AN OVERVIEW
OF THE
BENCHMARK COST MODEL
3.1 Preliminary discussion of the BCM
The goals of the Benchmark Cost Model are to (1) estimate the forward-looking cost of providing basic local exchange service to households and (2) to quantify the universal service funding necessary to cover the differential between that cost and a desired price level. The BCM uses Census Block Groups (CBGs) as a unit of analysis for computing household line costs for each of 49 states and the District of Columbia.[1] The United States Census Bureau defines a "Census Tract" as a subdivision of a county containing between 2,500 and 8,000 housing units. Within this division, a block is a small area bounded by "visible features" such as roads or streams.[2] A CBG (the area for analysis in the BCM) is a cluster of blocks having the same first three-digit identifying numbers which, according to the Census Bureau, are "usually between 250 and 550 housing units, with the ideal size being 400 housing units."[3] Census block groups vary enormously in terms of the land area that they cover, the population they encompass, and the relative presence or lack of household clustering.[4]
The BCM's cost results are expressed (1) on an average CBG per-line basis; (2) on an aggregate basis for each CBG; (3) on an average statewide per-line basis; and (4) on an aggregate statewide basis (i.e., the total cost for all household lines in the state).[5] The BCM arrives at these cost estimates by processing input in the form of individual CBG records for each state through three separate modules that each consist of an Excelreg. spreadsheet file.
The input to the BCM
The BCM is an incremental "scorched node" study, meaning that for the purpose of determining how to "deploy" theoretical telecommunications plant, the model is based upon the location of existing wire centers.[6] Although existing nodes (i.e., wire centers) are assumed for the purposes of the model, existing equipment is not assumed, and in fact the model reflects the deployment of state-of-the-art DMS 100 digital switches. The CBG records that serve as input to the BCM, are assigned to an existing central office switch and to an actual provider of local exchange services as well.[7] The BCM also assigns CBGs to one of four switch "quadrants." CBGs in the same quadrant are served by one of four main feeder segments which are assumed to leave from each switch in four directions (due east, north, west and south). The creation of these hypothetical wire center serving area boundaries is based upon V&H coordinates and NECA Tariff 4 CLLI codes. The CBG input records also include an estimate of the airline distance from the center of each CBG to the appropriate central office switch, a total household count, and the CBG's household density per square mile. Finally, the input records contain measures for bedrock depth and hardness, water table depth and a surface texture. These geological variables directly influence network design and cost algorithms throughout the BCM.
The Data Module
Although the BCM bases network development upon existing central office switches, the outside plant (feeder and distribution) is deployed without any reference to existing plant.[8] The Data Module characterizes the difficulty of constructing outside plant for individual CBGs on the basis of the geological and geographical variables mentioned above. For example, the "Data & Calcs" sheet of the Data Module includes formulas that derive the main feeder, subfeeder and total distribution distances for each CBG from the centroid distance and other geographic data provided in the input records. The BCM allocates sub-feeder segments to those CBGs not directly in the path of the main feeder route. Sub-feeder segments are assumed to branch off the main feeder route at a 90 degree angle and extend to the edge of the CBG. Distribution plant, which is always assumed to be copper, completes the connection to individual households in each CBG and is calculated on the assumption that households are evenly distributed throughout the CBG. Also, formulas in the Data & Calcs sheet transcribe geological variables such as rock hardness and surface texture into values that reflect for example the difficulty of deploying underground plant or of constructing new conduits or telephone poles.
The Data & Calcs sheet also includes formulas that generate terrain-based multipliers that are used later to calculate structure costs (or the costs of conduit systems, interduct, poles, etc.) for individual CBGs. The results of these formulas are pulled into columns labelled "Distribution Cable Multiplier," "Feeder Cable Multiplier," and "Fiber Multiplier."[9] The formulas in these columns "look up" multipliers in the "Weighted Cost Factor Table" located in the "Tables" sheet of the Data Module.[10] There are 54 weighted cost factors in this table ranging in magnitude from 0.233 (the multiplier for distribution plant in rural areas with the lowest household density and with "normal" surface texture) to 11.5456 (the multiplier for fiber plant in urban areas with the greatest household density and the hardest surface texture). The weighted cost factors themselves are generated by formulas that use as inputs 24 "unweighted" cost multipliers that are listed in four tables appearing in the same sheet.[11] The values in these tables are weighted by the percentage split between plant that is underground and plant that is aerial. Underground/aerial splits for distribution cable, copper feeder and fiber feeder plant for each of the six household density classes[12] are listed in three tables also found in the "Tables" sheet of the Data Module.
