In the Matter of ) ) Federal-State Joint Board ) CC Docket No. 96-45 on Universal Service )
GTE's COMMENTS IN RESPONSE TO QUESTIONS
GTE Service Corporation and its affiliated domestic telephone operating companies
Richard McKenna, HQE03J36
GTE Service Corporation
P.O. Box 152092
Irving, TX 75015-2092
(214) 718-6362
August 2, 1996 Their Attorneys
TABLE OF CONTENTS PAGE INTRODUCTION AND SUMMARY 1 RESPONSES 5
Before the
FEDERAL COMMUNICATIONS COMMISSION
Washington, DC 20554
In the Matter of ) ) Federal-State Joint Board ) CC Docket No. 96-45 on Universal Service )
GTE's COMMENTS IN RESPONSE TO QUESTIONS
GTE Service Corporation and its affiliated domestic telephone operating companies ("GTE"), responding to the FCC's Public Notice DA 96-1078, 1996 FCC LEXIS 3466, proposing seventy-two questions with reference to the captioned proceeding ("D.96-45") and recommended action pursuant to the Telecommunications Act of 1996 (the "1996 Act"), hereby offer the following comments in response to questions:
INTRODUCTION AND SUMMARY[1]
The Commission, taking into account the Joint Board's recommendations, must adopt and implement an integrated plan that is designed to achieve the universal service goals of the 1996 Act as well as the statute's pro-competitive and deregulatory goals. It must be stressed that GTE's plan, as described in its prior comments and in the responses to questions infra, is carefully structured as an entire program, with each part related constructively to each other part. GTE emphasizes the importance of avoiding arbitrarily picking pieces of this program and layering them onto incompatible concepts and schemes. In particular, nothing is more far-fetched than proposals that universal service support be available to any party that qualifies as an eligible telecommunications carrier under [[section]]214(e). Any such asymmetric plan would leave massive obligations on certain competitors while granting the same universal service support to largely unregulated competitors. This would: (i) be automatically ineffective; (ii) fail to preserve and advance universal service; (iii) destroy the workability of GTE's (or any) proposal; (iv) prevent the Commission from carrying out its assignment under [[section]]254 to develop an effective and sufficient universal service plan; and (v) be contrary to the pro-competitive and deregulatory goals of the 1996 Act.
Existing rates for local service can be considered affordable because they have been the subject of decades of attention from federal agencies, including the FCC, and from state regulatory agencies aimed at keeping local service prices low. However, in establishing a reference point as to what is "affordable," the Commission and the Joint Board should not base their finding on the rate levels currently in place, but should instead make an objective determination of how much customers should be expected to pay. Indeed, it is GTE's opinion that in many areas existing rates do not exceed, and in fact are well below the affordable level.
The "core" service rate in a given area should be considered to be "reasonably comparable" to rates in other areas if it does not exceed the national affordability guideline.
The initial level of universal service support should be based on the difference between the actual rate which Carriers of Last Resort ("COLRs") are allowed to charge in an area and a market price estimate developed using a proxy cost study. Once competitive bidding is conducted in an area, the results from the bidding would replace the estimates generated on a cost basis. All COLRs should be required to provide the defined "core" service package in order to qualify for support. This is necessary to ensure that all customers have available to them the services the Commission has determined to be essential, and to ensure that the plan is competitively neutral.
The telecommunications requirements of schools, libraries and rural health care providers vary widely, as does the current level of technology integration in any given locale. No single technology platform or list of services will fit all situations, nor will the judgments and preferences of school administrators always come out at the same point. Accordingly, the Commission should adopt a flexible framework. To impose on telecommunications carriers any obligation to provide inside wire or other such educational infrastructure would fall completely outside the letter and spirit of the 1996 Act. The 1996 Act serves to promote competition for services to educational and health care entities because it requires the creation of a new external funding source to assist eligible entities in obtaining telecommunications services. All telecommunications service providers are aware of this fact, and have strong business motivation to seek to provide services to eligible entities. The resale constraints in [[section]]254(h)(3) clearly prohibit an eligible entity from using universal support funding to obtain services that are resold in any fashion, whether for profit or "at cost." Funds that enable eligible educational entities and libraries to obtain network services at a reduced price should be directed to the states in the form of block grants, and state agencies should be used to distribute funds.
While the existing USF could have a place within the overall policy framework the Commission adopts, it is inconsistent with the requirements of the 1996 Act. Rather than attempt to modify the current USF, the Commission should develop a new framework which is better suited to a competitive environment. However, for a transitional period, the current USF mechanism could be retained for non-price cap companies serving rural areas.
Price cap companies cannot be excluded from eligibility for high-cost support as a matter of law. Indeed, any carrier that undertakes the COLR obligation established by a state regulatory agency must be treated in the same manner, i.e., it should have the same obligations and receive the same level of support.
Every practical effort should be made to resolve differences among the proxy cost models which have been presented to the Commission in this proceeding. GTE suggests instead that, wherever possible, groups of like-minded parties should work together voluntarily to narrow the differences between these models. However, it is likely that it will not be possible to produce a single model that will achieve a consensus among all parties in this proceeding; thus the Commission will have to make a final determination, using the best information on the record, within six months of the Joint Board's recommendation.
There is no need for proxy cost models to evolve to capture the costs of every technology that could be used to provide local service. The proxy model's purpose should be to provide an initial value for the support level in each area. After that, an auction mechanism will provide a better means for adjusting the support level over time.
RESPONSES
Definitions Issues
1. Is it appropriate to assume that current rates for services included within the definition of universal service are affordable, despite variations among companies and service areas?
Yes, existing rates can be considered affordable because they have been the subject of decades of attention from federal agencies, including the FCC, and from state regulatory agencies aimed at keeping local service prices low.[2] Moreover, the high subscribership levels that currently exist throughout the nation serve as one form of confirmation of the affordability of local service.
Indeed, it is GTE's opinion that existing rates in many areas are below the affordable level. Demand studies suggest that the demand for local service is extremely inelastic with respect to changes in price from current levels. If rates for local service could be increased toward market levels, the funding needed to maintain those rates could be greatly reduced, and any competitive distortions created by universal service policy could also be minimized. It is therefore important that the market intervention practiced on local service rates and terms should be the minimum necessary to achieve universal service policy goals.[3]
Rates for local service are regulated by state commissions, and changing them is outside the scope of this proceeding. In order to be sufficient and competitively neutral, universal service mechanisms must support the full difference between these rates and market rate levels. This should be accomplished through a combination of state and Federal mechanisms.
However, in establishing a reference point as to what is "affordable," the Commission and the Joint Board should not base their finding on the rate levels currently in place, but should instead make an objective determination of how much customers should be expected to pay. Such a reference point could serve several functions in the Federal plan. First, it could be used as a guideline for the level to which rates could be allowed to rise; incentives could be established within the Federal plan to encourage states to rebalance local rates toward this level. Second, a reference point (perhaps a second, higher one) could be used to divide responsibility for funding between the Federal plan and state plans. In areas where costs exceeded this level, Federal funding would be provided to ensure that local rates did not exceed this amount.
2. To what extent should non-rate factors, such as subscribership level, telephone expenditures as a percentage of income, cost of living, or local calling area size be considered in determining the affordability and reasonable comparability of rates?
The "core" service rate in a given area should be considered to be "reasonably comparable" to rates in other areas if it does not exceed the national affordability guideline. The 1996 Act does not require that rates be exactly equal between rural and urban areas, or be geographically averaged, in order to be "reasonably comparable."
See [[section]]254(b)(3).
In determining the "affordable" rate guideline, the Commission may wish to consider variations in income by area. This could be done by setting the guideline as a percentage of median household expenditure in an area. This approach does have certain pitfalls, however. If the areas used to measure income are large (such as states), then more extreme variations within those areas will not be captured. If smaller areas are used, the guideline will vary widely across these areas. However, even within a small area, customers are not homogeneous. Some customers with modest incomes may be harmed by this approach if they happen to live in areas where the median expenditure is high. It should be remembered that the most extreme differences in income across households will be addressed by income-based mechanisms such as Lifeline and Linkup.
The Commission should not consider other factors, such as subscribership, in establishing its affordability guideline. Subscribership is affected by many exogenous factors, as well as by policy instruments other than the local rate. The Commission has established a record on many of these other factors in its subscribership proceeding. The Federal plan should not attempt to compensate for geographic differences in these other variables by adjusting the local rate. Experience has shown that the local rate should not be relied on as the sole policy tool for addressing subscribership concerns; reductions in the local rate may be "pushing on a string." For example, the District of Columbia has a relatively low subscribership level, even though it has very low local rates, particularly for lifeline service.
The Commission is not in a good position to consider the effects of differences in calling scope on affordability. Local calling scopes, the structure of local calling plans, and rates for extended area and toll services vary widely from place to place. In many areas, several different service options are available. The most logical course for the Commission to follow in designing the Federal plan is to define the national "core" service without usage. Each state would then be free to augment the definition to include some package of usage as it sees fit, and to fund any such usage through its own state plan. State commissions would be in a better position than the FCC to sort through the specifics of geography, community of interest, and so on, in their own areas.
3. When making the "affordability" determination required by Section 254(i) of the Act, what are the advantages and disadvantages of using a specific national benchmark rate for core services in a proxy model?
There is no benchmark rate "in" a proxy model, as question number 3 implies. The benchmark rate is an affordable price level selected by the FCC and the Joint Board as the maximum price that consumers should be expected to pay for the "core" universal service. A proxy model (e.g., the Benchmark Cost Model or "BCM") is a cost model that generates estimates of the cost of service in different geographic areas. The estimates from such a model could be used to develop estimates of what the market price for the "core" service might be. GTE suggests that the initial level of support should be based on the difference between the actual rate which Carriers of Last Resort ("COLRs") are allowed to charge in an area and this market price estimate. Once competitive bidding is conducted in an area, the results from the bidding would replace the estimates generated on a cost basis.
The national affordability benchmark should not affect the estimate of the market price, and therefore is independent of the proxy model itself. Further, since the total amount (state and Federal) of support needed is based on the difference between the actual rate and the market price estimate, it would also be independent of the Federal benchmark or affordability guideline. However, the Federal benchmark would affect the proportion of the needed funding that would be provided by the Federal plan.
The use of the benchmark would thus give the Commission a policy tool for controlling how much of the overall funding requirement would be contributed by the Federal plan, and how much would be left to the states. This would allow the Commission to make an objective policy choice in this regard, rather than simply base the Federal plan on the results of the current separations process. The benchmark would also allow the Federal plan to establish limits on the variation in rates that would be allowed as a matter of national policy; by doing so, a plan so structured would fulfill the Commission's obligation under the Act to ensure that rates are affordable and reasonably comparable.
4. What are the effects on competition if a carrier is denied universal service support because it is technically infeasible for that carrier to provide one or more of the core services?
GTE has recommended that the FCC and Joint Board establish a definition of the "core" universal service based upon the functionality to be provided.[4] The instant question underscores the need for adoption of a technologically neutral definition so that a carrier's choice of technology is not unduly influenced by the potential availability of universal service support monies. Further, to keep the total support amounts to a reasonable level, and to allow the market to guide the development of new services, the "core" service should contain a limited number of features that are essential for basic telecommunications, rather than including a "laundry list" list of advanced functions.[5]
There is considerable variation today across geographic areas in the service package available to local customers. It is likely, therefore, that in some areas the service currently provided by the incumbent LEC will not satisfy the definition chosen by the Commission. An appropriate transition mechanism should be adopted which would allow carriers to adjust their services to meet the definition. This might include, for example, a transition to single party service in areas where it is not available today.
However, leaving aside such transitional issues, all COLRs should be required to provide the defined "core" service package in order to qualify for support. This is necessary to ensure that all customers have available to them the services the Commission has determined to be essential, and to ensure that the plan is competitively neutral. It would not be neutral to provide the same support to two competing carriers, while allowing one carrier to provide less than the full "core" service. Each carrier will then be free to chose the technology it will use to provide the specified functionality. A carrier may also choose not to be a COLR, and to adopt technology which cannot provide all of the COLR definition.
5. A number of Commenters proposed various services to be included on the list of supported services, including access to directory assistance, emergency assistance, and advanced services. Although the delivery of these services may require a local loop, do loop costs accurately represent the actual cost of providing core services? To the extent that loop costs do not fully represent the costs associated with including a service in the definition of core services, identify and quantify other costs to be considered.
Loop costs represent a large portion of the total cost of providing core universal service. However, other costs must be included if the total cost of core service is to be determined.
Schools, Libraries, Health Care Providers
6. Should the services or functionalities eligible for discounts be specifically limited and identified, or should the discount apply to all available services?
The 1996 Act requires the Commission to define a set of "special" services for which discounts would be available. GTE suggests that this definition should be broad enough to accommodate reasonable differences in the needs of different institutions.
The telecommunications requirements of schools, libraries and rural health care providers vary widely, as does the current level of technology integration in any given locale. No single technology platform or list of services will fit all situations, nor will the judgments and preferences of school administrators always come out at the same point. Accordingly, the Commission should adopt a flexible framework. However, special services should not include "core" services, which are supported at affordable levels by a separate program. Nor should it include items which schools may need to complete their plans, but which are not telecommunications services. These items would include customer premises equipment ("CPE"), computers, software, training, and inside wire.
Rather than offer discounts for services in the form of a percentage reduction in specific service prices, GTE recommends, in response to question number 12 infra, an approach that is more flexible and efficient, and more appropriate to a competitive environment, i.e., where schools and libraries can choose among numerous telecommunications service providers.[6] Under this suggested approach, an annual and predictable fund would be established and made available equitably to eligible schools and libraries in the form of either money or credits that could be used to purchase any special telecommunications service from any telecommunications provider.7
The only restrictions that should apply are: (1) schools and libraries should be able to order a telecommunications providers commercially available telecommunications services, or the additional telecommunications services that a provider voluntarily chooses to offer;[8] and (2) the service provider is chosen by the eligible entity through a process that is subject to periodic audit by the fund administrator.
