local competition;” and that allowing access to and resale of unbundled parts of an incumbent’s network will allow those parts of the local exchange market that can support competition to move forward with competition without being held back by those parts of the market that are still characterized by monopoly. Id. The Commission has jurisdiction over the terms and conditions under which telecommunications companies interconnect their networks. RCW 80.36.140 provides the Commission with authority to order unbundling. The second paragraph of this statute allows the Commission to determine the just, reasonable, proper, adequate and efficient practices to be observed and used if it determines after hearing that a company’s practices are unjust or unreasonable. The term “practice” is clearly broad enough to cover the offering of elements or services on a bundled or unbundled basis. Interconnection Order, at 51. RCW 80.36.300(5) provides additional statutory authority. There, the Legislature declared it state policy to “promote diversity in the supply of telecommunications services and products in telecommunications markets throughout the state.” In the Interconnection Case, the Commission found that “[I]t is clear from this record that unbundling and resale are key elements in fostering diversity in supply of services and products.” Interconnection Order at 52. The ultimate issue in this case, to be determined in the second phase of the proceeding, concerns the prices that will be charged for unbundled elements and resold bundled services. RCW 80.36.080 requires that these prices be just, fair, reasonable, and not unduly discriminatory. RCW 80.36.170-.180, in addition, prohibit unreasonable preferences and rate discrimination by any telecommunications company, and grant the Commission primary jurisdiction to enforce these sections. II. COST METHODOLOGY: PRINCIPLES A. The Goals of the Generic Proceeding The overall goal of this proceeding is to establish the prices that local exchange carriers will charge for unbundled network elements (UNEs) and resold bundled services. In this initial phase, however, the goal is simply to establish the costs for such elements and resold services. This exercise entails establishing the proper cost methodology, models, and inputs to be used for estimating the costs of unbundled network elements, and for estimating the avoided costs arising out of the resale of bundled services at wholesale rates. By conducting this docket as a generic proceeding, rather than a series of individual arbitrations, the Commission has afforded greater protection to the interests of all parties by allowing all to participate in the cost and price analysis, to comment, and provide alternative approaches. See Order on Sprint’s Petition to Intervene and to Establish Generic Pricing Proceeding, In the Matter of the Petition for Arbitration of an Interconnection Agreement Between AT&T Communications. of the Pacific Northwest, Inc. and GTE Northwest Inc., Docket No. UT-960307 et. al, at 4-5 (Oct. 23, 1996). The public interest is served by cost and price determinations that further efficient and effective competition. For this reason, it is important that the costs established in this phase of the proceeding accurately reflect total element long-run incremental costs (TELRIC). The costs set in this proceeding will serve as a price floor. Costs set too high will create artificial barriers to competition, depriving consumers of the lower prices available in an effectively competitive market and curtailing technological change. Costs set too low will allow for predatory pricing that will likewise create barriers to competition. Costs and prices appropriately set will provide incentives for economically efficient investment in the telecommunications network. Transcript, Vol. 15, at 1556 (Blackmon). TELRIC costs are determined through the use of cost models. While these models are important, the Commission should not spend an undue amount of time and resources attempting to resolve the hundreds of issues, many of them minute, regarding the arcane complexities of cost models submitted by the parties to this proceeding. The Commission, as more fully set forth below, should keep in mind that the purpose of the cost models is to determine price floors and should review those models with that purpose in mind. B. How Should Cost Models Be Used by the Commission? An analytical model is, in general terms, a simplified representation of some aspect of the real world. Analysts use models to organize the complex information that presents itself in the real world into some orderly form. Models are, by definition, simplifications or abstractions that omit some information. A model can be very powerful analytical tool. Like a microscope or a telescope, it permits an analyst to focus in on the key aspects of a situation and thereby solve problems that, in the absence of a model, would be hopelessly complex. The analytical models on the record in this case are computer models designed or used to estimate the cost of constructing and operating the existing public switched telephone network. That network is exceedingly complex, encompassing millions of access lines and thousands of switches, interoffice transmission facilities, signaling links, and other elements. Cost models are used to sort through the complexity of that network, organize it into similar elements having similar costs, and estimate the cost of those elements. These cost models lend themselves to two basic purposes. First, they can be used to measure the cost that would be incurred should the network be reconstructed under certain specified conditions, such as the "scorched node" assumption. Second, they can be used to disaggregate the otherwise undifferentiated costs of the network into various element costs, so that the cost of a loop can be separated from the cost of a switch, and even the cost of a 10,000-foot loop in an exchange of a certain size can be separated from the cost of a 10,000-foot loop in an exchange of different size. In other words, one might use a model to estimate what it would cost to build the whole thing over again or