[Federal Register Volume 78, Number 87 (Monday, May 6, 2013)]
[Rules and Regulations]
[Pages 26269-26277]
From the Federal Register Online via the Government Publishing Office [www.gpo.gov]
[FR Doc No: 2013-10565]


-----------------------------------------------------------------------

FEDERAL COMMUNICATIONS COMMISSION

47 CFR Part 54

[WC Docket Nos. 10-90, 05-337; DA 13-807]


Connect America Fund; High-Cost Universal Service Support

AGENCY: Federal Communications Commission.

ACTION: Final rule.

-----------------------------------------------------------------------

SUMMARY: In this document, the Federal Communications Commission 
(Commission) primarily addresses the model platform, which is the basic 
framework for the model consisting of key assumptions about the design 
of the network and network engineering. The Commission also addresses 
certain framework issues relating to inputs.

DATES: Effective June 5, 2013.

FOR FURTHER INFORMATION CONTACT: Katie King, Wireline Competition 
Bureau, (202) 418-7491 or TTY: (202) 418-0484.

SUPPLEMENTARY INFORMATION: This is a summary of the Commission's Report 
and Order in WC Docket Nos. 10-90, 05-337; DA 13-807, adopted on April 
22, 2013 and released on April 22, 2013. The full text of this document 
is available for public inspection during regular business hours in the 
FCC Reference Center, Room CY-A257, 445 12th Street SW., Washington, DC 
20554. Or at the following Internet address: http://hraunfoss.fcc.gov/edocs_public/attachmatch/DA-13-807A1.pdf.

I. Introduction

    1. In the USF/ICC Transformation Order, 76 FR 73830, November 29, 
2011, the Commission comprehensively reformed and modernized the 
universal service and intercarrier compensation systems to maintain 
voice service and extend broadband-capable infrastructure. As part of 
the reform, the Commission adopted a framework for providing support to 
areas served by price cap carriers known as Phase II of the Connect 
America Fund. An estimated eighty-five percent of the approximately 6.3 
million locations in the nation that lack access today to terrestrial 
fixed broadband at or above the Commission's broadband speed benchmark 
live in areas served by price cap carriers. The Connect America Fund 
will maintain voice service and expand broadband availability to 
millions of unserved Americans living in these areas within the next 
five years, and aims to close this gap entirely within a decade. 
Through Phase II, the

[[Page 26270]]

Commission introduced targeted, efficient support for broadband-capable 
networks in these unserved rural areas as part of its efforts to close 
the rural-rural divide and direct funding to parts of rural America 
where it is most needed. Specifically, the Commission will provide 
support through ``a combination of competitive bidding and a new 
forward-looking model of the cost of constructing modern multi-purpose 
networks.'' Using the cost model to ``estimate the support necessary to 
serve areas where costs are above a specified benchmark, but below a 
second `extremely high-cost' benchmark,'' the Commission will offer 
each price cap local exchange carrier (LEC) ``a model-derived support 
amount [for a period of five years] in exchange for a commitment to 
serve all locations in its service territory in a state that, based on 
the model, fall within the high-cost range and are not served by an 
competing, unsubsidized provider.''
    2. The Commission delegated to the Wireline Competition Bureau 
(Bureau) ``the task of selecting a specific engineering cost model and 
associated inputs that meet the criteria specified'' by the Commission. 
Consistent with the approach taken by the Commission when it 
implemented a forward-looking model known as the High-Cost Proxy Model 
(HCPM) to determine support amounts for non-rural carriers in the wake 
of the implementation of the Telecommunications Act of 1996, the 
Bureau's plan is to adopt a model to estimate forward-looking costs in 
two separate orders. In this first order, we primarily address the 
model platform, which is the basic framework for the model consisting 
of key assumptions about the design of the network and network 
engineering. We also address certain framework issues relating to 
inputs.

II. Discussion

    3. This order focuses on the platform components of the cost-to-
serve module. As detailed below, and consistent with the approach 
previously taken by the Commission in adopting its prior forward-
looking model for universal service support, we adopt a model platform 
that will allow the Bureau to estimate the full average monthly cost of 
operating and maintaining an efficient, modern network. Specifically, 
the model will begin by estimating all capital and operating expenses 
associated with a modern network. Those variously-timed expenditures 
will be converted to an average monthly cost, as described below. 
Because providers' support will be based on this average cost for five 
years, while many components of an actual network have much longer 
lives, using this average cost approach will not compensate providers 
for the full cost of a network within the five year Phase II timeframe. 
It will, however, estimate the cost of providing service in the way 
that best approximates the discipline of a competitive market.
    4. The average costs will be based on an efficient modern network, 
rather than a less efficient legacy network supplemented with 
incremental upgrades over time. That is, consistent with the 
Commission's directive to adopt a ``forward-looking'' approach, we will 
model the costs as if all providers were able to claim the efficiency 
advantages of a modern green-field build, rather than attempt to model 
costs of upgrades and inefficiencies associated with maintaining and 
upgrading legacy networks piecemeal (a ``brown-field'' approach). 
Although some commenters have argued that a ``brown-field'' approach 
would result in lower modeled costs, we find that this is only because 
the various brown-field estimates in the record have each improperly 
excluded certain costs.
    5. Following the assumption of a maximally efficient modern 
network, modeled costs will be based on an IP-based FTTP network of a 
wireline telecommunications provider, capable of providing both voice 
and broadband. Customer locations, both residential and business, will 
be placed in individual census blocks, and a network topology will be 
constructed to serve all of those locations. Consistent with the 
Commission's approach when it developed the HCPM in the 1990s, the 
model will calculate necessary interoffice transport (i.e., middle 
mile), which, in a modern network, would connect all central offices 
with internet gateways. The model will provide the capability to vary 
certain input values relating to the cost of construction based on 
physical geography within a given state. Costs will be calculated on a 
census block level.
    6. Although a large number of important decisions regarding input 
values and other issues remain, preliminary estimates based on the 
current version of the CAM suggest that this better calibrated approach 
results in more reliable cost estimates of an efficient provider. Using 
the platform decisions adopted in this Report and Order, we estimate 
that per-location costs for the highest cost areas (those potentially 
available for Phase II funding) are roughly 20-25 percent lower in the 
current version of the CAM than in the cost model submitted by the ABC 
Coalition prior to the Commission's adoption of the USF/ICC 
Transformation Order. The work done to date thus has modified aspects 
of the CQBAT model that led to an overstatement of the costs of 
providing broadband-capable infrastructure in Phase II areas.

