VI. Modeling Inputs and Other Changes
We now turn to a description of the modeling inputs that were most fervently debated by the parties in this proceeding. The sections below briefly discuss these disputes and resolve them by indicating the inputs that were selected for the Commission's run of HM 5.3.
A. Asset Lives and Depreciation
One of the expenses included in a TELRIC model is depreciation expense. In order to forecast depreciation expense, the models rely on assumptions regarding the economic lives of the assets used to provision UNEs, that is, the rate at which these assets depreciate.
The key dispute between the parties involves whether to adopt Verizon's proposed asset lives, based on the economic lives Verizon uses for financial reporting purposes, or the proposal of DOD/FEA and JC based on asset lives projected by the FCC. In 1996, the Commission endorsed the use of the economic lives used by Pacific Bell for financial reporting purposes as the appropriate forward-looking lives for UNE cost studies. In 2004, the Commission again considered whether to use the FCC's asset lives for SBC, and was not persuaded to change its earlier determination. (D.04-09-063, mimeo. at 137.)
The table below compares the opposing proposals for four key categories of asset lives, and also shows the lives adopted for SBC in 2004.
Table 3
Proposed Asset Lives
Asset |
DOD/FEA/JC Proposed Asset Life |
Verizon Proposed Asset Life |
SBC Adopted Asset Life |
Switching Equipment |
14 |
12 |
10 |
Circuit Equipment |
11 |
8 |
9 |
Metallic Cable (All) |
19-25 |
15-17 |
15 |
Non-Metallic Cable |
25 |
20 |
20 |
According to Verizon, the asset lives it proposes consider current network modernization strategies, the impact of technology and competition, regulatory commitments, state demographics, and wear and tear. (Verizon/Sovereign, 11/3/03, p. 9.) Verizon asserts that competition spurs technological development, shortens the economic life of existing assets, and makes them obsolete. Further, facilities-based competition diverts traffic from the ILEC's network to competitive local carriers' (CLCs) networks. (Id., p. 11.) Verizon compares its proposed asset lives to those forecast by Technology Futures Inc. (TFI), an independent research organization that specializes in technology market forecasts. Verizon indicates that its proposed lives fall within the range of lives proposed by TFI. (Id., pp. 20-21.)
In contrast, DOD/FEA propose that in order to set forward-looking UNE rates, the cost models should use economic depreciation rates based on the expected economic lives of newly placed plant, often termed the "projection lives." (DOD/FEA, 11/3/03, pp. 3-4.) DOD/FEA's witness Richard Lee supports use of the projection lives prescribed in 1996 by the FCC for Verizon-Contel as the most realistic and forward-looking estimates of plant lives. Lee contends that recent trends in depreciation reserve levels indicate that the FCC's projection lives are forward-looking. (Id.)
JC support DOD/FEA's proposal to use depreciation lives prescribed by the FCC and dispute Verizon's assertion that competition will necessarily shorten depreciation lives. (JC, 11/3/03, p. 24.) Several JC witnesses provide analyses of other network industries that, in the face of emerging competition, put older technologies to new and different uses to extend the economic lives of various assets. (JC, 8/6/04, p. 49.) TURN agrees with the proposal of DOD/FEA, and alleges flaws with Verizon's TFI study. TURN's witness Loube contends that if the Commission agrees with Verizon that customer demand for higher speed services shortens the useful life of copper plant, the depreciation cost caused by this demand should be allocated to those high-speed services. (TURN/Loube, 8/6/04, p. 15.)
In response to the DOD/FEA proposal, Verizon contends the Commission should not rely on the asset lives prescribed by the FCC in 1996 for Contel because they predate the competitive and technological developments that have occurred since the passage of the 1996 Telecommunications Act. (Verizon, 8/6/04, p. 88.)
In our SBC UNE reexamination, we found it was appropriate to continue to use asset lives developed for financial reporting purposes as the basis for determining depreciation expense. This was the approach we had initially adopted for Pacific Bell in D.99-11-050, and we reiterated that view in
D.04-09-063. We will not deviate from that finding here. Furthermore, we agree with Verizon that reliance on asset lives the FCC prescribed in 1996 for Contel is unwise given the competitive and technological developments since that time. Verizon's witness Sovereign adequately described the methodology he used and factors he considered in developing Verizon's proposed asset lives, including the effects of technological innovation. (Verizon/Sovereign, 11/3/03, pp. 9-16.) Finally, there is little rationale for different asset life assumptions between SBC and Verizon, since forward-looking technology assumptions for the two carriers should be similar, if not identical. The lives Verizon proposes are quite similar to the lives we adopted for SBC in D.04-09-063, as shown in Table 3 above. Therefore, we have run HM 5.3 using the asset lives proposed by Verizon.
We reject TURN's suggestion that any increased depreciation expenses from shorter asset lives should be assigned to advanced services that create the need for Verizon to update its plant. We find that depreciation expense based on Verizon's asset lives properly compensates Verizon for the TELRIC costs associated with providing the UNEs at issue in this proceeding.
B. Cost of Capital
A critical input to a TELRIC cost model is the estimated cost of capital, which is the cost a firm will incur in raising funds in a competitive capital market. The cost of capital is usually expressed as a weighted average of the cost of equity and the cost of debt for the firm, or a proxy group of firms, with a similar risk profile and in the same line of business as the firm. There are several key components used to calculate the weighted average cost of capital:
· Cost of equity -The Capital Asset Pricing Model (CAPM) and the Discounted Cash Flow (DCF) analysis technique are two quantitative financial models commonly used to estimate cost of equity, also called return on equity (ROE). These methods require assumptions regarding company growth rates, the premium that a stock of average risk commands over the risk free rate (market risk premium), the risk-free rate of return, and a measure of the risk of the company's stock (beta).
· Cost of debt - this involves estimates of the interest rates on long-term, and perhaps short-term, debt instruments.
· Capital structure of the firm - this refers to the ratio of debt and equity outstanding for the company, or proxy group.
· Proxy group - this key assumption involves the composition of the group of companies used as comparables to the ILEC's UNE business.
Federal regulations require that a "forward-looking cost of capital shall be used in calculating the [TELRIC] of an element." (47 C.F.R. 51.505(b)(2).)
In its Triennial Review Order, the FCC provides clarification on the cost of capital component of a TELRIC analysis. The FCC states that there are two types of risk that should be reflected in the cost of capital. First, a TELRIC-based cost of capital should reflect the risks of a competitive market. Specifically, the FCC says:
Because the objective of TELRIC pricing is to replicate pricing in a competitive market, [footnote omitted] and prices in a competitive market would reflect the competitive risks associated with participating in such a market, we now clarify that states should establish a cost of capital that reflects the competitive risks associated with participating in the type of market that TELRIC assumes. The Commission specifically recognized that increased competition would lead to increased risk, which would warrant an increased cost of capital. (TRO, para. 681.) (Footnote omitted.)
Second, the FCC states that a TELRIC-based cost of capital should reflect any unique risks (above and beyond the competitive risks discussed above) associated with new services that might be provided over certain types of facilities. The TRO specifies that states may establish UNE-specific costs of capital to reflect in UNE prices any risk associated with new facilities that employ new technology and offer new services. (TRO, para. 683.) Nonetheless, the FCC leaves states the option to adopt a single cost of capital for all UNEs. (TRO, para. 684.)
Table 4 summarizes the parties' proposals for the appropriate cost of capital to incorporate into Verizon's UNE prices. For reference purposes, the Commission adopted a cost of capital of 9.44% applicable to SBC's UNE rates. (D.04-09-063, mimeo. at 171.)
Table 4
Current and Proposed Cost of Capital for Verizon
Verizon Current |
Verizon Proposal |
Joint Commentors Proposal |
XO Proposal |
ORA Proposal |
TURN Proposal |
10.51% |
14.37%22 |
7.64%23 |
No more than 8.63%24 |
7.4%25 |
7.93%26 |
While these proposals differ by almost 700 basis points,27 the methods used by all parties are remarkably similar. Verizon, JC, and TURN offered the most commentary concerning cost of capital. These parties calculated a weighted average cost of capital based on their own unique assumptions regarding the cost of equity, cost of debt, and capital structure of the firm. We will discuss each of these components of the cost of capital calculation separately. But first we will give a brief overall description of each party's proposal.
According to Verizon, the FCC has emphasized that the cost of capital in UNE cases should reflect the risks of a competitive market in which all facilities-based carriers would face the risk of losing customers to other facilities-based carriers. (TRO at para. 680.) Therefore, it proposes a cost of capital of 14.37 %, which is based on a weighted average cost of capital of 11.64% plus a risk premium of 2.74%. (Verizon, 8/6/04, p. 81, n. 425.) Verizon explains that its risk premium proposal compensates for the regulatory risks inherent in providing UNEs, particularly the risk that UNE leases may be canceled on short notice. Since a cancelable lease has an economic value, Verizon has quantified that value and reflected it in the 2.74% risk premium that it adds to its cost of capital calculations. (Verizon, 11/3/03, pp. 7-8.) Other key assumptions in Verizon's proposal are a cost of equity based on a DCF analysis of the S&P Industrials, and a capital structure of 25% debt and 75% equity based on an average market value for a proxy group of S&P Industrial companies and a group of telecommunications companies.
JC's witness Murray proposes a cost of capital of 7.64% based on an analysis using data from a proxy group of three ILECs, including Verizon, that are subject to facilities-based competition. Murray uses the CAPM, with a beta of 1.0 to determine the cost of equity, and corroborated her results with a three stage DCF analysis based on data for the three RBOCs. She develops long and short-term debt costs using forward-looking yield to maturity for publicly-traded debt of the Verizon companies. For capital structure, she uses a target capital structure based on the average of market and book capital structure for the three firms in her proxy group. (JC/Murray, 11/3/03, pp. 69-70; JC/Murray, 11/9/04, p. 3.)
TURN proposes a cost of capital of 7.93%, derived using the same method as JC's witness Murray, with updated inputs. (TURN/Loube, 8/6/04, p. 85.) Specifically, Loube updates short and long-term interest rates, using those published by the Federal Reserve on June 1, 2004. (Id., pp. 70-86 and Exhibit RL-8.) TURN opposes Verizon's proposed cost of capital because it relies on a cost of equity method the FCC rejected in the Virginia Arbitration. (Id., p. 78.)
XO disagrees with Verizon's proposed risk premium, claiming that Verizon falsely characterizes the competitive and UNE pricing environment as posing unusual risk for the company. XO maintains Verizon faces no different conditions than any firm in a fully competitive market and has not provided any actual data to show Verizon faces risk from cancellation of UNE orders. In other words, XO contends Verizon has not shown CLCs purchase UNEs for a shorter time period than Verizon's retail customers. (XO, 8/6/04, p. 45.)
ORA recommends the Commission adopt a cost of capital similar to one adopted for SBC because Verizon and SBC are similarly situated companies. (ORA/Litkouhi, 11/9/04, p. 2). The Commission should reject Verizon's proposed cost of capital as unreasonably high, overstating the degree of demand and competitive risk, and inappropriately including a risk premium adder. ORA prefers the method suggested by JC's witness Murray, but removes short-term debt from Murray's weighted average cost of capital calculations.
Despite the large variance in cost of capital proposals, all parties essentially used the same financial modeling techniques, but with differing inputs and assumptions. We analyze each of their positions on the various components of the financial models below in order to determine the most reasonable inputs for financial modeling of the cost of capital. A summary of the financial modeling with the inputs we select is found in Section VI.B.5.
It is important to note that while we will review the financial modeling presented by the parties, particularly where it estimates the cost of equity, we will use judgment as well as the models to render our decision. As we stated in our order in 2002 where we established a return on equity for the four major energy utilities:
In the final analysis, it is the application of informed judgment, not the precision of financial models, which is the key to selecting a specific ROE estimate. We affirmed this view in D.89-10-031, which established ROEs for GTE California, inc. and Pacific Bell, noting that we continue to view the financial models with considerable skepticism. (D.02-11-027, mimeo. at 19.)
