The Evolution of Broadband Competition in Local US Markets: A Distributional Analysis

Abstract

The extent of competition among US broadband service providers in local markets has been a focus for scrutiny and policy debate since the beginning of the current century. The pessimistic view is that in most US residential service markets, the local market structure for broadband services basically remains a duopoly, with competitive choices for most residential consumers essentially limited to wireline services provided by a sole local cable and a sole DSL service network. The optimistic view, on the other hand, focuses on evident and highly visible examples of entry by new competitors in both rural and urban areas—including DSL resellers, innovative fixed wireless service providers providing line-of-sight service in both rural and urban areas, and well publicized new investments in building high speed, fiber-based networks in selected urban sub-markets. Which of these two conflicting perspectives best summarizes what is going on as competition has evolved in local US broadband markets? Our analysis leads us to conclude that there were highly statistically significant changes in the distribution of counts of local broadband service providers serving US census tracts and census blocks over the eight-year period from December 2008 to December 2016, with competition generally increasing after 2011. But surprisingly large numbers of US residential consumers still face what is effectively a duopoly market for residential broadband service. As late as December 2016, 60% of urban census blocks had two or less residential ISP competitors offering 3 Mbps download speeds or better, and a little more than 10% had a single residential ISP offering this quality of service. At this 3 Mbps speed threshold, 85% of rural census blocks had 2 or less ISPs offering terrestrial residential service, over 50% had one or less terrestrial ISPs offering this quality of service, and about 15% of rural blocks had no terrestrial ISP offering 3 Mbps or better service. One key insight from this analysis is that the degree of spatial disaggregation greatly influences conclusions about numbers of competing ISPs relevant to consumer choices. If we looked only at the roughly 72,000 populated census tracts tracked by the FCC in December 2013, we would conclude that roughly 10 percent of US census tracts—divided into urban, rural, mainly urban, and mainly rural groupings—had less than four ISPs offering slow (200 Kbps) or better internet service. If instead we analyze competition in the 6.2 million census blocks for which FCC data is available a year later, in December 2014, we would instead conclude that roughly 45 percent of urban census blocks had two or less terrestrial broadband providers of 200 Kbps or better service in 2016, with roughly 5 percent of urban census blocks having a single monopoly provider. Over 80% of rural census blocks had two or less terrestrial residential ISPs, and over half of rural census blocks had one or less residential ISPs in 2014. Over 15% of rural census blocks had zero terrestrial residential ISPs serving them at even this low, 200 Kbps data rate in December 2014. Our exclusion of satellite broadband providers from our 2014 counts would affect our conclusions about rural competition but is of limited relevance to urban competition. In urban areas, there is also a big difference between counts of all broadband providers (including both business and residential service), and broadband providers serving residential markets. Fully 60 percent of entirely urban census tracts had 3 or less residential service providers in June 2011, compared with only 15 percent with this small a number of providers when all ISPs (both business and residential) are counted. A similar pattern is evident for “mainly urban” tracts. Contrasts between urban and rural areas are also very different when analyzed at the census tract level, vs. census block level. Rural census tracts (which are vastly larger, on average, than urban census tracts) generally appear to have more residential ISPs serving them, at equivalent percentiles of their distribution, than urban census tracts. But when much smaller census blocks are analyzed, and satellite ISPs dropped from the comparison, that conclusion is reversed—urban census blocks have greater numbers of ISPs serving them than rural census blocks, at most percentiles of their distributions. A different pattern seems to hold in urban mobile wireless broadband competition, with new entrants generally increasing after 2008 in both urban and rural areas. After 2011, however, consolidation begins in markets with more than 4 wireless mobile competitors, and by the end of 2016, about 85% of both urban and rural census blocks have 4 or less mobile competitors, and about 40 percent of rural blocks have 3 or less mobile wireless competitors. In rural areas, a significant increase in mobile wireless competition, coupled with industry consolidation and exit that left diminished numbers of mobile wireless competitors in urban areas, leveled differences between mobile broadband availability in rural and urban areas. Finally, there was a general shift toward more competitors in higher quality service tiers (measured by faster advertised download/upload speeds), and higher quality service availability, over time. This is consistent with the view that improved service quality (as opposed to price competition, which we do not study) is the primary current dimension of competition among US fixed broadband service providers.

Keywords: broadband, competition, statistical distribution, spatial distribution, industrial organization

JEL Classification: L96, O31, R12