THE COST OF
UNIVERSAL SERVICE
A Critical Assessment of the
Benchmark Cost Model
Susan M. Baldwin
Lee L. Selwyn
April 1996
ECONOMICS AND TECHNOLOGY, INC.
ONE WASHINGTON MALL * BOSTON, MASSACHUSETTS 02108
PREFACE i
1 THE ROLE OF THE BENCHMARK COST MODEL IN ADDRESSING UNIVERSAL SERVICE FUNDING REQUIREMENTS 1
1.1 Background 1
1.2 General Overview of the status of regulatory USF proceedings 5
1.3 Organization of this report 7
2 ATTRIBUTES OF A USEFUL COST PROXY MODEL 9
2.1 Attributes of a reliable cost proxy model that would be useful for public policy purposes 9
Balancing of core attributes 12
2.2 Defining the costs to be estimated by a cost proxy model for universal service 13
3.1 Preliminary discussion of the BCM 17
The input to the BCM 18
The data module 18
The loop module 20
The output module 25
Summary of the uncorrected BCM results 27
3.2 The joint sponsors have, in some ways, facilitated public scrutiny of the model, and, in other significant ways, frustrated a close examination of some key parameters and algorithms 28
3.3 Purposes of the BCM analyses undertaken for this report 30
3.4 A careful analysis of some of the BCM's key variables and assumptions shows that there are certain areas where the BCM should be improved before it is used as a tool 34
Appendix 3A: States ranked by quantity of CBGs 37
Appendix 3B: Cost Factor Table 41
Appendix 3C: Density Zones Utilized by the BCM 45
Appendix 3D: Fill factors and outside plant costs 49
Appendix 3E: Surface texture table 53
Appendix 3F: Distribution of Washington state population within the BCM's six density
zones 61
4 AN ANALYSIS OF THE COST FACTOR AND PRICE THRESHOLD 65
4.1 The cost factor in a proxy model should not be based upon historical accounting data 65
What the model does 65
What the model should do 66
Historical depreciation expenses reflect estimated lives that are not indicative of the lives of plant necessary to offer basic telephone service 67
Return on investment 69
4.2 The price that policy makers choose to support will significantly influence the USF requirement 70
Appendix 4A: Preliminary analysis of LEC expense accounts 73
Appendix 4B: Distribution of switched access lines by local exchange carrier 77
5 DEVELOPMENT OF CORRECTED SWITCH COSTS 81
5.1 Outdated, unrealistically high switch costs overstate line costs 81
How the BCM reflects switch costs 81
How the BCM should be corrected 83
The model unrealistically deploys Nortel DMS 100 switches in small rural exchanges 86
6 AN EXAMINATION OF OUTSIDE PLANT COSTS 89
6.1 The BCM should be corrected to more accurately reflect the economy of scale and scope inherent in incumbent LECs' networks 89
The BCM only partially recognizes the presence of business lines 89
The use of CBGs as the relevant unit for high-cost analysis does not require it to be used as the basis for determining support requirements 92
By determining the need for USF support on a CBG basis rather than a wire center basis, the BCM significantly exaggerates USF requirements 93
Line Costs Compared Separately for Each CBG 98
CBG-Based Line Costs Aggregated and Compared for the Wire Center 99
Competitors will seek out opportunities to develop their own economies of scale and scope, and will not find it efficient to limit service to discrete CBGs 100
6.2 The failure to eliminate costs incurred to enable the provision of second (and additional) lines would grossly distort universal service support requirements 101
The costs of additional lines must be identified and removed from the aggregate cost of distribution plant to arrive at the economic cost of providing universal service 102
Universal service customers must obtain a reasonable share of the benefits arising from scale and scope economies 103
The additional costs of providing capacity for second lines are not insignificant 105
Any model used to determine universal service costs must correct for the overstatement of universal service costs incurred to accommodate demand for additional residential lines and other services 106
6.3 The fill factors for the feeder and distribution should be increased 107
What the model does 107