The 24 unweighted structure cost multipliers in the "Tables" sheet are calculated as a percent of cable costs based on ratios of a cost per foot to place the plant vs. the cost per foot of the plant itself. These estimates are provided for both underground and aerial copper and fiber plant in areas with different surface textures and different household densities. As such, they are among the most fundamental inputs to the BCM and yield the plant type structure multipliers described above. However, information on how these cost multipliers were determined is noticeably absent from the explanatory materials provided by the Joint Sponsors.
The Loop Module
The Loop Module is the second of three modules in the BCM and is perhaps the most important as it develops and costs out the outside plant portion of the network.[13] The Loop Module consists of five separate sheets, identified as "Input," "Main Logic," "Shared Allocation," "Costing," and "Output." The Input sheet consists of the output of the Data Module,[14] a column that divides the main feeder into individual segments so as to assign CBGs a new main feeder distance, many of the user-specified inputs, and costs for the outside plant.
The main feeder distance of the outermost CBG in a particular quadrant is set equal to its own total main feeder distance minus that of the CBG that is next closest to the central office switch.[15] Segmentation of the main feeder in this way is crucial as those segments closest to the central office switch will be engineered to carry the capacity of CBGs further out along the feeder. This becomes an important distinction later on as the cost of the main feeder is allocated among CBGs in the Shared Allocation sheet.
The default values for the user-specified variables in the "Input" sheet are shown in Table 3.1 below.
Table 3.1 User Inputs to the BCM in the Loop Module User Input Default Values Maximum Copper Feeder Cable Size 4200 Maximum Copper Distribution Cable Size 3600 Fill Factor for AFC Electronics .8 Fill Factor for SLC Electronics .8 SLC Cost per Access Line $500 AFC Cost per Access Line $550 Fiber Cable Discount 20% Copper Cable Discount 20% AFC Electronics Discount % 10% SLC Electronics Discount % 20%
The Input sheet also includes four new tables that are critical to the BCM's network design and costing functions. These tables include a Cable Fill Factor Table that assigns a feeder and distribution fill factor for each of the BCM's six household density classes; and three tables that provide per-foot costs for various sizes of copper distribution, fiber cable and copper feeder plant.[16] The costs in these tables are presented for both underground and aerial plant. Finally, the Input sheet includes discounted prices for both Advanced Fibre Communication's (AFC) and AT&T's "Subscriber Line Carrier" (SLC), the two fiber technologies employed by the BCM.[17] These discounted prices are derived from the baseline prices and percentage discounts described above.
The Main Logic sheet of the Loop Module assigns copper or fiber feeder plant of appropriate capacity to individual CBGs. As stated above, the distribution plant is always analog copper cable and is assumed to consist of four segments of equal size and capacity for each CBG. The main feeder route, however, may utilize (1) copper plant only, (2) fiber plant only or (3) a combination of copper plant and fiber plant. Copper plant is used for those CBGS where the total distribution distance (feeder and distribution) is less than 12,000 feet. Fiber plant is used when a CBG's total distribution distance exceeds 12,000 feet.[18] Where the model does deploy fiber, it then also deploys digital subscriber loop equipment -- either an AFC or an SLC.[19]
After a copper or fiber technology has been assigned to each CBG, the BCM begins the process of ensuring that each main feeder segment has the capacity to meet the traffic demand of CBGs further out along the main feeder segment. It does this by assigning a "Segment Type 2" and "Segment Type 3" to CBGs whose main feeder segments contain multiple technologies. A CBG will have a Segment Type 2 if another CBG further out along the main feeder route employs a main feeder technology different from its own. For example, as illustrated below in Table 3.2, if the first CBG in a sequence of 3 CBGs, CBG 1, is served by a copper main feeder segment, and if CBGs 2 and 3 are served by SLC, then the Segment Type 2 for CBG 1 would be SLC. In this case, CBGs 2 and 3 would not have a Segment Type 2. If in the above example, CBG 3 was served by AFC, then CBG 1 would have a third Segment Type -- AFC. Furthermore, CBG 2 would have a Segment type 2 of AFC and CBG 3 would have only a Segment type 1 -- AFC.