7. Does Section 254(h) contemplate that inside wiring or other internal connections to classrooms may be eligible for universal service support of telecommunications services provided to schools and libraries? If so, what is the estimated cost of the inside wiring and other internal connections?
No. The 1996 Act addresses the provision of telecommunications services. The definition of telecommunications services within the 1996 Act does not include inside wiring and "other internal connections."
The intent of the 1996 Act is to provide eligible entities with telecommunications services that are affordable. It is not concerned with end user or customer-owned equipment such as inside wire. Inside wire is an important element of the educational infrastructure, along with computers, software, local area networks ("LANs"), curriculum development, electrical power, ventilation and air conditioning. But Congress no more contemplated the furnishing of inside wire under the 1996 Act than it did air conditioning or software. To impose on telecommunications carriers any obligation to provide such educational infrastructure would fall completely outside the letter and spirit of the 1996 Act. The FCC determined years ago that inside wire fell outside the scope of telecommunications service.
More specifically, inside wiring is excluded from the scope of [[section]]254(h) for the following reasons:
1. Under [[section]]254(h)(1)(A) (concerning health care providers), the essential obligation of the telecommunications carrier is to provide to certain health care providers as requested "telecommunications services which are necessary for the provision of health care services in a State ... at rates that are reasonably comparable to rates charged for similar services in urban areas in that State."
(i) The furnishing of inside wiring is not a "telecommunications service" as defined ("the offering of telecommunications for a fee directly to the public, or to such classes of users as to be effectively available directly to the public, regardless of the facilities used") in [[section]]153(41) because inside wiring does not furnish "telecommunications" as defined in [[section]]153(38) inasmuch as inside wire does not constitute "the transmission, between or among points specified by the user, of information of the user's choosing, without change in the form or content of the information as sent and received", in fact does not constitute transmission of any kind.
(ii) To the extent a telecommunications carrier furnishes inside wire, it is not subject to FCC regulation inasmuch as, under [[section]]153(39), a "telecommunications carrier shall be treated as a common carrier under this Act only to the extent that it is engaged in providing telecommunications services...." and, as shown supra, inside wire does not constitute "telecommunications service."
(iii) Inside wire furnished by telecommunications carriers cannot be "necessary for the provision of health care services in a State" inasmuch as the FCC found that the public interest requires its removal from regulated offerings because: "Like CPE services, inside wiring installation and maintenance are severable from underlying common carrier transmission services, and are susceptible to being provided by a wide variety of competing firms."[9]
2. Under [[section]]254(h)(1)(B) (concerning educational providers and libraries), a telecommunications carrier serving a geographic area upon a bona fide request shall provide to specified parties at a discount "any of its services that are within the definition of universal service under [[[section]]154](c)(3)."
(i) Inside wire cannot be within the definition of universal service under
[[section]]154(c)(3) because, as discussed supra, it is not a "telecommunications service."
(ii) Inside wire cannot be deemed part of the existing services ("its services") of GTE or any other Incumbent Local Exchange Carrier ("ILEC") inasmuch as furnishing inside wire was deregulated by FCC decision.
8. To what extent should the provisions of Sections 706 and 708 be considered by the Joint Board and be relied upon to provide advanced services to schools, libraries and health care providers?
The Joint Board should place heavy reliance on the mandate of [[section]]706 of the 1996 Act that deployment of an advanced telecommunications capability should be encouraged "by utilizing, in a manner consistent with the public interest, convenience, and necessity, price cap regulation, regulatory forbearance, measures that promote competition in the local telecommunications market, or other regulating methods that remove barriers to infrastructure investment."
9. How can universal service support for schools, libraries, and health care providers be structured to promote competition?
The 1996 Act itself serves to promote competition because it requires the creation of a new external funding source that will assist eligible entities in obtaining telecommunications services. Thus, entities that could not previously afford desired services now have the opportunity to obtain federal financial assistance. All telecommunications service providers are aware of this fact, and have strong business motivation to seek to provide services to eligible entities.
Further, GTE's approach discussed in the response to question number 12 will promote competition because it is based on competitive neutrality. Schools, libraries and rural health care providers can choose any available special services from the telecommunications provider that best meets their needs. The price of the service would be the best price the institution can obtain from its chosen provider, independent of the discount.10 The institution would then receive the discount separately in the form of a credit or direct funding which would assist the institution in purchasing the service. This proposal is inherently neutral among carriers because it does not operate directly on the carrier's price, but instead supports the institution's purchase directly. This eliminates any possibility that the required discounts, as implemented by a particular carrier, could create either an advantage or disadvantage for that carrier, relative to other providers. It also eliminates the regulatory burden of establishing, reviewing, and tariffing different rates for this purpose.
10. Should the resale prohibition in Section 254(h)(3) be construed to prohibit only the resale of services to the public for profit, and should it be construed so as to permit end user cost based fees for services? Would construction in this manner facilitate community networks and/or aggregation of purchasing power?
The resale constraints in [[section]]254(h)(3) clearly prohibit an eligible entity from using universal support funding to obtain services that are resold in any fashion, whether for profit or "at cost."
GTE recommends that community networks be encouraged by providing federal funding for the non-profit portion of such networks by basing support upon the relative usage by the educational entity and other users, with provision for audit of reported percent of use by the fund administrator.[11]
11. If the answer to the first question in number 10 is "yes," should the discounts be available only for the traffic or network usage attributable to the educational entities that qualify for the Section 254 discounts?
See response to question number 10.
12. Should discounts be directed to the states in the form of block grants?
Yes, funds that enable eligible educational entities and libraries to obtain network services at a reduced price should be directed to the states in the form of block grants.[12] GTE submits that adoption of the following plan could satisfy the requirements of the 1996 Act, be administratively feasible, and enable the entire process to be managed in an efficient and consistent manner.
The first step in the administrative process GTE proposes would be to identify the total nationwide amount of funding needed for the network service component of the level of functionality chosen by public policy makers to be provided to eligible
educational entities and libraries.[13] Once this amount has been identified, each state would be allocated a "block grant" amount of funds.[14]
Next, each state would establish an administrator with responsibility to: (i) serve as a central focal point for information about available funds; (ii) review requests for support for network services to ensure the requesting entity is eligible under the 1996 Act, (iii) review each request to ensure it is "bona fide";[15] (iv) authorize dispensing of universal service funds either to the eligible entity or directly to the chosen network service provider;[16] and (v) perform audits, as found necessary, to ensure that funds were being efficiently used and/or to resolve complaints from service providers over the choice of supplier by an eligible entity. Upon approval by the administrator of an amount of funding, the eligible entity would be authorized to present a written request to the network service provider.[17] The administrator should also ensure that the total amount authorized in all of the plans within the state is no greater than the total amount assigned to that state.[18]
13. Should discounts for schools, libraries, and health care providers take the form of direct billing credits for telecommunications services provided to eligible institutions?
Under GTE's recommended approach described in the response to question number 12, the administrator could provide funds either directly to schools or libraries entities, or to network services providers themselves, whichever option proved to be more efficient.
However, because [[section]]254(h)(1)(A) establishes a different requirement for rural health care providers ("reasonably comparable to rates charges for similar services in urban areas in that State"), those entities should receive vouchers or credits if the price they pay is not reasonably comparable to the urban price.
14. If the discounts are disbursed as block grants to states or as direct billing credits for schools, libraries, and health care providers, what, if any, measures should be implemented to assure that the funds allocated for discounts are used for their intended purposes?
As described in the response to question number 12, as part of the process of obtaining funds, eligible entities should be required to attest or to certify that the support they receive is used only for the intended non-profit educational or health care purposes. Further, a condition of receipt of such support must be the right of the administrator to perform audits.
15. What is the least administratively burdensome requirement that could be used to ensure that requests for supported telecommunications services are bona fide requests within the intent of section 254(h)?
As described in the response to question 12 supra, each eligible entity desiring funding should be required to provide support materials that allow the central administrator to determine that the entity will effectively use the universal service support. The support materials should include:
(1) An attestation that the entity is eligible under [[section]]254(h).
(2) A telecommunications plan that describes how all network and non-network components fit together to create an effective program.
(3) A description of the process used to select the network services, the identity of the selected vendor, the services to be provided, the price to be paid for each service, and the amount of desired support funding.
(4) A budget showing that all of the necessary components other than telecommunications service (e.g., inside wiring, CPE, computers, educational application software and training in its use) are already present, or that commitments for their funding have been obtained from sources other than the universal service fund.
16. What should be the base service prices to which discounts for schools and libraries are applied: (a) total service long-run incremental cost; (b) short-run incremental costs; (c) best commercially-available rate; (d) tariffed rate; (e) rate established through a competitively-bid contract in which schools and libraries participate; (f) lowest of some group of the above; or (g) some other benchmark? How could the best commercially-available rate be ascertained, in light of the fact that many such rates may be established pursuant to confidential contractual arrangements?
The "base service prices" supported by the federal fund should be the price of the service quoted to the school by the provider it selects. That price could be the tariffed price or a price developed in response to an invitation to bid. This approach allows each eligible entity the maximum flexibility in choosing a service provider, and divorces the fund administrator from the price setting process. GTE's proposal effectively renders question 16 moot, since the FCC and state regulators would not have to establish or administer any discounted rates.
Options (a) and (b) are inappropriate in any case. The 1996 Act directs the Commission to determine what discount institutions should receive, relative to the price they would otherwise pay. This discount is provided to meet public policy goals, and has nothing to do with the structure of the provider's underlying cost. In determining the level of a Rhodes scholarship, does it matter what proportion of Oxford University's cost is fixed or variable? In any event, the discount should not depend on the identity of the supplier, but different suppliers will clearly have different cost structures.
Further, any process that relies on setting discounted rates for each carrier will inherently not be neutral, since different carriers are subject to different regulatory processes. If the rate-setting process has different effects on these carriers, then the competition among them for the institutions' business could be distorted. GTE's proposal obviates this concern.
17. How should discounts be applied, if at all, for schools and libraries and rural health care providers that are currently receiving special rates?
No further discount should be applicable to existing special rates. However, any eligible entity currently receiving a special rate that believes a better price might be available would be allowed to replace that price using the process described in the response to question number 12.
Moreover, any existing special prices mandated by state regulatory agencies that are applicable only to ILECs and that fail to offset such special price reductions with explicit, sufficient and predictable funding obtained in a competitively neutral manner from all telecommunications service providers, have been rendered null and void by the 1996 Act.[19]
18. What states have established discount programs for telecommunications services provided to schools, libraries, and health care providers? Describe the programs, including the measurable outcomes and the associated costs.
GTE provides local service in 28 states, but rather than attempt to document every program in each state, GTE herein discusses the single state in which GTE is the largest ILEC -- Hawaii.
At present, there is no regulatory or legislative requirement in Hawaii for discount programs for telecommunications services for schools, libraries and health care providers. In an approach similar to the one recommended herein, GTE initiated a program in Hawaii that provides free education credits of up to $2,000 each, as well as consulting services, to public K-12 schools and public libraries. Rather than taking a 'cookie cutter' approach, this plan provides technology to schools on educators' terms, allowing them to select the telecommunications services that best meet their needs.
19. Should an additional discount be given to schools and libraries located in rural, insular, high-cost and economically disadvantaged areas? What percentage of telecommunications services (e.g., Internet services) used by schools and libraries in such areas are or require toll calls?
GTE's recommended approach described in response to question number 12 supra would allow the administrator to provide a greater proportion of funds for schools and libraries in rural, insular, high-cost and economically disadvantaged areas. This approach would avoid a complex process of discounts on top of discounts.
GTE has no knowledge of the proportion of information services used by rural entities that require a toll call. However, question number 19 incorrectly equates
"Internet services" to the "Telecommunications Services" that are eligible for support under the 1996 Act. Information services, such as those available from an Internet access provider, do not fall within the definition of the telecommunications services to be supported under '254. Compare [[section]]153(a)(38) ("Telecommunications") and (41) ("Telecommunications Service") with [[section]]153(a)(20) ("Information Service"). In fact, the title of '254(h) that establishes special funding for educational entities is
"Telecommunications Services for Certain Providers" (emphasis added).
20. Should the Commission use some existing model to determine the degree to which a school is disadvantaged (e.g., Title I or the national school lunch program)? Which one? What, if any, modifications should the Commission make to that model?
No response.
21. Should the Commission use a sliding scale approach (i.e., along a continuum of need) or a step approach (e.g., the Lifeline assistance program or the national school lunch program) to allocate any additional consideration given to schools and libraries located in rural, insular, high-cost, and economically disadvantaged areas?
No response.
22. Should separate funding mechanisms be established for schools and libraries and for rural health care providers?
No, funding for entities eligible for support under [[section]]254(h) should be obtained in the same manner as funding for "core" universal service. The Federal fund administrator should add together all funding requirements and collect the necessary funds through a single mechanism. This should include funds for "core" universal service, low income individuals, educational and rural health entities, and any other universal service support the Commission adopts. The only competitively neutral collection mechanism available is a surcharge applicable to end user retail revenues. Any other collection mechanism would introduce distortions into the market.
As described in the response to question number 12, separate distribution processes are appropriate for educational entities and rural health care providers because of the different treatment of their support under the 1996 Act.
23. Are the cost estimates contained in the McKinsey Report and NII KickStart Initiative an accurate funding estimate for the discount provisions for schools and libraries, assuming that tariffed rates are used as the base prices?
GTE has no direct knowledge of the accuracy of the cost estimates, but without evidence to the contrary, the estimates must be assumed to be reasonable. Further, the study provides a template for estimating the relative amount of funding needed for different levels of services.
24. Are there other cost estimates available that can serve as the basis for establishing a funding estimate for the discount provisions applicable to schools and libraries and to rural health care providers?
No response.
25. Are there any specific cost estimates that address the discount funding estimates for eligible private schools?
No response.
High Cost Fund
General Questions
26. If the existing high-cost support mechanism remains in place (on either a permanent or temporary basis), what modifications, if any, are required to comply with the Telecommunications Act of 1996?
