Where Are Migrants Going? Part IV
No Evidence of Urban or Rural Migration Revivals
I’ve spent the last three days talking about metro areas. Next week, I will focus on state-level trends. But before I get to that, it’s important to dispel one particularly pernicious migration myth: the myth that trends in American migration have fundamentally changed, and Americans are, at long last, moving back to city centers. The reality is quite different. Much as it has for decades, migration in 2013 tended to favor the suburbs the most, and high density urban areas the least. Unfortunately, getting detailed data on the demographics of less urbanized migrants is much more challenging than urban migrants, so I won’t be able to provide the same level of disaggregated detail for these flows.
Data produced by the World Bank in collaboration with various national statistical offices indicate that urbanization is on the rise around the world, including in the United States. Additionally, the Census Bureau has confirmed that urban areas have seen the lion’s share of population growth in the US.
But hold on: how is the United States more heavily urban than, say, the European Union, which has far more densely populated cities? How is sparsely populated Canada as urban as the United States?
The reason is that “urban” in this context doesn’t mean dense cities as we think of them. Urban just means “not completely rural.” Europe (especially France and Eastern Europe) has a “dense countryside,” whereby even rural people live closer together than rural people in the United States. On the other hand, very few people live in the (to this southerner’s highly biased sensibilities) frigid ice-wastes that make up the vast majority of our charming northern neighbors’ territory. Thus urban population doesn’t mean high-density living.
For example, I grew up within a metropolitan statistical area, as part of an “urban” population. And yet I grew up surrounded by farms, with a horse pasture for a yard. This is what passes for “metropolitan” in the United States. Because I lived in a county with many commuters into the nearby city, I was “urbanized:” horses, tobacco barns, and swimmin’ holes notwithstanding.
Obviously, that’s not what we mean by “urban” when we talk about urban-centered migration and the renewal of city cores. There must be a better way to sort out the suburbs from the city.
Migration Varies Widely and Locally
The above map shows net migration’s contribution to population growth in 2013, by county. Red counties saw out-migration, green counties saw in-migration, and the blue-ish counties some roughly even migration. This map should hammer home the point I’ve already belabored many times: migration is about local factors, not multi-state regions. Sure, if you casually tilt your head to the side, squint, and look for a trend, you might see some broad patches of strength: west Texas, Florida, the mid-South, northern Nevada, the North Dakota-Montana oil fields, and some others. And maybe some areas look worse: New Mexico and southeastern Colorado, the deep South, and New York all seem to have a notable absence of green.
But this is clearly cherry-picking. There’s no seismic shift of the population southward that drives positive migration in north Georgia and negative migration in south Georgia. The alleged decline of the rust belt is troubled by the remarkable strength of migration flows around Indianapolis, Columbus, Springfield, Illinois, and northern Michigan. County data demonstrates the wide variety, and intense locality, of the American migration experience in 2013.
Migration is Highest in Suburbs and Mid-Sized Cities
I used population density data from the 2010 census to divide every county or county-equivalent (such as Virginia’s independent cities or Louisiana’s parishes) into one of four categories: low, medium, high, or very high density. The thresholds for each were 0, 100, 750, and 2500 people per square mile, respectively.
This data shows that, in fact, dense urban clusters are unpopular with American migrants. Contrary to what we hear about “Millennials” and “Creatives” migrating in droves to repopulate dense, mixed-use urban neighborhoods in Seattle, Portland, or other cities, the reality is that suburbs and mid-sized towns continue to win the migration battle.
Only 11 very dense counties had positive domestic migration, of which 5 were DC or its suburbs (and there is a methodological issue regarding Virginian migration patterns; see the note at the end). The others were Denver, Colorado; Lexington, Richmond, and Winchester, Virginia; and San Francisco, California (barely). Meanwhile, out-migration was highest in the densest parts of Detroit, New York City, and St. Louis, as well as Alexandria, Norfolk, and Portsmouth, Virginia.
In other words, reports of a migration-driven revival of the American city center are greatly exaggerated, at least among many of the biggest, densest cities.
Census-Designated Urban Areas Also Attract Fewer Migrants
The Census also designates certain areas as “urban centers.” Unfortunately, migration data for these areas is less widely available: the American Community Survey only aggregates this data for individuals migrating into urban centers, not out of them. As such, I will focus primarily on the county data I’ve assembled.
But, as far as we can tell, data on Census urban areas shows the same trend: migrants disproportionately move to non-urban areas instead of urban areas.
There’s an important caveat, though: interstate migration, especially by early-career and college-aged individuals, does skew somewhat towards urban areas. But in-state migration skews away from urban areas, especially for early-career and college-aged individuals. This paradox is easy to resolve: when Americans first move to an area, they like to be in a city center. But within a few years, they settle in the area by suburbanizing. City center populations are, then, highly transient, and represent a particularly volatile and capricious source of migration (and thus economic growth) for cities.
