[2] It is important to emphasize that the most important definitional issue in the design of a universal service plan is the definition of the service to be supported. This is the function the Commission wishes the universal service provider to perform. For lack of a better term, GTE has referred to it as the carrier of last resort ("COLR") obligation. The price ceiling imposed on the "core" service, to which this question refers, is one of the requirements that may be imposed on a COLR. There is no distinction between "high cost support" and support for the COLR function; they are one and the same. In order for the new universal service plan to meet the requirements of the 1996 Act, the plan must be built around a clear logical framework, of which the COLR obligation is an integral part. This is true regardless of whether the support level is determined on a cost basis, or through competitive bidding. Thus, while GTE will discuss the definition of the COLR obligation infra in its answers to the Commission's questions on bidding, this framework is also relevant here as a basic underpinning of the universal service plan, whether GTE's auction proposal is adopted or not. For further discussion of the COLR obligation and how it should be applied, see also Attachment 2, which contains an excerpt from a paper by Dennis C. Weller, Chief Economist for GTE Telephone Operations, presented at Rutgers University Ninth Annual Western Conference, July, 1996.

[3] See GTE's D.96-45 Reply Comments filed May 7, 1996, at 20-23, describing the price "rebalancing," on a revenue neutral basis, that is needed to move the prices for many access and business services closer to economic costs and remove the hidden support these services provide for local service prices.

[4] See GTE's CC Docket No. 80-286 Comments filed October 28, 1994, at 22-23.

[5] See GTE's D.96-45 Comments filed April 12, 1996, at 2-3.

[6] A separate process is needed for schools and libraries than that used for rural health care providers because of the different requirements of the 1996 Act.

Compare [[section]]254(h)(1)(B) with [[section]]254(h)(1)(A). See n. infra.

⁷ See GTE's D.96-45 Comments filed April 12, 1996, at 19-21.

[8] Indeed, [[section]]254(h)(1)(B) limits the services that a telecommunications provider must make available to educational entities to those already available ("its services"). Without such caveat, situations might arise where facilities are not available to provide requested services, and either the FCC or states would then have to consider the nature and consequences of construction costs that would be required to provide such service, and how the provider would be compensated for those costs.

[9] Detariffing the Installation and Maintenance of Inside Wiring, CC Docket No. 79-105, 1 FCC Rcd 1190, 1192 (1986).

¹⁰ The institution would be free to choose its supplier, and the price, using any method it finds to be best. Thus, the price could be the tariffed rate of a LEC, or it could be the offer selected by the school after issuing an RFP and soliciting bids from different carriers.

[11] See GTE's D.96-45 Comments filed April 12, 1996, at n.37.

[12] It could also be useful for each state to appoint an administrative agency to assist rural health care entities in obtaining service under the provisions of [[section]]254(h)(1)(A). This agency would determine if a requesting entity were eligible under the 1996 Act and review requests for discounted network services to ensure they were bona fide, using criteria similar to those discussed supra.

Compensation for price reductions provided to rural health care entities requires a comparison of the difference between an urban price and a price for similar services offered to rural non-health care providers. [[section]]254(h)(1)(A) The respective regulatory agencies should establish a "range of reasonableness" applicable to all carriers that seek universal service funding for services provided to rural health care entities so as to limit the variance between urban and rural prices. Adherence to such a price range should be required for Federal funding eligibility. Moreover, if no such rural services are currently offered, the involved regulatory agency should solicit competitive bids for use in establishing a comparison point for support calculation.

[13] A public policy decision that establishes the total amount of support to provide on a nationwide basis will yield a specific level of discount available to eligible educational entities, if apportioned equally among all eligible entities. This approach will also harmonize the state and Federal discount methodologies.

[14] The Joint Board could rely on a panel of educators to suggest a fair method for apportioning these funds among states.

