11. Cost Updating of High Cost Proxy
11.1 Introduction
As noted above, we conclude that the CHCF-B program needs to continue, albeit on a more limited and better-targeted basis applicable to those areas where cost levels may still warrant some level of subsidized support. As a basis for continuing to provide support in high cost areas on a more limited basis going forward, we must update the cost proxies to derive a reasonable level of support due for high cost lines.
In D.96-10-066, we defined high costs based on a "Census Block Group" (CBG) where the cost of basic telephone service exceeded the system average in the territories of the state's four large and mid-sized LECs. Since the adoption of CBG costs in 1996, over a decade ago, the makeup of the state's population demographics, technology and costs of providing service have changed considerably. As a result, the applicable level of costs included in the support calculation require updating accordingly.
Although raising the high cost benchmark to $36 per line will significantly reduce the number of lines to be subsidized, as shown on Appendix Table 1, certain areas will still reflect costs above the $36 threshold, at least based upon existing cost proxy levels. In the OIR, we tentatively concluded that we should still update the estimated cost of providing basic service to Californians to reflect current conditions. Based on our review of parties' comments, we affirm that conclusion.
By updating the cost data utilized for determining the support levels, we will have greater assurance that support is limited to reasonable cost levels needed to meet the goal of universal service. The cost updating process involves two primary determinations: (1) selection of a cost model to use in deriving high costs by region and (2) determination of the appropriate input values to reflect the applicable high costs to be modeled. As noted in the OIR, the original "Cost Proxy Model" (CPM) utilized to derive high cost proxies in D.96-10-066 is no longer available for use. As discussed below, we adopt an alternative model in this order for developing updated proxy costs for purposes of determining the appropriate level of support. In view of the limitations underlying the cost-updating approach that we adopt, however, we shall also pursue a longer term solution in the form of a reverse auction for determining applicable support levels.
11.2 Parties' Positions
As a basis for conducting updated studies of high cost proxies, the first step is to adopt an appropriate model to be used. AT&T identified various models that could possibly be used to perform updated calculations of cost proxies for purposes of setting the level of subsidy support in designated high cost areas. Such models include "CostPro" (originated by the developers of the original CPM) used to establish B-Fund support levels in 1996, the "Synthesis Model" (approved by the FCC, but not previously used within California), the Hatfield Model (HM 5.3) (previously used for modeling UNE costs for AT&T and Verizon, but not previously used for deriving universal service funds).
AT&T argues that while none of the available models is ideal, the HM 5.3 model offers the best choice for use in updating proxy costs for deriving high cost support levels. The advantage of the HM 5.3 model is that the Commission has already approved it for use in the UNE proceeding. The HM 5.3 model has the capability to compute costs for universal service. Although the universal service calculations in the HM 5.3 model are based on federal definitions, adjustments can be made to reflect California-specific costs.
Verizon proposes to utilize adopted UNE loop costs as a basis for updating basic retail service costs for deriving B-Fund subsidies. Verizon argues that UNE costs can be updated without the need to perform comprehensive cost studies. For Verizon, UNE costs were most recently adopted in D.06-03-025.166 The adopted UNE costs would require the addition of the port, usage and retail costs. Although UNE costs are identified only on a wire center (rather than a CBG) basis, Verizon argues that the wire center provides a sufficient level of granularity.
DRA argues that the current method for determining which areas are high cost and the amount of support required per area tends to overstate the amount of support required. DRA claims that the existing methodology for high cost support which is based on forward-looking costs to serve high cost areas produce estimates that are systematically overstated. DRA claims that the ILEC is likely to incur relatively minor costs to extend existing facilities short distances. DRA argues that forward-looking costs are meaningful only with respect to a new wireline competitor entering an area to serve customers. DRA argues that such costs are not relevant in the context of an incumbent carrier extending service using its preexisting facilities. DRA argues that the annual operating expense of the line is more relevant than the total service long-run cost to replace the line, and that the ILECs only need to cover any gap between ongoing incremental costs and revenues. DRA notes that the available models focus primarily on plant layouts and associated capital costs of a new plant, but provide little information as to relationship of ongoing operating costs in relation to population and geographic differences.
AT&T claims that DRA's position constitutes a repudiation of the "Consensus Costing Principles" which were previously adopted for use in developing wholesale rates.167 AT&T argues that these principles have been used in many subsequent proceedings including the proceeding that established the B-Fund, and that DRA arguments for changes to that standard are without any sound basis. Among other disagreements, AT&T disputes DRA's assumption that once facilities have been deployed to serve a new area, the only relevant costs applicable to determining B-Fund support levels are short-run incremental costs. AT&T argues that facilities periodically must be replaced, and depreciation of those facilities is a necessary part of service costs.
