TVA Rural Studies

Improving Rural Telecommunications Infrastructure

Bruce L. Egan
Columbia University

4. Rural Telephone Plant Characteristics and Costs

Based on RUS company cost characteristics, one broad gauge estimate of the total cost of providing rural telephone service in the US is $19B per year²⁰. This total assumes that all 22M non-MSA subscriber lines are classified as rural and an average annual cost of $871.08 or $72.59 per month.

There are significant differences in the physical characteristics of rural vs. urban telephone plants. RUS companies' markets are very thin, averaging only 4 subscriber lines per square mile of area served and only 6 lines per route mile of telephone transmission plants. For large telephone companies the average density of subscriber lines is greater by an order of magnitude²¹. Large LECs have five times more lines per switching office and almost five times less transmission facilities per line than small LECs (measured by sheath meters of copper cable - Table 2). The average length of subscriber connections to the LEC exchange switch for large LECs is about 10,000 feet vs. double that for small LECs. However, the net result is that the average investment and expense per subscriber line is only about 40% higher for the small LECs (Table 2).

Chart 2 shows a breakdown of small LEC total capital expenditures by major category of plant. Eighty-five percent of small LEC capital investment is represented by switching plant (31%) and cable and wire facilities (54%). Large LECs have 82% of total investment in switching plant(38%) and cable and wire facilities (44%). For both large and small LECs the remainder of the investment is primarily in land, building and support assets.

The average loop length for RUS companies is 20,330 feet, which is significant considering that access lines longer than 18,000 feet usually require special treatment to insure high quality basic service. The main problem is the attenuation of the analog signal, which may require boosting, using repeaters and amplifiers, or passive reduction of attenuation losses by loading coils, or both. Such loops pose a problem for the narrowband digital and new broadband services that require relatively high quality circuits for error free digital transmission. However, the mode loop length is less than the average for RUS companies. Consequently, 55% of the loops are less than 18,000 feet. The majority of RUS company loops are actually non-loaded, but many still receive treatment of some kind to improve transmission and signal quality. In contrast, about 90% of RBOC loops are less than 18,000 feet, and a large majority of those are non-loaded with an average length of only 7,500 feet.

On average, there are about 7,400 access lines per telephone company exchange in the US. Bell companies (BOCs) have about 12,000 lines per exchange²². Non-Bell Independent Companies (ICOs) have only about 3,000 lines per exchange. For 1993, the RUS reports an average of only 1,223 lines per exchange.

Average statistics regarding costs and network operations can be very misleading when considering any individual LEC or specific geographic region and caution must be used before ascribing average statistics to any company or group of companies. An examination of the RUS data for individual companies indicates some highly skewed distributions. Charts 3-5 illustrate the high variability in small company network characteristics including the number of exchanges, the number of subscribers and the average exchange size. For example, Chart 3 shows that the average number of exchanges per small LEC is 6 while the standard deviation is 8.5 and by far, most companies have only 1. Chart 4 shows that the average number of subscribers per company is 6,341 with a standard deviation of 14,000 with most companies having under 1000. Chart 5 shows that most RUS companies have between 200–400 subscribers per exchange, while the average is 1,223 and the standard deviation is 1,499. There are a considerable number of companies with over 2,800 subscribers per exchange.

Indeed, even within a single rural exchange area there are substantial differences in the physical characteristics of subscriber connections. This means that it is not only misleading to ascribe average company or exchange statistics to individual companies or exchanges, but that it is also problematic to apply average loop characteristics of a single exchange to individual subscribers. This has enormous implications for public policies that are trying to accurately target funding assistance to those subscribers who are truly in need.

Figure 1 is a stylized example of a representative local exchange area for a rural telephone company. The average exchange is comprised of about 1,200 households with a relatively dense downtown area containing 65% of total lines in the exchange area and 35% considered to be in the rural surrounding area of the exchange. The “typical” rural exchange as shown in Figure 1 has 768 households in the downtown area at a density of 256 subscribers per square mile, and 440 rural households with an average density of 6 per square mile. This example of a “typical” exchange shows that it is the rule rather than the exception to expect very different costs for individual subscriber connections within the same exchange area.

To illustrate the impact of subscriber density on the average cost per subscriber for rural LECs, Chart 6 provides cost estimates for the average urban and rural subscriber in the stylized exchange presented in Figure 1. The overall average per subscriber cost is $2,200. For the urban zone of the exchange the average cost is $800 and for the rural zone it is $6,000. As expected, the difference in cost is due primarily to the placement of longer loops for the rural subscriber.

A further examination of the variability of rural loop costs among small LECs can be found in Table 4 which provides a breakdown of total investment per subscriber for three density bands 1–10 lines per kilometer (km), 10–100 lines, and 100–500 lines. The per subscriber cost in the lowest density band (0-10/km) is about one third higher than for the second (10-100/km) and three times higher than the highest density band 100-500/km, with the average investment being $2,055 per line. Even within each density band, however, it would be misleading to ascribe the average cost result to any one company. For example, there could be drastic differences in topology and terrain which would dramatically affect costs but which do not appear in this data. One company may serve a relatively flat area with sandy soil, while another might be hilly or mountainous featuring solid rock. The spatial distribution of subscribers in a single exchange area could be exactly the same for both companies and yet the per subscriber costs for each could vary by an order of magnitude or more. The bottom line is that local conditions matter a lot.

Table 5 provides further support for the need to consider local conditions when assessing average cost characteristics. This Table displays statistical correlations between key publicly available measures of subscriber distance and density and investment and expense costs per line actually observed for 886 RUS companies. The subscriber density measures which were correlated with average cost per line were subscribers per route mile of cable, subscribers per square mile of serving area, and subscriber lines per switch. The very low values of the standard correlation coefficients demonstrate that there is no significant relationship between density measures and costs. Yet, it is well known that local factors like terrain notwithstanding, the primary engineering cost driver in local telephone networks is the distance of subscribers from the exchange. The second set of correlation coefficients is based on positioning all of the observed values for each variable in rank order from highest to lowest and correlating the rank ordered vectors. The very high rank correlation coefficients do indicate significant relationships, but now they have no meaning for any given company since the ranking of variable values were made without regard to which company the values belonged.

Kentucky is considered one of the most rural states in the US and Table 6 shows how small LECs average costs and revenues may vary within any given state. There are 16 rural Kentucky LECs that borrowed from the RUS in 1993. Table 6 (2 pages) provides operating and financial statistics for each of them. The weighted average revenue and cost per line and network density for the combined Kentucky rural LECs (second last row of Table 6) are fairly close to those for the national averages which appear in the last row of Table 6.

Conventional wisdom (at least to the layperson) is that rural telephone companies serve sparsely populated regions with little or no urban areas. This is not true. The available data makes it clear that inferences for any given company based on the average statistics for the group could be grossly misleading. Similar data is available for small LEC revenues and expenses. This data provides an important message for policy makers and regulators which may be tempted to develop competition policies and rural subsidy requirements based upon average cost and revenue statistics. There is no such thing as an “average” rural company, and no such thing as a “meaningful” average measure of the subsidy requirement.

Jump to Section:
Contents, 1, 2, 3, (4), 5, 6, 7, 8, Tables, Footnotes, Charts & Figures

Back to the 1996 Rural Telecommunications Workshop Homepage