Defining Broadband Coverage: It's Complicated

Broadband — defined as internet service at speeds of at least 25 Mbps download/3 Mbps upload — allows individuals, businesses, and communities to participate in the digital economy. Uneven access to or use of broadband results in a digital divide. But how serious is the digital divide across the country and in the Fifth District? It depends on who you ask.

There are numerous data sources that measure broadband access or use in the U.S., each with different methodologies, limitations, and implications for policy efforts to close the digital divide. The two most-cited sources of data on this issue are the Federal Communications Commission's (FCC) Fixed Broadband Deployment data (Form 477) and the Census Bureau's American Community Survey (ACS). However, these sources provide different estimates of broadband coverage. According to the FCC, 99.9 percent of Americans have access to one or more broadband provider; in contrast, the ACS shows that 78.1 percent of American households have a non-cellular broadband internet subscription. To further complicate the issue, neither measure fully aligns with other measures of internet usage from the Pew Research Center and Microsoft's Airband Initiative, for example.

These differences are even starker at the local level: According to the FCC data, all Fifth District counties have universal (or near-universal) broadband access, but according to the ACS data, broadband subscription rates were as low as 28.9 percent in some counties. (See maps below.)

The Heart of the Matter: Definitions and Methodology

What accounts for these discrepancies, and how do they shape our understanding of the digital divide? First, much of the difference stems from varying definitions of broadband coverage that are based on three distinct indicators: access, subscriptions, and usage. The FCC measures how many Americans have access to a broadband provider should they elect to subscribe. More specifically, the FCC estimate of broadband connectivity measures the number of Americans living in census blocks where at least one other household has a fixed broadband subscription. In contrast, the ACS collects data on whether the household has a subscription to a broadband provider. Rates of subscription to broadband can diverge from access, particularly in rural areas, since the decision to purchase a home internet subscription is also linked to factors such as consumer income and educational attainment. In addition to the FCC and ACS datasets, there are other datasets that measure broadband usage. These data capture the extent to which individuals and households actually access the internet, depending on the technology and services available to them. In particular, the Pew Research Center has authored reports since 2000 that summarize survey data on internet usage by demographic groups. Similarly, Microsoft's Airband Initiative reports include indicators of broadband usage derived from proprietary data of how users are accessing their software.

Second, the methodologies used to collect broadband data have implications for what the data show. To assemble its fixed-broadband deployment data, the FCC pulls information from a form that all broadband providers submit to the agency biannually (Form 477). In this form, facilities-based providers report fixed-broadband deployment at the census-block level. However, by this methodology, when a provider offers service to one household in a block, the entire block is considered covered. Further, Form 477 instructs providers to include census blocks on their list of service areas even if they do not currently provide service in the area, as long as they could provide service "without an extraordinary commitment of resources."

This methodology has led to overestimation of the number of individuals with access to a broadband provider. For example, in comments to the National Telecommunications and Information Administration regarding improving the accuracy of FCC data, the nonprofit Connected Nation detailed results from a study they undertook in Walton County, Florida, to verify FCC broadband availability. By cataloging all telecommunication infrastructure in the county, the organization found that the FCC's data overestimated access by 27 percent. By the FCC's own admission, because "[t]he calculations … treat every location [in a census block] as having service, [they] may … over-estimate broadband coverage, particularly in areas with large census blocks." This is particularly relevant for rural parts of the country, where census blocks can span hundreds of square miles.

While the FCC estimates make technical assumptions that likely lead to overestimates of broadband availability, they do provide detail on providers' speed and utilized technology. This contrasts with the ACS, which asks if a member of the household can access the internet using a "broadband (high speed) Internet service such as cable, fiber optic, or DSL service installed in this household." The question notably does not allow respondents to indicate what speed their subscription provides or which of the listed technologies they utilize. Although another option asks if respondents access the internet using a "cellular data plan for a smartphone or other mobile device" or a "satellite Internet service," both options are separate from language of "broadband (high speed)" and therefore could refer to any other type of low-speed internet connection.

