This post is the first in a series of four posts, which first explores the current infrastructure bills, diagnoses the problem inherent in the digital divide, explores broadband as an infrastructure problem, and finally lays out an algorithm for action in broadband.
President Biden recently laid out an agenda to spend $2 trillion on infrastructure, $100 billion of which is targeted to “bring affordable, reliable, high-speed broadband to every American.” Where the White House lacks specifics, bills in both the Senate and the House of Representatives have already filled in the gaps. If something happens on infrastructure, broadband will be included.
Acolytes claim that massive broadband expenditures will close the digital divide, but all of the plans are unabashedly focused on infrastructure. In reality, only a small part of the digital divide will actually be solved by making broadband infrastructure available everywhere.
Importantly, policymakers need to recognize the distinction between the total expenditures to plug the broadband gap and the level of government-provided support to reach that goal. Internet service providers don’t need to fully rely upon government funds to provide service, as past experience shows, which means $100 billion is probably too high.
Moreover, leaders should understand that,
- Completely closing the broadband availability gap will be costly;
- The FCC needs to build better maps to target the unserved;
- All reasonable attempts should be made to ensure this group is connected first before gigabit service; and
- More stringent rules need to be put into place to ensure monies are properly directed.
This first post examines the proposals being discussed in Congress and explains why they aren’t going to solve the problem they have been tasked to plug. The focus on broadband infrastructure has taken our eye off the real problems, failures, and successes of past policies and lacks a clear vision about the digital divide. A sketch of what is needed is laid out in the final section of this series.
Broadband infrastructure spending proposals
Biden’s $100 billion for broadband is a telling target since the number fits nicely with the “The Accessible, Affordable Internet for All Act” or the Internet for All Act. Representative Clyburn, sponsor of the Internet for All Act, was reportedly crucial to getting the White House to increase their broadband funding fivefold, from $20 billion to $100 billion. Given the Representative’s leadership on this issue, one would expect the final plan that passes Congress would closely echo the Internet for All Act.
Under the Internet for All Act, the federal government would,
- Authorize $80 billion to deploy high-speed broadband infrastructure, giving $20 billion to the states and the other $60 billion to the FCC for a nationwide auction;
- Allocate $5 billion for low-interest financing of broadband through a new secured loan program;
- Establish a new office within the National Telecommunications and Information Administration (NTIA) to ensure efficient use of federal money;
- Extend the Emergency Broadband Benefit, created last year, which gives families a $50 monthly discount on their Internet plans or $75 for consumers on tribal lands;
- Direct the Federal Communications Commission to collect data on prices charged for broadband service;
- Conduct a biennial study to measure the extent to which cost remains a barrier to broadband adoption; and
- Provide over $1 billion to establish two new digital equity plans to help close gaps in broadband adoption and digital skills.
The bulk of the spending comes in the form of the $80 billion that is to be split between two programs. The states, US Virgin Islands, Guam, and US protectorates are slated to get $20 billion right off of the top. The bill is written so that states spend their allotment through a competitive bidding process, similar to the national version. The table below approximates the funding each state will receive, as a one-time payment, under the state funding portion of the program.
The remaining $60 billion of the Internet for All Act will go towards a nationwide competitive bidding process that involves at least two rounds of funding. Evident in its structure, the Internet for All Act is an infrastructure program aimed at gigabit service. If indeed the bill were truly about the digital divide, the two-step process would be swapped.
In the first part, funding will go to proposals that can commit to gigabit service or projects that connect unserved anchor institutions like schools and libraries with superfast broadband. Up to twenty percent or $12 billion has been reserved for those projects that can specifically commit to gigabit service in four years. Up to $12 billion could be spent in this first round, on gigabit service, even though there is no requirement that a project targets needy areas.
After this first round, one or more further rounds will be held to award the rest of the funds, which could be around $48 billion, to unserved areas, areas with low-tier service or mid-tier service, and unserved anchor institutions. In these subsequent rounds, priority will be given to unserved areas, which is defined as an area where at least 90 percent of the population has no access to broadband service with a download speed of at least 25 Mbps and an upload speed of at least 3 Mbps. However, the Commission and the states might further prioritize projects if the projects have at least 20 percent matching funds from non-Federal sources, expand access on Tribal lands, offer high speeds, quickly deploy, or expand access to persistent poverty counties, among other features.
Still, the unserved should be a priority, and only after a reasonable attempt has been made to address this group would it be appropriate for the process to move on towards other goals like gigabit fiber. To its credit, the House’s “Leading Infrastructure for Tomorrow’s America Act” or the LIFT America Act prioritizes the unserved, a feature that should be adopted in any final infrastructure package.
