COMMENTS ON UNIVERSAL SERVICE

Federal-State Joint Board on Universal Service
CC Docket No. 96-45

by Kenneth Gordon and William E. Taylor

April 12, 1996

White Plains, NY Washington, DC / Los Angeles, CA / Cambridge. MA / Philadelphia. PA/ San Francisco. CA/ New York, NY/ Ithaca, NY/ Seattle, WA/ London/ Madrid
A MARSH & McLENNAN COMPANY

TABLE OF CONTENTS

I.   INTRODUCTION AND SUMMARY 

II.  BASIC ECONOMIC PRINCIPLES
     A. REPLACE IMPLICIT BY EXPLICIT SUPPORT AND MAKE ALL SUPPORT 
        COMPETITIVELY NEUTRAL
     B. RESTRICT SUPPORT ONLY TO ESSENTIL SERVICES THAT ARE IN THE PUBLIC 
        INTEREST AND CONSISTENT WITH COMPETITION AND ECONOMIC EFFICIENCY

III. APPLYING THE PRINCIPLES TO INTERSTATE ACCESS CHARGES
     A. THE BASIC PROBLEM
     B. REBALANCING AND REPLACING IMPLICIT WITH EXPLICIT SUPPORT
     C. BASING THE INITIAL LEVEL OF SUPPORT ON INCUMBENT's EMBEDDED COST

IV.  HIGHER FLAT RATE CHARGES FOR SUBSCRIPTION WILL NOT RETARD UNIVERSAL
     SERVICE 
     A. AS FLAT RATES FOR SERVICE ROSE (As RATES WERE REBALANCED), 
        SUBSCRLBERSHIP ROSE IN ALL INCOME SEGMENTS
     B. SUBSCRIPTION GROWTH IS ATTRIBUTABLE TO SMALL GROWTH IN REAL 
        SUBSCRIBER ACCESS PRICES, DECLINE IN OTHER PRICES, INCOME GROWTH, AND 
        OTHER FACTORS
     C. CONTINUED RATE REBALANCING WILL NOT HARM UNIVERSAL SERVICE

V.   ECONOMIC PRINCIPLES TO DEFINE/ANALYZE SERVICE ESSENTIALITY AND
     AFFORDABILITY
     A. ESSENTIALITY
     B. AFFORDABLE BASIC LOCAL RATES

VI.  OTHER ISSUES
     A. COMPETITIVE BIDDING FOR UNIVERSAL SERVICE SUPPORT
     B. PROXY COST MODELS

VII. CONCLUSION

COMMENTS ON UNIVERSAL SERVICE[1]

1. INTRODUCTION AND SUMMARY This report analyzes the economic issues associated with the Federal Communications Commission's (FCC's) Notice of Proposed Rulemaking (NPRM) on universal service. [2] We agree with the Commission that the Telecommunications Act of 1996 (the Act) requires replacement of the current implicit subsidies with "explicit," "sufficient" and "competitively neutral" universal service funding by all carriers. [3] We also agree that past FCC rebalancing efforts have brought substantial efficiency gains without harming universal service.4 Based on this experience and on fundamental economic efficiency principles, we recommend that the FCC and the Joint Board resume the rebalancing efforts begun in the mid-1980s. Furthermore, we conclude that additional rate rebalancing (i.e., increasing flat-rate charges and reducing usage-sensitive charges) together with targeted subsidies to reduce internal subsidies would not harm universal service. Such a shift would be entirely consistent with "a fundamental underlying principle of the 1996 Act ... the Congressional desire 'to provide for a pro- competitive and de-regulatory national policy framework...’’’[5] Such rate rebalancing would also be consistent with the observation in the NPRM that "the Act specifically provides that

____________________

1. The authors are Senior Vice Presidents of National Economic Research Associates, Inc. Dr. Gordon served as chairman of the Public Service Commissions of Maine and Massachusetts and at the Federal Communications Commission (FCC). Dr. Taylor has studied economic problems of telecommunications regulation at Bell Laboratories, Bellcore and NERA, filing testimony before the FCC, the Canadian Radio-Television and Telecommunications Commission and many state regulatory commissions.

2 Federal-State Joint Board on Universal Service, Notice of Proposed Rulemaking and Order Establishing Joint Board, CC Docket No. 96-45 (released March 8, 1996).

3. Id., at para. 113 citing the Act at sec. 101 (a), [[section]]254(d)-(e). Furthermore, according to [[section]]254(b)(4)-(5), there "should be ... sufficient Federal and State mechanisms to preserve and advance universal service."

4 Id., para. 113.

5 Id., para. 18.

telecommunications services - not just the narrow categorv of telephone exchange service - be affordable."[6] To the extent that rate increases for one service are offset by decreases for others, "telecommunications services" would be no less affordable after rates are rebalanced .[7]

We also recommend that the FCC deter-mine the initial level of universal service support based on the incumbent LECs' embedded cost. We show that this policv combined with "portable" universal service funding would be consistent with competitive, cost-based pricing. Thus, our proposal is consistent with competitive markets and would ensure that prices move towards competitive levels without the added cost and complexity of an externally- administered bidding process. Finally, we show that idealized proxy cost models produce, at best, hypothetical estimates that provide only a starting point to calculate actual network costs. Such estimates are inferior to incremental costs derived based on actual network configurations.

11. BASIC ECONOMIC PRINCIPLES

A. Replace Implicit by Explicit Support and Make All Support Competitively Neutral

The first economic principle is that explicit support should replace the internal subsidies that currently support universal service. This principle is consistent with the Act's requirement that "[t]here should be specific, predictable and sufficient Federal and State mechanisms to preserve and advance universal service."[8] Implicit subsidies (of the kind that are currently built into certain service prices) mis-state the actual relationship among prices and costs. They cause

____________________

6 Id., para. 14.

7 For example, increases in local exchange rates offset by decreases in toll prices could cause the average subscriber bill to fall, so that "telecommunications services" were more affordable even though local exchange prices increased.

8 The same point is repeated in the Act, [[section]]254(d) on telecommunications carrier contribution: "Every telecommunications carrier that provides interstate telecommunications services shall contribute, on an equitable and nondiscriminatory basis, to the specific, predictable, and sufficient mechanisms established by the Commission to preserve and advance universal service."

the wrong signals to be sent to the price system (on which efficient resource allocation in a market economy depends) and, as a consequence, distort the entry, exit, and expansion decisions of market participants. In contrast, explicit support (coupled with implementation of the principles discussed below) will foster more efficient relationships among prices and costs and send the proper signals to guide the decision-making of those participants.

The second economic principle is that universal service support should be recovered in a competitively neutral fashion from all telecommunications carriers so that the determination of the carrier that actually provides universal service would not depend on whether it was also the source of support. By de-linking the source of support from the provider of service, any impediment to the provision of service by only the most efficient providers would be removed. This principle is also recognized by the Act, which states that "All providers of telecommunications services should make an equitable and nondiscriminatory contribution to the preservation and advancement of universal service."[9]

In essence, these two principles imply that current universal service funding by internal and other implicit subsidies built into the carrier access rates charged by local exchange carriers (LECS) should be completely replaced by competitively neutral forms of support. If this is not done, then the universal service funding mechanism would not be "sufficient... to preserve and advance universal service. " [10]

Competitively-neutral support could be likened to the use of broad-based taxes in the wider economy for raising a certain amount of support. A general or broad-based sales tax does not distort relative price-cost relationships and, therefore, minimizes any distortion of

____________________

9 The Act, [[section]]254(b)(4) (emphasis added).

10 The-Act's [[section]]254(e)-(f) also support such reform at the state level by making intrastate universal service support explicit, competitively neutral and assessed on an equitable and non- discriminatory basis to all providers of intrastate telecommunications services, and not a burden on the Federal universal service support mechanisms.

consumption (or reduction of social or consumer welfare). [11] Thus, a broad-based method of paying for universal service would be more efficient than the present system that takes contribution from only certain services or service providers.

The economic efficiency of a broad-based method of payment for universal service could be further improved by taking account of the price elasticity of demand for various services. Usage services tend to be relatively more price-elastic than subscriber access services (for which costs tend not to vary with usage). Because of this, efficiency could be increased by recovering as much of the non-traffic-sensitive (NTS) costs as possible directly from final customers or "end-users" on a flat-rate basis. Usage services should contribute to universal services as little as possible, with the most price-elastic such services contributing the least. This approach would allow usage service prices to remain close to economic costs and thereby allow information age services to grow to their full potential.