A. Threshold Model Design/Platform Issues

1. General Approach to Cost Estimation
    7. Consistent with Commission precedent, the model platform that we 
adopt today will calculate a levelized cost that represents an estimate 
of the average monthly forward-looking cost of an efficient provider. 
Those costs include both capital and operating expenses. Recovery for 
each asset class, for example, poles, conduit, etc., will be spread out 
evenly over the useful life of the asset class according to empirical 
estimates of the rate at which elements of the asset class are retired. 
Costs will be levelized to produce a constant monthly cost throughout 
the life of each asset, which in many cases may exceed 20 years or 
more. Because a significant driver of network costs are assets with an 
accounting lifetime of 20 years or more, such as loop plant, the 
levelized cost calculated by the model will provide recovery for only a 
portion of the cost of the network over the five-year term of Phase II. 
In other words, as discussed more fully below, the model platform will 
calculate costs assuming that the supported network will retain 
significant value at the end of the five-year term of Phase II support.
2. Network Design
    8. In the USF/ICC Transformation Order, the Commission delegated to 
the Bureau the authority to select the specific engineering cost model, 
including the modeled network architecture. The Commission indicated 
that the Bureau's ``ultimate choice of a greenfield or brownfield 
model, the modeled architecture, and the costs and inputs of that model 
should ensure that the public interest obligations are achieved as 
cost-effectively as possible.''
    9. In the Model Design PN, 77 FR 38804, June 29, 2012, the Bureau 
sought comment on, among other things, the choice of a green-field or 
brown-field model; whether the model should estimate the costs of FTTP 
or Digital Subscriber Line (DSL) (including Fiber-to-the-Node (FTTN)) 
technology; and what terminal value to assign to the modeled network 
(e.g., book value or zero value). The Bureau also sought comment on 
whether the model should estimate the total costs of serving the entire 
service area so that shared costs

[[Page 26271]]

may be distributed between areas that are eligible and ineligible for 
support, or estimate only the standalone costs of areas eligible for 
support; how shared network costs should be distributed to the census-
block (or smaller) area; and whether the model should calculate support 
for areas to which broadband has already been deployed or only for 
unserved areas.
    10. As discussed below, we conclude that the Connect America Cost 
Model will be a green-field FTTP model with the terminal value of the 
network at the end of the five-year term determined by the book value 
of the assets. As explained in the Model Design PN, the issues of 
network technology (e.g., FTTP or DSL), design (green-field or brown-
field) and terminal value (e.g., book value or zero value) are 
interrelated. We conclude that using a green-field FTTP model paired 
with book value is the best choice for estimating the most efficient 
forward-looking cost of providing service over a voice and broadband-
capable wireline network in price cap areas.
a. Green-field vs. Brown-field
    11. We find that using a green-field model is more appropriate than 
using a brown-field model, for three principle reasons. First, a green-
field model is consistent with Commission precedent, including the USF/
ICC Transformation Order. Second, a green-field model provides an 
estimate of costs that creates appropriate incentives to invest--that 
is, it best approximates the discipline provided by a competitive 
market. And finally, a green-field model can be implemented in a 
straightforward and timely manner. Contrary to some commenters' 
assertions, we conclude that a green-field model does not over-
compensate providers. Indeed, a levelized green-field approach is 
likely to result in no more support than a properly calculated 
levelized brown-field approach because it approximates the average 
long-run cost of an efficient modern network optimized for voice and 
broadband, rather than the average long-run cost of a less efficient 
legacy voice network plus broadband upgrades.
    12. First, a green-field approach is consistent with Commission's 
determination in the USF/ICC Transformation Order in that it would use 
a forward-looking cost model to identify price cap areas eligible for 
Connect America Phase II support, as well as other Commission 
precedent. A green-field approach is forward-looking because it 
estimates the cost of the ongoing provision of specific services by 
developing a hypothetical efficient, modern network to calculate the 
minimum cost of providing such services now and in the future, given 
current technology and input costs. It does not take into account 
historic costs or whether the carrier historically recovered its 
earlier investments in the existing network, other than what is 
provided through the monthly levelized cost stream going forward.
    13. A green-field model is consistent with the approach taken by 
the Commission in developing and adopting its previous voice cost 
model, the HCPM. Even though legacy voice networks existed throughout 
the nation at that time, often including less-efficient older 
technologies or inefficient network routing, the Commission concluded 
that the appropriate way to determine support was to estimate the cost 
of an efficient modern network to provide voice service, assuming only 
the existence of incumbent central offices and current wire centers 
(referred to as the ``scorched node'' approach). Consistent with this 
longstanding precedent, the green-field approach we adopt will 
calculate (1) the minimum, levelized cost of a voice and broadband-
capable network today, using current, rather than historic, 
technologies and prices, and (2) the minimum costs of continued 
provision of voice and broadband services on that network, including 
the costs of maintaining the network's capabilities in each year going 
forward.
    14. Second, consistent with longstanding Commission precedent, we 
adopt a green-field approach because it estimates costs in a manner 
that provides appropriate forward-looking incentives to invest. A 
forward-looking approach to cost modeling does not ask whether or to 
what extent carriers' have recovered their costs from past investments. 
Instead, a forward-looking model calculates costs at a level expected 
to recover all network costs over the long term, accounting for 
investment risk and anticipated demand, comparable to a market with 
sustainable competition. In such a regulatory environment, recipients 
of support should receive appropriate forward-looking compensation for 
risks that are intended to mimic the risks that competitive firms face 
in markets where subsidies are not provided.
    15. We are not persuaded by the argument that using a green-field 
model for Connect America Phase II will over-compensate the price cap 
carriers over a five-year period because the actual replacement costs 
incurred over the next five years may in some instances be less than 
the green-field levelized cost. The Commission previously has concluded 
that forward-looking economic costs--not actual costs--are the proper 
framework for determining universal service support, and the Commission 
specifically directed the Bureau to use a forward-looking approach in 
the USF/ICC Transformation Order. Moreover, whether an individual price 
cap carrier would actually spend more or less than model-determined 
support over the course of the five-year term will depend on where the 
individual price cap carriers that make a state-level commitment are in 
their respective investment cycles. Carriers have made and must 
continue to make investments that last substantially longer than five 
years, incurring costs that do not, year-by-year, match their revenues 
(even for the case of commercially-viable investments). Those carriers 
that must undertake a relatively high level of asset replacement may 
therefore face higher costs than the modeled costs. Others will face 
lower costs. Allowing monthly recovery of the model's levelized cost 
means, on average, all carriers will earn an amount that would allow 
them to maintain the specified levels of service going forward over the 
longer term.
    16. Indeed, a green-field model may calculate costs lower than 
actual costs because it may overstate the degree to which carriers are 
able, in practice, to optimize their network. Carriers do not have the 
luxury of building their networks from the ground up to meet today's 
demand. Rather, they augment their networks piecemeal, with each 
upgrade subject to past investment decisions that may not always have 
been based on accurate forecasts of demand and technology developments. 
Consistent with Commission precedent in adopting a green-field model to 
estimate the forward-looking cost of voice service, we find that, on 
balance, the green-field approach should provide a reasonable overall 
approximation of costs for Phase II implementation.
    17. Third, a forward-looking green-field approach can be 
implemented in a straightforward and timely manner, allowing the 
fastest possible deployment of new broadband in price cap territories. 
Each version of the CAM released to date contains the capability to 
estimate the costs of a green-field FTTP network. Moreover, the ABC 
Coalition previously submitted into the record of this proceeding more 
than a year ago a green-field model. As a result, the public and Bureau 
staff have had ample opportunity to analyze the attributes and the 
usefulness of a green-field model for implementing the Commission's 
universal service policies. These submissions build on a substantial 
history of use of green-field

