Federal-State Joint Board on Universal Service

AGENCY:

Federal Communications Commission.

ACTION:

Final rule.

SUMMARY:

In this document, the Wireline Competition Bureau (Bureau) updates line counts and other input data used in the Commission's forward-looking economic cost model for purposes of calculating and targeting non-rural high-cost support beginning January 1, 2004. The Bureau denies a petition filed by the Maine Public Utilities Commission and the Vermont Public Service Board (Joint Commenters) seeking reconsideration of the Bureau's 2002 Line Counts Update Order.

ADDRESSES:

The full text of this document is available for public inspection during regular business hours in the FCC Reference Center, Room CY-A257, 445 12th Street, SW., Washington, DC 20554.

FOR FURTHER INFORMATION CONTACT:

Thomas Buckley, Attorney, Telecommunications Access Policy Division, Wireline Competition Bureau, (202) 418-7400, TTY (202) 418-0484.

SUPPLEMENTARY INFORMATION:

This is a summary of the Bureau's Order and Order on Reconsideration in CC Docket No. 96-45, DA 03-4070 released December 24, 2003.

I. Introduction

1. The Bureau, consistent with action taken in the past, updates line counts and other input data used in the Commission's forward-looking economic cost model for purposes of calculating and targeting non-rural high-cost support beginning January 1, 2004. In the Order on Reconsideration, the Bureau denies a petition filed by the Maine Public Utilities Commission and the Vermont Public Service Board (Joint Commenters) seeking reconsideration of the Bureau's 2002 Line Counts Update Order, 67 FR 3118, January 23, 2002.

II. Discussion

A. Switched Line Count Updates

2. Consistent with the framework adopted in the Twentieth Reconsideration Order, 65 FR 26513, May 8, 2000, and the 2001 and 2002 Line Counts Update Orders, 65 FR 81759, December 27, 2000 and 67 FR 3118, January 23, 2002, we conclude that the cost model should use year-end 2002 line counts filed July 31, 2003, as input values for purposes of estimating average forward-looking costs and determining support for non-rural carriers beginning January 1, 2004. We will adjust support amounts every quarter to reflect the lines reported by non-rural carriers. In addition, we will allocate switched lines to the classes of service used in the model by dividing year-end 2002 lines into business lines, residential lines, payphone lines, and single-line business lines for each wire center in the same proportion as the lines filed pursuant to the 1999 Data Request.

3. We disagree with BellSouth that line counts should not be updated unless the Bureau also updates road and customer location data. Updated line count data are readily available, whereas updated road and customer location data are not. As we have explained in the past, line count data must be updated to reflect cost changes and economies of scale associated with changes in line counts, consistent with the Commission's forward-looking cost criteria established in the First Report and Order, 67 FR 41862, June 20, 1997. Line count data also should be updated to avoid increasing the lag between such data and the quarterly line count data used to adjust non-rural high-cost support amounts. We are not persuaded that updating line counts is inappropriate because it may fail to reflect certain costs associated with serving new customer locations. The model's use of road surrogate data to determine customer locations ensures that the structure costs associated with serving new customer locations are reflected in model cost estimates unless such locations are along new roads. BellSouth contends that recent switched line decreases and new housing growth in its service territory undermine the assumption that most new lines are either placed at existing customer locations or along existing cable routes, but it submits no data in support of this contention. Switched lines nationwide decreased by 3.3 percent in 2002, and Commission data indicate that households increased by approximately one percent. Based on these data, we cannot conclude that the trends identified by BellSouth justify not updating line count data. On balance, we find that updating line count data is the best approach for estimating forward-looking costs and determining non-rural high-cost support amounts for 2004.

4. We also disagree with AT&T's argument that we should use projected lines for the end of the 2004 funding year, rather than the most recent reported year-end lines (end of 2002), to match the line count data used to estimate forward-looking costs with the quarterly line count data used to adjust non-rural high-cost support amounts. AT&T has not proposed a methodology for projecting lines. Verizon argues that any such methodology would be complex, difficult, and overly burdensome for purposes of estimating forward-looking costs. We also note that, as stated above, switched lines have declined recently, suggesting the difficulty of accurately projecting lines based on historical data. Consistent with the 2001 and 2002 Line Counts Orders, we find that year-end 2002 line counts are the appropriate data to use for updating the cost model's input values at this time.