B. Restrict Support Only to Essential Services that are in the Public Interest and Consistent with Competition and Economic Efficiency

Economic efficiency also requires that the subsidies be minimized in order that prices be distorted as little as possible. Even competitively neutral subsidies distort prices and choices among goods and services. Thus, the range of goods and services eligible for support should be chosen carefully and any necessary subsidies should be kept to a minimum. Such an approach is consistent with Congress' desire to provide for a procompetitive and de-regulatory national policy framework. It is also consistent with the Act's mandate to the Joint Board and the FCC

____________________

11 While such a tax would reduce purchasing power and, hence, likely curtail consumption, there would be none of the inefficient substitution of an untaxed good for a taxed good that could occur when the consumption of only some goods were taxed but not of others. In other words, while broad-based taxes impose only income effects on consumption, specific taxes - by distorting relative price-cost relationships - produce in addition more substantive substitution effects on consumption.

to consider "such other principles as [they] determine are necessary and appropriate for the protection of the public interest, convenience, and necessity and are consistent with this Act." [12]

In sum, to serve the public interest. regulators should limit the overall level of support and shift to more efficient forms of support, targeted more narrowly at low-income consumers and high-cost areas. As explained in Section IV, it is possible to design targeted support that would be sufficient to preserve universal service without conflicting with the public interest in promoting competition and economic efficiency.

Section 254 of the Act lists the services that are to be considered for universal service support and, in the process, reveals the tension between the desire to expand the concept of universal service and the economic efficiency, competitive and deregulatory goals of the Act. According to this section of the Act:

(c) DEFINITION- (1) IN GENERAL-....The Joint Board ... and the Commission in establishing, the definition of ... universal service ... shall consider the extent to which such telecommunications services- (A) are essential to education, public health, or public safety; (B) have, through the operation of market choices by customers, been subscribed to bv a substantial majority of residential customers; (C) are being deploved in public telecommunications networks by telecommunications carriers; and (D) are consistent with the public interest, convenience, and necessity.[13]

The operation of market choices in subsection (B) is indeed the key to efficiency; however, for those choices to be meaningful, they must be made on the basis of market prices. When certain services are offered at subsidized rates, the "operation of market choices" cannot be truly observed because if market forces set higher, cost-based rates, "the substantial majority of residential customers" might not subscribe to those services. Similarly, when other services are priced well above cost to provide universal service support, it is equally hard to observe true "market choices" because at lower, cost-based rates, those supporting services would be

____________________

12 The Act, [[section]]254(b)(7).

13 Id., [[section]]254(c)(1).

consumed in quantities greater than those observed. Subsection (B) would also have to be tempered by subsection (D) which requires consideration of the "public interest." Because the concept of public interest includes consideration of economic efficiency, excessive amounts of support provided without regard to efficiency could prevent deployment and consumption of new services and, thus, harm the public interest. As we show in Section V, however, these tensions can be resolved by (i) narrowly specifying the services to be supported and (ii) demonstrating that only "essential" services (those that possess additional social value not evident from market prices alone) be eligible for such support.

III. APPLYING THE PRINCIPLES TO INTERSTATE ACCESS CHARGES

A. The Basic Problem Applying the economic principles discussed above to derive an "interstate" universal service fund in the context of the Act, requires (i) identifying the services to be supported, (ii) calculating the amount of support needed by those services; (iii) determining the fraction of the support that should be recovered from interstate services, and (iv) determining the means for raising that support.

The NPRM identifies five categories of (essentially intrastate) services as being eligible for universal service support. These include voice grade access service, touch-tone, single part,, service, access to emergency services (91 1 and enhanced 91 1), and access to operator services. [14]

As explained in greater detail in Section C below, the level of support per residential line should be based initially on the difference between the incumbent LEC's per-line embedded cost and the rate that is set for the local residential services that fall within the aegis of universal service. Under current pricing and jurisdictional cost allocations. BellSouth's

____________________

14 NPPM, para. 9.

current conservative estimate of its annual cost[15] of providing universal service is about $4.8 billion, of which it recovers only about $2.8 billion from basic local residential service and interstate subscriber line charge (SLC) revenues. Thus, the support based on these data is almost $2.0 billion per year. BellSouth currently receives roughly half of this support (about $ 1.0 billion) from three specific interstate revenue streams: the carrier common line (CCL) charge, the interconnection charge ("RIC"), and the Federal Universal Service Fund. According to the NPRM:

(t)he current CCL charge appears to be inconsistent with the directives of the 1996 Act that universal service support flows 'be explicit' and be recovered on a ' nondiscriminatory basis' from all telecommunications carriers providing interstate telecommunications service.[16]

Thus, for BellSouth alone, at least $1.0 billion of support (currently raised through implicit subsidies) needs now to be recovered explicitly on an equitable, nondiscriminatory, and competitively neutral basis.[17]

The current system of interstate access charges has benefited from substantial reforms undertaken by the FCC, e.g., the phasing-in of a $3.50 SLC per residential line and changes in the local transport rate structure. However. the entire system is still based on an artificial allocation (of roughly 25 percent) of the LECs' NTS revenue requirement to interstate services. Only by pure coincidence would prices based on such arbitrary allocations resemble those based on proper economic principles. Also. because the FCC and state regulators each address only the portion of the revenue requirement over which thev have jurisdiction, it cannot be assumed that the correct application of economic principles in any one jurisdiction would lead

____________________

15 Comments of BellSouth Telecommunications, in re Federal-State Joint Board on Universal Service, CC Docket No. 96-45.

16 NPRM, I I 13.

17 The $1.0 billion of annual interstate support to Universal Service is conservative because (except for the interconnection charge) it does not take into account contribution (price less incremental cost) recovered from other carrier access services, including local switching and transport.

automatically to a correct overall result. As discussed above, any revenue shortfall for universal service should be recovered through a competitively neutral charge that can be imposed on all providers of telecommunications services. In principle, since jurisdictional distinctions have no economic meaning, this charge should be based on total interstate and intrastate revenues. However, since the bulk of the interstate revenues that the LECs currently receive are designed to support basic local residential service, these revenues should be excluded from those subject to the universal service surcharge. [18]

B. Rebalancing and Replacing Implicit with Explicit Support How should the principles of the Act be applied to replace the implicit interstate sources of support that were identified above? First, we should minimize the necessary support. That is, we should resume the FCC policies of the mid-1980s to move rates closer to costs. The rate structure should recover as much of the NTS costs as possible from the services that cause those costs, i.e., from basic access service, by means of flat NTS prices such as SLCS. Economic efficiency requires recovering costs of all services - including universal services from the causers of those costs to the greatest extent possible. Hence. NTS costs should be recovered from end-users on a flat-rate basis to the greatest extent possible.

Second, the remaining interstate revenue requirement - that not recovered through SLCs - could be recovered indirectly from those who use access lines to originate and terminate interstate calls by assessing a competitively neutral charge to all providers of interstate services. More specifically, we recommend that contributions be based on the end- user interstate revenues of those service providers. This allocation would avoid (i) the pricing distortions caused by per-minute or usage-based charges, (ii) the need to come up with arbitrary "'equivalency ratios' for calculating contributions owed by providers ... that were not

____________________

18 Ultimately, if it is possible to solve the jurisdictional issues, the Joint Board should develop the correct approach to collect the needed contribution when that amount has been determined and then apply that approach to overall telecommunications revenues. However, this cannot be done without coordinated state and FCC action to ensure recovery of the entire shortfall.

sold on a per-line or per-minute basis ... ," [19] and (iii) the need to determine how much an inter- exchange carrier (IXC) should contribute per line. Using revenues from sales of interstate services to end-users would also avoid unnecessarily taxing providers of universal service who are the intended recipients of the support.

C. Basing the Initial Level of Support on Incumbent's Embedded Cost The NPRM seeks comment on the methodology for "... determining the level of support required to assure that carriers are financially able to provide ... universal service ... in rural, insular, and high-cost areas."[20] It asks that the methodology proposed must be (i) simple to administer, (ii) technology-neutral, (iii) designed to produce the least amount of support needed to achieve the Act's universal service goals, (iv) equitable and nondiscriminatory in the burden it imposes on competitors, and (v) based on distribution procedures that are "direct, explicit, and specific."[21] More specifically, if the determination of the support should be based on some measure of the carrier's cost of providing universal service, the NPRM asks whether embedded costs or forward-looking costs should be used for that purpose .[22]

Once the appropriate basic local service rate ("basic rate" hereafter) has been determined,[23] the amount of support provided per end-user served (or, more precisely, per end- user line) should be initially set as the difference between the incumbent universal service provider's embedded cost per line and the basic rate. We emphasize the fact that this difference between the incumbent carrier's embedded cost and the basic rate be used to set only the initial

____________________

19 NPRM, para. 124.

20 Id., para. 27.

21 Id.

22 Id., para. 32.

23 Whether a single rate applies state-wide or the rate is deaveraged and allowed to vary by serving areas such as wire centers is not the issue here. Nor does this method depend on knowing whether support is limited to particular categories of users (e.g., residential and/or single-line business). All that matters is that support be provided, as the Act requires, to users in rural, insular, and high-cost areas, and that the support be based on both the cost of providing service and the rate set for it.

level of support because, as the example below will show, the level of support needed will subsequently come to depend on the forward-looking costs of competing universal service providers once the market mechanism takes over. In other words, our metmodology prescribes only the manner for setting the initial level of support but relies on the market mechanism for subsequent changes in that level.