[[Page 26272]]

models in a variety of regulatory contexts. In contrast, as discussed 
in more detail below, we are not satisfied that any version of the CAM 
has yet provided a reasonable way of estimating brown-field costs. We 
therefore conclude that adopting a green-field model platform now, so 
that parties can focus their attention on input values, will facilitate 
the timely conclusion of the Phase II cost model development process, 
and thereby accelerate the deployment of broadband-capable networks to 
unserved Americans.
    18. In contrast to a green-field approach, there are significant 
drawbacks to a brown-field approach. First, notwithstanding arguments 
to the contrary, a brown-field approach is not entirely forward-
looking. It represents a hybrid approach that falls between a true 
forward-looking approach, which a green-field model approximates, and a 
historic cost approach. A brown-field approach assumes existing 
infrastructure as of a point in time and adds the ongoing costs of this 
infrastructure to the cost of additional network upgrades necessary to 
provide a desired set of services in the future. As an example, 
existing fiber transport, and/or the last few thousand feet of copper 
terminating at an end-user location, could potentially be used to 
supply voice and broadband service. For these portions of the network, 
a brown-field approach would estimate costs based on the existing 
network facilities, rather than on a modern, efficient network.
    19. Second, there would be serious practical hurdles to overcome 
before we could implement such an approach. The Bureau considered two 
possible ways to implement a brown-field approach: one that identifies 
those assets actually in place, and then considers the incremental cost 
of making that existing network broadband-capable, and another that 
produces a hypothetical model of a voice-only network, and then 
considers the incremental cost of adding broadband capability to that 
network. Both approaches raise significant practical difficulties.
    20. The first approach to brown-field modeling has significant 
backward-looking elements not present in a green-field approach and is 
substantially more complicated than a green-field approach. In 
particular, this brown-field approach would require identification of 
the specific existing network assets that are assumed to be retained. 
Thus, we would need to develop a model that accurately represents the 
existing network infrastructure and determine what parts of the 
existing network can be used; we then would estimate the cost of any 
incremental upgrades required to meet the Commission's service 
obligations going forward, including the costs that would be necessary 
going forward to maintain the entire network's capabilities. In 
contrast to a green-field approach, this brown-field approach would 
require a substantial backward-looking exercise in which those 
components of the network that already exist must be identified and 
located, and characterized in terms of their age and capabilities going 
forward (e.g., gauge of copper wire, etc.). Additionally, this brown-
field approach would model the forward-looking costs of augmenting the 
existing network to make it broadband-capable. In comparison to a 
green-field approach, such an exercise would likely require far more 
data, because existing network investments would need to be catalogued, 
and it would present a more complex cost optimization, because the 
optimal network would be designed to account for the elements of the 
existing network that would be efficient to keep. This would be 
particularly complex, requiring the Bureau to make decisions about what 
assets should be retained, and what should be replaced.
    21. The second approach to brown-field modeling would be to 
estimate the green-field cost of the existing network and then estimate 
the incremental cost of making that network fully broadband-capable. 
This approach avoids the difficulties of cataloging existing network 
infrastructure, and of having to optimize taking historical investment 
decisions into account, but has the peculiarity of using a hypothetical 
optimized green-field cost model to estimate the cost of an existing 
network. While such an approach would limit the amount of data that 
would be required and would avoid some of the backward looking nature 
of the first approach, it only obliquely meets the ostensible objective 
of a brown-field approach, which is to assume that all existing 
infrastructure will be retained, with upgrades to make that network 
fully broadband-capable. In addition, taking this approach still would 
require the Bureau to make a substantial number of assumptions about 
the age and quality of existing assets and therefore significantly 
broaden the reasonable range of outcomes, compared to a green-field 
model. The Bureau first would have to determine which hypothetical 
assets are assumed to exist as the starting point, and then model the 
investments required to make that network capable of supplying 
broadband. In contrast, the green-field approach requires only modeling 
a current generation, modern network.
    22. We are not persuaded by ACA's argument that a brown-field 
approach would result in cost estimates substantially lower than a 
green-field model, and therefore expand the number of unserved homes 
that could receive broadband given the fixed budget for Phase II. ACA's 
attempts to estimate brown-field costs exclude some costs that should 
be included in a proper brown-field model. In response to the Model 
Design PN, ACA argues that ``the CQBAT model [submitted by the ABC 
Coalition] includes functionality to allow for the modeling of a 
brownfield DSL build-out.'' In fact, that function in CQBAT simply 
eliminated all capital expenditures for certain network elements, such 
as copper loops. ACA acknowledged that CQBAT did not adequately account 
for the operating expenses associated with the copper portion of the 
loop, copper replacement in cases where plant needs to be replaced, and 
loop conditioning costs on a granular level, but argued that adding 
these functionalities to the model should not be difficult. 
Subsequently, in October 2012, ACA filed additional estimates of brown-
field costs based on CQBAT runs under various scenarios, each of which 
excluded certain capital costs, such as copper loops, necessary for 
providing ongoing service from the calculations, and we find it would 
be appropriate to take these costs into account in a brown-field model. 
Therefore, we are not persuaded that the calculations provided by ACA 
appropriately reflect the cost estimates of a brown-field approach, and 
conclude that ACA does not provide a reliable estimate of the number of 
homes that would become served by broadband in Phase II.
    23. While CAM version 3.0 contains a feature that attempts to 
approximate brown-field costs, we still do not believe this approach 
fully corrects the issues associated with the CQBAT model's brown-field 
approach. This ``brown-field adjustment'' was intended to capture the 
replacement cost of existing plant as those assets are retired, but not 
to capture the cost of existing plant that is continued to be used to 
provide the existing services. That is, the calculation captures the 
cost of providing service when an asset is retired, but not of 
providing service until that point. We therefore conclude that 
additional costs would have to be added to this brown-field adjustment 
to properly take into account the existing assets necessary to provide 
and maintain voice and broadband services on an ongoing basis. In fact, 
we now are convinced that if all

