In the spring of 1936, in the middle of the Great Depression, the US Congress passed the Rural Electrification Act (REA), which helped to bring electricity to rural America. At the time about 90% of homes in urban and nonfarm rural areas had electricity, while only 10% of farms were connected. By 1960, that electricity gap was mostly closed, with electricity access nearly uniform across rural and urban areas. The REA is now hailed as one of the key successes of President Roosevelt’s New Deal program.
“Broadband internet is the new electricity,” President Biden said, when pitching his American Jobs Plan in March 2021. Indeed, broadband internet is increasingly recognized as a necessary, foundational part of American life, just like electricity or water. The experience of the pandemic highlighted how broadband was not only a key part of people’s livelihoods, but in some cases was the difference between life and death, with access to basic amenities, essential public health information and digital workplaces being available only to the connected.
With the common rhetoric comparing the REA to the current efforts to close the digital divide, it’s important to recognize that the landscape of the electrification divide in 1936 was drastically different than the landscape of the digital divide today. While only 10% of farms had electricity in 1936, compared to 90% of non-farm areas; 29% of rural areas have access to broadband today, compared to 67% in non-rural areas. When looking at absolute population, however, around 22 million of the digitally disconnected live in rural areas, while 4.5 times that—98 million—live in urban and suburban places.
Our past work, Uneven State of the Union, examined the digital divide across all 50 states, and focused primarily on broadband usage at the county and state level, with a racial and socioeconomic lens. A major finding of that analysis was that the true cost to close the broadband gap is closer to $240 billion—a multiple of the $65 billion allocated in the infrastructure bill.
While this budget is historically large, it is also limited in what is actually needed to bring broadband to every home in America. Given that this may be the best shot we have in closing the digital divide, the allocation strategy must shift away from covering the most miles to covering the most people.
Prioritizing People over Miles: The Danger of Stumbling into Digital Redlining
The current allocation strategy for the funding from the infrastructure bill risks exacerbating the racial digital divide. Rather than drawing parallels to the REA, a better historical lesson may be found in a different New Deal program: the National Housing Act of 1934. In short, faced with a national housing shortage, the federal government began a program to increase the country’s housing stock through making housing and mortgages more affordable. The Act created the Federal Housing Administration (FHA), which insured home mortgage loans, encouraging lenders to make more home loans. However, the FHA furthered segregation by refusing to insure mortgages in and around Black neighborhoods, and also refused to subsidize builders if they planned to sell homes to Black families. In other words, the policy was very successful at boosting homeownership rates for white people but had the opposite impact on Black and immigrant families. The impact of these policies—now known as “redlining” for the red color showing Black neighborhoods on the loan maps—is still felt today. The racial wealth gap stubbornly persists, and predominantly Black and brown neighborhoods still suffer from disinvestment—including in the broadband space.
In the map below of Detroit, we compare the historically redlined areas against internet usage. It is apparent that nearly a century later, we are seeing that the neighborhoods that did not have access to the wealth-building opportunities of mortgages do not have equal access to the opportunity of fast internet access today.
Unlike the redlining of the past, digital redlining is not based explicitly on race, but rather on calculations of whether investing in building out high-speed internet will be supported by the demand in an area. Investing in fiber is expensive, and ISPs tend to invest first in wealthier neighborhoods, while leaving poorer communities either without access or with slower legacy systems. Yet, poorer communities still pay the same for subpar service. Multiple studies show that areas on the wrong side of the digital divide are the same areas that were redlined in past.
The current administration stated its commitment to end digital redlining, and the FCC is kicking off a process with the goal of ending digital discrimination. Despite these good intentions, the current allocation process for broadband funding risks exacerbating digital redlining.
Part of the Infrastructure Investment and Jobs Act (IIJA), the BEAD program is the largest federal infrastructure program to date aimed at closing the digital divide in the US. BEAD allocates $42.5 billion to states to fund broadband network deployments, prioritizing unserved areas: areas that lack networks reaching 25 Mbps download/3 Mbps upload speeds. Underserved areas lack service above 100 Mbps download/20 Mbps upload speeds and are second in priority. A majority of the funding will be allocated by determining the total number of unserved locations in a state relative to the national total.
While at face value prioritizing the unserved areas may appear to be a logical way to sequence the funding, the policy tilts funding toward covering miles over covering people. This approach risks leaving many urban and suburban communities behind, while favoring rural areas with fewer people. In the chart below, for example, we plot the 10 largest cities in the US, with a line at the 25 Mbps download speed, delineating the border between the unserved and underserved. It is apparent that in the 10 largest cities, there are very few zip codes with unserved populations, while the bulk of lower-income, less white zip codes are found in the underserved areas. For those in the underserved areas, speeds are simply not adequate to meet current bandwidth needs for learning and working effectively online. Less white zip codes in the US face both an affordability and an accessibility challenge.
Explore your own city or state in the interactive below, which charts the dynamics of unserved and underserved communities and how these overlap with racial demographics and income levels.