SureWest agrees with DRA that the currently applied costing methodology for B-Fund support levels fails to capture SureWest's true cost applicable to serving customers in high cost areas. In contradiction to DRA, however, SureWest claims that incremental cost understates the actual costs involved in serving high cost lines. SureWest contends that "actual" cost of service data is preferred, rather than theoretical assumptions underlying "forward-looking" cost approaches, at least for purposes of a SureWest cost model to revise B-Fund support levels.
SureWest supports the creation of an entirely new model that is tailored to its specific costs of providing basic retail service in its territory, rather than relying on a model focused on the costs of a larger ILEC. In D.05-08-004, SureWest was previously directed to develop and submit a cost model for purposes of producing a cost proxy of its operations as a basis to derive B-Fund subsidy levels. The cost model was intended to be used to justify SureWest's interim annual draw of $11.5 million from the CHCF-B. SureWest subsequently filed a Petition for Modification of D.05-08-004, proposing to phase down the interim draw. SureWest was also granted an extension of time to submit the cost model required under D.05-08-004. SureWest argues that a company-specific model is no longer needed to resolve the interim draw, but that a revised statewide model is needed so that SureWest can receive an accurate level of B-Fund support.
Time Warner suggests that the updating of the cost proxy raises questions concerning what is the appropriate technology to be modeled for purposes of setting a cost proxy. Another question is who should pay for any higher quality or greater reliability features that may be associated with a wireline technology. Time Warner argues that any quality or reliability differences of wireline may be less important to customers in more remote service areas subject to high cost. As an alternative, Time Warner suggests merely capping the cost proxy at the rate level for basic wireless service.168 In this way, any incremental cost differences associated with a wireline network would be born only by those customers willing to pay more for the higher level of reliability. Time Warner argues that it would be unfair to require all consumers to subsidize the incremental improvement in reliability of wireline service if the recipients of that reliability are unwilling to pay for it themselves.
If the Commission chooses to utilize the ILEC network as a model for updating high cost proxies, however, Time-Warner proposes an approach based on comparing UNE Platform cost changes over distinct time periods. Specifically Time Warner proposes (1) calculating the percentage change in UNE costs between the mid-1990s and those adopted in D.04-09-063 (for SBC) and
D.06-03-025 (for Verizon) and (2) applying the percentage change by zone to the previously adopted costs for each CBG that exists within each respective zone.
TURN proposes a variation of the Time-Warner approach intended to produce a more accurate cost adjustment. As explained in the Declaration of Dr. Roycroft, attached to TURN's comments, TURN suggests that a similar scaling using a publicly available model such as the FCC Synthesis Model or Hatfield Model 5.3, with updated information provided by carriers to reflect forward-looking input prices. The updated model runs would then be compared with runs using input values from earlier periods, and a scaling ratio would be generated. Since these models generate cost estimates at the wire center level, a matching of wire centers and census block groups would be required to correlate cost model results with the CPM estimates.
11.3 Discussion
In weighing the various factors involved, we conclude that the cost proxy for B-Fund support must be updated given the passage of more than a decade since the costs were previously reviewed. The updating of costs raises a number of difficult questions. For example, given the competitiveness of the industry through intermodal technologies (i.e., wireline, wireless, VoIP, and cable networks), it is unclear as to what technology would be most relevant for purposes of modeling a competitively-neutral cost proxy. For example, wireless or broadband technology may be able to provide comparable local service to an area at a lower cost than the traditional copper-loop circuit-switched architecture as generally utilized by the ILEC. Yet, our current standards for basic service requirements and the associated existing cost models for identifying high cost areas are grounded in the traditional wireline network architecture. Thus, given current standards, we cannot rely upon basic wireless rate levels (as suggested by Time-Warner) as the basis for capping high cost proxy levels. None of the identified models, moreover, is capable of calculating the cost of a wireless or cable-based network, or voice service based on Internet Protocol. It would not be feasible to undertake a cost study based on data from intermodal providers, or to modify cost data to factor in service quality differences among different intermodal technologies.
While the ideal solution would be to identify costs based on the most competitive technology currently available, the resources are not currently available to identify and measure such costs. Since the ILECs currently serve the function of COLR, we conclude that the costs of the ILEC network continue to be acceptable, at least in the near term, as a basis for updating the high cost proxy. As discussed below, however, we intend to institute a reverse auction process whereby all competitors will be able to bid on obtaining rights to receive B-Fund subsidies. In this manner, the ultimate goal will be to let the marketplace determine the appropriate B-Fund support level based upon the least-cost technology delivered by the COLR that can most efficiently offer universal service access in a given area.