The market mechanism that we envision would eventually determine the level of support has three components, (i) competition among eligible telecommunications carriers (ETCs), (ii) full portability of the support, [24] i.e., availability of the support to any ETC that actually provides basic local service to a user, and (iii) price signals. The market environment in which such a mechanism would operate would likely include one or more ETCs in a given serving area, each willing to provide basic service at or below the set service rate.

In this environment, competition among ETCs would be necessary to ensure that end- users are served only by the lowest-cost service provider. In turn, this would ensure that society's scarce resources needed to provide service and the support needed to sustain below- cost pricing of that service are both minimized. Effective competition could be introduced in each serving area by making the support fully portable, a departure from the present state in which the incumbent carrier retains the support regardless of the abilities of alternative carriers to provide basic service at lower cost. That competition, moreover, would be truly dynamic if ETCs providing service at any given time were to be assured of continuing as incumbents only as long as they could match or beat the prices offered for basic service by their competitors. If price signals worked as envisioned in this market, a competitor that could charge a lower rate for service than the incumbent while receiving the same amount of support, or could match the

____________________

24 We envision a universal service fund that has three components: (i) the difference between embedded costs and rates, (ii) an amortization of current depreciation reserve deficiencies, and (iii) the cost of Lifeline and Linkup programs. While the first and third components should be fully portable, recovery of any reserve deficiency amortization should be confined to the firm that incurred the deficiency.

incumbent's rate while requiring a lower level of support, would successfully win the right to serve the end-user.

The following hypothetical example illustrates how an initial level of support based on the incumbent carrier's embedded cost would eventually result in least-cost provision of basic service. First, assume that the incumbent carrier's embedded cost of providing basic service in a given serving area is $25 per line per month. Second, assume that the retail rate for basic service in that area is $15 per line per month. Finally, assume that the incumbent's forward- looking long run incremental cost (LRIC) of providing basic service in that area is $20 per line per month. [25] The last assumption implies that the incumbent would save $20 per line per month when relieved of the burden of providing basic service or, alternatively, incur $20 per line per month when providing such service to an additional end-user. Under the proposed methodology, the initial level of support would be set, in this example, at $1 0 per line per month, i.e., the difference between the $25 embedded cost and the $15 basic rate.

To see what effects competition based on a fully portable universal service support can have in this hypothetical market, consider three alternative scenarios.

Scenario A: Potential competitors all have LRICs that exceed the incumbent's embedded cost of $25 per line per month.

Scenario B: At least one competitor has a LRIC that is higher than the incumbent's LRIC of $20 per line per month but below the incumbent's $25 embedded cost.

Scenario C: At least one competitor has a LRIC that is below the incumbent's $20 LRIC.

Scenario A: In this scenario. none of the potential competitors could expect to win the opportunity to provide universal service in place of the incumbent. The incumbent ETC would likely fullv recover its embedded cost of providing universal service, and the size of the support

____________________

25 In this discussion, we assume for simplicity that a competitive provider of basic service would be willing to set the price for that service (plus the support) equal to its LRIC. The LRIC, however, is only a price floor. In reality, even under competition, a multiproduct firm would generally set the price (including support) to equal LRIC plus a reasonable contribution toward its shared and common costs.

would change only to the extent that the basic rate or the incumbent's embedded cost changed in the future. The $15 basic rate - presumably deemed to be affordable - would continue in place and the incumbent, by virtue of being the least-cost provider, would require the least amount of support from any universal service fund.

Scenario B: If at least one competitor has a LRIC below the incumbent's embedded cost, the market mechanism would use portability and price signals to determine which of the two ETCs should serve the end-user. For sake of illustration, suppose that the competitor's LRIC is $22 per line per month, i.e., $3 below the incumbent's embedded cost but $2 above the incumbent's LRIC. Given its cost and portability of the support, the competitor could afford to offer basic service at a rate of $12 per line per month, i.e., the $22 LRIC less the SI 0 support. This rate would be $3 below the basic rate charged thus far by the incumbent. While the competitor would seemingly be assured recovery of its LRIC, it would, however. not be assured of the opportunity to win end-users away from the incumbent. This is because the incumbent would still have the incentive to lower its own rate in order to compete for the end-users. Although the incumbent would lose net revenue by lowering its own rate to $12 or below (to match or undercut the competitor's rate), that loss would be less than that which could occur from holding its rate at $15 and losing end-users in the process. By lowering the rate to $12 in order to keep the end-user, the incumbent would sacrifice $3 in net revenue. By losing the end- user altogether, the incumbent would sacrifice $5 in contribution to the excess of its embedded cost over its LRIC. [26] Thus, by choosing the former course of action, the incumbent could retain the end-user and still earn a $2 per line contribution toward the excess of its embedded cost over its LRIC (i.e., $22 in total revenue including support less the $20 LRIC).

In this scenario, the level of support toward the incumbent's embedded cost would decline and the attendant losses - if not offset by cost reductions - would be absorbed by the

____________________

26 By losing the end-user, the incumbent would "save" its LRIC, namely, $20. It would also sacrifice $25 in total revenue, namely, the S 15 rate plus S 10 in support. Therefore, it would make a net loss of $5 per line.

incumbent's shareholders. The rate for basic service, however, would fall below the level initially deemed affordable and necessary to support universal service objectives. Most importantly, this decline in the rate would occur ultimately not in response to the incumbent's embedded cost but to the potential competitor's LRIC.

The observed decline in the market price for local exchange service would signal to the regulator the fact that universal service support could be reduced. An alternative - but financially equivalent - result in this scenario could be that the basic rate would remain at the $15 level considered to be affordable but the incumbent would now only require (and receive) support per line of $7, i.e., $3 below the level initially determined. Either way whether the rate fell to $12 at a $ 10 level of support or the rate stayed at $15 but the support was reduced to $7 - the threat of competition from another carrier with a LRIC below the incumbent's embedded cost would have a salutary effect on the universal service system. Once again, the least-cost provider would serve the market and, in the process, minimize the amount of support needed. Even though the initial level of support was set using the incumbent firm's embedded costs, the driving force behind the reduction in support would be the presence of a low-cost competitor, not a measured reduction in the incumbent's embedded costs.

Scenario C: When at least one competitor's LRIC is below the incumbent's LRIC, that competitor would likely begin to displace the incumbent as the provider of universal service for manv customers. Again, for illustration, assume the competitor's LRIC is S 18 per line per month. i.e., $2 below the incumbent's LRIC. In this case, the competitor could afford to lower its rate to $8 per line and still recover its LRIC ($8 rate plus $10 in support). In contrast, the incumbent would fail to match the competitor' s rate because doing so would cause even larger losses of net revenue than those from simply losing the end-user. By giving up the end-user, the incumbent would only lose the $5 contribution toward the excess of its embedded cost over its LRIC. But by trying to match the competitor’s $8 rate, the incumbent would lose an additional $2. [27]

In this scenario, the basic rate could fall to $8 per line at a $10 per line level of support. Equivalently, the rate could stay at the affordable level of $15 while the needed amount of support could shrink to $3 per line. Either way, the least-cost provider would win out and competition would deliver the economically efficient outcome. The amount of support would be minimized and the driving force behind the competitive process would once again have been the forward-looking LRIC of the competitor, even though the initial level of support was set in relation to the incumbent’s embedded cost.

The three scenarios in this hypothetical example illustrate how the market mechanism could be relied upon to convey the proper price signals and for the market to be served only by the least-cost ETC. The example also demonstrates how the level of support could be adjusted down over time (at a constant basic rate) as more efficient competitors tried to enter the market or, alternatively, how the basic rate itself could be lowered over time (at a constant level of support). [28] In reality, costs, and possibly rates, could vary among end-users or serving areas. Moreover, the quality of service and the end-user’s taste for various services could also vary. Consequently, there could be competition for certain end-users or in certain serving areas, but not others. As a result of such competition, the incumbent could retain certain end-users or serving areas but lose others. The hypothetical example, nevertheless, conveys the efficacy of setting the initial level of support in relation to the incumbent’s embedded cost and then relying upon competition and the market mechanism for subsequent fine-tuning.

27 By matching the competitor’s $8 rate, the incumbent could expect to earn only the $8 plus $10 in support, i.e., $18 per line. That would be $2 below its LRIC and $7 below its embedded cost.

28 Reductions in the level of support could be accomplished in a number of ways. One possibility is for regulators to periodically lower the level of support by the amount of short-term rents that the most efficient entrant could earn by displacing the incumbent, i.e., by the sustainable difference between the incumbent’s LRIC and the entrant’s (lower) LRIC. Another possibilty is for the support to be capped and adjusted by some measure of inflation along with a built-in offset factor (akin to a productivity offset in price cap regulation).

