V. Interim Rates for Unbundled Loops

A. Joint Applicants Proposal

In their motion for interim relief, Joint Applicants request a statewide average loop rate of $7.51 for the basic (2-wire) loop.²⁵ This rate represents a 36% discount from the current statewide-average loop rate of $11.70.²⁶ In support of this request, Joint Applicants note a decline in forward-looking loop costs since 1994. Specifically, they cite reduced prices for DLC electronics that have dropped roughly 38% between 1994 and 2001. (Joint Applicants' Motion for Interim Relief, 8/20/01, p. 10.) They also report that Pacific's reported total of switched access lines grew from 15 million lines in 1994 to almost 19 million lines in 2000. (Id.)

Along with the motion, Joint Applicants provide testimony by witnesses Bryant, Mercer and Klick regarding estimates of Pacific's forward-looking costs for unbundled loops using the most recent versions of the HAI model and the FCC's Synthesis model. Dr. Bryant performed an analysis of the sensitivity of cost results calculated by the HAI model by changing two key input values, the cost of DLC equipment and demand levels. (Bryant Testimony, 8/20/01, p. 5-6.) According to his testimony, Bryant used the HAI model to simulate a 1994 view of forward-looking costs for California as constrained by the key input values that were adopted by the Commission in prior OANAD decisions. He then used this starting point to change DLC equipment and demand levels for 2000 and compared HAI's outputs for 1994 and 2000. Bryant states that the combined effect of these two input changes is a 36% decrease in average loop cost from 1994 to 2000. (Id., p. 6.) Joint Applicants claim that the Synthesis Model confirms this loop cost analysis. Based on this percentage change in the model output after changing only two inputs, Joint Applicants request a 36% reduction from the UNE loop rates adopted in D.99-11-050.²⁷

B. Response

On October 19, Pacific filed substantive comments in response to the proposed 36% loop rate reduction. Pacific maintains that the three cost drivers relied on by Joint Applicants -- line growth, DLC electronics costs, and
expenses -- do not support a 36% reduction in current UNE loop prices. We will discuss Pacific's criticisms of Joint Applicant's proposal, and responses to Pacific's criticisms, by issue below.²⁸

C. Discussion

1. Line Growth

Pacific states that Joint Applicants' line growth assumptions in the HAI model are flawed because they incorrectly treat special access facilities, particularly DS-1 and DS-3 lines, as ordinary copper loops. For example, Joint Applicants have attributed 24 lines to each DS-1 line and 672 lines to each DS-3 line because these lines carry 24 and 672 "voice grade equivalent" (VGE) channels. In contrast, Pacific notes that a DS-1 line consists of merely two copper loops, while a DS-3 line is provisioned over fiber so it does not involve any copper loops. According to Pacific, the net effect of Joint Applicants' use of VGEs is to overstate the number of loops in Pacific's network by about 10 million. Further, these inflated line assumptions produce illusory "scale economies," such as larger cable sizes and excess structure sharing, which understate Pacific's loop costs. (Pacific Loop Comments, 10/19/01, p. 4.)

Joint Applicants defend their modeling of line growth by claiming that the VGE method is well accepted and conservative. Joint Applicants cite examples of the FCC endorsing the use of line counts based on VGEs in its Synthesis Model, although they note that the FCC ultimately adopted a methodology that develops the network on a physical pair basis and divides the resulting total investment by the number of voice grade equivalents. (Klick Declaration, 10/30/01, p. 5.) Joint Applicants claim that using VGEs to model line growth is actually conservative because treating each channel on a DS-1 or DS-3 line as a copper line adds more cost per line than Pacific would actually incur to provision services using fiber. (Donovan Declaration, 10/30/01, p. 3.) They also note that Pacific has admitted significant volume growth for high capacity services provided over DS-1 and DS-3 facilities. Joint Applicants state that any analysis of line growth must be based on VGEs because DS-1 and DS-3 lines share outside plant structure with basic loop facilities. They allege that if DS-1 and DS-3 growth is not incorporated into the analysis, loop costs for basic unbundled loops will be overstated and this will shift shared costs to basic loops and force basic service to cross-subsidize business service.