[[Page 26273]]

these costs are properly accounted for, brown-field modeling should 
provide cost estimates no lower than, or potentially higher than, a 
green-field approach.
    24. In sum, we find that a green-field cost approach is the 
preferable approach to calculate the cost of a forward-looking network. 
It is more consistent with the Commission's directive and prior 
precedent, and we conclude that there are no persuasive arguments that 
using a green-field approach would result in overpayments to the price 
cap carriers. In contrast, development of a suitable brown-field model 
would likely take a considerable amount of additional time and delay in 
implementation of Connect America Phase II, because it is a much more 
complex undertaking with little precedent to guide staff efforts.
b. FTTP
    25. We also conclude the best approach to meet the Commission's 
directive that we adopt a forward-looking cost model is to estimate the 
costs of a FTTP network rather than a twisted copper pair DSL network. 
As explained in the Model Design PN, a DSL network ``is only forward 
looking from the perspective of decisions made a decade or more in the 
past,'' and ``has higher expected operating expenses and is more likely 
to require significant additional investment to make faster broadband 
offerings available.'' Although some price cap carriers may choose to 
extend broadband to unserved areas in the near term by shortening 
copper loops, rather than deploying FTTP, the most efficient wireline 
technology being deployed today in new builds is FTTP. Network 
construction costs are essentially the same whether a carrier is 
deploying copper or fiber, but fiber networks result in significant 
savings in outside plant operating costs over time. If an efficient 
carrier were to design a new wireline network today, it would be an all 
Internet protocol (IP) fiber network, not a circuit switched copper 
network, because such a network would be cheaper and more scalable over 
time. Indeed, an IP fiber network would be the appropriate choice for a 
wireline network even if there were no service obligation to extend 
broadband. Therefore, FTTP is more consistent with a forward-looking 
approach.
c. Methodology for Determining Terminal Value
    26. The model platform that we adopt today provides capital 
recovery through what is termed depreciation. We conclude that the 
model should determine the terminal value of the network based on 
``book value'' calculated as the difference between investment and 
economic depreciation, which takes into account the economic life of 
the equipment and infrastructure. Specifically, the model will 
calculate book depreciation expense based on equal-life-group 
methodologies, using Gompertz-Makeham survivor (mortality) curves and 
projected economic lives. The model will adjust the survivor curves, 
however, so that the average lifetime of the asset falls within the 
range of expected accounting lifetimes authorized by the Commission. 
This approach is consistent with the methodology used in the 
Commission's previous cost model used to determine support amounts for 
the non-rural LECs, HCPM, and supported in the current record.
    27. In the virtual workshop, the Bureau sought comment on whether 
any of the projected lives used in HCPM are outdated and should be 
modified. The ABC Coalition recommended that the Bureau uses the same 
economic lives for assets as HCPM, while ACS suggested the Commission's 
economic lives are too long and should be updated. Based on our review 
of the record, we now conclude the model will utilize the same economic 
lives for assets as specified by the Commission previously when it 
adopted the HCPM, when determining the monthly cost of capital 
investments. As the ABC Coalition notes, for more than a decade, these 
economic lives for assets have been widely used in cost models in state 
regulatory proceedings. We are persuaded that it would be 
administratively burdensome to establish new values, which would 
unnecessarily delay implementation of Connect America Phase II. We 
recognize that to the extent economic lives are overstated for 
particular assets that would result in a systematic understatement of 
costs, but no party has submitted any evidence in the record 
demonstrating that this effect would result in a material change in 
support levels thwarting achievement of the Commission's universal 
service objectives.
    28. As the Bureau explained in the Model Design PN, the annual cost 
and support values are highly dependent on the terminal value, because 
the five-year support period is much shorter than the average lifetime 
of all of the asset classes in the model. At the end of five years, a 
FTTP network would have significant commercial value. Because 
estimating commercial value at the end of the five-year term would 
require making a number of assumptions about the evolution of 
technology and the marketplace, we conclude that using book value is 
the best approach. Using a terminal value of zero, as some parties 
advocate, would permit carriers to recover the entire cost of the 
network over five years, and assume the network had no future 
commercial value. We find that to be an unreasonable assumption and 
would over-compensate carriers, so we decline to use a zero terminal 
value in CAM.
3. Assigning Shared Network Costs
    29. The Commission concluded in the USF/ICC Transformation Order 
that it would use a forward-looking model capable of determining ``on a 
census block or smaller basis, areas that will be eligible for CAF 
Phase II support.'' As a threshold matter, we conclude that the model 
will calculate costs at the census block level, except in those 
instances where a census block is split between two service providers. 
The model will calculate costs at a significantly more granular level 
than the Commission's prior forward-looking model, HCPM, which 
calculated costs at the wire center level. There are approximately 11 
million census blocks, compared to approximately 20,000 wire centers. 
We therefore conclude that calculating costs at the census block level 
will be sufficient to meet the Commission's objective of targeting 
support to high cost areas.
    30. The Commission also concluded that ``it would be appropriate to 
exclude any area served by an unsubsidized competitor'' that meets the 
Commission's initial performance requirements. Most costs in a network 
are shared costs. As a result, the method used to attribute the costs 
of shared plant to eligible and ineligible areas and among census block 
or smaller areas will have a significant effect on the relative cost of 
serving different areas.
    31. In the Model Design PN, the Bureau asked how shared network 
costs should be assigned between eligible and ineligible areas. 
Specifically, the Bureau asked whether costs should be modeled for the 
entire service areas and then allocated between eligible and ineligible 
areas or costs should be estimated only for the eligible areas on a 
standalone basis.
    32. We conclude that the Connect America Cost Model will model the 
total cost of serving an entire service territory within a state, 
rather than calculating the standalone costs of serving only eligible 
census blocks, and then, as more fully discussed below, allocate the 
shared costs between eligible and ineligible census blocks. Modeling 
the costs associated with a complete network (i.e., including both