Table 3.2
Main Feeder
Segment Types
for CBGs in
the Same
Quadrant
Office Quadrant Block Group Segment Segment Segment
Seqnc. # Type 1 Type 2 Type 3
ABCDSTMA 1 1 Copper SLC --
ABCDSTMA 1 2 SLC -- --
ABCDSTMA 1 3 SLC -- --
ABCDSTMA 2 1 Copper SLC AFC
ABCDSTMA 2 2 SLC AFC --
ABCDSTMA 2 3 AFC -- --
The BCM then assigns to each CBG an aggregate number of households for each main feeder technology, again for the purpose of calculating the capacity requirements of each main feeder segment. For example, as illustrated below in Table 3.3, the number of households "on copper" for the first CBG in a sequence of three CBGs that are all served by copper would be the total number of households in the three CBGs. The number of households on copper for the second CBG in this case would be equal to its own households plus those in the third CBG. Returning to our first example where CBG 1, CBG 2, and CBG 3 were served by copper, SLC and AFC respectively, the households on copper for the first CBG would equal the number of households in CBG 1, the households on SLC for CBG 1 would equal the number of households in CBG 2, and the households on AFC for CBG 1 would equal the number of households in CBG 3.
Table 3.3 Main Feeder Segment Types and Household Count for CBGs in the Same Quadrant Segment Households Segment Segment HH on HH on HH on Type 1 in the CBG Type 2 Type 3 Copper SLC AFC Copper 250 -- -- 400 -- -- Copper 100 -- -- 150 -- -- Copper 50 -- -- 50 -- -- Copper 300 SLC AFC 300 150 100 SLC 150 AFC -- -- 150 100 AFC 100 -- -- -- -- 100
The household totals for each main feeder technology are used to calculate the number and size of copper feeder pairs required by each CBG and the number and size of SLC and AFC fibers as well. In the case of CBGs served by copper main feeder, the BCM divides the number of households on copper by the feeder fill factor appropriate to the CBG's household density.[20] For example, a CBG with 6,000 households on copper and a household density of 1,000 households per square mile would, according to the BCM's default fill factors, have a feeder fill factor of .8 and thus require 7,500 copper feeder pairs.[21] In the case of copper plant, the BCM then translates the capacity requirements for each CBG into the number of maximum size cables that would be employed (4200 for copper feeder and 3600 for copper distribution by default) and the cable size necessary to carry any remaining fraction of total capacity. Sub-feeder and distribution plant carry only the traffic of their associated CBGs and so their capacity requirements are calculated on the basis of the number of households in that particular CBG as opposed to the total households served by that CBG's main feeder segment.
Formulas in the Main Logic sheet labelled "Cable Structure %," "SLC Structure %," and "AFC Structure %" return percentages that add up to 100% for each CBG and serve to weight the copper feeder and fiber multipliers previously calculated in the Data & Calcs sheet of the Data Module.[22] For example, the cable structure percentage for a CBG that was served by cable feeder but which had a Segment Type 2 of SLC and a Segment Type 3 of AFC -- meaning that other CBGs further out along the main feeder route are served by SLC and AFC feeder -- would be 80% while the SLC Structure percentage and AFC Structure percentage would be 10% each. The copper multiplier for that CBG then, would be weighted by 80% and the fiber multiplier would be weighted by 20%.
The Shared Allocation and Costing sheets of the Loop Module cost out the network that has been developed in the Main Logic sheet by matching the copper and fiber cable sizes calculated in Main Logic with per foot plant costs for copper feeder, copper distribution and fiber. These costs are located in three tables in the Costing sheet that include per foot costs for each plant type and cable size for each of the six household density classes. For example, column H in the Costing sheet looks up the appropriate per foot cost for the maximum size copper feeder calculated in Main Logic and multiplies this cost by the number of maximum size cables employed and the length of the Main Feeder segment. The cost for other plant types is calculated in much the same way. Formulas in the Shared Allocation sheet then assign a proportion of plant costs to each CBG and then to each household using the "Segment Type" and "Household Count" columns described above. Structure cost percentages from Main Logic are applied to each plant type and incorporated into final cost calculations for total copper, total SLC, and total AFC feeder. These three columns are combined in a Total Feeder Cost column and added to the Total Distribution Cost to yield a Grand Total Loop Cost. The Output sheet of the Loop Module then, includes this Total Loop Cost as well as a Loop Cost per Household column, a household density range, and a measure of the average Total Loop Length for each CBG.
