The 15 Cities Where Migration Changed in 2014
Volatility Is Everywhere, But Significance Is Rare
I wrote last week about the new American Community Survey migration data. In that post, I focused on nationwide changes in gross rates, as well as net rate changes for states. For today, I’ve put together a dataset on migration by metro area. Before I go any further, let me say, I’m happy to share the data I use. It’s a bit too big for a useful Datawrapper chart to include all of it. I’ve compiled every metro area for 2013 and 2014 with intrastate and interstate gross, net, and flows migration data. Leave a comment or Tweet at me if you want the data for yourself.
Rather than talk about metro areas generally, I’m going to look at just 15 metro areas. But these 15 metro areas are special. These 15 metro areas stand out from among the rest. These 15 metro areas saw large changes in their net migration rates, large enough that their 2014 rates were outside the margin of error for their 2013 rates. In other words, for all the volatility in migration, there are just 15 metro areas for which we can say with any real confidence that migration in 2014 really was different from migration in 2013.
So which cities are we talking about?
Metro Migration 2014
A Significant Map of Metros
The map’s a bit small and not labeled, I know. But as you can see, the significant changers seem to be out west, in the midwest, or in the midsouth and Piedmont. There were no significant metro-level changes in net migration anywhere in New England, the Southeast outside of Athens, Georgia, and just one in the Mountain West. None in the Great Plains either.
The significant changers vary in population, migration balance, and size of change. The largest, Richmond, has over 1.2 million residents. The smallest, Cape Girardeau, has just 95,000. Still, there are no major metropolises here, and likewise the smallest geographic areas don’t show up either. Big cities tend to have pretty stable migration trends little impacted by yearly idiosyncracies, while very small areas tend to have small sample sizes and thus large error bands. Most of the cities listed are on the small-ish size, but not too small.
Metro Migration 2014
Estimating Changes in the Significant Cities
The above chart shows 2013 net migration rates in brown, and 2014 rates in yellow. The 2013 and 2014 bars are clearly very different, as one would expect for metro areas with statistically significant changes. As should be clear, the “big changers” had a wide variety of initial net migration rates in 2013, and changes have little correlation with 2013 values. This is important: it suggests that these aren’t just individually significant regressions to the mean. Some metros with positive migration got more positive. Some with negative got more negative. Others switched balances, or moderated. For these metros, 2013 migration is a poor predictor of 2014 migration.
But just how much does migration change?
Changes range from a 6 percentage point decline in Cheyenne to a 5 percentage point rise in Bloomington, Indiana. These are genuinely big changes: net migration rates with absolute values in the 0.5% to 1.5% are the usual range, so these changes have exceptional magnitudes.
But what also stands out is the structure of change. For the most part, most big changes arise from disproportionate changes in one component part. For example, Cheyenne’s big drop is wholly due to a worsening interstate balance, while Redding’s big gain is wholly due to an improving intrastate balance. These different migration ranges don’t necessarily move together.
Large changes in net migration usually come from very specific sources.
This tells us something interesting: if different migration ranges have different paths, then we’re looking at some kind of change that impacts certain groups or decisions, but not other groups or decisions. This gives us some hints for the types of “causes” we should look for in each area.
But can we say anything more detailed? Can we dig a little deeper?
The above chart shows the share of the change in net migration caused by changes in a given component flow: intrastate inflows and outflows, and interstate inflows and outflows. Blue is inflows, red is outflows. Dark is interstate, light is intrastate.
On average, the biggest bar for each metro accounts for 52% of the change. That means that changes in just one component flow account for over half of the change. If you add in the companion direction for the other range, you get 72%.
Just a single migration flow usually accounts for about half of changes to net migration.
So, for example, in Las Cruces, 80% of the change can be accounted for by changes in Intrastate Outflows. In other words, the big increase in net migration is Las Cruces was 80% because outflows to the rest of New Mexico declined. Meanwhile, for Jackson, Mississippi, 77% of the change comes from Intrastate Inflows. Because intrastate inflows fell, Jackson’s overall rate fell, even though other flows changed relatively little.
But if you’re like me, you’ve got another question: is any of this unusual? Maybe these are just really volatile cities. So, to answer that question, I’m going to zoom in on 6 of the cities: Athens, GA; Durham, NC; Richmond, VA; Beckley, WV; Jackson, MS; Cheyenne, WY; and El Centro, CA.