The methodology proposed here may now be evaluated in terms of the five requirements placed by the NPRM.

Administrative simplicity: The administratively simplest methodology should minimize the burden on regulators on three fronts: (i) determining the initial level of support, (ii) adjusting the support or the basic rate over time, and (iii) determining the least-cost ETC for a given serving area. The methodology proposed here succeeds on all three fronts. The initial level of support can be easily determined from two readily available figures - the basic rate which is set on the basis of an appropriate affordability criterion and the incumbent's embedded cost which can be readily inferred from information in the carrier's own books of account and from publicly available ARMIS data.[29] Adjustments to the level of support and/or the basic rate would be driven by potential or actual entry by equally or more efficient competitors and the promise of full portability of the support. Price signals emitted by competitors would reflect their underlying forward-looking costs of providing service and would eventually ensure service by the least-cost provider. Regulators would have to neither make arbitrary picks among competing ETCs nor conduct expensive and unnecessary competitive bids[30] in order to determine the providers of universal service.

Technological neutrality: The proposed method of determining the level of support would encourage more efficient carriers to come forward and compete. Because the method tilts the likelihood of serving the market in favor of neither the incumbent nor its competitors, it would not influence the technology choices of the competitors. By assuring that the least-cost ETC would have the opportunity to serve the market. the proposed method would encourage potential competitors to seek out technologies that do not necessarily imitate that of the

____________________

29 Automated Reporting Management Information System (APMIS) reports containing actually incurred cost data are filed annually with the Federal Communications Commission by the larger local exchange carriers.

30 We examine this issue in detail below.

incumbent. Instead, competitors would have the incentive to minimize the costs of providing service at a certain acceptable level of service quality and innovation.

Minimize needed support: The hypothetical example showed how, at a given level of the basic rate, the amount of support needed would be set initially at the minimum level and reduced over time by competitive forces.

Equitable and non-discriminatory burden: The proposed method would provide equal incentives to all competitors (incumbent and entrants alike) to compete for the opportunity to serve the market. No category of competitor would be unduly burdened by, e.g., having to recover certain costs that its competitors would not have to, or being denied the universal service support that its competitors would be entitled to.

Direct, explicit, and specific distribution: While the method of distributing the support is, in principle, a separate issue than the method of determining the appropriate level of support, the methodology proposed here makes it easy to size any explicit universal service fund and to distribute from it only the amounts per line that an ETC serving a particular serving area would be entitled to at any given time. The proposed method simplifies the process of determining how much support is needed and who should receive it when both demand and cost conditions van, across end-users and serving areas.

IV. HIGHER FLAT RATE CHARGES FOR SUBSCRIPTION WILL NOT RETARD UNIVERSAL SERVICE.

A. As Flat Rates for Service Rose (as Rates Were Rebalanced), Subscribership Rose in All Income Segments

The impact of telephone rate changes was a source of substantial controversy as the divestiture of AT&T was being implemented. At that time, some consumer advocates made dire predictions regarding the impact of higher residence rates on subscription to telephone service. 3 1 The actual results since that time show that the FCC's partial rebalancing of rates through the phasing-in of SLCs and the reduction in toll rates beginning in 1985 and ending in 1989 did not harm telephone subscription. Thus, despite a rise in monthly flat rates from about $13.35 in 1984 to $17.53 in 1989 (largely attributable to the FCC increasing the SLC to $3.50 and tax increases totaling $0.45), 32 telephone penetration rose from about 91.4 percent in November 1983 to about 93.0 percent in November 1989. [33] Viewed another way, over this period, the percentage of households not subscribing to telephone service declined from 8.6 to 7.0. This 19 percent reduction in the percentage of households without service represents substantial progress during the very period in which rates were rebalanced by the FCC. [34]

Further progress has occurred since 1989. By November 1994, the percentage of subscribing households had risen to 93.8, while the percentage without service had declined to 6.2 percent or by another 9 percent, bringing the cumulative decline in households without service to about 28 percent since divestiture. By October 1994, flat rates (not including touch- tone) had increased by another $1.47 to $19.00 per month. Substantial subscription gains were also made by households below the poverty level. We estimate that 77 percent of households below the povertv level subscribed to telephone

____________________

31 L.J. Perl and W.E. Taylor. "Telephone Penetration and Universal Service in the 1980s," in B. Cole (ed.), Divestiture Five Years Later, New York: Columbia University Press, 1989.

32 Residential unlimited rates, excluding these increases rose from $12.10 to $12.30 in the same time period. FCC Statistics of Communications Common Carriers, 1994-95, Table 8.4, at 340.

33 These penetration figures are based on the Census Bureau Current Population Survey (CPS). The CPS estimates are similar to but somewhat lower than those based on data gathered during the decennial census. The CPS figures appear to be lower because of minor differences in the questions asked in the two surveys and in sampling techniques. According to the FCC, the actual figure may be between the two. See the FCC's Telephone Subscribership in the United Slates (by Alexander Bellinfante), April 1995, at 2.

34 Note that progress was also made according to the data reported in the decennial census of households. Using those data, which use a slightlv different measure of telephone penetration, subscription rose from 92.9 percent in 1980 to 94.8 percent in 1990. During this period, flat rates increased from $8.74 to $17.79 in nominal dollars. Adjusted for inflation, the increase was only about $3.12, from $1 1.02 to $14.14 in 1984 dollars. Data from FCC Telephone Subscribership, April 1995, at 2; FCC Trends, February 10, 1995 (Updated) Table 8, at 13; Statistical Abstract of the UnitedStates, 1995, Table 762, at 493, and 1991, Table 769, at 476.

service (i.e., 23 percent did not subscribe) in 1983, and in 1994 the corresponding figures were 84 percent and 16 percent. Non-subscriber households below the poverty level declined by about 29 percent between 1983 and 1994, implying, in fact, a somewhat greater improvement in the subscription performance of these households than of those above the poverty level or in the general population. [35] Table 1 [36] summarizes the progress by these groups since 1983.

Table 1. Percentage of Households With and Without Telephone Service, 1983-1994

			Household Income Group

      Below Poverty Level  Above Poverty Level        Total
	 With  Without	      With  Without       With  Without
1983     77.1    22.9         94.4     5.6        91.4     8.6
1994     83.7    16.3         95.8     4.2        93.8     6.2
% Change  8.6   -28.9          1.6   -26.1         2.6   -27.9

B. Subscription Growth is Attributable to Small Growth in Real Subscriber Access Prices, Decline in Other Prices, Income Growth, and Other Factors

Residential demand for telephone access service is related to real (i.e., inflation adjusted) prices for flat-rated and measured service, connection charges, toll rates, income and

____________________

35 Because poverty levels are given by household size, we first estimated the average size for an impoverished household from The Statistical Abstract of the United States, 1995, Table 65, at 57, and Table 753, at 484. Then we used this figure, 2.97, to find the average poverty threshold level, in household income, using linear interpolation between the poverty threshold for 2-person households and 3-person households for each year from 1983 to 1994. Finally, we applied the income levels to FCC data on penetration to find the average penetration rate for households below the poverty threshold. Note that this approach is somewhat imprecise. The income distributions used for the poverty levels by the Statistical Abstract is more coarse than those used for the FCC telephone penetration. However, it is clear from the FCC data that progress has been achieved for all income groups.

36 Data based on The Statistical Abstract of the United States, 1995, Tables 65, 726, 746, and 753, at 57, 471, 48 1, 484 respectively; The Statistical Abstract of the United States, 1986, at 447, and the FCC Telephone Subscribership Report, April 1995, Table 4.

other demographics, the uses to which telephone service can be put, and prices of complementary and competing services.

Since real prices, not nominal prices, affect the demand for telephone services, it is important to assess how inflation-adjusted prices have changed. Since divestiture, inflation has essentially offset increases in flat rate charges. Using 1984 as the base year, the SLC increased flat-rated charges from $13.35 to $16.85. Adjusted for inflation from 1984 to 1994, $16.85 translates to about $11.81 in 1984 dollars. Inflation has also offset the effect of tax and intrastate price increases, including the SLC. In 1994, the flat-rated monthly charge (including tax and intrastate increases) averaged $19.00. Adjusting for inflation, this was equivalent to $13.32 in 1984 dollars, or slightly lower than the 1984 rate.

Declines in other telephone service prices offset nominal increases in flat rates.

Adjusted for inflation, from 1984 to 1994: [37]

* Interstate toll rates declined by about 47 percent,

* Intrastate toll rates fell by about 40 percent;

* Connection charges fell by 33 percent; and

* Overall telephone service rates declined by 20 percent.

Figure 1 38 shows the inflation-adjusted prices described above. [39] The figure reflects both rate rebalancing and the impact of inflation on rates for residential telephone services.

____________________

37 Information on rates obtained is derived from The Statistical Abstract of the United States, 1991, Table 769, at 476, The Statistical Abstract of the United States, 1995, Table 762, at 493, and The Economic Report of the President, February 1996.

