5. Brookline has not made any final decision on the extent of underground CATV system required, but it is likely that 20 percent of system will be laid underground in existing utility ducts.

In short, Mitchell's "documentation" of underground cable percentages is substantially in error. City officials in each of the four cities cited are frank to admit that little new trenching will be undertaken in their communities because of its prohibitive private and social costs.

What is required is substantial documentation of the extent of underground construction in "typical" systems--especially those in the largest, most dense markets. In order to shed light upon this question, we attempted to contact all of the systems which are either operating or under construction in the central city of the top 50 markets with the exception of New York City. The responses which we received by telephone are documented in Table C-8.

20-344 O 73 23

Outside California and New York City, underground cable

percentages rarely exceed 5, and in California--where new housing construction is especially important--underground cable costs are often much lower than those specified by Mitchell because the television cable is buried at the same time that other utility lines are laid. Thus, there is little sound data from actual or prospective cable systems that underground percentages will "typically" average 10 or 20 in even the largest markets. In cities such as New York, Washington, or San Francisco, where topography and other unusual circumstances dictate burying cable at great expense, higher subscriber fees will be paid by television homes. But, these cities should not be used as the model for "typical" systems in calculating rates of return. To do so would bias the results severely. Timing of Capital Expenditures. A second assumption

b.

which leads Mitchell to underestimate the internal rate of return on cable systems is his assumption (and the assumption of Comanor-Mitchell) that the entire plant is constructed at the beginning of the first year. In virtually every system, construction is phased out over more than one year, and in many completion requires three or more years. In their study of the Dayton-Miami Valley system, the Rand researchers assumed that the distribution system would be built in a three-year period, with 21 percent completed in the first year, 44 percent built in the second year, and 35 percent in the third year. We utilize an intermediate pattern in our calculations in the next section, a pattern which dictates higher rates of return since it reduces the present value of capital expenditures at any calculated discount rate.

Mitchell assumes that his large underground cable percentages will be achieved at a cost of $15, 480 per mile because he implicitly assumes that large cities will allow and even require cable companies to trench and lay conduit throughout the core city area. In fact, this is not likely

to occur in many large cities because of the costs and discomforts created by the trenching and filling process. Many cities require, instead, that the telephone or electric utility company provide conduits for other purposes. These conduits are laid when other utility lines are laid, and cable operators may be required to use them.

In cities such as Seattle, San Jose, and Los Angeles,

underground cable is laid at the same time utility companies bury their cables. This leads to much lower costs than independent de novo construction hypothesized by Mitchell. Even where underground construction is undertaken independently, the cost of laying cable can vary enormously. Oakland is laying 38 channels of cable underground at an estimated cost of $19, 000 per mile while San Francisco, across the bay, is encountering costs of up to $50,000 per mile and more. These differences are reflected in different monthly charges--$4.45 for Oakland and $6. 25 for San Francisco.

21,

Where cable is simply strung through underground ducts, the cost per mile may even be lower than aerial construction. One study conducted by the Stanford Research Institute' found that this type of construction cost only $3,000 per mile, or at least 25 percent less than aerial construction. Maintenance costs may be greater for this alternative, however, since major repairs or alterations may require the assistance of other utility companies who share the same ducts.

In Mitchell's rate of return calculations, it is assumed that the entire plant is rebuilt in each 15-year generation. Moreover, because of the assumption noted above, this plant is replicated at the beginning

21Stanford Research Institute, Business Opportunities in Cable Television, March 1970.

of each generation. Clearly, these assumptions overstate actual capital expenditures. Even if most of the cable plant has a 15-year life in the face of extensive maintenance built into operating costs, not all of the plant requires rebuilding. Trenches dug in the first year and conduit laid in these trenches do not have to be replicated in year 16. Many of these conduits will survive for several generations, and some may not require rebuilding in the foreseeable future. Moreover, tower expenses need not be replicated every 15 years. In some cases, new technology will dictate replacement of capital equipment but only if operating costs are so reduced by the improvement that average costs of operation are less than incremental costs with the older equipment. Thus, to the extent that rebuilding is dictated by new developments, operating costs should be reduced accordingly. Mitchell does not do this; he simply reproduces the plant in toto each 15 years--a methodology which obviously lowers the realized rate of return.

In our calculations in Section E we shall make the

reasonable assumption that the underground percentage in the typical system is 5 percent. Moreover, we shall phase the initial (and subsequent replication of) investment over two years, a pattern which the Rand Dayton Study uses for each of the sectors of its enormous prospective system. Unfortunately, we cannot present very firm data on the percentage of cable which will be laid in utility ducts nor on the share of plant investment which will not require replication. Therefore, we utilize Mitchell's data on aerial and underground costs per mile and, like him, assume that the entire plant is rebuilt each generation even though we know that these assumptions will lead to conservative estimates of the rate of return.