DETERMINACIONES ANALÍTICAS

When demand is uncertain and, in any event, significantly lags network con- struction, the decision to invest in capital-intensive construction projects, like fixed-wireline telecommunication networks (whether for initial investment, upgrade, or enhancement), is an extremely difficult one. That is especially true of telecommunications relative to other capital-intensive industries (e.g., auto- mobile manufacturing and petrochemical processing) because many telecom- munication services are public in the sense that demand directly depends on others also being hooked up to the network. Therefore, until the network itself is widely deployed, demand cannot take off.

Demand uncertainties notwithstanding, there are some advantages to pub- lic communication network investments compared to other capital-intensive industries. Multiservice communication network firms feature production

processes characterized by very high joint and common costs. Telephone net- work operators are somewhat unique in that even though capital assets are fixed and immobile (“sunk”), they are also widely dispersed geographically in the network. In many situations, local subscriber distribution plant comprises half, or more, of the total network investment. That provides unique oppor- tunities for flexible technology-adoption strategies. In other capital-intensive industries, such as auto, steel, and energy production, adoption of a new tech- nological paradigm involves major lump-sum investments for large centralized production facilities. Contrast that with telecommunications, for which the spatial distribution of network nodes and transmission facilities is relatively decentralized, allowing for selective upgrading and modernizing of facilities without major production interruptions. However, the slower the rate of net- work technology adoption in response to such fundamental changes as substi- tuting analog copper and coaxial cable with digital fiber optics, the more it will raise the cost of interworking the two technologies during the transition phase. Nevertheless, it is useful from a capital-budgeting perspective to have the flexibility.

It is much easier and less expensive to alter, postpone, or cancel a tele- communication network construction project in response to market conditions than it is to do so for, say, a nuclear power plant. The primary reason is that portions of the network are useful for providing advanced communication serv- ices to customers in a certain geographic area without having to make those services available everywhere. It is not necessary to have the entire network converted to be able to provide new services. New telecommunications services often are phased in gradually, even though, from a demand perspective, it would be preferable to offer new communication services to everyone at the same time.

The risk associated with future demand is lower for telcos than for other capital-intensive firms because of much shorter lead times and greater flexibil- ity in construction. Construction time horizons for telecommunication network facilities are only 1 to 3 years for trunk and switching equipment. Likewise, the financial implications of reaching a point of no return in construction are less treacherous. Half a network node or trunk group, in terms of fungible plant ca- pacity, is better than none at all, but half a nuclear power plant is not. That provides a certain amount of comfort for telco management. Should future cash flows fall short of expectations, heavy borrowing from external sources is not necessarily required. The bottom line is that no investment costs are sunk until actually committed.

Many possible engineering tradeoffs in broadband network design and construction serve to complicate costing out each physical network alternative in a business case analysis. Most important are the tradeoffs among fixed and variable network cost components. For example, two classic alternatives for ca- pacity additions are: (1) to lay more fibers in an initial cable installation; or (2)

to plan for increases down the road by adding more electronic and photonic equipment as needed, thereby reducing up-front fixed investment in favor of future incremental investment in electronic devices. Such considerations alter the mix of fixed, sunk, incremental, avoidable, and variable costs for any given network architecture over the planning horizon. In turn, the regulatory cost al- locations of telcos are affected since (mostly fixed)non traffic sensitive (NTS) costs (e.g., fiber optic trunk cable) are allocated to state and interstate regulatory jurisdictions in different ways than thetraffic sensitive(TS) costs for electronic devices. Even in the case of unregulated cable television firms, such tradeoffs among fixed and variable costs can alter pricing and capital recovery strategies prospectively, thus affecting the demand side of a business case.

The embedded base of network assets also may affect marginal investment decisions. Any given mix of embedded, fixed, and variable network facilities costs alters the decision parameters for future capacity additions. That is true even though, in theory at least, the embedded base of fixed costs became sunk and nonavoidable (irrelevant) for future decisions (or future business case analyses) immediately after they were incurred. The past selection of network technology and architecture can (and likely does) affect future marginal tech- nology and capacity decisions forever.

It is, therefore, important to anticipate the impact of current business deci- sions, such as choosing the next-generation network architecture, on future de- cisions. In a world of rapidly changing technology it is important not to simply adopt the least expensive short-run network architecture alternative (especially if it might not be robust enough to accommodate anticipated future demand and technology changes). In addition, how robust current investment decisions are to unanticipated demand changes also must be considered. In the case of residential broadband networks, the unknown factors of new digital wireless loop technology and satellite technology may come into play and make the land-based wireline loops prematurely obsolete. No doubt other new alterna- tives for local telecommunication will appear on the scene. It is important to avoid the potentially large financial mistake of deploying dedicated fiber optic subscriber loops if, indeed, alternative technologies around the corner can pro- vide similar service at lower cost.