The Output Module
The Output Module adds the cost of switching plant to the outside plant calculations completed in the Loop Module. There are four sheets in the Output Module titled, "Data & Calcs," "Tables & Inputs," "Results," and "Summary."[23] The BCM also includes a formula that multiplies the number of households in each CBG by a "Business Gross Up Factor" of 1.75 which is a user input in the Tables & Inputs sheet.[24] This factor is described in the Joint Submission as the ratio of total lines to residential lines and is used to account for economies attributable to the presence of non-residential lines.[25] The "new" grossed up number of households is converted into a total switched line count for each CBG by dividing the grossed up number of households by the switch fill factor.[26]
The switched fixed cost per line is calculated as follows: The formula first multiplies the BCM's default fixed cost per switch of $647,526 by the "Basic Local Service Factor" or the percent of the switch cost that is allocated to the line.[27] This product is divided by the total number of lines in the wire center to yield a total fixed switch cost per line. The BCM then uses a fixed common cost per switch of $238.87 that was generated through a regression analysis of per line costs for a Northern Telecom DMS 100 switch for 20 different line sizes.[28] This fixed common cost per switch is added to the switched fixed cost per line described above and multiplied by the BCM's default Switch Land and Building Factor of 1.043 to yield a total Switched Investment per Line.
The Tables & Inputs sheet also includes the two cost factors used by the BCM. The first cost factor of 31.6765% was calculated on the basis of historical accounting data and total expense levels of Tier 1 LECs utilizing 1994 ARMIS Form 43-01, while an alternate cost factor of 22.97% is based on the Hatfield/MCI study approach and reflects limited expense categories and amounts.[29] Both of these cost factors are applied to the Total Investment per line to yield two separate annual costs. Annual cost figures are converted to monthly costs which are then compared to the BCM's three benchmarks for high cost support of $20, $30 and $40 to yield the total monthly support requirement if any for each CBG. Finally, the Summary sheet presents the aggregate support at the three benchmark support levels for all of the CBGs in the state under the two different cost factor scenarios. Statewide totals for the sum of households, average total loop length, average loop cost per household, and average total investment per line for each of the six density classes are also presented in the Summary sheet.
Summary of the uncorrected BCM results
The Joint Sponsors' Submission of December 1, 1995 contained average monthly costs for local exchange service for the entire nation (except Alaska) of $23.04 and $16.71 respectively for the Annual Cost Factor 1 (the embedded cost factor) and the Annual Cost Factor 2 (the Hatfield/MCI cost factor). These results and the national universal service support requirements for the three benchmark support levels are presented below in Table 3.4.
Table 3.4 Summary Results
of the BCM National Total
(excluding Alaska)
Annual Cost Factor #1 Annual Cost Factor #2
Annual Benchmark Cost $25,377,893,663 $18,402,608,162
Support at $20 $8,082,313,345 $3,977,572,193
Support at $30 $4,916,517,444 $2,203,441,910
Support at $40 $3,208,565,853 $1,372,205,121
Average Monthly Cost $23.04 $16.71
Source: Joint Submission
of December 1, 1995 at
II-2.
As Table 3.4 highlights, the BCM's aggregate national support requirement ranges between $1.4-billion and $8.1-billion, depending upon the cost factor selected and the price level being supported. If one considered the entire country to be the relevant "study area" -- this is not an approach we recommend -- and used the Annual Cost Factor 2, universal service support would be required only if the price threshold were set below $16.71.[30] In fact, one of the key questions facing policy makers is to define the appropriate boundaries of the "study area." This issue is discussed in more detail in Chapter 6, below.
The data shown in Table 3.4 are based upon the printed document submitted by the Joint Sponsors to the FCC on December 1, 1995. However, the results that are generated by actually running the most recent public version of the BCM (i.e., September 12, 1995) with the most recent public copies of the state files (i.e., December 1, 1995) do not coincide with the results submitted to the FCC in the Joint Submission.[31] The Joint Sponsors updated the BCM since the September filing with, at a minimum, an expanded list of soil types.[32] We incorporated these new soil types into ETI's analyses, which may explain the discrepancy of $0.08 between the Average Monthly Cost for cost factor #2 for Washington State, under the ETI run of the BCM and the Joint Sponsors' run of the BCM.[33] In Chapters 4 through 8, below, where we discuss the results of our analyses, we use as our "status quo" benchmark the results that we obtain by running the model, even though these differ slightly from those published by the Joint Sponsors.
