In the calculations which are presented in the next section, we substitute an annual fee of $63 plus $4.92 for the second connections, or $67.92 for Mitchell's low figure of $62.40. While our typical current monthly fee of $5.66 ($5.25 + $0.41) per month is $0.46 higher than Mitchell's, it is considerably below the price charged by many systems currently in operation in major markets. In New York City, one system charges $9.95 per month while another-in the more modest section of Manhattan--charges but $6.00. In San Francisco, the monthly fee is $6.25 while in Oakland it is only $4.45 and in Los Angeles it is $7.00. These differences persist despite Mitchell's claim that systems face unitary elastic demand and that increases in the subscription fee to above $5.00 per month will not raise net revenues of the system. 14 A final fragment of evidence supporting the notion that subscription fees can be raised with salutary effects upon profits is the number of price changes sought by cable operators. Between January and September 1972, 40 systems were reported to have sought increases in their monthly fees and only one asked for a decrease. 15 An important omission from both our calculations and Mitchell's is the installation fee exacted by cable systems for a household's initial connection to the cable. The random sample of cable systems which we collected for the purpose of testing alternative demand functions lists an average installation fee which is slightly more than $10, but it is well known that many systems waive the installation fee as an incentive to gain additional subscribers. Unfortunately, precise data on the frequency of billing for installation are not available. 14His deductions are theoretically invalid, for he presumes that a firm confronting a demand curve of constant unit elasticity cannot increase price profitability. In fact, it can and should--until it serves but one customer (as long as the second and ensuing customers add to operating expenses of the firm). 15Source: TV Factbook Addenda. The omission of installation fees creates a certain asymmetry in the calculation of rates of return because we have followed Mitchell and included a charge of $7.50 for each disconnect and reconnect in the "house drops" cost category. Since households are assumed to move once every four years, this adds $3.75 per year to costs in a steadystate environment. If the installation fee of $10 were included in our revenues, it would add $2.50 per year to revenues, in large part offsetting the disconnect-reconnect charges, but without substantive data on the frequency of the levying of installation charges we have decided to ignore this revenue category altogether--leading to a downward bias in calculated rates of return in the next section. b. Ancillary Revenues. To his modest $62. 40 in subscriber fees, Mitchell adds an equally conservative $2.20 for advertising revenues. This estimate is also permitted to grow at a rate of zero percent a year, and it is the only allowance for ancillary revenues. In our calculations, we choose to ignore the category of advertising revenues altogether, substituting for them a unitary category of income from all sources, including one which appears likely to dominate advertising revenue for cable operators in a very few years--leased subscription channels. How remunerative will leased channels prove to cable owners? At this time precise evidence on this question is impossible to muster. One study by the Rand Corporation of the potential Dayton-Miami Valley 16 system projects the leasing of one motion-picture channel and ten educational channels. With no firm basis for their estimate, the Rand researchers simply posit a gross revenue of $350,000 per year for the motion picture channel--equal to ten times estimated costs. Since the Dayton system is projected to reach an average of 62, 830 subscribers during the first ten years, this revenue averages approximately $5.50 per subscriber per year. In addition, Rand projects another $350,000 in annual revenue from the leasing of the educational channels at cost, but there is no projection of other leased noneducational channels. 16L. L. Johnson, et al., Cable Communications in the Dayton Miami Valley, The Rand Corporation, F-943 KF/FF, January 1972, p. 2-10. The Rand estimate is conservative given the economics of motion-picture production and distribution, and the likelihood that other entertainment vehicles will be very attractive when offered by leased channels. Given theater admission prices which average more than $1.25 per seat at present, it would not be surprising if the average cable subscriber would be willing to pay several times this amount for the privilege of viewing new feature films via a leased channel in his home. The Rand estimate of leased-channel revenue is roughly consistent with a monthly charge of $4.50 for 15 new films with one repeat of each during the month and a subscriber enrollment of 50 percent. Alternatively, it may be viewed as being consistent with the sale of $4.50 per subscriber per month in advertising time on the leased channel and it is even more conservative as an average estimate of the attractiveness of this option over the next ten or fifteen years. Similar examples could be constructed for sports events, cultural offerings, or numerous other services, but all would rest heavily upon conjecture at this point. We can only assert that at this moment extra subscriber revenues for cable systems are not very important but that in the very near future they are likely to become significant. Moreover, it is clear from the behavior of buyers and sellers of cable systems that industry personnel expect considerable revenue growth in the near future. Therefore, in the analysis of Section E below, we shall always assume that current revenues per subscriber are derived solely from subscription fees of $67.92 per annum, but we allow these revenues to grow over time as ancillary revenues assume increasing importance. To summarize, we have suggested that future revenues per subscriber will rise because of modest monthly fee increases and because of the very considerable potential of ancillary services. Nevertheless, we perform the calculations in the next section under three sets of assumptions about price: Price is equal to $67.92 per year for all 45 years. Price is equal to $67.92 and grows by 2 percent per year. Revenues per subscriber are equal to $67.92 and grow by 4 percent per year. All three estimates are above Mitchell's gloomy $64.60 per year, ad infinitum, and they are more consistent with the anticipations being discounted by current buyers of franchises as we shall see in Section D below. 17 Despite the great effort undertaken by Comanor and Mitchell to fit a statistical demand function to a large sample of cable systems' data, Mitchell now drops the "best" equation from the earlier Comanor-Mitchell study in favor of a more recent study by Rolla Edward Park of Rand. He argues that since Park utilizes data for communities having reception capabilities similar to those in the top 100 markets, allows for differences in reception capability, and has gone to the trouble to verify his data with cable operators, his results are superior to those obtained in earlier Comanor-Mitchell study. Park's study is based upon a sample of 63 systems in markets generally served by three or more commercial signals. He does not and cannot divulge the indentity of these systems since data were obtained on a confidential basis, but he has indicated that the mean size of system is 4, 300 subscribers with mean age less than six years. Few very large systems are included--20, 000 being the largest system in the sample. He chooses these systems to be "representative" of top100 market systems, but he excludes all areas in which signal obstruction is deemed to be a problem. Such problems, however, are not uncommon 17R. E. Park, op. cit. in markets such as New York, Chicago, and San Francisco. These three markets alone account for almost a quarter of the potential subscribers in the top 50 markets. His regression equation is similar to that utilized by ComanorMitchell with one important difference: each off-the-air signal is weighted by its distance from the B contour, expressed as a fraction of the radius of the B contour. Thus, distance is taken as an invariant measure of signal reception by households. This assumption is quite dubious on its face for two reasons. First, a station's signal strength is allowed to vary considerably by the FCC and the procedure for estimating the location of the B contour is known to be imprecise. Second, the quality of the local signal is in large part a reflection of the household's investment in antennas. In older areas with older television stations, these antennas are likely to be larger and more sophisticated. In future years, normal attrition of these antennas will make cable more attractive, but examination of any of these markets at present will underestimate future cable penetration. In estimating his demand function, Park attempts to estimate an exponential maturity factor similar to that attempted by ComanorMitchell. He finds that the "best" estimate of this growth factor is -3.3/t equal to e where t is system age in months. Comanor-Mitchell, -450/t on the other hand, discovered that e was the best fitting maturity factor. These two estimates--neither of which is utilized by Mitchell in his simulations - are quite different and give rise to very different paths to eventual system maturity. Surprisingly, Park then proceeds to estimate his equation under the assumption that a system approaches maturity at a linear rate of t/18 for the first eighteen months, reaching maturity at a mere eighteen months. He argues that this gives him his best fit in the penetration equation, and all of his estimates are dependent upon imposing this maturity path upon the penetration expression. |