What the model should do 107
6.4 The BCM makes an uneconomic choice between deploying copper and fiber in the feeder plant 110
6.5 The assumption of uniform household density should be revisited 117
Appendix 6: Comparison of wire center and CBG based support 119
7 INTERPRETING AND APPLYING THE RESULTS OF THE BENCHMARK COST MODEL 123
7.1 The appropriate benchmark for determining USF eligibility and quantifying the level of support 123
7.2 Recommendations regarding the establishment of a benchmark for determining eligibility and the method for computing USF support 126
7.3 Federal and state policy makers should coordinate USF plans to ensure that in no event is more than 100% of "high cost" recovered by USF support 127
7.4 If a new universal service funding mechanism is established, other existing sources of support should be either eliminated or reduced 129
7.5 Revenues from other services linked to the provision of residential dial tone lines will offset much of the USF requirement 129
Appendix 7: State summary data for 1993 134
8 THE MAGNITUDE OF THE UNIVERSAL SERVICE FUNDING REQUIREMENT 139
8.1 Aggregate effect of correcting the BCM 139
Without adjustment for subscribership: 139
With adjustment for subscribership: 140
Appendix 8A: Summary of major ETI runs of the BCM 147
Appendix 8B: ETI's partial corrections and sensitivity analysis BCM:
Washington State 149
Appendix 8C: Cost comparison of copper/fiber crossover point at the CBG level 163
9 THE RELATIONSHIP OF THE BCM TO STATE USF PROCEEDINGS 167
9.1 Analyses of USF issues at the state level are instructive to federal USF policy development 169
9.2 A detailed examination of the Cost Study Methodology and Universal Service Funding Issues under investigation in California 169
CPM software and costing algorithms 172
User-modifiable assumptions and inputs 175
10 CORRECTING THE SHORTCOMINGS OF THE BENCHMARK COST MODEL 177
10.1 Recommendations regarding the use of the BCM in USF policy deliberations 177
10.2 Summary of the effects of the ETI corrections to the BCM 178
10.3 The use of the BCM in policy making proceedings 180
TABLES
Table 3.1 User Inputs to the BCM in the Loop Module 21
Table 3.2 Main Feeder Segment Types for CBGs in the Same Quadrant 23
Table 3.3 Main Feeder Segment Types and Household Count for CBGs in the Same Quadrant 24
Table 3.4 Summary Results of the BCM National Total (excluding Alaska) 27
Table 3.5 Percentage of Washington State CBGs Receiving USF Assistance at Different Thresholds as Calculated by the BCM 31
Table 3.6 Percentage of Washington State Households Receiving USF Assistance at Different Thresholds as Calculated by the BCM 32
Table 3.7 Summary Results of the BCM Washington State 33
Table 4.1 Selected Data for Tier I LECs 67
Table 5.1 Implausibly High Switch Costs Exaggerate USF Requirements 85
Table 5.2 Comparison of Switch Costs for the BCM Washington State Average and for an Illustrative Rural Wire Center 87
Table 6.1 Costs Are Too High if Based upon CBG Averages - Illustrative Analysis of Wire Center AVONCOMA, Colorado Assumes Cost Support at $30 Per Line Level 98
Table 6.2 Costs Should Be Aggregated to the Wire Center - Illustrative Analysis of Wire Center AVONCOMA, Colorado Assumes Cost Support at $30 Per Line Level 99
Table 6.3 Cable Sizing Means that Actual Fill Will Be Significantly Less than Objective Fill Illustrative Analysis (95% Objective Fill) 108
Table 6.4 BCM Fill Factors for Cable Should Be Increased 109
Table 6.5 Average Monthly Cost by Household Density Class Using Various Copper/Fiber Crossover Points Generated by the BCM 113
Table 6.6 Analysis of BCM Main Feeder Selection and Alternative "Total Loop Cost" Assuming Copper Main Feeder 114
Table 6.7 Approximation of SLC Electronics Cost Necessary for Economic Copper/Fiber Crossover Point at 12,000 Feet 116
Table 7.1 Benchmarks Vary for Determining Eligibility for High Cost Support 125
Table 7.2 Methods for Computing Amount of High Cost Support Vary 125
Table 7.3 Existing FCC High Cost Fund 126
Table 8.1 The BCM Overstates the USF Requirement Data for Washington State without Adjustment for Subscribership 142