The 1996 Act requires the Commission to adopt a new Federal universal service plan, taking into consideration the recommendation of the Joint Board. The current USF mechanism cannot serve this purpose, because it is inconsistent with the requirements of the 1996 Act. Rather than attempt to modify the current USF, the Commission should develop a new framework which is better suited to a competitive environment. However, the existing USF could have a place within the overall policy framework the Commission adopts.
Specifically, GTE suggests that all price cap companies, and all companies serving non-rural areas, should transition to the new Federal universal service plan. The current USF mechanism would be retained for non-price cap companies serving rural areas. These companies would have a one-time option to switch to the new Federal plan at any time; once a company had exercised this option, it would not be permitted to return to the USF. At the end of some period (say five years) the Commission could review the USF to determine whether it should be continued, or whether companies then remaining on USF should be brought into the new Federal plan.
In principle, it makes sense for each area to be treated the same, regardless of the company that serves it. As a practical matter, however, the circumstances of the rural telephone companies differ from those of the large ones, and vary significantly as well from one small company to another. The task of developing a new Federal plan for adoption in May, 1997, will be made much more difficult if the Commission attempts to make that plan fit all of the incumbent companies, and there will be a risk as well of unreasonable shocks to certain companies. The most pressing need for a new Federal plan to meet the requirements of the 1996 Act is in the areas now served by the large companies. These are the areas where competition is developing most rapidly, and where the incumbent LECs have been obliged to rely most heavily on implicit sources of support from their own services. The Commission should focus its efforts in the coming months on developing a new Federal plan for these areas that is targeted to small geographic areas, that is sufficient, that is competitively neutral, and that will replace the current implicit support flows, so that distortions in the larger ILECs' rates can be corrected.
While the current USF is far from perfect, it is more targeted in rural company areas simply because these study areas are small. It is also more sufficient in rural company areas, so that these companies' reliance on implicit support is less. It is therefore reasonable, and consistent with the requirements of the 1996 Act, for the USF to be continued in its present form for small companies within the context of a larger policy that includes a new plan for price cap and nonrural companies.
27. If the high-cost support system is kept in place for rural areas, how should it be modified to target the fund better and consistently with the Telecommunications Act of 1996?
The Commission should not expend effort attempting to "fix" the current USF. It is better to develop a more competitively neutral plan, and then, once that plan has been proven as applied to larger companies, consider whether small companies should adopt a version of it as well.
28. What are the potential advantages and disadvantages of basing the payments to competitive carriers on the book costs of the incumbent local exchange carrier operating in the same service area?
The objective of a cost-based universal service plan should be to use the cost measure as an estimate of what the market price of the "core" service would be in a competitive market.[20] This is necessary to ensure that the price the COLR sees -- the sum of the rate the customer pays and the support payment -- is the right price signal for entry and investment decisions by prospective COLRs. In order to be competitively neutral, the support level should be the same for any carrier in a given area that undertakes the same COLR obligation.
This price estimate should be based on ILEC costs, since ILECs are the current COLRs, and because they provide the bulk of the supply capacity in the industry today. The question then arises as to whether embedded ILEC costs should be used, or a forward-looking estimate based on ILEC technology and network practices.
In general, the market price in a given market will be based on the average cost of the suppliers in the industry. In this sense, the average cost of capacity in the industry is relevant; this would argue for using ILEC average (i.e., embedded) costs to form the universal service cost estimate. If technology makes possible a lower cost on a forward-looking basis, this fact, in and of itself, will not change the market price. The price will change only as firms actually install capacity to produce at the lower cost. Indeed, the first firm to adopt the technology will generally not affect the price greatly, if it supplies only a part of the demand; instead, the price will continue to be based on the average cost, and the low-cost firm will earn rents, for a time, as a reward for its innovation. Then, as more firms adopt the new technology, the price will be driven down. At the end of this adoption process, there will be enough capacity to serve the entire market at the lower cost. By this time, the price will have been driven to the new (lower) average level, the transitory rents will have been competed away, and all of the benefits of the new technology will have been passed on to consumers. The value of any older plant will have been reduced. But all of this would have happened only as firms could actually supply at the lower cost. A strong argument can therefore be made that the estimate of market price at the outset of the new universal service plan should be based on the cost of the (ILEC) capacity that is actually supplying the market at that time. This would provide the correct incentives for new firms to enter if they can supply at lower cost.[21]
Several objections have been raised to basing support on ILEC embedded cost levels. The first is that if such estimates are adjusted over time to reflect ILEC book costs, an incentive will be created to inflate those costs, since this would create additional support. GTE agrees. However, no such incentive would be created if support levels are established at the outset of the plan, and not adjusted to reflect changes in book cost from that time forward. In GTE's proposal, the need to adjust the cost estimate over time is avoided, since the bidding mechanism would be the means for correcting the level of support over time.
The second objection is that current ILEC book costs have already been distorted because of past regulation. Rate of return regulation may have encouraged ILECs to overinvest or may have imposed mandates that entailed a higher level of investment than a nonregulated firm would have chosen. Further, it is broadly accepted that depreciation allowances by state and federal regulators have been inadequate, so that costs incurred in past periods were not fully recognized at the time; this would cause current ILEC revenue requirements to include recovery of some of these past and underdepreciated costs. However, these concerns do not justify disregarding real ILEC costs , as some would suggest.
GTE suggests that the effects of any overvaluation of ILEC plant because of insufficient depreciation is an issue that should be addressed separately from the main Federal plan to support COLR obligations on a going-forward basis. This valuation problem is inherently asymmetric, since it affects only incumbent LECs, and represents unrecovered costs of past COLR obligations, not of future ones.
Once this problem has been addressed, then it may be most reasonable for the Federal plan to be based on a proxy model that generates a forward-looking estimate of investment, reflecting (i) current ILEC network topology, and (ii) practices currently used by ILECs to place new equipment. The patterns of current ILEC operations should not be disregarded, as some parties suggest; they contain information about factors affecting ILEC costs which may not be fully captured by a model that attempts to design a network from scratch. The estimate should also include directly attributable expenses, as well as contributions toward shared and common costs; current ILEC levels for these expenses are the best available estimators.
The initial levels of support should be developed by comparing the rate COLRs are required to charge with this cost measure. Once new entrants are willing to undertake the COLR obligation in a given area, then the auction process -- described infra -- should supersede this cost-based approach.
29. Should price cap companies be eligible for high-cost support, and if not, how would the exclusion of price cap carriers be consistent with the provisions of section 214(e) of the Communications Act? In the alternative, should high-cost support be structured differently for price cap carriers than for other carriers?
Any carrier that undertakes the COLR obligation established by the state regulatory agency for a given area should be treated in the same manner. It should have the same obligations, and should receive the same level of support. This is necessary for the plan to be competitively neutral, as the 1996 Act requires; for it to be sufficient; and to reduce the current level of implicit support in ILEC rates.
There is no basis for excluding some companies, or some areas, from universal service support, simply because they may be subject to a different form of regulation. As long as a market intervention is imposed on the rates and terms for local service, then a mechanism must be maintained to compensate carriers subject to that intervention in a competitively neutral way.
Price cap companies cannot be excluded from eligibility for high-cost support as a matter of law, for:
(1) Price cap ILECs are "Telecommunications Carriers" as defined by [[section]]F153(39).
(2) Price cap ILECs come squarely within the criteria for eligibility spelled out by [[section]]214(e)(1), i.e., they: (i) offer the services that are supported by Federal universal service support mechanisms under [[section]]254(c), using either their own or a combination of their own and resold services; and (ii) advertise the availability of such services using general media.
(3) As COLRs that currently offer services subsuming the "core" universal service, price cap ILECs have been de facto recognized by state regulatory agencies as eligible telecommunications carriers.
(4) Any universal service plan that precluded the price cap ILECs, representing service areas that embrace more than ninety percent of the people of the United States, would by definition fail to comply with the statutory mandate because, among other reasons, it would not be sufficient. See [[section]]254(e).
Reducing the amount of support provided to price cap companies as compared to that available to a similarly situated non-price cap company would be arbitrary and capricious because:
(1) [[section]]254(b)(5) requires "specific, predictable and sufficient" mechanisms to preserve and advance universal service.
(2) The underlying intent of the 1996 Act is to foster genuine competition among all telecommunications service providers.[22].
(3) Commission action that would arbitrarily reduce the amount of support available to a price cap ILEC would result in compelling that ILEC to contribute to universal support while denying that ILEC a reasonable opportunity to receive support -- as contemplated by Congress -- in respect of additional costs imposed by virtue of its COLR status..[23]
30. If price cap companies are not eligible for support or receive high-cost support on a different basis than other carriers, what should be the definition of a "price cap" company? Would companies participating in a state, but not a federal, price cap plan be deemed price cap companies? Should there be a distinction between carriers operating under price caps and carriers that have agreed, for a specified period of time, to limit increases in some or all rates as part of a "social contract" regulatory approach?
There is no reason why the new Federal plan should apply differently based on the form of regulation applied to that carrier. All carriers subject to Federal price caps, and all carriers serving nonrural areas (whether price cap or not) should adopt the new Federal plan. Other incumbent LECs should have the option of remaining under the current USF plan. The purpose of this distinction, however, would not be to account for any inherent difference between price cap and rate of return regulation as they relate to universal service. Rather, the intent of GTE's proposal is simply to use these categories to distinguish companies to which the new plan can readily be applied from the many small companies for which the old plan may be more appropriate, at least for a time.
31. If a bifurcated plan that would allow the use of book costs (instead of proxy costs) were used for rural companies, how should rural companies be defined?
The 1996 Act's definition of "Rural Telephone Company", [[section]]153(37), must be used in interpreting and applying the 1996 Act. This definition does not preclude rural study areas served by large holding companies. However, GTE suggests that areas served by price cap companies should be included in the new Federal plan, rather than under the current USF, regardless of whether a study area is "rural."
32. If such a bifurcated approach is used, should those carriers initially allowed to use book costs eventually transition to a proxy system or a system of competitive bidding? If these companies are transitioned from book costs, how long should the transition be? What would be the basis for high-cost assistance to competitors under a bifurcated approach, both initially and during a transition period?
Any company that continues on the USF plan under a bifurcated approach should have a one-time option to switch to the new Federal plan. After some period, say five years, the Commission should review the status of the USF to determine if it should be continued, or if USF should be ended, and LECs still remaining on USF transitioned to the new plan. If the Commission determines after five years that USF should be ended, it can decide at that time what additional transition mechanism might be reasonable for companies moving from USF to the new plan.
If another COLR is authorized by the state commission to operate in the serving area of a rural, non-price cap company under this bifurcated arrangement, then competitive neutrality would require that the new COLR should receive the same level of support, on a per-customer basis, that the incumbent LEC receives from USF.
33. If a proxy model is used, should carriers serving areas with subscription below a certain level continue to receive assistance at levels currently produced under the HCF and DEM weighting subsidies?
No. The level of universal service support in an area should not depend on the level of subscribership there. As the record in the Commission's CC Docket No. 96-115 clearly shows, subscribership levels are affected by many more factors than price alone.24 The COLR's obligation should be to provide service at the price and terms specified by the Commission.
A proxy model should not be used to distribute funds under the current USF structure. Under GTE's proposal, a rural, non-price cap company would have the option of remaining under USF if it believed that its ability to provide universal service would be harmed if it adopted the new plan. This should be the case, regardless of the level of subscription in the serving area.
Proxy Models
34. What, if any, programs (in addition to those aimed at high-cost areas) are needed to ensure that insular areas have affordable telecommunications service?
Funds for "core" universal services for insular areas should be provided through the same universal service mechanism used for "core" services in mainland areas. Should the Commission determine that additional telecommunications services should be supported only for insular areas, funding requirements should be determined separately, if necessary, but the same mechanisms as used for "core" mainland service should be used to obtain and distribute the additional funds.
35. US West has stated that an industry task force "could develop a final model process utilizing consensus model assumptions and input data," US West comments at 10. Comment on US West's statement, discussing potential legal issues and practical considerations in light of the requirement under the 1996 Act that the Commission take final action in this proceeding within six months of the Joint's Board's recommended decision.
GTE agrees that every practical effort should be made to resolve differences among the models which have been presented to the Commission in this proceeding. However, GTE does not believe that an industry task force brought together by the Commission, and representing widely diverging interests, would be a useful method for resolving differences. GTE's recent experience in state workshops suggests that such a task force would expend more time on rhetoric than on resolution of substantive issues.
GTE suggests instead that, wherever possible, groups of like-minded parties should work together voluntarily to narrow their differences. GTE is working to promote such cooperation among parties who have proposed proxy models. If successful, these efforts will narrow the range of choices among competing models. However, it is likely that it will not be possible to produce a single model that will achieve a consensus among all parties in this proceeding; thus the Commission will have to make a final determination, using the best information on the record. GTE urges the Commission to make such a determination as part of the final action it must take within six months of the Joint Board's recommendation.
36. What proposals, if any, have been considered by interested parties to harmonize the differences among the various proxy cost proposals? What results have been achieved?
GTE has worked actively with companies who have sponsored proxy models to determine how the differences between them could be minimized, or whether it might be possible to produce a single model reflecting the best features of each of the existing models.
37. How does a proxy model determine costs for providing only the defined universal service core services?
Each of the proxy models estimates the cost of those network components that would be required to provide the defined "core" service. They exclude from consideration services which are not included in the definition. However, the models do attempt to take into account economies of scope made possible by the provision of some other services, such as local business lines. Further, in GTE's experience, none of the currently available proxy models corresponds exactly to any well-defined economic cost concept. For example, none of the models produces a reasonable estimate of average-incremental (or "TSLRIC") cost.
38. How should a proxy model evolve to account for changes in the definition of core services or in the technical capabilities of various types of facilities?
The proxy cost model should not evolve. Its purpose is to provide an initial value for the support level in each area. After that, the auction mechanism will provide a better means for adjusting the support level over time.