Finally, although the FCC's TRO discusses the option to set unique costs of capital for each UNE, we will establish one cost of capital for all UNEs because we have no record to do otherwise.
We now turn to an examination of the inputs to the financial models used by the parties.
a) Cost of Equity
Verizon, JC and TURN use different approaches to estimate the cost of equity. Verizon uses the DCF methodology to estimate the cost of equity, while JC and TURN use the CAPM approach.
According to Verizon, the DCF model is often used by economists to estimate a firm's cost of equity based on the assumption that the market price of a firm's stock is equal to the present value of the stream of cash flows that investors expect to receive from owning that stock. (Verizon/Vander Weide, 11/3/03, p. 17.) The DCF method requires assumptions about the future dividends and growth rate of the companies being studied to forecast the future cash flows that will accrue to shareholders into the indefinite future. (JC/Murray, 11/3/03, p. 59.)
Verizon uses the DCF methodology applied to a proxy group of over 100 S&P industrial companies. Verizon justifies using this proxy group because there are no publicly-traded companies solely in the business of operating a telecommunications network to provide UNEs. Verizon believes a
well-known sample of companies operating in competitive markets is the best available proxy. The proxy group includes a broad array of S&P companies as diverse as 3M Company, Avon Products, Coca Cola, Gap Inc., Halliburton, Marriot International, and Procter & Gamble. According to Verizon, although the proxy companies do not have subsidiaries in the UNE business, this is not important as long as they are of similar risk to the entity whose cost of capital is being estimated. (Verizon/Vander Weide, 8/6/04, p. 33.) Using the DCF method, Verizon calculates a weighted average cost of equity of 13.46% for its proxy group of S&P Industrials. (Id., p. 70.)
JC, ORA, TURN and XO claim Verizon's proxy group and DCF analysis are flawed. JC's witness Murray notes the extreme diversity in the proxy group and claims Verizon's witness Vander Weide offers no sound explanation for why such diverse firms form an appropriate proxy group for a cost of capital analysis. (JC/Murray, 8/6/04, pp. 49-50.) XO echoes these concerns noting that Verizon's proxy group is based on a subset of S&P 500 companies, none of which are telecommunications carriers or are primarily involved in the telecommunications industry. Further, Vander Weide excludes S&P companies that he defines as outliers, including those with negative growth and companies with low equity costs. (XO, 11/9/04, p. 25.) ORA claims Verizon's approach contradicts the outcome of the SBC UNE decision, where the Commission looked to a group of companies in a similar line of business to determine capital structure, cost of equity, and cost of debt for SBC. (ORA/Litkhouhi, 11/9/04, p. 3.) JC and TURN criticize Vander Weide's constant growth assumptions that underlie his DCF analysis. According to TURN, Vander Weide's average growth assumptions are 11.26%, while current estimates of the growth of the U.S. economy are only 6%. (TURN/Loube, 8/6/04, p. 79.)
JC propose a cost of equity of 9.14%. (JC/Murray, 11/9/04, p. 43.) To arrive at this cost of equity, Murray employs the CAPM, which requires assumptions regarding the historical and forward-looking market risk premium, risk-free interest rates, and beta.28
Murray performs several CAPM analyses with differing assumptions for these three main inputs and averages the results. For all her CAPM calculations, Murray uses a beta of 1.0 because the FCC's Virginia Arbitration rejected the use of ILEC specific betas below 1.0 as outdated and from the period when ILECs were predominantly near-monopolies whereas they now face increased competition for long-distance, local services, and broadband markets. (JC/Murray, 11/3/03, p. 53.) For her CAPM analysis based on historical inputs, Murray uses an estimate of the market risk premium from Ibbotson Associates ranging from 7.2% to 8.6%, resulting in a cost of equity of 11.17%. She then performs a second CAPM analysis based on an average of forward-looking estimates of the market risk premium, resulting in a cost of equity of 7.11%. The average of these two CAPM analyses results in Murray's proposed cost of equity of 9.14%. (JC/Murray, 11/9/04, Exh. TLM-REB-3.)
As a check on the reasonableness of her analysis, Murray performs a three stage DCF analysis using data for Verizon and two comparable ILECs, BellSouth and SBC. The three stage DCF model is a common alternative to Verizon's one-stage DCF model, and it assumes three stages with distinct growth rates that converge toward the future rate of overall economic growth. Murray's three stage DCF model provides a cost of equity of 9.41% (JC/Murray, 11/3/03, pp. 61-63.)
Verizon responds that Murray's three stage DCF is inferior to the single stage DCF model that Vander Weide used, primarily because it produces counterintuitive results wherein higher risk companies garner lower returns. (Verizon/Vander Weide, 8/6/04, p. 51.) Regarding Murray's CAPM analysis, Verizon criticizes the interest rates Murray uses as too short-term. Verizon also contends Murray should use a beta greater than 1.0, in line with the betas of other competitive telecommunications firms such as Level 3 and AT&T, with betas in the range of 1.5 to 2. (Id., p. 62.) Finally, Verizon expresses general reservations with CAPM as not capturing all the risks that affect cost of equity and containing significant problems in estimating the model's basic inputs, i.e. the risk-free rate, the beta, and the expected return on the market portfolio. (Id., p. 63.)
TURN reviewed both Murray's and Vander Weide's cost of equity analyses and contends that Murray's analysis is closer to FCC guidance from the Virginia Arbitration because it uses CAPM and a beta of 1.0. TURN performs its own analysis using CAPM, updating the short and long-term interest rates in Murray's analysis to arrive at a cost of equity of 9.04% (TURN/Loube, 8/6/04, Exh. RL-8, Table 5.)
The debate over cost of equity first hinges on whether we should use CAPM or DCF. We have already faced this issue in the SBC UNE decision, where we discussed our reservations with the DCF model, which relies heavily on growth forecasts for firms. The growth forecasts can lead to a large disparity in DCF results depending on the time period and forecasters selected. We found the DCF model too dependent on this one forecasted input and opted instead to rely on CAPM results to set the cost of equity for SBC. (D.04-09-063, mimeo. at 159.) Here, we find the same issue arises again. Verizon proposes a cost of equity based solely on the DCF approach, using growth forecasts for a group of 100 S&P Industrials, and providing little explanation of how the growth forecasts were selected. Furthermore, Verizon provides scant information on why this particular proxy group of non-telecommunications firms was deemed to have the same risk profile and growth forecast as Verizon.
JC's witness Murray performs her own variation of a DCF analysis, using growth forecasts for her proxy group of three telecommunications firms, and arrives at a significantly lower cost of equity 9.41%, 405 basis points lower than Verizon's. Thus, we again see huge disparities in DCF results depending on the proxy group and growth forecasts. As in the SBC case, we will not rely on any DCF results to set the cost of equity for Verizon. We are comfortable ignoring the DCF analyses provided by both parties, because as TURN points out, the FCC Wireline Competition Bureau also chose to ignore DCF analyses, noting that "Verizon's use of the constant growth DCF model to estimate the cost of equity...stretches the reasonable limits of its use." (Virginia Arbitration, para 73.)
Even if we were open to using the DCF approach, we disagree with Verizon's choice to use a proxy group for its analysis that has no connection at all to the telecommunications industry. Verizon assumes that the companies in its proxy group have the same risk as a firm building a telecommunications network. Even though the companies in Verizon's proxy group are in competitive industries, Verizon has not provided any plausible support for its assumption that the risk of building a facilities-based telecommunications network equates to the risks faced by Marriot, Coca Cola, Halliburton, or any of the firms in the proxy group. When choosing a proxy group for a cost of equity analysis, we find it significantly more reasonable to choose a proxy group of firms in the same industry as Verizon. This allows comparison of what investors require for returns on telecommunications firms facing the same technological and regulatory conditions. Verizon's choice of a non-telecommunications industry proxy group leads us to reject Verizon's cost of equity DCF analysis.
Now that we have rejected the parties' DCF analyses, we turn to the CAPM analyses provided by JC and TURN. The CAPM is the approach we relied on in the SBC case to set a cost of equity of 11.78% for SBC.29 In the SBC case, we relied on an historical estimate of the market risk premium to arrive at this result.
After reviewing the CAPM analyses provided by JC and TURN, we find it appropriate and reasonable to use SBC's 11.78% cost of equity for Verizon as a starting point. If we update our analysis from the SBC proceeding to use a beta of 1.0 rather than .93, the resulting cost of equity is 12.3%. The beta of 1.0 was used by JC's witness Murray in her various CAPM analyses, and by the FCC in its Virginia Arbitration Order. (Virginia Arbitration, para. 90.) We find it reasonable to use our CAPM analysis from the SBC proceeding, which is similar to the analyses performed here by JC's witness Murray, although we will update the beta in our analysis to 1.0.
The CAPM analyses provided here by JC and TURN corroborate and support this 12.3% figure. JC's Murray calculates a 12.03% long-term CAPM cost of equity, based on an historical market premium of 7.2%. This is almost identical to the CAPM analysis in D.04-09-063 except that Murray now uses a beta of 1.0. However, Murray takes the additional step of averaging her 12.03% projection with further CAPM analyses involving short-term and forward-looking inputs. We have less confidence in Murray's short-term and forward-looking CAPM results because we find a longer-term, historical approach more reliable and consistent with our analysis in the SBC case. Plus, forward-looking estimates of interest rates and market returns vary greatly and are highly disputed among financial experts, and a long-term projection of the cost of equity better matches the long-term investments required for a telecommunications network. The effect of Murray's additional CAPM runs is to water down her overall projection with lower estimates of the cost of equity based on disputed inputs. Therefore, we will use a cost of equity of 12.3% for Verizon, noting that JC's historical and long-term CAPM results of 12.03% support this outcome.30
In comments on the draft decision, the Joint CLCs31 contend the Commission ignores Murray's CAPM analysis using an intermediate horizon historical risk premium, as described in her rebuttal declaration. (JC/Murray, 11/9/04, para. 91.) Murray's intermediate horizon proposal relies on five-year Treasury bond yields in the CAPM equation, which Murray contends reasonably match the time horizon over which the UNE prices adopted in this proceeding will be in effect. Verizon responds that the cost of capital should reflect investor expectations over the life of Verizon's assets, not simply until Verizon's UNE rates are reset. (Verizon, 12/21/05, p. 7.) Verizon requests we modify the beta in our CAPM analysis to 1.0. We agree with Verizon that CAPM analyses based on short term investor expectations are not appropriate when setting a cost of capital for a longer life telecommunications assets. We modify the beta in our CAPM from the SBC order to 1.0 for a revised cost of equity of 12.3%.
b) Cost of Debt
Verizon's witness Vander Weide uses a 6.15% cost of debt for his analysis, based on the yields on Moody's A-rated industrial bonds as of April 2004. (Verizon/Vander Weide, 8/6/04, p. 70.) He contends this benchmark interest rate best approximates what Verizon would actually pay on debt issued to finance the construction of a new telephone network. He does not include short-term debt because Verizon does not generally finance investments in its long-term assets in this manner. Murray criticizes Verizon's approach as not industry specific and based on a bond maturity of 30 years, substantially longer than the average 17-year asset life Verizon has used in its cost studies. (JC/Murray, 11/9/04, pp. 75-77.)
In contrast, JC's witness Murray developed long-term and short-term debt costs using the forward-looking yield to maturity for publicly-traded debt of the Verizon companies, similar to the approach used in the Virginia Arbitration. (JC/Murray, 11/3/03, p. 64.) She uses a long-term debt rate of 4.99% and short-term debt of 2.77% (Id., 11/9/04, p. 43.) Murray supports her use of short-term debt by noting that SBC and BellSouth both recently announced or completed debt issuances that are quite similar to the data on Verizon and include a mixture of short and long-term debt maturities. (Id., p. 78.)