[[Page 26274]]

eligible and ineligible census blocks) and then assigning shared costs 
between the eligible and ineligible census blocks has significant 
benefits. First, it more accurately depicts an economically efficient 
network and provider. An economically efficient network would cover all 
or most locations in a given service territory, rather than only 
serving a small subset of locations that lack broadband. Indeed, 
building a network to only serve those locations that lack broadband 
would likely result in higher cost estimates for those areas than 
otherwise would be the case, because the service provider would have to 
deploy less than optimal routing to reach those pockets of customers 
that are in eligible census blocks. Moreover, an economically efficient 
provider would not generally cede a large fraction of customers within 
its service territory to unsubsidized competitors; rather, it would 
seek to compete in those areas where a positive business case exists. 
Modeling the entire network and then making adjustments to determine 
support for particular census blocks where there is no unsubsidized 
competitor is a reasonable way to proceed. Finally, the Bureau notes 
that this approach has broad support in the record. For these reasons, 
the Bureau finds that it is appropriate for the Connect America Cost 
Model to model the total cost of serving the entire state, not the 
standalone costs of only serving eligible census blocks, and then 
allocate shared costs between eligible and ineligible census blocks.
    33. In the Model Design PN, the Bureau also asked how to allocate 
shared costs consistent with the requirement in the USF/ICC 
Transformation Order that the model be capable of determining ``on a 
census block or smaller basis, areas that will be eligible for CAF 
Phase II support.'' Shared costs need to be allocated not only between 
eligible and ineligible areas, but among census blocks in eligible 
areas so that the costs of serving each individual census block can be 
estimated. The Bureau sought comment on two potential options: (1) A 
subtractive method, in which the model would estimate only those costs 
to serve eligible areas that are over and above the costs of serving 
the ineligible areas, and (2) a pro rata method, in which costs would 
be assigned to eligible and ineligible areas on some pro rata basis or 
using some other formula. The Bureau indicated a general preference for 
the subtractive method, but acknowledged that the computational 
complexity of the subtractive method might make it difficult or 
impossible to implement in practice. Subsequently, as part of the 
virtual workshop, the Bureau sought comment on a possible approach to 
the subtractive method.
    34. Based on our review of the record and our development of CAM to 
date, we now conclude that the model will use a pro rata method for 
assigning shared costs. The Bureau gave significant consideration to a 
subtractive approach for assigning costs, and there was support in the 
record for such an approach. Ultimately, however, we find that the 
computational complexity and the novelty of the subtractive approach 
renders it too difficult to implement. The cost-causation approach 
contained in the current version of CAM (CAM version 3.0) provides a 
practical method of assigning shared costs in a reasonable manner. 
Specifically, the model will use a ``cost causation'' method that 
assigns a fraction of the costs associated with a shared network 
facility according to the relative number of customers in each area 
using the facility. Using cost causation to allocate costs is 
consistent with the current High-Cost Proxy Model, the model submitted 
by the ABC Coalition and the National Broadband Plan modeling. For that 
reason, the Bureau concludes that the cost-causation approach for 
sharing costs between eligible and ineligible census blocks is 
appropriate for use in the Connect America Cost Model.
4. Calculation of Costs for Price Cap Carriers' Currently Served 
Locations
    35. We conclude the model platform will estimate the costs of 
serving locations irrespective of whether they are currently provided 
broadband by the ILEC. We find that this approach is consistent with 
the Commission's goals and directives in the USF/ICC Transformation 
Order. While the Commission sought to ``extend[] broadband to millions 
of unserved locations,'' it also recognized the importance of 
``sustaining existing voice and broadband services.'' We therefore 
reject the Joint Michigan Competitors' claim that the model should 
exclude broadband-served areas because the Commission's focus is on 
deploying broadband to unserved areas, and ACA's claim that broadband-
served areas should only receive ongoing support for maintenance and 
operational expenses--not for capital expenses.
    36. We will presume, consistent with the Commission's direction and 
predictive judgment, that locations that exceed a specified cost 
benchmark, which will be determined in a future order, will require 
support on an ongoing basis based on the total levelized cost of 
sustaining existing voice and broadband services at reasonable end-user 
rates. As we noted in the Model Design PN, carriers may have deployed 
broadband in certain areas based on past universal service support and 
intercarrier compensation revenues. Even where carriers may have 
deployed broadband to fulfill merger commitments, because they received 
another source of funding, or for other reasons, such carriers still 
may require funding to sustain the previous broadband deployment. And 
as we explained above, providing support for only maintenance and 
operational expenses would not cover the entire cost of sustaining 
service.
    37. Moreover, treating locations currently served by the incumbent 
differently from completely unserved locations is inconsistent with a 
using a green-field approach to estimate the costs of an efficient 
modern network optimized for voice and broadband. Treating served and 
unserved locations differently would require modeling actual historical 
network deployment, rather than an efficient forward-looking network. 
This is functionally similar to the first approach to brown-field 
modeling, which would require an extensive data collection, while 
unnecessarily delaying implementation of Phase II.
    38. Accordingly, we reject commenters' claims that areas already 
served by broadband do not require ongoing support, (or only require 
limited ongoing support), and we conclude that the model will include 
and calculate ongoing support for high-cost locations above the cost 
benchmark that are both served and unserved by broadband. We note that 
this is consistent with the Commission's approach when it adopted HCPM; 
it calculated the cost of an efficient provider to provide voice 
service throughout the territory of a non-rural LEC, even though those 
LECs already provided voice.
5. Treatment of Non-Contiguous United States
    39. The Commission has ``direct[ed] the [Bureau] to consider the 
unique circumstances of [Alaska, Hawaii, Puerto Rico, the U.S. Virgin 
Islands and Northern Marianas Islands] when adopting a cost model.'' 
The Commission further directed the Bureau to determine whether the 
cost model provides sufficient support to these areas, and if, in the 
Bureau's determination, the model does not provide these areas with 
sufficient support, the Commission granted the Bureau the discretion to 
``maintain existing support levels, as modified in