Table 8.2 The BCM Overstates the USF Requirement Data for Washington State Including Adjustment for Subscribership 143
Table 8.3 Comparative Summary Results of the BCM and the ETI Partially Corrected BCM Washington State 144
Table 8.4 Comparative Summary Results of the BCM and the ETI Partially Corrected BCM National Total (excluding Alaska) 145
PREFACE
THE COST OF
UNIVERSAL SERVICE
The need for an objective measure of the cost of providing basic local exchange service has been identified by state and federal policy makers as they pursue the simultaneous goals of universal service and local competition. The recent enactment of the Telecommunications Act of 1996 underscores the importance of establishing universal service funding mechanisms that allow the efficient development of competition in the local market while ensuring that all households have access to affordable telecommunications service. The Benchmark Cost Model represents a commendable effort by four telecommunications carriers to develop an objective measure of the cost of providing basic telecommunications services to households throughout the country. The BCM is a potentially valuable and timely tool that can assist in federal and state investigations of universal service. The BCM's potential to contribute to universal service funding decisions, however, depends upon the incorporation of several important modifications to the cost proxy model, which we present here.
This report was prepared by Economics and Technology, Inc. on behalf of the National Cable Television Association in order to provide a critical assessment of the strengths and weaknesses of the BCM, and, where possible, to offer affirmative recommendations for improvement. The project was conducted under the overall direction of Susan M. Baldwin and Dr. Lee L. Selwyn. Contributing to this work were Helen E. Golding, John T. McDermott, Michael J. DeWinter, Irena V. Tunkel, Scott C. Lundquist, and Susan M. Gately. The project also benefitted from the suggestions and ideas of Richard L. Cimerman, Director, State Telecommunications Policy, NCTA. The views in this report are those of ETI and do not necessarily reflect the views of the NCTA.
EXECUTIVE SUMMARY
THE COST OF
UNIVERSAL SERVICE
The prospect of real competition in local telephone service has the promise of bringing to this sector the kinds of benefits that have been realized in other competitively impacted telecommunications markets -- lower prices, increased pace of technological innovation and diffusion, acceleration and expansion in the introduction of new products and services, and enhanced industry productivity and growth overall. As with the competitive initiatives that occurred in the customer premises equipment and long distance service markets in the 1970s and 1980s, incumbents have sought to slow the pace of competitive incursions into what had been single supplier markets by seeking to portray the disbanding of their monopolies as a threat to "universal service" -- the goal of assuring maximum connectivity throughout the nation to the public switched telecommunications network. With respect to competition in the local exchange market, incumbent carriers have resurrected the "cream skimming" argument, claiming that new entrants would select only the lowest cost areas and markets to serve, leaving the incumbents -- and their customers -- with the prospect of escalating costs to maintain universal service availability. According to incumbent local exchange carriers, their traditional ability to offer below-cost service to customers in "high-cost" areas stems from their ability to fund such shortfalls internally, by charging above-cost rates to other customers and for certain high-margin services. With selective competitive entry, the LECs contend, these internal sources will be diminished, and alternate funding sources and mechanisms will be needed. Most prospective new entrants accept this possibility, but disagree that the magnitude of the universal service support that is needed will be anywhere near as extensive as that contended by the incumbents.