B. Special Access Line Count Updates

5. Consistent with the 2002 and 2001 Line Counts Update Orders, we will use year 2002 ARMIS special access line count data as model inputs to estimate forward-looking costs and determine non-rural high-cost support amounts in 2004. On balance, we conclude that this approach is consistent with the Commission's criteria for estimating forward-looking costs and with applicable universal service principles. We also will continue to divide the updated special access lines among wire centers in the same proportion as the special lines from the 1999 Data Request. As discussed below, we conclude that this methodology is a reasonable approach for estimating special access line growth to determine non-rural high-cost support amounts in 2004.

6. Based on our examination of the record, we continue to find that it is appropriate to update special access lines for purposes of determining non-rural high-cost support in 2004. The First Report and Order requires that the model reflect the economies of scale of serving all business and residential lines, including special access lines. Consistent with this criterion, the Bureau always has included special access line count data within its cost estimates. Removing special access line count data from the model's cost calculations would ignore the demand for special access services. We find that removing special access lines would be inconsistent with the Commission's criteria requiring that the model reflect the economies of scale of serving all business and residential lines, including special access lines.

7. We also conclude that updating special access line count data for purposes of determining non-rural high-cost support in 2004 is consistent with the principle set forth in section 254(b)(5) of the Act that the universal service support mechanism should be specific and predictable. Because different states have different percentages of special access lines, removing them has differential effects on costs and, therefore, support among states. We decline to adopt an interim approach to estimating costs that would significantly change support in some states outside the context of a Commission proceeding to address the underlying model design issues raised by commenters. We conclude that it would be more appropriate to maintain continuity of support until these issues can be addressed comprehensively in a future Commission proceeding.

8. The current record is insufficient to permit us to reach a conclusion as to what adjustments may be needed, if any, to the model's process for counting high-capacity special access lines. Although some commenters argue that the model understates costs by counting high-capacity lines as voice-grade equivalents, it may overstate costs by deploying high-capacity lines on copper instead of fiber. Some commenters also argue that the model overstates costs because it does not include inputs for non-switched services such as digital subscriber lines. In other words, to the extent that adjustments to the model may be needed, such adjustments may increase some costs and reduce others. Consequently, we believe that the most prudent approach is to wait for further action by the Commission to consider several model improvements, specifically including the process for estimating special access demand. In the meantime, we conclude that updating special access line count data for purposes of determining non-rural high-cost support in 2004 is consistent with the Commission's forward-looking cost criteria and with applicable universal service principles.

9. We reject BellSouth's contention that special access line count data should be removed from the model's cost calculations for purposes of determining non-rural high-cost support based on the Bureau's decision to remove special access demand to set unbundled network element (UNE) prices in the Virginia arbitration proceeding. Different rules and principles apply in this proceeding that warrant a different approach. In that proceeding, the Bureau was faced with two proposals for accounting for special access lines and their associated costs in setting Verizon Virginia, Inc.'s UNE rates. Under total element long range incremental cost (TELRIC) principles, the Bureau had to choose the methodology which would result in UNE rates within a range of reasonableness. Here, in contrast, we must determine how to treat special access lines for purposes of calculating non-rural high-cost support. Whereas the Bureau's decision in the Virginia arbitration proceeding affected UNE rates in one state, non-rural high-cost support is determined based on the relationship between each state's average cost per line and the nationwide average. Because different states have different percentages of special access lines, removing special access lines from the model's cost calculations may significantly change support in some states. Our decision here is guided in part by the section 254(b)(5) principle that universal service support should be specific and predictable. Under the circumstances, we conclude that a different approach is warranted for the purpose of determining non-rural high-cost support.

10. We also reject Verizon's request that we publish model cost estimates with and without special access demand at the study-area level before deciding this issue. Verizon argues that it cannot determine whether zeroing out special access lines would produce reasonable results because the Commission has not provided adequate data to allow interested parties to “run the latest version of the model to remove special access demand.” Contrary to Verizon's claim, the Commission provides all the necessary tools and data to run the model without special access lines. Specifically, both the model and ARMIS special access line data are made available to the public on the Commission's Web site. Further, switched line count data are available to the public under a protective order.