to estimate what it would cost to build only a portion of the existing network. The most common application of cost models in the regulation of utilities has been to estimate the long-run incremental cost of a particular service. Such estimates have been used by regulators and others to guard against uneconomic and predatory pricing of individual services. It is well established in economic theory that a firm should offer a service only if its price covers the long-run incremental cost of that service. It is a longstanding practice of this commission and others to require that regulated companies demonstrate that prices cover cost as measured in this fashion, and cost models have been used to make that estimate. The cost model provides an estimate of the price floor, i.e., the lowest price at which the firm should be allowed to offer a service. It is crucial that the Commission recall how cost models and their estimates of long-run incremental cost have not been used: to establish revenue requirement or prices. For example, in the U S WEST rate case concluded last year, long-run incremental cost estimates for virtually every service were introduced on the record and critiqued and analyzed by various parties. No party in that case suggested that the revenue requirement of the company should be based on those incremental cost estimates, so that the price of each service was equal to its long-run incremental cost estimate. Incremental cost pricing was proposed for some individual services, such as AT&T’s recommendation that access charges be priced at incremental cost, but such proposals were largely rejected by the Commission. In short, cost models and the cost estimates they produce have not and should not be used as a substitute for establishing a revenue requirement and prices, as is traditionally done in the regulation of a monopoly telephone company. Both the value and the limitation of cost models continue in this proceeding, where the Commission has the task of establishing cost-based prices for unbundled network elements. A good cost model will enable the Commission to unbundle the costs of the telephone network just as the network itself is being unbundled. Where there is no good information on the actual cost of constructing and operating individual network elements, a cost model can assist in disaggregating costs, an important step to the unbundling of elements. Moreover, a good cost model will enable the Commission to estimate the forward-looking economic cost of various elements. This forward-looking, incremental measure of cost remains a key constraint on the pricing of network elements. It is the floor below which the price of an element cannot go without violating rules of fair competition and producing economically inefficient investment and consumption decisions. One may well ask, if cost models are only being used to set price floors and not to set prices, what is the point of expending so much effort on them? Does one really have to worry about incumbent companies setting their prices too low for bottleneck facilities? The answer to the second question is, yes, the Commission should be concerned about predatory pricing of unbundled network elements. While incumbents may appear to have a lock on virtually every part of the network today, competitive alternatives are likely to appear in some areas before others. Switching and interoffice transport, for example, have to date experienced more competitive entry than the provision of loops. In this circumstance incumbents might perceive an advantage to themselves from pricing switching and transport below cost, and the Commission has a responsibility to ensure that such predatory pricing does not occur. As to the first question -- why so much effort on cost models if their purpose is limited in this way -- Staff believes an excessive amount of effort and resources has gone into the estimation of forward-looking economic costs of unbundled network elements. This Herculean effort began with the FCC’s requirement that prices be based on such cost estimates. If prices were to be set by rigid formula based on a model’s calculation of incremental cost, then millions or even billions of dollars were at stake in the determination of which model and which inputs should prevail. While those rules were stayed and then vacated by the U.S. 8th Circuit Court of Appeals, momentum has propelled the cost model juggernaut to this point, where the Commission has before it a record of arcane disputes whose complexity far exceed their importance. Ironically, the FCC criticized embedded cost approaches because, “The legislative history demonstrates that Congress was eager to set in motion expeditiously the development of local competition and intended to avoid imposing the costs and administrative burdens associated with a traditional rate case.” In the Matter of Implementation of the Local Competition Provisions in the Telecommunications Act of 1996, “First Report and Order, FCC 96-325, August 8, 1996, ¶704. It was Staff’s concern about this excess that led to the introduction of Dr. Blackmon’s testimony on pricing principles that would otherwise be addressed in the costing phase of this proceeding. Staff sought, in essence, to de-escalate the battle of cost models and place them in the proper context. State policy has long been to encourage competition as a more effective substitute for regulation as a check on market power. Passage of the federal telecommunications act escalated that policy to the national level. The Commission should ensure that this policy is not thwarted by the substitution of regulation by cost model for the more traditional regulation by income statement and balance sheet. Staff strongly encourages the Commission to keep cost models in the proper perspective. Cost model sponsors and critics have expended a huge level of effort and expense and have raised countless disputes in this proceeding. It would be a mistake for the Commission to match that effort by carefully resolving each of the countless disputes and then to assume that the result was suitable for anything more than disaggregating costs and establishing price floors. To