In an attempt to resolve this dispute as to the aggregate magnitude of universal service support that must be generated from other services, four telecommunications carriers -- NYNEX, US West, MCI and Sprint -- collaborated on the development of an economic model of the local exchange service industry. The goal of this effort was to definitively establish the actual costs and support levels needed to assure that no diminution of residential connectivity will occur, and to provide a basis for a formal funding mechanism that would assess all telecommunications providers for a fair share of the required universal service support. The results of this effort, known as the Benchmark Cost Model, were submitted to the Federal Communications Commission on September 12, 1995. The BCM represents what may be the most comprehensive attempt to date to develop an objective national model of the cost of providing basic residential local exchange service. The BCM applies a "scorched node" philosophy in modelling the architecture of the local exchange carrier networks, in that it "deploys" a state-of-the-art telecommunications network assuming that all existing locations of network nodes -- wire centers -- are unchanged and that the locations and numbers of distribution nodes are established exogenously rather than optimally.[1] The BCM relies upon census data in order to determine the quantity of households that are to be served and the average household density within each of some 220,000 subregions known as Census Block Groups. The BCM is also based upon the simplifying assumption that all households are uniformly distributed within each of the CBGs. By design, the BCM is not intended to replicate a carrier's actual or embedded costs, but rather is intended to simulate the forward-looking cost of providing basic local exchange service based upon factors such as the terrain being served, population density, and equipment costs.
The BCM satisfies many of the essential attributes of a useful cost proxy model, and can be a valuable tool for achieving the universal service and local competition policy goals and mandates of the Telecommunications Act of 1996. The BCM relies upon publicly available data, generally incorporates reasonable network engineering assumptions, and models forward-looking costs. However, the BCM has several serious shortcomings that have the cumulative effect of substantially exaggerating the aggregate cost of basic local exchange service and of the universal service funding requirement. To be useful, the BCM must be corrected to eliminate these deficiencies and to generate more accurate and realistic results.
This report analyzes and corrects several of the key engineering/economic assumptions and input data upon which the BCM is constructed, and details the effect of these corrections on the model's results. Because replication of the entire national model was not feasible within the time frame available for this examination,[2] we selected one state -- Washington -- as the basis for our analysis. Washington ranks 17th in the number of CBGs, and includes a diverse and representative mix of natural terrain and population densities. Where correction of a defect was not feasible, we have attempted to identify the problem and to propose specific methodological remedies. Finally, the report discusses how the results of the BCM should be used in universal service funding deliberations at the state and federal levels as well as at the Joint Board. Among the report's key findings are these:
The BCM overstates the average cost per residence line and overstates the universal service funding requirement:
* Corrections to the model reduce the national average monthly cost as determined by the BCM by at least 25%, and the national aggregate universal service funding requirement by at least $1.4-billion.[3] The results of rerunning the BCM to incorporate some but not all of ETI's corrections produced an average national cost of approximately $12.37 and a national USF requirement of approximately $749-million. If it had been possible for us to implement all of the corrections that we have identified, the actual results would have been even lower. These partial results do not yet reflect (1) a correction for the BCM's implausible assumption of uniform density of households within a CBG; (2) replacing the arbitrary and fixed copper/fiber crossover rule with an economic algorithm; and (3) assessing the need for universal service support at the wire center level rather than at the CBG level.
The existing sources of universal service support are not in imminent jeopardy:
* The numerous implicit existing sources of revenues that incumbents currently enjoy, combined with the various existing explicit sources of USF support, are not jeopardized by the entry of local competition. The largest such revenue source -- yellow pages -- has historically been used for this purpose, was assigned to the Bell Operating Companies in the MFJ expressly for this purpose, and is not affected by the entry of competing local exchange carriers. Therefore, policy makers can move forward in a timely but comprehensive manner to resolve the funding issues while still achieving the goals of promoting universal service and competition in the local market.
The BCM results should be considered together with an examination of various existing implicit and explicit sources of USF support
* The BCM does not purport to address many key questions relating to the establishment of an explicit universal service fund. The BCM is potentially a valuable tool that can contribute substantially to the USF debate, but the Joint Sponsors have neither raised nor answered a number of important and highly relevant questions. In addition to evaluating the BCM, this report also undertakes to address several of the more challenging questions that are key to ultimately sizing and establishing explicit universal service support programs.