1. Prioritize people over miles
The current funding mechanism fails to prioritize people over miles, and risks exacerbating racial and socioeconomic inequalities. The NTIA, responding to stakeholder feedback, is offering more leeway to states in the sequence in which they can use funding in unserved vs. underserved areas. However, they are still requiring states to determine that they have a plan to reach all unserved areas before using funds for highly populated underserved areas, and the funding allocation mechanism is based heavily on the number of unserved locations. State-level stakeholders would be wise to realize that the BEAD money will be more favored to unserved, primarily rural areas, and look to other funding sources, like the American Rescue Plan Act (ARPA) to fill the gap in underserved locations in cities and suburbs. While awaiting updated FCC maps, states should also be preparing themselves for the challenge process, through creating their own maps with the best data available. The end goal must be to get high-speed internet to as many people as possible. Stakeholders should pay close attention to where funding is going, and look not only at number of locations, but number of people served.
2. Ensure accountability in affordability programs
The infrastructure bill not only is focused on building out broadband and improving access, but also contains some key funding to improve affordability. Affordability and accessibility are closely intertwined, as ISPs are more likely to prioritize investments in areas where they believe enough households will sign up for internet and help recoup their investment.
The Affordable Connectivity Program (ACP) passed as part of the infrastructure bill, offers lower-income households a $30/month subsidy to access broadband. The Biden administration also recently secured commitments from 20 internet service providers to cut prices and improve speeds for those receiving the ACP benefit. Advocates and policymakers would do well to track which ISPs are following through on these public commitments.
3. Reliable, verified, and timely data is imperative to understanding the true state of America’s digital divide
Policymakers from opposite sides of the aisle, advocates, and the FCC chair herself agree that the current FCC maps are inadequate. While new, more geographically detailed maps are in progress, they are not expected to be released until late 2022 at the earliest. States and other stakeholders will have the opportunity to challenge the veracity of these maps, and the final outcome of the areas mapped as unserved will determine what portion of funding is allocated to each state. State and local practitioners would do well to get their own cartographical house in order ahead of the FCC release, working with local groups and stakeholders to ensure they are not overlooking unserved and underserved areas. Given the history of the FCC’s mapping errors, it is no surprise that several states and cities are moving ahead with their own mapping efforts. For organizations looking for a source for mapping, our interactive, which maps the affordability and accessibility challenges in states and cities at the zip code level, offers a solid baseline to begin planning.
As this funding is spread over several years, and is enhanced by other programs and funding sources, timely updates to mapping will be necessary to track progress and keep a close eye on whether the current funding allocated will be adequate to close the digital divide.
Percent of households that can afford the cheapest internet
Our calculation for the ability to afford internet for each zip code included in our interactive leverages data from two sources: BroadbandNow and the US Census Bureau’s 2015-2019 American Community Survey (ACS). To begin, we multiplied the percentage of households within ten income brackets, ranging from households making less than $10,000 annually to households making more than $200,000 annually, by the total number of households in a zip code. This gives us the number of households within each income bracket in each zip code. Next, we used BroadbandNow data on the monthly cost of the cheapest internet-only plan available in a zip code, and multiplied it by 12 to estimate the cheapest annual cost for internet for a household in each zip code.
In our research, we found a target of spending no more than 2% of income on entry-level broadband services annually, as communicated by the United Nations Broadband Commission for Sustainable Development. Using this target, we calculated a breakeven annual household income needed in each zip code with data on the annual cost of the cheapest internet-only plan by dividing the cost of the internet plan by two percent. For example, a zip code in which the lowest priced monthly internet costs $69.99 equates to yearly spending on internet of $839.88. For a household to spend 2% or less on internet in this zip code, they would need to make $41,994 or more per year.
Finally, using data on the number of households in a given zip code and the percentage of households in that zip code that fall within ten different income brackets (ranging from less than $10,000 annually to more than $200,000 annually), we can create an estimate of the number of households in the zip code that cannot reasonably afford the cheapest internet in their zip code. Using the example above in which a household must make $41,994 annually to afford internet, all households in the income brackets that make less than $10,000, $10,000 to $14,999, $15,000 to $24,999, and $25,000 to $34,999 cannot reasonably afford the cheapest internet. An assumption was made that the distribution of income within a bracket was perfectly even. This allows us to say that (41,994-35,000/49,999-35,000) = 46.63% of households in the $35,000 to $49,999 bracket cannot afford internet. From this, the number of households in each of these income brackets—with only 46.63% of households in the $35,000 to $49,999 included—are aggregated and divided by the total number of households in the zip code to calculate the percentage of households that cannot reasonably afford the cheapest internet. To create our final metric that measures the percentage of households that can afford the cheapest internet, we subtracted the percentage of households that cannot reasonably afford the cheapest internet from 1.
Defining unserved, underserved, and served zip codes
In the White House’s Building A Better America guidebook, the administration outlines the order of priority for funds to be administered to zip codes regarding equitable access to sufficient broadband. The administration states that “the first priority for deployment is for providing broadband to projects that primarily reach unserved locations (those below 25/3 Mbps), followed by those that primarily reach underserved locations (those below 100/20 Mbps), and then serving community and then serving community anchor institutions. Using data on average download speed from BroadbandNow, we characterize zip codes with average download speed less than 25 mbps as “Unserved,” zip codes with average download speed less than 100 mbps as “Underserved,” and zip codes with average download speed greater than 100 mbps as “Served” (referred to by the White House as “community anchor institutions”).