Attempting to adapt the cost models over time would be a difficult process. None of the currently proposed models is an optimizing model; the models do not select the best technology for each situation on an objective basis. Rather, the models are designed to implement a given set of engineering practices or rules of thumb. In general, these practices are those now being used by the ILECs for the placement of new facilities. As GTE has explained in its previous comments, attempting to optimize over different technologies would make these models even more complex than they already are, and, by calling for a greater degree of extrapolation, would reduce their ability to estimate costs reliably.
Two years of effort to model the current LEC costs has yet to yield a single, agreed-upon modeling process. While GTE recommends that the Commission should select a model to provide the necessary starting point, the plan should be designed to avoid the need for revising the model over time. A plan that required updating the model would involve the Commission and the parties in an ongoing modeling process for the indefinite future, with a constant expenditure of resources and unnecessary contention among the parties. This can be avoided by adopting a plan which relies on a market-based approach, such as the auction process proposed by GTE, to update the support level over time.
39. Should a proxy model account for the cost of access to advanced telecommunications and information services, as referenced in section 254(b) of the Act? If so, how should this occur?
Section 254(b)(2) establishes the principle of "access to" advanced services, but does not include the usage of such services within the definition of "core" universal service. Stated another way, a person can use "core" universal service to access advanced services, but separate charges for such services are not required to be supported.
The cost proxy models should not address access to more advanced telecommunications and information services. By their nature, the proxy models are intended to be simplified replications of the process by which costs are developed for various core services. By adding more and more complex decision nodes to the modeling process, the algorithms get exponentially more complicated, the development and refinement process becomes more difficult, and the computational system requirements increase exponentially as well. The current proxy models are reasonably effective because they are only required to address relatively simple questions.
40. If a proxy model is used, what, if any, measures are necessary to assure that urban rates and rates in rural, insular, and high-cost areas are reasonably comparable, as required in Section 254(b)(3) of the 1996 Act.
Use of a proxy cost model has no direct relationship to the comparability of prices. The proxy cost model simply produces a cost output that can be compared to an affordable price level chosen by the FCC and the Joint Board to determine an amount of support to be made available in high cost areas. As long as the core service price is held to an affordable level on a nationwide basis, the price between rural and urban areas should be considered reasonably comparable.
41. How should support be calculated for those areas (e.g., insular areas and Alaska) that are not included under the proxy model?
To the extent that such areas are not served by Rural Telephone Companies, as discussed in the response to question number 26 supra, the selected proxy cost model should be modified to accommodate any relevant geographic area.
42. Will support calculated using a proxy model provide sufficient incentive to support infrastructure development and maintain quality service?
Yes, but only if the model is designed to provide a reasonable estimate of the costs that would determine a market price level. This will only be the case if : (i) the investment and expense estimates are realistic; (ii) the model algorithm reflects contemporary network design; (iii) the model reflects the fact that networks grow over time in response to demand growth; (iv) the cost of capital and depreciation reflect a competitive environment; and (v) the model captures all of the costs that a market price would recover, including a contribution toward shared and common costs of the firm. Models such as the Hatfield Model that use unreasonably low investment and expense estimates and pretend that perfect networks can materialize overnight will produce such low support amounts that there will be no incentive for network investment.[25]
43. Should there be recourse for companies whose book costs are substantially above the costs projected for them under a proxy model? If so, under what conditions (for example, at what cost levels above the proxy amount) should carriers be granted a waiver allowing alternative treatment? What standards should be used when considering such requests?
See response to question number 26, supra. Use of a bifurcated approach that would allow rural companies to remain under the current structure for a transitional period would serve as a safety net to reduce the possibility that a company would experience a devastating change in support level.
See also the response to question number 42, supra. If a model is properly designed and allows each company to use expense, investment and other data representative of its circumstances, the errors in the proxy estimates can be minimized. Nonetheless, it is reasonable to expect that there will be significant errors in the cost estimates.
The auction process proposed by GTE would provide a mechanism for adjusting the support level to correct errors in the estimates. [26] This would certainly be the case in areas where the estimate is too high, since this would tend to attract bidders to those areas. The auction structure proposed by GTE would also allow the support level to rise, within limits, to correct estimates which are too low.[27] However, there may be cases where the cost estimate is too low, and the auction process is unable to correct the problem because of a lack of bidders in the area. For this reason, GTE supports the inclusion in the plan of a "safety valve" mechanism through which a LEC could demonstrate that the estimate for a given area is too low, and seek an increase.
44. How can a proxy model be modified to accommodate technological neutrality?
There is no need to modify a proxy model to accommodate every conceivable technology. A model that accommodates the prevalent technologies used by ILECs will provide a reasonable estimate of the cost of providing "core" service today. There is no need to include all new technologies in a proxy model in order to ensure that the plan is neutral toward those technologies. Indeed, basing support on the current technology will provide a correct price signal to any firm considering the use of a new technology that might be cheaper. That firm will have the same incentive to innovate that a competitive firm would have.
If there is a concern here, it is not that the plan should be technology-neutral, or that the adoption of new technology should be encouraged. These objectives can be achieved using a model based on current technology. The concern might be that, over time, the support level might be too high. In a competitive market, the benefits of lower costs will be passed on to consumers as the capacity in the industry converts over time to the new technology. That is likely to occur in this case as well, since competition among providers will probably lead them to pass along their cost savings in the form of price reductions to consumers. However, the support level which produced the original prices will remain unchanged unless the model changes. For reasons described supra, it is undesirable for the plan to rely on changing the proxy model over time.
The auction process proposed by GTE will provide a method for reflecting in the support amount changes in technology over time. When an area is first bid, and each time it is rebid thereafter, the bids will reflect the carriers' best estimates of the cost of service they expect to face during the commitment period. This will happen automatically, without the need for the Commission to estimate any costs.
45. Is it appropriate for a proxy model adopted by the Commission in this proceeding to be subject to proprietary restrictions, or must such a model be a public document?
To be reasonably accurate, the models must undoubtedly contain some proprietary information, such as prices for equipment that reflect volume discounts. Further, in competitive markets, there is no justification for requiring routine disclosure of competitively sensitive proprietary information. Such proprietary information should not be made public, but should only be available for scrutiny by regulators, and to competitors under a protective agreement.
The algorithms used by the models themselves should be open to public scrutiny. At the minimum, there should be a full set of documentation which allows any potential user to view each key calculation made by the model.
46. Should a proxy model be adopted if it is based on proprietary data that may not be available for public review?
See response to question number 45, supra.
47. If it is determined that proprietary data should not be employed in the proxy model, are there adequate data publicly available on current book costs to develop a proxy model? If so, identify the source(s) of such data.
No response.
48. Should the materiality and potential importance of proprietary information be considered in evaluating the various models?
No, the degree of reliance upon proprietary information should not impact the acceptance of a model's output. Proprietary information represents the real cost that firms pay for equipment, and those real costs must be used in any model adopted by the Commission. Further, as discussed in the response to question number 45 supra, since proprietary information can be reviewed for validity, its use is not unreasonable.
Competitive Bidding
49. How would high-cost payments be determined under a system of competitive bidding in areas with no competition?
GTE has worked with Professor Paul Milgrom to revise and extend its proposal on competitive bidding for universal service. Professor Milgrom describes his recommendations in Attachment 1, "Statement of Paul R. Milgrom."
GTE proposes that the level of universal service support provided to the incumbent LEC to support its COLR obligation should be based, at the outset, on a comparison of the rate the COLR is allowed to charge and the estimate of the market rate derived from a proxy cost model.[28] A procedure would then be established which would allow other firms who wish to become COLRs in a given area to submit a Notice of Intent to bid to the state commission. This would trigger an auction process for that area; GTE proposes that these be held at regular intervals, perhaps twice a year. The process, carried out on a pre-announced schedule, under which areas would be noticed and auctioned is referred to as a "bidding cycle"
Thus, in areas where no party has yet prompted an auction, the incumbent's support would be determined on a cost basis. The plan provides a flexible mechanism which would introduce auctions in each area as circumstances there permit. The auction process would also be designed to determine endogenously the number of COLRs that should be supported in each area.
Professor Milgrom explains why, given the structure of the auction he proposes, two qualified bidders will be sufficient to ensure a successful auction. He further specifies rules which would govern circumstances in which that requirement is not met - - that is, where an auction is held but only one qualified bid is received. This essentially involves canceling the auction, and returning to the previous support level. Given this design, there would never be a case in which a support level was determined by an auction with an insufficient number of bidders.
50. How should a bidding system be structured in order to provide incentives for carriers to compete to submit the low bid for universal service support?
The form of the auction proposed is a single round, sealed-bid auction. Professor Milgrom explains why this format would make collusion among the bidders difficult to sustain. Each bidder would know that it could be excluded if any of the other parties defects, and bids aggressively. In a single round format, no other party would have the opportunity to match the defector's bid, or to punish the defector as a means of enforcing the collusion. Given these circumstances, the only way for a bidder to ensure against being excluded is to bid aggressively.
More generally, the structure of the plan must establish a clear framework which defines the purpose of the plan. Such a clear structure is necessary, regardless of whether the level of support is determined on a cost basis, or through a competitive bidding process. Central to this framework is the definition of the COLR obligation. Attachment 2, which excerpts from a paper by Dennis Weller, explains why a COLR obligation is an efficient means for distributing universal service funding. This is true because the customers in an area are heterogeneous, so that not all of them would be served voluntarily at an averaged level of support. Lacking the perfect information that would be required to optimize support for each customer individually, the Commission must rely on a n averaged level of support for customers in each area. To assure that all customers would be served at that support level, an obligation to serve must be imposed on the carrier as a condition for accepting support. This basic premise is incorporated in the Act in the requirement that a carrier must be an "eligible telecommunications carrier" ("Eltel") in order to be eligible for universal service support. The question facing the Commission, then, is what specific requirements should be adopted in order to implement this framework.
Failure to associate a COLR obligation with the receipt of funding would cause the plan to fail to meet the statutory requirements of the Act. The plan would not be competitively neutral if one carrier (the incumbent) is required to perform the COLR function, while another carrier can receive the same funding without performing that function. Further, such a plan will never be sufficient, because the COLR would never be able to sustain its obligation to serve in the face of entry by other carriers who could selectively serve only the customers they wished, and yet could receive the same funding.[29]
The universal service plan clearly must define the COLR obligation in more specific detail than is provided for in the Act. The Act specifies that an Eltel must hold itself out to serve all of the customers in an area, and advertise its rates. Yet without specification as to the terms and conditions of this obligation, it is without meaning. For example, carrier A could announce a basic service price, at which it will serve any carrier, of $200 per month. The bill does not specify the price at which the Eltel is obligated to serve; yet clearly the maintenance of an affordable price is crucial to meeting the objectives of the Act. The carrier could further offer service to its "preferred" customers at $15 per month; the Act does not specify that the Eltel must charge everyone the same price. If it advertised these prices, the carrier A would technically meet the obligations specified for Eltels in the Act; yet in fact it would be able to serve selectively, and would receive support for serving customers it would have chosen to serve anyway. If carriers could receive support on this basis, the objectives of the Act would be undermined, and the Commission would be unable to create an effective plan.
There is nothing in the Act that precludes state commissions, which are charged with certifying Eltels, and which have traditionally certified and regulated local carriers, from establishing specific requirements for the receipt of universal service funds. [30] "Eligibility" does not guarantee that a carrier will actually receive funds. Here, as with other aspects of the universal service plans, commissions will provide the specifics to implement the general framework set forth in the Act. State commissions also retain the ability to regulate local service, which they would be precluded from doing if the Act prevented them from establishing any requirements for COLRs. Further, the Commission can set guidelines for the states, in its Federal plan, as to how states will structure their COLR requirements, as a condition for the provision of Federal funding to carriers in each state. COLR requirements might include the ceiling on the rate the COLR can charge, terms and conditions of service, any quality standards, and limits on the carrier's ability to exit.[31] Perhaps the most important guideline the Commission should establish for these state COLR requirements is that the state must apply the same obligations to all COLRs in a given area. This is necessary to ensure that the Federal plan is competitively neutral.
A clearly defined COLR obligation is also necessary as the basis for the structure of an auction mechanism. In order to have an auction, there must be a commodity to be auctioned -- namely the COLR obligation. If a party wins the auction, it must be possible for that firm to take on the COLR obligation in return for the support level it has bid. This means that commissions must be able to assign the obligation they have defined to a carrier other that the ILEC. As GTE has shown supra, the Act clearly allows for this to occur.
Similarly, it must be possible for a carrier -- including the ILEC, to lose the auction. In this case, the carrier should lose access to the COLR funding. But it should also be relieved of any COLR obligations, an be treated symmetrically with other carriers who are not COLRs. Note that in the auction structure GTE proposes, the incumbent LEC would only be able to lose the COLR auction in the event that at least one other firm wins it. Thus, as the Act requires, the availability of service would be guaranteed even if the COLR designation changes from the LEC to another carrier.
If other parties entered the auction knowing that the LEC could not "lose", it would certainly interfere with the auction outcome. Each party could prepare its bid secure in the knowledge that the LEC would be required to provision it as an underlying carrier. Not only would this be unfair to the LEC; it would also prevent the auction from revealing information about the bidders' costs. Instead, the bids would simply reflect the rates which had been established for resale of the LEC's facilities or bundled services.
Fortunately, the 1996 Act clearly provides the Commission with the authority it needs to define a COLR auction process in such a way that the incumbent LEC can lose.
1. Subsections 254(a) and (b) place firmly on the FCC -- acting so as to take account of Joint Board recommendations -- responsibility to develop and implement a Universal Service Plan that is effective in maintaining and promoting universal service throughout the country. Specifically, one of the subsection 254(b) "principles" upon which "the Joint Board and the Commission shall base policies for the preservation and advancement of universal service" is stated in subsection 254(b)(5): "There should be specific, predictable and sufficient Federal and State mechanisms to preserve and advance universal service." Even more significant is the mandate of subsection 254(e), which says, "Any such [Federal universal service] support should be explicit and specific to achieve the purposes of this section [254]."
2. Under subsection 160(a) (subsection 10(a) of the 1996 Act), Congress gave the FCC broad power to "forbear from applying any regulation or any provision of this Act to a telecommunications carrier or telecommunications service [or a class thereof] in any or some of their geographic markets...." Under subsection 160(a)(1), (2) and (3), exercise of this power depends upon an FCC determination that amounts to saying regulation is not necessary and forbearance is consistent with the public interest.
