1. Network Design Efficiencies for Service Providers. In maximizing the cost effectiveness of their deployments, BSPs seek to aggregate individual locations where the build is economic. As a general matter, BSPs will be more willing to bid and bid aggressively for BEAD funds to the extent a project would align with their business plans. From a network design perspective, BSPs largely ignore boundaries established by governments; rather, cost per location and potential revenue are the key factors in deciding where it would be economic to build and in driving project design decisions.
2. Maximizing Fiber Broadband Deployment While Ensuring All Eligible Locations Are Served. Eligible Entities, each with different funding allocations, are charged with ensuring that eligible locations receive BEAD funding, prioritizing unserved and then underserved locations. They must also uphold BEAD’s preference for fiber connectivity by setting the Extremely High Cost per Location Threshold (EHCT) at a level that ensures fiber is deployed as widely as possible. For administrative convenience, Eligible Entities may want to apply the threshold at the minimum biddable geographic level. For example, if the minimum biddable unit were a census block, each census block would be deemed either above or below the threshold based on the average costs of serving eligible locations within that census block. However, the larger the geographic area at which the threshold is applied, the greater the possibility that many of the very highest cost locations will be included, which would lead to the unfortunate result that “fiber viable” locations will not receive fiber because they are “grouped in” with these highest-cost locations.
3. Administrative Feasibility. Eligible Entities may prefer define minimum biddable units that align with existing administrative boundaries in order to simplify the subgrantee selection process. However, this concern may be overstated, as sophisticated computer models are readily available that can handle “deconfliction” of any applications that partially overlap geographically, i.e., that include some but not all of the same locations.
As noted above, Eligible Entities are likely to apply the EHCT at the level of the minimum biddable geographic unit. In other words, all locations within the chosen geographic unit (e.g., county, census block) would be deemed to fall above or below the threshold based on the average cost to serve all the locations, irrespective of the costs to serve any specific location. Where average cost per location is calculated across a large geographic area that includes a significant number of extremely high-cost locations, it will lead to larger proportions of relatively low cost locations falling above the EHCT – contrary to the Program’s goal of deploying fiber as widely as possible. Conversely, the choice of smaller geographic units would tend to minimize this effect.
Figure 1 is an example showing the impact of setting boundaries by census tract or census block group for two example counties. As the figure demonstrates, setting boundaries at the more granular level of census block group results in a far greater share of BEAD-eligible locations falling below the threshold. (Note we have assumed that only BEAD eligible premises will be included in the boundary.) And, if we examined census blocks, an even greater number of locations would fall below the threshold.
At a high level, Eligible Entities face the choice of (1) defining a minimum geographic unit for deployment projects, most likely along existing administrative boundaries (e.g., counties, census tracts, census block groups, census blocks) or (2) inviting applicants to construct project areas on a location-by-location basis1. These options include various permutations, several of which we examine below.
1. Subgrantees Set the Project Areas
The most straightforward approach from the standpoint of a prospective applicant is to allow each applicant to determine its project area, i.e., to decide the individual locations it wishes to serve. BSPs and other relevant parties could submit their preferred areas, to be assessed and consolidated by the Eligible Entity, with processes on the back end to address overlap and any “stranded” locations. This approach would be most likely to generate interest and maximize bids from a broad range of potential Subgrantees.
2. Census Block Areas
Likely the smallest administrative boundary that could be used as a minimum biddable geographic unit is the census block. The number of census blocks per state ranges from ~6,000 (District of Columbia) to 670,000 (Texas), although many census blocks have no BEAD-eligible locations and can thus be ignored. As this is the most granular unit, it would be the most appealing to smaller providers and least likely to disadvantage fiber builds. BSPs could submit applications using aggregations of census blocks. Eligible Entities, either on their own or at the recommendation of BSPs, could remove the very highest cost locations from a census block to enable more cost-effective bids.
3. Census Block Group Areas
The next largest unit is the census block group. There are ~230k census block groups in the US and an average of ~3 census block groups per census tract. Opting for this boundary size versus larger sized boundaries may further encourage the participation of smaller providers, which in turn increases competition in the bidding process and would likely increase the share of fiber over alternate technologies. However, an even smaller boundary would produce more optimal outcomes, with minimal additional administrative burdens.
4. County or Census Tract Areas
There are ~3k counties and ~80k census tracts in the US with an average of 25-30 census tracts per county. Eligible Entities may find it tempting to define project areas at the county or census tract level to minimize their administrative burden. However, these larger areas carry a higher risk of the inclusion of very high-cost locations or a natural boundary, such as a river or mountain range, that could sub-divide the area and make it prohibitively costly for any single applicant to serve. Furthermore, using larger areas would tend to limit viable fiber builds (See Figure 1, above). Finally, smaller BSPs would be deterred from submitting bids for such large areas – reducing competition in the funding process. In sum, these larger areas present risks of poor outcomes, undermining the BEAD program’s aims.
In recent research (download required), Cartesian further demonstrated that smaller areas tend to be best at minimizing the number of locations above the high-cost threshold. As individual locations have different costs, the choice of area will determine the average location cost for each project. Cartesian compared project areas defined at the level of census block, census block group, and census tract to investigate the effect of project area granularity on average location cost. Cartesian compared the average location costs against an assumed $12k high-cost threshold (including provider match), which Cartesian used in its BEAD Framework cost model. The chart below shows, for each State and each of the relevant geographies, the percentage of those geographies in the State that have an average location cost of $12k or less.
Grouping project areas by census block produced the highest percentage – i.e., the best result – in 25 cases. It is worth noting, however, that this analysis considers each geographic unit (e.g., each individual census block) in isolation. The efficiency of census blocks as biddable units would be further improved if applicants were permitted to construct project areas from census blocks of their choosing.
The smaller the biddable unit, the greater the likelihood that bidders, where permitted to aggregate these units, will submit applications with partially overlapping project areas. This prospect should not deter Eligible Entities from making census blocks or even individual locations the minimum biddable unit, as there are a variety of ways for an Eligible Entity to address these overlaps, especially given the availability of sophisticated software for “deconflicting” proposed project areas. Using such software, the Eligible Entity could conduct the following process:
It is important that project areas are defined such that Eligible Entities can deliver fiber broadband cost-effectively to as many unserved and underserved locations as possible. Defining project areas granularly – such as at the location or census block level – will maximize participation from prospective bidders and best ensure that “fiber-viable” locations actually receive fiber, especially if the very highest cost locations are removed and treated separately. We encourage Eligible Entities to collaborate with potential Subgrantees to facilitate participation from a broad spectrum of BSPs.