[[Page 26275]]

this Order, to any affected price cap carrier, without exceeding the 
overall budget of $1.8 billion per year for price cap areas.'' The 
Bureau has sought comment to further develop the record on these two 
options for areas outside the contiguous United States, and the 
associated service obligations.
    40. The decisions we make herein do not prejudge whether 
modifications to the model platform or input values should be made with 
respect to the non-contiguous United States, or support levels for 
those areas should be frozen. We will address those arguments at a 
future date.

B. Customer Locations and Outside Plant Design

    41. As the Commission recognized when it adopted the model platform 
for HCPM, outside plant--namely, the loop facilities between switches 
and the customer premises--constitutes the largest portion of total 
network investment, and the design of outside plant facilities depends 
heavily on the location of customers. Business customer information is 
important not only for locating business customers, but also for 
scaling the network infrastructure to ensure that the costs of shared 
resources are appropriately shared among all users. The placement of 
customer locations thus is an important element of the CAM platform.
1. Customer Locations
    42. In the Model Design PN, the Bureau proposed to use a commercial 
data set for residential customer location data, but also sought 
comment on two alternatives: Using official government census data, 
which would provide the number of housing units in a census block but 
no geocodes, and collecting actual customer location data from 
providers. For business locations, the Bureau proposed using government 
data from the U.S. Bureau of Labor Statistics (BLS) Economic Census, 
but also sought comment on using commercial data sources. The Bureau 
sought further comment via the CAM virtual workshop on methods for 
determining customer locations.
    43. Few commenters offered any comments about customer locations 
data. In the absence of actual geocode information, the ABC Coalition 
supports using a methodology that uses a combination of data sources to 
estimate the number of customer locations by zip code and then 
distribute those locations randomly along roads in the census block. 
The only commenter suggesting an alternative source for customer 
location data is the National Association of State Utility Consumer 
Advocates (NASUCA), which proposed the Commission obtain E911 databases 
and translate the addresses into geocodes that can be used in the cost 
model. If the Commission uses census data, NASUCA argues that these 
data should be augmented by geocoded data provided by the carriers in 
census blocks above a certain size.
    44. We adopt a model platform that will use a combination of 
commercial data set (GeoResults Q3 2012) and census data to determine 
residential and business locations. Specifically, the model will use 
GeoResults Q3 2012, which provides an address-based residential data 
set of households. To the extent there are discrepancies between the 
location counts from GeoResults and 2011 census housing unit estimates, 
the GeoResults count will be adjusted upward or downward to conform to 
the census, with the records for the requisite number of locations to 
be added or subtracted selected in a random manner. We conclude the 
model also should use GeoReults for business location data, because 
those data are more current and include more businesses than the BLS 
economic census data. GeoResults also provides a national building 
file, which is used to identify buildings that have both residential 
and business customers. The model will use additional data sources to 
identify the locations of community anchor institutions and cell 
towers.
    45. The CAM will use geocoded locations wherever possible, and 
place locations that cannot be geocoded randomly along the roads within 
the census block. This is an improvement upon the approach previously 
taken by the Commission when it implemented HCPM. By using geocoded 
data where available, the model will estimate with greater precision 
the amount of feeder plant necessary to reach all locations, which 
should result in more accurate cost estimates than the prior forward-
looking cost model utilized by the Commission, which assigned all 
locations randomly along roads using Topologically Integrated 
Geographic Encoding and Referencing (TIGER) data.
    46. We find that using these data is preferable to using E911 data, 
supplemented by carrier-provided data, as suggested by NASUCA. First, 
NASUCA does not specifically identify the E911 database(s) that it 
contends should be used. Moreover, an approach based on E911 databases 
would potentially introduce inconsistencies in the model across states, 
because each state and, in many instances depending on state and local 
regulations, individual Public Safety Answering Points (PSAPs), are 
responsible for their E911 databases, and these databases differ in 
methodology, completeness and accuracy. Using a consistent methodology 
throughout the nation will lessen the likelihood of inconsistencies in 
cost estimates among states, which could skew the relative distribution 
of support in unknown ways among the states.
    47. We conclude that it is not feasible to develop a model platform 
that incorporates actual customer locations for all locations. There is 
no publicly available source of nationwide geocoded location data, and 
commercial data sources do not provide geocodes for all locations. Even 
if the price cap carriers provided the Commission with their geo-coded 
customer database, or address list if they do not have geo-coded 
customer locations, these data bases would only include the incumbent 
local exchange carriers' customers and not all the housing units in the 
census block. Doing a mandatory data collection that collected customer 
location information from cable operators and other non-incumbent 
providers would be a significant Commission undertaking, and it would 
impose burdens on those providers. Nothing in the record before us 
suggests that the incremental improvement in precision of locations 
that would result from such a mandatory data collection would be worth 
the costs in terms of burden on both the Commission and outside 
parties. Accordingly, we conclude that GeoResults, trued-up with Census 
data for residential locations, is the best source of customer 
locations because of the number of locations that are geocoded. The 
final model will use the methodology in CAM version 3.0 for assigning 
included locations that cannot be geocoded along road segments.
2. Clustering
    48. We adopt a clustering approach that uses road-based routing to 
determine the maximum size of the clusters. Once customer locations 
have been identified, the model must determine how to group and serve 
those customers in an efficient and technologically reasonable manner. 
Consistent with past Commission precedent for forward-looking cost 
models, the objective is to group customers into serving areas in an 
efficient manner to minimize costs, while maintaining a specified level 
of network performance equality. Like HCPM, our model platform will 
design clusters consistent with engineering constraints, grouping 
customers so that they are no further away than allowed by network 
design to deliver services