International Migrants, Higher Birth Rates Drive Very Dense Population Growth
Rather than domestic migration, very high density areas rely on international migration. San Francisco, DC, Boston, and New York are all major international migration hubs, or “gateway cities.” They are called “gateways” because there’s a conventional belief that many migrants come to these cities, but don’t stay long: they use them as ports-of-entry for subsequent internal migration. As I’ll show tomorrow and next week, that’s actually not an entirely accurate characterization, but it is true that dense cities depend on international migration.
High density urban areas also have lower death rates and higher birth rates, leading to higher natural population growth. This is a reversal of a once-widespread American norm, whereby rural households had the most children. However, rural America is now much older than urban America, largely due to young, international migrants living in cities, and out-migration by young people in rural areas for school and work. This relative youthfulness of cities drives up birth rates, just as the graying of American farm-country drives up rural death rates.
There’s enormous diversity among counties with densities between 750 and 2500 people per square mile. Medium- and higher-density counties saw the highest net migration on average as Americans continue to prefer suburban or town living over high-rise apartments. For example, El Paso County, Texas is in this group, with among the highest out-migration rates in the country. But the actual award for the highest out-migration in the “high density” cluster is in Clayton County, Georgia. Clayton County is also a “majority-minority” county, like El Paso County. Meanwhile, the highest in-migration among high density counties was in Forsyth County, Georgia, a northern (and mostly white) suburb of Georgia.
Some high density areas also succeed in attracting international migrants. While Miami-Dade County, Florida suffers high domestic out-migration (-0.8%), it also enjoys high international migration (1.2%). Lee County, Florida manages to draw both international migrants (0.4%) and domestic migrants (2.2%).
The high and medium density counties with highly positive internal migration are largely southern, including many cities and suburbs in North Carolina, Texas, Georgia, and Virginia. However, there are exceptions like Bristol County, Rhode Island, or Macomb County, Michigan.
Likewise, areas suffering high-outmigration do include many “Rust Belt” or “Frost Belt” counties like Camden County, New Jersey; Providence County, Rhode Island; or Lake County, Indiana. But the list of low-migration counties also includes Shelby County, Tennessee; Miami-Dade as listed above; and many other southern areas or western areas. Suburbs and mid-sized cities succeed or fail based on their own local economies, amenities, existing migrant networks, and demographic features: not implacable economic destiny or a regional curse. Local experiences create migration trends, not vice versa.
Rural Areas Experience Wide Range of Migration
In the years after WWII, the United States underwent rapid (sub)urbanization. Farmboys home from fighting Hitler got an education then moved into the suburbs, even as rising racial tensions led to “white flight” from the highest-density urban centers. This resulted in a major decline in rural areas’ share of the population. This de-ruralization hasn’t reversed, but it has nearly stabilized.
Higher density areas saw virtually no change in their share of the population between 2012 and 2013. But the low-density share fell by 0.1%, and the medium-density share rose by 0.1%.
However, while low-density areas did see a smaller share of the population on the whole, some counties saw population booms. McKenzie and Williams Counties, North Dakota, for example, saw greater than 10 percent population growth due to migration in a single year. Other rural areas experiencing natural resource booms, like Esmerelda County, Nevada or McMullen County, Texas also saw high migration rates. Booming oil and gas production have helped provide rural America a major inflow of migrants (though sometimes at great social cost).
Even some areas without such resources saw strong migration. McPherson County, Nebraska; Rawlins County, Kansas; Corson County, South Dakota, and others all saw robust migration. In many cases, this is due to another natural resource: good soil. High agricultural prices boost farm incomes, which in turn slows out-migration. Since 2000, as Chinese demand for food has steadily grown, US farm exports have risen dramatically, and with them farm incomes. Rising farm incomes have slowed the flow of rural laborers into cities and suburbs, and thus play a major role in reduced migration since 2000. On the other hand, rising farm revenues may drive up demand for migrant farm laborers (especially foreign-born workers), so the total effect on migration is hard to estimate.
At the other end of the spectrum, many rural areas have been less lucky. Areas that haven’t benefited from robust farm export revenues for one reason or another, or that don’t have oil and gas reserves to be exploited, continue to exhibit the age-old out-migration. In my home state of Kentucky, for example, 88 counties fall in the low-density category and of those only 24 had positive migration. Just 2 had over 1% net in-migration. Kentucky’s smallest-in-the-nation average farm size and isolated Appalachian communities make it challenging for the farmers there to enjoy the increasing returns experienced by many larger farmers in the midwest, California, and the plains states, and the area’s mineral wealth is all but exhausted. What coal reserves remain are increasingly uneconomical either due to fundamental constraints or regulatory costs.
When they lack a high-value product to be traded beyond the rural community, rural areas continue to see the same out-migration they’ve seen since WWII: but some communities have found pockets of strength, leading to an enormous variety of recent rural migration experiences.