Another important strategic consideration on the supply side of the broad- band network equation is to accurately identify the life-cycle stage of each customer service that is contemplated. Forecasts of demand growth and con- struction of network capacity requirements must be consistent with the place on the S-shaped life-cycle curve for each service. This obviously can affect the target for capital recovery or payback period. For example, even though many entertainment video services represent a brand new growth business for telcos, they are well along in their respective life cycles, and full market penetration may be around the corner. Growth in traditional mobile cellular service is also going to slow down at some point, and it is critical to accurately forecast that

point. Network investment strategies for both network hardware and software must be geared toward capturing new customer service markets that are early in their growth phase.

Currently, only two major players are anticipated to have a significant role in the future game of network deployment for advanced two-way residential broadband communication networks: cablecos and local telcos. On the other hand, as might be expected, there is a host of potential players in the business broadband communications market, including cablecos, telcos, private satellite and radio network vendors, private fiber optic networks, LANs,wide area net- works(WANs), andmetropolitan area networks(MANs). Due to relatively early demand drivers, such as those required for high-resolution imaging to support health care, high-speed computing, computer-aided design (CAD), computer- aided manufacturing (CAM), computer-aided engineering (CAE), and other business applications, the development of the market for business broadband services will proceed quite differently from that for residential services. The differences could be minimized if government authorities chose to promote an infrastructure approach to broadband network technology deployment. Indeed, unless an infrastructure approach is taken, in which government authorities promote early retirement of old technology, it may take several decades for resi- dential switched broadband networks to be widely deployed (because of the long-lived embedded base of traditional copper phone lines).

No one knows for certain what path is socially optimal for residential adoption of broadband network technology. Predictions, to date, vary from one or two decades to never. Whatever the predictions, business decisions must be made in both the public and private sectors to determine the next step. The world does not stand still; telcos and cablecos have to continue to upgrade their networks both to achieve network efficiency and to compete in the near term for new—and advanced—communication services. Furthermore, from the per- spective of the nation as a whole, our performance for domestic economic and social productivity and international competitiveness may not allow us the lux- ury of simply waiting for the existing base of public telephone network technol- ogy to physically wear out before replacing it with a broadband alternative. 3.6.1 Cash Flow Analysis

The results of standard calculations of the net present value (NPV) of future cash flows, properly applied to the life-cycle costs of network investments can be used to evaluate the financial viability of both short- and long-term network upgrade alternatives for telephone and cable network companies.16 _Life-cycle 16. For a description of the capital budgeting decision process and the role of NPV analysis, see R. Brealey, S. Meyers, and A. Marcus, Fundamentals of Corporate Finance (New York: McGraw-Hill, 1995), chaps. 6–11.

costs include both the upfront network construction costs (installed first costs, or IFC) and the annual ongoing expenses associated with operating the network expressed as an annual or monthly annuity amount.

For a very high level analysis, some rules of thumb exist for evaluating the costs of telecommunication network investments. For example, in the tele- phone industry, an annual charge factor of 40% is used to reflect the annual costs (sometimes referred to as carrying charges) of new network investment. There are four components of the 40%:

• The annual depreciation charge of 10% (indicating an average service life of 10 years);

• Annual interest charges, or “cost of money,” of 10% to reflect that funds tied up in network investments could be earning interest (or that interest has to be paid on the borrowed funds for the network investment);

• Annual income tax expenses of about 4% (composite corporate tax rate times the 10% return);

• Annual operating expenses and allocated overhead of 16%.

Using those four components, for every additional $1 of network invest- ment cost, a telco would need to increase its annual revenue (immediately in the first year and every year thereafter) by $.40 to recover its original capital investment over 10 years’ time. Thus, for every $1,000 of broadband/multime- dia network upgrade costs per subscriber, a company would need to start col- lecting $400 more a year ($33 monthly) from every subscriber just to recover its original investment costs.