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
The Joint Sponsors have, at one level, made a commendable effort to develop and make available a "public" cost proxy model. The demonstration version of the model and data for six states has been available for public examination since September 1995, and data for the entire country has been available since December 1995. The Joint Sponsors have also held four workshops to facilitate use of the model by interested parties.[34] Furthermore, the Joint Sponsors have been responsive to various criticisms of the model, and where they concur and believe feasible, have indicated their intent to correct the BCM accordingly.[35] Finally, because all of the algorithms, inputs, and formulas that are in the full model are also in the demonstration model, many of the attributes of the model can be readily evaluated through use of the demonstration model alone.[36]
There is, however, one significant aspect of the BCM that belies its characterization as an "open" model and that frustrates efforts at pursuing a comprehensive and objective analysis. The Main Logic, Shared Allocation, Costing, and Output sheets of the Loop Module are password protected and cannot be adjusted by the user.[37] The Loop Module, as described above, is perhaps the most important of the BCM's three separate modules as it assigns plant types and costs to the outside plant portion of the network. Among the types of analyses that cannot be readily performed because of the password protection are the following:
* Adjustment of the 12,000 foot crossover point for the deployment of copper or fiber feeder plant.
* Alteration of the allocation of plant and structure costs among CBGs in the same quadrant.
* Adjustment of the plant costs associated with different size cables of all plant types.
The overall credibility of the BCM is diminished by the Joint Sponsors' decision to "lock" these aspects of the model and to prevent their modification by other users. As we discuss below in Chapter 6, preliminary efforts to modify one of these three "locked" parameters -- the copper/fiber crossover point -- suggests that the BCM has adopted a fundamentally uneconomic decision rule that appears to result in a significant overstatement of the costs that are required by LECs to furnish primary residential access lines.
3.3 Purposes of the BCM analyses undertaken for this report
There are several distinct purposes of the analyses of the BCM that we have undertaken here. They are to:
* Gain hands-on familiarity with the algorithms used in the model, e.g., the relationship of engineering assumptions to cost results;
* Identify the sensitivity of the results to some of the individual parameters, variables and assumptions; and
* Analyze the implications of changing certain parameters to derive what we believe are more accurate results.
For the purposes of this report, we ran the model for illustrative samples of CBGs that we selected from the full-state data input files, and we also ran the model for the entire state of Washington for most of our analyses. The analyses of very small samples are clearly not intended to be either statistically significant in any fashion nor comprehensive, but rather to illustrate some probable results and to frame future analyses that should be undertaken by or on behalf of the Joint Board. We believe that the analyses undertaken in this report of an entire state are, however, indicative of the effect of changing the BCM parameters upon the national results.
Washington State has 1,871,765 households and 4,542 CBGs (which places it 17th in the number of census block groups). Washington State includes a diversity of natural terrain and population densities, ranging from Seattle (a metropolis with a population of over 500,000 and a density of over 6,000 people per square mile) to rural areas, including 21-million acres of forested land.[38] Other than the cities of Seattle, Spokane and Tacoma (the latter two each having populations in the 175,000 range), no other city in Washington has a population approaching 100,000. The densest CBG has an average of 33,713 households per square mile,[39] and the most sparsely populated CBG has an average of 0.03 households per square mile.[40] According to the BCM, within the State of Washington, the monthly average line cost, as measured within a CBG, ranges from a low of $5.70 per month to a high of $544.61 per month.[41] Thus, the lowest cost CBG is in the highest density category, and conversely, the highest cost CBG is in the lowest density zone. To place this in perspective, the highest cost CBG serves a mere seven households which require (according to the uncorrected BCM) aggregate annual support of $44,068 at the $20 threshold level or over $6,000 of support per household per year. Table 3.5 and Table 3.6 below indicate separately for each of the six density zones the CBGs and the households that would receive USF assistance according to the BCM.