Whether it’s neo-rustics calling for more rural and earthy living, or urbanists arguing for greater and greater density, the vogue trends in migration are almost all opposed to what most migrants actually do. And contrary to what pundits might think, migrants aren’t wrong or acting irrationally when they prefer mid-density living. They’re choosing the lifestyle that they think is best for them. As I’m an economist, I tend to trust markets: with millions of Americans reliably suburbanizing year after year after year, it’s unlikely that all of these people are all wrong. Far more likely that the critics of suburbia are missing something.
The reality is that most Americans haven’t revealed any preference for high-density mixed-used neighborhoods, or for a vast Jeffersonian republic of small farmers. Rather, they want neighborhoods, pools, malls, parking, vacation routes, and other amenities often best-provided in mid-density settings.
Whether suburbs are environmentally and financially sustainable can be debated, but whether they’re what most migrating Americans prefer can’t be. And until city managers and cultural trend setters find a way to make high-density urban areas more attractive for the hundreds of thousands of Americans still leaving the city centers every year, the endless rhetorical trench-warfare between urban apologists and the suburban middle class will continue.
Start the series from the beginning!
Follow me on Twitter. Follow my Medium Collection at In a State of Migration. I’m a grad student in International Trade and Investment Policy at the George Washington University’s Elliott School. I like to write and tweet about migration, airplanes, trade, space, and other new and interesting research. Cover photo from Unsplash.
A Note on Methods: There’s no Perfect Way to Assess Urbanization
The method I use to identify various densities does produce some debatable results: geographically large counties with one small, high-density cluster would fall entirely under my classification of “rural,” for example. The most heavily populated county in my rural category is Tulare County, California. It is a Census-designated metropolitan statistical area, and does have some notable towns like Visalia.
The picture at left shows Tulare County. It includes Sequoia National Park and several other parks, and, as this satellite photograph reveals, is a major agricultural center. Such “urban” areas as exist aren’t high-density city cores, but operate in many regards as service centers for the local agriculture and tourism sectors. Perhaps Visalia itself is not rural, but it’s reasonable to lump it in, as most people migrating to Tulare County probably understand the area as being more rural and agricultural.
A second complaint could be registered with my selection of the “medium density” or “suburb” cutoff. Technically, many medium-density areas aren’t suburbs: they’re just towns that didn’t make my “city core” cut. For example, Polk County, Iowa includes Des Moines. The county had 452,000 residents as of 2013, and the Des Moines metro area includes areas beyond Polk County as well. Downtown Des Moines is fairly dense, and there are definable suburbs within Polk County, so my inclusion of Polk County among “medium density” areas doesn’t sit well with my characterization of those areas as mostly suburbs. However, the reality is that living in a mid-sized or small city (like Lexington, Kentucky, where I attended college) is probably more like living in the suburbs than in Manhattan. Especially if the city has minimal public transit and most housing is single-unit, and if many houses have driveways or garages, it’s reasonable to lump such cities in with suburban areas.
High-density areas tend to be suburbs or mid-sized cities. At left, I have Denton County, Texas. The large lakefront and rapid transition to rural land in the northwest cause it to barely cross the threshold for the “high density” area, but it does do so.
Finally, one major issue I encountered is that states vary in how they define counties and county-equivalents. In my home state of Kentucky, for example, counties have been called “little kingdoms,” and county officials have enormous authority. Kentucky has 120 counties, more than any state except Texas and Georgia. Local lore suggests our numerous small counties (by which most Kentuckians identify, rather than by a city) were established to ensure that everyone lived within a day’s walk of their local courthouse, enabling easy redress of grievances. So Kentucky’s counties tend to be small, though geographically roughly even. By comparison, counties in many western states can be over 5 times as large, on average.
Mostly, these differences don’t matter much: as long as county lines are drawn based on roughly similar rules, and as long as what it means to be a county is similar, varying size of county across states shouldn’t change many results.
But in one state in particular, county- or county-equivalent definitions follow different rules. The Commonwealth of Virginia has counties and independent cities. Within the core and suburban regions of the Washington, DC metro area, for example, are the District of Columbia, 4 counties in Maryland, 4 counties in Virginia, and 5 independent cities: Alexandria, Fairfax, Falls Church, Manassas, and Manassas Park.
By defining cities separately from counties, Virginia creates artificial densities. The cities are denser than their equivalents in other states while the counties are less dense than their equivalents. Furthermore, due to their eccentric urbanization profiles, Virginian counties and independent cities tend to have more extreme measured migration rates: somebody can migrant across county lines by moving a genuinely miniscule distance in some cases. Much migration across county lines in Virginia is more similar to in-county migration in other states.
So all numbers quoted from Virginia should be understood as not strictly comparable. They are likely to be more extreme and to show many local counterbalancing flows that wouldn’t show up in other states.
In sum, measuring urbanization isn’t easy. The Census Bureau has numerous different options of geographies for measuring urban areas and aggregating the data. Doubtless others will prefer a different measure. I use county population density simply because, in most states, it creates comparable units for which there is uniform data covering virtually the entire nation, and allowing more granular subdivisions than some other geographic options.
