XII. Comments on Alternate Proposed Decision
Following the issuance of the Alternate Proposed Decision, comments were received on September 16, 2004 from MCI, CWA, SBC-CA, AT&T, ORA/TURN, Pac-West Telecomm, Inc. (Pac-West), XO, CALTEL-Navigator, Telescape Communications, Inc. (Telescape), Covad Communications Company (Covad), MPower, and DOD/FEA. Reply comments were received on September 21, 2004 from AT&T, MCI and JA (filing jointly), SBC-CA , XO, Mpower, Pac-West, ORA/TURN, and CALTEL-Navigator.
As is the practice of this Commission, after reviewing the comments and replies, we made changes and corrections throughout the text. In addition, we discuss in this section the major issues raised by commenting parties.
CALTEL-Navigator objects vigorously to the use of SBC-CA's LoopCAT as part of the calculation of local loop rates, arguing that splitting the difference is inappropriate. AT&T similarly argues that it is inappropriate to rely on LoopCAT without correcting its "serious, well-documented overstatements." Among other things, AT&T argues that LoopCAT double counts loop installation costs. XO also comments adversely on the decision to use LoopCAT. ORA/TURN argue that rates that use LoopCAT are not "TELRIC compliant."
We find these arguments unpersuasive for a number of reasons. LoopCAT is superior to HM 5.3 in its compliance with TELRIC. For example, contrary to AT&T's assertions, LoopCAT reflects real-world considerations much more accurately than HM 5.3. HM 5.3 does not "model loop plant to the specific locations of every SBC-CA customer." As the maps presented in the June 2003 workshop make clear, not only does HM 5.3 discard all of SBC-CA's "actual" customer locations before its hypothetical network is modeled, but many of these so-called "actual" customer locations are not real at all, but are surrogates.
Second, this decision distinguishes between current costs (which are proper under TELRIC) and embedded (i.e., accounting) data -- a distinction that is fully supported by the record. LoopCAT's cost estimates are not embedded, but instead recognize that certain network characteristics (with appropriate forward-looking adjustments) represent the most sensible measurements of the physical characteristics of a forward-looking network capable of providing reliable service at existing customer locations and facility routes.
As a factual matter, we note that the estimates produced by SBC-CA's LoopCAT and HM 5.3 are virtually identical, with LoopCAT providing slightly lower cost estimates than those yielded by HM 5.3. It is therefore particularly difficult to find credible the arguments of the proponents of HM 5.3 that LoopCAT systematically overestimates costs when it yields estimates that have lower costs than those produced by the model that the opponents of LoopCAT sponsor or support. An examination of the appendices to this decision will show that there is little merit to the claims that LoopCAT produces outrageous results. Its results are as reasonable as those provided by HM 5.3.
It is well for all to realize that cost modeling is an exercise fraught with error. We found that SBC's LoopCAT better models the local characteristics of SBC-CA's service territory, but we also found that it is flawed. Similarly, although we found that HM 5.3 was flawed, we elected to rely upon it. We do not believe that the factual outcome - both models producing almost identical results for loop costs - undercuts our decision to use and average both models. Given the flaws documented for both models, this remains a prudent approach. We take the fact that both models produce almost similar costs as an indication that in general errors tended to offset each other, not that either is correct or that we could substitute one for the other.
AT&T claims that we have committed error in modeling the costs of residential premises termination equipment are wrong. AT&T's claim that the "material inputs" for a drop relate solely to material are false. The material inputs include both labor and materials. Second, 90% of the drop costs are labor related, and thus are insensitive to whether the drop consists of 2-pair or 6-pair. Furthermore, drop wire is considered "exempt" in this exercise, and a shift from 6-pair to 2-pair has no effect on the model. In addition, line protectors are required for all lines, even if not in use. Thus, AT&T's objections are in error.
Concerning copper distribution fills, we corrected the error pointed out in comments. Thus we now use an achieved fill of 41.7% in HM 5.3.
On other matters, parties reargued their previous positions, and we see no need to repeat the reasoning that led to our conclusions.
1. In D.95-12-016, the Commission adopted a set of Consensus Costing Principles that it has applied in TSLRIC and TELRIC cost proceedings.
2. Pursuant to federal regulations, the Commission must comply with the FCC's TELRIC methodology when setting UNE rates for SBC-CA.
3. The Commission established cost modeling criteria for this proceeding in a June 2002 Scoping Memo.
SBC-CA Models/LoopCAT
4. The SBC-CA models contain many inputs based on the characteristics of SBC-CA's current network operations.
5. The SBC-CA models as presented do not readily allow the Commission to isolate and determine SBC-CA model inputs related to loop length assumptions, structure sharing assumptions, and labor crew and installation time assumptions.
6. LoopCAT uses current cabling characteristics rather than cable-sizing conventions to optimize network design.
7. In its First Report and Order, the FCC rejected the inclusion of embedded and retail costs in TELRIC cost studies.
8. The factors in the SBC-CA models can be modified.
9. LoopCAT factors can be traced to SBC-CA internal accounting data, but this record does not match this data to publicly-available cost data, such as ARMIS, or other public sources of information.
10. Because LoopCAT uses a top-down approach and HM 5.3 uses a bottom-up approach to estimate installation costs, it is difficult to compare installation crew sizes in LoopCAT's factors to input assumptions in HM 5.3.
11. It is difficult to test the sensitivity of LoopCAT to differing forward-looking assumptions or network configurations.
12. LoopCAT approximates loop lengths for each distribution area assuming that all loops in that distribution area are one-half the "design point," which is defined by SBC-CA's loop planning guidelines as the longest loop that might be built in the next twenty years for existing or potential customer locations.
13. Approximately 100,000 loops in LoopCAT are longer than 18,000 feet, which is less than 0.5% of SBC-CA's loops.
14. Copper loops exceeding 18,000 feet will not work properly without additional equipment such as load coils, which are not modeled in LoopCAT.