In support of Joint Applicants' use of VGEs, ORA notes that Pacific itself measures wire-line growth in terms of VGEs. ORA maintains that line growth should be based on VGEs because it is not appropriate for the Commission to only consider the costs of copper loop plant when that plant is being replaced with less expensive and more cost effective fiber transport and distribution. (ORA Loop Comments, 10/30/01, p. 3.)

TURN contends that Pacific's analysis is contradicted by its own public data. The FCC's ARMIS database indicates that the total number of access lines in 2000 was 29.6 million. Further, TURN claims that all services that share facilities such as cables, conduit, trenches and remote terminal facilities should benefit from the economies of scale that have resulted from Pacific's substantial line growth. According to TURN, the net effect of Pacific's approach of excluding VGEs from any estimate of line growth is to "unreasonably shift costs away from the telecommunications lines utilized by large business customers and onto the loops utilized in the provision of residential and small business basic exchange services." (TURN Loop Comments, 10/30/01, p. 2.) In other words, if line growth is understated, this has the effect of causing higher per line costs for basic exchange loops.

For this interim rate setting exercise, we prefer to adopt a more conservative approach rather than a modeling technique that admittedly overstates the number of copper lines in Pacific's network. Although the FCC used VGEs for its Synthesis Model, parties admit that this model was not developed for UNE cost purposes but for universal service purposes. The goals of a model for UNE costing and universal service are quite different, and the FCC has suggested that state or company specific values should be used for UNE costing and pricing purposes.²⁹ As we develop interim estimates of costs for Pacific's loops, we are concerned that overstating the number of copper lines could create assumptions of scale economies in Pacific's network that are not realistic. Although we agree with TURN and ORA that we should not ignore the undisputed growth of special access services, we prefer to account for it on a physical pair basis at this interim phase. We will not adopt a modeling convention that assumes this growth is provisioned entirely over copper when it quite clearly is not. We are troubled by the notion that it is acceptable to overestimate the number of copper lines in the model simply because they are more expensive. Although Joint Applicants, TURN and ORA are concerned that residential users may cross-subsidize business customers, the VGE method would have the opposite effect of allocating the higher costs of a copper-based network to users of fiber-based special access services, potentially violating the TELRIC methodology. We want to avoid creating cross-subsidies in either direction and prefer to take a more careful look at this issue in the next phase of this proceeding.

Joint Applicants acknowledge that the FCC ultimately adopted a methodology that develops the network on a physical pair basis. We are persuaded to adopt that approach for this interim exercise as well rather than inflating copper line counts to reflect special access lines using the VGE method. We will assign the cost of shared facilities such as conduit, poles, and trenches commensurate with a service's physical use of that facility. Therefore, we will alter the line counts in the HAI model to reflect physical facilities. In other words, we will account for DS-1 lines as two access lines since they are comprised of two pairs of copper, and we will account for DS-3 lines as one access line since they are provisioned over a single strand of fiber. This results in a line count of 16.3 million in 1994, growing to 20.0 million lines in 2000. When we insert these adjusted line counts into the HAI model and perform a comparison of 1994 and 2000 model runs, the net result is a decrease in the loop discount proposed by Joint Applicants from 36% to 25%, and an increase in the proposed interim loop rate from $7.51 to $8.73. (See Appendix B.)

In comments on the draft order, Joint Applicants claim that the draft decision's use of a physical line count methodology is flawed and produces illogical results. They claim that the physical line approach implies that DS-1 loops should be priced at twice the basic loop rate and that DS-3 loops should be priced at the same as the basic loop. They also claim that the methodology ignores the demands that DS-1 and DS-3 loops place on loop electronics. TURN and ORA echo these comments, while Pacific supports the draft decision on this issue as written.