be sure, there are legitimate issues that must be resolved concerning the structure of these models and the inputs used therein. Regardless of how these models are perfected, their role in the establishment of prices should be limited as described above. C. What Criteria Should the Commission Follow in Examining Cost Models? Several cost models have been submitted for consideration in this proceeding: (1) the Hatfield proxy model, sponsored by AT&T/MCI; (2) the Benchmark Cost Proxy Model (BCPM), sponsored by United/Sprint; (3) the proprietary in-house COSTMOD system, sponsored by GTE; and (4) the proprietary in-house models (including RLCAP, SCM as well as signaling and transport models) sponsored by US West. Cost models are sets of mathematical equations designed to estimate the long-run incremental costs of unbundled network elements. They generally do so by calculating an amount of investment required to provision the element, and then applying an annual cost factor to the investment to determine the annual costs of the element. All the models use simplified network designs to estimate costs. Ex. 104, Direct Testimony of Thomas L. Spinks, at 17. A cost model should first, adhere to sound engineering and economic principles. It should use the most current technology and efficient design in building a telecommunications network. It must also be flexible and transparent. The latter characteristic includes a model’s openness to inspection and criticism and its ability to accommodate many different inputs and assumptions. Id. at 17. This Commission has previously indicated its frustrations with cost models that it has described as “cryptic” or a “black box”. In the recent US West rate case, the Commission stated that any acceptable costing methodology must be “a transparent, rational, stable, consistent, and understandable approach, that will continue to be viable and applicable in determining costs for services in the foreseeable future.” Ninth Supplemental Order Granting Respondent’s Oral Motion for Continuance, Washington Utilities and Transportation Commission v. US West Communications, Inc., Docket No. UT-950200, at 2 (Oct.19, 1995). The Commission continued that there must be “adequate information to allow parties to proceedings involving cost issues to have the ability to understand assumptions used, to review and analyze the effect of inputs and outputs, and to modify and model different inputs and assumptions.” Id. Finally, cost models should not be evaluated upon their ability to duplicate the actual costs incurred on the books of the incumbent LECs. The present exercise involves a determination of forward-looking economic costs, not simply a replication of the past. This aspect is elaborated upon below. In summary, a cost model should be adopted by this Commission that shows exactly where and how costs are developed, the assumptions made, and the formulas and inputs used. It should be an open model that can be shown to the public, rather than one which depends upon large amounts of proprietary information that must be kept concealed. Staff disagrees with the positions of certain parties (for example, TCG) that the Commission need not select any cost model at all in this proceeding. Transcript, Vol. 15, at 1718-20 (Spinks). D. TELRIC Total Element Long Run Incremental Cost (TELRIC) is, when applied to unbundled network elements, the same basic measure of cost that has been used by the Commission for several years for services. TELRIC is the term that was coined by the FCC in its interconnection/unbundling decision of August 8, 1996. “In the Matter of Implementation of the Local Competition Provisions in the Telecommunications Act of 1996,” First Report and Order, FCC 96-325, August 8, 1996 (hereafter, “First Report and Order”). When applied to services, the acronym is TSLRIC, but there is no difference at a conceptual level. Of particular importance are that: (1) Both TELRIC and TSLRIC involve the application of forward-looking economic costs; and (2) Neither TELRIC nor TSLRIC include any common or overhead costs. Ex. 107, Direct Testimony of Glenn Blackmon, at 4. The FCC’s First Report and Order summarized the application of TELRIC as follows: The increment that forms the basis for a TELRIC study shall be the entire quantity of the network element provided. As we have previously stated, all costs associated with the providing the element shall be included in the incremental cost. Only forward-looking, incremental costs shall be included in a TELRIC study. Costs must be based on the incumbent LEC’s existing wire center locations and most efficient technology available. Id. at ¶690. The TELRIC methodology should employ what is referred to as the “scorched node” approach. That is, one assumes that the LEC’s current wire center locations will remain in place, but that the remainder of the network would be built from scratch, using the most efficient technology. As the FCC explained, this approach mitigates LEC concerns that a forward-looking methodology will ignore existing network design, while relying upon new, efficient technology that is compatible with the existing infrastructure. See id. at ¶685. Consistent with TELRIC, one should attempt to attribute costs to individual elements wherever possible. However, simply splitting a service into its various elements may not account for all costs that are incurred in the provisioning of those elements, because some costs of services are not attributable to any unbundled elements in particular. Staff has recommended that a factor of 20% be added as part of the TELRIC loop estimate to account for additional attributable costs. Ex. 104, Direct Testimony of Thomas L. Spinks, at 26-27. The 20% attributable cost figure is derived from US West’s response to Commission Staff Data Request No. 43. Staff initially expressed reservations concerning the use of this figure if it in fact included “shared residual costs”--that is, certain common costs previously rejected by this Commission in Docket UT-950200 (the recent US West rate case) as not properly part of a TSLRIC calculation. Staff is satisfied that such costs are not included in the 