AN ANALYSIS OF THE
COST FACTOR AND
PRICE THRESHOLD
4.1 The cost factor in a proxy model should not be based upon historical accounting data
What the model does
The BCM computes a monthly per-line cost by multiplying the total investment per line by a factor which is intended to reflect operating expenses (including depreciation) and an after-tax return on investment. The BCM provides two sets of results that reflect two very different cost factors:
(1) A factor of 31.6765% reflecting historical accounting data and total expense levels of Tier 1 LECs based upon 1994 ARMIS Form 43-01; and
(2) A factor of 22.97% reflecting a forward-looking estimate of expenses and overheads using the MCI/Hatfield methodology.[1]
The selection of a cost factor clearly has a material impact on the aggregate estimate of the costs of providing universal service:
* For the national results (without ETI corrections), the model yields an average monthly cost of $23.04 if the embedded cost factor is used and yields an average monthly cost of $16.71 if the forward-looking cost factor is used.[2] Also, assuming a price threshold of $30, if the embedded cost factor is used, the BCM computes a national USF requirement of approximately $4.9-billion, whereas when the forward-looking cost factor is applied, the BCM computes a national USF requirement of approximately $2.2-billion. The model documentation fails to provide details of the calculation of the two different cost factors.[3]
Given the importance of this variable, the Joint Sponsors should be encouraged to provide detailed documentation and justification of their respective computations.
What the model should do
The cost factor that is based strictly upon ARMIS reports of embedded expense levels -- without any apparent modifications -- should be rejected for several reasons, which are discussed below and in Appendix 4A. We have taken preliminary steps to develop an alternative to the ARMIS-based cost factor by examining the data in Table 2.9 in the FCC's Statistics of Common Carriers[4] for all reporting local exchange carriers.[5] These data include total cost figures (i.e., they include "nonregulated" items and are thus slightly in excess of the amount subject to separations). In Appendix 4A, we revised figures in certain expense accounts to more accurately reflect expenses associated with the provision of universal service, however, even these figures should be considered as upper bounds in determining which expenses are appropriately associated with the provision of basic local exchange service.[6] We also examined the six-volume Cost of Service Study (COSS) submitted by NYNEX to the Massachusetts Department of Public Utilities for the 12 months ending November 1992 for more detailed descriptions of the accounts and for data regarding the distribution by NYNEX of the expenses among subcategories of individual accounts, and between residence and business classes. For example, an analysis of the Massachusetts COSS narrative and data for Account 6611 (Product Management) reveals that 84% of the expenses in this account support market management and planning for business customers and, indeed, only 5.3% percent of the expenses in the account support residence services.
Table 4.1 below, summarizes some of the relevant accounting data from the Statistics of Common Carriers.
Table 4.1 Selected Data for Tier I LECs (1994) (000s Omitted) Account Amount TPIS $267,443,392 Total Plant $272,474,927 Total Depreciation and Amortization $115,703,078 Net Plant $156,771,851 Total Operating Revenues $ 92,927,905 Total Operating Expenses Depreciation $ 70,263,301 Expenses $18,655,947 Net Operating Revenues $ 22,664,599 Source: SOCC, 1994/1995 Edition, Table 2.9, Column 3
Historical depreciation expenses reflect estimated lives that are not indicative of the lives of plant necessary to offer basic telephone service
Depreciation expenses account for approximately 27% of the LECs' total operating expenses and thus should be examined most critically. The digital switches, distribution, and feeder that are of all the LEC expenses, necessary for basic telephone service need not be replaced for at least 20 years and thus depreciation expenses should be less than the historical depreciation expenses reflected in the ARMIS reports. Depreciation expenses during the last decade reflect a time period when local exchange carriers accelerated their depreciation of analog switches for diverse reasons, many of which are unrelated to the ongoing provision of the primary residential access line. Although local exchange carriers have sought to portray modernization plans as "business-as-usual," in fact the plans have typically caused the premature retirement of telecommunications plant that is/was in all other respects adequate and efficient to satisfy the needs of primary residential access line subscribers. The acceleration in the replacement of existing plant results in retirements in advance of the originally anticipated mortality curve. This, in turn, impacts depreciation expenses in several ways, including the creation of reserve deficiencies and a rationale for increases in the depreciation rates for the affected plant categories. Moreover, the acquisition of new plant creates additional depreciations charges; to the extent that those acquisitions/replacements were not economically justified on their own merit (in terms of incremental revenues and avoided costs), upward pressure is placed upon embedded cost levels overall.