This lack of specificity has consequences for how we understand broadband access. If a household has internet access at below-broadband speed, they may still self-identify in the ACS as having broadband internet service, even though that is technically not true. Further, as elucidated below, the specific technology a household uses has implications for how often they use the internet and their satisfaction with their service. Some technologies provide services with greater likelihoods of delay and service interruption than others, even when technically providing the same speed. Therefore, while the ACS estimates likely capture households' self-perceived internet connectivity more accurately than the FCC figures, they lack the specificity needed to understand households' real access to broadband. To compound this issue of specificity, ACS estimates for geographies with fewer than 65,000 residents (including census tracts, census blocks, ZIP codes, and some rural counties) represent a five-year moving average, making it difficult to understand the current state of broadband access.

Microsoft's Airband Initiative has attempted to rectify the lack of specificity about speed by estimating the portion of Americans who connect to the internet at speeds of at least 25 Mbps/3 Mbps to use Microsoft programs. By their estimates, fewer than half (49 percent) of Americans utilize the internet at these speeds. Of course, these data have limitations as well. As with the ACS estimates, Microsoft's figures are aggregated and therefore do not provide information on the technology that individuals are using to access the internet. As another limitation, the estimates are derived from internal Microsoft data. Non-users of Microsoft products are not included in the company's dataset, and it is unclear how or if Microsoft reconciles this limitation.

Complicating Factors: Reliability and Consumer Satisfaction

There may be other reasons to believe that current broadband estimates do not capture the full picture. Even if consumers have access to services through a broadband provider, they may not subscribe if they do not perceive that the connection is high quality. The mere presence of service does not necessarily translate to use.

According to a 2019 Morning Consult survey, the broadband service characteristic that Americans believe is most important is dependability. One technology that rates lowest on dependability is satellite technology. In a 2018 report measuring performance by different broadband technologies, the FCC found that out of all technologies, satellite internet has the greatest variation between peak-hour (7 p.m. to 11 p.m. Monday-Friday) and off-peak-hour speeds. This higher risk of connection delays has repercussions for consumers' satisfaction and likelihood of adoption. Indeed, a 2018 BroadbandNow survey found that users of satellite internet are the least satisfied with their service out of all internet users. In fact, the study found that in 2018, consumers who switched internet service had higher satisfaction than those who did not switch — except for purchasers of satellite internet service, who became less satisfied.

This matters for estimations of broadband access because the FCC data include satellite internet, which requires significantly less infrastructure but provides less desirable service options. Indeed, when one excludes satellite technology, the FCC's estimations of internet access in Fifth District counties begin to resemble patterns in the ACS subscription data. (See map below.) When satellite technology is excluded, broadband access in the FCC data falls to below 25 percent in 10 counties (nine of which are rural ¹), indicating that satellite technology is the only broadband technology offered for the vast majority of residents in those areas. Four of these counties (King and Queen, Surry, Brunswick, and Cumberland counties, Virginia) are also among the 10 counties with the lowest subscription rates as measured by the ACS. This indicates that access to a broadband provider does not guarantee the absence of a digital divide — the quality of available broadband services matters as well.

Conclusion

Broadband data from the FCC, ACS, and other sources paint conflicting pictures of the digital divide in the U.S. and the Fifth District. These differences spring from how broadband coverage is defined – in terms of access, subscriptions, or utilization – and how data are collected. Despite their respective limitations, the data sources discussed above can add value to ongoing policy conversations and decisions about broadband investment – but, like all data, they must be analyzed with a recognition and understanding of their constraints.

For additional information about broadband coverage in the U.S. and the Fifth District, see our Broadband topic.

Have a question or comment about this article? We'd love to hear from you!

Views expressed are those of the authors and do not necessarily reflect those of the Federal Reserve Bank of Richmond or the Federal Reserve System.