Although broadband internet access is a functional prerequisite for modern civic and economic life, significant inequities in broadband access remain. Commercial, government, and healthcare services are all offered online — but accessing these offerings requires reliable and affordable broadband access. The Federal Communications Commission’s most recent report on the state of broadband in America finds that millions of Americans lack access to reliable broadband service, and that millions more must procure broadband service from a local monopolist.
U.S. policymakers have deployed a range of programs to address these concerns. For example, Congress recently allocated over forty billion dollars to the Broadband Equity Access and Deployment (“BEAD”) Program in order to ensure ubiquitous broadband access across the nation, and fourteen billion dollars to the Affordable Connectivity Program (“ACP”) to address affordability-related concerns (though ACP’s funds have since been depleted, forcing the FCC to wind-down the program).
Before BEAD, the Federal Communications Commission (“FCC”) allocated over ten billion dollars to the Connect America Fund (“CAF”), a similar program aimed at ensuring broadband access in underserved areas, with a particular focus on rural areas. Specifically, the FCC offered CAF subsidies to internet service providers (“ISPs”) in exchange for deploying broadband infrastructure: A single ISP was eligible for the subsidy in each region without broadband access, and ISPs accepting the subsidies were required to satisfy certain regulatory obligations, including speed and cost requirements.
Given the scale and importance of policy interventions like CAF and BEAD, it is critical to assess their efficacy. Effective program review can help policymakers both evaluate the success of these programs and design new and better policy interventions that tackle our nation’s stubborn digital divide. In a peer-reviewed study, accepted to the 2024 ACM SIGCOMM Conference, we take one initial but substantial step in this direction. We find that the serviceability rate — i.e., the weighted fraction of CAF-funded addresses that are actually served by subsidized ISPs — is only 55 percent, dropping to as low as 18 percent in some states. Moreover, the compliance rate — i.e., the weighted fraction of CAF-funded addresses where ISPs offer service that satisfies the FCC’s program rules — is only 33 percent. In short, over two-thirds of addresses “certified” as served by the ISPs do not, in fact, appear to receive service that meets the FCC’s program rules.
These results offer an important measure of the extent to which ISPs have lived up to their promises under CAF, and the extent to which CAF has helped improve broadband access in its target regions. So long as policymakers continue to allocate funds to improve broadband access nationwide, empirically grounded analysis of these programs remains critical. Regulators should interrogate ISPs’ service claims, improve audit transparency, and implement stringent requirements to ensure better service quality and consumer value. And officials might even consider past compliance with funding programs when deciding how to award subsidies in the future.
Evaluating the Connect America Fund
We divide our evaluation of CAF into three questions. First, we determine whether ISPs genuinely offer broadband access at the addresses that CAF sought to serve. Second, we assess ISPs’ compliance with the rate and service standards specified for the program. And third, we evaluate the monopoly controls implemented by CAF. Specifically, because CAF creates regulated monopolists in its program areas — single subsidized ISPs in areas previously lacking service — we evaluate whether regulated monopolists offer consumers better value than unregulated ones.
Our measurements leverage the Broadband-Plan Querying Tool (“BQT”), a tool that we have built and that has become a part of the computer science’s internet measurement community. BQT can take, as an input, a dataset of street addresses and will return, as an output, the set of broadband plans (i.e., download speeds, upload speeds, and corresponding prices) offered by BQT-supported ISPs at each input location. It does so by mimicking the behavior of human users interacting with the ISPs’ websites.
We use CAF’s public dataset as our input: Subsidized ISPs were required to “certify” to the Universal Service Administrative Corporation (“USAC,” a non-profit entity designated by the FCC to administer CAF monies) the residential addresses they served under the program. Those certifications included descriptions of the broadband plans — speed and price — offered. In theory, the public dataset could be enough to evaluate CAF’s successes. Indeed, USAC itself relies on this data to evaluate ISPs’ compliance with CAF program rules. But both federal and state policymakers have raised questions about this self-reported data — and these concerns, alongside widespread reports about problems in CAF, led us to a more independent audit of the program. By combining BQT’s automated approach to broadband data collection with the millions of addresses in CAF’s public dataset, we can measure the efficacy of this policy intervention at a scale that is orders of magnitude greater than any comparable previous study.
Service Availability
We begin by exploring whether service is genuinely available to addresses certified as served by the ISPs through the CAF program. To answer this question, we calculate a “serviceability rate” — the percentage of addresses that are actively served.
We queried over 537,000 street addresses across 15 selected states (selected to account for a range of geographic conditions) and 4 selected ISPs (the top 3 ISPs, in terms of addresses served and subsidies received under CAF; as well as one smaller provider), to measure serviceability. (Our strategy for sampling these addresses across states and census block groups is detailed at length in our underlying study.)
We observed an aggregate serviceability rate of 55 percent — meaning that, based on our estimates, only 55 percent of these ISPs’ CAF address can receive service in these states. This finding is deeply concerning. It suggests that broadband access is still lacking for well over 40 percent of the addresses that CAF intended to serve (and that taxpayers spent over ten billion dollars to help bring online).
Disaggregated by ISP, we observe serviceability rates of 32 percent, 71 percent, 90 percent, and 84 percent for AT&T, Frontier, CenturyLink, and Consolidated Communications, respectively. That CenturyLink has the highest serviceability rate is one encouraging glimmer, given that it received the most CAF funding (1.84 billion dollars) of any ISP. But, by the same token, AT&T’s staggeringly low serviceability rate is especially concerning, given that it ranked second among all ISPs in terms of total funds received, and first in terms of addresses “served.”
We also analyzed the geospatial distribution of serviceability. We found that, in many states, areas further from large city centers tended to have lower serviceability rates — even though CAF was intended to support rural connectivity. Figures 1 and 2, for example, offer two visualizations of the serviceability rates in California and Georgia, respectively, where these trends are most striking.