Verizon criticizes Murray's use of short-term debt, contending it would not use it to finance the construction of a new telecommunications network and claiming Murray ignores the reality that Verizon's current long-term debt is near maturity and trading as short-term debt. (Verizon/Vander Weide, 8/6/04, p. 45.) Further, Verizon contends Murray ignores the average asset life of 17 years assumed in Verizon's TELRIC analysis. (Id., p. 44.)
TURN urges the Commission to reject Verizon's cost of debt because Vander Weide relies on the industrial cost of debt rather than debt associated with the telecommunications industry. (TURN/Loube, 8/6/04, p. 79.) For its own analysis, TURN proposes a debt rate of 5.69%. This rate is calculated using Murray's short-term debt rate and an updated long-term debt rate of 6.57% based on 2004 telephone bond rates. (Id., p. 83.)
We find it most reasonable to use the long-term debt cost for industrial companies proposed by Verizon, which is the 6.15% rate on Moody's A-rated industrial bonds cited by Verizon's witness Vander Weide in his update of August 6, 2004. Vander Weide's declaration attests that telecommunications companies such as Verizon are considered A-rated industrial companies and the 20-year average term to maturity on the bonds is close to the 17 year average economic life of Verizon's forward-looking network. (Verizon/Vander Weide, 11/9/04, p. 26-8.) We prefer this rate because the longer term of this debt is a closer match to the asset life assumptions incorporated into our model runs.
We decline to use Murray's analysis, which includes short-term debt costs because, as we stated in the SBC UNE case, we are not convinced that short-term debt has a place in a TELRIC-based cost of capital analysis where we prefer to use long-term financing assumptions to match asset lives. (D.04-09-063, mimeo. at 166.) Similarly, we decline TURN's proposal because it includes short-term debt.
c) Capital Structure
Verizon recommends use of a market value capital structure, similar to the approach used in the Virginia Arbitration. According to Verizon's witness Vander Weide, a market value capital structure is more forward looking than book value because investors look only to the future to determine the value of their stocks and bonds, whereas book value is based on the embedded or historical costs of a company's assets. (Verizon/Vander Weide, 11/3/03, p. 40.) Based on his review of market value capital structures for both a proxy group of S&P Industrials and a group of telecommunications companies, he recommends a capital structure for Verizon of 25% debt and 75% equity. (Id., p. 41.)
Murray opposes using a market value approach, stating it does not provide the best guide to Verizon's forward looking target capital structure. (JC/Murray, 11/3/03, p. 65.) Generally, Murray criticizes Verizon's approach on the basis that a historical review of market valuations of S&P Industrials does not represent the best available information concerning investors' expectations about how Verizon would finance forward-looking investments in UNE operations. (Id., 8/6/04, p. 61.)
For her own analysis, JC's Murray cites several financial economists' views that ideally, a firm's target or optimal capital structure should be used in weighting the cost of equity and cost of debt. (Id., 11/3/03, p. 66; 8/6/04, p. 62.) Murray says respected researchers, such as Prof. R. Glenn Hubbard and Dr. William H. Lehr, have found evidence that, in the long run, market equity tends to move toward book equity. (Id., 8/6/04, pp. 62-63.) On the other hand, high market-to-book ratios predict future book profitability. Thus, on balance, this suggests the best prediction of a firm's target capital structure incorporates both book and market information. Murray, therefore, gives equal weight to the market and book capitalization of the companies in her proxy group. She recommends a capital structure of 66.44% equity, 28.63% long-term debt and 4.93% short-term debt. (Id., p. 43.) Murray provides information that this capital structure is highly consistent with the publicly stated target capital structures of other major ILECs, corroborating the reasonableness of her approach. (Id., 11/3/03, p. 69 and 8/6/04, p. 63.)
Verizon opposes Murray's approach, stating that economic research does not support the theory that market and book values of companies converge. (Verizon/Vander Weide, 11/9/04, p. 34.) Further, Verizon argues that target capital structures of other ILEC's have been misinterpreted by Murray. (Id., 8/6/04, p. 42.)
TURN supports use of the target capital structure, as proposed by Murray, rather than a capital structure based solely on market value. (TURN/Loube, 8/6/04, p. 77.) ORA and XO support JC's proposal to average market and book value capital structure for a comparable group of companies, excluding short-term debt. (ORA/Litkouhi, 8/6/04, p. 11.) Both ORA and XO claim that Verizon's witness Vander Weide gives no support for his proposed capital structure of 75% equity and 25% debt. (Id., p. 4; XO, 11/9/04, p. 25.)
Similar to the approach we used in setting a cost of capital for SBC, we adopt JC's approach of averaging market value and book value information for a proxy group of companies. As stated in D.04-09-063, we reject a capital structure based entirely on market value as too volatile and subject to fluctuations in stock prices. Rather, we have previously found that a forward-looking capital structure for a firm is based on a firm's target capital structure, and the best predictor of target capital structure for a firm uses both market and book value information, just as investors might do in valuing a company's assets.
(D.04-09-063, mimeo. at 169.) JC have convincingly shown that the target capital structures of other telecommunications companies compare reasonably to the proposed capital structures developed from book and market value information for Murray's proxy group. Based on this evidence, we find Murray's proposed capital structure reasonable, although we will exclude short-term debt. As we did in the SBC case, we decline to adopt a capital structure that includes short-term debt. Instead, we will make a simplifying assumption that all debt is held at the long-term rate, consistent with our assumptions regarding asset lives. Therefore, we adopt a capital structure of 66.44% equity and 33.56% debt.
d) Risk Premium Adder
Verizon proposes an adder of 2.74% to its cost of capital to compensate for the regulatory risks in providing UNEs. Vander Weide arrives at this adder amount by estimating the risk Verizon assumes in providing UNEs through a cancelable lease arrangement. According to Verizon, cancelable operating leases involve significantly higher risk for Verizon because its network investment is large, long-lived and largely sunk and its investments and operating expenses will remain the same even if CLCs are able to cancel their UNE leases as lower-cost substitutes become available. This increases the risk that Verizon will be able to earn a fair return on its UNE investments. (Verizon/Vander Weide, 11/3/03, pp. 48-49.) Verizon provides examples of the risk involved in facilities-based network investments by citing investments by WorldCom, Global Crossing, Qwest, Teligent, and Covad where these companies have found telecommunications demand was overestimated and the companies have lost 80% to 100% of their market values. (Id., p. 51.)
JC counter Verizon's UNE risk adder by maintaining there is no need for such an adder and that similar proposals by Verizon and other ILECs have been rejected numerous times. JC provide key reasons that Vander Weide's analysis of the risks of UNE leases is faulty, including (1) investors have presumably reflected this risk in the prices they are willing to pay for Verizon securities, (2) any risk, if it did exist, is retail rather than wholesale, because Verizon does not incur sunk investment costs specifically for UNEs, (3) if there is no risk adder to Verizon's retail cost of capital, there is no need for one for the UNE cost of capital, and (4) Verizon's network assets have other revenue generating uses that Vander Weide has ignored in his lease cancellation analysis. (JC/Murray, 8/6/04, p. 66.) Moreover, Murray notes that other agencies have found that any regulatory risk is captured by the market-based cost of capital. Murray cites as examples several FCC orders, this Commission's earlier OANAD decision for SBC (D.99-11-050), and decisions by other state commissions. (Id., pp. 67-69.)
TURN opposes Verizon's risk premium adder because cost of equity analyses already incorporate assumptions about competition. (TURN/Loube, 8/6/04, p. 80.) Further, Loube contends Verizon's justification for requiring a risk premium is speculative and unsupported, relying on assumptions that may not occur in the future. (Id., p. 77.)
ORA and XO both note the Commission has already rejected the idea of a risk adder for Pacific Bell in the earlier OANAD proceeding. (See D.99-11-050, mimeo. at 37-43.) In D.99-11-050, the Commission described how Pacific Bell's arguments for a "sunk cost" adder were really a collateral attack on the TELRIC methodology and inconsistent with a federal court ruling finding that an adder such as the one proposed by Pacific in OANAD was inconsistent with the basic pricing standards contained in Section 252(d)(1) of the Act. (Id., pp. 37 and 43.) Further, the decision found that Pacific Bell had not shown that an adder for future stranded plant is appropriate. (Id., pp. 42-43.)
We agree with Murray that Verizon has not justified a premium over the market based cost of capital calculated using CAPM and a weighting of the portion of debt and equity in the company's capital structure. We maintain the view that quantitative models, such as CAPM, do a reasonable job of capturing investors' views of the risks facing Verizon in the UNE market. Further, we agree with Murray that any risk from UNEs is no greater than the risk Verizon faces in its retail operations, particularly since Verizon does not have to incur "sunk investments" solely for UNE purposes. As pointed out by ORA and XO, the Commission has already rejected previous risk adder proposals. Verizon's arguments here echo Pacific Bell's proposal that was rejected in D.99-11-050. Therefore, we reject Verizon's proposal for a risk adder of 2.74% to its cost of capital.
The results of our cost of capital analysis are summarized in the table below. In short, we derive the capital structure for our analysis based on Murray's proposed 50/50 weighting of market and book values for her proxy group of firms, although we exclude Murray's use of short-term debt and will consider all debt as long-term. We use a 12.3% cost of equity based on Murray's long-term, historical CAPM analysis and our findings in the SBC UNE case updated to a beta of 1.0. We give no weight to the parties' DCF analyses. The 6.15% cost of debt is based on Moody's A-rated industrial bond yields. Altogether, these inputs result in a weighted average cost of capital for Verizon of 10.23%.
Table 5
Weighted Average Cost of Capital
Component |
Percent of Total |
Cost |
Weighted Cost |
Equity |
66.44% |
12.3% |
8.17% |
Debt |
33.56% |
6.15% |
2.06% |
100% |
10.23% |
C. IDLC/UDLC
A key modeling input involves the technology choice for digital loop carrier electronics. Digital loop carriers (DLCs) are the electronics that connect fiber feeder cable to copper distribution cable, and which allow telecommunications services to pass from copper to fiber and back, and between the fiber feeder and the switch.
JC propose that all DLC systems should be modeled as "integrated" or IDLC systems. In an IDLC system, voice signals remain digital all the way from the remote terminal to the switch. JC contend that IDLC is the more recent and forward-looking technology, and is more efficient and reliable than "universal" (UDLC) systems. (JC/Donovan-Pitkin-Turner, 8/6/04, p. 116.) According to JC, an IDLC system can be used to provision a stand-alone unbundled loop at the DS-1 level using an interface known as GR-303. (Id., p. 118.) JC claim that this capability exists today in the DLC systems Verizon has deployed throughout its network. (Id., p. 119.)
In contrast, Verizon has modeled a portion of its DLC systems as UDLC. Specifically, Verizon assumes use of 90% IDLC systems, and 10% UDLC systems. (Verizon Recurring Costs Testimony, 11/3/03, p. 50.) In a UDLC system, voice signals are converted from analog to digital at the remote terminal, then converted back to analog at the central office. Verizon incorporates some UDLC into its model under the theory that a forward-looking network must allow a carrier to provide unbundled loops to its competitors and it is not technically feasible in a multi-carrier environment to provision a single, or "stand-alone" unbundled loop using an IDLC system. (Verizon, 8/6/04, p. 57.) Verizon contends that various unresolved problems prevent provisioning of stand-alone loops over IDLC systems, including operational, security, and administrative concerns. (Id., p. 57.) Essentially, Verizon says it is unclear how different switches owned and operated by competing carriers can connect to one DLC system.
In the SBC UNE case, we found that IDLC was the forward-looking technology choice to include in our models runs, but that we should incorporate a portion of UDLC to account for operational issues yet to be resolved with provisioning single unbundled loops to CLCs. We used a mix of 60% IDLC and 40% UDLC, assuming that 40% of loops would need UDLC available for unbundling purposes. Here, Verizon has assumed that a forward-looking network can operate with only 10% UDLC equipment. JC continue to propose, as in the SBC case, that the network can operate with 100% IDLC.