Metro Migration 2014
Athens, Georgia
Viewed by component across a longer time horizon, Athens’ trajectory is fairly clear. Interstate inflows have risen significantly since 2011, before which they were fairly stable. Interstate outflows, meanwhile, have fallen. For intrastate migration, inflows have risen, but show very wide fluctuations year to year. Meanwhile, intrastate outflows have shown a somewhat more consistent decline.
Every migration flow combined in 2014 to boost Athens’ net migration rate.
All four flows moving in that way creates a perfect storm of positive migration, and implies that migration must be fueling lots of growth in Athens. And so it is: net migration of all kinds rose from -1.5% in 2009 to 8% in 2014. 8% net migration is huge to begin with, but coming from -1.5% just 5 years ago makes the turnaround even starker.
I’m not going to venture detailed explanations here for why these changes may occur. Athens’ status as a college town undoubtedly impacts migration, and if university offerings or recruitment strategies have changed, then migration could certainly change with it. Rising college costs nationwide may cause more students to defer to in-state schools, which could explain some of the rise in in-state inflows, but it doesn’t explain the falling outflows or rising interstate inflows.
Metro Migration 2014
Durham-Chapel Hill, North Carolina
The story for Durham-Chapel Hill is quite different from Athens, even though both are university cities. I’ve written about migration in the Research Triangle before, so if you want my detailed thoughts on the area (and the villainous institution of Duke University), see my previous work.
The first standout point in the above chart is how stable it is: migration in the Durham-Chapel Hill metro is not volatile the way migration in Athens, GA is. It’s fairly stable.
Migration in Durham is slowly getting more localized.
But 2007, intrastate outflows exceeded inflows. This means more people moved to the rest of NC that moved in from the rest of NC. This, in turn, means either Durham was depopulating (no), or it had a very high birth rate (mostly no), or it was serving as a conduit for interstate net inflows to be transformed into intrastate net outflows. In other words, from at least 2007 to 2010, Durham gained people from other states, then passed them along to the rest of NC (likely after training and educating them). But that system is changing.
In 2011, intrastate migration became positive: Durham-Chapel Hill gains more people from NC than it loses. Net interstate migration became much more volatile after 2010 as well, but remained around 0.75%. But then in 2014, as you can see, interstate inflows fell and interstate outflows rose. On average then, as of 2014, Durham-Chapel Hill draws in North Carolinians, then launches them out of the state. Whether this is good or bad can be debated, but it does reflect a recent shift.
Metro Migration 2014
Richmond, Virginia
Richmond’s migration flows are even more unperturbed than Durham’s. Notably as well, while Richmond has a similar overall gross intrastate migration rate, is interstate migration rates (inflows and outflows) are much lower. Durham-Chapel Hill has a uniquely strong connection to out-of-state locations that Richmond does not share.
Rather, Richmond’s migration has been quite stable over the period in question. By and large, Virginia’s capital has gained people from the rest of the state, and experienced small net outflows beyond the state. This trend was interrupted in 2012 and 2013 when interstate outflows fell and inflows rose, but the “significant change” in 2014 seems to be mostly just a return to normalcy.
Metro Migration 2014
Beckley, West Virginia
The above chart shows migration flows for Beckley, West Virginia. Many readers may not know where Beckley is. So, here you go:
Highlighted and circled there in the middle, Beckley is in southern West Virginia. It is not in immediate proximity to an interstate border, but is connected via major interstates. There are no major metro areas adjoining Beckley. Indeed, as you drive in any direction from Beckley, the countryside gets very rugged and empty very fast. Southern West Virginia is beautiful, but sparsely populated. Meanwhile, I-64 is one of the few major thoroughfares through the Appalachians. The other options are I-68 several hours north in Maryland, or I-81 which heads south to Tennessee. Both of these options are a long ways away. Finally, the road between Charleston (the nearest substantial in-state metro area) and Beckley has $6 of tolls along it, in three toll stops.
All of these details add up to explain one startling oddity in Beckley’s migration profile: interstate and intrastate migration are about equal in terms of gross flows. And they are equal at fairly low levels: for the most part, Beckley is a low-migration area.
Geographic isolation gives Beckley low migration.