20% calculation, and notes that other parties have not testified otherwise. This is consistent with the FCC’s approach. See First Report and Order, at ¶682. Attributable costs do not, however, include common or overhead costs. The latter are not properly included in any definition of TELRIC. E. Common Costs Common or overhead costs should not be considered by the Commission in this phase of the generic proceeding. They are not relevant to the determination of TELRIC-based price floors for unbundled network elements. Nor should TELRIC include any uniform allocator of common costs. Staff notes that the FCC, in its pricing rules recently vacated by the 8th Circuit Court of Appeals Iowa Util. Bd. v. Federal Communications Comm’n, Nos. 96-3321 et al., slip op. at 6-7, 19-20 (8th Cir. July 18, 1997)(abbreviated slip opinion)., stated that prices should equal TELRIC plus an allocation of common costs. First Report and Order, at ¶672. This Commission should not employ this approach, because a uniform allocation of common costs will lead to arbitrary prices that are economically inefficient. Staff recognizes that common costs are real costs that cannot be assumed away, and that a firm must recover these costs in order to stay in business. The Commission should address this issue by setting some or all of the rates for unbundled network elements above incremental cost. That, however, is an issue for the pricing phase of this proceeding. Ex. 107, Direct Testimony of Glenn Blackmon, at 4-7. AT&T witness Dr. Thomas Zepp recommended that a uniform factor be applied to TELRIC results to cover what are generally considered common costs. Ex. 163, Rebuttal Testimony of Thomas M. Zepp, at 5. He cites a regression analysis indicating that common costs vary with the scale of a company’s operations. Staff does not disagree with the argument that some costs that may be accounted for as administrative or overhead should be directly attributed to specific elements. As described above, we recommend that those costs be attributed using the 20% factor. However, Staff believes Dr. Zepp goes too far in suggesting that the Hatfield model’s TELRIC estimates (including a 10.4% overhead factor) “include everything that would be needed to provide unbundled network elements.” Id. at 4. A TELRIC estimate is, by definition, the additional cost that a firm would incur to provide a particular network element, if that firm is already providing every other relevant network element. While Staff believes the Hatfield model does a good job of producing TELRIC estimates, we cannot agree that those estimates represent cost ceilings, as Dr. Zepp suggests. If it is true that “a new facilities-based entrant would never incur costs higher than those produced by the Hatfield model,” id. at 5, then it is hard to see what harm would come to competitors from allowing incumbents to price above that level. Any effort by incumbents to price above the level sustainable in a competitive market would simply result in loss of market share to new facilities-based entrants. Ex. 107, Direct Testimony of Glenn Blackmon, at 8. The very existence of this proceeding suggests that, in fact, the Hatfield model results do not represent the stand-alone costs or “price ceilings” of unbundled network elements. It would be a disservice to the Hatfield model to overstate its capability in this fashion. F. Actual or Embedded Costs and Their Utility Actual or embedded costs are not a proper component of a TELRIC-based cost calculation. In fact, the suggestion that the forward-looking competitive environment should be determined by the past monopoly environment is self-contradictory. It is true that current costs and prices may sometimes be the best estimate of forward-looking costs and prices. This exception should not be confused with improper proposals to validate cost models by using historic, embedded data. Staff disagrees entirely with the apparent position of GTE in this proceeding that embedded costs are the sine qua non of a proper costing methodology. Yet GTE’s prefiled testimony, as well as its depositions and cross-examination of Staff witnesses, all feature the repetition of this erroneous theme. As Staff witness Mr. Spinks points out: [GTE witness] Dr. Duncan’s position regarding the purpose of the proxy models is not clear. He criticizes the [Hatfield] model for being independent of past ILEC investment decisions, simulating a network different from the actual ILEC network. Yet none of the discussion of TELRIC principles suggests that forward-looking economic costs should be dependent on past ILEC investment decisions, or that a model should build a network like “the actual ILEC network” with the exception of the current switch locations. Dr. Duncan states that the Hatfield model assumes the present network will be scrapped. That is not true unless one assumes the present ILEC network is, or should be, the starting point for estimating forward looking network costs. But the purpose of the costing exercise in estimating forward looking economic costs is to answer the question of what it would cost to provision the network today if the company were not locked into past investment decisions . . . not whether the model duplicates the existing design used by GTE. Ex. 106, Rebuttal Testimony of Thomas L. Spinks, at 2-3 (citations omitted). On cross-examination, counsel for GTE persisted in its contention that Staff lacked “empirical evidence” of GTE’s actual past experience, and took Staff to task for not simply plugging into the cost models the numbers found on GTE’s books. Yet, the task at hand necessarily involves the use of thought, judgment and estimation of what forward-looking economic costs will be. There is no shame in this. It will not suffice simply to plug in the past figures generated by a monopoly provider, since, as Mr. Spinks explained, Necessarily when you’re looking to the future, the empirical data doesn’t exist yet, since companies have not operated in a competitive environment [and] structure that would be the necessary condition that you would need to generate empirical data. That condition has not existed yet. . . . All I can tell you is t