38 Ibid, Base Year = 1984.

39 Prices were adjusted for inflation with CPI data from The Statistical Abstract of the United States, 1995, Table 762, at 493, The Statistical Abstract of the UniledStates, 1991, Table 769, at 476, The CPI for each service was divided by the overall CPI for all goods and services. The data were then adjusted so all services had a 1984 base year.

A study by Hausman, Tardiff, and Belinfante suggests that rate rebalancing (i.e., lowering toll rates and increasing flat-rate charges) will in fact stimulate demand for telephone access service. [40] A 1988 Southwestern Bell study comparing telephone bills from a sample of 500,000 customers with another sample of 500,000 customers in low income areas found that

... the reductions in toll rates since the introduction of SLCs have more than offset the amount of those charges for the average customer in both samples, resulting in a lower toll bill; the reduction of toll rates has greatly stimulated toll usage since divestiture, the growth rate of toll usage has been about twice as great for low-income subscribers as for subscribers in general, resulting in toll usage paners that are now nearly equal for both groups; and the SLC constitutes a small percentage of the average subscriber's total bill (including subscribers in low-income areas). This study provides evidence [that] the reductions in toll

____________________

40 Jerry Hausman, Timothy Tardiff, and Alexander Belinfante, "The Effects of the Breakup of AT&T on Telephone Penetration in the United States, American Economic Review, 83, 1993, 178-184.

rates have provided significant benefit to low-income households ... by making toll calls more affordable ... and ... by reducing the total bill of average and above average users of interstate toll service .[41]

According to Belinfante, disconnect studies performed by the Regional Bell Operating Companies and GTE in the Monitoring Docket during 1988

... [found that] virtually no households disconnected due to the SLC increase, most of the households disconnected for economic reasons were involuntarily disconnected due to nonpayment of their bills, and most involuntarily disconnected households were heavy users of telephone service, including toll service. These findings led me to conclude that there are far more households without phone service today because of their inability to pay for toll charges than because of their inability to pay for SLCS. This conclusion was reinforced by the observation that involuntary disconnects declined after toll rates were reduced .[42]

These studies were complemented by a survey which found that 56 percent of respondents said that they do not have telephone service because of cost, i.e., 44 percent do not have service for reasons other than Cost.[43] The reasons varied widely. Some respondents wanted to avoid bothersome incoming calls, some felt no need to call anyone, some preferred to live in remote areas, etc. The basic monthly cost ranked near the bottom (only 23 percent of respondents) among reasons for finding telephone service hard to afford. Thus, only for a small subset of households that did not subscribe would any change in residential basic rates possibly have affected the affordability of telephone service. Further, of those for whom affordability was an issue, the most frequently cited impediments to subscribing were not the monthly rate but the "cost of calls outside the U.S." (49 percent) and the "cost of calls within the U.S." (40

____________________

41 Belinfante, in B.Cole (ed.), at 379, op cit., supra, note 31.

42 Id., at 378.

43 Field Research Corporation, Affordability of Telephone Service: A Survey of Customers and Non- Customers, 1993.

percent). Evidently, these respondents either could not control their own toll calls, or they could not control the toll calling behavior of other household members. [44]

Increases in the flat-rate charges have also been offset by the increased availability of lower-cost measured rate and Lifeline options. Options such as local measured service (LMS) and message rate service are typically offered at substantially lower access prices than flat rate service. For instance, in 1993, the average residential rate for flat rate local service was $18.82, while the lowest generally available rate was $11.27. 45 The availability of LMS has increased in recent years, from about 51 percent of lines in 1989 to 67 percent in 1995 for the BellSouth region .[46] Lande's stud), of local service finds that availability in his sample of 95 cities throughout the US has increased from 80 percent in 1987 to 89 percent in 1992. 47

The two federal subsidy programs created to assist low-income households have also grown substantially. The Lifeline program provides subsidies to offset the SLC. Lifeline has grown since its inception in 1985 to 38 states in the United States. Federal funding for Lifeline has increased from $12 million in 1987 to $123 million in 1994, as both the size of the support grew with the SLC increases and the number of subscribers under the plan grew from about I million households in 1987 to 4.4 million in 1994. [48] The Link-Up program provides one-time assistance to low-income households to help pay for the initial installation fee. Link-Up

____________________

44 See also "Phone Plan is Attracting Immigrants in New York", by Randy Kennedy, New York Times, March 18, 1996, page BI The story explains how a firm is making a niche for itself by providing resale phone-service for customers in New York City who have been disconnected from NYNEX for failure to pay large toll bills among other reasons. The new company only allows customers to make long distance and overseas calls that are pre- paid, something NYNEX does not currently have the capability to do.

45 Federal-State Joint Board, Monitoring Report, CC Docket No. 80-286, May 1995, Table 5.7. These figures both include the SLC.

46 Data from BellSouth. Figures apply to residential subscribers.

47 James L. Lande, Reference Book.: Rates, Price Indexes, and Household Expenditures for Telephone Service, FCC Common Carrier Bureau (Industry Analysis Division), at 14. The presence of LMS is governed by two major factors: regulatory approval and the availability of stored program switches. The higher presence of stored program switches in urban areas (Lande's sample is composed of cities) explains the greater availability of LMS within the Lande sample.

48 Federal-State Joint Board, Monitoring Report, CC Docket No. 80-286, May 1995, Tables 2.5 and 2.6.

currently exists in 51 states and territories (including Puerto Rico, the Virgin Islands, and the District of Columbia), and has helped over 4.4 million households to subscribe .[49]

Real income growth also stimulates demand for subscriber access. Real income has grown modestly during recent years. Median household income, adjusted for inflation, has grown from about $22,415 in 1984 to about $22.620 in 1994.[50]

Although overall rates for telephone service have decreased by 20 percent in inflation- adjusted terms, average monthly expenditures per household have increased by $3.00 per month in constant dollars in the decade following divestiture.[51] Demand for most residence telephone services is inelastic; thus, this expenditure growth suggests that the demand curve has shifted over time. This shift appears to have been caused by the availability of a growing number of telecommunications products and services, particularly beyond pure voice communication services, e.g., fax, data, voice mail, Internet, etc. It might also reflect lower prices for these services and higher prices for substitutes.

C. Continued Rate Rebalancing Will Not Harm Universal Service

A $1 increase in local rates would have a much smaller effect today than when the SLC transition began. There are three reasons for this. First, a $ 1.00 increase todav would represent a smaller percentage increase in a person's telephone bill than it would have in 1984. For example, the first dollar of the SLC was initiated in 1985 and gradually raised to $3.50 by 1989. Were a charge of similar magnitude to be introduced today, the consequent increase in local rates would be less dramatic. This point can be seen in Table 2 where we compare rates between 1984 and 1994, the most recent year for which such data are available.

____________________

49 Id., Table 2.7

50 The Statistical Abstract of the United States, 1995, Table 724, at 469 and the Bureau of the Census, Income, Poverty and Labor-Force, Statistics Branch, fax, figures in 1984 dollars.

51 From FCC Trends, February 10, 1995 (Updated) Table 8, at 13.

Table 2. Percent Increase in Nominal Local Rates Due to Increases in LSC
(Current Dollars)

		1984				1994

Base monthly local rate=$13.35		Base monthly local rate=$19.00

Percent increase in local rate due 	Percent increase in local rate due
to $1.00 in new SLC charges=7.6%	to $1.00 in new SLC charges=5.3% 

Percent increase in local rate due	Percent increase in local rate due
to $3.50 in new SLC charges=26.2%	to $3.50 in new SLC charges=18.4%

					Percent increase in local rate due
					to $2.50 in new SLC charges (which
					would raise total SLC to $6.00)*=13.2%

*This hypothetical $6.00 SLC would match the FCC's original proposal
in CC Docket No. 78-72, Phase 1

Table 3. Real Local Rates Due to Increases in SLC (Constant Dollars)

		1984					1994
						(in constant 1984 dollars)

Base monthly local rate=$13.35		Base monthly real local rate=$13.32
					(nominal local rate=$19.00)

Local rate due to $1.00 SLC=$14.35	Real local rate due to an additional
					$1.00 SLC=$14.02

Local rate due to $6.00 SLC=$19.85	Real local rate due to an additional
					$2.50 SLC (which would raise total
					SLC to $6.00 in 1994)=$15.07

Table 4. Comparison of Local Rates Over Time When SLC Charges Are Increased

Year	Local Rates With SLC Raised to	Local Rates With SLC Raised to
	$6.00 in 1996			$6.00 by 2000 (Phased-In)

	Nominal Rate	Real Rate      	Nominal Rate	Real Rate 
		      (1994 Dollars)		      (1994 Dollars)

1996	   $19.00	  $12.56	   $19.00	  $12.56

1997	   $21.50	  $13.79	   $19.50	  $12.51

1998	   $21.50	  $13.39	   $20.50	  $12.77

1999	   $21.50	  $13.00	   $21.00	  $12.70

2000	   $21.50	  $12.62	   $21.50	  $12.62

Note: Assumes 3.0 percent inflation annually between 1994 and 2000.