The cost of funding broadband infrastructure
While it is not explicitly mentioned, the $80 billion target in the Internet for All Act comes from a 2017 cost estimate of Paul de Sa, Chief of the Office of Strategic Planning & Policy Analysis at the Federal Communications Commission (FCC). de Sa estimated the total capital expenditures “to increase and accelerate profitable, incremental private-sector investment to achieve at least 98% nationwide deployment of future-proofed, fixed broadband networks.” Closing the broadband availability gap with fiber or cable would total $40 billion for 98 percent coverage, and another $40 billion will get the last two percent connected for a total of $80 billion.
The estimate is rooted in the unique economics of networks. Like all other investment decisions, new network investments are made when the project is cash-flow positive. Deployment decisions are thus highly sensitive to service-area economies of scale, which can vary greatly depending on the population density, the cost of building in an area, and the regulatory regime. Most of the costs of building broadband networks come upfront. Maintaining and operating the infrastructure are generally inexpensive. That means some areas where investments don’t make sense today might be able to develop a network if the startup costs are subsidized.
The de Sa cost estimate assumes that “approximately 14% of the ~160m U.S. residential and small-and-medium business locations lack access to [25 Mbps download and 3 Mbps upload] capable fiber-to-the-premise (FTTP) and/or cable service.” But the broadband market has changed significantly since the whitepaper was drafted. The total number of households in the U.S. without a cable or fiber provider in this speed tier stands at 9.3 percent as of December 2019. In other words, the current funding gap is sure to be lower than the 2017 forecast, tentatively in the range of $8 billion less.
Importantly, de Sa suggested that only the first $40 billion should be targeted (in 2017 dollars) because the last 2 percent of locations, which tack on another $40 billion, have unique characteristics. This is where the bills differ since the LIFT America Act only spells out $40 billion to close the broadband availability gap compared to the White House plan of $100 billion. The last two percent are the most costly group to connect. It costs just as much to connect the last two percent as it costs to connect the first 12 percent.
Adjusting de Sa’s estimates to 2021 dollars, it will cost on average $96,000 per location to build broadband infrastructure accessible for the last two percent. That adds up to a total cost of $43 billion for only the last two percent of locations. More important still, this group of locations, which I’ve dubbed economically stubborn, is unique because subscriber revenues won’t be sufficient to pay for yearly operational costs. In the world of finance, these projects will be cash flow negative on an operational basis, meaning they lose money in their operation.
As written, both bills tackle the locations currently without broadband up until the point where locations become economically stubborn. In this range, projects should be expected to have high initial capital costs that could be offset by funding, but they should also have the cash flow to maintain the operations once they are built. Both bills all but guarantee that economically stubborn projects won’t be funded after they are put in place because the funding methods assume that the network will be able to maintain operational costs once the initial investment is made. This isn’t a death blow to the bill. Rather, leaders need to be clear-eyed about the limits of getting the last 2 percent of locations connected.
Reverse auctions leverage private funds
Still, there is a deep disconnect between the work of de Sa and the current discussions over funding. Given that there is a gap in broadband expenditures, policymakers need to ask as a follow-up, should the federal government completely foot the bill, or are there mechanisms that can be used to leverage private money to get everyone a connection?
Traditionally, the FCC has relied upon reverse auctions, which solicit bids from multiple companies to provide broadband service in an area, to plug this gap. Both the Rural Digital Opportunity Fund (RDOF), which ended late last year, and the Connect America Fund Phase II (CAF II), which ended in 2018, were reverse auctions that provided federal support for unserved areas of the country. Moreover, language found in both bills suggests funding will occur through a reverse auction.
Reverse auctions make sense. Research from economist Sarah Oh finds that these auctions are typically the most efficient way to support broadband subsidies because they leverage private and public funds. Leaders need to embrace this. If there truly is an $80 billion hole, the government support levels to fill it will be much smaller than $80 billion. A lot of housework needs to be completed.
Assuming that the process proceeds like RDOF, the next auction will consist of timed rounds where bidders will be asked to provide service to an area, at a specified performance tier, in return for a specific support amount. The support amount is only a small part of the overall investment needed to build a network, but it signals the cost needed for a private entity to enter into a market. In each subsequent round, the support will drop in price, which will cause bidders to drop out. This reverse auction is known as a multi-round, descending clock auction because it descends to the lowest price for a service area.