[[Page 26276]]

meeting the Commission's performance requirements. CAM will improve the 
approach previously used by the Commission in HCPM, however, as it will 
use road-based routing to determine the maximum size of the clusters. 
Thus, clusters defined by CAM are likely smaller, but more realistic 
estimates of cluster size, resulting in more accurate cost estimates. 
By using road segments in clustering, the CAM model avoids the problem 
of having the length of some loops modeled along roads exceed the 
maximum loop length necessary to provide service meeting specified 
standards. The ABC Coalition supported this approach, and no party 
objects to using this clustering methodology for modeling costs in the 
contiguous United States. We conclude that the model will include the 
clustering methodology currently incorporated into CAM version 3.0.
3. Routing
    49. We adopt the routing methodology used in CAM, which builds 
plant along roads and uses a minimum spanning tree algorithm. Although 
HCPM allowed for minimum spanning-tree optimization of routes, it did 
not use the road network. CAM, on the other hand, represents an 
enhancement to the approach taken by the Commission in developing a 
forward-looking model in the 1990's, as it lays loop plant along actual 
road segments and utilizes a spanning tree algorithm to find the lowest 
cost route to serve all customer locations along road paths. The ABC 
Coalition supported this approach, and no party objects to using this 
routing methodology for modeling costs in the contiguous United States. 
We conclude that the model platform will include the CAM version 3.0 
algorithm for routing loop plant and feeder network.
4. Sizing Network Facilities
    50. We adopt a model platform that will size network facilities 
such that there is sufficient capacity at the time of peak usage. The 
model platform accomplishes this by ensuring that the size of each link 
in the network is sufficient to support peak usage busy hour offered 
load, taking into account subscriber usage capacity (GB/month/
subscriber) as well as throughput (Mbps) and take-rate. This method is 
basically the same approach that was taken in the National Broadband 
Plan modeling. Because voice is the supported service, the model also 
takes into account peak demands associated with voice service in the 
sizing calculations. No party objects to this general approach to 
network sizing. The ABC Coalition agrees that sizing broadband 
facilities based on throughput required at the time of peak usage is 
reasonable, while noting that the peak demands associated with voice 
service should be included in the sizing calculations if voice 
capability is to be added to the model. We will address the specific 
input values the model will use for busy hour under load in a future 
order.

C. Switching and Interoffice Facilities

1. Voice Capability
    51. In the USF/ICC Transformation Order, the Commission determined 
that ``voice telephony service'' is the service supported by federal 
high-cost universal support. All recipients must offer voice telephony 
service. In addition, as a condition of receiving support, all 
recipients must offer broadband service.
    52. We adopt a model platform that estimates the cost of an IP-
enabled network capable of providing voice service. The cost is modeled 
on a per-subscriber basis and takes into account the cost of hardware, 
software, services, and customer premises equipment to provide carrier-
grade Voice over Internet Protocol (VoIP) service. No party objects to 
this general methodology for including voice capability to serve the 
contiguous United States, and the ABC Coalition supports this approach. 
We conclude that the appropriate forward-looking way to model a network 
today that provides voice service is to design an all-IP network. The 
specific inputs used to calculate the per-subscriber cost will be 
addressed in a future order.
2. Interoffice Facilities
    53. We adopt a model platform that ties central offices to the 
nearest tandem location, ties tandems together, and uses efficient 
routing paths for all connections, using information from the Local 
Exchange Routing Guide database. The model platform assumes Ethernet-
based fiber connections among wire centers and between wire centers and 
tandem switches, including the use of wave division multiplexing 
gateways. Additionally, the model platform connects each hierarchy to 
the nearest (lowest cost) Internet access point regardless of 
ownership. The model platform also uses routing along roads to 
determine the cost of deploying fiber to make connections, and includes 
Broadband Remote Access Services and/or gateway costs. No party objects 
to this general approach for the contiguous United States, and the ABC 
Coalition supports this approach. This is consistent with the HCPM, 
which also included the middle mile costs of providing service. We will 
address cost inputs related to interoffice transport in a future order.