3. A further qualification on this power of the FCC serves to stress the scope of the power. Subsection 160(d) says the Commission (putting aside subsection 254(f), not relevant to this discussion) "may not forbear from applying the requirements of [sub]section 251(c) or 271 under subsection (a) of this section [160] until it determines that those requirements have been fully implemented." This makes it unmistakable that
the FCC may forbear from applying the subsection 251(c) and section 271 requirements once there has been a showing of full implementation.
4. Driving home the force of the FCC's broad power of forbearance is subsection 160(e), which makes the FCC's forbearance decision preemptive. This subsection says a "State commission may not continue to apply or enforce any provision of this Act that the Commission has determined to forbear from applying under subsection [160] (a)."
5. In harmony with the overall thrust of the 1996 Act looking to deregulation as competition becomes established, subsection 214(e)(4) is concerned with relinquishment of a universal service obligation. This provision contemplates a case where more than one eligible telecommunications carrier (Eltel) serves a particular area. It sets out the procedures by which a State commission, upon advance notice by an Eltel that it wishes to relinquish a universal service obligation, may permit such relinquishment provided the remaining Eltel(s) are required to ensure that all customers served by the relinquishing carrier will continue to be served. Such notice must be "sufficient ... to permit the purchase or construction of adequate facilities by any remaining [Eltel]," and the purchase or construction should be completed within a year of State commission approval.
6. Subsection 214(e)(4) nicely accommodates the GTE auction scheme. The duties of an ILEC under subsection 251(c) are the counterpart of Carrier of Last Resort (COLR) obligations. As proposed by GTE, an existing COLR/ILEC could be discharged of the COLR/subsection 251(c) obligations in favor of a successful bidder qualified and willing to assume these obligations, in exchange for which the successful bidder would receive universal service support. Alternatively, the existing COLR/ILEC might choose to continue in that capacity, but then it would have to share universal service support with the successful bidder and that support would be set at the per-customer amount of the successful bid. The FCC's broad power of forbearance, discussed supra, provides ample authorization for FCC action -- preempting the State commission as necessary -- discharging a COLR/ILEC from subsection 251(c) obligations in accordance with GTE's scheme.
7. Still another indication in the 1996 Act that Congress contemplated a relinquishment of COLR/subsection 251(c) obligations upon a new ILEC assuming such obligations is contained in subsection 251(h)(2), under which the FCC "may, by rule, provide for the treatment of a local exchange carrier (or class or category thereof) as an [ILEC] for purposes of this section [251] if:
"(A) such carrier occupies a position in the market for telephone exchange service within an area that is comparable to the position occupied by a carrier described in paragraph (1) [which defines ILECs];
"(B) such carrier has substantially replaced an [ILEC]; and
"(C) such treatment is consistent with the public interest, convenience, and necessity and the purposes of this section."
8. Thus, Subsection 251(h)(2) contemplates that, under the FCC/Joint Board Universal Service Plan, an ILEC would be "substantially replaced" by another firm occupying a position in the market comparable to an ILEC's position. Accordingly, it provides for the new ILEC/COLR to assume those obligations; while discharge of the replaced ILEC/COLR is accommodated under the FCC's broad and preemptive forbearance power discussed supra.
9. These provisions hang together. They offer a coherent statutory plan well suited to implementing the GTE proposal, which was extensively discussed with congressional staffers as well as Representatives and Senators.
51. What, if any, safeguards should be adopted to ensure that large companies do not bid excessively low to drive out competition?
See response to question number 49, supra.
52. What safeguards should be adopted to ensure adequate quality of service under a system of competitive bidding?
The task of monitoring and enforcing COLR performance of their obligations rests in the first instance with the states.[32] Further, under the plan proposed by GTE, no additional measures are needed to ensure that COLRs use support only for the intended services, because payment of support, tied to the performance of the obligations established for COLRs by the state, is just sufficient to compensate the COLR for that performance.
In any event, there is nothing about the auction process that creates any concerns with respect to quality enforcement that would not be present in the same degree if the support is determined on the basis of cost. If the concern is that the firm would be motivated to shirk its obligations because the support is set too low, GTE responds that a too-low support level is more likely to result from error or misspecification in the cost estimate than it is from a bid that the carrier itself has submitted. Note also that the proposed auction scheme would base support for all COLRs on the highest accepted bid, rather than on the lowest bid.
53. How is collusion avoided when using a competitive bid?
As Professor Milgrom explains, the single-round, sealed bid format proposed here is not vulnerable to collusion among the bidders.
54. Should the structure of the auction differ if there are few bidders? If so, how?
See response to question number 49, supra.
55. How should the Commission determine the size of the areas within which eligible carriers bid for universal service support? What is the optimal basis for determining the size of those areas, in order to avoid unfair advantage for either the incumbent local exchange carriers or competitive carriers?
GTE proposes that the geographic units used for the auctions, and for the assignment of COLR obligations, should be small, standard units, such as CBGs. This allows the bidding to establish separate support levels which will capture differences in cost across areas. Further, because each area is small, the requirement to serve the entire area will not create an unreasonable barrier to entry for a prospective COLR. Each entrant can determine the set of CBGs that would comprise the area it wishes to serve, and submit NOIs for those CBGs. The framework is thus very flexible, and adaptable to the business plans of the entrants. GTE submits that there is no arbitrary grouping of CBGs the Commission could establish in advance that would accommodate firms' business plans as well as the groupings they would create for themselves through the NOI process.
Because the bidding cycles would group CBGs together to be auctioned at the same time, and because the proposed single-round auction is inherently simple, GTE believes that it has structured the proposal to allow the large number of CBGs to be auctioned efficiently, over time, as they are notice by the carriers.
Benchmark Cost Model (BCM)
56. How do the book costs of incumbent local exchange carriers compare with the calculated proxy costs of the Benchmark Cost Model (BCM) for the same areas?
GTE's prior submissions in D.96-45 and CC Docket No. 80-286 addressed many of the issues within the questions pertaining to the Benchmark Cost Model. GTE will provide more specific comments on August 9, 1996, in response to the Commission's Public Notice DA 96-1094, released July 10, 1996.
57. Should the BCM be modified to include non-wireline services? If wireless technology proves less costly than wireline facilities, should projected costs be capped at the level predicted for use of wireless technology?
No, there is no need to modify the BCM to include non-wireline services. The starting point for initiation of a new universal service support plan should be today's ILEC network design and the associated costs because that is the network used to provide local service today. When new entrants, perhaps using wireless technology, desire to serve a high cost area and be eligible for universal service support, the auction process proposed by GTE obviates the need to accommodate wireless and other technologies in a cost model. The bidding process itself will provide its own estimates of service costs through the support levels bid by each potential provider.
58. What are the advantages and disadvantages of using a wire center instead of a Census Block Group as the appropriate geographic area in projecting costs?
GTE's prior submissions in D.96-45 and CC Docket No. 80-286 addressed many of the issues within the questions pertaining to the Benchmark Cost Model. GTE will provide more specific comments on August 9, 1996, in response to the Commission's Public Notice DA 96-1094, released July 10, 1996.
59. The Maine PUC and several other State commissions proposed inclusion in the BCM of the costs of connecting exchanges to the public switched network through the use of microwave, trunk, or satellite technologies. Those Commenters also proposed the use an additional extra-high-cost variable for remote areas not accessible by road. What is the feasibility and the advisability of incorporating these changes into the BCM?
GTE's prior submissions in D.96-45 and CC Docket No. 80-286 addressed many of the issues within the questions pertaining to the Benchmark Cost Model. GTE will provide more specific comments on August 9, 1996, in response to the Commission's Public Notice DA 96-1094, released July 10, 1996.
60. The National Cable Television Association proposed a number of modifications to the BCM related to switching cost, fill factors, digital loop carrier subscriber equipment, penetration assumptions, deployment of fiber versus copper technology assumptions, and service area interface costs. Which, if any, of these changes would be feasible and advisable to incorporate into the BCM?
GTE's prior submissions in D.96-45 and CC Docket No. 80-286 addressed many of the issues within the questions pertaining to the Benchmark Cost Model. GTE will provide more specific comments on August 9, 1996, in response to the Commission's Public Notice DA 96-1094, released July 10, 1996.
61. Should the support calculated using the Benchmark Cost Model also reflect subscriber income levels, as suggested by the Puerto Rico Telephone Company in its comments?
Under the program recommended by GTE, the level of support is not directly "calculated using the Benchmark Cost Model." High cost support would be calculated based upon a comparison of the output of the BCM with a benchmark affordable price established by the FCC and the Joint Board.
Individual subscriber characteristics, such as income or similar customer- or area-specific factors, should not be considered in distributing high-cost support. The need to support individuals whose income falls below a specified level should be addressed separately through a program of income-based support, similar to the current Lifeline program. Such a program would entail means-testing to determine eligible customers, so it would only apply to a relatively small subset of customers.
While the concept of tailoring support generally to fit each individual's need is attractive, it is not administratively feasible.[33] Means-testing every customer would be unreasonably burdensome, and would involve an unacceptable intrusion in customer privacy. Even limiting means-testing to those requesting assistance would be complex and expensive and would involve private firms in an activity that is properly the role of a governmental agency. Further, the determination of the support needed for each customer would also require information on customer-specific costs and other factors, which are unlikely to be available. In the absence of this information, support would still have to be averaged over some defined area.
There appears to be no useful way to use area-specific income data to adjust the average support amount provided in an area. If income data for a large area is used, it will not reflect variations in customer income very closely. If median or average income for a small geographic area is used, it will create very large variations in the amount of support provided. The problem with the latter approach is that even small areas are not very homogeneous with respect to household income. For example, a minimum-wage employee in a restaurant in Aspen may have to pay a very high price for telephone service if the "affordable" rate is based on the average or median income in Aspen. On balance, GTE recommends that income not be used to determine the amount of support provided in an area.
62. The BCM appears to compare unseparated costs, calculated using a proxy methodology, with a nationwide local benchmark rate. Does use of the BCM suggest that the costs calculated by the model would be recovered only through services included in the benchmark rate? Does the BCM require changes to existing separations and access charge rules? Is the model designed to change as those rules are changed? Does the comparison of model costs with a local rate affordability benchmark create an opportunity for over-recovery from universal service support mechanisms?
GTE's prior submissions in D.96-45 and CC Docket No. 80-286 addressed many of the issues within the questions pertaining to the Benchmark Cost Model. GTE will provide more specific comments on August 9, 1996, in response to the Commission's Public Notice DA 96-1094, released July 10, 1996.
63. Is it feasible and/or advisable to integrate the grid cell structure used in the Cost Proxy Model (CPM) proposed by Pacific Telesis into the BCM for identifying terrain and population in areas where population density is low?
GTE's prior submissions in D.96-45 and CC Docket No. 80-286 addressed many of the issues within the questions pertaining to the Benchmark Cost Model. GTE will provide more specific comments on August 9, 1996, in response to the Commission's Public Notice DA 96-1094, released July 10, 1996.
Cost Proxy Model Proposed by Pacific Telesis
64. Can the grid cell structure used in the CPM reasonably identify population distribution in sparsely-populated areas?
GTE will provide more specific comments on August 9, 1996, in response to the Commission's Public Notice DA 96-1094, released July 10, 1996.
65. Can the CPM be modified to identify terrain and soil type by grid cell?
GTE will provide more specific comments on August 9, 1996, in response to the Commission's Public Notice DA 96-1094, released July 10, 1996.
66. Can the CPM be used on a nationwide basis to estimate the cost of providing basic residential service?
GTE will provide more specific comments on August 9, 1996, in response to the Commission's Public Notice DA 96-1094, released July 10, 1996.
67. Using the CPM, what costs would be calculated by Census Block Group and by wire center for serving a rural, high-cost state (e.g., Arkansas)?
GTE will provide more specific comments on August 9, 1996, in response to the Commission's Public Notice DA 96-1094, released July 10, 1996.
68. Is the CPM a self-contained model, or does it rely on other models, and if so, to what extent?
GTE will provide more specific comments on August 9, 1996, in response to the Commission's Public Notice DA 96-1094, released July 10, 1996.
SLC/CCLC
69. If a portion of the CCL charge represents a subsidy to support universal service, what is the total amount of the subsidy? Please provide supporting evidence to substantiate such estimates. Supporting evidence should indicate the cost methodology used to estimate the magnitude of the subsidy (e.g., long-run incremental, short-run incremental, fully-distributed).
The CCL recovers all non-traffic sensitive loop costs that are not recovered through direct end user Subscriber Line Charge. Thus, all of the traffic sensitive CCL charge can be viewed as a subsidy to local service.
70. If a portion of the CCL charge represents a contribution to the recovery of loop costs, please identify and discuss alternatives to the CCL charge for recovery of those costs from all interstate telecommunications service providers (e.g., bulk billing, flat rate/per-line charge).
GTE recommends recovery of all loop costs through a combination of charges to the end user and Federal and state universal service support. Under GTE's plan, universal service support provided to an ILEC would be used to eliminate the CCL.[34]
Low-Income Consumers
71. Should the new universal service fund provide support for the Lifeline and Linkup programs, in order to make those subsidies technologically and competitively neutral? If so, should the amount of the lifeline subsidy still be tied, as it is now, to the amount of the subscriber line charge?
Yes, the new universal service fund should generate funding for the Lifeline and Link Up America programs. Support should be made available to all low-income individuals that provide proof of meeting income level criteria established by a state regulatory agency[35]
The amount of Lifeline subsidy should not be tied to the subscriber line charge or be linked with any other FCC accounting, separations and access charge rules. This is necessary to be competitively neutral so that support may be available to Eltels that are not required to use the FCC's accounting, separations or access charge rules. Specifically, each Lifeline customer should receive a credit to offset the charges the customer selects[36] This program should not be tied to the interstate EUCL as it is today because only incumbent LECs assess such a charge.