RDOF intended to bring “high speed fixed broadband service to rural homes and small businesses that lack it.” At its close in late 2020, $9.2 billion in support was given out to 180 different groups which will bring service to over 5.2 million unserved homes and businesses, almost 99 percent of all locations targeted. Phase II of the auction, which has yet to be announced, has an additional $11.2 billion in support that will be allocated for the same purpose.
Originally, the FCC estimated that $16 billion would need to be given out, but the real amount came out to be nearly $7 billion under this mark because bidders were able to deliver service at a support level much lower than estimates first suggested. Since RDOF was technology-neutral, it selected the least costly structure for service, allowing high-speed wireless technologies to get support in areas where it is competitive.
A similar sort of result should be expected with a new auction. While it is an outcome that should be embraced, a political compromise will probably come in the form of the Internet for All Act since there is a push to build fiber. Wireless tech cannot provide gig service while fiber can, so the two-part system funds fiber first by supporting gig service in the first round and then runs additional rounds which would allow for tech neutral bidders. It is a political solution but a convoluted method to achieving ubiquitous broadband.
Still, a trend of declining support comes into clear focus when RDOF is considered alongside a similar auction, the Connect America Fund Phase II (CAF II). In 2014, CAF II estimated that support would roughly total $2,400 per location to serve the hardest to reach locations, but the actual auction secured support at only $2,000 per location. The RDOF continued that trend and pushed down support to $1,700 per location only four years later.
Given this trend, it remains an open question whether $80 billion, estimated four years ago, is the right amount to bridge the availability gap. RDOF is only now being distributed and the FCC hasn’t conducted any recent studies to recalculate the cost of filling the broadband availability gap or the support levels needed to bring service into an area. The maps on which those costs, and therefore an auction, would be based are only now being updated with more fine-grain data; a perennial stumbling block to proper targeting of monies. Additionally, there is nearly $40 billion sitting in the bank from a completely separate wireless auction that hasn’t yet been attached to any spending. There is still a lot of work that needs to be completed to ensure a successful auction.
The time component to rolling out the broadband infrastructure plan presents its own challenges. Once the bill is signed into law, the Commission and the states will have 18 months to get 75 percent of the monies out the door. While agencies shouldn’t be sitting on a pile of cash, the year and a half window will tend to favor those with knowledge of the FCC process and existing businesses for whom it is easiest to make a business case. The losers in this process will be new entrants into the broadband market seeking to serve the unserved. Even now, cities are lining up their infrastructure projects to get funded.
An injection of $80 billion should be expected to accelerate business plans that have already been in the works, instead of opening up new opportunities in places where networks would not have been built. None of the bills prioritize new networks despite this being critical to solving the broadband gap. While duplicate networks serving the same community should be avoided, priority should be given to new builds that can prove they are new via a certification process, as de Sa suggested.
Among all of the issues with the proposed bills, the most concerning is that they make no effort to limit duplicate efforts in a region. Overbuilds, as they are called, are new networks that are built in areas already serviced by an incumbent Internet service of the same type. Recent CGO research from economists Michael Kotrous and James Bailey found that cable overbuilds tend to have significantly lower average download speeds than cable broadband in blocks with no overbuild. Crowding out in these regions is a real phenomenon, recognized even by de Sa. For this reason, it would be best to implement a rule that limits overbuilds. The success of a broadband infrastructure plan hinges on its funding mechanism. Simply put, more groundwork is needed to ensure success.
A better way forward
With the Biden Administration has come a fresh look at policy and some are hopeful that the digital divide might receive the attention it deserves. As it currently stands, however, the $80 billion slated for broadband won’t be spent wisely.
A better path forward for closing the digital divide would mean first getting our house in order. Policymakers interested in effectively and quickly closing the digital divide should take four additional measures:
- First, the FCC should estimate the cost of support to get the unserved connected to broadband, built on up-to-date maps;
- Using new cost data, the FCC should oversee a reverse auction where unserved locations are prioritized first;
- As part of this auction, projects that would not have been executed under existing business plans should be prioritized; and
- Rules should be put into place limiting overbuilds.
At the end of the day, the broadband plan being discussed by the administration and in Congress is still an infrastructure plan, and a limited one at that since it does not tackle the cost of construction or permitting. To faithfully tackle the digital divide, a more comprehensive approach is needed. Without it, the funds meant for underserved areas will likely go towards projects that would have happened anyway. That will leave the underserved in the same position that they’re in now, disconnected from the country’s increasingly digital economy. The next three posts in this series lay out that approach, by explaining the nature of the problem, exploring broadband infrastructure, and finally laying out an algorithm for action.