The Daily Yonder's coverage of rural economic issues, including workforce development and the future of work in rural America, is supported in part by Microsoft.
Although most of the nation’s rural counties lost population from 2010 to 2020, our analysis of new Census data shows that rural counties with better broadband access tended to do better with population change than counties that lacked access.
We can’t say whether broadband access caused the population change or whether it was the other way around – counties with growing populations may have attracted more broadband service. But it’s clear that there’s a relationship between broadband and population change, our study shows.
Factors That Affect Rural Population
Since the Census released their 2020 county-level numbers in August, a lot has been written about how the country’s population evolved from 2010 to 2020. We know that large metro areas lead the way for overall population growth, while roughly two-thirds of non-metro (rural) counties shrunk. Rural counties had an overall population loss of around 280,000 individuals over the last decade. We also know that specific types of rural counties fared worse than others: nearly 80% of farming-dependent counties lost population, while only 40% of recreation-dependent ones did.
Academic research has long told us that a variety of factors influence non-metropolitan population growth, including natural amenities, economic dependency (like the categories for farming or recreation noted above), and proximity to a major metropolitan area. But we’ve also heard a lot (both from a while back, and more recently) about the importance of broadband access in rural areas. This is particularly true as we continue to deal with the pandemic. So we wondered, did rural counties with relatively better broadband access early on during the 2010s fare better with population growth?
Thankfully, statistics give us a way of figuring out which factors are most important for explaining a particular outcome. It’s called regression analysis, and many social scientists (like us) use it to find those important factors – and to argue for policies that focus on them. In our case, we’re interested in trying to explain the factors influencing population change between 2010 and 2020. We have several variables that we know should have an impact: the base (2010) population, the economic dependency categories, the presence of natural amenities, and proximity to a metropolitan area. We also have a hypothesis: that rural counties with better broadband access in the early years of the 2010s fared better with population change. Thankfully the National Telecommunications and Information Administration (and later the Federal Communications Commission) began collecting data in the early 2010s that can be used to determine the percentage of county residents who had access to different broadband speeds.
So, we set out to build a dataset of all 3,142 U.S. counties with each of these variables included. We then ran a regression analysis to see which of these variables was more strongly associated with 2010-2020 population change. If broadband showed up as being “statistically significant,” we can conclude it has a meaningful relationship with population change.
The regression results in the table below show that most of the variables we included had the anticipated relationship with population change. For example, a county classified as recreation dependent saw a population change rate that is 2.9% higher than a similarly-sized county without the recreation designation. Counties dependent on manufacturing or government employment saw 1.1 to 1.5% decreases in population change compared to otherwise-similar counties. Amenities and metro proximity also matter – higher scores on the ERS’ amenity index are associated with positive population change, and each extra mile away from a metropolitan area is associated with population decline. Counties that started off with bigger 2010 populations also had higher growth rates over the 2010-20 period. Surprisingly, farming dependency and mining dependency were not significant in predicting lower growth rates – but this may be because many of those counties also had lower 2010 populations, lower amenity scores, and / or were further away from metropolitan areas.
Evidence that Broadband Availability Matters
But what about broadband access? Even after all of these other factors are included, there is strong evidence that early broadband availability mattered. The results show that as more residents had access to broadband as defined by the FCC in 2011, the county population increased nine years later. (The FCC defines broadband as having internet speeds of at least 25 megabits per second [Mbps] download/3 Mbps upload, often expressed as 25/3.)
Putting a firm number on the potential impact of broadband requires a little more calculation. Consider two identical rural counties, with the only difference being that County A had broadband available to 50% of its residents in 2011, while County B only had 10% with access. In this case, County A would be expected to see a 1.0% increase in population compared to County B [(0.026)*(0.50 – 0.10) = 0.01]. A 1.0% increase in population is a big deal, when you consider that the total nonmetro population change was just -0.6%.
Most counties did improve their broadband situation as the 2010s continued, as the graph at the top of this article shows. Broadband access grew as the decade progressed for both kinds of counties – those that lost population and those that gained. But the importance of broadband access only continued to increase – another regression shows the broadband coefficient doubling to 0.052 if 2015 availability is used.
Other models that only include rural counties or that use rural-urban continuum codes instead of the distance measure had similar results. One interesting finding is that access to faster (100 Mbps) broadband speeds showed an even stronger relationship with population growth over this period.
However, the data does not tell us which caused what. In other words: did good broadband access in rural areas cause higher population growth, or did faster population growth in some rural areas enable them to get better broadband access? The tools we are using here simply show the correlation between population change and broadband after accounting for other factors. And the takeaway is that better rural broadband access does seem to be an important characteristic.
Brian Whitacre is professor and Neustadt Chair in the Department of Agricultural Economics at Oklahoma State University. Roberto Gallardo is the director of the Purdue Center for Regional Development.