For example, in 1993, Southern New England Telephone Company (SNET), in a revenue requirement investigation, sought to increase depreciation rates for its analog switch account from 7.8% to 7.9% and for its digital switch account from 5.0% to 7.2%.[7] Under the embedded cost approach, such increases in depreciation charges on embedded plant (whether of a current or older generation) are being attributed to universal service, although there is no specific linkage between the plant replacement decision and the plant actually required to supply primary residential access line service. The use of embedded cost factors is fundamentally at odds with the TSLRIC concept, and has the effect of transforming what is facially offered as an "incremental cost" study into an historical embedded cost analysis. Once efficient technology is assumed for the provision of basic telephone service, there is simply no reason to incorporate into the model an expectation that such equipment will be prematurely retired unless such an expectation was itself incorporated into the capital budgeting decision upon which the LEC relied in justifying the plant replacement in the first place.[8] Lives of at least 20 years should be assumed.[9]
The projected average service lives included in SNET's 1993 represcription was 14 years for the digital switching account;[10] however, this average life reflects the replacement of early vintage digital switches.[11] Longer lives should be assumed, as Hatfield/MCI have done, thereby producing a lower forward-looking cost factor than the embedded cost factor advocated by the LECs. For example, the depreciation expenses reported for all Tier I LECs for the year ending December 31, 1994 totalled approximately $18.7-billion.[12] Total reported expenses for this time period were approximately $70.3-billion,[13] and thus depreciation expenses accounted for more than one quarter of all operating costs. These depreciation expenses should be scaled down to reflect longer estimated lives.
Table 4.1 shows that depreciation expenses comprised 7.0% of total plant in service (TPIS) for all reporting LECs. As a rough rule of thumb, altering this percentage to 5.0% (i.e., approximately 20 years), would reduce depreciation expenses to $13.4-billion.
Return on investment
The cost factor of 31.6765% most likely reflects the LECs' authorized return of 11.25%. The return on investment should instead be computed using 10.3125% to reflect (1) a jurisdictional weighting of 25% of the FCC's authorized return of 11.25%[14] and a 75% jurisdictional weighting of recent state PUC decisions, which have resulting from comprehensive investigations of LEC capital structures, cost of debt, and cost of equity.[15]
4.2 The price that policy makers choose to support will significantly influence the USF requirement
The model allows a user to compute the USF requirement for three different monthly price levels: $20, $30, and $40. The price being supported is one of the major factors influencing the size of a USF.[16] The establishment of a price threshold, i.e., a monthly rate above which USF support would be provided, is a key policy consideration. The national average rate for basic local exchange service should be viewed as setting a lower bound of affordability. The national average residential rate for unlimited local exchange service and the subscriber line charge is approximately $16.76.[17] However, the range of monthly local exchange service rates and the size of the local calling area are diverse, with some paying $23.56 for this same service.[18] It would thus be misleading to consider simply the average rate as the measure of affordability. Furthermore, subscribership rates are also critical sources of relevant information about affordability. When federal and state policy makers address the establishment of affordability levels, they should examine penetration rates as well as monthly local exchange prices and prices for other relevant services (e.g., touchtone). Finally, as is well recognized, household income is a major factor in subscribership,[19] and therefore universal service support should be targeted to those most in need.