D. Framework for Capturing Variations in Cost

    54. As discussed more fully below, the CAM will utilize differing 
assumptions for certain input values based on three geographic density 
zones, and will adjust certain input values for labor and materials 
based on the three-digit zip code.
1. Plant Mix Based on Density Zone
    55. The cost of a modern broadband network varies significantly 
based on the type of infrastructure used to deploy the wires--
specifically whether the wires are underground, buried or aerial. Most 
networks rely on all three types of plant in varying degrees, with the 
precise mix of plant dependent on many factors. A model used to 
estimate the costs of deploying a network must therefore make 
assumptions regarding the mix of plant used in the network.
    56. We adopt a model that assumes that each state is made up of 
three density zones--urban, suburban, and rural. For each density zone, 
the model will assume a specific plant mix for each of three different 
parts of the network--distribution, feeder, and inter-office transport. 
As a result, each state will have a matrix of nine different density 
zone/network component combinations, each of which has its own mix of 
underground, buried, and aerial plant. In addition, the model will 
include a nationwide set of plant mixes for each density zone and 
network component, which may be used in any state for which specific 
inputs may not be available.
    57. The Bureau concludes that this methodology will provide 
sufficiently granular variation in the mix of plant in the entire 
network. We recognize that the HCPM varied cost by nine density zones, 
but no party in the current proceeding objects to using three 
geographic zones. The ABC Coalition notes there was no variation in the 
plant mix between the least dense zones in HCPM, which together 
correspond to the rural zone in the model we are evaluating.
    58. No commenter objected to the general principle that plant mix 
should vary according to density zones, with different plant mix values 
in different areas. Rather, the parties that addressed this issue 
argued there should be a process to document the development of the 
specific input values to be used. The source and specific percentages 
of plant mix to be used in the matrix will be

[[Page 26277]]

determined in a future order addressing inputs.
2. Material and Labor Cost Adjustments Based on Location
    59. We adopt an approach that utilizes uniform input values for 
various capital costs, with adjustments for regional variations in 
labor and material costs. We conclude that this approach to development 
of a forward-looking model is consistent with past precedent. In the 
HCPM Inputs Order, 64 FR 67372, December 1, 1999, the Commission 
determined nationwide default values are generally more appropriate 
than company-specific input values for a forward-looking model. It 
noted that the universal service support mechanism is ``based on the 
estimated costs that an efficient carrier would incur to provide the 
supported services, rather than on the specific carrier's book costs.'' 
It concluded that ``it would be administratively unworkable to use 
company-specific values in the federal nationwide model.'' At the same 
time, however, the Commission recognized the desirability of having 
data that accurately and objectively reflect ``variations in forward-
looking costs based on objective criteria,'' and it stated that it was 
open to additional modifications of inputs in the future. Thus, 
although the Commission did not adjust costs for regional variation in 
adopting HCPM, it expressly recognized that a forward-looking model 
could appropriately recognize variations in cost.
    60. Our forward-looking model will use regional cost adjustment 
factors to capture variation in labor and materials costs by three-
digit ZIP codes. Those regional adjustments are based on data obtained 
from a national survey of the costs of construction in various areas of 
the United States by R.S. Means. The ABC Coalition supports this 
approach of using nationwide average values with regional adjustments, 
noting that the R.S. Means data is widely recognized and used in 
numerous contexts. No party objected to the use of this methodology for 
areas in the contiguous United States.

III. Procedural Matters

A. Paperwork Reduction Act

    61. This document does not contain new or modified information 
collection requirements subject to the Paperwork Reduction Act of 1995 
(PRA), Public Law 104-13. In addition, therefore, it does not contain 
any new or modified information collection burden for small business 
concerns with fewer than 25 employees, pursuant to the Small Business 
Paperwork Relief Act of 2002, Public Law 107-198.

B. Final Regulatory Flexibility Act Certification

    62. The Regulatory Flexibility Act of 1980, as amended (RFA), 
requires that a regulatory flexibility analysis be prepared for 
rulemaking proceedings, unless the agency certifies that ``the rule 
will not have a significant economic impact on a substantial number of 
small entities.'' The RFA generally defines ``small entity'' as having 
the same meaning as the terms ``small business,'' ``small 
organization,'' and ``small governmental jurisdiction.'' In addition, 
the term ``small business'' has the same meaning as the term ``small 
business concern'' under the Small Business Act. A small business 
concern is one which: (1) Is independently owned and operated; (2) is 
not dominant in its field of operation; and (3) satisfies any 
additional criteria established by the Small Business Administration 
(SBA).
    63. In this Report and Order, we adopt a model platform for the 
Connect America Phase II cost model that will calculate a levelized 
cost that represents an estimate of the average monthly forward-looking 
cost of an efficient provider. A model platform is the basic framework 
for the model consisting of key assumptions about the design of the 
network and network engineering. We also address certain framework 
issues relating to inputs for the model. These decisions are not 
anticipated to have a significant economic impact on small entities, 
insofar as the model produces high-cost support amounts for price cap 
carriers and their affiliates that accept the right of first refusal 
pursuant to Connect America Phase II. This is primarily because most 
(and perhaps all) of the affected carriers are not small entities. 
Moreover, the decisions made about the model platform in this Report 
and Order are not anticipated to systematically increase or decrease 
support for any particular group of entities as compared to possible 
alternatives discussed in the record. Therefore, we certify that the 
decisions made in this Report and Order will not have a significant 
economic impact on a substantial number of small entities. The 
Commission will send a copy of the Report and Order, including a copy 
of this final certification, in a report to Congress pursuant to the 
SBREFA. In addition, the Report and Order and this certification will 
be sent to the Chief Counsel for Advocacy of the SBA, and will be 
published in the Federal Register.

C. Congressional Review Act

    64. The Commission will send a copy of this Report and Order to 
Congress and the Government Accountability Office pursuant to the 
Congressional Review Act.

IV. Ordering Clauses

    65. Accordingly, it is ordered, pursuant to the authority contained 
in sections 1, 2, 4(i), 5, 214, 254, 303(r), and 403 of the 
Communications Act of 1934, as amended, and section 706 of the 
Telecommunications Act of 1996, 47 U.S.C. 151, 152, 154(i), 155, 214, 
254, 303(r), 403, and 1302, sections 0.91, 0.201(d), 1.1, and 1.427 of 
the Commission's rules, 47 CFR 0.91, 0.201(d), 1.1, 1.427, and the 
delegations of authority in paragraphs 157, 184, 186, 187, and 192 of 
the USF/ICC Transformation Order, FCC 11-161, that this Report and 
Order is adopted, effective thirty (30) days after publication of the 
text or summary thereof in the Federal Register.

Federal Communications Commission.
Carol E. Mattey,
Deputy Chief, Wireline Competition Bureau.
[FR Doc. 2013-10565 Filed 5-3-13; 8:45 am]
BILLING CODE 6712-01-P