15. The record does not contain information on SBC-CA's actual loop lengths to modify the design point distance to exclude potential loops.
16. LoopCAT assumes separate drops for each residence and equipment to terminate up to six lines for every residence.
17. The SBC-CA models calculate costs for 2-wire, DS-1, and DS-3 loops separately.
SICAT
18. SBC-CA's SICAT module calculates switching costs based on SBC's average purchases over a five-year period (1998 through 2002) under its multi-state contracts with switch vendors.
SPICE
19. SBC's SPICE model assumes that a forward-looking interoffice network would mirror SBC-CA's existing interoffice network characteristics.
20. SPICE does not produce a total investment figure, but instead calculates "node investment."
21. SPICE does not allow the user to segment demand for different interoffice services such as voice and high capacity services.
22. SPICE estimates costs using factors that incorporate structure sharing data, pole and conduit investment, and EF&I costs that are based on SBC-CA's historical network data.
23. One cannot readily extract individual input information from the factors used in SPICE in order to understand the underlying input, compare it to other public information, or test the effect of different input assumptions.
24. It is not possible to identify the demand level that the SPICE model is designed to serve.
25. The Commission cannot readily modify demand assumptions and factor inputs in SBC-CA's SPICE model without knowing the assumptions embedded in SBC-CA's factors.
ACFs and Expenses
26. The SBC-CA models use ACFs to convert investments into annual costs and expenses.
27. The expense factors in the SBC-CA models do not readily allow the Commission to isolate or understand individual input assumptions, compare and verify inputs to public information, or test differing assumptions.
28. In this proceeding, SBC-CA and Joint Applicants use a different cost methodology than the prior OANAD proceeding.
29. SBC-CA did not reconcile shared and common costs from the prior OANAD proceeding with the direct UNE costs calculated through the cost studies proposed in this proceeding.
30. SBC-CA relied on total expenses and investments when calculating its per unit expense factors, which means that its ACFs include expenses related to unregulated activities.
31. SBC-CA's ACFs include expenses related to transactions between SBC-CA and its affiliates.
32. SBC-CA's ACFs include expenses for Project Pronto incurred in 2001.
33. The TBO accrual is a liability for future retiree medical costs already earned by current and former employees.
34. SBC removed one-third of its estimate of TBO expenses from its ACF study.
35. The record does not contain sufficient information allowing the Commission to modify SBC-CA's expense assumptions to remove potential shared and common costs and Project Pronto costs.
36. SBC-CA incorporated inflation into its cost models through inflation factors for capital investments and operating expenses.
37. As a result of D.89-10-031, the Commission's New Regulatory Framework incorporates inflation and productivity adjustments.
38. BLS data shows telephone utility worker productivity has exceeded inflation from 1996 through 2000.
39. SBC-CA did not provide information related to actual installation times or material costs, except for DLC costs.
HM 5.3
40. Many of the inputs to HM 5.3 can be modified, such as fill factors, plant mix, structure sharing, switching investment assumptions, and some labor installation times and crew sizes.
41. HM 5.3 uses a customer location database created by a third-party vendor, TNS, as an input.
42. The Commission staff could not modify HM 5.3 inputs related to labor costs in all areas.
43. The HM 5.3 interoffice transport module underestimates demand.
44. In defining TELRIC, the FCC has accepted modeling based on "a reasonable amount of excess capacity to accommodate short term growth.
45. Loop lengths based on right-angle connections are longer than straight line connections because the two sides of a right triangle, when added together, are longer than its hypotenuse, but ignore real-world obstacles such as those posed by California's topology and urban characteristics.
46. LoopCAT is more accurate than HM 5.3 in modeling California geographic and urban characteristics because it follows existing feeder routes .
47. TELRIC favors but does not mandate the use of existing outside plant routes.
48. The Supreme Court has rejected basing UNE costs on an incumbent carrier's historical costs, as opposed to using current costs as a basis for developing forward-looking costs and network data.
49. SBC-CA provided actual DLC installation cost information that was lower than the DLC installation costs used in LoopCAT.
50. HM 5.3 uses SBC-CA customer location information to identify SBC-CA's current customer locations and cluster them into distribution areas.
51. The clustering algorithm used as an input to HM 5.3 imposed three engineering restrictions relating to maximum copper length, maximum lines served, and maximum distance between two points in the cluster.
52. SBC-CA ran its own clustering scenario with a maximum line size of 1,800 lines, although Commission staff was unable to run its own clustering scenarios.
53. The Commission staff could not fully replicate the preprocessing steps used in either HM 5.3 or the SBC-CA models.
54. LoopCAT assumes that SBC-CA's current customer groupings are forward-looking and efficient and does not regroup customers into different distribution areas based on current population characteristics.
55. TELRIC allows the reconstruction of the network using existing wire centers, and favors the use of actual outside plant routes because they likely represent the most efficient, forward-looking plant design.
56. HM 5.3 made simplifying assumptions about customers with the same address where it did not know the square footage "footprint" of a building.
57. HM 5.3 assumes distribution areas can accommodate a CEV up to 6,451 lines, which is larger than the CEV size SBC-CA normally installs.
58. The Commission could not run a scenario with a lower assumption regarding the maximum lines per distribution area.
59. Equipment to serve 7,200 pairs in a distribution area is available, but impractical to use on a widespread basis in California,
60. LoopCAT assumes that distribution areas serve a maximum of 200 to 600 households based on guidelines that have been in place for approximately 25 years.
61. HM 5.3 uses many inputs that are based on expert judgments and relies on vendor quotes that are not always documented. Some of these inputs can be modified and some cannot.
62. The SBC-CA models rely on judgments of engineers and subject matter experts for many inputs, such as design point assumptions, ACFs, SICAT, and SPICE inputs.