We will use Verizon's input assumptions of 90% IDLC and 10% UDLC for our model runs for the same reasons elaborated in the SBC case, namely that some portion of UDLC may be required for unbundling purposes until operational issues are resolved. If Verizon believes that 10% UDLC is adequate in the forward-looking environment, we see no reason to use the 40% level adopted in the SBC case. We do not agree with JC's assessment that the network can operate with 100% IDLC for the same reasons articulated in the SBC UNE decision. (D.04-09-063, mimeo. at 173-175.)
D. Fill Factors
The parties have varying proposals for the amount of spare capacity that should be designed in a forward-looking local exchange network. In TELRIC cost models, designing a network with spare capacity entails use of a "fill factor," or utilization level, as a modeling input. For example, a fill factor of 40% means that 40% of the physical plant is in use, while 60% is available for maintenance, customer churn, and growth. (See D.96-08-021, mimeo. at 23.)
As the FCC stated in 1996 in its First Report and Order:
We conclude that, under a TELRIC methodology, incumbent LECs' prices for interconnection and unbundled network elements shall recover the forward-looking costs directly attributable to the specified element, as well as a reasonable allocation of forward-looking common costs. Per-unit costs shall be derived from total costs using reasonably accurate "fill factors" estimates of the proportion of a facility that will be "filled with network usage); that is, the per unit costs associated with a particular element must be derived by dividing the total cost associated with the element by a reasonable projection of the actual total usage of the element. (First Report and Order, para. 682.)
Key fill factors in HM 5.3 determine the appropriate investment for copper distribution cable, fiber feeder facilities, copper feeder facilities, DLC equipment, serving area interfaces (SAIs), and premise termination equipment. These fill factors are usually hotly disputed in TELRIC models because the lower the fill factor, the more spare, or excess, capacity will be included in the cost study. If fill factors include more spare capacity than is needed for a reasonable projection of forward-looking demand, plant costs will be inflated. Conversely, if modeling assumptions minimize excess capacity and lead to high achieved fill rates, costs of excess plant may be minimized at the expense of adequate spare plant to achieve reasonable service quality and service to new connections.
In the SBC UNE case, we reviewed the proposed fill factors in great detail. The fill factors that were adopted after that review are largely the same as those proposed here by JC, except for DLC and SAI fill levels. SBC proposed lower fill factors, which were largely rejected because they were derived from SBC's current network operations and were not considered forward-looking.
(D.04-09-063, mimeo. at 183.) The table below shows the "achieved" fill factors32 adopted in the SBC UNE proceeding, those proposed here by JC and Verizon, and a summary of the fill factors adopted in this order.
Table 6
Comparison of Fill Factors33
SBC Adopted Achieved Fill Factor |
JC Proposed Achieved Fill Factor |
Verizon Proposed Achieved Fill Factor |
Adopted Fill Factors | |
Copper Distribution |
51.6% |
51.7% |
38.76% |
52% |
Fiber Feeder |
79.6 |
79.6 |
95.5 |
80 |
Copper Feeder |
76 |
77.3 |
65.99 |
76 |
DLC Common |
62 |
78.8 |
83 |
65 |
DLC Plug In |
75 |
89.9 |
100 |
75 |
SAI |
67.8 |
55.9 |
42.3 |
56 |
We briefly discuss the fill levels we adopt for our HM 5.3 model run below.
Copper distribution fill relates to the amount of copper facilities, or line pairs, that are modeled in the distribution network. JC propose almost the identical 52% fill level that we adopted for SBC in D.04-09-063. Verizon proposes a fill factor of 38.76%. JC contend that Verizon's low fill is at odds with what was considered forward-looking for SBC and very likely reflects the historical practice of substantially overbuilding the distribution network. (JC/Donovan, 11/9/04, pp. 92-93.) In contrast, JC maintain their proposed inputs result in an achieved fill of approximately 50%, which provides facilities to serve almost twice the current demand level. (JC/Donovan, 11/3/03, p. 23.) Verizon alleges JC's sizing inputs provide minimal spare copper distribution plant that will lead to longer downtimes and poor service quality. (Verizon, 8/6/04, pp. 61-62.)
A fill level similar to the one Verizon proposes here was deemed to include excessive spare capacity in the SBC UNE case, while a fill level identical to the one JC now propose was considered reasonable to provide adequate spare capacity for customer churn, maintenance, and growth. (D.04-09-063, mimeo. at 189-90.) We continue to find that a fill factor that reserves close to 50% of capacity as spare is reasonable, particularly in light of Verizon's admission that Verizon, SBC and BellSouth have seen business and consumer access lines fall 3.6%, 4.1% and 3.2%, respectively, in 2002, with wireless substitution seen as a significant factor. (Verizon, 11/9/04, p. 5.) Therefore, we will adjust JC's proposed modeling inputs to achieve a fill factor of approximately 52%.
JC propose sizing factors for HM 5.3 that achieve a fill level for fiber feeder of approximately 80%, identical to what was adopted in the SBC case. As they explain, the inputs are based on an assumption of 4 fibers per DLC site, or two redundant fibers for each two fibers in service. (JC/Donovan 11/3/03,
pp. 27-28.) This is identical to the modeling we adopted in the SBC case, and we will use it here as well. Verizon proposes six strands per DLC terminal, which we consider excessive spare capacity.
HM 5.3 sizing factor inputs provide an achieved fill rate of 77.3% for copper feeder, almost identical to the 76% adopted in the SBC UNE case.
(D.04-09-063, mimeo. at 192.) Verizon uses almost identical sizing factor inputs in its model, but the achieved fill in the Verizon model is 66%. We see no reason to deviate from the 76% achieved fill rate that we adopted in the SBC UNE case, and we have adjusted HM 5.3, as we did in the SBC case, so that the sizing factors achieve that fill level.
JC propose a fill level for DLC plug-in equipment, i.e. line cards, of 89.9%. This is the same fill level that we rejected in the SBC UNE case as too high. Instead, for SBC UNE pricing, we assumed a 75% fill factor based on the finding that a level approaching 90% ignored real world constraints such as inventory management and travel time. (D.04-09-063, mimeo. at 199.) In Verizon's model, DLC equipment achieved fill factors are outputs, not inputs. Verizon did not use fill factor or breakage assumptions for DLC plug-in equipment. Thus, its achieved fill factor is 100%. Because of the different modeling approaches of Verizon and JC, it does not make sense to substitute Verizon's achieved DLC fill factors into HM 5.3. Rather, for the same reasons we discussed in the SBC case, we prefer to assume a lower fill level than the one proposed by JC. It is reasonable to assume Verizon would face the same inventory management and travel constraints that we assumed for SBC's utilization of DLC equipment. Therefore, we will run HM 5.3 to achieve a fill level of 75% for DLC plug-in equipment.
In the SBC UNE case, we learned that fill level for DLC common equipment is a modeling output that depends on the chosen level of DLC plug-in equipment. (D.04-09-063, mimeo. at 198.) Since we have adopted a 75% plug-in fill factor, the resulting DLC common equipment fill is 65%, nearly identical to our results in the SBC UNE case.
Premise termination equipment refers to the equipment that terminates a local loop at each customer location and includes the drop-wire from the distribution network to the "network interface device" (NID) on the customer premise.
The HM 5.3 model assumes a 2-pair NID for each residence that is not in a multiple dwelling unit, and a 6-pair NID for each business location. Verizon contends that a 2-pair NID assumption for each residence is inadequate because many customers will ultimately demand more than two lines and extra costs will be incurred when additional field visits are made to replace the 2-pair device when more lines are ordered. (Verizon Rebuttal Panel on VzCost and VzLoop, 11/9/04, p. 73.)
In the SBC UNE proceeding, we found that the assumption of a 2-pair NID per residence left no room for spare, but that a 6-pair device inflated loop costs by installing more equipment than necessary. To resolve the dispute, we increased the labor component of NID installation by lengthening the assumed install time for the NID to one hour to conservatively account for travel and set up time for multiple NID installations. (D.04-09-063, mimeo. at 204.) As we discuss in greater detail below in Section VI.H, we have chosen to run HM 5.3 with specific labor input changes suggested by Verizon. These changes include modified NID labor inputs and achieve the same result as the changes we made in the SBC case, namely increasing NID installation costs. While we will use Verizon's NID labor inputs, we will adhere to the 2-pair NID assumption, as we did in the SBC case. We consider this reasonable because by increasing labor assumptions, we account for the possibility that a second visit to enlarge the
2-pair NID may be required in certain circumstances.
A Serving Area Interface (SAI) is the equipment in the loop network that connects feeder and distribution facilities. JA initially proposed a fill factor for SAI equipment of 90.3%, then adjusted this fill factor on rebuttal to 56% in response to criticism by Verizon that the SAIs modeled in HM 5.3 did not have sufficient capacity to terminate all feeder and distribution pairs. (JC/Mercer-Pitkin-Turner, 11/9/04, p. 118.) Verizon's model achieved an SAI fill factor of 42.3%, based on the distribution areas in Verizon's model. Since we are not using the Verizon model for costing purposes, Verizon's fill level is not relevant.
The JC's SAI fill factor results from modeling assumptions of 3.5 lines per residential living unit, and 2 lines per business. (JC/Mercer, 11/3/03, Exh. RAM-4, p. 41.) These are identical assumptions to those we adopted in the SBC UNE proceeding, although the resulting fill factor is lower than what we adopted for SBC because SAI sizes and clusters are unique to Verizon. We will adopt the inputs for HM 5.3 that lead to an SAI achieved fill of 56%, based on the changes presented by JC in their rebuttal version of HM 5.3.
E. Structure Sharing
"Structure sharing" refers to the modeling assumption that poles and conduit modeled in a forward-looking network may be shared with other utilities. It also refers to the assumption that even within one company's network, feeder, distribution, and interoffice facilities may share the same poles and conduit. In the cost models, a lower structure sharing percentage indicates less costs are borne by Verizon because more structure costs are shared with other utilities.
For its model, Verizon assumes that forward-looking structure sharing will match the levels that are reflected in its current network experience. (Verizon Recurring Cost Testimony, 11/3/03, p. 55.) Specifically, Verizon assumes pole sharing reflects Verizon's actual inventory of poles it owns versus those it shares, and it assumes no sharing of buried placement costs based on its current experience. (Id., pp. 54, 57.) In contrast, JC contend that state regulatory commissions and the general public may require more structure sharing among utilities in the future, to reduce costs and prevent disruptions from excavation and other construction. Thus, JC contend that on a forward-looking basis, Verizon's engineers will implement more structure sharing than Verizon's current network experience. (JC/Donovan, 11/3/03, pp. 54-55.) HM 5.3 also reflects sharing of structure between feeder and distribution cable by assuming a default value of 55% for sharing of feeder and distribution facilities. (Id., p. 55.)
JC criticize Verizon's structure sharing assumptions as dramatically understated and merely invoking its embedded network. (JC/Donovan-Pitkin-Turner, 8/6/04, p. 146.) TURN also criticizes Verizon's structure sharing inputs for assuming there will be little opportunity to increase sharing percentages. TURN recommends the Commission look to structure sharing inputs used by the FCC in its Virginia Arbitration, noting both AT&T and Verizon accepted elements of the FCC inputs as reasonable in that case. (TURN/Loube, 8/6/04, p. 51.) Specifically, TURN recommends a 50% sharing assumption except in the three lowest density zones, where TURN recommends the Commission use the FCC inputs. (Id., p. 52.)
In contrast, Verizon criticizes JC's structure sharing assumptions because they ignore Verizon's actual experience and rely on speculation by JC's witnesses. JC erroneously assume that all networks, including those of utility and cable providers, are rebuilt simultaneously, so that each provider would be ready and willing to share structure costs with the hypothetical new entrant in a TELRIC model. Specifically, JC assume that other service providers will finance up to 75% of pole costs, up to two-thirds of Verizon's underground construction costs, and 75% of the cost to bury cable. (Verizon, 8/6/04, p. 63.)
