Broadband adoption rates and gaps in U.S. metropolitan areas

The Brookings Institution
There is no question that the Internet is a huge boon to the economy and society, but maximizing its potential is only possible if all individuals are online. As a result, it is critical that policymakers closely track broadband adoption rates: the share of households with a DSL, cable, fiber optic, mobile broadband, satellite, or fixed wireless subscription.4

Until now, public, private, and civic leaders have frequently concentrated on broadband adoption at a national or international scale, looking at how rates vary across large segments of the population.5 However, new survey questions from the U.S. Census Bureau enable analysis at the metropolitan scale, creating new ways to measure and understand where America falls short in getting people online. This subnational approach is especially important because local and state governments play a lead role in guiding Internet policy, including infrastructure deployment, public outreach, skills development, and affordability programs.6

This brief uses 2013 and 2014 American Community Survey data to track current and changing broadband adoption rates at the metropolitan scale, while using a combination of other Census and Internet speed data to model what factors affect metropolitan adoption rates. In turn, the results of this analysis have clear implications for efforts to address the significant gaps in American Internet adoption.

Further information on the  analytical approach and methods is available in the appendix (see report PDF).

While 75.1 percent of American households had a broadband Internet subscription in 2014, there is enormous variation in U.S. digital connectivity across demographic groups and between metropolitan areas.

In 2014, more than 87 million households—or three-quarters of all households nationally—had a broadband Internet subscription, speaking to the importance of having a reliable, efficient connection to digital information networks across the United States.7 There’s little question broadband has evolved into the defining infrastructure of the 21st century.8

Yet despite its central role in promoting economic prosperity, broadband adoption remains highly uneven among specific groups, limiting opportunity for many people. For example, only 46.8 percent of households with incomes under $20,000 annually had a broadband subscription in 2014, compared to 88.8 percent of households earning $50,000 or more. Likewise, while 54.1 percent of individuals with less than a high school diploma had a broadband subscription, 91.5 percent with a bachelor’s degree or higher did. Relatively low adoption rates also appear among older age groups—64.5 percent of individuals 65 years and older—while those not in the labor force (69.7 percent) subscribe to broadband at marginally lower rates relative to the national average.

Broadband adoption also varies substantially across different U.S. markets, including the country’s 100 largest metropolitan areas. Although these areas tend to have higher adoption rates (77.8 percent) than the country as a whole (75.1 percent), these rates can still differ by up to 30 percentage points or more in some cases. Tech centers like San Jose (88.2 percent), Seattle (84.8 percent), and Boston (82.7 percent), for instance, far exceed the shares seen in Lakeland (64.1 percent), Greensboro (64 percent), and McAllen (58.1 percent). Table 1 includes the top and bottom performers across all metro areas, demonstrating the enormous range seen across the country. However, as revealed in Figure 2, the level of variation among broader geographical regions in the West, South, and East is less clear, suggesting local dynamics may be a more important indicator of broadband adoption.

Multiple factors—including higher levels of income, educational attainment, and telecommuting—all have a positive and significant effect on broadband adoption rates in metropolitan America.

National surveys—conducted by the Census Bureau, the Pew Research Center, and other organizations—have consistently found large variation in household broadband adoption across many demographic characteristics, including age, income, race, and education.9 While such cross-tabulations provide valuable context, they do not reflect the simultaneous influence of other relevant factors. This brief aims to addresses this shortcoming by developing a regression model to assess those variables of interest while controlling for other factors.10

Consistent with findings from national surveys, the model provides further evidence that income, education, and telecommuting had the largest effects on metropolitan broadband adoption in 2014 (Figure 3). Wealthier and more educated metro areas tend to have higher broadband adoption rates. Specifically, the model estimates a 1 percent increase in either the share of households making over $50,000 annually or individuals with at least a high school diploma is associated with a 0.3 percentage-point increase in metro adoption rates. Telecommuting has an even larger estimated effect, where a 1 percent rise in the share of all commuters working from home is associated with a 1 percentage-point increase in broadband adoption.

 On the other hand, the model failed to find statistically significant evidence of an association between broadband adoption and two traditional demographic indicators of interest. Neither the share of population 65 and older nor the Hispanic share of total population registered a significant effect. In no way do these results contradict the demographic splits found in the ACS and previous surveys. Instead, they suggest top-line findings of earlier surveys may reflect a combination of other underlying factors. Older individuals tend to have lower broadband adoption, for instance, but these modeled results suggest that could be due to lower incomes or less education.11 Yet not all traditional demographic splits were insignificant: The share of black population had a significant and small negative effect on broadband adoption. Overall, these results invite deeper investigation of demographic factors at a more local level.
The model also helps clarify the extent to which local broadband quality, development patterns, and industries can affect adoption rates. While media narratives often suggest deployment of higher-speed networks will boost local subscribership—and current data does show a wide range in average speed across the country—the model finds that faster download speeds are not yet associated with higher subscription rates.12 Instead, the share of a metro’s population living in urban areas is positively associated with adoption rates. This association may speak to a combination of different factors, including greater infrastructure deployment in denser areas, lower costs of service relative to rural areas, and potential network effects from neighbors using computers and the Internet.13 In addition, with broadband connectivity of growing importance to many service-related industries, the share of workers in technology-, management-, and education-focused occupations registered a moderately significant (at the 5 percent level) effect.14 The relative importance of industry was sensitive to model specification, though, suggesting the need for further research into the relationship between occupations and broadband.15 Still, the kinds of jobs people hold and where they live are important indicators of projected subscribership at the metro level.Collectively, the model points to many different factors influencing broadband adoption rates in metro areas in 2014. Markets boasting higher levels of income, educational attainment, and urban density with many tech workers—like Seattle or university towns like Iowa City—tend to have the greatest rates of broadband adoption. On the other hand, markets with lower levels of income, educational attainment, and density with fewer tech workers tend to lag behind in adoption, including places like Memphis and Laredo, Tex. In turn, viewing broadband adoption at a metro scale is often tightly linked to several variables, many of which will need to be examined in greater depth and over longer timeframes to gauge their relative importance.