Furthermore, the national average rate represents a composite of pre- and post-rate rebalancing of local and toll rates.[20] The scope of "local service" varies widely from state to state and, for that matter, as between urban and rural areas within the same state. For customers with small local calling areas, toll calls to nearby points are equivalent to extended area service and metropolitan exchange calling areas insofar as meeting the customer's local community of interest communications needs. Customers in nonmetropolitan areas generally spend far more on toll calls to nearby localities (intraLATA toll) than do customers in large cities. If some national average "affordability" level is to be developed, it must focus not on tariff definitions of "local" service as these may exists from place to place, but upon a defined basket of services that includes both access and usage within a defined region. A "national average" rate for local exchange access service necessarily understates the actual amounts being paid by customers to meet their basic telecommunications needs.
[2]Id., at II-2.
[3]Id., at IV-28.
[4]This table is based upon the Automated Reporting Management Information System (ARMIS) Reports (FCC Report 43-02). Table 2.12 is based upon ARMIS Reports 43-01, however, Report 43.01 lacks the detail provided in Report 43-02.
[5]See Appendix 4B for the list of reporting (i.e., Tier I) local exchange carriers.
[6]ETI's examination is intended to highlight some key areas that merit scrutiny, but does not represent a complete examination of all accounts. The purpose of the ETI analysis is to expose some illustrative fundamental flaws with the ARMIS-based approach. The analysis below represents an upper bound for an estimate of the expenses because, in some instances, it reflects all types of residence activities, not just basic local exchange service.
[7] Connecticut DPUC Docket No. 92-09-19, Application of the Southern New England Telephone Company to Amend its Rates and Rate Structure, SNET Response to OCC-595 ("SNET Response to OCC-595"). The prescribed depreciation rates for SNET's digital switch account in 1987 and 1990 were 4.9 and 5.0, respectively. Id..
[8]A critical parameter in a discounted cash floor (DCF) type of capital investment analysis is the revenue-producing life of the asset under examination. All other things being equal, the more years that the asset is expected to remain in revenue-producing service, the higher will be its net present value (NPV). If the life of an asset is shortened after the decision to acquire it has been made, the asset may well become non-performing (in the financial sense) on the basis of its newly reduced life expectancy. That is, had the revised life been used in the original DCF analysis, the NPV may well have turned out to be negative, indicating that the investment should not be pursued. The financial consequences of this type of revisionism of previously-made management decisions properly belong to the LEC's management and shareholders, and not to its customers. Of course, by utilizing embedded cost factors, the economic loss is imposed entirely upon customers and, in the instant case, upon universal service in particular. It is also worth observing that (in the context of incentive regulation), were the initial life expectancy to prove unduly pessimistic, shareholders, not customers, would enjoy the financial gains arising from the longer actual service life that would then ensue.
[9]See, for example, SNET Depreciation Study for 1993, Account 2212. A life span of 19 years is estimated for six of the ten DMS 100 and DMS 100/200 switches and a life span of 15 years is estimated for the other four switches.
[10]SNET Response to OCC-623, op.cit., footnote 97.
[11]For example, SNET planned to replace 12 early vintage digital switches in the years spanning 1993 through 1996. SNET 1993 Depreciation Rate Study, December 29, 1992, Account 2212, at 15-17.
[12]SOCC, Table 2.9, line 250, column 3.
[13]Id.
[14]In the Matter of Represcribing the Authorized Rate of Return for Interstate Services of Local Exchange Carriers, CC Docket No. 89-624, Order, 5 FCC Rcd 7507 (1990).
