In this day and age, when someone exclaims something to the effect of “The government really ought to do something about this issue,” the reality is that it probably already is. The so-called “digital divide” is one such issue.
The digital divide refers to the gap between people with and without access to high-speed broadband Internet service. With state and local governments essentially forcing much our daily lives online during this crisis, from work to entertainment, it is only natural for the digital divide to be top of mind. Again, however, an issue’s newfound prominence before voters and policymakers does not necessarily mean nothing is being done—or that more should be done.
Concern for disparities in Internet access predate this crisis and as a consequence, several programs targeting the digital divide already exist. In fact, according to a June 2017 report by the Department of Commerce, there are at least 17 different programs across six different federal agencies that provide some kind of subsidy toward broadband infrastructure development alone. In total, there are at least 28 programs across seven agencies aimed at addressing the digital divide and that “may fund projects involving broadband infrastructure, adoption, access, planning or research.” This is before considering countless state and local programs aimed at the same objectives.
Most federal broadband spending goes through programs administered by the Federal Communications Commission (FCC) and the U.S. Department of Agriculture (USDA). According to the Congressional Research Service, these programs spent a combined $9.1 billion in 2018. Over 90 percent of the spending came via FCC administration of the Universal Service Fund (USF). USF funding is largely automatic and not at risk of expiring on an annual basis. Per a December 2019 Congressional Research Service report:
USF programs are not funded via annual appropriations, but rather from fees the FCC receives from telecommunications carriers that provide interstate service. The FCC has discretion to spend these fees without congressional appropriations.
If Congress does nothing, billions of dollars will still be poured into expanding broadband service. What’s more, the FCC recently finalized a plan to ensure that more USF broadband funding is targeted at truly unserved areas, directing the bulk of the new $20.4 billion Rural Digital Opportunity Fund to this effort over the next decade.
The FCC highlighting the fact its new program targets, as Commissioner Brendan Carr put it, “truly unserved communities,” illuminates one of the many problems with existing government broadband programs: overbuilding.
Overbuilding occurs when subsidies create a new competitor in a market where private investment has already provided service. Dr. Mark A. Jamison, a visiting scholar at the American Enterprise Institute, explained the overbuilding problem in testimony before the Senate last year:
It is normal in competitive markets for rivals to duplicate each other’s infrastructure. This duplication is important for rivals to be able to compete for customers and for companies to test alternative services for the future. This duplication of infrastructure can be wasteful at best, and perhaps even destructive, when the government is subsidizing one of the providers.
The distinction between private-market competition and overbuilding is critical. The destructive effect to which Dr. Jamison refers occurs for a few reasons when government is involved. First, private services that aren’t backed by subsidies have a hard time competing with subsidized entities that can effectively socialize their risk. With the backing of government, subsidized providers have much greater leeway to undercut existing private providers. Overbuilding, therefore, is no boon to competition or consumers because private providers will eventually be driven from the market.
In addition, overbuilding causes private services and investors to be hesitant in deploying future networks. This means Americans in unserved regions that are potentially profitable go needlessly without service until government steps in, ultimately wasting tax dollars solving a problem largely caused by government itself.
Overbuilding is not a hypothetical problem. FCC Commissioner Michael O’Rielly has repeatedly expressed concerns about overbuilding across federal broadband programs, not just those overseen by the FCC. Last month, Commissioner O’Rielly wrote a letter to the USDA Rural Utilities Service (RUS) Administrator, expressing concern that new qualifying thresholds for RUS’s ReConnect program will inevitably lead to overbuilding. O’Rielly wrote:
RUS [has] decided to abandon its sound decision to limit 100 percent grants to areas that were 100 percent unserved, and instead lower an area’s unserved threshold to 90 percent. Rather than targeting grant money to those most remote and expensive to serve areas, this revised threshold will likely result in upgrading service in lower cost areas that are not in need of broadband subsidies and leaving the hardest to reach areas without service.
Of particular concern during the COVID-19 crisis and the resulting government stimulus interventions is that increased broadband subsidies and subsequent overbuilding problems were results of stimulus spending after the Financial Crisis of 2007-2008. Dr. Jamison explained in the testimony referenced above:
Other federal subsidy systems in the past have not been as focused on eliminating subsidies for overbuilds. For example, Jeffrey Eisenach and Kevin Caves examined the effectiveness of the $2.5 billion provided in the American Recovery and Reinvestment Act (ARRA) to the Rural Utilities Service (RUS) for the ARRA created Broadband Initiatives Program (BIP). Focusing on three case studies, Eisenach and Caves found that “more than 85 percent of households in the three project areas are already passed by existing cable broadband, DSL, and/or fixed wireless broadband providers. In one of the project areas, more than 98 percent of households are already passed by at least one of these modalities.” This was a waste of taxpayer dollars, so much so that the study estimated the cost of providing service to a home that had no service before was $30,104 if existing coverage by mobile broadband providers is ignored, and $349,234 if mobile broadband coverage is taken into account.
In short, even a decade ago, stimulus spending on expanding broadband service ended up going to areas where virtually all households were already being served—and since then billions more have been spent annually by non-stimulus programs to expand service.