63. SBC-CA did not provide an assessment of new input values for many of the HM 5.3 inputs it criticized.
64. In many cases, it is not possible to make direct comparisons between HM 5.3 and SBC-CA model inputs.
65. It was not possible for us to change labor rate assumptions in HM 5.3 related to SAI investment, terminal and splice investment, buried drop installation, underground and buried trenching and riser cable investment, because they were often embedded with material cost and other input assumptions.
66. Neither SICAT nor HM 5.3 models the characteristics of individual switches.
67. HM 5.3 does not model an interoffice network that can accommodate all of SBC-CA's current interoffice high capacity demand.
68. TELRIC requires the modeling of forward-looking costs attributable to UNEs, taking as a given the incumbent LEC's provision of other elements.
69. HM 5.3 allows the user to adjust inputs to model varying levels of spare capacity.
70. Both HM 5.3 and the SBC-CA models adjust investments to current cost before calculating E/I ratios.
71. Verizon has higher investments per line than SBC-CA.
72. TURN compared HM 5.3, the SBC-CA models and SynMod using a uniform platform of loop-related and general input values from SynMod.
73. HM 5.3 produced lower costs than SynMod when run with SynMod's default inputs.
74. JA changed eight categories of inputs to HM 5.3, which resulted in a significantly higher loop rate.
75. HM 5.3 can be modified to use different input and engineering assumptions, spare capacity can be increased and expense assumptions can be modified, but it is not possible to modify HM 5.3 with regard to certain labor inputs, the customer clustering process, and demand assumptions in the interoffice transport module.
Resulting UNE Rates
76. When HM 5.3 and the SBC-CA models are run with similar inputs to match Commission precedent, federal requirements, and additional rationales developed herein, the resulting costs tend to converge, with some rates converging to the same value.
77. The SBC-CA models require time intensive efforts to modify, are prone to errors due to complex input modification requirements.
78. The LoopCAT module reasonably models California local loop costs.
79. HM 5.3's greatest weakness is its modeling of local loop costs.
80. The SBC-CA cost factor module is an integral component in all other SBC-CA cost modules and modifications to the cost factor module require numerous manual input changes to flow the results into other SBC-CA cost modules.
81. Both HM 5.3 and LoopCAT assume uniform distribution of customers throughout the distribution area.
82. Both HM 5.3 and LoopCAT include a mixture of real and hypothetical assumptions.
83. HM 5.3 uses actual customer locations, but clusters these locations into reconfigured, or hypothetical, groupings. The cluster sizes are greater than one can reasonably envision in California.
84. LoopCAT uses some existing plant routes, particularly for feeder, but designs distribution loop lengths based on estimates of customer locations.
85. HM 5.3 uses the TNS clustering database as an input, which the Commission is not able to adjust.
Asset Lives
86. New technological developments such as VoIP will likely decrease the economic life of copper distribution loops.
87. The asset lives adopted by the FCC do not match the financial asset lives proposed as modeling inputs by SBC-CA and are not appropriate for a forward-looking TELRIC.
Cost of Capital
88. Since 1994, several mergers and acquisitions have impacted the telecommunications industry including Pacific Telesis' merger with SBC, and SBC's subsequent merger with Ameritech.
89. SBC-CA proposes a proxy group of seven.
90. Both SBC-CA and JA use the CAPM and DCF methods to estimate cost of equity.
91. SBC-CA uses growth estimates from 1999 for its DCF analysis.
92. SBC-CA's interest rate adjustment to the market risk premium differs substantially from other measures of the market risk premium.
93. In prior cost of capital reviews, the Commission has adjusted cost of equity for interest rate changes after completing its CAPM review rather than incorporating interest rate changes into the CAPM model.
94. SBC-CA proposes a risk-free rate of 5.8% based on 1999 government bonds.
95. The CAPM computes a cost of equity for SBC-CA of 11.78% when it is run with a 7.4% market risk premium, a beta coefficient of .93, and a risk free rate of 4.9%.
96. The 11.78% cost of equity used to derive SBC-CA's cost of capital is slightly higher than the cost of equity adopted for California's energy utilities.
97. The Commission has generally excluded short-term debt when setting the cost of capital for utilities.
98. SBC-CA's proposed capital structure uses market values of equity and debt from 1998.
99. The firms in SBC-CA's proxy group have substantially increased their debt levels in recent years.
100. Ibbotson Associates has stated that a firm's target or optimal capital structure should be used in weighting the cost of equity and debt.
101. The capital structure proposed by JA, which mixes book and market values, does not comport with the FCC's TRO mandates.
IDLC/UDLC
102. SBC-CA's engineering guidelines call for greater deployment of IDLC systems when economical.
103. A CLC cannot gain access to an unbundled IDLC.
104. UDLC loops are required for circuits that cannot be provisioned over an IDLC system, such as ISDN, DS-1, and burglar alarms.
105. At present, there are no stand-alone loops provisioned over IDLC anywhere in the U.S.
DLC Costs
106. SBC-CA proposes a factor-based approach to estimate DLC installation costs in LoopCAT, based on the ratio of installation to material costs.
107. The cost data sample provided by SBC-CA shows lower DLC installation costs than those estimated by the factors in LoopCAT.
108. SBC-CA incurs DLC installation costs above and beyond those included in its contract with Alcatel.
109. Actual DLC installation costs in the sample data provided by SBC-CA are lower than the costs produced by DLC EF&I factors used in the SBC-CA models.
Fill Factors
110. There is a wide disparity between the fill factors SBC-CA proposes in its models and those used in its TSLRIC studies for pricing flexibility.
111. HM 5.3 uses SBC-CA's temporary engineering guidelines to design cable sizes to provide 1.5 to 2 lines per living unit for residential customers.
112. SBC-CA engineering guidelines call for 2.25 lines per lot.
113. A fill factor for copper distribution plant of 41.7% comports with the current experience of SBC-CA, and is likely to characterize its future experience.
114. In California, low density areas generally have higher fill levels, and this experience informs the LoopCAT model's higher fill levels in low density areas.