____________________

52 The $2.50 increase assumed in our example would bring the SLC to $6.00 by 2000, in increments of.- $0.50 in 1997, $1.00 in 1998, $0.50 in 1999, and $0.50 in 2000. This phase in is loosely based on the actual period in which the $3.50 SLC was implemented.

increases in flat-rate charges for telephone service during the last decade (e.g., wider availability of low-price access service, lower toll rates, lower connection charges, and an exogenous shift in the demand curve in response to increased quality and uses of telephone service) have also reduced the elasticity of demand with respect to flat rates. That is, as prices of complements decrease, the desirability of the service increases, and income levels rise, the sensitivity of access demand to price will also decrease. This means that faced with an equivalent percentage price increase, fewer households would discontinue service today than would have discontinued service during the 1980s. Therefore, a given increase in service charges today will likely have less of an effect on the average household today than it did during the 1980s.

It is important to note that we are not simply making the case for raising the SLC while leaving other rates unchanged. We also believe that rate rebalancing - by reducing rates - would offset the impact of a rising SLC on total telephone charges. Based on our experience in the last decade, it seems clear that rate rebalancing that leads to lower usage charges, expanded use of Lifeline programs, and increases in flat-rate charges for telephone service or in the SLC would have little. if any, adverse impact on universal service. [53]

V. ECONOMIC PRINCIPLES TO DEFINE/ANALYZE SERVICE ESSENTIALITY AND AFFORDABILITY

A. Essentiality The FCC seeks comments on how it "should evaluate whether a service or feature is ,essential to education. public health, or public safety."[54] In this respect. the FCC specifically

____________________

53 Quantitative evidence of this is presented in Alexander Larson, Thomas Makarewicz, and Calvin Monson, "The Effect of Subscriber Line Charges on Residential Telephone Bills," Telecommunications Policy, 13, 1989, and T. Makarewicz, "Efficient Telecom Pricing: Who Stands to Benefit?" Public Utilities Fortnightly!, March 15, 1996.

54 NPRM, I9.

solicits comments for the following services: voice grade service, touch-tone, single party service, access to emergency services (91 1 and enhanced 91 1), and access to operator services.[55]

Historically, not all telecommunications services have been deemed sufficiently important or "essential" - to warrant regulation to the same degree as services more closely associated with universal service, such as the services listed above. For example, even though certain vertical services offered by LECs (call waiting, call return, etc.) are not exposed to a high level of competition, those services have traditionally not been accorded the. same regulatory treatment as services considered to be more essential.[56] That is, even among non- competitive services, there are some services that are deemed to be essential and the others non- essential. Therefore, whether or not a service should be considered essential does not appear to depend solely on whether it is supplied competitively or non-competitively. Accordingly, the FCC seeks guidance on what should qualify a service to be essential, and we propose the following principles in response.

We consider an end-user service [57] to be essential if it is vital for promoting and sustaining a minimal, socially acceptable standard of living. Which service qualifies for that status may be judged by asking whether (i) non-availability of the service would cause the end- user to be deprived of a means for meeting a minimum threshold of his or her social or survival needs, and (ii) availability of the service would provide a benefit to society that exceeded the direct benefit that would accrue to the end-user.

____________________

55 Id., para. 16-22.

56 It is readily evident that such services have not been either (i) subscribed to by a substantial majority of residential subscribers or (ii) universally deployed by telecommunications carriers, i.e., have not met two of the criteria established by the Act ([[section]]25 1 (c)(1)(A)-(D)) for a service to qualify for universal service status.

57 We refrain here from defining essentiality for services not sold to end-users, e.g., facilities, inputs, or intermediate services that would be considered essential to the further production of services but are not, in and of themselves, of any consumption value to end-users.

By the first criterion, an essential service can have no feasible substitutes, i.e., functional alternatives available at equal or comparable economic cost [58] to end-users. The dependency of end-users on a particular service deemed to be essential could be explained, at least in part, by the fact that those end-users have few options with respect to meeting their minimum social or survival needs. Therefore, subscribership to the public switched network has traditionally been the most effective means for retaining access to and communicating with communities of interest considered vital to the end-user's private and social well-being.

The second criterion - of social value exceeding private value - is frequently referred to in economics as a "positive externality" and is usually accompanied by the recommendation that the pricing of the service in question be done in a manner that recognizes the value of that externality. Consistent with this principle. we have seen a long tradition of pricing basic local residential service "low," even below its incremental cost. The economic defense of pricing that service below cost and, hence, requiring a subsidy from other services has traditionally been that the economic surplus (or social welfare) lost from pricing the service below or away from cost is compensated by the additional social welfare generated by the externality arising from the increase in network subscribership. [59] Note, however, as shown above, rate rebalancing (i.e., raising basic local rates and lowering other rates) appears to avoid the welfare losses associated with keeping basic local rates substantially below cost, without compromising subscribership. In fact, this pollcy combined with focused universal service support may actually stimulate demand for residential access . We evaluate the eligibility of the services listed in the NPRM for "essential service" status bv applylng these two criteria to each service. The NPRM already contains some of the

____________________

58 That is, not just the price paid for an alternative of comparable functionality but also costs incurred in seeking out or obtaining the alternative.

59 See, e.g., the discussion in John T. Wenders, The Economics of Telecommunications.: Theory and Policy, Cambridge, MA: Ballinger, 1987, especially Ch. 4.

reasoning that we would consider sufficient for according essential service status to the listed services.

Voice grade access to the public switched network: Voice grade access, whether by wireline or wireless means, satisfies both criteria for the same reasons that basic local residential service (of which voice grade access is a component) has traditionally been considered an essential service. There are no comparable-cost alternatives to wireline/wireless access to the network and, when a new subscriber gains such access, the value of the network as a whole (i.e., value to all existing subscribers) increases more than the value to the single subscriber. Not only would the new subscriber receive value from gaining access to educational and medical facilities, emergency services, and other communities of interest, but additional value would also be created to those very same facilities and communities from being able to access the new subscriber through the same network.

Touch-tone: As the NPRM itself states,[60] touch-tone is becoming "increasingly indispensable for subscribers" because it speeds up access to emergency services and is the predominant means for interacting with automated information systems. Although most current phone sets can easily switch between pulse and tone d'aling to allow use of such services by those who do not subscribe to touch-tone service, doing so is less convenient and slower than using touch-tone service per se. Furthermore, the value to societv of adding another subscriber to touch-tone service conceivably exceeds the private value to that subscriber because of increased awareness and utilization of touch-tone accessible services.

Single party service: This is the usage counterpart of voice-grade access. Again, the reasons cited by the FCC in the NPRM[61] and the reasons provided above for regarding voice grade access as essential qualify this service to be essential as well.

____________________

60 NPRM, para., 19.

61 Id., para. 20.

Access to emergency services: Traditionally, these services have been regarded as essential because of obvious public health and safety reasons. However, they appear to meet our criteria as well. While home alarm monitoring systems may represent a potential alternative to 91 1 and E91 I services, they are also considerably more expensive (or not universally affordable) and do not, in many cases, provide the ability to communicate live with emergency and public safety officials. Hence, alarm systems may not be a feasible alternative to 91 1 and E91 I services. In addition, access to these services may make possible increased community awareness and security, not just greater security for the subscriber. Therefore, its social value likely exceeds its private value.

Access to operator services: Traditionally, these services have been deemed essential because of public necessity and convenience reasons. However, they too appear to meet our criteria. Operator services are usually demanded whenever individuals require alternative methods of payment (credit card calls, collect calls, etc.). The method of access itself does not matter - wireline or wireless. Because such services greatly enhance the potential to use the public switched network and the communication reach of individual subscribers themselves, the social value created by such services may well exceed the private value. While other communication options to traditional sent-paid calls may exist (e.g., mail, courier service, etc.), none can match the immediacy of the communication made possible by telecommunication services, albeit operator-assisted. In that respect, no feasible alternatives may exist.

We conclude that the five services listed in the NPRM qualify to be called essential services by our two criteria.

B. Affordable Basic Local Rates

The Act requires the FCC and the states to ensure that service rates nationwide are "just, reasonable, and affordable."[62] Accordingly, the FCC seeks comment on how rate levels should

____________________

62 The Act, sec. 101(a), [[section]]254(i).

be considered to be affordable. [63] Specifically, the NPRM asks for (i) criteria or principles for determining affordability, (ii) methods for evaluating rate levels, and (iii) procedures to recalibrate the rate levels should changes in inflation and other factors make such recalibration necessary from time to time. [64]

We start by noting that no specific guidelines for determining "affordable rates" exist today. States have used a variety of justifications to set rates for basic local residential service while allowing for different (but averaged) rate levels for different rate groups. Some states depart from the traditional flat-rated rate structure to allow measured (or usage-based) rate plans under which subscribers typically pay a combination of smaller monthly charges and usage-related charges. In addition to the state-set rates, subscribers pay an FCC-assessed SLC, currently $3.50 per residential line per month.