[15] To be considered a bona fide request by the state administrator, a school's plan would specify each of the components required to create an effective program. The network services provider would be pre-selected by the school on the basis of a bidding or similar process. The plan's budget would also show that all of the necessary non-network components (e.g., inside wiring, CPE, computers, educational application software and training in its use) are already present, or commitments for their funding have been obtained from sources other than the universal service fund. The plan would also authorize access to all documents that might be necessary to perform an audit of the use of funds.

[16] Under this approach, the network services provider would not be required to tariff an entirely new and separate set of "discounts" for eligible entities. Rather, the eligible entity would purchase services either at the "normal" rate, or as part of a package offering in response to the entities' request for bid. The difference between the price obtained and the funds provided by the administrator would yield the discount percentage.

[17] Unless the services were part of a previously negotiated package, a written request for tariffed services should clearly state the needed network services, including the desired installation dates, quantities of services by bandwidth, signaling protocols, interface requirements, points of origination and termination, relevant traffic load information, and other information needed to ensure the request can be fulfilled efficiently and expeditiously.

[18] [[section]]254(h)(1)(B) provides that discount levels will be established by the FCC for interstate services and by the states for intrastate services.

[19] See [[section]]254(b)(4), (b)(5); and (f), which says (in part) "A State may adopt regulations to provide for additional definitions and standards to preserve and advance universal service within that State only to the extent that such regulations adopt additional specific, predictable and sufficient mechanisms to support such definitions or standards that do not rely on or burden Federal universal support mechanisms.". An ILEC could, of course, voluntarily choose to continue such programs.

[20] This is not a complete measure of the compensation that should be provided to COLRs, because it does not include the value (positive or negative) of any non-price aspects of the COLR obligation. A bidding process would measure all aspects of the COLR obligation the bidders themselves believed to be relevant.

[21] WELLER TO PROVIDE CITE from CA PUC comments.

[22] The preface of the Telecommunications Act of 1996 clearly states that the intent of the bill is "to provide for a pro-competitive, de-regulatory national policy framework" by "opening all telecommunications markets to competition."

[23] See Conference Report at 16: "To the extent possible, the conferees intend that any support mechanisms continued or created under new section 254 should be explicit, rather than implicit as many support mechanisms are today."

²⁴ See GTE's Comments filed in CC Docket No. 96-115, September 27, 1995, at 2-6.

[25] The shortcomings of the Hatfield Model have previously been brought to the Commission's attention. See Ex Parte of GTE, CC Docket No. 96-98, July 11, 1996, providing an evaluation of the Hatfield Model conducted by Timothy J. Tardiff of NERA.

[26] This approach would also capture any differences in cost resulting from a change in the definition of "core" service over time, or changes in technology, or in input prices.

[27] See attached Statement of Paul R. Migrom. As Professor Milgrom explains, if the auction mechanism failed to allow the support to increase, it would create a sample selection problem as bidders would be drawn only to those areas where the errors in the cost estimates were positive.

[28] As described supra, the sum of Federal and state funding should cover this amount. The division of responsibility between the Federal fund and state funds could be established by choosing the desired level for the federal benchmark above which Federal support would be provided.

[29] In a recent ex parte presentation, Ameritech emphasized the need to define the COLR obligation carefully as a prerequisite for a successful universal service plan. GTE strongly agrees. Ameritech provided a framework for analyzing such obligations, in which it distinguished between two basic types of obligations into which firms might enter with the government. Unilateral obligations are imposed on firms without compensation; an example might be health and safety regulations. Bilateral commitments are entered into by the firm and the government on a voluntary basis, with an exchange of considerations on both sides; an example might be a contract to supply military vehicles to the Pentagon. Where unilateral obligations are necessary, the distortion they impose on the market can be minimized by applying them symmetrically to all firms in an industry; for example, all automobile manufacturers must meet the same safety standards. The challenge facing the Commission and state regulators is to transform a COLR obligation which today is unilateral and asymmetric (it applies only to the LEC) into a bilateral commitment (that is, one for which the COLR is compensated) which is symmetric (the same for all COLRs). See Ameritech ex parte of 31 July 1996.