In Georgia, for example, we see that communities on the outskirts of urban centers like Atlanta and Savannah show higher serviceability rates, whereas areas in less densely-populated regions evince lower serviceability rates. Likewise, in California, areas near San Francisco and Los Angeles exhibit higher rates than areas in California’s Central Valley, or in areas approaching the state’s northern border with Oregon. While we might ordinarily expect an ISP to deploy more service in dense urban centers, one purpose of CAF is to counter these prevailing market forces to ensure a more equitable distribution of broadband access. These persistent trends suggest that CAF has not completely succeeded in doing so (though, of course, the difference might have been even greater but for CAF).
Program Rule Compliance
Next, we explore program compliance by comparing the speeds and prices of offered plans to the FCC’s CAF program standards.
On price, we find that the prices charged by ISPs are well below the FCC’s price caps. This is good: ISPs meet the FCC’s rate requirements. We note, however, that the FCC’s rate requirements have been criticized as too lenient. The FCC is statutorily required to ensure that the rates charged are “reasonably comparable” to those charged in urban areas. And the FCC considers a price “reasonably comparable” if it falls within two standard deviations of the average urban rate. This standard means that consumers served through CAF may receive broadband value that is one hundred times worse than the median urban consumer (as we measured in a previous study).
On speed, CAF program rules require that ISPs offer download speeds of at least 10 Mbps. All four studied ISPs certify that they offer service that meets or exceeds this benchmark. For instance, AT&T has certified that it offers at least 10 Mbps to all the addresses in our sample. Our results, however, which draw upon the plans advertised to consumers on the ISPs’ websites, tell a very different story. We estimate that over five percent of the addresses served by AT&T can get, at most, 5 Mbps, and another five percent are offered AT&T’s “Internet Air” product, which does not guarantee any minimum speed. And this is on top of the 67 percent of addresses that do not receive any service at all.
Table 1, below, describes these results, across all four ISPs, in greater detail. Non-compliant plans are highlighted in red, and the “0 Mbps” entries indicate the percent of addresses that remain unserved.
Overall, our analysis indicates that 67 percent of CAF addresses in these states do not receive service from these ISPs that complies with the FCC’s program rules (including those addresses where service is not available at all). Again, these findings are deeply concerning: ISPs have accepted public subsidies on the condition that they offer broadband access meeting certain minimum service standards, but those ISPs are not living up to their end of the bargain.
Monopoly Regulation
Finally, we consider whether consumers at CAF-subsidized addresses receive better value than those consumers served by an unregulated monopolist. Our aim here is to quantify the efficacy of CAF regulations at curbing a subsidized monopolist’s monopoly powers. To do so, we compare the plans available at CAF addresses with those available at those neighboring addresses (i.e., within the same census block) where that same ISP (i.e., the one that received the CAF funding) operates as an unregulated monopolist (i.e., faces no competition and is not subject to CAF program rules).
We find that, in the majority of cases, the plans offered at CAF-subsidized addresses are the same as those offered at other, nearby monopoly-served addresses. Of course, we do not interpret these results to necessarily mean that the CAF program rules have had no effect: It may be, for example, that, in some areas, ISPs improved the offerings to all consumers in the area. Or, alternately, given how lax the CAF program rules are, it may be that the ISPs’ existing offerings already satisfied program rules and no changes were needed.
We also find that CAF-subsidized addresses receive better service in 27 percent of those census blocks where a single ISP acts as a monopolist, subject to program rules in some locations and unregulated in others. Moreover, the degree of the improvement is substantial: The median percentage increase in service quality (measured by download speed) is 75 percent, and the 80th percentile percentage increase is 400 percent. These results suggest that, where competition is lacking, monopoly regulation can help to improve consumer value overall, albeit inconsistently.
We can also compare the efficacy of CAF program rules to market competition. We do so by comparing the plans available at CAF addresses with those available at neighboring addresses where that same ISP faces competition. Here, our results are more evenly divided: In 37 percent of cases, the plans are the same across addresses; in 32 percent of cases, CAF-regulated plans are better than those subject to market competition; and in 31 percent of cases, competition is best at improving consumer value.
Implications and Conclusions
Our analysis sheds light on the effectiveness of CAF. Our findings tell a dim story: Over 40 percent of CAF-funded addresses appear to remain unserved, contradicting the ISPs’ own certifications. Moreover, even among those consumers who receive service, that service often falls short of the FCC’s minimum standards.
Our study is the first to comprehensively evaluate such a large policy intervention at this scale. In so doing, we not only call attention to the failures — and the (limited) successes — of the program, but we also emphasize the need for better frameworks for assessing program compliance and success. Given that the federal government is, through BEAD, on the verge of spending over 40 billion dollars to do what CAF tried to do — to ensure greater broadband access nationwide — the importance of such empirically grounded analyses is pressing. For one, regulators need to more thoroughly vet ISPs’ claims, and they should make their audits more transparent to the public and other policymakers. Indeed, state officials responsible for selecting BEAD grantees might consider past compliance with funding programs (including CAF) when deciding how to allocate these new funds. Moreover, such policymakers should consider implementing more stringent rate and service quality requirements to ensure that the regulated providers offer greater value to the communities served by these programs. And, finally, policymakers should remain committed to data-driven program development, adapting, as necessary, to the evidence that is uncovered.
In conclusion, our study emphasizes the need for rigorous oversight and empirical evaluation in broadband policy. Given the substantial public investments in a wide range of broadband subsidies, from CAF to BEAD to ACP, we deserve to know whether these programs are working as designed, and how they are being improved for the future.
The post <strong>Measuring Broadband Policy Success</strong> appeared first on Harvard Law Review.