In configuring the model network, the BCM adopts an unduly expansive definition and scope of universal service that greatly exceeds the statutory requirements:
* The statutory goal of universal service refers to connectivity to the public network, which is satisfied by the provision of one primary access line per dwelling unit. However, the BCM does not confine its network design criteria to this standard, but instead models a network capable of supporting demand for multiple access lines as well as for certain premium services. The cost levels developed by the model thus overstate those that would be required to satisfy the minimal connectivity requirement. Although local telephone switching and distribution infrastructures are designed to satisfy more than the stand-alone demand for primary residential access lines, the model needs to confine itself to those costs that would be incurred solely to achieve the more limited universal service objective.
Certain key variables should be corrected before the BCM is adopted:
* Outdated and overstated switch costs exaggerate the potential funding requirement. The model should be corrected to reflect current switch costs, including the often substantial discounts that LECs routinely receive from the switch manufacturers.
* By determining the need for USF support on a CBG, rather than on a wire center, basis, the BCM fails to recognize many economies of scale and scope and thus significantly exaggerates funding requirements. Universal service costs and support requirements should be evaluated on a wire center basis.
* The fill factors in the model should be corrected to reflect the fact that the service that should be modelled -- single line basic residence local exchange service -- is a stable, predictable service that does not require the excess capacity needed by the LEC to offer other local exchange services (e.g., additional residential access lines and business service) that are characterized by more volatile demand. This correction reduces the amount of excess outside plant capacity and results in a lower cost per working subscriber line.
* The model's algorithm for determining when to deploy fiber rather than copper in the feeder plant is not economically based, and overstates the cost of feeder plant.
* The costs for the digital subscriber loop equipment do not reflect manufacturer discounts and should be reduced accordingly.
* The BCM computes a monthly per-line cost by multiplying the total investment per line by an expense factor, which is intended to reflect operating expenses and a return on investment. Expenses should be based upon forward-looking costs that reflect those expenses legitimately associated with the provision of primary residential exchange access service.
The BCM's approach for determining USF support is more appropriate than that used by the existing High Cost Fund:
* The need for and size of universal service support should be determined by comparing the cost of providing the primary residential access line in each wire center district with the price level that policy makers determine to be affordable. This approach differs from the existing high cost fund (which compares a carrier's average cost with the national average cost), and also differs from some state proposals that would compare the price of local exchange service in given areas with the statewide average price.
If properly corrected, the Benchmark Cost Model can support the objectives of the Telecommunications Act of 1996:
* The Telecommunications Act of 1996 underscores the need to quantify the level of support required to ensure network connectivity for all households throughout the country, while developing support mechanisms that will enable competition to evolve efficiently in local markets.[4]
* Reliance upon embedded cost studies would not satisfy the important purpose of a cost proxy model, which is to provide an objective, forward-looking measure of the ongoing cost of supporting a correctly-specified universal service goal based upon efficient engineering and design.
* A cost proxy model, by reflecting objective measures of providing basic residential local exchange service, will allow the FCC and the state PUCs to size and to target assistance where it is needed, without unnecessarily burdening consumers and providers of basic local telecommunications service.
* The BCM, with the corrections identified in this report, should be adopted by federal and state policy makers as a valuable tool for addressing and resolving universal service funding issues.
[2]We estimate that each full national run of the BCM would require approximately 110 hours on a Pentium 100 MHz PC, plus an additional 20-30 hours of manual intervention. Our analysis of the data for Washington state by itself involved not less than 20 individual runs of the model, each one of which required roughly 2.5 hours of computer time.
[3]For illustrative purposes, these data assume the "Cost Factor 2" and a support threshold of $30.
[4]Telecommunications Act of 1996, Pub. L. No. 104-104, 110 Stat. 56 (1996) ("Telecommunications Act"), Sec. 254.