11. We also will continue to divide the updated special access lines among wire centers in the same proportion as the special access lines from the 1999 Data Request. We conclude that allocating year 2002 ARMIS special access lines based on the 1999 Data Request remains a reasonable approach for estimating special access line growth for purposes of calculating and targeting non-rural high-cost support for 2004. In this regard, we have analyzed the Verizon data submitted by the Joint Commenters. Based on our analysis, we are not persuaded that the Bureau's allocation methodology is unreliable or produces biased results.

12. The Joint Commenters submitted an analysis comparing model cost estimates based on (1) Verizon data reflecting the number of high-capacity special access lines in each Maine and Vermont wire center served by Verizon at the end of 2001 and (2) year 2000 ARMIS data allocated to wire centers using the Bureau's methodology. They contend that their analysis demonstrates that the Bureau's allocation methodology produces “significant errors” (defined as line count data requiring a correction of 25 percent or more) for 78 percent of the wire centers. They further contend that this methodology overestimates special access lines within 83 percent of wire centers with less than 3,000 switched lines, and underestimates special access lines in 67 percent of wire centers with more than 10,000 switched access lines. As a result, they claim that the data used by the Bureau to allocate special access lines are “unreliable for both urban and rural areas.” The Joint Commenters also calculated an “average cost correction” for wire centers in five size groups (based on switched access lines). They contend that the correction factors vary according to wire center size, and that their application to 2002 support amounts increases support by $0.49 per line for Maine and by $0.50 per line for Vermont. They argue that the Bureau should use special access line count data used to estimate costs for the 2000 funding year, or provide non-rural carriers with the greater of the amount calculated with updated data or the amount provided in 2000.

13. As an initial matter, we disagree with the premise of the Joint Commenters' analysis that the goal of the allocation methodology is to achieve an exact correspondence between the lines assigned to a given wire center in the model and the actual number of lines served. Rather, the goal is to achieve reasonable results that are consistent with the Commission's forward-looking cost criteria using the best available data. For example, the 1999 Data Request required carriers to report intrastate “private lines” with special access lines, pursuant to the criterion that the model estimate the cost of serving all businesses and households, including the cost of special access and private lines. The Commission has never used the number of private lines as model inputs, however, because nationwide private line data had not been available until this year. The Bureau's methodology assigns updated ARMIS special access lines to a wire center based on the proportion of special access and private lines reported for that wire center in the 1999 Data Request. Thus, we would expect differences between the number of lines the allocation methodology assigns to a given wire center in the model and the number of special access lines a carrier serves in that wire center.

14. In addition, because it compares model lines and Verizon lines from two different time periods, the analysis is not the “apples-to-apples” comparison that the Joint Commenters set out to achieve. The Joint Commenters compared model lines based on year 2000 ARMIS special access line count data with year 2001 special access lines obtained from Verizon. Furthermore, the analysis focuses on the number of special access lines assigned to wire centers, rather than the percentages of lines in a study area that are assigned to wire centers. Even if the Joint Commenters had compared model and Verizon data from the same year, as explained above, we would not expect the number of special access lines assigned to a wire center to be the same. The Bureau's methodology assigns special access lines to wire centers using fractions calculated based on the 1999 Data Request. Thus, a more appropriate comparison for evaluating the Bureau's methodology would be to compare the percentage of special access lines in a study area that are assigned to a wire center using the Bureau's methodology with the percentage of total special access lines in the study area that are identified in the Verizon data as serving that wire center.