115. A fill factor for copper feeder plant of 66.2% comports with the current experience of SBC-CA, and is likely to characterize its future experience.
116. HM 5.3 models 4 fibers to each DLC site for redundancy, which results in a fiber fill rate of 79.6% that includes duplicate facilities. This approach is consistent with the approach used by the FCC in its universal service cost modeling.
117. SBC-CA proposes a 16.22% fiber feeder fill based on its actual utilization experience and the percentage of fiber strands that are actually in use, as well as channel utilization.
118. Because of an additional "channel fill," SBC-CA's fiber feeder fill rate is less than half its copper distribution fill rate.
119. For the DLC common equipment fill factor, HM 5.3 unrealistically incorporates a choice of DLC system sizes from 24 lines up to 8,064 lines, which exceeds the range of sizes deployed, or likely to be deployed, in California. SBC-CA models its four most common DLC sizes, which is less than the range of sizes SBC-CA actually deploys.
120. SBC-CA's DLC common equipment fill factor is based on its actual network operations and allows for ten years of spare capacity.
121. LoopCAT does not use the correct line capacity for a 6x16 CEV because of California real-world requirements that commonly necessitate below-ground placement.
122. SBC-CA models a fill factor for the CEV structure and a fill factor for the DLC equipment housed in the CEV.
123. SBC-CA engineering guidelines stress minimization of spare DLC plug-in equipment.
124. Placement of DLC plug-in equipment involves travel time to the DLC site and the inability to manage DLC channels on a single pair basis.
125. HM 5.3 models an SAI equipment fill level based on 3.5 lines per living unit.
126. SBC-CA admits an error in developing the SAI fill factor in its model.
127. The SBC-CA models allow the user to assume a linear relationship between maintenance costs and fill factors, so that maintenance costs rise at higher fill levels.
128. SBC-CA's analysis shows maintenance costs for copper distribution rise with fill levels above 50%.
Structure Sharing
129. The FCC's SynMod and SBC-CA's loop deployment guidelines assume sharing of structure by feeder and distribution cable.
Plant Mix
130. HM 5.3 uses averages of ARMIS data from the last eleven years to develop plant mix assumptions.
131. The SBC-CA models assume forward-looking plant mix matches current plant mix and will reflect the forward-looking mix as well.
Labor Costs
132. Labor costs in HM 5.3 and the SBC-CA models involve inputs for hourly wage rates, crew sizes, and task times.
133. Labor cost inputs in HM 5.3 involve witness opinion and national vendor quotes circa 1997. Not all labor costs can be easily adjusted in HM 5.3.
Crossover Point
134. In D.96-08-021, the Commission adopted an economic crossover point of 12,000 feet.
135. Copper loops longer than 12,000 feet are not consistently capable of supporting many services and loops longer than 12,000 feet present compatibility problems for UNEs.
Switch Vendors
136. SICAT switching investments are based on contracts with Lucent and Nortel, while HM 5.3 models investments based on Siemens switch prices.
137. At present, SBC-CA does not deploy Siemens switches in its California network.
138. The Siemens switch modeled in HM 5.3 with SONET based optical interface capabilities is not currently available in North America.
139. In D.98-12-106, the Commission rejected the assumption that 90% of lines could be purchased at the new line price.
Vertical Features
140. Vertical feature hardware costs are included as part of the per-line cost for new switch lines.
141. SBC-CA's feature cost study does not incorporate memory and processor costs, or costs for feature software for usage above caps in the switching contracts.
142. SBC-CA data shows that average processor utilization is well below 100%.
Switching Rate Structure
143. SBC-CA's switch contracts charge a flat price per line up to a given CCS, or usage per line, level.
144. Current statewide average CCS levels are well below the upper limit of CCS quantity provisioned under the switching contracts.
145. Based on SBC-CA's current switching network the price per line does not change as long as usage does not rise above the maximum CCS levels in the switch contracts.
146. Other states that have adopted a flat, per port rate structure for switching have retained usage-sensitive rates for reciprocal compensation purposes.
Switch Fill Factors
147. The SBC-CA models use switch fill levels from the current network.
148. Digital fill levels on the current network are low because digital technology is newly deployed.
DS-3 Loop Rates
149. The SBC-CA models calculate DS-3 loop costs on a deaveraged basis.
Annual Reexamination Process
150. In D.99-11-050, the Commission established a process for the annual review of UNE rates.
1. Neither the HM 5.3 nor the SBC-CA models allow the Commission complete flexibility to modify inputs and test various outcomes.
2. Both HM 5.3 and the SBC-CA models are flawed because the Commission is unable to modify key structural elements of either model.
3. The SBC-CA models do not fail the Commission's modeling criteria to such a significant extent that we cannot reasonably rely on them to set UNE rates.
4. It is reasonable to rely on the local loop costs produced by the SBC-CA LoopCAT model because it captures costs that arise from California's unique topology and regulatory environment.
5. The complexity of the SBC-CA models make it unreasonable to rely on them to model UNE costs when a more user-friendly model produces similar results.
6. HM 5.3 inputs, as filed, are at the low end of what we consider reasonable.
7. The results of the SBC-CA models provide estimates of forward looking UNE costs that we consider reasonable, albeit a bit high.
8. It is reasonable to rely on both the HM 5.3 and the SBC-CA LoopCAT models to set UNE-L rates.
9. It is not reasonable to rely on the HM 5.3 model alone to set UNE-L rates because of its inherent flaws in the cluster component of its model, which produces a hypothetical network model insensitive to California topographic and regulatory factors and incapable of providing service.
10. It is not reasonable to rely on the SBC-CA model as an input into UNE rates other than UNE-L. For these rate elements, the SBC-CA produces results similar to HM 5.3 and using the model requires great effort.
11. It is reasonable to use the HM 5.3 model alone to set UNE rates other than UNE-L, despite its flaws, because the model is easy to use and appears to produce results similar to that of the SBC-CA models.