Administration of Universal Service Support
72. Section 254(d) of the 1996 Act provides that the Commission may exempt carriers from contributing to the support of universal service if their contribution would be "de minimis." The conference report indicates that "[t]he conferees intend that this authority would only be used in cases where the administrative cost of collecting contributions from a carrier or carriers would exceed the contribution that carrier would otherwise have to make under the formula for contributions selected by the Commission." What levels of administrative costs should be expected per carrier under the various methods that have been proposed for funding (e.g., gross revenues, revenues net of payments to other carriers, retail revenues, etc.)?
GTE cannot offer estimates of the administrative costs under the various scenarios posed in question number 72. The most relevant information will be obtained directly from potential administrators through a competitive bidding process.
The overriding principle that should guide the Commission in examining any claims of exemption from payment into the fund is that of competitive equity. Such equity requires that all potential payers report the relevant data each reporting period. This will ensure that new entrants with high growth rates will not escape their fair
contribution when their size warrants contribution. Thus, requests for exemption from payment should be granted only for one reporting time period.
Respectfully submitted,
GTE Service Corporation and its affiliated domestic telephone operating companies
By___________________________________
Richard McKenna, HQE03J36
GTE Service Corporation
P.O. Box 152092
Irving, TX 75015-2092
(214) 718-6362
August 2, 1996
Their Attorneys
As will be apparent from the discussion below, the Commission confronts a number of trade-offs in designing an auction. The comment period in the Commission's Notice is not sufficient for me to recommend to the Commission the optimal way of making those tradeoffs. For that reason, this statement should be considered an outline describing some of the main features that should be included in a COLR auction, rather than as a final, fixed proposal.
When there are two or more potential carriers of last resort (COLRs), auctions have several important advantages over industry cost models as a means of determining the support payments for meeting universal service obligations. First, an auction uses an actual market process to set support levels. That is desirable not only to avoid the controversies that inevitably accompany cost modeling and estimation but also because even the best cost models are both biased and incomplete as a basis for setting support levels. Support payments based on cost models overestimate the actual level of support needed to attract a COLR when the LEC technology and facilities locations on which the models are based are not the least cost way to meet the COLR obligation. Also, when the LEC technology is the cheapest way to meet COLR obligations but competition in the provision of services is desired, support payments based on LEC costs may be too low to attract and sustain the desired competition, or perhaps any competition at all. Further, it is reasonable to assume that the firms' actual bids will be based on even more detailed cost estimates than could be reflected in an industry cost model and will be reduced to reflect the profit opportunities on any incidental or complementary services that the firm expects to sell along with basic services. No model that the Commission could plausibly implement would include so many factors or be based on such detailed cost analysis as the bids in an auction.
A second advantage is that auctions can determine how many COLRs should be supported and who they should be. Competition among potential COLRs can be of two kinds: "competition in the market" - in which several carriers accept COLR obligations and compete to acquire subscribers and the associated support payments - or "competition for the market" - in which companies bid for the right to serve as the exclusive COLR (or as one of a limited number of COLRs). "Competition in the market" is likely to lead to more innovative and responsive service to consumers and to reduce the severity of "hold up" problems that come from reliance on a single supplier. However, competition in the market can also result in duplicated facilities costs and burdensome support payments that necessitate imposing surcharges on other communications services. Competition "for the market" in a traditional auction can lead to lower support payments as the bidders vie aggressively for the exclusive (or at least limited) right to serve as a COLR, reducing the burden on other services. Auctioning a fixed number of COLR designations would require the FCC to determine the fixed numbers: it must decide how many COLRs to authorize in each area. That determination would be a difficult and costly one for any regulator to make well because it would require extensive and reliable cost information and, possibly, market and technology forecasts.[37] By contrast, my proposal permits the number of COLRs to be an outcome of the auction itself, as auction participants place bids based on what will be inherently better cost information and on what they believe is the best information on future market and technological developments.
Third, by establishing actual market prices for universal service in the various service areas, the auction provides useful information to potential entrants. Market prices are useful for determining which markets may be ripe for entry and what cost targets need to be reached to make entry profitable in these markets. COLR auctions would also be likely to generate statistical information about service costs that the FCC might find useful in other proceedings and at other dates. For example, the FCC might use the auction results in markets with substantial competition to assess standards for LECs in regions where there is no competition.
Another important advantage arises when service areas are re-auctioned over time, as I propose. A series of auctions allows the support payments to respond to changing technologies, population densities, and other factors. Probably, there will initially be some geographic areas in which only a single COLR operates but for which changing circumstances will eventually make competition among multiple COLRs feasible and desirable or in which reduced costs call for reduced support payments. The auction system can respond flexibly to changing circumstances, allowing entry to occur when the time is ripe and encouraging support payments to fall in tandem with the falling costs of service.
The auction proposal developed here calls for sealed tender auctions that would allow multiple COLRs to be selected if the several lowest bids are close enough together. The support levels would be the same for each COLR serving an area and would be set equal to the highest accepted bid.
This is a novel auction design, constructed to meet the novel challenges posed by the universal service context. While the FCC's simultaneous multiple round auctions have proved themselves to be effective for the spectrum sales with fixed numbers of licenses, I shall argue that such a design is less well suited to determine the extent of competition that should prevail among COLRs in each market area.
Section II of this statement examines theoretical considerations that apply in designing an auction to determine the amount of support and the level of competition simultaneously. Section III contains a specific proposal and a discussion of both the basic auction design and related practical details.
It is important to set realistic expectations about what a good auction design can and cannot achieve. Most importantly, auctions cannot resolve all the problems that may arise when there is a single facilities based universal service provider. If a single COLR with large sunk costs is the inevitable practical outcome in any particular geographic region, no auction, however cleverly it may be designed, can substitute for effective continuing regulation of the monopoly COLR.[38]
Second, an auction system cannot be effective unless the bidders have something to win. If one allows providers other than auction winners to provide basic service with support from the universal service fund, then that eliminates the bidders' incentives to bid for a low support levels,[39] leading to undesirable increases in the surcharge needed to fund universal service.
The mathematical analysis of this section accounts explicitly only for the first of these differences, but the way the mathematical results are applied takes some account of the second, third and fourth differences as well.[42] That is, we seek an auction design that is simple for the bidders and the administrators, that generates uniform levels of support for all COLRs in a market area, and that is resistant to collusion while still taking proper account of the benefits arising from competition after the auction among COLRs in the market.
To derive principles to guide the design of an auction for carrier of last resort obligations, I first consider a scenario in which there is just one region in which universal service needs support. The main problem in this scenario is to use the bids to determine how many COLRs there should be and what level of support to pay. The principal qualitative finding of the analysis is that the auction outcome should specify that the COLR obligation is shared only when the bidders' service costs are sufficiently close. This may be reflected by sufficiently close bids in a sealed bid auction. Of course, the detailed quantitative conclusions of the analysis, including how many COLRs to authorize for any particular cost or bid levels, depend on the detailed assumptions of the model, but the general conclusion reported here is sufficient to help us distinguish some poor auction designs from more desirable ones. For example, I find that multiple round auctions such as those used for the PCS auctions, even in the trivial case where there is just one COLR service area for sale, cannot generally implement the optimal auction outcomes, but that certain sealed bid auctions can implement the optimal outcomes.
The theoretical analysis cannot specify how many COLRs should be assigned in any particular situation, but it can identify the relevant considerations. Generally, the number of COLRs should depend on the gains to increased competition in the ensuing market, the magnitude of the duplicated fixed costs (greater duplication favors fewer COLRs), the differences between the COLRs in the levels of their variable costs (smaller differences favor more COLRs), and the social loss associated with paying unnecessarily high support payments (larger losses favor fewer COLRs).
I make the simplifying assumption that the fixed costs of service are the same across bidders.[44] Also, at this stage, I assume that at least one COLR must be selected for each area.[45] The solution to this problem can be characterized using the methods of optimal auction theory.[46]
The optimal auction problem is to choose the rules and the behavior of the bidders, subject to the constraints described above, to maximize the following three-term objective:
Expected Benefits to Consumers
- Expected Costs Incurred by the COLRs
- __Expected Support Payments to COLRs
where _ is a parameter indicating the costs of distortions created by the support payments to the COLRs.[47] The benefit to consumers is assumed to be B1 if there is just one COLR; B1+B2 if there are two COLRs, and so on, with Bn denoting the incremental benefit of introducing an nth COLR to compete in providing universal service.
The analysis characterizes the optimal auction in terms of the outcomes that ensue. To avoid technical problems, we limit our analysis here to what the modern economic auction theory literature calls the "regular case."
Then, an auction design that always selects at least one winner is optimal if and only if its outcomes have these two characteristics: (1) bidders with sufficiently high costs cannot expect to profit from participating in the auction and (2) for any profile of actual costs, the set of bidders selected to be COLRs maximizes the expected benefits to consumers minus the expected costs incurred, minus _ times a "virtual cost" (which is a theoretical construct consisting of the actual cost adjusted upwards to account for bidding incentives). If the bidders are otherwise symmetric, multiple COLRs are most likely when the low cost bidders' cost levels are close together.
One immediate implication of this characterization is that multiple round auctions, which the FCC has used successfully in other contexts, are not well adapted to this context. To see why, consider the simplest case with just two bidders. An efficient multiple round auction would then need to specify that a support payment near the reserve is paid to both bidders if the auction ends immediately after opening bids near the reserve. With such rules, it is often consistent with rational behavior by both bidders for neither to lower the bid below the reserve even if the two bidders' costs are very different and much lower than the reserve.[48] In plain English, a multiple round auction that tries to implement the efficient outcome rule is exceptionally vulnerable to both explicit and implicit collusion. Such collusion is undesirable because it would be likely to result in unnecessarily high support payments and the inclusion of inefficient COLRs among the winning bidders.
An auction design that does encourage efficient outcomes in case there are just two bidders is the sealed tender auction in which two COLRs are assigned if the second lowest bid is close enough to the lowest bid. The support payment may be set equal to the highest accepted bid (although, as we shall see later, other payment rules are also permitted by the theory). An important advantage of the proposed sealed tender auction compared to the multiple round design is that it creates a powerful incentive for each bidder to defect from any pre-auction collusive agreement by undercutting its rival's bid in order to acquire the exclusive right to receive support payments for COLR services.
This analysis implies that an auction can be used to encourage competition both for the market and in the market even when there are only two bidders. Of course, the idea can also be extended to apply when there are more than two bidders. For a simple
(though unrealistic) example, suppose B2=B3=... (meaning that the incremental benefit of additional competitors is the same for each extra competitor). Let us assume for the cost calculation that the COLRs would share the market equally. Then, in the optimal auction, the nth lowest bidder should be included as a COLR only if the n-1 lower bidders are included and the cost of the nth lowest bidder does not exceed the average of the costs of the n-1 lower bidders by more than a specified amount c.[49] In the interests of simplicity, one might use an "approximation" of this outcome rule by specifying that all bidders whose bids are within some amount c_ of the lowest bid are included.
Generally, with more than two bidders, the form of the optimal auction depends on several things, including prominently the relative magnitudes of B2, B3, etc. On the basis of economic theory, it is reasonable to suppose that the benefits of additional competition decline as the number of competitors increase, that is, B2>B3>B4>.... The theoretically optimal rule in this case depends on the likely market shares of the bidders as determined by their various costs. If one assumes that the COLRs will eventually have roughly equal market shares, the optimal rule would be to include the nth bidder as a COLR if its cost is not too much higher than the average of the cost of the n-1 lower cost bidders. As a practical approximation of the actual optimal outcome rule, one might set the outcome rule in an actual auction as follows.
[--- Unable To Translate Graphic ---]
The parameters in this auction design - including the use of just three cases and the 15% and 25% cut-offs - are merely illustrative and not based on any detailed analysis. The illustrative rule shows how the auction is constructed to facilitate the presence of at least two actual COLRs in the market when the inefficiency from doing so, in terms of supporting a relatively inefficient competitor, are not too high. A more restrictive standard is set for including competitors beyond the second, because they are expected to contribute less to consumer welfare.
According to theory, the outcome rule described here could be used with any of several different payment rules without affecting the optimality of the auction. The payment rule, however, should be set to respect the other considerations not included in the optimal auctions model. For example, as described earlier, it is desirable to have the same level of support payments for each COLR, for that avoids creating distortions in the subsequent competition among them. One such rule would set each bidder's support payment at the level of the highest accepted bid. Yet another variation would specify that, in case 3 only, the support payment would be set at the level of the second lowest bid.[50]
Each of these variations would change the bidders' strategic problem and lead to different levels of bids being submitted, making cost comparisons among the various rules appear difficult. One of the surprising conclusions of optimal auction theory, however, is that contrary to simple intuition, the expected size of the support payments to the winners is unaffected by the form of the payment rules (among the set of payment rules that always produce the same set of winners). A rough explanation for this conclusion is as follows: If one payment rule leads to systematically higher support payments corresponding to any particular bids than another rule, the bidders will offset that difference by submitting systematically higher bids for the rule that calls for the lower support payments.
In practice, the proposed auction would consist of a large number of simultaneous sealed bids for the job of being the COLR. The main difficulty with this proposal is that it fails to allow bidders to account fully for "cost synergies," that is, for the possibility that it is cheaper to provide COLR services in one market when they are already providing COLR services in related markets. Such synergies might arise because the related markets used shared switching, transmission or other facilities.
However, permitting combination bids would add significantly to the complexity of the auction design, which is quite important given the possibly large number of small auctions to be conducted. To evaluate the potential benefits of combination bids, one needs to assess the importance of cost synergies.
The need for COLRs arises only in markets where it costs more to serve some potential subscribers than the established maximum basic service rate. If these high cost customers are subscribers who are distant from a town center, then the main cost complementarity may be between serving customers close to town and those at a greater distance from the town center. In that case, if service for the core town will be established anyway, then there are no important cost complementarities in serving two outlying areas bordering the town. If the core town will be served by the COLR in any event, then the model used to study the optimal auction adequately characterizes the basic auction design problem.