15. After analyzing the two data sets on which the Joint Commenters base their analysis, we cannot conclude that the Bureau's allocation methodology produces unreliable or biased results. We first analyzed the data sets for differences between the percentages of total special access lines assigned to individual wire centers, using the Joint Commenters' wire center size categories. We found that for the 45 wire centers with less than 3,000 lines, the Bureau's methodology assigns a higher percentage of lines than Verizon's special access lines in most cases (consistent with the Joint Commenters' contention), but the average difference between the model percentages and the Verizon percentages is very small—only −0.1 percent. For the 24 wire centers with over 10,000 switched lines, we found that the Bureau's methodology assigns a lower percentage of lines than the Verizon data in only 33 percent of the wire centers. Contrary to the Joint Commenters' findings, the Bureau's methodology assigns a higher percentage of lines than the Verizon data in most wire centers from this group. We also analyzed the correlation between wire center size and percentage differences between model lines and Verizon lines. Although we found an overall correlation of +0.541, this correlation is caused mainly by two outlier data points. Thus, although our analysis reveals differences between model lines and Verizon's special access lines that are on average negative in small wire centers and positive in large wire centers, the differences are very small—less than 1 percent—and do not reveal a pattern that supports the Joint Commenters” allegation of substantial systematic bias.

16. Furthermore, our analysis of the Joint Commenters' cost results does not show a consistent pattern in the data that would support their allegation of bias. Again, for purposes of our analysis, we used the Joint Commenters' wire center size categories. As stated above, they contend that the differences in model cost estimates based on Verizon lines and model lines correlate to wire center size: higher-density (urban) wire centers have lower costs and lower-density (rural) wire centers have higher costs based on Verizon lines. Although this is true, on average, most of the wire centers within their groups do not conform to this pattern. For small wire centers with 0 to 1,000 lines, the Joint Commenters found that the average difference was +$0.11. Twenty-eight of the 34 wire centers in this group have lower costs using Verizon data, however. For wire centers with 1,000 to 2,500 lines, the Joint Commenters found that the average difference was +$0.23, but 57 out of the 77 wire centers in this group have lower costs using Verizon data. Thus, the majority of small, rural wire centers show differences that are counter to the Joint Commenters' allegation of bias.

17. We also analyzed the cost results when the Verizon data are adjusted to match the vintage of the other line count data used in the Joint Commenters' analysis. As discussed above, they compared two vintages of special access lines: year 2000 ARMIS line count data and 2001 line count data obtained from Verizon. To obtain cost results, they used these data in combination with year-end 2000 switched line counts. The Bureau runs the model using switched and special access lines from the same year, however, which is important for purposes of analyzing cost results because it allows one to distinguish between effects due to changes in the overall number of lines and changes due to the allocation of lines. Accordingly, Bureau staff factored down the Verizon year 2001 special access data to reflect the total year 2000 ARMIS special access line data, and combined this data with year-end 2000 switched line count data to obtain adjusted cost results. Comparing these adjusted results to results based on model lines, we again found that although the average differences were consistent with the Joint Commenters' findings, most wire centers showed differences counter to the allegation of bias. As shown in Attachment B, the overall result of our analysis of the relationship between wire center size and differences in cost results based on adjusted Verizon lines and model lines was a slight statistical correlation of −0.085 percent. Given the slight correlation between costs and size in the two states and the various directions of cost corrections for wire centers within each group, we cannot conclude that the Joint Commenters' cost correction factors are reliable. In sum, therefore, we conclude that allocating year 2002 ARMIS special access lines based on the 1999 Data Request remains a reasonable approach for estimating special access line growth for purposes of calculating and targeting non-rural high-cost support for 2004, and that the Joint Commenters' analysis does not establish that this methodology is unreliable or produces biased results.

18. Finally, the Joint Commenters do not establish an alternative methodology that would provide fairer or more reasonable results. Even if their cost correction factors were reliable for Maine and Vermont, there is no reason to believe the same factors would be reliable nationwide. The differences in costs based on special access lines and costs based on model lines are likely to differ significantly by state given the diversity of terrain, population density, and size. Because support is determined in relationship to the nationwide average cost, we would have concerns about applying cost correction factors derived from two states to the nation as a whole. Moreover, if state-specific cost correction factors were used, it is not clear that the states of Maine and Vermont would see a “substantial” increase in support. Depending upon the “corrected” costs in other states, their support could also decrease.