12. The Commission should set the UNE-L rate for SBC-CA based on the midpoint of the HM 5.3 model and SBC-CA's LoopCAT as run with the inputs described in this order and as set forth in Appendix A. For all other UNE rates, the Commission should set prices for SBC-CA based on the HM 5.3 model as run with the inputs described in this order and as set forth in Appendix A.
SBC-CA Models
13. SBC-CA's existing cabling inventory, which reflects incremental network growth over many years, is most likely the way a real carrier would build a network in response to California's growth, regulatory requirements, and topological characteristics on a forward looking basis. .
14. SBC-CA's models generally comply with TELRIC standards because they estimate the cost to rebuild the network SBC-CA has in place today, with some changes for forward-looking technology.
15. The fact that SBC-CA has operated under incentive regulation for over ten years suggests that its models are forward-looking, even when individual modeling inputs, such as labor installation times, crew sizes, material prices and structure sharing, cannot be determined or modified to test differing assumptions.
16. Because LoopCAT relies on current cable and network characteristics, the model follows FCC guidance that TELRIC should assume reconstruction of the least-cost network configuration.
17. It is reasonable to rely on current accounting information used in SBC-CA's factors.
18. SBC-CA has not always shown why the cost data related to its current network should be considered forward-looking.
19. LoopCAT's use of the design point to calculate loop lengths results in a network configuration that is likely both least-cost and forward-looking.
20. LoopCAT overstates loop costs by not modeling multiple dwelling units, but instead assuming that each residence requires a separate drop and termination equipment for six lines.
21. Issues raised by parties regarding SBC-CA's SICAT model can be addressed by changing SICAT input assumptions.
22. SICAT is not irreparably flawed because it incorporates some non-California switching information based on SBC's multi-state switch vendor contracts.
23. The SPICE model is difficult to use.
24. The factors used throughout the SBC-CA models cannot be readily disaggregated to extract individual inputs, compare them to other public information, or modify them to test the effect of differing assumptions. This diminishes our confidence in the results of the model.
25. SBC-CA's ACFs may contain some portion of shared and common costs, because SBC-CA uses a different costing methodology than the prior OANAD and did not analyze whether ACFs it now proposes include expenses previously categorized as shared and common costs.
26. Based on the current record, there is no evidence one way or the other whether SBC-CA's ACFs double-count shared and common costs.
27. The inclusion of expenses for its unregulated businesses, such as inside wire maintenance and billing services, or for services SBC-CA performed on behalf of its affiliates in calculating expenses related to UNE operations diminishes our confidence in the results yielded by the SBC-CA model.
28. SBC-CA has not met its burden of proving that all of its Project Pronto expenses are appropriately allocated to UNEs.
29. SBC-CA's ACFs should not include the TBO accrual because it is not a current operations cost.
30. It is not possible, given the current record, to isolate and remove expenses for Project Pronto and shared and common costs from the SBC-CA models, but the midpoint approach, in combination with the fact that the SBC-CA models produce lower costs that HM 5.3, resolves any potential problem.
31. The SBC-CA models should not include inflation adjustments.
HM 5.3 Model
32. Some criticisms of HM 5.3 cannot be addressed by input modifications.
33. The clustering assumptions in HM 5.3 are worse than the loop input assumptions in the SBC-CA models because they have no tether to real California conditions or engineering standards. HM 5.3 has aspects of loop modeling that the Commission was unable to modify to its satisfaction.
34. The loop modeling and customer location process in HM 5.3 was not filed and lacks transparency and its vast amounts of preprocessed data limit the Commission's ability to test various scenarios.
35. Both HM 5.3 and SBC-CA's LoopCAT can be faulted for the accuracy of customer locations because HM 5.3 ignores customer locations when modeling the distribution plant and LoopCAT does not attempt to locate all existing customers and assumes they are all evenly dispersed throughout the distribution area.
36. HM 5.3 accounts for a reasonable level of growth in network demand and sizes the network to provide appropriate service quality.
37. HM 5.3's use of right-angle routing is not based on SBC-CA's actual outside plant routes, and there is no basis to conclude that it realistically reflects that networks cannot always follow either straight line or rectangular routes.
38. The fact that SBC-CA actual costs are multiples of the costs produced by HM 5.3 suggests that HM 5.3 is flawed in its grasp of California operating realities.
39. HM 5.3 is not more reasonable and forward looking than LoopCAT with regard to loop design because it creates customer clusters in sizes not practical for California
40. HM 5.3 reconstructs SBC-CA's network within TELRIC guidelines and given existing wire center locations, but fails to follow TELRIC guidelines and CCP #6 because it does not follow existing outside plant routes.
41. When HM 5.3 is re-run with clusters limited in size to 1,800 lines per cluster, the results demonstrate no tradeoff between feeder and distribution costs. This outcome is fundamentally illogical, and suggests that HM 5.3 is insensitive to basic changes in network design.
42. It is highly suspect for a forward-looking network configuration to size distribution areas much larger than SBC-CA has sized them.
43. It is unreasonable for HM 5.3 to assume that all distribution areas in California could accommodate a CEV to serve 6,451 lines.
44. The customer location and loop modeling assumptions in HM 5.3 are an unfiled "black box." This limits the Commission's ability to fully understand or replicate the preprocessing steps and decreases confidence in the model.
45. The proposal of MCI to model labor costs in HM 5.3 is a reasonable way to overcome the model's inherent weaknesses on this point.
46. HM 5.3 and SICAT have taken a similar modeling approach that does not analyze the characteristics of individual switches.
47. Because HM 5.3 does not incorporate optical interface equipment, HM 5.3 would not allow the provisioning of the high capacity services SBC-CA provides today.
48. Because HM 5.3 appears insensitive to demand changes, it is unclear how it derives its SONET ring structure to set interoffice transport rates.
49. The HM 5.3 interoffice transport module is flawed because it underestimates demand, may not incorporate optical interface equipment, and is insensitive to demand changes.