From 2013 to 2014, the share of households with a broadband Internet subscription rose by 1.7 percentage points nationally, including a statistically significant increase among 58 of the country’s 100 largest metropolitan areas.

Over the two years the American Community Survey has collected data on broadband adoption, approximately 2.6 million more households gained a broadband Internet subscription in 2014 compared to 2013, driving up the national share from 73.4 percent to 75.1 percent. At the same time, the share of households without a computer dropped from 16.2 percent to 14.9 percent. Collectively, these year-to-year changes point to a growing a reliance on high-speed Internet access—and digital hardware in general—that has resulted in an increased demand for broadband, in line with previous studies.16

However, broadband adoption rates have continued to differ markedly among certain households and individuals, confirming two indicators of significance from the previous finding: income and educational attainment. For instance, while households with annual incomes under $50,000 tended to see their adoption rates increase the most from 2013 (1.8 percentage points), their overall shares remained considerably below households earning more than $50,000 (59.3 percent compared to 88.8 percent). Individuals with less than a high school diploma made similarly high gains—jumping from 51.7 percent to 54.1 percent, or 2.4 percentage points— but they still lagged far behind individuals with a high school diploma or higher (81.6 percent). Although it’s promising to see lower-income and less-educated groups increase their subscribership, these growth rates will need to accelerate even more to achieve adoption across the entire population.

The country’s 100 largest metropolitan areas followed many of these national trends, with most markets seeing a statistically significant increase in broadband adoption. Almost 1.8 million more households in these metro areas had a broadband Internet subscription in 2014 versus 2013, accounting for 68 percent of the entire U.S. gain. While widespread, the largest increases overall were concentrated in the most populated metro areas, including 426,000 more households with a broadband subscription in New York, Los Angeles, Houston, Seattle, and Chicago.

In total, 58 of the 100 largest metro areas saw a statistically significant increase in their broadband adoption rates compared to 2013, including 39 metro areas that exceeded the U.S. increase over this span (1.7 percentage points). The biggest gains—of 3 percentage points or more—tended to occur in metro areas with relatively low adoption rates to begin with, such as Fresno (66.1 percent to 71.5 percent), Youngstown (63.6 percent to 68.9 percent), and Tulsa (70 percent to 74.2 percent), which are building off a number of initiatives.17 Nonetheless, significant gains have also taken place in markets with higher adoption rates—those that were above 80 percent or more in 2013—like San Diego, Honolulu, and Ogden, Utah.The 42 remaining metro areas saw little to no change. Many of these markets registered increases under 1 percentage point or less, with 11 other markets seeing a slight and insignificant decline. The largest drops were concentrated in the south, led by Greensboro (4.8 percentage points), Augusta (1.3 percentage points), and Lakeland (1 percentage point). In this way, while many metro areas are recording significant gains in broadband adoption, several others are declining or remain stuck in place, reinforcing the continued need to get more Americans online.

Implications

National and local broadband adoption rates confirm the country’s transition to a digital economy is well underway. Private subscribership rates at the national level continue to grow each year, connecting millions of more Americans to digital information networks. Meanwhile, leading metro areas like San Jose and Washington, D.C. demonstrate how rates have the potential to get even higher around the country.

However, completing the transition to an all-digital economy will be impossible until broadband adoption looks ubiquitous like water and electricity infrastructure.18 And much like electricity development in the 20th century, ensuring every American has reliable online access is a clear 21st century mandate to maintain the country’s global economic preeminence.

The results of this brief begin to sketch a roadmap to get there.

Targeted income assistance programs are a clear priority. Some of these policies will be set at the federal level, including efforts already underway; a reformed Lifeline program and the newly-announced ConnectHome will both improve broadband availability and affordability for low-income households.19 However, regional programs must also address local needs. In addition, there will need to be greater coordination across all government levels with private Internet service providers, most of whom already have a demonstrated commitment to community investment, despite lingering questions over civic transparency.20

Expanding digital skills curricula and training are equally important—and they should extend across entire metro areas.21 Digital literacy classes and training programs, for instance, can help prepare young students for their digital future and offer opportunities for adults to improve their digital skills today. Community assets like libraries are especially important in this respect, by providing public Internet access and representing centers for training.22 Fortunately, the recently-expanded federal eRate program will help communities build capacity at their schools and libraries, in particular.23

Public and private sector employers should also continue to incentivize telecommuting for both broadband and transportation benefits.24 Many of the largest metro areas with the highest broadband adoption rates also have some of the country’s worst roadway congestion—places like Washington, D.C. and San Jose. Telecommuting could both help get more households online and better maximize the transportation capacity already built in these regions.

Finally, continuing development of those federal and local policy roadmaps will also require continued research into broadband adoption. As demonstrated by the wide range in metropolitan adoption rates, it’s especially important to understand the local neighborhoods that drive high or low metropolitan performance, and how other variables may impact neighborhood-scale adoption. Considering the macro importance of getting the entire country online, the results of this study are really just a start.