However, it may be the case that the bidder, possibly not the LEC, fails to win the COLR designation for the core town and rates for basic service are so low that support payments are required for service to all the potential subscribers in a particular town or other geographic area. In this alternative scenario, a firm's decision to provide any service to the area may depend on its ability to acquire business in the town core, or even throughout the related areas. If the relevant areas are the same for all bidders, one might try to avoid the problem by specifying larger areas for the universal service obligation. However, different customers within any large area may have very different costs of establishing service. That creates a problem as the COLRs avoid offering service to the highest cost customers. This "cherry picking" problem is discussed in more detail in the next section. Even without cherry picking, if the areas with synergies vary among bidders, then the way the areas are carved up is another tricky problem that needs to be resolved in the auction. These cases, which may be called the cases of "complex cost synergies," are the most difficult ones for simple auction designs to treat successfully.[51]
My central proposal is based on the presumption that complex cost synergies are of secondary importance, especially in areas where there are to be multiple COLRs, and that it is not worthwhile to adopt the more complex auctions necessary to account fully for cost synergies. In my judgment, the complexity of the combinatorial auction in this context are even greater than was found to be the case in the PCS spectrum auction. Partly, this additional complexity arises from the need to provide uniform pricing in each separate market after the auction, and partly it derives from the very large number of small areas that need to be combined. This complexity suggests that such combinatorial bidding schemes should only be considered where the strength of the synergies means the likelihood of very inefficient outcomes from any non-combinatorial scheme is very high. Even in that case, one might first consider the use of a simultaneous multiple round auction, weighing the risk of collusion against the desire to allow bidders to assess the values of combining service areas.
In the next section, to account in a highly imperfect way for cost synergies, I will propose a rule allowing winning COLRs to withdraw bids. The ability to withdraw bids allows the potential COLRs to avoid being forced to provide service in a patchwork quilt of geographic areas. These proposed withdrawals will be subject to penalties, as in the spectrum auctions, to discourage frivolous bidding
In summary form, the auction would be conducted as follows. Auctions would be conducted twice annually on specified dates. For each Census Block Group (CBG), the FCC or state PUCs would first establish a maximum support rate (the "reserve") based on a multiple of the predicted cost under an adopted cost model.[52] A notice process in which potential bidders nominate areas in which they are interested in providing service would fix the CBGs for which COLR obligations are to be auctioned. Those making nominations would be required to establish their qualifications to satisfy the COLR obligation. If a party indicates an intention to bid on one particular area for an auction, other parties may nominate additional adjacent areas to auction with that particular area. On the auction date, sealed bids would be submitted indicating the support levels that the bidders require.
In the initial auction for each area, if there are no bids submitted at or below the reserve, the LEC is designated the COLR at an "official" support level determined by the FCC or state PUCs and based upon a cost model (such as the BCM or CPM).[53] This would be treated as if no auction had transpired and the are would remain eligible to be noticed for auction.
Once a new COLR (instead of or in addition to the LEC) has been established in any CBG, the obligations would be fixed for a period of three years, subject to performance standards. After the initial three year term, any qualified entity could notice the area for an auction. If no one notices these areas, then the incumbents would continue to receive the same level of support payments but without extending the period of protection.
In order to mitigate the complex cost synergies problem described earlier, I suggest that any bidder be permitted to withdraw its bid from one or more areas. If a bid is withdrawn, the outcome of the auction will be determined as if the withdrawn bidder had never participated in the auction for that area. To discourage frivolous bidding and withdrawals, the FCC and/or state PUCs should establish withdrawal penalties similar to those adopted for the PCS auctions. The penalty might be equal to the larger of any increase in (e.g.) the twelve-month support obligation of the government as a result of the withdrawn bid or, say, $20 per subscriber in the CBG.
In what follows, I describe how these components will serve to ensure that the objective of providing universal service is efficiently attained.
a. The size of the service area.
It is very difficult, if not practically impossible, to define service areas that are homogeneous in terms of the costs of serving subscribers. Heterogeneous costs in a single service area lead to several costly effects. First, the COLRs may have an incentive to avoid serving the higher cost subscribers and to focus their marketing efforts solely on the relatively low-cost subscribers.[54] This problem is compounded when there is competition among COLRs, each of whom may hope to force its competitors to serve the subscribers for whom costs are highest. Second, support payments distort competition between COLRs and non-COLRs to serve subscribers for whom service can be provided at relatively low cost. The more heterogeneous the costs of service in an area, the worse these problems are likely to be. Smaller service areas therefore tend to reduce these costs.
An additional advantage of small service areas is that different service providers can assemble groups of areas that fit their technological capabilities. Larger service areas that include geographic areas outside the reach of a potential entrant may dissuade the entrant from bidding.
In economic terms, the choice between small and large service areas is governed by a comparison of the costs of cherry picking plus the costs of the monitoring and regulation needed to mitigate it, the costs of conducting auctions for a multitude of small areas, and the tendency of large service areas to block entry by some service providers. GTE has proposed the use of CBGs (which are quite small service areas) to control the costs of cherry picking and its regulation. If adopted in combination with my proposal for relatively simple, inexpensive sealed bid auctions, the package would constitute a coherent and workable plan for developing market competition.
Question 58 in the Commission's Public Notice asks whether wire centers rather than CBGs should be used as the basis for cost projections. The considerations already discussed above suggest that wire centers have two disadvantages. First, they are relatively large, encouraging cherry picking. Second, they are a natural area only for the incumbent LECs. A new entrant might be able to serve many CBGs but unable to serve the entire wire center, giving the LEC an artificial cost advantage in serving as the COLR. The use of CBGs would be technologically neutral because the definition of a CBG is unrelated to the provision of telephony. Thus, the use of CBGs would tend to avoid the possibility of biasing the auction outcomes towards one technology (or one incumbent).
b. One-shot sealed bids.
The simultaneous multiple-round auction format used in the FCC's spectrum auctions has a number of advantages. Foremost among them is that it permits bidders to take into account the possibilities of substitutability and complementarity among the licenses for which they bid and to adopt back-up strategies (for example, to acquire substitute licenses) in case their primary strategies fail.
In theory, the simultaneous multiple round format should be particularly good at accounting for substitutes, and the FCC experience has borne that out. In the paging auctions, for example, some bidders switched between bidding on the high capacity 50/50 licenses and the lower capacity 50/12.5 licenses during the auction to account for the changing levels of bidding activity. Similarly, in the PCS A and B block auctions, bidders frequently switched between the very similar A and B blocks, substituting between them. The simultaneous design also has important advantages over the sealed bid design in dealing with complementarities when those are important.
Substitution and "back-up strategies" are likely to play much smaller roles in the COLR auction than in the spectrum auctions, because the COLR obligations to service various areas are not technological substitutes. As in the PCS auctions, some substitution possibilities could be generated by a firm's service capacity limitations. Limited budgets could also lead bidders to seek a limited number of COLR obligations. However, the important technological substitution possibilities will be missing.
As against these advantages for the simultaneous multiple round auction, the sealed bid auction has advantages of simplicity and reduced vulnerability to collusion. Any pre-auction collusive agreement among bidders will tend to collapse in the sealed tender auction proposed here because each bidder has a straightforward and powerful incentive to defect from it.
Even if collusion were not an issue, the costs of administering a simultaneous multiple round auction for both the regulator and the bidders may not be worth the benefits. In the PCS auctions, the values of the individual licenses were substantial in comparison to the administrative costs of running the auction and the problem of collusion appears to have been of minor importance. The benefit-cost analysis in this case thus looks quite different than that of the PCS auctions.
c. Determining the support paid to winning bidders.
According to the optimal auction analysis in section II, if the bidders respond "rationally" and competitively to one another's strategies, then a variety of rules can be used to determine the support payment without affecting the efficiency of the overall design. Choices among these support rules must therefore be determined by factors apart from those built into the optimal auction model. These factors include (1) the ease or difficulty for bidders of determining their best ("rational") bid, (2) the vulnerability of the rule to collusive behavior, and (3) public perception of the rule as fair and reasonable.
Among the payment rules that might be acceptable according to the optimal auction theory are: (1) the payment is set equal to the lowest rejected bid or to the reserve if all bids are accepted and (2) the payment is set equal to the highest accepted bid. The first of these rules performs poorly in the public perception (as the experience of the New Zealand spectrum auctions demonstrates) and is vulnerable to some collusive bidding patterns.[55] The second rule is readily perceived as fair and reasonable, since it allows the bids to be interpreted straightforwardly as the lowest level at which the bidder offers to supply service. For that reason, I favor it.
d. The number of COLRs.
I would propose that the Commission permit the designation of multiple COLRs for any particular area, the number depending on the differences in the bid amounts. Lacking any quantitative basis for the assignment rule, I tentatively propose the rule described in the previous section. To repeat, that rule is as follows.
[--- Unable To Translate Graphic ---]
There are three advantages of a rule such as this. First, it encourages competition within the market for the patronage of potential subscribers. Second, the presence of multiple COLRs may ease the Commission's burden of monitoring and enforcing the performance of the COLRs after the auction, for several reasons. If some COLR is tempted to avoid serving the highest cost subscribers in a service area, the other COLRs will be led to detect and report that in order to avoid being forced to serve a disproportionate share of those subscribers. Multiple COLRs also provide the regulatory authorities an opportunity to compare the performance of several COLRs in the same market, making it easier to detect false claims about the impossibility of providing some promised services. Moreover, the Commission's threat to impose sanctions, including possible termination of a company's COLR status, is more credible if there are alternative COLRs available to protect consumers against service disruptions.
Third, the approach I have proposed accounts for both the declining benefits from designating multiple COLRs and the cost increases that may accompany a larger number of COLRs. When the bids of the participants are relatively close, the cost disadvantages from multiple COLRs will be correspondingly small, resulting in greater net benefits from multiple COLRs. In this case, the rule would designate multiple COLRs. When the cost differences are larger, the net benefits from multiple COLRs will be smaller, and the proposed rule would limit the number of COLRs designated.
e. The "official" reserve and the auction initiation.
For each CBG, the Commission should establish a maximum support level or "reserve" equal to the difference between the standard rate for the basic service package and a multiple[56] of the cost estimate of providing that package based on an estimation model such as the CPM or BCM. The primary purpose of the reserve is to limit the required support payment in areas where only the LEC can provide economical service. However, the ceiling created by the reserve will also encourage somewhat lower bids in the auction.
After the official reserves have been set, the Commission (or the state PUCs) should allow bidders to nominate CBGs for inclusion in the next auction. This could be done by asking interested parties to submit a Notice of Intent by some specified date before each auction. If the auction for a particular CBG attracts any valid bids from any bidder besides the incumbent LEC, the auction is held; if it attracts no bidders or if only the incumbent LEC submits a valid bid, the incumbent would retain the COLR obligations at the previously established support level based on a multiple of estimated costs.[57] Similarly, in any area where an auction has not yet been held, the incumbent LEC would retain the COLR obligation at the previously established support level.
For those CBGs for which auctions are held, the designated COLRs would be obliged to provide service beginning, say, one year or eighteen months after the COLR designation. This delay is to permit new entrants whose business plans call for additional facilities investments to make those investments after winning in the auction. This encourages the widest feasible participation in the auction.
f. Exploiting synergies in adjacent CBGs and withdrawal penalties.
Participants in the auction may bid on as many CBGs as they choose, thus permitting bidders some limited flexibility to account for economies of density and scale in their CBG-specific bids. Thus, if a particular entity bids for only one CBG and there are scale and density economies in serving that CBG and adjacent CBGs, then another entity can underbid the first entity in the one-shot auction format.
Some winning bidders may discover after the auction that the aggregation of the particular CBGs won would not permit the bidder to attain all of the expected synergies. This is likely to be a serious problem only if both of the following two conditions apply:
(1) the bidders' overall cost levels are similar and (2) the synergies are strong. The first condition makes it more likely that each bidder wins a COLR role in several areas, which is a pre-requisite for the problematic "checkerboard pattern," and the second is necessary for the consequences to be economically costly. To help remedy this problem when it is most severe, I propose that a winning bidder be permitted to withdraw its bid for some period after the auction. In effect, a bid withdrawal substitutes partially and quite imperfectly for combinatorial bidding.
When a winning bidder withdraws its bid for a CBG, the auction outcome would be determined by the remaining bids as if the withdrawn winner had never bid. (If only the incumbent LEC remains as a bidder, the auction is canceled, and the incumbent LEC receives support payments at the previously determined level.) This rule prevents any participant from using withdrawals strategically to trigger a new auction, thereby effectively turning a one-shot auction into a multiple-round auction.
Although withdrawals should be permitted, they also need to be penalized. There are two important reasons. First, the withdrawals may disrupt the outcome of the auction and the plans of other bidders and so need to be discouraged. Second, the lack of any penalty may encourage frivolous bidding, in which the bidder attempts to assemble unrealistic combinations or tries to mislead competitors about its future intentions. If there are no penalties, this sort of disruptive bidding behavior is riskless to the bidder.
To assist in maintaining the integrity of the auction, I would propose that the Commission establish moderate withdrawal penalties to deter frivolous bidding, as it did
in the PCS auctions. To determine the withdrawal penalty, the Commission would assume that in the future, the winning COLRs would have equal market shares in the CBG. The penalty for a withdrawn bid might be equal to the larger of any increase in the twelve-month support obligation of the government as a result of the withdrawn bid or, say, $20 per subscriber in the CBG. The penalty protects the government from any increases in its support costs and provides some compensation for any loss in post-auction competition resulting from the frivolous bid.
g. The length of the COLR designation.
The length of the time period for which an entity is designated a COLR has several effects. First, a long period ensures that what a bidder wins by making a low bid is of significant value. Second, the period affects the pattern of investments that may be undertaken to provide COLR services.
Encouraging efficient investment is a subtle matter. Optimal investments require that today's COLRs properly anticipate the likelihood that superior technologies will become available tomorrow, replacing the COLR or cutting into its profit margins. Setting too long a period of protection discourages or even blockades entry when the new technology becomes practically available. Setting too short a period may require large initial support payments to allow the investor to recover its investment in a short period. Such support payments may exceed the reserves or be embarrassing to the regulator.