Very few comprehensive capital budgeting analyses of broadband net- work upgrades have been made public, but the results of quite a few general analyses have, by now, been published without revealing the detailed (often proprietary) underlying analysis.17 _{One of the most comprehensive studies to} attempt a standardized financial comparison of the NPV of both telco and ca- bleco broadband upgrade alternatives was performed by a consortium of Euro- pean researchers employing a host of technical and financial evaluation tools developed by the European RACE project [7]. That study juxtaposed the likely per-subscriber costs for both telcos and cablecos for: (1) new construction; 17. For example, see P. Shumate, “Broadband Access Networks,” “Broadband Subscriber Access Architectures and Technologies,” “First Costs, Operations Costs, and Network Comparisons,” and “First Costs, Operations Costs, and Network Comparisons—Summary” (Bellcore, 1995); B. Egan, “The Case for Residential Broadband Telecommunication Networks: Supply-Side Considerations” (Columbia Institute for Tele-Information, Research Working Paper Series, Co- lumbia Business School, February 1991); and “Overview of New Technology Deployment Model: Broadband and Associated Depreciation and Overheads” (Telecommunications Indus- try Analysis Project, University of Florida, Gainesville, FL,Quarterly Bulletin, Vol. 16, No. 4), pp. 569–573.

(2) relatively modest broadband network upgrades using ADSL and HFC alter- natives; and (3) relatively aggressive upgrade scenarios for switched integrated broadband service. The study also projected demand for each scenario and cal- culated the NPV of cash flows for both telcos and cablecos for the 15-year period 1996–2010.

The analysis indicates that the broadband network upgrade costs for both telcos and cablecos are roughly comparable to the investments already made in their current networks (actually somewhat higher for telcos and somewhat lower for cablecos). The NPV and the payback period of the upgraded networks in the modest upgrade scenario were similar to that for the base case scenario of no network upgrade at all (i.e., present method of operations to the year 2010). In fact, viewed in isolation, when the incremental costs of the network up- grades themselves were compared to the incremental revenues from new inter- active broadband services that the upgrade made possible, the investments never paid for themselves.

Another comprehensive analysis was conducted by Bellcore to evaluate the NPV of cash flows associated with various alternatives for upgrading the telco POTS network to provide VDT service [8]. The results of the Bellcore study are similarly pessimistic. The average discounted payback period for re- covery of the initial upgrade investment costs was 6–7 years for ADSL type sys- tems and 11–15 years for HFC and FITL systems, with most of those being closer to 15 years. Even under some fairly aggressive demand assumptions, the NPVs (and corresponding internal rates of return) are never very high.

From the viewpoint of telco management, in an age of increasing competi- tive alternatives for multimedia distribution systems, 15 years is a long time to wait to recover an initial investment. Despite the early pessimistic financial forecasts, given that the future high-growth telecommunications markets will be for digital data and interactive multimedia services, broadband network up- grades remain a strategic imperative for most incumbent telcos and cablecos. After all, the alternatives—gradually losing market share18_{or abandoning the} public network business by selling out—are not so rosy.

References

[1] Shumate, P., “Broadband Access Networks,” “Broadband Subscriber Access Architectures and Technologies,” “First Costs, Operations Costs, and Network Comparisons,” and “First Costs, Operations Costs, and Network Comparisons—Summary,” Bellcore, 1995.

[2] Salamone, S., “Higher Data Speeds Coming for Plain Phone Lines,”BYTE, January 1996, p. 30; Shumate, P., “Broadband Access Networks,” Bellcore, 1995.

[3] Darcie, T., AT&T Bell Laboratories, “Broadband Subscriber Access Architecture and Tech- nologies,” OFC ’96, San Jose, CA, Feb. 28, 1996.

[4] Hobbs, J. (Kessler Marketing Intelligence), “Cost Factors Affecting Fiberoptic Deployments in Communications Networks,”Proc. OFC ’96, San Jose, CA, February 28, 1996.

[5] Jones, J., “Video Dialtone: Choosing the Right Network Architecture,” Broadband Technolo- gies, Inc., April 1993.

[6] Reed, D.,Residential Fiber Optic Networks: An Engineering and Economic Analysis, Nor- wood, MA: Artech House, 1992. Also see Reed, D., “The Prospects for Competition in the Subscriber Loop: The Fiber-to-the-Neighborhood Approach,” FCC Office of Plans and Policy, presented at 21st Annual Telecommunications Policy Research Conference, Solomons, MD, September 1993.

[7] Olsen, B., et al., “PNO and CATV Operator Broadband Upgrade Technology Alternatives: A Technoeconomic Analysis,”Proc. OFC ’96, San Jose, CA, February 28, 1996.

[8] “Residential Video Dial Tone Network Service Prospectus,” Special Report SR-TSV-002373, Issue 1, Bellcore, Livingston, NJ, December 1992.