50. HM 5.3's demand assumptions incorporate a low level of growth.
51. The use of E/I ratios in HM 5.3 should be adjusted to remove comparisons to Verizon expense levels.
52. It is reasonable to use recent data from SBC-CA's ARMIS expense information to estimate forward-looking expenses with the HM 5.3 model.
53. SBC-CA has not shown that its current costs are forward-looking, and it would be unreasonable to reject HM 5.3 merely because its results are lower than SBC-CA current costs.
54. HM 5.3 is not structurally biased to produce low results for loop costs.
55. Both HM 5.3 and the SBC-CA models fail the Commission's cost modeling criteria, as set forth in the scoping memo of this proceeding because neither allows us to reasonably understand all inputs or modify inputs and assumptions in all areas.
Resulting UNE Rates
56. The Commission should not rely solely on the results of either HM 5.3 or the SBC-CA models, but should use the results of both models to estimate the UNE-L rate.
57. It is burdensome to use the SBC-CA models because the models lack flow-through capability and require extensive manual manipulation.
58. HM 5.3 is less burdensome to operate because we can reasonably understand how to make necessary modifications.
Asset Lives
59. The asset lives proposed by SBC-CA are reasonable, in part because of technologies such as VoIP make it unwise to lengthen depreciation schedules for copper cable to provide broadband services.
60. The economic asset lives proposed by DOD/FEA based on an analysis by the FCC are not reasonable to use as inputs for our TELRIC cost modeling because they do not account for facilities-based competition as required by TELRIC.
Cost of Capital
61. The cost of capital originally adopted for SBC-CA in 1994 should be revised because financial conditions today are vastly different than they were at that time.
62. It is reasonable to use the proxy group of three companies proposed by JA to analyze the cost of equity, debt, and capital structure for our cost of capital analysis.
63. SBC-CA's growth estimates used in its DCF analysis are outdated and not reasonable, and its updated growth estimates in the "b x r approach" are excessive.
64. JA's three stage DCF analysis, based on more current growth rates than SBC-CA's analysis, is more reasonable than assuming all telecommunications firms will grow continuously at a faster rate than the whole economy.
65. SBC-CA's interest rate adjustment to the market risk premium is not reasonable because of updated assumptions regarding interest rate effects on equity premiums and SBC-CA's updated market risk premium of 9.86% is based on an unreasonably short study period.
66. A market risk premium of 7.4%, based on Ibbotson Associates study of equity premiums from 1926 to 2001, is reasonable to use in our CAPM analysis because it is based on documented equity returns rather than disputed expectation of future returns.
67. It is more reasonable to base a risk-free rate on 30-year bonds, rather than 10-year bonds, to match the longer investment horizon in our market risk premium figure.
68. A risk free rate of 4.92% is more reasonable than SBC-CA's outdated risk free rate.
69. We should adopt SBC-CA's updated beta coefficient of .93 because it is based on recent data for the same proxy group that we use for our other cost of capital inputs.
70. When setting the cost of equity, we should give no weight to the DCF model results because DCF relies heavily on widely disparate growth forecasts for telecommunications firms.
71. It is reasonable to adopt an 11.78% cost of equity based on the conservatively higher CAPM results.
72. It is reasonable to determine a cost of capital by looking at the returns investors require in capital markets.
73. It is reasonable to assume that capital markets have already figured the relative risk of UNEs into the equity returns they require for SBC's stock.
74. SBC-CA's UNE business is subject to regulatory risk regarding the accuracy of UNE prices and competitive risk.
75. The Commissions' cost of capital analysis should incorporate long-term debt costs that match UNE asset lives, and are less volatile than short-term debt costs.
76. It is reasonable to assume a long-term debt cost of 6.34% for our analysis.
77. A capital structure based solely on market values is appropriate for TELRIC pricing.
78. It is not reasonable to base a capital structure on a firm's target capital structure, which includes a mix of market and book values.
79. The Commission should use a capital structure of 74% equity and 26% debt.
IDLC/UDLC
80. UDLC is the forward-looking technology choice for network design.
81. The Commission should assume a mix of 5% IDLC and 95% UDLC in its model runs because UDLC is the forward-looking technology that permits the unbundling of loops.
DLC Costs
82. It is reasonable to incorporate DLC installation costs above and beyond those listed in the Alcatel contract in our TELRIC model runs.
83. SBC-CA could not reasonably explain how LoopCAT's DLC installation factor was derived.
84. SBC-CA has not shown that its DLC installation cost factor is reasonable and forward-looking because it is greater than actual cost information it provided.
85. The Commission's model runs should incorporate SBC-CA's DLC installation costs of $22,814 for RTs and $49,569 for CEVs, rather than LoopCAT's factors or the estimates proposed by JA.
Fill Factors
86. Fill factors derived purely from current network operations are not automatically forward-looking, but SBC-CA's experience in operating under incentive regulation has provided strong incentives to deploy spare capacity efficiently.
87. Fill factors should reflect accurate projections of investment to accommodate growth and a reasonable estimate of demand.
88. A fill factor for copper distribution of 41.7%% provides an adequate level of spare capacity to accommodate a reasonable projection of future demand, and is therefore, reasonable.
89. The fact that SBC-CA's fill factors may remain constant over time suggests that these fill levels are optimal.
90. SBC-CA's distribution fill factor is a reasonable proxy for forward-looking utilization.
91. It is reasonable to use the 66.2% copper feeder fill factor that SBC-CA experiences today.
92. A 79.6% fiber feeder fill rate is reasonable because it is similar to the approach used by the FCC in its modeling and it provides full redundancy and spare for growth.
93. A fiber feeder fill rate of 16.22% is not forward looking because it incorporates 80% spare capacity and it contradicts SBC-CA's statements that optimal fill rates for feeder plant are higher than for distribution.