The comments do not persuade us that the physical line count approach is flawed. This approach is used here solely for the purpose of allocating certain shared facility costs between basic loops and DS-1 and DS-3 facilities. We are not using this methodology to develop UNE costs for DS-1 and DS-3 loops so any extrapolation of the results for that purpose, as Joint Applicants, TURN and ORA suggest, is improper. We do not agree that our adopted approach ignores the demands of DS-1 and DS-3 facilities on loop electronics; thus we decline to make any changes to the order in this regard.

2. Infill Growth vs. Plant Extension

Pacific claims that the Joint Applicants' line growth analysis is flawed because it assumes that 100% of growth in Pacific's network since 1994 has been "infill" growth, i.e., growth in already developed areas. Pacific maintains that 70% of the growth in its network over the last several years has been growth to previously unserved areas, or "plant extension" growth, and only 30% has been infill. (Pacific Loop Comments, 10/19/01, p. 5-6.) Pacific alleges that the manner in which HAI models customer growth guarantees lower loop costs because it packs more lines and customer locations into hypothetical local distribution areas, or "clusters." (Tardiff Declaration, 10/19/01, p. 6-7.) Pacific asserts that in reality, plant extension growth tends to be more expensive because it involves the placement of new feeder and distribution facilities and longer cables. Accordingly, Pacific asserts that the costs of plant extension growth more than offset any potential per loop savings from infill growth.

Joint Applicants respond that Pacific's criticisms of the way HAI models growth are not consistent with accepted forward-looking costing principles. According to Joint Applicants, a cost model should not look at the cost of "infill" vs. "plant extension" growth because that approach only looks at the cost to augment existing plant to serve a new increment of demand since the prior OANAD costing exercise. Instead, a proper forward-looking methodology considers the cost to serve total demand in the most efficient manner possible, constrained only by Pacific's current wire center locations. Joint Applicants claim that HAI uses this latter approach and therefore, Pacific's criticisms are meaningless. (Joint Applicants' Loop Reply Comments, 10/30/01, p. 11-12; Murray Declaration, 10/30/01, p. 3-4.)

In addition, Joint Applicants defend the placement of customers under the HAI model by explaining that HAI uses precise geocoded customer location data to place approximately 65% of the customer base. For the approximately 35% of customer locations that are unknown, the model distributes customers uniformly along all roads within the census block. Joint Applicants maintain that this approach conservatively over-disperses customers and potentially increases loop costs by overestimating loop plant. (Joint Applicants' Loop Reply Comments, 10/30/01, p. 18-19; Mercer Declaration, 10/30/01, p. 7-8.) Joint Applicants directly dispute Pacific's assertions that 70% of growth involves costly plant extensions by citing statistics that suggest the majority of growth is infill instead.³⁰ Joint Applicants also contradict Pacific's assertions that loop costs have increased by providing ARMIS data showing decreases in total loop investment per line from 1994 to 2000. (Joint Applicants' Reply Comments, 10/30/01, p. 23.)

ORA challenges Pacific's contention that plant extensions are more costly by noting that Project Pronto and fiber fed "next generation" DLC technology extend central office functions throughout Pacific's outside plant network without long runs of costly copper.

We have already found that because Pacific has not provided us with a model that we can use to test undisputed line growth, we must use the HAI model for this interim pricing effort. While Pacific alleges certain shortcomings in HAI such as potential problems with how it locates customers, this problem is not insurmountable because it pertains only to the one-third of customers that cannot be located using geocoded information. Indeed, HAI places two-thirds of its lines based on actual customer location information. We believe that any customer location problem is somewhat mitigated by our adjustments to HAI to back out the use of VGEs for line counts. When we base line counts on physical facilities rather than VGEs, as discussed above, we reduce the extent to which HAI "crams more customers" into existing areas for the one-third of customer locations that HAI must model without customer location information.