19. In the absence of new data, the Joint Commenters urge the Commission to revert to the special access line counts used to distribute support in 2000, that is, year 1998 ARMIS special access lines. Using these line counts would provide demonstrably less reliable results than the current methodology for two reasons. Prior to ARMIS reporting year 2000, some carriers were under-reporting their special access lines by reporting special access circuits terminating at multiple customer premises as a single special access line, rather than as multiple special access lines. As part of its ongoing effort to improve data consistency, the Bureau subsequently clarified how special access lines should be reported in a consistent fashion. As a result, Verizon's special access lines increased substantially between year 1999 ARMIS reports and year 2000 ARMIS reports. Second, the method used to allocate special access lines to wire centers in the model's first year of operation was not as reliable as our current method. Because we had not yet developed a methodology to use the 1999 Data Request to allocate lines to wire centers, we used the only data available at the time to allocate lines: the wire center line counts developed by PNR Associates, trued-up to year 1998 ARMIS line counts. The allocations in the 1999 Data Request are more reliable because the data were filed by the carriers, rather than being estimated by PNR's National Access Line Model.

C. Other Issues

20. Consistent with the 2002 Line Counts Update Order, we will update the model with year 2002 ARMIS data used to compute general support facilities (GSF) investment so that the model's cost estimates take into account the current costs of GSF investment associated with supported services. In addition, we will update the model with the most recent traffic parameters available from the National Exchange Carrier Association (NECA) to determine the percentage of the switch allocated to supported services and the switch port requirement for interoffice transport. We also will use the methodology employed in the 2001 and 2002 Line Counts Orders to match wire centers reported by non-rural carriers in their quarterly line count data used to adjust non-rural high-cost support amounts with the wire centers found in the 1999 Data Request and in the model's customer location data. Commenters generally support these input updates.

21. Some commenters express concerns regarding reporting of unbundled network element (UNE) lines that are sold or leased to competitive LECs for purposes of calculating and targeting non-rural high-cost support amounts. In particular, AT&T urges that leased lines and UNE lines must be reported to ensure that the model's cost estimates reflect the demand for total lines. The Maine and Vermont Commissions state that some non-rural carriers do not include UNE lines in their ARMIS reports, a practice which could reduce support amounts by exaggerating per-line costs in urban areas with substantial UNE-based competition relative to per-line costs in other areas. We clarify that the model uses lines reported to NECA pursuant to section 36.611 to estimate switched line demand, and that NECA requires that carriers report both leased lines and UNE lines that are sold to competitive LECs for purposes of § 36.611 reporting.

22. AT&T urges the Commission to initiate a proceeding to consider improvements to the model's platform and inputs, arguing that the model has “well-known deficiencies” and that recent developments confirm the inaccuracy of certain model platform and input assumptions. Such a proceeding is beyond the scope of the Bureau's delegated authority. The Commission has expressed its intention to initiate a proceeding to study proposed changes to the model inputs and model platform in a comprehensive manner.

III. Petition for Reconsideration of the 2002 Line Counts Update Order

A. Discussion

23. We do not address Petitioners' arguments that the model input data used by the Bureau pursuant to the 2002 Line Counts Update Order was unreliable, because these arguments are fully addressed in the foregoing Order. As demonstrated in the foregoing Order, there is no merit to Petitioners' contention that the Bureau's methodology for allocating updated special access lines in the model is unreliable or produces biased results. As also explained above, and contrary to Petitioners' assertion, it is appropriate to use data sources from different years in the model when these are the best available data to achieve reasonable results that are consistent with the Commission's forward-looking cost criteria and with applicable universal service principles. Below, we conclude that Petitioners' contention that the Bureau failed to provide adequate notice of its decision to update data in the 2002 Line Counts Update Order is without merit.

24. Petitioners argue that the Bureau's 2002 Line Counts Public Notice, 66 FR 48259, September 19, 2001, seeking comment on updating line counts for 2002 did not provide adequate notice that “routine updating of line counts would substantially reduce the support available for Verizon customers in their states.” We disagree. The Bureau clearly stated in the 2002 Line Counts Update Public Notice that it was considering updating line count data in the model using the same methodology as the Bureau used in the 2001 Line Counts Update Order. In particular, for purposes of determining support for the year 2002, the Bureau sought comment on updating the switched line counts in the model with year-end 2000 wire center line count data, updating special access line counts with year 2000 ARMIS data, and using the Bureau's 1999 Data Request to allocate the updated lines. In the 2002 Line Counts Update Order, the Bureau then applied these methodologies to estimate switched line and special access line count growth. Therefore, the Bureau provided adequate notice in the 2002 Line Counts Public Notice of the method it used to update model inputs in the 2002 Line Counts Update Order.