94. SBC-CA's DLC equipment fill factor is reasonable because it is based on its current experience with this technology.
95. A 47.4% fill factor for DLC common equipment is reasonable because it allows for growth and comports with SBC-CA's current experience.
96. SBC-CA's fill factor for DLC plug-in equipment is based on its current experience.
97. A DLC plug-in equipment fill factor of 53.1% is reasonable given SBC-CA's current experience..
98. It is reasonable to use an SAI fill factor of 67.8% given the admitted errors in SBC-CA's fill factor.
99. HM 5.3 undersizes premise termination equipment by modeling only two pairs per residence, which leaves no room for a third line.
100. SBC-CA overestimates premise termination equipment by modeling equipment for 6 line terminations at each residence, which is greater than forward-looking estimations of lines per residence and ignores the economies of serving multiple dwelling units.
101. Neither HM 5.3 nor the SBC-CA models determine appropriate NID sizes based on multiple dwelling units. It is reasonable to modify HM 5.3 and the SBC-CA models to run with similar assumptions regarding the NID and premise termination fill factors for residential customers.
102. SBC-CA's proposed linkage of fill factors and maintenance expenses is not reasonable because it has only analyzed the effect of fill levels on one aspect of loop costs rather than total loop costs.
103. SBC-CA has not shown a linkage between higher fill levels and higher maintenance for feeder, DLC equipment, or switching equipment.
104. The linkage of fill factors and maintenance expenses is not reasonable because SBC-CA has not shown the linkage applies to anything other than copper cable with distribution fills above 50%.
Structure Sharing
105. The structure sharing percentages between utilities assumed in HM 5.3 are not reasonably supported.
106. The structure sharing percentages in the SBC-CA models are not reasonable because they cannot be identified.
107. It is reasonable to use the structure sharing percentages adopted by the FCC in its Synthesis Model.
108. It is reasonable to assume 55% sharing of feeder and distribution facilities given findings of the FCC on this subject and SBC-CA's own guidelines.
Plant Mix
109. It is reasonable to adopt SBC-CA's plant mix assumptions for our model runs rather than assumptions based on ARMIS data dating back 11 years.
Labor Costs
110. It is more reasonable to use SBC-CA's proposed fully loaded hourly wage rates than expert judgment, whenever possible.
111. HM 5.3 underestimates crew sizes in certain circumstances, such as cable placing.
112. The labor loading adjustments suggested by JA's witness Flappan are not reasonable because they are based on nationwide information for companies that are not reasonably similar to SBC-CA.
Crossover Point
113. Loops longer than 12,000 feet cannot provide other UNEs as required by TELRIC unless additional equipment is added.
114. The Commission should model a crossover point of 12,000 feet.
Switch Vendors
115. It is more reasonable to model Lucent and Nortel switches in SBC-CA's network because Siemens switches may not provide all the functions and capabilities provided by the switches currently deployed in SBC-CA's network.
New and Growth Lines
116. It is reasonable to average the JA forecast concerning the percentage of lines that can be purchased at the new line discount with the SBC- CA forecast.
117. SBC-CA has not adequately supported its "other replacement costs" that it models with switching investments. These should be removed from Commission model runs because they appear to relate to SBC-CA's embedded switching network.
118. It is reasonable to include upgrade costs in our switching investment modeling.
Vertical Features
119. Feature hardware and software costs that are incurred through per line, per switch, or buyout charges should be modeled in the monthly port price.
120. SBC-CA's feature cost study double counts feature hardware costs that are already included in the per line switching price.
121. Costs to upgrade the switch memory and processor are included in switch upgrade costs as part of the per line switching investment.
Rate Structure
122. It is unreasonable to assume SBC-CA will exceed the capacity limitations in its switch vendor contracts because switches are provisioned based on a 10-year forecast of capacity requirements and average utilization is below the minimum switch capacity that is provisioned under the contracts.
123. UNE Switch pricing should be a flat price per line because SBC-CA switching costs do not change given today's switch usage, as long as usage does not rise above the maximum CCS level in the switching contract.
124. A flat per port price for switching usage is consistent with TELRIC guidance that rate structures should reflect the manner in which costs are incurred.
125. The flat per port switching rates adopted in this order should not apply in the context of reciprocal compensation between carriers because changes to reciprocal compensation rate structures are beyond the scope of this proceeding. The usage sensitive rates shown in Appendix C can be used for reciprocal compensation purposes.
Switch Fill Factors
126. It is not reasonable for SBC-CA to apply a low digital fill factor when it contends its customer base is declining.
127. SBC-CA has not shown why analog and digital switch fill rates should differ so drastically.
128. Switching equipment is highly modular and can be expanded in less than a year, so it is reasonable to use higher fill rates for switching equipment.
129. It is reasonable to apply an 82% fill factor to analog and digital lines in the models.
DS-3 Loop Rates
130. It is reasonable to adopt deaveraged DS-3 loop rates because the models can calculate costs to support this result.
131. We should take official notice of DS-1 and DS-3 loop cost calculations proposed by JA in the Verizon UNE Phase of R.93-04-003/I.93-04-002.
Annual Reexamination Process
132. Absent action by the FCC, the Commission should suspend further reexamination of UNE prices until February 2007 to provide wholesale pricing stability in the local exchange market.
IT IS ORDERED that:
1. The recurring prices for unbundled network elements (UNEs) offered by Pacific Bell Telephone Company d/b/a SBC California (SBC-CA) that are set forth in Appendices A and C to this decision satisfy the requirements of Sections 251(c)(2), 251(c)(3), and 252(d)(1) of the Telecommunications Act of 1996 and are hereby adopted.