Further, we are not persuaded that potentially costly plant extension growth outweighs other cost reductions. We agree with Joint Applicants that it is improper for a cost model to consider only the cost of infill or plant extension growth. Instead, the cost model should consider the cost to serve total demand as set forth in our adopted Consensus Costing Principles.³¹ Even if we were to consider Pacific's approach, Pacific's claims are disputed by Joint Applicants with demographic, line growth, and ARMIS investment data. Given this material that contradicts Pacific's claims regarding growth, it would be improper to accept Pacific's unsupported assertions that the cost of plant extension growth exactly counteracts loop cost reductions. Therefore, we will rely on the HAI model for the interim, irrespective of Pacific's comments in this area. We reiterate that our use of HAI for interim rates in no way prejudges whether to use HAI for setting permanently revised UNE loop rates.

Pacific comments that the draft decision commits legal and factual error by assuming that all line growth has been in developed areas (i.e. "infill growth"). Pacific claims that the record is undisputed that 70% of growth has been higher cost plant extension growth, and that the HAI model used in the draft decision does not allow Pacific to test its assertions regarding line growth. Joint Applicants respond that Pacific is merely rehashing the same arguments rejected by the draft decision and that Pacific's claim of an undisputed record on higher cost plant extension growth is inaccurate.

We agree with Joint Applicants that Pacific is, for the most part, rearguing its earlier positions. We disagree with Pacific's contention that we have ignored undisputed evidence regarding line growth. The record was indeed disputed on whether 70% of growth is plant extension and whether the cost of plant extension counteracts other loop cost reductions. Pacific's assertions that plant extension growth completely offsets other loop cost reductions are not supported by the record. The Commission will resolve this dispute in the final phase of this case rather than delay interim relief.

Furthermore, the draft decision explains that reductions in line counts mitigate the extent to which HAI models infill growth. Thus, we did not ignore Pacific's criticisms in this regard. Pacific's concerns were addressed appropriately, given the fact that the dispute is over only one-third of the lines modeled and the total number of lines was reduced from Joint Applicants' initial model runs.

3. DLC Equipment Costs

Pacific asserts several flaws in Joint Applicants' analysis of DLC equipment cost reductions. Pacific does not dispute that DLC equipment prices have fallen in recent years.³² Nevertheless, Pacific claims that Joint Applicants incorporated flawed assumptions into their DLC analysis. These assumptions include 1) an analysis that all remote terminals (RTs) are above-ground while ignoring allegedly higher cost underground controlled environmental vaults (CEVs), 2) allocations of DLC site preparation and installation costs that are too low, and 3) unsupported reductions in non-equipment DLC costs. Pacific contends that all of these items overstate the cost savings attributable to falling DLC equipment prices and are not justified.

Joint Applicants respond that they modeled RTs above-ground because that was the assumption Pacific itself used in the adopted OANAD studies. In addition, they claim that CEVs are less costly than RTs on a cost per line basis. (Joint Applicants Loop Reply Comments, 10/30/01, p. 13.) In other words, if HAI had modeled underground CEVs rather than above-ground RTs, the proposed interim prices might be even lower.

Joint Applicants address Pacific's other allegations by stating that site preparation, installation, and non-equipment DLC costs were held constant in the 1994 and 2000 runs of HAI. Therefore, Joint Applicants maintain that those costs play no part in the trend analysis supporting the 36% proposed loop reduction. (Id., p. 13; Donovan Declaration, 10/30/01, p. 8-9.)

We find that Pacific's criticisms of Joint Applicants' assumptions have no merit. First, Pacific does not dispute that RTs were modeled as above ground in OANAD.³³ Indeed, above-ground RTs were the reality in the network at that time. Today, however, comments by both Joint Applicants and Pacific indicate that CEVs are replacing RTs in many locations. Unfortunately, the record is disputed on whether CEVs are more or less expensive than RTs on a per line basis. We find flaws with the analysis offered by both Pacific and Joint Applicants. Both parties appear to mix costs and line capacities from various size CEVs in their calculations and we are not confident in the calculations of either party.³⁴ The record is incomplete in this area and it is unclear what CEV cost we would use if we wanted to change assumptions used in the prior OANAD decision. We cannot accept Joint Applicants' assertion that CEVs are less costly on a per line basis than RTs. Similarly, we cannot accept Pacific's assertions that CEVs are more costly than RTs. In keeping with our overall desire that interim rates reflect primarily undisputed cost changes, we will not change the original OANAD inputs with regard to RT and CEV costs. This is an area we can explore in setting final rates. For now, we will not change the original OANAD assumptions regarding above-ground RTs. We note that because we make no changes to prior RT assumptions, the loop cost trend analysis using HAI is not impacted in this area.