Table 3.5
Percentage of
Washington State
CBGs Receiving
USF Assistance
at Different
Thresholds as
Calculated by
the BCM
Density Zone Number of CBGs $20 Support $30 Support $40 Support
Level Level Level
<=5 275 100% 100% 99.3%
5 to 200 1099 75.3% 23.4% 6.9%
200 to 650 642 0.9% 0.0% 0.0%
650 to 850 242 0.0% 0.0% 0.0%
850 to 2550 1548 0.7% 0.0% 0.0%
> 2550 736 0.4% 0.0% 0.0%
TOTAL 4542 24.7% 11.7% 7.7%
Table 3.6
Percentage of
Washington State
Households
Receiving USF
Assistance at
Different
Thresholds as
Calculated by
the BCM
Density Zone Number of $20 Support $30 Support $40 Support
Households Level Level Level
<=5 62645 100% 100% 99.8%
5 to 200 372988 78.2% 21.6% 5.6%
200 to 650 273086 0.7% 0.0% 0.0%
650 to 850 109294 0.0% 0.0% 0.0%
850 to 2550 689169 0.2% 0.0% 0.0%
> 2550 364583 0.1% 0.0% 0.0%
TOTAL 1871765 19.1% 7.6% 4.5%
Our preliminary analysis of the BCM's Washington State data file[42] shows a mean of 412 households per block group. Yet, further analysis shows that the standard deviation of this mean is 269, with a minimum value of one household[43] and the maximum value of 3,489 households per CBG.[44] (Nationwide, 3,608 of 220,000 CBGs have densities of less than 1 household per square mile.[45]) The largest CBG in Washington is approximately 1,300 square miles and includes 35 households.[46] Businesses are not considered households under the census definition, and therefore, particularly in those areas where there is a disproportionate number of businesses, the density of the CBG will be understated,[47] as will the potential impact of the business demand upon the overall unit cost of serving the CBG and its associated wire center.[48] Therefore, the cost factor multipliers that are related to the density zone assignment may inadvertently be applying "more rural" cost factors than are appropriate. The BCM's results for the state of Washington are shown in Table 3.7.
Table 3.7 Summary Results
of the BCM Washington State
Annual Cost Factor #1 Annual Cost Factor #2
Annual Benchmark Cost $524,623,612 $380,427,268
Support at $20 $158,350,839 $77,846,835
Support at $30 $97,982,543 $50,692,630
Support at $40 $72,368,201 $37,662,589
Average Monthly Cost $23.36 $16.94
Source: ETI run of the
BCM, without corrections.
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
Although the BCM provides a reasonable foundation for a cost proxy model, there are certain assumptions and algorithms that should be corrected before the model is adopted for use in policy making proceedings. The following sections of this chapter analyze and, in many instances, recommend corrections to these key variables:
* Cost factor, i.e., the way to translate total investment into annual carrying costs (the percentage by which to multiply the total investment in order to compute an annual figure to reflect operating expenses and a reasonable return on investment).
* The price threshold, i.e., the monthly price above which the BCM computes USF requirements.
* The cost of the switches.
* Variables that relate to a LEC's economy of scale and scope. These variables include:
(a) The manner in which the BCM accounts for the existence of business lines.
(b) The area for which the eligibility for and level of USF support is evaluated.
* The scope of the service for which the model is yielding a cost proxy. The low fill factors used in the BCM suggest that the model fails to distinguish the costs of providing one primary basic residential exchange service access line per household from the costs of additional residential lines and, for that matter, of all other loop-using LEC services. This issue in turn affects the fill factor used.
* The equipment prices and discounts for the SLC and AFC equipment.
* The assumption of uniform distribution of households within a CBG.
* The lack of SAIs.
* Overstatement of incremental costs to CBGs most distant from the central office.
This report does not evaluate each and every critical variable, and thus silence should not necessarily be construed as acquiescence or endorsement. For example, this report has not provided any examination of the costs of fiber and copper cable that are assumed, although such data would clearly influence the BCM results. In this example, and for similar situations, we urge the Joint Board, the FCC, and PUCs to seek back-up support from the Joint Sponsors for the data provided.
[2]Definitions from this section are summarized from the 1990 Census of Population and Housing, Summary Population and Housing Characteristics, New York, at A-3 to A-5.
[3]Id., at A-4.
[4]See Appendix 3A for the quantity of CBGs in each state.
[5]The census data in the BCM (e.g., quantity of households and CBG boundaries) are based upon 1990 information. See Joint Submission, at 10.
[6]By contrast, a so-called scorched earth model would not presume any information about the location of switches.
[7]Each CBG is mapped to the closest existing wire center, regardless of whether the carrier that serves that wire center actually serves the geographic area encompassed by the particular CBG.
[8]This is necessary for modelling purposes since the locations of CBGs do not correspond to existing wire center serving areas. Indeed, if one assumes that the existing wire center serving areas are "efficient," then modelled areas based upon arbitrary (from the perspective of telephone network design) CBG boundaries are likely to be less efficient.
[9]The "Distribution Cable Multiplier," "Feeder Cable Multiplier," and "Fiber Multiplier" columns are AB, AC and AD respectively in the Data & Calcs sheet of the Data Module.
