IV. Overview of Cost Models

In order to establish the forward-looking incremental cost of Verizon's UNEs, in compliance with the CCPs and TELRIC guidelines described above, Verizon and JCs each offered a separate cost model. The Commission has typically relied on cost models to estimate the costs to construct a forward-looking local exchange network. This allows the Commission to take a holistic view of the costs to construct a network as an integrated system, with all of the economies of scale and efficiencies derived from modeling an entire network's operations rather than the cost of a piece of equipment in isolation.

A. VZ Cost

Verizon's UNE cost proposals are based on a new model known as VzCost. Verizon describes VzCost as "a state of the art, Internet-based model that allows users to view its underlying data, assumptions, algorithms, inputs and outputs." (Verizon Panel on Recurring Costs, 11/3/03, p. 11.) According to Verizon, VzCost users can easily run scenarios with varying assumptions and data and save these results for later analysis.

VzCost has four basic modules - the investment calculators, the investment generator, the costing generator, and the report and documentation generator. The loop investment calculator within VzCost is known as VzLoop, but VzCost also includes investment calculators for switching, interoffice facilities, and Signaling System 7. (Id., p. 17.)

The investment calculators within VzCost, such as VzLoop, generally begin by determining the costs of materials and equipment that are needed to provide various UNEs. The model then adds costs for engineering, installation and power. Next, annual costs for operations and maintenance, capital carrying costs such as interest, depreciation, and income taxes, as well as an allocation for common overhead costs are modeled through the use of Annual Cost Factors (ACFs) and expense loadings. (Id., p. 10.)

According to Verizon, VzLoop develops a forward-looking loop network by modeling from the ground up all of the facilities required in the forward-looking exchange network, along with the investments for those facilities. (Id., p. 35.) Verizon explains that VzLoop "relies on an unprecedented collection of plant records, designed to ensure that the model takes adequate account of both the constraints of the real world (such as bodies of water and zoning requirements) and the efficiencies of existing rights of way." (Verizon Rebuttal, 11/9/04, p. 1.) Furthermore, Verizon contends VzLoop takes advantage of the context specific judgments of Verizon CA's engineers, and clarifies that "... while VzCost begins with these real-world data, it then adjusts them in significant ways to ensure that the investments and expenses relied upon in the cost studies are forward-looking and rely on the most efficient technologies that are currently available." (Id., p. 2.)

Verizon maintains its cost studies are forward-looking and comport with the FCC's TELRIC principles. It alleges inputs are based on forward-looking assumptions about the network plant mix and improved operational methods using the most efficient, currently available technology mix if the network were rebuilt from the ground up. (Verizon Panel on Recurring Costs, 11/3/03, p. 13.) According to Verizon, VzCost fulfills TELRIC requirements for numerous reasons, including that it attributes costs to specific elements to the greatest extent possible, maps costs to those elements that cause the costs to be incurred, and measures incremental costs of providing a UNE based on the total quantity of the service provided. (Id., p. 23.)

Finally, Verizon claims its general approach is designed to avoid double recovery of costs by identifying the discrete assets dedicated to each element and calculating the investment associated with those assets. Where some assets are used by more than one element, Verizon used modeling methodologies to assign relevant investments to each of the different elements. (Id., p. 15.)

B. HM 5.3

JCs offer the HAI Model, Version 5. 3 (HM 5.3), which they describe as a "bottom-up economic-engineering costing model" that models the local exchange network, assuming existing wire centers, and allows the user to change more than 2100 inputs and assumptions. (JC/Mercer Declaration (Decl.)), 11/3/03, pp. 11 and 27.) HM 5.3 begins with information provided by Verizon on the location of its business and residential customers, then constructs a network to serve the identified locations using granular information as to service demand, network component capacities and costs, and expenses. (Id., p. 10.)

Through this process, HM 5.3 estimates the investments required for each component of the network, and the costs associated with the investments using what JC contend are conservative assumptions regarding applicable costs. These costs include capital carrying costs, plant-specific costs, general support and overhead costs. HM 5.3 assigns these costs to UNEs according to the manner in which these UNEs use different network components, then determines a cost per unit for each UNE. In this manner, HM 5.3 calculates the forward-looking costs Verizon would incur to provide "plain old telephone service," as well as various narrowband, wideband, and broadband loops and broadband interoffice circuits. (Id., p. 4.) JC contend that a key asset of HM 5.3 is that it deals with UNEs associated with all of the components of the local exchange network, and thereby recognizes the relationships and synergies between the different components of the network. (Id., p. 11.)

One of the key attributes of HM 5.3 is its customer location process. According to JC, HM 5.3 is designed to model a least-cost, forward-looking network with all necessary components to provide local exchange service and UNEs. The model does this by incorporating the most specific and detailed demand data available from Verizon. (JC/Murray Decl., 11/3/03, p. 26.) One of the inputs to HM 5.3 is a customer location database prepared by a third-party vendor, Taylor Nelson Sofres (TNS). TNS created the database by taking Verizon's current customer address information and "geocoding" the precise location of these customers by assigning each a longitude and latitude. Once geocoded, TNS grouped these current customers into logical serving areas, or "clusters." (JC/Mercer, 11/3/03, Attachment RAM 4, pp. 19-24.)

JC claim that HM 5.3 complies with the Commission's CCPs and the FCC's TELRIC costing standards in several ways. We discuss only a few of these below. First, in compliance with CCP 3 and TELRIC, HM 5.3 models the total demand for network elements from both Verizon and other sources, including competitors that lease UNEs. According to JA, HM 5.3 captures all economies of scale and scope in the provisioning of retail services, UNEs, universal service, and interconnection services. (JC/Murray, 11/3/03, pp. 25-26.) HM 5.3 assumes a network that can accommodate both current and reasonably foreseeable demand by assuming sufficient capacity to allow for defective equipment and some "churn" in the locations at which demand will occur. (Id., p. 24.)

Second, JC contend that HM 5.3 is a forward-looking approach in compliance with CCP 6 and TELRIC rules because it reflects reasonable expectations of actual, achievable forward-looking costs savings that can be expected to occur as Verizon moves from today's embedded facilities to a more efficient technology mix and a forward-looking network design. (Id., p. 33.) While JC admit that HM 5.3 does not use actual outside-plant routes as alluded to by CCP 6, they contend that the FCC's TELRIC rules, issued after the Commission's CCPs, do not require the use of actual plant routes and only constrain cost models to the use of existing wire center locations. (JC/Murray, 11/9/04, p. 13.)

Third, JC maintain that HM 5.3 uses inputs and assumptions that reflect long-run costs, as required by the CCPs and TELRIC, because it does not treat any of Verizon's existing facilities as fixed other than the location of Verizon's wire centers. (JC/Murray, 11/3/03, p. 22.)