In the SBC UNE decision, we found fault with the proposals of both SBC and the competitive carriers with regard to structure sharing input percentages. Instead, we found it reasonable to use the percentages relied on by the FCC for its Synthesis Model. (D.04-09-063, mimeo. at 210.) In this proceeding, Verizon has proposed structure sharing percentages that match its current network experience, whereas JC propose inputs based on assumptions on the amount of sharing in a forward-looking environment. After reviewing these inputs, we once again prefer to adopt a middle ground between Verizon's historical experience and JC's optimistic assumptions. TURN has suggested a 50% sharing assumption coupled with inputs used by the FCC in the Virginia Arbitration for the lowest density zones. Upon review, we find TURN's approach is overly simplistic and assumes more sharing of buried and underground plant than our SBC inputs. Further, it appears TURN's proposal does not exactly mirror all the Virginia Arbitration sharing inputs, just some of them.
We find it reasonable to use the structure sharing inputs we used for SBC in D.04-09-063, which are based on the FCC's Inputs Order, rather than TURN's interpretation of and modifications to the Virginia Arbitration. We find this approach a reasonable compromise to the other structure sharing proposals on the record.
With regard to intra-network structure sharing, we find that JC's assumption of a 55% sharing percentage between feeder and distribution networks is realistic on a forward-looking basis, and within the range of percentages adopted in other states and by the FCC. Verizon produced evidence that only 12% of cable segments share feeder and distribution, and aerial sharing between feeder and distribution is no higher than 41.2%. (Verizon Loop Rebuttal, 11/9/04, pp. 150-52.) We do not consider Verizon's historical practice appropriate to use for TELRIC modeling purposes. As TURN notes, the 55% sharing assumption in HM 5.3 is based on evidence that 75% of feeder routes follow distribution routes. (TURN/Loube, 8/6/04, p. 55, citing JC/Mercer, 11/3/03, Attachment RAM-5, p. 56.) We adopted this percentage in the SBC UNE case and we continue to find it reasonable to assume that on a forward-looking basis, an ILEC would make efforts to maximize sharing on networks that it controls. We will adopt this assumption for our runs of HM 5.3.
In comments on the draft decision, Verizon criticizes the use of FCC sharing inputs, stating that only inputs based on California data are appropriate. We disagree. We are not precluded from using the FCC's national data on inter-utility sharing of poles, trenches and conduit for guidance. Verizon itself has used or referred to nationwide data for other input assumptions, such as DLC and switching costs, and it is reasonable for the Commission to use its discretion and look to other states for recent analyses of forward-looking practices, depending on the input involved.
F. Plant Mix
"Plant mix" assumptions refer to the percentages of aerial, buried, and underground plant assumed in the loop network.
Verizon contends HM 5.3 assumes a plant mix that could never be achieved in California because it assumes away the constraints faced by providers operating in the real world. (Verizon, 8/6/04, pp. 67-68.) According to Verizon, JC rely on statewide ARMIS34 data, then allocate it across density zones based on the opinion of JA's witness Donovan. Further, JC rely on averages of data dating back eleven years rather than more recent data. This results in JC understating the amount of underground facilities that could reasonably be expected on a forward-looking basis given new local ordinances that mandate "out-of-sight" placement of new telecommunications outside plant construction. JC's assumptions are counter to recent trends toward greater use of underground facilities throughout California. (Id.)
In response, JC note that Verizon erroneously assumes the HM 5.3 plant mix is based on eleven year old ARMIS data, as was proposed in the SBC case. JC clarify that in this proceeding, HM 5.3 plant mix assumptions are based on percentages that Verizon supplied to JC in discovery, and Verizon is attacking its own current data. (JC, 11/9/04, p. 59; JC/Donovan, 11/3/03, para. 112.) Furthermore, JC defend their plant mix inputs as recognizing that dense areas will have a higher percentage of underground structure. (JC/Donovan, 11/9/04, paras. 44-47.)
In the SBC proceeding, we used plant mix assumptions provided by SBC, noting that we were uncomfortable relying on HM 5.3 inputs based on eleven year old ARMIS data. Here, JC have updated the HM 5.3 inputs with current information from Verizon. Therefore, we find it reasonable to use JC's proposed plant mix assumptions in our run of HM 5.3.
G. DLC Costs
Both models assume a forward-looking design that incorporates digital loop carrier (DLC) electronics into the loop plant. The parties dispute the installation costs for DLC systems that serve as inputs to the TELRIC model.
In the SBC proceeding, the Commission found it could not rely on the DLC installation costs provided by SBC, nor those suggested by AT&T and MCI. Instead, the Commission based HM 5.3 inputs on an average of actual cost information provided by SBC for 50 recent DLC installations, 25 for remote terminals (RTs) and 25 for controlled environmental vaults (CEVs). (D.04-09-063, mimeo. at 180.)
In this proceeding, Verizon contends the DLC costs in HM 5.3 are understated because more DLC systems are needed if the network employs the IDLC technology. Furthermore, the DLC inputs in HM 5.3 assume labor costs that are too low, and ignore costs of site acquisition, site preparation, and testing. (Verizon, 8/6/04, pp. 59-61.) For example, Verizon notes DLC labor costs adopted in D.04-09-063 are four times higher than those assumed in HM 5.3. (Id., p. 72.) As an alternative, Verizon develops an average DLC installation cost per dollar of material investment based on data from DLC installations it has performed over a two year period from its nationwide footprint. (Verizon, 11/9/04, p. 77.) It then confirmed this average cost based on a sample of 17 recent Verizon California DLC work orders. (Id.) Verizon proposes replacing the DLC and SAI costs in HM 5.3 with inputs based on these average costs. (Verizon/Tardiff-Murphy-Dippon, 11/9/04, Attachments TMD-8 and 9.)
JC deny Verizon's claim that an IDLC assumption requires the installation of more DLC systems. (JC, 11/9/04, p. 55.) In addition, JC respond that its DLC installation cost inputs are reasonable because they are based on a bottom-up approach that identified the engineering labor for each type and size of DLC configuration, similar to the approach used by the FCC for its Synthesis Model. (JC/Donovan-Pitkin-Turner, 8/6/04, p. 126.) JC contend Verizon's witness Richter exaggerates the complexity of DLC installation by customizing each installation at a higher cost and inflating the length of time for each installation project. For example, Verizon bases cost estimates on the total days it might take to get an engineering document processed through a department rather than the actual engineering review time. JC's witness Donovan contends that standardized DLC installations can be installed more quickly and efficiently, and the actual install time is much lower than the padded wait times Verizon has assumed. (JC/Donovan, 11/9/04, pp. 73-75.)
Moreover, JC maintain Verizon's DLC cost inputs based on its nationwide sample are flawed. Specifically, they charge Verizon has created "linear loading factors" that are multipliers applied to material costs. Hence, Verizon's approach inappropriately assumes installation costs increase proportionately with material cost increases. (JC/Donovan-Pitkin-Turner, 8/6/04, p. 121.) Further, JC contend Verizon's analysis inappropriately relies on embedded data and piece-meal replacement projects rather than new installations and larger-scale projects that are customarily associated with a TELRIC study. (Id., pp. 122-125.) With regard to Verizon's California sample of DLC costs, JC maintain Verizon developed its cost factor using the wrong type of DLC equipment. (Id., p. 129.)
We do not find it reasonable to rely on Verizon's proposed DLC installation costs. Verizon's costs rely on the assumption that installation cost is directly related to material cost, and we do not find this a reasonable assumption. Moreover, JC have pointed out numerous other pitfalls in Verizon's DLC cost factors, leading us to question whether these factors are based on the correct equipment and appropriate projects. We are also wary of Verizon's contention that the use of IDLC technology requires installation of more DLC systems. We did not find it necessary to increase the number of DLC systems modeled in the SBC case even though we assumed the use of IDLC technology there. Further, JC have presented reasonable arguments that unbundling over IDLC is unlikely to require the installation of additional DLC systems.
On the other hand, we will not rely on the DLC cost inputs proposed by JC for HM 5.3. In the SBC case, we found that HM 5.3 inputs assumed least cost installation scenarios that were below SBC actual costs for a sample of 50 recent installation projects. Given that similar or lower inputs are proposed in this case by the same witness,35 we find it more reasonable to use the actual costs developed in the SBC case as a proxy for forward-looking DLC installation costs. We find this a reasonable approach because SBC and Verizon are similarly-sized ILECs operating in the same state and, thus, facing similar labor and other local installation costs. Although Verizon provided national data on DLC installation costs, the flaws in its analysis enumerated by JC dissuade us from relying on Verizon's loading factors. Because SBC operates in the same California labor market as Verizon, we find SBC's installation costs present a reasonable proxy. Therefore, we will use the DLC installation cost inputs from D.04-09-063 in our run of HM 5.3.36
In comments on the draft decision, Verizon claims it is arbitrary and capricious for the Commission to use SBC's DLC costs rather than Verizon's own national data. (Verizon, 12/14, 05, pp. 8-10.) The decision was augmented to explain why we found Verizon's analysis using national data flawed and why it is reasonable to use data from a similarly situated California carrier.
A secondary issue related to DLC costs involves Verizon's claim HM 5.3 misallocates DLC costs to DS-1 service based on the relative space occupied by the DS-1 plug in unit, rather than the relative proportion of circuit capacity used. (Verizon, 8/6/04, p. 60.) As a result, Verizon contends HM 5.3 subsidizes DS-1 service by shifting costs away from DS-1 loops onto basic loops. (Id.) JC respond that Verizon proposes shifting DLC costs to DS-1 from basic loops using a voice grade equivalent (VGE) approach that assigns costs based on capacity used rather than space for the equipment. JC maintain this VGE approach was rejected by the Commission in the SBC UNE proceeding. (JC, 11/9/04, p. 56.) We agree with JC that the VGE approach to DLC cost allocation was rejected when setting UNE rates for SBC. (See D.02-05-042, mimeo. at 26-28.) For consistency, we will not use Verizon's proposed VGE method here either.
H. Labor Costs
A critical input in TELRIC modeling exercises involves the forward-looking cost of labor to install, operate and maintain the network. Labor costs are generally manifested in TELRIC models through hourly wage rates and assumptions regarding crew size and the time it takes to perform a given task. We now address the key criticisms of the labor cost assumptions in HM 5.3.
Verizon claims HM 5.3 understates labor costs by relying on the expert judgment of JC's witness Donovan without backup documentation or other explanation of the derivation or reasonableness of the proposed inputs. Verizon questions why Donovan used his opinion and quotes from outside vendors spread throughout the country when Verizon-specific data was available. (Verizon, 8/6/04, pp. 30-31.) Verizon alleges the resulting HM 5.3 labor inputs do not remotely resemble values that actual carriers operating real world networks can attain. For example, Verizon contends HM 5.3 undersizes the crews necessary to install network equipment and assumes they can achieve unrealistic productivity levels, particularly for cable placement and splicing. (Id., p. 70.) Moreover, Verizon asserts JC's witness Donovan misuses data from Verizon's Engineering Construction Records Information System (ECRIS) database, which is the system Verizon uses to estimate construction costs. Verizon contends Donovan inappropriately looks at crew sizes for discrete tasks rather than an entire placement project. (Verizon, Richter Surrebuttal, 1/28/05, p. 6.) In addition, Verizon complains HM 5.3 labor cost inputs are reduced from prior versions of HM, without sufficient justification for the assumption that labor costs in California have declined. (Id., p. 71.)
JC respond that Verizon unfairly attacks Donovan's use of expert judgment when Verizon's own modeling inputs rely to a significant degree on the undocumented opinions of "subject matter experts." (JC, 11/9/04, p. 60.) Further, JC claim that Verizon's witness Richter misinterprets and selectively uses ECRIS data to support his contentions that JC's HM 5.3 inputs are inadequate. (JC/Donovan, 11/9/04, pp. 64-65.) In contrast, JC's analysis shows that non-selective use of ECRIS data affirms HM 5.3 inputs. (JC, 11/9/04, pp. 60-61.)