[30] GTE's Comments in D.96-45, filed April 12, 1996 at 5 et seq. stresses that it would be a grave misinterpretation of the Act to assume that an Eltel that does not preserve and advance universal service will receive any funds at all.

[31] See Ameritech Ex parte at 6.

[32] There is no need for new Federal quality monitoring activities or performance-based measurements inasmuch as: (i) quality standards should be part of the obligations established by the state agency for receipt of support; (ii) state agencies already have a wide variety of service quality criteria and measurement mechanisms in place; and (iii) the ARMIS 43-05 report already provides service quality information to the FCC.

_

_

[33] The attraction of customer-specific support is that it would eliminate the need to average support in an area. This, in turn, would eliminate the need for rate intervention and a consequent COLR obligation, or for any intervention in the rate, since the support provided would allow each customer to pay enough to induce a carrier to provide service.

[34] See GTE's D.96-45 Comments filed April 12, 1996, at 14-16.

[35] Self certification should not be adopted due to the possibility for misuse of support funding.

[36] GTE suggests that the amount of the credit should be at least equal to the EUCL, and that it be linked to an inflation index so the passage of time does not dilute the effectiveness of the program.

[37] I note that the recent Telecommunications Act appears to be largely premised on the presumption that the benefits of promoting entry will usually outweigh the costs, but the extent of entry will still vary among service areas and the auction design needs to be cognizant of that.

[38] If an exclusive franchise is efficient but large sunk costs are not required, then there can be effective "competition for the market" each time the franchise is available for auction.

[39] An auction could conceivably be designed in which the winner receives a cash bonus but no advantage in the subsequent market competition. However, our analysis in section II implies that such a scheme is never optimal.

[40] In the PCS auctions, the market structure was determined primarily by restrictions on the amount of spectrum that individual licensees are permitted to control. These restrictions were the same for all areas of the country.

[41] The US Treasury uses a discriminatory auction to sell T-bills, but the individualized prices in that auction do not distort subsequent competition because the bids become sunk costs before the buyers engage in resale.

[42] The last difference is a matter to be solved primarily by pre-qualification of the bidders and by specifying that the support payments are made on a per subscriber basis rather than by lump sums (at least when there is competition in the market). It is not a matter to be resolved directly through the auction design.

[43] That is, the strategies are assumed to form a Nash equilibrium of the auction game.

[44] This is not an assumption I make happily. I make it because it makes the analysis tractable and leads to intuitively sensible results. Also, the auction obtained from the analysis has at least some robustness: identical recommendations are obtained when the ratio of fixed to variable costs are the same across bidders.

[45] This assumption sets aside the question of reserves, i.e., maximum opening bids. As we shall see later, the franchises offered for auction are determined by a nomination process with a workable reserve determined as part of that process.

[46] Myerson, Roger, "Optimal Auction Design," Mathematics of Operations Research 6 (1981): 58-73.

[47] More exactly, the distortion is created by the surcharge or tax used to finance the subsidy.

[48] That is, strategies incorporating this behavior may comprise a Nash equilibrium.

[49] If the shares are not equal, the relevant comparison is between the cost of the n^th bidder and the weighted average cost of the n-1 lower cost bidders, weighted according to the number of customers taken from each bidder by the n^th bidder.

[50] Another rule would specify that the support payment is the level of the highest accepted bid multiplied by 1.15 in case there are two winners and by 1.3 in case there are three or more winners. Again, the percentages are arbitrary and intended for illustrative purposes only. What is illustrated is that the payments can be made to depend on the number of COLRs selected.

[51] In the paging, PCS, and SMR auctions, besides any cost synergies, there were important additional synergies from demand side effects. Buyers of PCS services, for example, find the service more valuable when the phone works over a wider geographic area. In contrast, there appear to be no important demand side synergies in meeting universal service obligations.

