USC Annenberg has releases a report (BEAD program targeting: A preliminary analysis) that looks at potential impact of BEAD with recommendations on how to reach the folks who were originally priority households for the funding. High level remarks on the report…

Overall, the results suggest that BEAD will primarily benefit low population density areas where adoption of high-speed broadband is lagging, with limited expected impacts in other areas. A particularly important finding is that, after controlling for other demographic factors, the share of households in poverty is not correlated with priority BEAD areas in plans submitted to the NTIA. This finding raises concerns about whether BEAD will meet its mandate of reducing income-based disparities in high-speed broadband access. At the same time, the findings also point to vast differences in program targeting across states, suggesting that these impacts will be highly dependent on local contexts.

Getting into details of how low is the population density…

A key demographic characteristic of BEAD clusters is low population density: the average density across these block groups is ~300 people per square mile (PPSM), compared to approximately 6,500 PPSM in other block groups. It is worth noting this figure is lower than the 500 PPSM threshold used by the Census Bureau to distinguish urban from rural areas. In total, approximately 93% of BEAD clusters fall below the 500 PPSM threshold. This suggests that the impacts of the BEAD program are likely to be very limited in urbanized areas and large population centers.

The report also looks at broadband access, income disparity and poverty and comes to finding the characteristics that are more strongly associated with BEAD clusters…

In order to examine this question, we create two regression models to predict whether a block group falls within our definition of a BEAD cluster conditional on a set of demographic characteristics including education, age, race/ethnicity, employment and poverty levels (see table A1 for full model results). The two models are identical except that in model 1 the share of households with any broadband service is used as a predictor, while in model 2 the share of households with high-speed broadband is used as a predictor. As expected, population density is one the strongest predictors, with the probability of being flagged as a BEAD cluster (holding other variables constant) dropping sharply as density increases (Figure 9). The probability also drops as education (share of population with bachelor’s degree or higher) and labor force participation increases. Conversely, the predicted probability increases with the share of White-only residents in the block group but is unaffected by the share of Hispanic population or the share of foreign born residents.

Their takeaways…

The BEAD program offers states a unique opportunity to address enduring gaps in high-speed broadband access. The program allows significant flexibility for each state to establish priorities and select grantees, recognizing that the drivers and characteristics of such gaps depend on local factors. While the roll-out of the program has been delayed by lapses in coordination between NTIA and state policymakers, BEAD retains bipartisan support and is on track to begin disbursing funds in early 2025.8

As a matter of first principles, appropriate program targeting is critical to ensure that BEAD funding primarily reaches the areas and residents in most need of support for high-speed broadband access. State policymakers will also need to factor in the end of the Affordable Connectivity Program (ACP), a key complement to BEAD that provided direct support for service subscriptions to a broad group of eligible households.9 The absence of ACP support weakens service demand among low-income households and thus raises concerns about the long-term financial viability of BEAD-supported networks. The large number of defaults in the FCC’s Rural Digital Opportunity Fund (RDOF) program, a comparable program designed to connect rural locations through reverse auction bidding, provides a cautionary note worth heeding.10

Based on the demographic analysis of BSLs preliminary designated by states as eligible to receive BEAD funding, we find that BEAD will primarily benefit low population density areas where adoption of high speed broadband is lagging, with limited expected impacts in other areas. A particularly important finding is that, after controlling for other demographic factors, the share of households in poverty is not correlated with priority BEAD areas in plans submitted to the NTIA. In fact, in one model specification, the probability of classification as a BEAD cluster falls as poverty levels increase. This finding raises concerns about whether BEAD will meet its mandate of reducing income-based disparities in high-speed broadband access. At the same time, the findings also point to vast differences in program targeting across states, suggesting that these impacts will be highly dependent on local contexts.

It is worth recalling the limitations of the analysis presented above. First, eligible locations will not necessarily receive BEAD funding. Rather, they represent the universe of locations from which network operators will selectively submit funding bids to state policymakers. Further, our data predates the BEAD challenge process, and preliminary analysis suggests major adjustments to location eligibility in some but not all states following the challenge process. This is an issue we address in a forthcoming policy brief.

Overall, these preliminary findings call for continued monitoring of the demographic factors that correlate with the allocation of BEAD funds. To this end, the availability of data on a timely basis and full transparency in decision-making from federal and state policymakers remain critical. Given the historic level of investment in broadband in the IIJA of 2021, it is imperative to continue promoting independent evaluations of the funding priorities and the long-terms impacts of the BEAD program.