We propose that any affordability criterion that is adopted should pay particular attention to what subscribers or households on the margin would consider affordable. These consumers may respond to very small changes in basic local rates by changing their status, e.g., from subscriber to non-subscriber, or vice versa. For these consumers. federal and state assistance programs like Lifeline and Link-Up America are currently in place to reduce their costs of subscribing. Affordability in this sense, however, is generally not an issue for the vast majoritv of subscribers for whom small or even moderate increases in rates do not typically induce a change to non-subscriber status .[65]

____________________

63 NPRM, para. 25.

64 Id.

65 Some analysts believe that telephone services as an expenditure category are generally income- inelastic, i.e., not very responsive to changes in income. See, e.g., B.M. Mitchell and 1. Vogelsang, Telecommunications Pricing.- Theory and Practice, New York: Cambridge University Press, 1991, at 226, note 4. However, at least one study suggests that end-user responsiveness to changes in local rates does vary by income groups. See P. Cain and J.M. MacDonald, "Telephone Pricing Structures: The Effects on Universal Service," Journal of Regulatory Economics, 3, 1991, 293-308. This study found that households with annual income below $I 0,000 tend to be considerably more responsive(1O-15 times)than household seaming inexcess of $25,000 annually. However, even for the most vulnerable, lowest-income segments, the risk of dropping telephone service was minimized by allowing households to opt for measured, rather than flat-rated, service plans.

We propose that a lower bound for affordability be constructed for low-income households on the basis of the purchasing power of those households. For this, we first focus on households that qualify for various forms of public assistance, e.g., Social Security Insurance (SSI), Aid to Families with Dependent Children (AFDC), and Food Stamps. Recipients of these forms of public assistance also qualify for Lifeline assistance in most participating states. [66] In other states (e.g., Arizona. California, Michigan, etc.), eligibility for Lifeline is based on household income being up to 130-150 percent of the poverty level. Accordingly, we propose that, for present purposes, affordability be based on household income of 125 percent of the poverty level.

Earlier, we showed that the average size of a household living at or below the poverty level in 1993 was 2.97 .[67] By linearly interpolating between poverty threshold incomes of impoverished households of sizes two and three ($9,646 and $11,807 respectively), [68] we calculate that the annual poverty threshold level of income in 1994 for a household of size 2.97 was $11,745. Furthermore, at 125 percent of the poverty threshold, that household had an annual income of $14,68 1. We also note that the average annual income in 1993 of households at the top of the lowest quintile (20 percent) of the income distribution was $16,952 ,[69] i.e., above the 125 percent of poverty level.

Recent data show that while average annual expenditures of households on all forms of telephone service are about 2.0 percent of expenditures on all items, there is some variation in that percentage between low and high income households. For example, while only 1.4 percent of annual expenditures was directed at telephone service bv the highest income quintile in 199 1. up to 3.1 percent of annual expenditures was so directed bv the lowest income quintile in

____________________

66 Federal-State Joint Board, Monitoring Report, CC Docket No. 80-286, May 1995, Table 2.4.

67 See supra, note 35

68 Constructed from Stalistical Abstract of the United States, 1995, Tables 746 and 76 1.

69 Statistical Abstract o the United States, 1995, Table 73 3.

that year.[70] Applying the latter percentage to the average household income at 125 percent of the poverty level (i.e., $14,68 1), we estimate that household's average expenditure on all telephone services to be $455 per year or $37.90 per month.

Other data indicate that, for the lowest income quintile, a household's expenditures on basic local service has remained between 43 and 48 percent of its expenditures on all forms of telephone service. [71] Accordingly, assuming that 45 percent of such expenditures were on basic local service in 1994, we estimate that the average household at 125 percent of the poverty level would have spent $17.06 per month on basic local service. This estimate can serve as a lower bound on the affordability threshold for basic local rates for low-income households.

It is noteworthy that the affordability benchmark of $17.06 per month is based on actual out-of-pocket household expenditures on basic local service. That is, a household that receives Lifeline assistance or other basic rate subsidies could, in principle, afford a gross basic local service bill of at least $7 more (at current SLC levels) or nearly $24. Also, many, if not most, of these households could reduce their out-of-pocket costs of basic local service by opting for measured rate local service plans which often cost far less. For example, in 1992, while the national average monthly flat-rated local service cost households $18.66 per month (including SLC but not Touch-tone), the national average measured rate was onlv $11.12 per month (including SLC). [72]

It is instructive to evaluate our proposed lower bound for the affordability threshold in light of national average flat-rated local rates. According to a recent survey of 95 cities typically sampled by the Bureau of Labor Statistics for constructing its CPI values. flat-rated basic local service was priced on average at $19.74 (with touch-tone service) per month in

____________________

70 These percentages are remarkably stable over time and, hence, can be assumed to be true in the latter half of the 1990s as well. See James L. Lande, Reference Book. Rates, Price Indexes, and Household Expenditures for Telephone Service, FCC Common Carrier Bureau (Industry Analysis Division), May 1993, Table 7.

Id., Table 8. Lande provides annual data from 1984 to 199 1.

72 Id., Table 4.

1992. This was down slightly from $19.83 in 199l.73 As stated above, measured rate plans were priced on average (not including usage) at $11.12 per month in 1992. Both sets of average rates included a federal SLC of $3.50 per month. In light of these figures and the very small increase in local residential rates in recent years, we surmise that there is room for those rates to rise further. For example, if the federal SLC were to rise another $2.50 per month (to $6 per month), the average flat-rated service - starting with the 1992 base - would be priced near or just above $22.33. This would be affordable because assistance-eligible households that can afford at least $17.06 per month could receive an additional $7 to $12 in total Lifeline assistance (assuming current plans remain in effect) .[74] If the SLC were to rise from its present level of $3.50 to $6 per month, the total Lifeline support could rise from $7 to $12 per month. Households not eligible for Lifeline assistance would also not be disadvantaged by a flat-rate near $22.33 because of their greater purchasing power. [75] Since the affordability threshold here is calibrated to the purchasing power of households that qualify for Lifeline and other assistance, it follows that such a threshold would keep rates affordable for non-assistance- eligible, i.e., higher-income, households as well. We conclude that raising the monthly residential subscriber line charge to $6 per line would still keep basic residential local rates affordable for all households. We also conclude that our affordability threshold of nearly $17.00 per month would receive substantial support from any move to make overall telephone service affordable, especially for the most economically vulnerable households. For example, institution of toll

____________________

73 Id., Table 2. Even though prices of other components of local residential service have trended up slightly since 1990, reductions in touch-tone rates have offset or more than offset those increases.

74 Except for California, eligible households in every other Lifeline-participating state receive at least twice the level of the interstate SLC in support, of which half is federal support and the other half is matching state support.

75 These households typically have a lower percentage of their expenditures on telephone services directed at basic local service - 38.5, 36.4, 33.3, and 26.8 percent for the second, third, fourth, and highest income quintiles in 1991, respectively. Lande, supra, note 70, Table 8.

limitation services [76] and/or reductions in rates for toll or discretionary services would amount to rate rebalancing that consumers on the margin of subscribership would find to their advantage. This form of rate rebalancing would offset the higher SLCs with lower charges for other services, with little or even favorable impacts on overall bills.

VI. OTHER ISSUES

A. Competitive Bidding for Universal Service Support

The NPRM seeks comment on whether competitive bidding (using a form of Dutch auction) should be used to determine which carriers should provide universal service and receive universal service support.[77] Specifically, it wishes to know whether competing carriers should be allowed to bid to set the level of universal service support, with the lowest bidder in a given serving area winning the right to serve that area.

The overarching objective of ensuring that the least-cost provider serves a particular area is indisputably in the public interest. The question, however, is whether a competitive bidding process is the best way to achieve that objective. In our opinion, the answer is "no." We believe instead that the methodology that we have proposed for determining the initial level of support is more effective and economical and administratively simpler than the competitive bidding process.

The competitive bidding process is unnecessary for the stated purpose. Our proposed methodology for setting the level of support, besides being administratively simpler, would also let the "Invisible hand" of market competition determine the least-cost providers. As we explained before, the per-line support would be set initially as the difference between the incumbent's per-line embedded cost of basic service and the basic rate. With that support

____________________

76 NPRM, II54-55.

77 NPRM, II35-37.

available to any competitor that actually serves final consumers, the market would send universal service support to carriers that needed less per-line support (because of their inherent relative efficiency) and could lower the basic rate to consumers. This process would continually match the competitor's incremental cost against the incumbent's embedded cost (and, eventually, the incremental costs of all competitors) and ensure that only the least-cost provider served the market and received universal service support to the extent needed.