[10] See Appendix 3B.
[11]These four tables are labelled "Urban Copper Cable Table," "Rural Copper Cable Table," "Urban Fiber Table" and "Rural Fiber Table".
[12]See Appendix 3C for a listing of the six density classifications that are used in the BCM.
[13]The BCM labels these "loop" costs, but for sake of clarity, we refer to this component of the cost that is being modelled as the outside plant cost.
[14]Mechanically, to "run" the model, output data from the Data Module are copied and pasted into the Loop Module.
[15]See Column M of the Input Sheet.
[16]See Appendix 3D
[17]These are in cells Q89 and Q90, respectively.
[18]The algorithm that determines the feeder plant type (Column F in Main Logic) does not consider capacity in the decision to use copper or fiber. In fact, fiber may be the correct economic choice for shorter distances in situations involving large blocks of capacity, and copper may be less costly even for longer distances if the capacity requirement is modest. Significantly, in Pacific Bell's Cost Proxy Model, the Company assumes a crossover point at 9,000 feet (see Chapter 9, below), but like the BCM does not allow that to vary based upon capacity. A major factor in the determining of the economic crossover point is the relationship between the cost of copper and the cost of the digital subscriber loop equipment that is used to "light" the fiber and to derive individual voice-grade (DS-0) channels therefrom. Unfortunately, the prices of the electronics are among the items that are considered highly proprietary, and are subject to heavy discounting from their respective manufacturers. Accordingly, not only is it impossible to verify the copper vs. fiber crossover assumptions incorporated into the BCM, it is also not possible to verify the pricing data that was used to drive that assumption.
[19]An AFC (American Fibre Communications Next Generation Digital Loop Carrier System) is electronic gear that the BCM deploys only in the least dense of the six density zones (where there are fewer than five households per square mile) and the SLC (Subscriber Line Carrier Series 2000) is deployed in the other five density zones.
[20]Column W in Main Logic.
[21]6,000/.8 = 7,500 copper feeder pairs. The corresponding calculations for SLC and AFC fiber requirements are found in columns L and O respectively. Column L calculates the number of SLC fibers for each CBG based on the assumption that each CBG with SLC main feeder utilizes a minimum of 4 dedicated fibers to provide up to 672 voice grade paths. Column O calculates the total number of AFC fibers for each CBG by dividing the number of households on AFC by the product of 672 and the fill factor for AFC electronics. This value in turn is multiplied by 4 to yield a final result.
[22]These are in columns Q, R, and S, respectively.
[23]As before with the output of the Data Module, the output from the Loop Module is copied and pasted into the Data & Calcs sheet of the Output Module.
[24]See Column M of Data & Calcs, which is labelled "Households Including Business."
[25]Joint Submission, September 12, 1995, at 35.
[26]The BCM's default fill factor of 80% is located in cell H17 of Tables & Inputs.
[27]See column C of the Tables and Input sheet. The percentage of the switch cost that is allocated to the line is a combination of the percent of switch costs that are non-traffic sensitive and the percent of traffic sensitive costs that is local. The BCM's default values for these percentages are 70% and 30% respectively and yield a total switch cost allocation figure of 79%.
[28]Op. cit., footnote 65, at Attachment 1.
[29]Id., at 4.
[30]Therefore, on average, any price above $16.71 would exceed the average cost. Furthermore, a household that subscribes to basic local exchange service also typically subscribes to discretionary services (such as call waiting) and makes toll calls. Thus on average, the telecommunications revenues associated with providing any given household with basic local exchange service are significantly above the cost for basic local exchange service.
[31]In fact, running the most current public version of the BCM with the state data files generates error messages because the September version of the BCM does not account for all soil types. For example, in the Washington State WADTIN_1.XLS File, these are the unspecified soil types: BY-FSL, BY-LS, CBV-MUCK, CBX-SIL, ST-VFSL, STV-LS, STX-FSL, and STX-LS, which do not appear in the soil "look-up" tables sheet (beginning at row 38 column A) of the data60~1.XLS file. These abbreviations correspond to soil types bouldery and fine sandy loam, bouldery and loam, very cobbly and muck, extremely cobbly and silt, stony and very fine sandy loam, extremely stony and fine sandy loam, and very stony and loamy sand. These omissions were overcome by obtaining a more up-to-date Surface Texture Table from the Joint Sponsors and inputting this new information.