In the SBC proceeding, the Commission chose to substitute SBC's fully loaded hourly wage rate, wherever possible, rather than rely on the opinion of AT&T and MCI's witness Donovan regarding wage rates. Thus, SBC hourly wages were used for cable installation, SAI investment, and terminal and splice investment. (D.04-09-063, mimeo. at 216.) The Commission also agreed with SBC that certain crew size assumptions in HM 5.3 were understated. Therefore, crew sizes for cable installation were increased, although crew sizes for splicing and NID installation were not modified. The Commission accepted Donovan's assumptions regarding time estimates for installation activities such as cable installation per day and splicing time. (Id.)
In this proceeding, we are relying on the same model we used in the SBC case, and JC have proposed labor inputs similar to, and in some cases lower than, those we modified in D.04-09-063. For example, JC proposed lower hourly engineering labor rates, and lower total labor costs for pole labor and DLC vaults. (JC/Donovan, 11/9/04, Attachment JCD-REB-14.) It is not reasonable to rely on JC's proposed HM 5.3 labor inputs when we found similar, and sometimes higher, inputs unsatisfactory in the SBC case. Specifically, since we increased crew sizes in the SBC case and used a higher hourly wage rate, we should do the same here. The reasonable approach would be to modify the hourly wage rate similar to our modifications in the SBC case, using a fully loaded hourly wage rate supplied by Verizon. Unfortunately, Verizon's model does not provide us with a fully loaded hourly wage rate. Instead, it uses factors for labor costs. Verizon, however, attempted to transform its model inputs and translate them into an alternative set of selected inputs for HM 5.3. (Verizon/Tardiff-Murphy-Dippon, 11/9/04, p. 26.)
Without accepting all of Verizon's proposed labor changes, we can use certain categories of Verizon's alternative inputs for our run of HM 5.3. In particular, we will use Verizon's proposed labor inputs for installed copper and splice, installed fiber and splice, installed poles and spacing, SAI investment, installed manhole, pullbox and spacing, and drop and NID. (Id., Attachment TMD-8.)37 We find it reasonable to use these inputs provided by Verizon, based on the labor costs it used in its own model, and run HM 5.3 with these changes. By using these suggested inputs, we are in effect modifying the labor rate and crew sizes in HM 5.3 similar to our modifications in the SBC case. Therefore, we will run HM 5.3 with the labor inputs suggested by Verizon in TMD-8 and 9 for the categories listed above.
Finally, we will not rely on references to Verizon's ECRIS data for crew size or other labor information. We find the selective presentations made by the witnesses on both sides do not provide a sufficient basis on which to make an informed decision.
In comments on the draft decision, Joint CLCs criticize the Commission's use of Verizon's proposed HM 5.3 labor inputs as inappropriate for several reasons. First, Joint CLCs claim the inputs increase both labor and material costs even though the decision only discusses the intent to make labor cost changes. Joint CLCs maintain that using these inputs "smuggles" much higher materials costs into the HM 5.3 model, often higher than material inputs used in the SBC proceeding, without any discussion or justification and without adequately describing how these inputs were developed. (Joint CLCs, 12/14/05, pp. 2-8.) Second, Joint CLCs claim Verizon presented these inputs as rebuttal to a TURN proposal, and never recommended their use to set UNE rates. (Id., p. 4.) Finally, Joint CLCs maintain that using these inputs, which were only introduced to highlight alleged flaws in TURN's inputs, imports the flaws of Verizon's model into HM 5.3. (Id., pp. 8-12.) TURN joins in criticizing the use of Verizon's labor inputs to HM 5.3 and suggests revisions to remove inflated material costs for poles, NIDs, drops, copper manholes and pullboxes. (TURN, 12/14/05, pp. 6-8.)
Verizon defends the Commission's decision to use inputs Verizon presented in rebuttal testimony, stating Verizon criticized material and labor inputs and it is appropriate for the Commission to revise both. (Verizon, 12/21/05, p. 2.) Verizon disagrees that it "smuggles" revised material costs into the inputs, stating its filings present separate material and labor inputs for various inputs, and its HM 5.3 inputs are based on the material and placement costs shown in the VzLoop model. Verizon contends its inputs are based on its actual contracts for installation labor costs which are documented on the record of the proceeding and are superior to generalized labor rates, crew sizes, and activity times assumed by JC's witnesses. (Verizon, 12/21/05, p. 3.)
We have made minor modifications to the labor inputs based on comments. We acknowledge the criticism of Joint CLCs that using Verizon's HM 5.3 inputs modifies material as well as labor costs. We were aware of this with regard to cable materials, but apparently the effect was more widespread than we initially understood. In comparing the various options, we find it preferable to use Verizon's inputs, even though they bring with them some material cost changes, because they are based on Verizon specific information from its contracts rather than the speculation of JC's witnesses. We find it preferable to use Verizon specific information for labor costs, even if the inputs come with some material cost increases, in lieu of speculative information offered by JC. This type of modeling adjustment is within our discretion and parties had adequate notice the Commission would adjust modeling inputs and assumptions. The fact that some inputs were pulled from rebuttal testimony does not limit our ability to use them, just as we use JC's DS-3 and interoffice modeling inputs from rebuttal filings in Section VI.K below. TURN suggests revisions to remove allegedly inflated material costs from Verizon's rebuttal inputs. We have made the changes TURN suggests to poles, NIDs, drops, copper manholes and pullboxes.
I. Cross-over Point and Maximum Copper Loop Length
The cross-over point refers to the feeder route length at which fiber feeder facilities become less costly than copper feeder. In the SBC UNE case, we modeled a cross-over point from copper to fiber at 12,000 feet. (D.04-09-063, mimeo. at 218.) In other words, we ran HM 5.3 assuming copper feeder loop segments longer than 12,000 feet convert, or "cross-over," to fiber after 12,000 feet. We find no reason to deviate from this modeling approach and we will employ 12,000 feet as the cross-over point in our model run of HM 5.3 in this proceeding.
The parties also dispute the maximum copper loop lengths in HM 5.3. According to Verizon, copper loops in excess of 12,000 feet are not consistently capable of supporting many services such as DSL, and longer loops introduce inefficiencies into the provisioning process. (Verizon, 8/6/04, p. 41.) Verizon claims that an 18,000-foot loop, as modeled in HM 5.3, cannot provision all the UNEs at issue in this proceeding and would present compatibility problems by not adhering to industry equipment standards. (Id., p. 42.) JC maintain that a limitation of 12,000 feet is unnecessary for the UNEs modeled in this proceeding and inefficiently increases loop costs. (JC, 8/6/04, p. 60.)
In the SBC UNE proceeding, we resolved this same dispute by finding that FCC rules require a TELRIC model to design a network that assumes the provision of other ILEC services. (See 47 C.F.R. Section 51.505(b).) Therefore, we ran HM 5.3 for SBC with a maximum copper loop length of 12,000 feet. For the same reasons, our run of HM 5.3 will use the same modeling assumption of a 12,000 foot maximum copper loop length.
J. Switching Inputs
In modeling forward-looking costs for the unbundled switching UNE, HM 5.3 uses as an input the cost of switching investment on a price per line basis. The price per line depends on the vendors from whom switches are purchased and whether the lines relate to new switch installations or growth to existing switches.
JC's witness Pitts develops a price per line based on information provided by Verizon California regarding its actual switch purchases for new lines and growth hardware. Pitts also includes a "switch installation multiplier" that accounts for additional investment associated with the main distribution frame, power, ILEC engineering and installation costs, and sales tax. (JC/Pitts, 11/3/03, pp. 4-5.)
Pitts claims her analysis is conservatively high because it is based on switch purchase data from the 1996 and 1998 time frame that most likely overstates costs compared to current switch prices. Further, the information relates solely to digital circuit switching, thereby failing to reflect the economies of forward-looking packet switching technology. (Id., p. 16.) Pitts explains that she analyzed the percent of remote switches and host or standalone switches in Verizon's California network to calculate a melded new host/standalone-remote price per line for new switch lines. She then takes this result and melds it with price estimates for growth lines to arrive at a melded new/growth price per line. (Id., p. 11.) Pitts assumes 92.6% of lines purchases are new, while 7.4% of line purchases are for growth. (Id., p. 13 and Attachment CEP-6.) Pitts contrasts her assumptions with what she considers Verizon's unreasonable assumption that 100% of switch lines are purchased at the growth price. (JC/Pitts, 8/6/04, p. 2.)
Not surprisingly, Verizon counters that Pitts's assumption that 92.6% of switches will be purchased at the discounted "new" switch price is unrealistic. According to Verizon, it would not be able to purchase virtually its entire switching network at the new switch discount because vendors only offer the new switch discount for a small portion of lines, expecting carriers to purchase a larger percentage of growth additions at relatively higher prices. (Verizon, 8/6/04, p. 73.) Verizon suggests it is more realistic to assume 64% of switching equipment is purchased at growth prices, and 36% is purchased at new prices based on Verizon's actual switch purchases from 1997 through 2002. (Id., p. 74.)
Similar to JC, TURN disagrees with the high percentage of growth lines assumed in Verizon's switch cost studies. TURN proposes the Commission rely on the analysis of the FCC's Wireline Bureau in its Virginia Arbitration order, where it assumed new line installations at 88% and growth additions at 12%. (TURN, 8/6/04, p. 46.)
In the SBC case, we found that HM 5.3 had assumed too high a percentage of lines could be purchased at the new switch discount. In both D.99-11-050 and D.04-09-063, the Commission frowned on the assumption that switch vendors would sell over 90% of the lines needed for a forward-looking network at the discounted "new" switch price that is offered to large incumbent carriers such as SBC and Verizon for only a small percentage of purchases. (D.09-4-09-063, mimeo. at 223.) For SBC's UNE switching prices, we found it reasonable to rely on a mix of new and growth lines that included a higher percentage of growth line purchases to reflect that in a forward-looking environment, a carrier would not be able to purchase all of its switches at the new line discount and would incur upgrade and growth costs. (Id.) Thus, we ran HM 5.3 in the SBC case using a weighting of new and growth lines based on SBC's actual purchases over a recent five year period. (Id.)
For the same reasons discussed in the SBC UNE case, we will not adopt Pitts' assumption that over 90% of switch lines can be purchased at the new switch discount. Instead, we will use the mix of new and growth purchases that Verizon has shown in the last five years, which is the same measure we used for SBC. We prefer to rely on actual switch purchase information rather than the assumptions of either JC or TURN that the majority of lines will be purchased at a highly discounted price. Thus, based on Verizon's information, we will apply a mix of 36% new and 64% growth purchases. This mix of new and growth purchases will then be applied to the prices per line that Pitts has developed in her testimony. Although we will not use Pitts' recommended mix of new and growth purchases, we find Pitts' price per line methodology logical and reasonable to rely on, in contrast to the price per line developed from the Verizon model.
Our rationale for not using the Verizon switching models is discussed at length in Section V.A.5 above, but a key reason we will not use Verizon's price per line is that it is dominated by GTD-5 switch costs, and we have found the GTD-5 is not a forward-looking switch. Specifically, Verizon's switching cost studies assumes a mix of 63% GTD-5 switches, 22% Lucent supplied switches, and 15% Nortel switches. (Verizon Switching Rebuttal, 11/9/04, p. 23.) In contrast, Pitts uses only a mix of Lucent and Nortel switches to match the switch purchase data provided by Verizon. We find the assumed mix of Lucent and Nortel switches more reasonable.
Finally, with regard to switch feature pricing, JC's witness Pitts contends the switch prices she used in her modeling include feature-specific hardware costs. Thus, feature costs are included as part of the switching port cost. (JC/Pitts, 11/3/03, p. 8.) We find this explanation satisfactory and for this reason, we will not set individual feature prices.
In comments on the draft decision, TURN contends the FCC rejected Verizon's switch purchase assumptions in favor of the assumption of 88% new switch purchases and 12% growth. (TURN, 12/14/05, p. 9.) TURN also notes that prices for circuit switching has dropped precipitously with the introduction of packet switching. Verizon responds that TELRIC circuit switching costs should not depend on the cost of a new technology that is not required to be unbundled. (Verizon, 12/21/05, p. 10.)
