Calculating Average Per-Unit Costs/Take Rate

Please provide comments to the issue below as part of the 2012 WCB cost model virtual workshop for inclusion in the record. Comments are moderated for conformity to the workshop's guidelines.

Background

Hybrid Cost Proxy Model: The HCPM calculated the average cost-per-line in use by dividing the total cost of serving current customer locations within a wire center by the number of lines in use, as determined by National Exchange Carrier Association (NECA) data. Because it used current lines, the Commission used the number of households (i.e., occupied housing units) as a proxy for current residential customer locations, rather than the number of housing units (i.e., occupied and unoccupied housing units). At the time, the Commission observed that "as long as there is consistency in the development of total lines and total cost, it makes little difference whether households or housing units are used in determining cost per line."

The Commission found that if it were to calculate the cost of a network that would serve all potential customers, it would not be consistent to calculate average cost-per-line by using current demand, and it declined to estimate future demand because "[t]he level and source of future demand . . . is uncertain." The Commission was "concerned that including such a highly speculative cost . . . may not reflect forward-looking cost and may perpetuate a system of implicit support."

Although the HCPM did not explicitly use a take rate, it effectively assumed a 100 percent take rate, because unit costs were calculated by taking total costs (which were calculated based on deploying plant to the number of occupied locations) and dividing by the number of lines in use, as determined by NECA data. At that time in the late 1990s, the number of lines in use was equal to or greater than the number of occupied locations due to the prevalence of second lines for dial-up Internet and fax machines.

Connect America Cost Model: The first version of the CACM (CACM v.1) has been designed to calculate the average cost-per-subscriber by dividing the total cost (the fixed cost of passing all locations in a given area plus the variable cost associated with serving active subscribers) by the number of active subscribers. Unlike the HCPM, CACM v.1 has the ability to calculate variable cost (e.g., the amount of success-based capital investment and the capacity of network links) based on an assumed level of demand.

Questions for Comment

1. The Bureau believes that it is reasonable to design the CACM so that it includes the total cost of passing all locations and the cost of actually serving subscriber locations in its calculation of per-unit cost. The Bureau proposes to modify CACM v.1 to calculate all unit costs on a per-location-passed basis, rather than on a per-subscriber basis. We seek comment on this approach. The Commission's objective is to ensure access to modern networks to all homes and businesses in a funded area; and the cost-per-subscriber, given the impact of take rate, may not always provide a means of comparing those costs. We note that as take rate varies, whether across different areas in a given model run or across different model runs, the cost-per-subscriber can change dramatically even if the total cost changes very little. As shown in the simplified example below, using cost-per-location-passed would lead to unit costs that are more stable and more reflective of the total cost of the network (i.e., varying little based on take rate).
   
   To illustrate the point, imagine an area with 10 homes (locations) in it. In this area, the fixed cost to build a network to pass all those homes is $1,000 while the variable cost to connect each home is $10. If one assumes a 90 percent take rate, there would be nine subscribers, the total cost would be $1,090 and the cost-per-subscriber would be $121 ($1,090 divided by nine subscribers). If, on the other hand, one assumed a 50 percent take rate, there would be only five subscribers; the total cost would be $1,050 and the cost-per-subscriber would be $210. Thus while total cost decreased by less than 4 percent with the lower take rate, the cost-per-subscriber would increase by more than 70 percent. Comparing these unit costs across different runs, or between similar areas with different take rates, would be complicated by the dramatic change in per-subscriber costs. On the other hand, if one determines the cost-per-location-passed, the difference in unit costs is small, similar to the change in total cost. In the first case, with a 90 percent take rate, the cost-per-location is $109 ($1,090 total cost divided by 10 locations); and in the second case, with a 50 percent take rate, the cost-per-location is $105.
2. The Bureau intends to explore adjustments to the take rate in CACM v.1 to determine the impact of varying assumptions on calculated support per-location. Setting the take rate at 90 percent of all residential and business locations passed (as proposed by the ABC Coalition in the CQBAT model) would effectively assume that all occupied locations take both voice and broadband service. The Bureau recognizes that in areas eligible for Connect America Phase II support, some number of customers could opt for mobile services for voice, while relying on a Phase II-supported carrier for fixed broadband. In addition, even in the presence of latent demand, it likely would take some time for customers to adopt a newly available service like fixed broadband. If the CACM ultimately adopted by the Bureau were to calculate costs on a locations-passed basis, is a 90 percent take rate a realistic assumption for price cap territories? How, if at all, should the Bureau take into account the fact that more than 30 percent of households do not subscribe to landline voice service at all? How, if at all, should the Bureau take into account the possibility that not all locations that have broadband newly available will subscribe to that service, unlike historical trends for voice service, and therefore variable costs will actually be lower? What adjustments to the take rate, if any, should we make to reflect the possibility that some people may choose to subscribe to mobile broadband, but not fixed broadband services, and some may not subscribe to broadband service at all during the five-year term of Phase II support? Would a blended-across-time take rate of 50 percent reflect a reasonable average for demand for broadband services over a five-year period? How, if at all, should the Bureau take into account that different locations may subscribe over time, leading to incremental additional cost for the provider? Would it be appropriate, for example, to over-estimate take rate on the assumption that while only 50 percent of subscribers might purchase a service at any one time, that closer to 1.5 x 50 percent of locations will subscribe at some point over the five-year period?