How Migration Changed in 2016
Data from the 2016 American Community Survey
We’ve got new ACS migration data! So what does it say?
Well, as always, we start with the headline migration numbers.
The red lines are the new data. Interstate migration rose slightly, while intercounty-but-within-state migration fell slightly. These changes are all quite small. For journalists out there, the appropriate summary verbiage is a “marginal,” “small,” or “slight” change, with a bias towards saying that migration has been roughly “even” or “stable” the past few years.
But that’s boring! Let’s break out the demographic cuts!
The Migration Life Cycle
Certain times in life are more associated with migration than others. For example, the college years often see residential relocations. So here’s migration by age since 2006:
And here it is presented another way, with the 2006, 2010, and 2016 migration life cycles shown:
So since 2010, migration has risen for most groups, except people ages 5–24. And since 2006, migration has fallen for every age group except people aged 65–74, who it’s basically stayed the same for. Broadly speaking, the migration life cycle curve is a little bit flatted in 2016 than in either 2006 or 2010, suggesting that the “life cycle” is less extreme than it used to be, though only a bit.
Finally, we can calculate a “total interstate migration rate” that is age-independent, exactly how we would calculate a “total fertility rate.” This metric shows us a measure of migration that is stabilized with respect to age, and shows how many times a person who is born today will migrate over their lifespane if current rates stay the same over their whole life. That “if” never occurs of course, but it’s still a useful way to conceptualize age-independent migration rates.
As you can see, TMR did not budge from the 2015 to 2016 ACS. In other words, changes in aggregate migration can be fairly completely decomposed into changes in the age composition of the population.
Another way to look at the life cycle is to consider education. Here’s interstate migration rates by education.
Here we see a fascinating trend. Migration continued to rise for the very most educated, while it fell for the very least educated. In-between groups show less consistent changes, but the result is that we see a growing gap in migratory propensities across education. The highly-educated are finding ways to keep moving to opportunity, while the less-educated are more stuck in place.
This is interesting because this has not always been the case: historically, migration rates by education have tended to move in tandem.
It is possible that as society becomes more educated, the most migration-prone of lower educational groups are also the most marginally likely to obtain higher levels of education, meaning that rising levels of education would reduce migration rates for lower-education-groups without representing any actual change in gross migration rates. I suspect that such a trend is likely, but it’s not clear why it would kick in right in 2010 and not before. I think some kind of event around 2010 or after is more likely, whether it’s recession, skill-biased recovery, changing educational financing, the opioid epidemic, changed healthcare policy, or something else, I don’t know.
This trend is sufficiently interesting that I wanted to look at more than just interstate migration. Let’s include in-state intercounty migration.
Here the trend is, if anything, even clearer. I would disregard 2005 for this chart, by the way, as the ACS methodology changed markedly in 2006. But since 2010, we’ve see much more recovery in educated migration rates than for the less educated. Indeed, graduate degree holders have about the same gross migration rates they displayed in 2006, and bachelor’s holders are similar! But for the less educated, full recovery is nowhere in sight.
This is an unusual kind of bifurcation in migration trends, and it is large. College graduates have seen 6 years now of rising migration, by nearly a full percentage point, while the less educated have just flatlined.
Other Demographic Factors
We can also look at at other key factors. For example, here’s migration by race:
I’m not sure we should make too much of these rates, but I do want to suggest one thought. Hispanics have much lower migration rates than other large racial or ethnic categories. I have to wonder if this is connected to low migration among less-educated people. It could see causality flowing either way: perhaps something about being Hispanic induces lower migration, or perhaps something about being less educated. But either way, if Hispanics are a growing share of the less-educated, and if the less educated are an above-average share of Hispanics, some of the trend we’ve observed by education level may be closely bundled up with changing U.S. racial demographics. But that’s fairly speculative.
We can also, relatedly, look at migration by citizenship status.
Gross migration rates rose for all three citizenship status categories, with non-citizens showing particular growth in recent years.
Census also gives us some more economically-related factors. For example, we can look at migration by income level…
And see that it isn’t super interesting. Lower income folks move somewhat more, although the over-$75k category has risen relative to the groups right below it in recent years. The lowest-income continue to be the highest-migration, largely thanks to student-migration.
But perhaps more interestingly, we can look at migration by housing tenure.
Renters are moving less, owners are moving more.
Now, quick caveat: this is designated by current housing status. So if a large number of renters become owners, then they will report their status as an owner, and as having migrated, given they’ll have changed residence. So flows between these categories can be misleading: if many renters move up in the world and buy homes, it will look like owner migration rising, and probably renter migration falling, if the new-owners were the marginally most migration-prone (i.e. young and educated) among renters.
