Which City Won Migration In 2015?
Oh, Come On, You Know I Hate the Winner/Loser Stuff… But, Okay, West Virginia Cities Lost; Anchorange, Alaska Won-ish.
UPDATE: I’m seeing a lot of traffic on this post from direct links. If you came here from such a direct link, maybe on an email chain or something, would you mind letting me know where this is being shared? I enjoy knowing how my work is being used, and it helps me make it even more useful. You can tweet at me, or just leave a comment here! Thanks!
I am slowly catching up on a backlog of issues I wanted to address in all our various new datasets. For today, it’s ACS time! Now, unlike the IRS migration data, the ACS comes from a survey-based sample, not a whole population. This means that we are bound by margins of error. And this means that we can’t say very much of interest about small geographic regions, like a specific county. So I can’t give you the county maps I gave for the IRS data.
But we do have relatively complete data on all metro areas. So I can give you data on that! So that’s what we’ll do.
Let’s dive right in. What did metro-level net migration rates look like in the 2015 ACS?
As you can see I’m using a county map, but for multi-county metros I assign the whole-metro rate to every county in the metro area.
We can see clear reddish-splotches across the eastern Megalopolis, around the Great Lakes, around the Bay Area in California, and a few other spots. Meanwhile, we can see positive migration to many of the Rocky Mountain or Great Plains metro areas, as well as much of Florida, Texas, and the mid-south.
But is that any different than in the past? Well, we can look at 2014 this way too:
So on face value this looks pretty similar. Maybe the south looks more solidly blue in 2014 than 2015, and California’s metros look a little bluer, but by and large, fairly similar.
Let’s make the comparison easier, however, by making a direct map of change!
Here, we can see that migration worsened along basically the entire Atlantic and Gulf coasts except the beach towns of Delaware and southern New Jersey, as well as Florida. We can see that the Megalopolis on the whole was fairly unchanged. The mountain west improved some, while Texas shows many areas with lower net migration. California also seems to be showing us lower net migration.
But on the whole, these regional trends don’t seem extremely strong or compelling. Local-area explanations are probably more interesting than a regional trend-story for metro-area migration. If what you see when you look at that map is a Sunbelt resurgence or the Rust Belt making a comeback, or any of those stories, then I’d say check your prescription. Look for metro-specific trends here, not wider macro-regions.
But when we zoom in on metros, the story gets complicated. See, as you know, ACS provides us with margins of error for their estimates. So we can easily make the error bands for each city last year and see if they overlap with the error bands this year. We make an error band by taking the estimate for inflows, then adding/subtracting the MOE to get a range for inflows. Then we do the same for outflows. Then we can get the maximum possible net migration by subtracting the minimum outflow from the maximum inflow. We get the minimum possible net migration by subtracting the maximum outflow from the minimum inflow. And just like that, we’ve got our range of possible net migration rates. On average, the range is about 6.5%. That is an enormous range.
Now, in a large city like, say, Atlanta, the error-band is smaller: it runs from -0.63% to 1.16%. But in a small city like Ames, Iowa, the error band is huge running from -2.5% to 8.75%. The true net migration rate might be anywhere within that range! And, just to make this clear, the most deeply-negative net migration rate in 2015 was probably in Hinesville, GA, at -5.5%. But the error margin was huge, running from -18.4% to 7.43%! To be clear, that error band covers the entire sample of “central estimates” of net migration (i.e. headline ACS net migration rates). The same is nearly true of Walla Walla, WA, where the error band runs from -8.2% to 5.3%. In other words, error bands can sometimes give as wide a range of possibilities for one metro as exists within the entire sample of central estimates for metros. These are big-freakin-error-bands.
But, again, the error bands do shrink by a large amount for the largest metro areas.
Once we’ve come up for 2015 and 2014 net migration error bands for each metro area, we can compare. If two error bands overlap, then we can’t rule out the possibility that net migration has not changed.
So to really see how migration changed, we need to do this significance test.
Spoiler: only one city in the nation passes this test — Anchorage, Alaska. No city other than Anchorage experienced a change in net migration large enough to conclusively show that net migration really-absolutely-did change. So outside of Anchorage (and actually in Anchorage too: I won’t get into the stats here on why, but basically there’s reason to be skeptical of Anchorage’s estimate too), we can’t authoritatively say that migration changed at all.
But who want’s authoritative information anyways? We all just want reckless speculation, so let’s lower our statistical standards and have some fun!
Let’s just compare the 2015 “central estimate” to the 2014 error band. This isn’t true statistical significance because the 2015 value is a new sample, not a true “value” pulled from the 2014 sample, but this statistically inappropriate method does still show us a list of places where the net migration estimate is at least surprising, given 2014 statistical parameters, if not necessarily significant.
Under this method, a different pattern emerges. We can see clusters of decline around Lake Michigan and central Appalachia, which improvements are more widely scattered.
Again, this map doesn’t show truly statistically significant changes, just metro areas where the 2015 central estimate doesn’t seem extremely likely given the 2014 parameters. Of course, the 2015 estimate may itself be wrong, meaning that these changes mean nothing! But this map helps us cut through the noise at least a little bit and observe some metro areas that may have meaningful stories.
I won’t focus on telling those stories today. I’m sure others can think of reasons for some of these changes. But for now, it’s enough for me that these are more-or-less the cities you should think of as the “big changers” in 2015.
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 graduate of the George Washington University’s Elliott School with an MA in International Trade and Investment Policy, and an economist at USDA’s Foreign Agricultural Service. I like to learn about migration, the cotton industry, airplanes, trade policy, space, Africa, and faith. 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. More’s the pity.