To balance these competing concerns, I have tentatively proposed a three year period for the COLR obligation. To account for cost increases during the interim, the Commission could periodically raise the support rate by an exogenous index of costs, in the same way that the Commission currently implements its price cap policies.
Further, to allow new entry to occur when it is ready, the three year period of protection might not apply to auctions in which the set of COLRs serving an area does not change, or changes by the exit of a COLR. The three year period of protection would then apply only when a new COLR is introduced into the group serving a particular CBG. The justification is that only a new COLR might be regarded as needing an initial period of predictable competition during which it amortizes its investment.
At the end of the three year period, the areas for which the COLRs were selected via an auction would be eligible to be nominated by qualified parties for a new auction. The rules for these auctions would be nearly identical to those for the original auctions, but taking into account that the COLR for an area may no longer be the LEC. Simply put, the FCC (or state PUCs) would once again announce an official reserve and call for bidders. If no notice of intent is received for a CBG or if there are no valid bids for it, then the incumbent COLRs retain the obligation to provide basic service at the original support rate.
h. Default penalties.
If a bidder defaults, the outcome could be determined as if there had been a withdrawal, as discussed above. However, the costs to the government and consumers will be more substantial the longer the time between the initial auction and the default. This is because the plans of other potential COLRs may have been seriously affected. Consequently, any replacement for the defaulted COLR is likely to demand a higher support level for the shorter-term obligation than for the initial obligation.
Because the COLRs are likely to be parties with continuing relations with the regulators, there are many ways for the Commission to discourage default. The Commission should explore whether it may modify any of its current regulatory penalties for the purpose of deterring the default of a COLR.
i. Transferability of the COLR obligation
As already noted, the proposed auction mechanism has only a limited ability to accommodate synergies in service provision across CBGs. To permit COLRs to realize greater economies after having some experience with the COLR obligation, I would permit a COLR to sell its COLR status to any other qualified company (for example, one that is a COLR in some CBG) that is a non-COLR in that particular CBG. That is, sale would permitted be to a qualified firm (as evidenced by its COLR obligations elsewhere) provided it does not reduce the number of competing COLRs in the affected service area.
Permitting the COLR to sell the obligation after the auction also permits a bidder whose costs are unexpectedly high to transfer the obligation to a more efficient provider.
Attachment 2
Excerpt from
An Auction Mechanism for Determining
Universal Service Support
Dennis Weller
Chief Economist
GTE
Rutgers University
Ninth Annual Western Conference
July 1996
2. The Carrier of Last Resort Obligation
In the past, the regulator has been able to rely on the incumbent as the "carrier of last resort" ("COLR"). It could require the incumbent to serve any customer on request, and could establish prices, terms, and conditions as needed, through its pervasive regulation of the incumbent, to meet its policy needs. This asymmetric requirement would appear to be incompatible with a competitive market. They represent a burden on the incumbent that other carriers do not share. To the extent that the required prices and terms could be sustained, they would be a barrier to entry; to the extent they could not, the universal service objective itself would not be achieved.
At first blush, it would appear that the way to deal with the problem is to eliminate the carrier of last resort obligation, and to simply distribute funding from a new mechanism to all carriers. However, after some more thought, I have come to the conclusion that linking support to an obligation to serve is in fact an efficient design, given a real world of limited information. This is so because customers are heterogeneous.
Evidence from recently developed cost models suggest that the cost of providing local service varies dramatically from one place to another. Even within a small rural town, the cost of a customer on the main street may be an order of magnitude less than the cost of a farmer several miles outside of town. One tool for dealing with this variation is to calculate support amounts for geographic units which are relatively small.[58] However, even within a small area, some variation in cost will remain. Further, customers will differ with respect to other variables of interest to a prospective local carrier, such as their demand for other services (toll, vertical services, video, etc). It is useful to think of an ordering of customers according to their attractiveness to a carrier. This could be expressed in terms of the amount that would be required, as a side payment from the regulator, to induce an efficient provider to serve a given customer at the price and terms the regulator wishes to maintain.[59]
Given perfect information, the regulator could simply announce, for each customer, the amount of support -- the side payment -- that would be just sufficient. This would obviate the need for any obligation to serve, since supply would be forthcoming on a voluntary basis for every customer. Unfortunately, the regulator does not have perfect information, and must therefore make do with some average level of support for all of the customers within a given area. If carriers receiving such support are able to serve selectively, then not all customers will be served. The only way to ensure supply for all customers in an area would be to set the average support amount at the level carriers would require to serve the least desirable customer in the ordering. This would be expensive, since the regulator would pay too much for all but one of the customers.[60]
The carrier of last resort may therefore be seen as performing an averaging function for the regulator. The regulator "bundles" heterogeneous customers in an area together, and announces an average support amount for a carrier willing to serve any of them. In effect, the regulator delegates its information problem to the COLR. There is a limit, however, to the degree of heterogeneity that is reasonable to expect this structure to deal with; if the range is too great, the COLR will not be able to sustain its obligation in the face of entrants who are not so encumbered, nor will the regulator be able to police the COLR effectively to avoid shirking. For this reason, I believe that the carrier of last resort obligation should be assigned, and the support level set, for small geographic areas.
In the past, there has been only one carrier of last resort -- the incumbent. There has been considerable debate on the issue of how many carriers should be designated as COLRs in a new universal service plan designed for a competitive market. Several alternative structures could be adopted in a new universal service plan:
1) The regulator could continue to designate the incumbent as the sole COLR. This would not appear to be consistent with the objectives of the plan.
2) Some method, such as competitive bidding, could be used to select a single COLR. This would permit a priori competition for the market among prospective candidates for the COLR designation. However, it could limit ex post competition in the market, since support would be available to only one carrier. Of course, other carriers would be free to enter, and to provide packages of service to selected customers without support.
3) Support could be provided to more than one COLR, each of which would be obligated to provide the basic service on request to any customer in the area. The support could then be provided on a per-customer basis, so that support would be distributed among the COLRs by the customers' own choice of carrier. This would allow the COLRs to compete in the market ex post on an equal footing. It would also eliminate the possibility that the provision of support to a single COLR would deter entry by another efficient provider.[61] On the other hand, provision of support to several COLRs could induce inefficient entry, and thus raise the cost of supply in the market generally.
4) All carriers in the market could be treated as COLRs. I do not believe that this option is attractive, since it would establish an obligation to serve as a condition of entry. Thus would effectively force all firms to be ubiquitous providers, and would preclude niche entry.
Option 2 would appear to be preferable to option 1, since it would at least ensure a priori competition for the right to be the COLR. Option 3 would, in addition, promote more effective ex post competition among the COLRs, at the possible cost of some inefficiency of supply.
Option 3 would result in a sort of two-tier market. Anybody can enter and provide any kind of service under whatever minimal regulation the regulator decides to apply to all carriers. All of these carriers are treated alike. A subset of these chooses to perform a specific function defined by the regulator -- serving as carrier of last resort - - for which each is compensated by an average amount per customer served in the (small) area.[62] All of these COLRs are also treated alike. This scheme is thus as symmetric as it can be and still employ a COLR obligation.
It is useful to think of the COLRs simply as suppliers who perform a specific function the government wants done, and are compensated at a market level for doing so. Viewed in this way, the COLRs are no different from many other vendors who enter into contracts to supply services to the government. Government agencies can, and often do, select multiple suppliers for a contract; but other firms can enter the market and do business without becoming such a vendor.
The risk that this system will bias the market outcome can be minimized if the symmetry I have described is maintained, if the COLRs are self-selecting, and if the level of compensation can be set as closely as possible to the level that would be needed to induce firms to self-select in an open market.[63] I therefore find it natural to think of an auction process as the means for selecting the carriers of last resort, and for determining the support level.
My priors are that the value of enhanced ex post competition that option 3 would permit is likely to outweigh any possible inefficiency that may be created, relative to option 2, by supporting multiple COLRs. Ideally, the number of COLRs would be determined endogenously through the auction process, rather than determined in advance by the regulator. The auction would also determine which firms should serve as COLRs, and what the level of support, per customer, should be.
3) Elements of a Universal Service Plan
Before discussing the auction process itself, I will first fill in some of the basic elements of the framework in which the auction might be employed.
First, I suggest that the regulator should determine what it expects carriers of last resort to do. It should publish a clear list describing the obligations a COLR would undertake. This would include such items as the definition of the basic service package the COLR must provide, a ceiling on the price to be charged, any other requirements as to terms and conditions, quality standards, tariffing requirements, resale obligations, and so on.[64] In itself, this process would be an important step in the transition from traditional regulation of the incumbent, since it would transform universal service from a sometimes ill-defined function that the incumbent performs under pervasive regulation to a specific function that could be assigned to any carrier(s). In effect, the list would serve as the basis for a Request for Proposal, or bill of particulars, that bidders would respond to. For this purpose, the regulator should also specify a time period for the COLR responsibility -- effectively the term of the contract the successful bidders would be entering into with the regulator. This would be necessary for firms to prepare their bids, and for the regulator to know commitment it was buying from the carriers.
Support would be provided to each COLR on a per-customer basis. Each COLR in an area would be obligated to serve any customer that chose it. COLRs would thus be forced to compete among themselves, and the distribution of support would be determined by the customers in selecting their carriers. When a customer switched providers, the support would be portable, if the new carrier was also a COLR.
Second, the regulator should develop cost estimates to serve as a basis for estimating initial levels of support. At the outset, the support the incumbent receives, per customer, in each area would be the difference (where positive) between the cost measure and the price ceiling set by the regulator. This is necessary because, at the outset, in many areas there will not be firms willing to bid for the carrier of last resort function. The cost-based approach would serve as an alternative means of determining support.
The cost-based approach would attempt to estimate the market intervention in the carrier's price. However, even if it could do this well, it would not be able to capture any other factors which might affect a carrier's willingness to serve as a COLR. These might include any nonprice requirements imposed as part of the list published by the regulator, such as quality standards. They could also include any relevant complementarities, of demand or supply, with other services. Finally, they could simply include the difficulty the firm expects in dealing with the regulator. The bidding process, if successful, could put a price tag on any such factors in a way that a cost model cannot.
In fact, a number of models have been developed to estimate local service costs, by small geographic area, for universal service purposes. For the reasons I have just mentioned, I believe that such models are necessary. However, the experience with these models has shown that they are difficult to develop, and the subject of considerable controversy. For example, I have been spending a good deal of time recently in California, where the public utilities commission will soon be selecting a cost model for use in its universal service plan. After many months of workshops, hearings, and legal briefs, we have managed to narrow down the range of estimates for the size of the fund. The low estimate is zero; the high estimate is 1.7 Billion dollars. This experience suggests that it would be a mistake to think of these cost models as precision tools. Each party has proposed cost estimates influenced by its own interests, leaving the commission with a wide range of discretion in selecting the fund size it will find to be "sufficient". Further, as the Commission in California is beginning to appreciate, any change over time in cost levels, technology, or the definition of the basic service[65] would require new models to be developed, and the contentious regulatory process to be repeated. An auction approach offers remedies for the deficiencies of the cost approach. It would provide a mechanism for correcting errors in the initial cost estimates. It would require a firm wishing to influence the support amount to submit a bid, which would represent a binding offer to actually perform for that amount, rather than simply to hire a consultant to say what the incumbent's costs should be. I believe that this is a more promising approach for the regulator to induce firms to reveal their true cost expectations.
An auction would also remove from the regulator the discretion to set the support amount; the regulator would set the COLR requirements, and the firms, through the bidding process, would set the support. This would impose a certain discipline on the regulator, since it would know, when it set the requirements, that it would not be able to escape the funding need those requirements would imply. Finally, bidding would provide a means for the plan to adapt over time, without the need to update cost models.
Third, where firms have entered the market and are willing to bid to become carriers of last resort, a mechanism could be established to initiate bidding processes in those areas. The results would supersede the cost-based approach. In areas which were not nominated for bidding, the cost-based support would continue. I will discuss the design of an auction for this purpose below.
Fourth, the plan should raise the necessary funds through a minimally distorting mechanism. In principle, since universal service is a broad policy goal, there is a good argument for funding it from general revenues. As a practical matter, the next best approach is a sector-specific contribution mechanism which would be applied symmetrically to all telecommunication providers. I suggest a surcharge on all retail transactions for telecommunications services. This surcharge, like an excise tax, would avoid double-counting transactions at the wholesale level. By putting a specific charge on customer's bills, it would allow customers to see what they are contributing toward universal service. It would also avoid any asymmetry that might result from a "tax" that carriers would have to recover through their own service rates, when the incumbent's rates are still regulated, while entrants' rates are not.
Fifth, the funds received by each incumbent should be applied toward offsetting reductions in rates for other services which are contributing implicit support for universal service today. This would leave the incumbent with the same total revenue, but a price vector closer to a competitive market outcome.
There are several inconsistencies in the incumbents' rates that a universal service mechanism can help to address.
_ The first is the obvious one: the difference between the local rate the regulator finds "affordable" and the market rate for the service. The funding mechanism allows the subscriber to see the affordable rate, while the carrier sees something like a market rate -- the sum of the rate the customer pays and the support. This is necessary to allow correct price signals to prospective local carriers.
_ The second rate anomaly is the fact that some service prices may have been set at artificially high levels to support universal service; the offsetting reductions proposed here would help bring these rates closer to where they should be. By adjusting access rates downward, this approach would also help to align them more reasonably with other rates for interconnection to the LEC's network.
_ The third problem with rates is that of reconciling the rates for unbundled elements with the rate for the local service package. If the package rate is set below market, then it will not be possible to choose market-level rates for the unbundled elements which will sum to the package price. In this case, again, the universal service support provides the necessary degree of freedom. The COLR is the entity that buys inputs at market prices (whether the COLR "makes" these inputs with facilities of its own, or "buys" some of them as unbundled elements) and turns them into output in the form of the package at the socially desired price. The support payment is the element that makes this possible; it also assures that each carrier, whether it is a COLR or not, faces undistorted prices in its "make or buy" decision.