[32]Conversation with James Dunbar (Sprint) on March 25, 1996. See Appendix 3E for the up-to-date, complete list of soil types.
[33]The printed results presented by the Joint Sponsors in the December 1, 1995 filing showed Washington State's average monthly cost to be $17.02 for cost factor #2. After we incorporated the new soil types into the September 12, 1995 version of the BCM, our analysis yielded an average monthly cost of $16.94.
[34]Joint Submission, at I-3.
[35]Ex parte submission in CC Docket No. 80-286 by Glenn Brown, Executive Director--Public Policy, US West (" Ex parte submission"), January 26, 1996. Ex parte submission, February 21, 1996.
[36]The difference between the demonstration model and the full model is simply the size of the database, which, in turn, affects the hardware required to run the model. The "Demo" can be run on an ordinary personal computer, where as the full model requires substantial computer requirements. The full model is designed for use with up to 600,000 CBG input records while the demonstration model includes space for only 50 CBGs.
[37]The Joint Sponsors consider the password proprietary to the developers of the model and thus will not divulge it to others. Conversations with Mark Bryant, MCI, March 27, 1996; Peter Copeland, US West, April 1, 1996.
[38]See Appendix 3F for the classification of households in Washington among the BCM's six density zones.
[39]There are a total of 885 households in a 0.03 square mile area in this CBG which is in Seattle (CBG #530330074006).
[40]There are a total of 22 households in a 711 square mile area in this CBG which is in Twisp, a rural area in the northern part of the State located in Okanogan County (CBG #530079601001).
[41]The low cost one is CBG 530330033005 of the STTLWASU (Seattle) wire center. The high cost one is CBG 530050116002 of the KNWCWAXB wire center.
[42]WADTIN_1.XLS.
[43]CBG 530050105005, Row 2380 of WADTIN_1.XLS
[44]CBG 530530729021, Row 3865 of WADTIN_1.XLS.
[45]Joint Submission, at V-1.
[46]CBG No. 530599501001; Row 594, WADTIN1.XLS.
[47]In their Ex parte submission of January 26, 1996, op. cit., footnote 75, the Joint Sponsors indicated that the suggestion to identify CBGs which are primarily business (i.e., low number of households in a small geographic area) is "desirable but difficult" which the Joint Sponsors further indicate means that the change would enhance the usefulness of the model and that if there is sufficient interest in making the change, they would be willing to attempt to make the modification. In a later ex parte filing, the Joint Sponsors indicate that they plan to identify situations where the CBG area is less than "x" and households are less than "y" and that the model will assume such cases are primarily business. The model will assign a default business line count of 400 for network design counts. See, Ex parte submission of February 21, 1996, op. cit., footnote 75. We ranked the Massachusetts CBGs by density and identified several CBGs that might satisfy such a test. For example, in Malden (a city with a population of 53,884 and a land area of 4.8 square miles) there is a CBG of 0.19 square miles with one household (CBG 250173413004); and in Lexington (a town with a population of 28,974 and a land area of 16.64 square miles), there is a CBG of 0.75 square miles, and 6 households. (Bureau of the Census, 1990 Census of Population of a Place; Malden Chamber of Commerce and Lexington Chamber of Commerce).
[48]The presence of business demand will permit the LEC to utilize higher capacity cables and switches that offer lower unit costs. By ignoring business demand, the BCM thus overstates the cost of the residential-only demand that it purports to examine. Ironically, although the salutary effects of business demand upon unit cost are omitted in the model, the BCM relies upon company-wide plant utilization factors that are driven downward by the very business demand that the model otherwise ignores, thereby compounding its erroneous treatment of business customers. The correct approach, which we employ here, is to limit the model to the specific services at issue (i.e., the primary residential access line) and to utilize the plant utilization factors that are applicable for this specific service. Gains from scale and scope economies resulting from the inclusion of business service and additional residential access lines are identified separately and should inure to all service categories.
Zone Number Households Per Square Mile
1 0 - 5
2 5 - 200
3 200 - 650
4 650 - 850
5 850 - 2550
6 > 2550
Note: The definition of density zones is located in the Tables tab of the DATA60~1.XLS file. There is a special cost multiplier of 1.28 for density zone 6.
Density Zones Households Percent of Total
<=5 62645 3.3%
5 to 200 372988 19.9%
200 to 650 273086 14.6%
650 to 850 109294 5.8%
850 to 2550 689169 36.8%
> 2550 364583 19.5%
TOTAL 1871765 100.0%