Second, we find that Joint Applicants only made adjustments for DLC equipment cost reductions and did not vary any other DLC-related costs such as site preparation, installation, and non-equipment costs in their trend analysis using the HAI model. (Donovan Declaration, 10/30/01, pp. 8-9.) Again, we find that because non-equipment DLC costs were held constant, they do not impact the trend analysis. We have no basis on which to increase these costs, as Pacific has suggested. Hence, we will reject Pacific's protests on this issue.

In comments on the draft decision, Joint Applicants contend that the Commission errs in not accepting the evidence they presented allegedly showing that CEVs are less costly on a per line basis than RTs.

In contrast, Pacific claims the draft decision errs in ignoring evidence that roughly half of the RTs in Pacific's network are housed below ground in CEVs. Pacific alleges that it has documented the real costs of CEVs while Joint Applicants' presentation of CEV costs is flawed and should be ignored.

We reiterate that we cannot agree with either side of this dispute without developing further record evidence on this issue. That effort is more appropriate for the next phase of this case. The comments clearly validate the conclusion in the draft decision that the record is disputed on RT and CEV costs. The results of the draft decision are left unchanged on this issue.

4. Loop Expenses

Pacific claims that Joint Applicants have made unsupported reductions in the expenses for maintaining and repairing loops. Pacific asserts that the reason for this decline in expenses is the application of an "investment/expense factor" embedded in the HAI model. Essentially, Pacific claims that for each dollar decrease in capital expenditures in the HAI loop model, HAI automatically decreases loop expenses by a certain amount. Pacific cites language where the Commission stated that this "investment factor approach is inconsistent with TSLRIC Principles No. 4...." (D.95-12-016, mimeo., p. 10) and that simple common sense dictates that even if DLC equipment costs decline, repair expenses are not automatically reduced. Further, Pacific claims that expenses included in current loop costs are not based on 1994 data but on repair expenses that were trended downward for 1996 and 1997 to reflect forward-looking technology.

Joint Applicants defend their expense analysis by stating that Pacific's expenses have fallen considerably on a per loop basis since 1994. (Joint Applicants' Loop Reply Comments, 10/30/01, p. 14, footnote 36.) Joint Applicants' contend that HAI results track with actual trends and are a realistic reflection of forward-looking loop expense reductions (Klick Declaration, 8/20/01, p. 4, 8-11; Murray Declaration, 8/20/01, p. 5, 26-29, 35-37, 40-41.) Joint Applicants claim that Pacific has not addressed this substantial evidence of expense reductions and does not adequately support its claim that expenses have not dropped in the face of the actual reported data.

Joint Applicants defend their use of expense to investment ratios because they replicate forward-looking expense adjustments without requiring a data-intensive review of each expense account. Joint Applicants also note that the ratios used in HAI are those developed by the FCC for use in its Synthesis Model. Joint Applicants further maintain that Pacific uses this same FCC Synthesis Model to support its proposed switching discount in the Section 271 proceeding.³⁵

ORA responds that it is reasonable to assume maintenance expenses have fallen for loops given Pacific's statements that implementation of Project Pronto would pay for itself in maintenance savings. ORA states that the forward-looking trends anticipated in the earlier OANAD calculations likely have not fully captured the expense savings associated with Project Pronto.