We will not alter our determination to base the new/growth mix on data from Verizon's last five years of purchases. Although TURN alleges it is unjustified to assume a higher mix of growth purchases than new switches, we find it unsupported to assume switch vendors would sell 88% of their switches at the new switch price that JC propose. Instead, we use the percentage mix from Verizon's recent purchase history coupled with prices developed by JC. This is intended to approximate the composite price that switch vendors would offer in a forward-looking environment based on the prices they offered in the past. Therefore, our approach is different from the Verizon proposal the FCC rejected.
Verizon comments that the draft decision errs by relying on the "switch installation multiplier" in HM 5.3. Verizon claims the multiplier is completely undocumented and understates costs. Further, Verizon contends the Commission cannot rely on inputs from other states without sufficient evidence that Verizon has similar costs. We disagree with Verizon's claims. We find JC's witness Pitts provided adequate documentation of how she derived the multiplier, using data from the FCC's Synthesis Model. (JC/Pitts, 11/3/03, pp. 14-15.) Further, we are not limited to using only Verizon California inputs, particularly when witnesses demonstrate that Verizon's inputs reflect costs that are not forward-looking. (JC/Pitts, 8/6/04, pp. 59-63.) Verizon's stance is puzzling given its assertion the Commission should use nationwide DLC installation cost data rather than information from another California carrier. We have used Verizon's nationwide data for new and growth switch purchases, thus we see find it reasonable to use national data for the switch installation multiplier.
JC propose a flat-rated port pricing structure as more representative of the way Verizon incurs switch costs. (Id., 3.) According to JC's witness Pitts, the current generation of end office digital switches has little or no equipment that can exhaust based on usage and the vast majority of switch costs do not vary with respect to minutes or usage. (Id., p. 19.) While earlier generations of analog and some digital switches did have switch processors that were limited and could exhaust their call processing capacity, the current generation of digital switches have processing capacity far exceeding the volumes that lines or trunks could generate. The current Lucent 5ESS can handle 2.5 million call completions per hour, and Verizon's data shows statewide average processor utilization is far below this level. Further, forecasted subscriber calling behavior on landline switches is stable or declining. Thus is it not expected that the current generation of digital switches will exhaust. (Id., p. 19-20.)
Verizon opposes the idea of a flat port rate, claiming Pitts ignores the fact that switches are engineered up front to avoid exhaust situations. (Verizon, 8/6/04, p. 76.) Verizon contends that switches are traffic limited and the design efforts to avoid exhaust should not be construed as evidence that costs are not usage based. (Id.) Moreover, Verizon argues that a flat switching rate violates the principle of cost causation by subsidizing competitors who target high usage business customers, allowing them to avoid usage charges, while competitors who supply low volume carriers will pay a higher flat rate than might otherwise be necessary. (Verizon, 11/9/04, p. 88.)
TURN agrees with Verizon in opposing a flat port charge rather than minute of usage charges. In TURN's view, network engineering has been driven by the needs of high volume users and it is reasonable to impose usage charges to recover the costs of providing excess capacity from those who most benefit from it. (TURN, 8/6/04, p. 48.)
In the SBC proceeding, AT&T and MCI made the same proposal for a flat monthly price per port to cover switching costs formerly collected in minute of use rate elements. In D.04-09-063, the Commission found that since switch costs incurred by SBC were set based on a flat price per line based on a 10-year usage forecast, and since it was unlikely SBC would exceed that usage forecast, it was reasonable to set switch rates on a flat per port basis. (D.04-09-063, mimeo. at 239-241.) In the SBC UNE case, we also retained a usage-based rate that interconnecting carriers could rely on, if needed, where interconnection contracts specified compensation for using another carrier's network, otherwise known as "reciprocal compensation." (Id., p. 242.)
For the reasons articulated in D.04-09-063, and given JC's data showing Verizon's statewide average processor utilization levels and the low probability of switch exhaustion, we will adopt a flat-rated port pricing structure for Verizon as we did for SBC. As in the SBC case, we will run HM 5.3 to calculate usage rates for reciprocal compensation purposes. Those rates are shown in Appendix B.
K. High Capacity Loop and Transport Inputs
One particularly thorny area of cost modeling involves the inputs and assumptions relating to high capacity loops and interoffice transport. Verizon questions the JC's expert opinions on inputs for these portions of the HM 5.3 model. Specifically, Verizon maintains that the transport and high capacity loop modeling in HM 5.3 is premised on faulty engineering assumptions, unrealistic network designs, and inappropriate demand assumptions. (Verizon/Murphy, 8/6/04, p. 98.) For example, Verizon contends HM 5.3 overlooks the total demand associated with high capacity loops and the total volume of interconnection trunks. According to Verizon, total demand is essential to the proper sizing and design of high capacity loop and transport systems. (Id., pp. 97-100.) In addition, Verizon criticizes HM 5.3 for omitting certain optical equipment. (Id.) Finally, Verizon's own analysis indicates the interoffice ring architecture in HM 5.3 is insensitive to both demand and costs for fiber cable and electronics. (Verizon/Tardiff, 8/6/04, pp. 89-90.) Thus, Verizon questions whether the model truly optimizes the interoffice ring architecture when demand and material inputs are changed.
JC respond that Verizon misunderstands the approach used in HM 5.3 to derive interoffice demand and that HM 5.3 does include the proper optical interface equipment in the switches that are modeled. (JC/Mercer-Pitkin-Turner, 11/9/04, pp. 169-171.) In response to the charge HM's ring architecture modeling is flawed, JC maintain Verizon's analysis is based on the faulty assumption that the architecture will change when demand changes. JC contend the predominant costs in the ring architecture are not subject to variation based on demand. (JC, 11/9/04, p. 70.)
Many of the identical criticisms were made in the SBC UNE proceeding with regard to the modeling of transport and high capacity services. There, we found flaws with HM 5.3 interoffice transport and DS-3 loop rates and were unwilling to rely on them solely to set SBC's UNE rates. (D.04-09-063, mimeo. at 100.) Indeed, we found that for DS-3 related UNEs, HM 5.3 inexplicably yielded cost results higher than those requested by SBC. Therefore, we adopted SBC's proposed rates for DS-3 related UNEs. (Id., p. 245.)
In this proceeding, our early efforts to run HM 5.3 also indicated inexplicably high cost results for DS-3 loops and interoffice transport, more than double the DS-3 loop rates adopted for SBC in 2004. Specifically, our early run produced a DS-3 loop rate of $1352, compared to SBC's rate of $573. When we compared these results to the rates proposed by Verizon, we found Verizon had proposed even higher rates.
Although we used SBC's DS-3 loop rates as a fallback in the SBC UNE proceeding, we will not take a similar approach here. We will not rely on Verizon's transport and high capacity loop rates because JC's criticisms in this area convince us that would be unwise. First, JC contend that Verizon's methodology for transport and high capacity loop modeling does not reconstruct an efficient, forward-looking interoffice network. According to JC, Verizon relies on an over-simplified "capacity costing" approach that ignores the use of the most efficient technology and does not reflect the demand associated with a properly sized interoffice network. (JC, 8/6/04, p. 61.) Second, JC allege Verizon's modeling results in proposed increases for transport and high capacity loop rates of as much as 451% over current rates. JC maintain any cost increases for these services are ludicrous given industry trends toward lower costs. (Id., p. 62.) We agree that rate increases for transport and high capacity loops, as proposed by Verizon, are not in keeping with industry trends.
Given that we reject Verizon's modeling, and that HM 5.3 produces results that we consider unreasonable, we must consider modifying inputs in HM 5.3 related to high capacity loops and transport. Conveniently, the rebuttal version of HM 5.3, which we decline to use for our model run, contains updated inputs in this area. We can extract those updated inputs, examine them, and if reasonable, insert them into the version of HM 5.3 that we are using to set UNE rates.
JC provided several updated inputs in the rebuttal version of HM 5.3 in response to Verizon's extensive criticism of HM 5.3's high capacity and interoffice modeling. MCI contends these updated inputs are derived from data provided by Verizon to JC after the initial November 2003 cost filings in this proceeding. In MCI's motion requesting hearings, it describes four recommended input modifications to HM 5.3 based on data provided by Verizon. (MCI motion, 5/5/05, pp. 7-8.) MCI admits JC "inadvertently failed to describe" these changes in the rebuttal filing of HM 5.3. (Id., p. 6.) While we do not accept any of the modeling revisions in the rebuttal version of HM 5.3, we can isolate these four updated equipment inputs and consider placing them into the earlier version of HM 5.3. MCI enumerates four input categories, but we will only consider changing the first two input categories MCI describes which relate to revised equipment costs. The two input categories we consider are (1) revised prices for SONET multiplexer and Digital Cross-Connect (DCS) equipment based on current Verizon supply contracts, and (2) a revised input fraction for interoffice circuits requiring inter-ring connections based on Verizon's data for interoffice equipment costs.38 Verizon responds that these input changes were not adequately described in JC's rebuttal testimony and should be stricken. (Verizon response to MCI motion, 5/24/05, p. 3.)
While we are troubled that JC did not provide a more detailed description of these input changes in rebuttal testimony, we find that the changes were made in response to Verizon's criticisms of HM 5.3 high capacity and interoffice transport modeling. Further, the input changes are based on updated equipment price information provided by Verizon to JC, and which JC did not receive in time to use in its initial cost filings. In using these revised inputs, we do not accept any of the modeling or algorithm revisions in the rebuttal version of HM 5.3. Rather, we merely substitute updated equipment cost inputs, provided by Verizon, into the initial HM 5.3 model. We find the revised inputs reasonable and filed in response to Verizon's criticisms. We adopt these modified inputs for our HM 5.3 model run. When these modified inputs are inserted into our adopted version of HM 5.3, the resulting DS-3 loop rate is $ 592.73. This result compares favorably to the DS-3 loop rate of $573 adopted for SBC. While we are not able to adjust all aspects of the HM 5.3 interoffice transport modeling and address all of the concerns cited by Verizon, we will use the HM 5.3 results for interoffice transport and high capacity loops because it is appropriate to use one model throughout for consistency in inputs and assumptions.
In comments on the draft decision, Verizon contends use of these updated inputs violates due process because the changes were submitted on rebuttal and cherry-picking input changes from the rebuttal version is arbitrary and capricious. Verizon contends it was denied discovery and hearings on these input changes. (Verizon, 12/14/05, pps. 10-11.) Furthermore, Verizon alleges that in using these inputs, the Commission inappropriately excluded installation costs relating to these inputs. (Id., pps. 12-13.)
XO/Cbeyond respond that the input changes in dispute are based on Verizon's own contracts as reflected in Verizon's own cost study. The JC rebuttal filing containing these inputs provides precise references to Verizon's own cost study as the source of the input changes. (XO/Cbeyond, 12/21/05, p. 7.) Moreover, XO/Cbeyond claim the inputs include installation costs as shown in JC's rebuttal testimony. (Id., p. 8 citing JC/Donovan, 11/9/04, Exh. JCD-REB-14, "Input Tables for HIP" tab, cells D455:F466.)
We disagree with Verizon's claim of a due process violation. The inputs at issue were provided in rebuttal based on Verizon's own cost data and in response to Verizon's criticism of HM 5.3. While description of these input changes was not provided, cites were provided indicating Verizon as the information source. Again, we emphasize that we are not relying on any new modeling algorithms or formulas in accepting these inputs. We merely substitute these cost inputs for those in the initial HM 5.3 filing because they represent the newest cost information for Verizon based on actual Verizon data. Verizon does not dispute that this is its actual cost information, although it claims installation costs were removed. After review, we find that installation costs are included in these inputs and we can use them.