But, this is still really interesting. Interesting enough that I wanted to see if it holds up when we include in-state migration.
So yes, the trend holds when we include intercounty migration. If anything, it’s even stronger. Now, I’m willing to believe that a substantial part of this is driven by renters converting into being owners, but I’m inclined to think that much of it is also a “real” change among existing renters and owners too.
Winners and Losers
Because I am a slave to your clicks, I will offer you net migration rates by state in the 2016 ACS, showing “winners and losers,” as you awful people like to call them.
Lots to unpack here. Several “frost belt” states do rather well: Maine and New Hampshire. North Dakota’s oil boom has ended as far as migration is concerned. “Sunbelt” states like Tennessee, Virginia, Mississippi, Louisiana, and New Mexico fair poorly, worse than many “frosty” states like Wisconsin or Indiana. Delaware continues to be a quiet champion. The mountain west still rocks, as does the Pacific Northwest; all of those states benefit significantly from migration coming off the Great Plains, New England, and Rust Belt, but they particularly receive flows from California.
So, cool! But hold on. ACS migration estimates have an error band. How confident are we really about these states? Well, the next map will show each state color coded by whether its net migration rate is statistically different from zero, and the direction of that difference, positive or negative.
Much of the regional noise vanishes, and we’re left with some very high-level trends. California is losing big (as is Alaska), and it’s spilling into the other western states. The “Eastern Megalopolis” is losing lots of people, and it’s spilling into the Southeast. Texas and Illinois sort of stand alone; Illinois loses to Texas, the west, and the southeast, while Texas makes gains from all three big losing regions.
This graph helps illuminate a key point: “Frostbelt” and “Sunbelt” are nonsense terms. Rather, we see semi-regional migration complexes driven by a variety of factors: jobs, cost of living, culture, policy, demographic fundamentals, climate, etc. We aren’t seeing just 1 or 2 macro-regions for the whole United States. We’re seeing an interlocking network of discrete sub-networks. If you have a framework that calls Seattle a Sunbelt but makes New Orleans frosty then, well, you’re full of crap. Different forces are at work to different extents in different places.
Now, we can also look at 2015 migration. We’ll start by doing a simple change map.
You can see substantial improvements in New Mexico, Michigan, Washington, and especially Maine, and see substantial worsenings in North Dakota, Wyoming, Delaware, Rhode Island, and Alaska. Or at least, it looks like that’s what you can see.
We can look at the error bands for these estimates in 2015 and 2016 and see if there are cases where the 2015 and 2016 error bands are completely non-overlapping, indicating we have a great deal of confidence that migration really did increase/decrease, and also for places where the central estimate in 2016 is outside of the 2015 error range, indicating places where it’s pretty reasonable to think that migration did indeed increase/decrease. Here’s a map of that.
The only state where we can say with an extremely high degree of confidence that the ACS picked up a real change in migration is North Dakota, where migration definitely fell. Migration probably fell in Rhode Island, Oklahoma, and Tennessee as well. Meanwhile, migration probably rose in Michigan, Pennsylvania, and Maine. For other states, we should be very cautious about making strong statements vis-a-vis changes in gains and losses.
ACS data shows more-or-less stagnation in gross migration rates, although there are some substantial changes among certain subgroups, especially across different educational and housing groups. For the most part, changes in net migration by state were not extremely large, although there is some sign of improving migration in several traditional losing-states, and worsening migration in the North Dakota oilfields as well as possible some of the upper-south.
CPS ASEC migration data is released in microdata format, but I’m waiting for it to be available in a more easily usable format. That should be within a few weeks or months, and at that time we’ll get a first glimpse of 2017 migration data!
Check out my Podcast about the history of American migration.
If you like this post and want to see more research like it, I’d love for you to share it on Twitter or Facebook. Or, just as valuable for me, you can click the recommend button at the bottom of the page. Thanks!
Follow me on Twitter to keep up with what I’m writing and reading. Follow my Medium Collection at In a State of Migration if you want updates when I write new posts. And if you’re writing about migration too, feel free to submit a post to the collection!
I’m a native of Wilmore, Kentucky, a graduate of Transylvania University, and also the George Washington University’s Elliott School. My real job is as an economist at USDA’s Foreign Agricultural Service, where I analyze and forecast cotton market conditions. I’m married to a kickass Kentucky woman named Ruth.
My posts are not endorsed by and do not in any way represent the opinions of the United States government or any branch, department, agency, or division of it. My writing represents exclusively my own opinions. I did not receive any financial support or remuneration from any party for this research.