Given the limitations involved, we conclude that the HM 5.3 model offers the best choice among the available options for purposes of updating the B-Fund high cost proxies. In previous Commission proceedings, we have acknowledged that the HM 5.3 model is not ideal as a cost estimating tool, but contains certain flaws. Nonetheless, we previously concluded that HM 5.3 model results were sufficiently acceptable to use in the most recent UNE cost proceedings for AT&T and Verizon (see D.04-09-063 and D.06-03-025). As stated in D.06-03-025, we concluded that adoption of the HM 5.3 model for determining Verizon's UNE rates was "reasonable given the enormous complexity involved in TELRIC modeling exercises. It is reasonable to use a model with some flaws when the alternative is another model with more significant flaws that is also difficult to operate and modify." (P. 56.)
We therefore adopt the HM 5.3 model for use in this proceeding for purposes of the development of updated cost proxies to derive B-Fund support levels prospectively. The HM 5.3 model offers a relatively current customer location database and model inputs that have already been reviewed by the Commission for AT&T and Verizon. In the next phase of the proceeding, we shall conduct additional inquiry concerning whether or how the cost data for AT&T and Verizon may serve, or be adapted, as proxies for B-Fund support in the SureWest and/or Frontier service areas.
Although some code changes are required, HM 5.3 does include CBG-level cost data that may be extracted. We conclude that costs should be disaggregated at the CBG level in order to provide a sufficient level of granularity for identifying high cost lines. Using the adopted version of the HM 5.3 model is superior to updating the 1996 cost study for year 2000 U.S. Census data because the HM 5.3 model contains actual customer location inputs that are more current. Since the rights to the model are controlled jointly by the model developers and the former stand-alone AT&T and MCI, however, provisions will have to be made for appropriate access for third-party review.
The other potential modeling sources suggested by parties do not provide as useful a basis for deriving a cost proxy as does the HM 5.3 model. The Cost Pro model is not suitable in view of the fact that no party, including AT&T appears to have any working knowledge of the model's capabilities in estimating retail costs. Likewise, the Synthesis Model is not a suitable choice. As noted by DRA, this model is aging, not regularly maintained, and is cumbersome to use. The Synthesis Model also requires a proprietary third-party input database which is not publicly available. The existing database is over a decade old and numerous new company-specific inputs would need to be obtained and processed through the third-party database developer. As another disadvantage, the Synthesis Model only calculates costs at the wire center, rather than at the CBG level.
We decline to adopt the modeling approach suggested by Verizon using UNE-P - based loop costs as the basis for updating high cost support levels. The UNE model only calculates costs at the wire center level, and thus would not provide sufficiently granular delineation of costs at the CBG level. As a result of such an approach, costs would be averaged across broad geographic areas rather than being targeted to the specific CBG areas where high costs exist. Moreover, the model only provides costs for AT&T and Verizon (which would still require some updating), but no recent cost data for either SureWest or Frontier.
We also decline to adopt the Time-Warner/TURN proposals for scaling the costs from the original Cost Proxy Model using the zone-wide shifts in the Commission-adopted UNE results. These proposed approaches would be unnecessarily complex without yielding meaningful projections to reflect real population and demographic changes, or technology and market changes that have affected costs since 1996.
DRA argues that it would involve a major undertaking to litigate and adjudicate the appropriate level of updated costs for each of the COLRs. TURN, however, believes that the inputs to the model can be updated relatively easily. AT&T proposes that the principal changes for cost proxy updating be limited to:
(a) Updating the locations of customers and CBGs;
(b) Adjusting the model to calculate costs per CBG rather than per wire center;
(c) Reflecting geographical and political obstacles to the construction of networks;
(d) Updating inputs for costs of labor and equipment; and
(e) Incorporating retail cost inputs not previously derived.
Our intent is to avoid relitigation over the previously adopted methodology or sources used to calculate input prices, such as depreciation rates or costs of capital. Rather, our goal is to limit the updating process merely to reflect the cost inputs to that were used in the most recent UNE cost proceedings for AT&T and Verizon. We shall set as an initial priority that cost updates and model runs be performed for AT&T and Verizon. We recognize that no previous costs have been adopted for SureWest or Frontier utilizing the HM 5.3 model. We shall separately consider the process for updating of cost proxies for SureWest and Frontier to follow after costs are updated for AT&T and Verizon. We shall schedule a workshop in the next phase of this proceeding for the parties to meet and confer with the goal of reaching consensus, or at least minimizing, disputes concerning the manner in which cost inputs should be updated for the limited purpose of computing the high cost proxy to derive subsidy levels in accordance with the approach adopted herein.
166 See D.06-03-025, Appendix C, "Wire Centers by Zone."
167 See D.95-12-016, Interim Opinion Adopting Cost Methodology Principles and List of Basic Network Functions for which Cost Studies are to be Performed, 62 CPUC 2d 575, Appendix C (Dec. 6, 1995).
168 Time-Warner Comments of 4/27/07 at 20.
