CORRECTING THE
SHORTCOMINGS OF THE
BENCHMARK COST MODEL

10.1 Recommendations regarding the use of the BCM in USF policy deliberations

In the preceding chapters of this report, we:

* Described the criteria by which one should evaluate any cost proxy model;

* Summarized the BCM's essential algorithms, assumptions, and cost data;

* Identified specific flaws and shortcomings in the BCM, and suggested ways to remedy those deficiencies;

* Suggested a framework for the consideration of the revenues that are relevant to any universal service funding determination;

* Computed partially corrected data for (1) the cost of basic residential local exchange service and (2) the level of universal service support required; and

* Discussed some of the salient aspects of several state universal service proceedings that relate to the use of the BCM.

We have identified many key aspects of the model that should be modified and, where feasible, we have suggested either specific corrections or a detailed framework by which corrections can be made. However, although the BCM needs to be changed in some fundamental ways, these corrections are all "doable" and, indeed, if these modifications are incorporated, the BCM can serve as a valuable tool for policy makers as they meet the challenge of creating a workable, effective, and competitively neutral universal service support mechanism. The BCM should thus serve as the foundation for a comprehensive cost proxy model and, if the corrections that are identified in this report are made, it should be adopted as a policy making tool.

Our analysis also demonstrates that claims of universal service requirements are typically inflated. Even the results of the uncorrected BCM demonstrate that universal service support typically is needed only in sparsely populated parts of the country, and that for most parts of the nation no funding or support is needed at all. As Appendix 7A demonstrates, present high cost funds of approximately $750-million, in combination with the existing income-based Lifeline, Link Up, and TRS programs, provide targeted support. There is absolutely no reason for expanding universal service support more broadly.

10.2 Summary of the effects of the ETI corrections to the BCM

In some instances we have corrected individual flaws in the BCM in isolation by revising the input data and rerunning the model. In other cases, we have run the model with several simultaneous changes. In the following discussion, we summarize the effect of some of these corrections. Summary tabulations produced by the BCM for some of the many ETI runs are included in Appendix 8B.

As is described in Chapter 3, the statewide average cost for residential local exchange service in Washington State, as computed by the uncorrected BCM, is $16.94. In Chapter 5, we demonstrated that the switch cost data are flawed. The result of correcting the cost data for switches is to lower the average monthly cost by approximately $2.50.

In Chapter 6, we identified the flaw in the BCM concerning the extremely low fill factors that are assumed in the outside plant. The low fill factors (ranging between 25% and 75% in the distribution plant, and between 65% and 80% in the feeder plant) are indicative of network engineering associated with volatile, unpredictable demand for services other than the primary residential access line. A cost proxy model that is being used for assessing universal service requirements should instead reflect substantially higher fill factors to reflect the fact that demand for single-line residential local exchange service is stable and highly predictable. Therefore, we corrected the flawed BCM fill factors and related structure cost multipliers: The result of making this correction in isolation is to lower the average monthly cost by approximately $2.50. Furthermore, the costs and discounts for digital loop carrier (DLC) subscriber equipment that the BCM assumes are excessive, in part because they fail to reflect large discounts that are routinely offered by vendors. However, because of the difficulty in obtaining more accurate price data, we conducted a sensitivity analysis based upon cost information developed by Hatfield Associates for use in the California USF proceeding. The result of changing these data (and making no other corrections to the BCM) is to lower the average monthly cost by approximately $5.00.

Not all households subscribe to local exchange service. However, a network design intended to accommodate the universal service goal must assume 100% penetration. Accordingly, we incorporated a revision to adjust for the actual subscribership rate of 96.0% in Washington State. This correction raises the average monthly cost by approximately $0.40.

The BCM incorporates an uneconomic decision as to when to deploy fiber in the feeder plant. We analyzed the implications of changing the crossover point from the default value of 12,000 feet to various alternative distances ranging between 9,000 feet and 27,000 feet. Increasing the distance at which the crossover decision is made decreases the average cost, confirming our belief that the BCM's fiber/copper decision rule is not economically based.[1] For example, increasing the distance to 27,000 feet lowered the average cost by approximately $2.00[2]

These corrections are interrelated, so their combined effect must be determined through the model itself rather than by a mere aggregation. As discussed in Chapter 8, ETI ran the BCM with corrected switch costs, revised fill factors for outside plant, and an adjustment for the subscribership rate. The result of these changes - expressed on a national basis -- is to lower the average monthly cost by approximately $4.34, i.e., the ETI corrected result is a monthly average cost of $12.37. There are also significant corrections whose effect we were not able to quantify that should nevertheless be addressed. These changes will significantly affect the average cost as determined by the BCM (and in almost all instances reduce it), and correspondingly decrease the overall universal service funding requirement:

* As is shown by our sensitivity analysis, the assumptions about the costs and discounts for digital loop equipment substantially influence the cost of basic local exchange service. Our partially corrected final numbers reflect the BCM's unsupported and likely excessive costs for digital loop equipment. The use of more realistic cost data would significantly lower the final results.

* Correcting the over-simplified assumption of uniform household density with the road buffer approach that the Joint Sponsors have identified will further lower the cost.

* The fiber/copper crossover point, if corrected to reflect engineering economic cost evaluations (rather than being based upon considerations other than the provision of primary residential access lines), would result in lower costs.

* The deployment of wireless service in those high-cost areas where the costs of wireless are less than those for wireline service would lower the average cost.

* The stand-alone cost of single-line basic residential service and the stand-alone cost of all other local exchange services should be computed, so that the benefit of the economies can be shared with universal service. This would lower the average cost.

* The BCM should be revised to include the costs of service area interfaces (SAIs), which would increase the average cost by less than $0.50.[3]

Based upon our comprehensive analyses of the BCM, we conclude that the average national monthly cost for basic residential local exchange service is less than $12.50 per month. Also, once corrected costs are computed for each CBG, reflecting the corrections identified above, the BCM should then assess the need (if any) for universal service support by evaluating such need on a wire center basis. Because the BCM does not, in its present form, readily permit such an evaluation, we have not yet quantified the effect of this correction to the model. However, such a correction will clearly lower the universal service funding requirement.

Correcting the switch costs and the fill factors lowers the national universal service requirement computed by the BCM from approximately the range $1.4- to $4.0-billion range[4] to a range of $400-million to $1.5-billion.

10.3 The use of the BCM in policy making proceedings

As presented by the Joint Sponsors, the BCM overstates -- and by a significant amount -- the costs incident to the universal and ubiquitous provision by LECs of primary residential access lines. As such, the BCM cannot be used in its present form unless the various logical and factual assumptions and data upon which it is based are addressed and corrected. The design of the BCM is, however, sufficiently flexible so as to accommodate all of the specific corrections that we have identified. With these modifications, the revised BCM can satisfy the need for a comprehensive cost proxy model that will be capable of informing the complex universal service policy issues currently before state and federal regulators.

[2]As we explain in Chapter 6, reducing the digital loop equipment costs to reflect vendor discounts will likely affect both the crossover distance itself as well as the ultimate magnitude of this correction.

[3]California PUC, Universal Service Proceeding, Pacific Bell and INDETEC International, CPM -- California Universal Service Subsidy, at 5.

[4]These data assume Cost Factor No. 2.