While Verizon has not been granted hearings or additional discovery to further litigate these inputs, it has been given the opportunity to comment on their use. Indeed, Verizon defends our use of changes it presented on rebuttal to labor rates. (Verizon, 12/21/05, p. 5.) We see little difference between the updated labor inputs and updated DS-3 cost inputs, both provided on rebuttal, and continue to believe that use of either is appropriate in this proceeding. Moreover, the Commission frequently exercises discretion in choosing modeling inputs as long as they are reasonable and supported by the record.39 Verizon should not be surprised the Commission is now choosing from competing modeling inputs since the Commission's modeling criteria specifically required that "cost studies must allow parties to modify inputs and assumptions." (ALJ's Ruling Revising Schedule, 7/23/02, p. 4.)
Although we will rely on HM 5.3 interoffice transport modeling, we will use the rate design suggested by Verizon for these UNE rate elements. JC proposed flat-rates rather than per-mile charges for all interoffice transport UNEs. According to JC's witness Mercer, since interoffice circuits are now implemented on rings rather than as point-to-point circuits, there is only an indirect relationship between the air distance and the route miles required to provide the circuit. (JC/Amended Declaration of Mercer, 2/6/04, para. 60.) Verizon modeled these UNEs with a fixed plus a per-mile rate component.
In the SBC UNE proceeding, we adopted rates that were a blend of fixed and per-mile charges. Here, we are using the same model that we used to set SBC's UNE rates. We are not persuaded to suddenly shift to a flat rate design for interoffice transport UNEs, particularly when the ring architecture assumed in both proceedings is largely the same. Verizon's proposal is similar to the rate design we adopted in the SBC proceeding, and we prefer to keep a somewhat consistent rate design for these rate elements.
L. Miscellaneous Adjustments to HM 5.3
In addition to the input modifications already discussed, we made other minor adjustments to HM 5.3 based on our familiarity with the model from the SBC UNE proceeding. These minor modifications were made to mirror changes we made to HM 5.3 in setting SBC UNE rates, and described in D.04-09-063. The changes are the following:
· Pole Spacing-we modified HM 5.3 to assume pole spacing of 150 feet for all density zones of the distribution network.
(D.04-09-063, at 125.)
· Drop Terminal-we modified HM 5.3 to assume 85% buried drop terminals and 15% aerial. (Id.)
· Switch Fill-we adjusted the switch port administrative fill to 82%. (Id.)
In comments on the draft decision, Joint CLCs claim the draft erred by ignoring the pole spacing testimony of JC's witness Donovan. (See JC/Donovan 11/9/04, pp. 45-50.) Verizon responds that its witness Richter showed the spacing suggested by Donovan does not meet safety guidelines. (Verizon/Richter, 8/6/04, p. 38-39.) Our decision to use a pole spacing assumption of 150 feet does not mean that we find Verizon's witness more credible than JC's witness. Both witnesses present credible arguments that pole spacing depends on the weight of the cable and the height at which cable is attached to the pole. Both witnesses have the relevant qualifications to express opinions on this topic. Rather than dissect each distribution area for cable weight and pole height requirements, we prefer to use a blanket assumption of 150 foot pole spacing throughout the model as a conservative measure. We realize this may inflate pole costs to some extent, but it avoids analyzing each modeling segment to ensure safety guidelines are not violated.
Regarding Drop Terminal, TURN comments that Commission staff erred in its model run of HM 5.3 by altering the buried drop sharing fraction rather than the buried drop structure fraction. (TURN, 12/14/05, p. 6.) We have corrected this error in the final decision.
M. Shared and Common Cost Markup
TELRIC based UNE prices are designed to recover both the costs directly attributable to UNEs and a "reasonable measure" of forward-looking overhead costs. (FCC First Report and Order, para. 336.) Thus, a critical component of final UNE rates is an adder to recover overhead costs. This overhead component has come to be known as the "shared and common cost markup," or simply "markup." It is generally a percentage added to TELRIC costs to recover costs attributable to a group of UNEs but not specific to any one UNE, as well as costs that are common to all outputs offered by the firm. (See D.95-12-016, Appendix C.)
Verizon proposes a markup of 14.5 %, which is comprised of a 9.08% common overhead loading, a 1.68% marketing loading, and a 3.25% other marketing support loading. (Verizon/Jones, 11/9/04, p. 72.) These loading percentages are derived from various categories of expenses in Verizon's general ledger. (Verizon Recurring Costs Testimony, 11/3/03, p. 148.) According to Verizon, 12.5% of total company costs were attributed to support and common costs and form the basis of the loadings it proposes. (Verizon/Jones, 11/9/04, p. 69, n. 109.)
JC claim Verizon's proposed markup exceeds direct costs to an unreasonable extent. Specifically, JC witnesses Brand and Menko claim Verizon's proposed overhead loadings are an unreasonably high percentage of the total expenses allocated to wholesale services. (JC/Brand-Menko, 8/6/04, p. 42.) JC provide corrections to Verizon's proposed cost studies revising the shared and common cost markup to 9.12%, only a slight increase from Verizon's proposed common cost markup of 9.07%. (JC, 8/6/04, p. 76.) Verizon rebuts JC's analysis by claiming Brand and Menko base their calculations on the wrong set of numbers and mischaracterize how Verizon develops its overhead cost loadings. (Verizon/Jones, 11/9/04, p. 68.)
TURN raises concern with the process Verizon uses to develop its proposed markup. Specifically, TURN comments that Verizon does not reduce its general support costs to take retail services into account. (TURN, 8/6/04, p. 41.) TURN also expresses concern with Verizon's method of forecasting overhead expenses based on current costs. TURN alleges that Verizon's method ensures Verizon will recover its current expenses no matter what is varied in the rest of the model. This fixed recovery does not allow corporate overhead to fluctuate as the network increases or decreases. (Id., p. 43.)
For HM 5.3, JC propose a markup of 8.93%. (JC/Brand-Menko, 11/3/03, para. 89.) They note this is comparable to the 8% markup adopted by the FCC's Wireline Competition Bureau in the Virginia Arbitration. (Virginia Arbitration, para. 143.) Moreover, they explain that this markup factor is not comparable to markup factors adopted in prior Commission UNE pricing proceedings because the costs to which HM 5.3 applies the markup already include a portion of costs that prior UNE cost studies recovered through the markup. In other words, HM 5.3 assigns more costs directly to UNEs, leaving fewer costs "left-over" to be considered overhead. (JC/Brand-Menko, 11/3/03, p. 42.)
JC forecast overhead costs based on the relationship between corporate operations expenses and total operating revenues less corporate operations expenses. They derive their markup using data specific to Verizon's California operations. (Id., pp. 42-43.) JC's 8.93% markup excludes retail, non-recurring costs and other non-UNE costs. (Id.) JC also removed what it identified as extraordinary one time charges that were primarily merger related and not expected to recur in the future for an efficiently operating firm. (Id., p. 48.)
Verizon asserts an 8.93% markup is grossly understated and calculates overhead expenses of about one-quarter Verizon's actual overhead expenses in 2003. (Verizon, 8/6/04, p. 81.)
We find JC have provided a more rational and coherent explanation of how they developed their overhead common cost markup. While Verizon describes the various cost categories it includes in its loadings, it then provides the generic statement that "the expenses are adjusted to make them forward-looking before they are used in the calculation of each loading." (Verizon Recurring Costs Testimony, 11/3/03, p. 148.) Verizon fails to provide an adequate explanation of this forward-looking "adjustment." Later, in the final round of comments, Verizon explains that 12.5% of its total company costs were used to develop the loadings, leading to a total markup of 14.5%. This leads us to wonder why Verizon's forward-looking overhead expenses would be higher than today's overhead costs. The answer might lie in the forward-looking adjustments that Verizon fails to adequately describe. Furthermore, if today's 12.5% of costs that cannot be attributed directly to UNEs was used as a proxy starting point for calculating a markup, the final markup should be lower than 12.5% once retail, non-recurring, and other non-UNE common costs are removed. As TURN notes, Verizon does not provide assurance that its common and support expenses are adjusted to remove retail service costs. Overall, we are not satisfied with Verizon's explanation of how it calculated its three loading factors that together form the 14.5% markup it proposes.
In contrast, JC's witnesses Brand and Menko give a thorough explanation of the data they used to calculate their 8.93% markup, using data specific to Verizon California. They also provide reasonable explanation and support for the adjustments they make to their data, primarily to reflect unique one-time merger expenses. Interestingly, the 8.93% markup proposed by JC is remarkably close to Verizon's proposed common cost loading of 9.08%, and similar to Brand and Menko's 9.12% restatement of Verizon's calculations. The chief difference between the proposals of Verizon and JC is that Verizon proposes its common cost loading of 9.08%, then it layers on two separate markup factors for "Marketing and "Other Marketing Support," for a total markup of 14.5%. Verizon fails to provide assurance that retail-related marketing and marketing support have been removed from its loadings. Therefore, we reject Verizon's proposals and instead rely on the analysis of Brand and Menko to adopt a common cost markup of 8.93%.
22 Verizon initially proposed a cost of capital of 15.95%, which it adjusted to 14.37% in its reply comments on 8/6/04.
23 JC initially proposed 7.12%. The proposal was updated to 7.64% on 11/9/04. (JC/Murray, 11/9/04, p. 42.)
24 XO, 11/9/04, p. 29.
25 ORA's proposal is based on Murray's methodology, but excludes short-term debt. (ORA/Litkouhi, 8/6/04, p. 12.) XO calculates that ORA's methodology would result in a 7.4% cost of capital. (XO, 11/9/04, p. 28.)
26 (TURN/Loube, 8/6/04, Exhibit RL-8, Table 7.)
27 A basis point equals one one-hundredth of a percent.
28 The CAPM formula is:
Cost of equity = Risk free rate + (Market risk premium) x (Beta)
29 The cost of equity for SBC was calculated as follows: (7.4% market risk premium x .93 beta) + risk free rate of 4.9% = 11.78%. (See D.04-09-063, mimeo. at 159.)
30 Although Verizon criticizes Murray's CAPM analysis for not using a higher beta coefficient for telecommunications carriers, we find a beta of 1.0 in line with FCC guidance from the Virginia Arbitration and similar to the beta we used for SBC. Further, Verizon's criticism lacks credibility given it did not use comparisons with telecommunications firms for its own DCF analysis.
31 The Joint CLCs are Cbeyond Communications LLC, Covad Communications, DMR Communications Inc., MPower Communications Corp, Navigator and XO, as explained in more detail in Section XI, Comments on the Draft Decision.
32 The term "achieved fill" represents the spare capacity "achieved" after the model is run, as opposed to the "input fill," or sizing factors, which are model inputs. These inputs size the network for spare and growth and lead to an output, or "achieved fill."
33 See Joint Comparison Exhibit, 9/2/05, admitted into the record by ALJ's Ruling on Additional Exhibit and Submission of Case, November 8, 2005.
34 ARMIS refers to the FCC's "Automated Reporting Management Information System" that was initiated in 1987 for collecting financial and operational data from the largest carriers.
35 See JC/Donovan, 11/9/04, Attachment JCD-REB-14.
36 Remote terminal costs are $22,814 per site and controlled environmental vault installation costs are $49,569 per site. (D.04-09-063, mimeo. at 180.)
37 In using Verizon's labor inputs, we had to adjust cable prices in HM 5.3 to avoid double counting. HM 5.3's labor inputs were zeroed out and its cable material cost inputs were replaced with Verizon's combined cable material and labor inputs. Joint CLCs comment that using Verizon's cable inputs inappropriately changes all cable to 24-gauge size. (Joint CLCs, 12/14/05, p. 6.) Verizon responds that its cable gauge assumptions avoid maintenance problems and do not limit deployment of advanced services. (Verizon 12/21/05, p. 9.) We find Verizon's cable gauge assumptions reasonable and we will use them.
38 MCI motion, 5/5/05 p. 7-8, and HM 5.3 Rebuttal version, 11/9/04, Switching IO Module, Tandem and STP investment worksheet, cell P30.
39 See D.04-09-063, mimeo at 249, which describes how "...the initial Proposed Decision exercised a great deal of judgment in reviewing the models' flaws, correcting them where possible, and selecting numerous modeling inputs."