2. Pursuant to Commission Resolution ALJ-181 (adopted October 5, 2000), SBC-CA shall prepare amendments to all interconnection agreements between itself and other carriers. Such amendments shall substitute the recurring UNE prices set forth in Appendices A and C for the UNE prices set forth in such interconnection agreements. Such amendments shall be filed with the Commission's Telecommunications Division, pursuant to the advice letter process set forth in Rules 6.1 and 6.2 of Resolution ALJ-181, within 30 days after the effective date of this order. The amendments do not require a signature of the carriers involved as long as the amendments are limited to substituting the UNE rates adopted in today's order. Unless protested, such amendments shall become effective 30 days after filing. The flat per port switching rates adopted in this order shall not apply in the context of reciprocal compensation between carriers. The rates shown in Appendix C shall be used for reciprocal compensation purposes.
3. The UNE prices adopted in this order shall be effective on the date this order is effective. SBC-CA shall make all billing adjustments necessary to ensure that this effective date is accurately reflected in bills applicable to these UNEs. SBC-CA shall have 60 days from the date of this order to complete the billing program changes necessary to reflect in bills the recurring prices for UNEs adopted in this order. Upon completion of said billing program changes, SBC-CA shall notify the Director of the Telecommunications Division in writing that all of the necessary billing program changes have been completed.
4. Within 90 days of the effective date of this order, SBC-CA shall calculate any billing adjustments owed to or by interconnecting carriers based on the modification of interim rates set in Decision 02-05-042 and Decision 02-09-052 to the rates in this order, but payment of any billing adjustments, or "true-up," is stayed pending the outcome of further proceedings in this docket to consider payment options or other mitigations to lessen any negative effects of the true-up and whether and how to implement any shared anc common cost markup revisions along with the true-up. The Administrative Law Judge shall issue a ruling within 30 days of this order setting a prehearing conference to initiate these proceedings.
5. The annual nomination procedure set forth in Ordering Paragraph 11 of Decision (D.) 99-11-050 is suspended until 2007 or until the FCC revises its methodology for calculating TELRIC, whichever comes first. SBC-CA or carriers with which SBC-CA has interconnection agreements, may file nominations of UNEs for review, as described in D.99-11-050, between February 1 and March 1, 2007.
6. Official notice is taken of the DS-1 and DS-3 loop cost calculations proposed by AT&T Communications of California and MCI-WorldCom Inc. in the Verizon UNE phase of Rulemaking 93-04-003/Investigation 93-04-002.
7. Application (A.) 01-02-024, A.02-02-035, A.02-02-031, A.02-02-032, A.02-02-034, and A.02-03-003 shall remain open pending resolution of true-up payment issues.
This order is effective today.
Dated _____________________, at San Francisco, California.
Appendix D
Glossary of Acronyms
ACF Annual cost factor
ARMIS Automated Reporting Management Information System
BLS Bureau of Labor Statistics (U.S. Dept. of Labor)
CAPM Capital asset pricing model
CCPs Consensus Costing Principles
CCS Centi-call second
CEV Controlled environmental vault
CLC Competitive local exchange carrier
DA Distribution area
DCF Discounted cash flow
DEM Dial Equipment Minutes
DLC Digital loop carrier
DSL Digital subscriber line
EF&I Engineer, furnish and install
FCC Federal Communications Commission
HM 5.3 HAI Model, Version 5.3
IDLC Integrated digital loop carrier
ILEC Incumbent local exchange carrier
IOF Interoffice facilities
LEIS Loop engineering information system database
LoopCat Loop Cost Analysis Tool
MDU Multiple dwelling unit
MST minimum spanning tree
NID Network interface device
NPRM Notice of Proposed Rulemaking
OANAD Commission Rulemaking 94-04-003 regarding "Open Access and Network Architecture Development"
POTS Plain old telephone service
RBOC regional bell operating company
ROE return on equity
RT Remote terminal
SAI Serving area interface
SICAT Switching Cost Analysis Tool
SONET Synchronous optical network
SPICE SBC's Program for Interoffice and Circuit Equipment
SS7 Signaling System 7
SynMod FCC's Synthesis Model
TBO Transitional benefit obligation
TELRIC Total element long run incremental cost methodology
TSLRIC Total service long run incremental cost methodology
TNS Taylor Nelson Sofres
TRO FCC's Triennial Review Order
UDLC Universal digital loop carrier
UNE Unbundled network element
UNE-P Unbundled network element platform
APPENDIX E
LIST OF APPEARANCES
Applicants: Goodin, MacBride, Squeri, Richie & Day, LLP, by John Clark, Attorney at Law, for The Telephone Connection Local Services, LLC; David Discher and Stephanie E. Krapf, Attorneys at Law, for Pacific Bell Telephone Company; William C. Harrelson, Attorney at Law, for MCI/WorldCom, Inc.; Preston, Gates, Ellis & Rouvelas Meeds, LLP, by Christopher S. Huther and Megan Troy, Attorneys at Law for Pacific Bell Telephone Company; David J. Miller and W. Clay Deanhardt, Attorneys at Law, for AT&T Communications of California, Inc.; Earl Nicholas Selby and Karen M. Potkul, Attorneys at Law, for XO California, Inc.
Interested Parties: Goodin, MacBride, Squeri, Richie & Day, LLP, by John Clark, Attorney at Law, for Z-Tel Communications, Inc.; William J. Cobb III, Attorney at Law, for Covad Communications Company; Christine Mailloux, Attorney at Law, and Regina Costa, Representative, for The Utility Reform Network; Adams, Broadwell, Joseph & Cardozo, by Katherine S. Poole and Lonnie Finkel, Attorneys at Law, for Communications Workers of America, District 9; Terrance A. Spann, Attorney at Law, for United States Department of Defense and All Other Federal Executive Agencies; Glenn Stover, Attorney at Law, for Sage Telecom, Inc., Tri-M Communications, Inc., and Anew Telecommunications Corporation; Morrison & Foerster, LLP, by James M. Tobin, Mary E. Wand and Theresa L. Cabral, Attorneys at Law, for Pac-West Telecomm, Inc.;
Office of Ratepayer Advocates: Natalie Billingsley, Representative, and Natalie Wales, Attorney at Law.