[15]The ETI figure relies on the 10% return authorized by the California Public Utilities Commission. California PUC, Consolidated A.92-05-002, A.92-05-004, I.87-11-033 and A.87-05-049, I.85-03-078 and A.85-01-034, Applications of GTE California Incorporated (U 1002 C) and Pacific Bell (U 1001 C) for Review of the Operations of the Incentive-based Regulatory Framework Adopted in Decision 89-10-031, and Related Matters, Decision 94-06-011, June 8, 1994, at 52. Also, the Vermont Public Service Board authorized an effective rate of return of 8.5%. Vermont PSB, Dockets No. 5700/5702, Investigation of Proposed Vermont Price Regulation Plan and Proposed Interim Incentive Regulation Plan of New England Telephone and Telegraph Company and Petition of Department of Public Service for an Investigation of New England Telephone and Telegraph Company Rates, Order, October 5, 1994, at 82. The Massachusetts Department of Public Utilities authorized an effective rate of return of 9.63%. Massachusetts D.P.U. 94-50, Petition of New England Telephone Company d/b/a NYNEX for an Alternative Regulatory Plan for the Company's Massachusetts Intrastate Telecommunications Services, May 12, 1995, at 507.
[16]Although the price thresholds are not user inputs, per se, one can effectively alter the price thresholds that are modelled simply by changing a single formula and copying that new formula to the other 5,000-plus rows in the model. It appears that it would be fairly straightforward for the Joint Sponsors to modify the BCM to allow this essential variable to be altered by changing a number in a single cell. Assuming that it is feasible, such a modification would be a useful improvement to the BCM.
[17]FCC Monitoring Report, 1995, Table 5.7 (1993 Data).
[18]New York Telephone Company, P.S.C. No. 901 Telephone Tariff, Section D., 33rd Revised Page 29; The NYNEX Telephone Companies, Tariff F.C.C. No. 1, Effective September 1, 1995, Section 31.4.3, 19th Revised Page 31-2.
[19]Common Carrier Bureau, Federal Communications Commission, "Preparation for Addressing Universal Service Issues: A Review of Current Interstate Support Mechanisms", February 23, 1996, at 15-16.
[20]For example, as a result of a directive by the Massachusetts Department of Public Utilities in 1990, NYNEX gradually increased local exchange rates and decreased intraLATA toll charges in a series of revenue-neutral filings. Massachusetts D.P.U. 89-300, NET, June 29, 1990; Massachusetts D.P.U. 91-30, NET, September 9, 1991; Massachusetts D.P.U. 92-100, NYNEX, October 28, 1992; Massachusetts D.P.U. 93-125, NYNEX, January 13, 1994. In 1994, the DPU determined that: "There has been no statistically significant change in the Massachusetts telephone service penetration rates in the years 1989 to 1992. ... Thus we find that through 1992 the transition to cost-based rates has not negatively impacted universal service, and the current proposed increase is unlikely to have an adverse impact on universal service." Massachusetts D.P.U. 93-125, NYNEX, January 13, 1994, at 58 (footnote omitted). This rate rebalancing experience in Massachusetts as well as similar experiences in other state jurisdictions can be instructive as policy makers define "affordable."
Distribution of Switched Access Lines by Local Exchange Carrier (1993) Company Total Lines Served* Share Lines Served Seven RBOCs 112,215,811 75.7% Cincinnati Bell 838,006 .6% SNET 1,937,185 1.3% Citizen Communications 79,134 .1% Commonwealth Telephone 207,995 .1% GTE 15,289,500 10.3% Lincoln Telephone 251,610 .2% Puerto Rico Telephone 1,159,565 .8% Rochester Telephone 492,512 .3% Sprint Corporation 5,504,431 3.7% Total for Non-RBOC Tier 1 LECs 25,759,938 17.4% Tier II LECs** 10,214,671 6.9% Total # of Local Loops*** 148,190,420 100% *SOCC, 1993/1994 Edition, Table 2.10. **Note: Share of loops served by Tier II LECs is equal to: Total # of Local Loops-# of Loops served RBOCs-# of Loops served by Tier I LECs ***FCC Monitoring Report, May 1995, Table 3.6 Notes: Lines as reported in the SOCC include residential, business, public and mobile access lines. Loops are approximately comparable to lines. Numbers do not add to 100% due to rounding.