In addition, we believe that competitive bidding could be a complicated, costly, and potentially contentious process that either federal or state regulators or some other neutral entity would have to administer and oversee. Moreover, as technology changes and different service providers enter or exit the market over time, competitive bidding would have to be conducted repeatedly in order to ensure that only the least-cost providers qualified for support at all times. A similar process may be needed as high cost areas are periodically refined with respect to their geographic and/or demographic characteristics.

In sum, under our proposed method, the competitive market would ensure that prices ultimately move towards cost without requiring an externally-administered bidding process to generate competition.

B. Proxy Cost Models

The NPRM seeks comment on the usefulness of proxy cost models (such as the Benchmark Costing Mode,[78]or BCM) for determining the level of universal service support in different serving areas.[79] As the NPRM states:[80] The [BCM] produces a benchmark cost range for a defined set of residential telecommunications services assuming efficient wireline engineering and design,

____________________

78 Jointly sponsored by MCI, Sprint, NYNEX, and US West in CC Docket No. 80-286 (December 1, 1995).

79 NPRM, para. Para.31-32.

80 Id.

and using current technology. It is not based upon the costs reported by any company, nor the embedded cost to a company of providing service today.

This description also helps explain why the BCM, in its present form, is not sufficient for determining the appropriate level of universal service support. The BCM's purpose is to identify geographic areas which are relatively high or low cost to serve, i.e., to provide benchmarks of how much more or less expensive one area is relative to another. Its purpose, however, is not to determine the absolute level of cost for any area.

By construction, this model would not produce the forward-looking costs of any particular carrier that would likely compete in the local exchange market in a particular state. The BCM uses nationwide values for critical cost inputs such as network equipment costs and installation costs, and assumes engineering practices that cannot be attributed to a particular carrier and might not be feasible or optimal in particular circumstances. Also, since it only produces relative costs, the BCM cannot help to determine the absolute size of the proposed universal service fund. Instead, the model only indicates which serving areas (census block groups or CBGs) are more costly to serve than others.

There are other specific problems with the BCM in its current state. First, the BCM focuses specifically on the investment portion of local telephone service. It accounts for the operating expense portion of costs through assumed annual cost factors on which even the sponsors are not in complete agreement. Second, the BCM does not always accurately represent the locations of existing or planned facilities or assign the CBGs to the correct wire centers. For example, in a recent regulatory proceeding in Kentucky, it was argued that 16 percent of the CBGs in that state were incorrectly assigned by the BCM. [81] In our opinion, this is a compelling reason for caution in using CBGs as the geographic units in cost models. [82]

____________________

81 Kentucky Public Service Commission, in re: An Inquiry into Local Competition, Universal Service, and the Non-Traffic Sensitive Access Rate, Administrative Case 355. Rebuttal Testimony of Peter F. Martin for BellSouth Telecommunications, March 11, 1996.

82 The NPRM, I34, asks whether CBGs are the "best geographic units for developing a proxy model."

Third, certain cost assignments may depart from the best engineering practices followed in a state. Assumptions made about loop lengths, switch types for rural and urban areas, feeder lengths at which fiber is placed, etc. may not be representative of that state and, hence, produce incorrect estimates of cost.[83]

We believe that the BCM is not yet sufficiently reliable for determining proxy costs for specific serving areas in a state. Even its sponsors are unable to agree on specific assumptions and parameter values under which it should be run. Its predictions would be all the more unreliable as the cost of capital and depreciation rates both rose with the advent of local competition. The BCM is most troubling because it does not depict the actual costs of an actual local exchange carrier. Moreover, the "scorched node" assumption [84] that underlies its optimized network model pays little attention to the characteristics of a real-world carrier. With a technology and a network already in place, an existing carrier's options for future technology choice and network optimization would be quite different from - and more constrained than - those faced by a new entrant into the local market. Hence, the BCM's claimed strength that it does not represent costs of a particular carrier but rather that of a representative optimized carrier - is also its biggest drawback.

In general, there are two problems with using optimization models for estimating incremental costs: (i) the optimization process usually succeeds only at providing the lower bound on incremental costs, and (ii) the models ignore real-world details which cause actual incremental costs to exceed the models' estimates.

____________________

83 Timothy J. Tardiff reports that in California, when California-specific assumptions and corrections to the BCM were made, costs rose by about 20 percent: see T. J. Tardiff, Universal Service Funding and Cost Modeling, prepared for Pacific Bell, January 19, 1996 and Evaluation of the Benchmark Cost Model, prepared for Pacific Bell, December 1, 1995.

84 " Scorched node" describes an idealized model in which the most efficient engineering practices are adopted and state-of-the-art loop and switching technology are deployed but the currently existing local exchange network topology (locations of switches, etc.) is accepted. An even more idealized model, often called "scorched earth," would replace the existing network topology with an "optimal" topology, i.e., one in which locations of network components could be optimally and costlessly reconfigured without regard to the actual prior history of the network.

First, as we have argued for the BCM, an optimization model that calculates forward- looking incremental costs for a network that deploys optimal technology at all times cannot depict the actual incremental costs of a real-world network. While such a model may well serve as a predictor of costs for a new network, it cannot possibly depict costs of an existing network with its inherent rigidities. An example will show the type of problem created by the assumption of constant optimization:

Suppose there are I 0,000 households uniformly distributed in a square around a wire center, so that each household is served by one of 4 feeder routes. Ignore the need for spare capacity and growth and assume that an efficient network would supply only one loop to each household. Suppose also that cable sheathes come in units of 100, 500, 1000, 2000 and 2500 pairs and that unit costs are much smaller with larger sheath sizes. An optimization model would first calculate the total cost of the network at 80 percent of its current demand, and the feeder network for this calculation would contain four 2,000-pair cables. Next, the model would recalculate the optimal network to serve I 0,000 households; the feeder part of this network would then contain four 2500-pair cables. The difference in total costs of these optimized networks - divided by 2,000 loops - would constitute the incremental cost of serving each household.

In the real world, however, suppose we begin with an idealized network serving 8,000 households today. If that network experienced growth of 2,000 households, it could not simply scrap its 2,000-pair cable and serve its customers with the 2,500-pair cable that an optimization model would assume. Instead, the network would most likely augment its 2,000-pair cable with an additional 500-pair cable in each direction. The net result would be a higher incremental feeder cost per circuit than that calculated by a model based on the assumption of constant optimization.

Optimization models also typically use simplifying assumptions to keep the computations and simulations tractable. In particular, such models (i) assume a uniform geographic distribution of customers within a wire center, (ii) ignore the real-world complications of mountains, rivers, unusual or difficult terrain or climatological events, and

(iii) fail to account for demographic and population shifts within a network's serving jurisdiction which may necessitate adjustments that an idealized network could disregard.[85] Because optimization models cannot typically account for these elements, the estimates of total and incremental costs that they produce are not relevant measures of those costs incurred by an actual efficient carrier.

We conclude that idealized, optimization-based network cost models tend, at best, to produce idealized minimum cost estimates that provide only a starting point to calculate the actual costs of an actual network. Incremental costs in a network that is optimized at every point in time must underestimate the true incremental costs of network operation because real- world networks have fewer options than continually-optimized networks. Apart from the lumpiness of its network capacity (because it lacks the ability to make fine, marginal adjustments), a real-world network cannot simply resize its facilities to serve more or fewer customers or to take advantage of the latest trend in technology. Unlike optimized networks, therefore, real-world networks can only "optimize" on a going forward basis, conditional on the networks as they currently exist, their past histories and peculiarities notwithstanding.

VII. CONCLUSION

The Commission and the Joint Board face a challenging responsibility - balancing the multiple (sometimes disparate) goals and requirements of the Act - to promote both universal service and more efficient competition. We believe that, in this pursuit, the past is prelude: the FCC and the states have shown that it is possible to rebalance rates to promote both economic efficiency and universal service. Following the principles we articulate above would help the Commission and the Joint Board reach the appropriate balance. As we demonstrate, this can be done by (i) phasing in modest increases in the SLC (to reduce the current implicit internal

____________________

85 This responds to the NPRM's request at (134) for comment on "whether the assumption of uniform population distribution adequately reflects the possibility that in some rural areas, despite the theoretical sparsity, all lines are clustered near a single location" and "whether the [BCM] could be improved by the addition of other variables, such as climate or slope."

subsidies developed under past regulatory policies), and (ii) replacing the remaining implicit subsidies with explicit, targeted, and competitively neutral support mechanisms. This approach would promote the universal service goals of the Act, with minimal harm to another "fundamental underlying principle of the Act ... to provide ... a pro-competitive and de- regulatory national policy framework."[86] Resuming the rate rebalancing effort initiated, but not completed, in the mid -I 980s would capitalize upon the fresh opportunity for substantive and pro-competitive reform offered by the 1996 Telecommunications Act.

____________________

86 NPRM, para. 8.