We agree with Pacific that the use of investment to expense factors in HAI may not be reasonable. The fact that investments in certain technologies may have decreased in price does not mean that all other expenses, such as maintenance, have also dropped. Nevertheless, we will not go so far as to state that an investment factor approach violates the forward-looking cost principles. Indeed, the same decision cited by Pacific as denying an investment factor approach suggests that "partial use of investment factors may help to reduce the possibility of `gaming' in the assignment of maintenance expense." (D.95-12-016, mimeo, p. 12.) Because we are setting interim rates that will be subject to true-up, we will use a conservative approach and remove the effects of the investment/expense factor approach from the trend analysis to avoid the risk of overstating any loop cost decreases. We think that Pacific has raised valid criticisms of the factor approach so we will not use it to adjust rates for the interim. After we rerun the HAI model keeping expenses constant in the 1994 to 2000 runs, we see that this removal of the factor approach, coupled with our removal of the VGE line count method, has the effect of reducing the relative change in loops costs from 1994 to 2000 from 36% to 15 %. As a result, the interim loop rate proposed by Joint Applicants increases from $7.51 to $9.93.

Nevertheless, we find that Joint Applicants have provided preliminary evidence of expense cost declines based on actual data that we will need to explore further when we set final rates for loops. Thus, loop expenses will undergo further scrutiny in the next phase of this proceeding.

In comments on the draft decision, Joint Applicants, TURN, and ORA comment that the draft uses an improper methodology to remove the impact of expense-to-investment ratios and this error leads to an incorrect calculation of loop expenses. The draft decision holds expenses per loop at a constant level. At the same time, the analysis assumes an increase in the number of loops served, which causes total loop expenses to increase. Joint Applicants claim that it is undisputed that Pacific's total expenses have decreased, or at least remained flat, from 1994 to 2000. Using this reasoning, Joint Applicants argue that the Commission should hold total expenses constant rather than expenses per loop, which effectively results in a decrease in expenses per loop.

Pacific disputes Joint Applicants' comments on this point and instead argues that loop expenses were calculated correctly in the draft decision. Pacific states that Joint Applicants use flawed logic to suggest that expenses per loop should decline simply because the number of loops served has grown. For illustration, Pacific suggests that if it serves 100 loops at $10 per loop in 1994, the model should assume costs of $10 per loop in 2000 no matter how many loops are served. If the number of loops served has grown, then total loop expenses would also increase although per loop expenses would be held constant at $10. Pacific claims that the draft decision uses the correct methodology in holding expenses per loop constant. Pacific also contends that ARMIS data cited by Joint Applicants actually shows an increase in total loop expenses.³⁶

In the trend analysis in the draft decision, we intended to reverse the effect of the investment-to-expense factors embedded in the HAI model because we did not agree with the assertion that expenses automatically decline when investment levels decline. To remove the investment-to-expense factors, we held expenses per loop at a constant level for the 1994 and 2000 model runs. We are not persuaded to alter this methodology because the record thus far does not convincingly support a lowering of expenses per loop for two reasons. First, Pacific is correct that certain ARMIS data indicates an increase in total loop expenses from 1994 to 2000. Second, the record on expenses per loop is far from clear given disputes over line counts and growth assumptions in the HAI model. Therefore, we find that leaving expenses per loop constant for this interim rate analysis is the proper middle ground. Parties may make their case for a change to expenses per loop in the next phase of this case.

5. Summary of Loop Changes

We have modified Joint Applicants' HAI trend analysis to remove line counts using the VGE methodology and to remove the effects of the investment/expense factors embedded in HAI. These changes are shown in Appendix B and reduce Joint Applicants' requested loop discount from 36% to 15.1%, for an interim unbundled loop rate of $9.93.

²⁷ Joint Applicants contend that further circumstances most likely lead to an even lower rate, and therefore the 36% reduction that they request is likely conservative. Joint Applicants maintain that existing UNE loop costs are based on assumptions regarding "fill factors" and the amount of structure that is shared (e.g. poles, trenches) that the FCC has found to be inappropriate for a forward looking analysis. If fill factors and structure sharing assumptions were increased to levels that the FCC has found to be forward looking, Joint Applicants claim that the discount from current rates would be higher than the proposed 36%. (Joint Applicants' Loop Reply Comments, 10/30/01, p. 21.) "Fill Factor" is a manner of expressing the percentage of Pacific's loop plant capacity that is in use as opposed to spare capacity. If a network has 30% spare capacity, then the network's fill factor is 70%.