[52] A multiple greater than 100% of the estimated cost should be used, with the extent of the mark-up dependent on the amount of error in the cost estimates. The mark-up is needed to compensate for "selection bias": auctions will be most likely to be conducted for those areas where the model overestimates the costs and will be least likely where the model underestimates the costs. Consequently, a simple 100% rule would leave the LEC receiving the model cost estimate most often when the model most underestimates the actual cost. A reasonable allowance for upward movement also needs to be made when an area is reauctioned to allow for changes that may increase costs over time, such as a change in the definition of the "core" service.

[53] If the LEC believes that the official rate is too low, it may seek a higher rate from the FCC or state PUC. Of course, the higher rate may encourage other potential COLRs to petition for an auction of some or all of the LEC's COLR service areas.

[54] In general, if an area is sufficiently homogeneous, the COLR will find this kind of discrimination unprofitable because (1) even a subscriber that is more expensive to serve than the average subscriber may make a positive contribution to covering the system's fixed costs and (2) when the heterogeneity is not too great, the cost of discriminating between relatively high- and low-cost subscribers may exceed the profit from successful discrimination.

[55] If the reserve is known to the bidders to be very high, there is a Nash equilibrium in which the bidders each bid zero and receive the reserve as their subsidy. This outcome leads to the same kinds of losses that we identified earlier for other forms of collusive behavior.

[56] As I have already explained, the reserve needs to be based on a multiple of the estimated cost in order to allow the auction to correct errors - both overestimates and underestimates - in the cost estimates and to mitigate the "selection bias" that would be otherwise created.

[57] Any other rule would allow a non-COLR to affect the support price in an area merely by nominating a CBG for auction and without actually bidding, possibly encouraging mischievous nominations.

[58] For example, several of the proposals in the US are designed around a standard unit reported by the US Census Bureau, which is called a census block group, or CBG. The boundaries of each CBG are chosen to encompass from 200 to 500 households. Except in the most rural areas, these units are usually small enough to capture differences in cost between town centers and outlying areas. Another model, developed by Pacific Bell, has been built around "grid squares", each of which is 1/100th of a degree, or about 3,00 feet, on a side. Because these "grid squares" are constant in size, they may provide improved precision in rural areas. In the state of California, for example, about one third of the grid squares have no households at all.

[59] Note that, since the cost of serving each customer is not independent, neither would be the amount of the required side payment.

[60] It is of course quite possible that all customers in an area may be attractive to serve without support; the market price may be less than the "affordable" price set by the regulator.

[61] The probability that an efficient carrier would be deterred would depend on the size of the market intervention the regulator decides to impose, and hence on the magnitude of the necessary support.

[62] Note that the use of small areas, besides limiting heterogeneity, also minimizes the size of the obligation an entrant must take on to qualify for support.

[63] I have used the term "support" to describe the side payment, because it is a usefully bland term without another economic meaning. I have avoided calling it a "subsidy" because it has nothing to do with whether the basic local service is being subsidized in the economic sense. If economies of scale and scope are large, the range of rates which are subsidy-free might be quite large. This does not mean that the firm would be indifferent with respect to where in the range the rate is set. Conversely, the rate the regulator chooses may be within the subsidy-free range, or it might not. It really doesn't matter. The issue is whether or not there has been a market intervention, and what payment is needed to compensate for that intervention -- not whether a subsidy exists in the economic sense.

[64] The 1996 revision to the Telecommunications Act establishes a link between an obligation to serve and the receipt of universal service funding. It defines an "eligible" carrier ("Eltel") as one that commits to serve a given area, and to advertise its price. I believe that the Act gives ample scope for state and federal regulators to flesh out the details of these requirements, just as it leaves other implementation specifics to those regulators. In any event, an obligation to serve would mean little if the price were not specified; if the carrier were free to vary its price, it could still target only the customers it wished to serve, while ostensibly standing ready to serve everyone.

[65] The 1996 telecommunications Act requires the FCC to review the basic service definition periodically to determine if newer or more advanced functions should be added.