Defining urban, suburban, and rural zip codes
We leverage the US Department of Agriculture’s Rural-Urban Commuting Area (RUCA) codes to define our data as urban, suburban, or rural at the zip code level. More granular detail on RUCA codes can be found described in the “Data Sources and Defining Data” section below.
Zip codes that have a primary RUCA code of 1 are considered urban. These zip codes, as defined by the USDA, lie at the center of a metropolitan area. Zip codes with a primary RUCA of 2, 3, and 4 are considered suburban, as they are either directly outside of the metropolitan center and commute into it (codes 2 and 3), or are at the center of a micropolitan area (code 4). Zip codes with a RUCA ranging from 5 to 10 are all coded as rural.
If a zip code has a RUCA code of 99, it is undefined by the USDA. In these cases, and other zip code cases in which RUCA is blank, we defined a zip code as urban if its population density is above 500 people per square mile, suburban if it is between 100 and 500 people per square mile, and rural if it is less than 100 people per square mile.
Defining Cities in our city interactive
To create our list of cities for the City Interactive, we used a crosswalk from FIPS County ID to the Census Bureau’s defined Core Based Statistical Areas (CBSAs). Because of the methodology, not all zip codes included in the City Interactive are defined as urban, if they fall into a county that matches a CBSA but are on the outskirts of that county.
Data Sources and Defining Data
Sources: BroadbandNow, Microsoft, American Community Survey 2015-2019, U.S. Department of Agriculture.
Average Download Speed: BroadbandNow collects average download speed, measured in megabits per second, from the website Measurement Lab. The final average download speed per zip code is a rolling 12 month average of users testing their download speed on the website between November 2019 and October 2020.
Lowest Priced Terrestrial Broadband Plan: BroadbandNow collects pricing data on the cheapest available internet-only plan that provides broadband speed (25 mbps download/3 mbps upload) as of the third quarter of 2020. This data comes from publicly-available sources from over 2,000 sources across the country.
Broadband Usage: Data is collected by Microsoft and is as of October 2020. This metric measures the percentage of internet users in a given zip code that use internet above broadband speeds (25 mbps download). Microsoft estimates this metric by combining data from multiple Microsoft services: every time a device receives an update or connects to a Microsoft service, Microsoft is able to estimate the throughput speed of that machine. It can then determine the zip code level location of the data via reverse IP. For the purposes of these interactives, Digital Planet calculates the percentage of population below broadband speeds by subtracting Broadband Usage from 1. Then, we multiply the estimated population of the zip code, taken by ZCTA from the 2015-2019 American Community Survey (ACS), by the percentage below broadband speed in the zip. This gives us the population of people in each zip code that are using internet below broadband speeds.
Racial Data: The 2015-2019 ACS has estimates on the racial profile by ZCTA. Digital Planet aims to measure the racial digital divide by using estimates of the percentage of a zip code that is White and non-White using this data.
Total Population: The 2015-2019 ACS has an estimate of the total population by ZCTA.
Percentage of Households in various Income Brackets: The 2015-2019 ACS estimates the percentage of households in a zip code that fall into ten ranges of annual household income. These brackets include income less than $10,000, between $10,000 and $14,999, between $15,000 and $24,999, between $25,000 and $34,999, between $35,000 and $49,999, between $50,000 and $74,999, between $75,000 and $99,999, between $100,000 and $149,999, between $150,000 and $199,999, and income greater than $200,000.
Number of Households: This is an estimate of the number of households in a given ZCTA. We use this statistic in the calculation for the percentage of households that cannot afford internet. We also this statistic to create a ratio of the number of people per household in a given zip code; multiplying this by the total number of households and, subsequently, the percentage of households that cannot afford internet gives us the total population in a zip code that cannot reasonably afford the cheapest internet.
Population Density: Population density is calculated by taking data on the total population in a zip code from the 2015-2019 ACS and dividing it by the land square miles per zip code from the 2010 US Census. It is interpreted as the number of people per land square mile.
Primary RUCA Codes, 2010: Defined by the US Department of Agriculture, primary rural-urban commuting area (RUCA) codes give us the most granular data on the primary flow of the population on a zip code level. The level of RUCA codes used in our urban-suburban-rural dummy variable delineates metropolitan, micropolitan, small town, and rural commuting areas. A value of 1 means that the zip code is within the core of a metropolitan area. Values of 2 and 3 represent zip codes that are within an urban cluster, but commute into the metropolitan center. Values of 4 and 7 represent the core of a micropolitan area (population between 10,000 and 49,999) and the core of a small town (population between 2,500 and 9,999), respectively. Values of 5 and 6 commute into the core of a micropolitan area, and values of 8 and 9 commute into the core of a small town. Finally, a RUCA value of 10 dictates a rural area, in which those who commute do not have primary flow to any RUCA zip codes of 1, 4, or 7.