25. As the Bureau informed the public that it was considering the same framework for 2002 updates as it had in the past, we also disagree with Petitioners that they lacked adequate notice of the potential impact of input updates on 2002 support distributions. Consistent with the Commission's criterion that “[t]he cost study or model and all underlying data, formulae, computations, and software associated with the model must be available to all interested parties for review and comment,” the model was posted on the Commission's website, and the input data used by the Bureau was available to the public either on the website or under a protective order or licensing agreement. Petitioners were therefore capable of determining the support distributions for 2002 based on the model's cost calculations before the 2002 Line Counts Update Order was adopted. If Petitioners believed the support distributions were inappropriate, they had the burden of identifying why specific inputs should not have been updated, but Petitioners did not meet this burden. We therefore find that Petitioners had adequate notice of the potential impact on non-rural high-cost support amounts of the model input updates proposed in the 2002 Line Counts Public Notice.

26. Petitioners further argue that the 2002 Line Counts Public Notice failed to notify parties that the Bureau would count special access lines as voice grade equivalent channels in the model's inputs, special access lines would increase in various non-rural wire centers, and updated line counts would be matched with older data for purposes of assigning such lines to wire centers. We reject these claims for the following reasons. First, in the 2002 Line Counts Update Public Notice, the Bureau stated it was considering updating special access lines as it had done in the past, which was to count special access lines as voice grade equivalent channels. In the comment cycle in that proceeding, Verizon requested that the Bureau count special access lines as facilities for purposes of calculating support for 2002. The Bureau, however, noted in the 2002 Line Counts Update Order that such an alteration would require a platform change outside the scope of the proceeding, and deferred consideration of this issue until a future proceeding on possible improvements to the model platform and inputs. Similarly, because Petitioners were notified that special access lines would be updated using the same methodology as in the past, Petitioners could access year 2000 ARMIS special access filings for the non-rural carriers in their states on the Commission's website to find out whether special access lines increased or decreased for 2002 cost estimates. Consequently, we reject Petitioner's argument that the 2002 Line Counts Update Public Notice failed to apprise interested parties of the methodology used to update special access lines in the 2002 Line Count Updates Order. We find that the 2002 Line Counts Public Notice was clear in seeking comment on whether to update the model's inputs consistent with past practice.

27. Petitioners also argue that the Bureau did not make available line count data at the time of release of the 2002 Line Counts Update Public Notice due to proprietary treatment of these data. This claim is incorrect. In the First Report and Order, the Commission established, as one of the criteria in developing a forward-looking economic cost model to determine universal service support, that “all underlying data, formulae, computations, and software associated with the model should be available to all interested parties for review and comment.” Consistent with this principle, the Commission has determined that line count data used for wire centers that receive high-cost support should be publicly available. In addition, line count data for wire centers that do not receive high-cost support are available pursuant to the Bureau's Interim Protective Order, April 7, 2000. Year-end 2000 line count data used to estimate high-cost support for 2002 was filed by non-rural carriers by July 31, 2001, and therefore was available to Petitioners at the time of the release of the 2002 Line Counts Public Notice on September 11, 2001.

IV. Ordering Clauses

28. Pursuant to the authority contained in sections 1-4, 201-205, 214, 218-220, 254, 303(r), 403, and 410 of the Communications Act of 1934, as amended, and § 1.108 of the Commission's rules, this order is adopted.

28a. Pursuant the authority contained in sections 4, 201-205, 218-220, 303(r), and 405 of the Communications Act of 1934, as amended, and 405 of the Communications Act of 1934, as amended, and §§ 1.106 and 1.429 of the Commission's rules, that the petition for reconsideration filed February 25, 2002, by the Maine Public Utilities Commission and Vermont Public Service Board is denied.