Naturally this raises another point not considered by those who instinctively back increased government broadband spending: Just how many people in the United States are truly unserved? Will Rinehart, formerly of American Action Forum and now with the Center for Growth and Opportunity at Utah State University, took this question head on in 2018. Rinehart’s analysis reinforces the old saying, “There are three kinds of lies: lies, damned lies, and statistics.”
Rinehart found three critical problems with the current approach to broadband subsidies. The first two relate to how broadband service is defined.
Broadband and Internet service are not interchangeable terms. As the FCC notes, “The term broadband commonly refers to high-speed Internet access that is always on and faster than the traditional dial-up access.” High-speed, of course, is a relative and subjective measure and how one defines it can radically alter the picture of broadband service penetration. Rinehart explains:
In 2015, the Federal Communications Commission (FCC) increased the technical threshold for Internet to be considered broadband from 4 megabits per second (Mbps) download and 1 Mbps upload to a new standard of 25 Mbps download and 3 Mbps upload, sometimes abbreviated as 25/3. By increasing the limits for the definition of broadband, the number of regions with broadband service decreased. This shift is important, as it effectively expanded the number of areas that are the focus of federal policies. In the 2018 Broadband Deployment Report, the FCC noted, “over 24 million Americans still lack fixed terrestrial broadband at speeds of 25 Mbps/3 Mbps.” … The definition of broadband matters. For example, if the FCC were to shift the broadband speed threshold slightly, to 24 Mbps download and 2 Mbps upload, then nearly 1.5 million more people would have access to broadband. If the FCC lowered the standard to 20 Mbps download and 2 Mbps upload, then another 1.3 million people would have broadband.
Further blurring the picture is what kinds of technology are included in defining broadband Internet service. As Rinehart also notes, satellite-based broadband services were not included in the 2018 Broadband Deployment Report:
When satellite is included in the broadband mix, official statistics jump up. Just over 14 million Americans don’t have access to broadband-speed Internet when satellite is included—a drop of over 10 million people, translating into 95.6 percent coverage of people in the United States instead of 92.3 percent.
Satellite’s role is vital to consider in addressing the digital divide. As CEI noted earlier this year, SpaceX’s Starlink project has thus far launched nearly 300 satellites of its initial 1,584-satellite network and has plans to launch as many as 42,000 to beam high-speed Internet directly to consumers from space. As with any industry, some firms are more successful than others, but satellite broadband remains a viable market. As reported late last year, SpaceX may soon face competition from Apple in the satellite broadband business.
Satellite broadband cannot be excluded from the picture going forward because, quite simply, it has the potential to render much of the conversation regarding the digital divide moot. While the initial costs are high, the marginal cost of connecting a new customer to a satellite broadband network is significantly lower than that of connecting truly unserved households via entirely new terrestrial infrastructure such as cable/fiber lines and cellular towers.
Finally, Rinehart illuminates a “lead a horse to water” problem with broadband subsidies. Current policy is based primarily on providing broadband service access under the premise that such access promotes all sorts of positive economic outcomes. However, Rinehart demonstrates that the mere availability of broadband service has no statistically significant impact on several metrics of economic health:
[T]he percentage of the population with access to 25/3 broadband doesn’t explain the unemployment rate, median household income, the change in employment, or the rate of population change in rural regions. An analysis with the 4/1 standard similarly turns up no connection. This analysis calls into question whether the 25 Mbps download and 3 Mbps upload metric, the older 4 Mbps download and 1 Mbps upload metric, or any availability standard for that matter, provides a reasonable understanding of the underlying economics of rural communities.
To be clear, none of this is to say that access to broadband cannot have a positive economic impact. What matters, Rinehart’s study concludes, is if Americans actually use it:
[A]doption explains economic trends four times better than the 25/3 threshold. … Internet access [is] correlated with economic growth, but only when people adopted broadband.
The data on access versus adoption is illustrative of the broader point regarding government broadband subsidies: Simply spending more taxpayer money is, at best, unproductive and likely counterproductive.
The federal government has spent and continues to spend billions of dollars on broadband subsidies on an automatic, annual basis in addition to untold state and local expenditures. The government has additionally appropriated billions of dollars toward broadband subsidies in recent years through economic stimulus legislation. These subsidies often have the effect of begetting more subsidies through overbuilding, driving out and deterring new private investment in potentially profitable, yet underserved areas. The necessity of any form of broadband subsidies is further obscured by subjective, incomplete, and somewhat arbitrary measures of what constitutes broadband service, while new technologies undercut the case for government involvement altogether. Finally, additional government spending on broadband infrastructure has proven economically irrelevant absent sufficient consumer utilization.
The COVID-19 crisis is an unprecedented emergency that has produced record government spending to salvage the parts of the economy that cannot move online. The impulse to expand Internet access in this situation is therefore rational. However, simply throwing more money at extant government broadband programs in any forthcoming stimulus legislation would amount to nothing more than expensive veneer and impede the long-term effort towards closing the digital divide.