Abstract We study how US broadband has been deployed from 2010 to 2016, at the block and block-group level, focusing on access technology development and upgrades, expansion in rural and non-rural areas, demographics, housing starts, and ISP growth. Since broadband definitions have changed over time, we provide all statistical analyses using both the 25/3 Mbps and the 10/1 Mbps thresholds. We combine data from the FCC 477 forms, the 2010 Census, and 2010-2014 ACS data to analyze the technology, performance, provider and service territory characteristics of broadband service offerings.
Starting with the infrastructure-centric analysis, we analyze the most common technologies in new broadband deployments and then proceed to determine the speed distributions for each. We follow up to see how the access technologies have changed. From 2014-2016, terrestrial fixed wireless, fiber to the home, and cable modem (DOCSIS 3.0 ), respectively, were the most common technologies associated with new deployments of broadband by population for 25/3 speeds. For new 10/1 deployments, the general trends were similar except ADSL 2/2 was also heavily used. As expected, speed increases from 10/1 to above 25/3 speeds were most common for fiber and terrestrial fixed wireless.
In terms of the rural/non-rural divide in expansion, we investigate the change in speeds of new offerings in rural regions relative to urban ones and on the deployment of higher-speed technologies (particularly, fiber and DOCSIS 3.0 ) in rural areas. As expected, from 2014-2016, urban blocks saw both the most new 25/3 offerings and the largest increase in average speeds relative to rural ones. We take a closer look at zero population blocks that have broadband access, finding that over 80% of these blocks have a large fraction of seasonal homes.
We use binary logistic regression to develop correlation coefficients between broadband availability as well as new deployment and income, population density, population change, education levels, labor force participation, and home ownership percentages.
Finally, we explore how provider coverage across the nation has changed, focusing on the importance of new providers such as rural electric cooperatives, the impact of provider mergers, the disappearance of providers, the (lack of) territorial expansion of cable providers, and competition.
Currently, 45% of households for 25/3 and 67% for 10/1 have access to two or more providers. We describe what forms of competition are most common in rural and non-rural areas.
We also consider providers showing large gains or losses of broadband access during the study period and explore whether this time-aligns with changes in USF policies or provider merger commitments.
These results may allow policy makers to distinguish the organic growth of broadband from growth stimulated by universal service support, and create predictors of where and when broadband is most and least likely to arrive.