The lack of nearby intrastate population and poor travel conditions within West Virginia yield a very low intrastate rate, while interstate migration is nearer the rates we see for all metros, possible because Beckley has fairly direct interstate connections to other cities.
Does this explain why, in the last 2 years, Beckley has seen a massive uptick in people moving out to the rest of West Virginia? No. I don’t have an answer for that. But Beckley’s unique circumstances at the outset may expose it to unique volatility, especially for short-range migration.
Metro Migration 2014
Jackson, Michigan
Jackson, Michigan had three great years for intrastate migration in 2011, 2012, and 2013. But in 2014, intrastate inflows returned to their previous levels, close to even with intrastate outflow levels. Meanwhile, interstate migration remained extremely stable at breakeven levels, a pattern going back to 2012 when interstate outflows dropped significantly.
Jackson from 2007 to 2010 was a city that about broke even within the state, but lost people to the rest of the nation. From 2010 to 2012, there seems to have been a transition. Jackson no longer loses very much to the nation on the whole and, for several years, saw large inflows from the rest of Michigan. But in 2014, it was approximately breaking even on all fronts.
So what’s going on with Jackson? As a coworker of mine likes to say:
Beats the hell out of me is a perfectly acceptable answer.
I dunno. I know next to nothing about Jackson, Michigan. I originally wrote this blog post thinking I was talking about Jackson, Mississippi, and only realized my error during a last read-through. So you’ve got the charts and graphs, but I’ve got no idea what’s going on!
Metro Migration 2014
Cheyenne, Wyoming
Cheyenne’s interstate migration dropped precipitously in 2014. Outflows rose and inflows fell even more. The rest is that net interstate migration fell from 2% to -4.5% in one year. The next lowest since 2007 is -2.5% in 2011. So what’s going on?
First of all, note that Cheyenne’s interstate migration dwarfs intrastate migration in total volume. This is because, wait for it… Cheyenne is in Wyoming. There’s not really a lot of other people in the state for migration to circulate, and there’s certainly not the kind of metro-to-metro migration we see within most states. Meanwhile, Cheyenne is close to the border with Colorado, which does have several metro areas closely connected with Wyoming’s economy.
Finally, Wyoming is special in another way. Along with Alaska, West Virginia, and North Dakota, Wyoming is one of the most natural-resource dependent states for its economic activity and government revenue. Wyoming’s big resource is coal. As it turns out, 2014 wasn’t a great year for coal. I can’t say absolutely for certain, but if I had to guess, I would say Wyoming’s sharp decline in interstate migration relates to weakness in the mining industry. Notably, migration by workers in the mining industry generally also fell sharply in 2014.
Metro Migration 2014
El Centro, California
El Centro is hemorrhaging people within California. But in 2014, the bleeding slowed as inflows ticked up and outflows dropped some. Many readers will recall that El Centro has among the highest unemployment rates of any metro area in the country. This pretty handily explains the outflows. Furthermore, people leaving unemployment likely have relatively few liquid assets, so long-range migration is often out of the question. This explains why the major net outflows are shorter-range intrastate flows.
But why did we see migration improve for all 4 flow categories?
The above chart shows El Centro’s unemployment rate. It fell substantially over 2011, 2012, and especially 2013, before stabilizing in 2014. Crucially, the relevant period for migration decisions manifested in 2014 is changes in the 2013–2014, or even 2012- or 2011–2014 period. People move to seek work based on lagged perceptions.
If improving employment dynamics are why El Centro’s migration improved, then it may worsen again in 2015.
Unemployment in January and February 2015 was higher than in January and February of 2014 in El Centro. Peak unemployment in the summer months doesn’t appear to be as high, but at least so far, it seems unemployment might be worsening in El Centro. If that’s true, then we can expect migration to stop improving in 2015, or perhaps even get worse (though probably these changes will, again, we lagged).
Metro Migration 2014
Conclusion
Although many metro areas saw migration changes in 2014, just 15 of these changes were statistically significant. Many large changes can be explained by small sample sizes and statistical noise. For the significant changes, most can be explained by volatility in just one or two subcomponents of migration. This illustrates how aggregated migration balances are heavily dependent on highly specific economic and social relationships and geographies. By looking at detailed migration information for specific cities, we can often explain changes and, in some cases, even predict future outcomes. On the other hand, for many cities, it’s very challenging to make heads or tails of a given set of migration indicators. The underlying specificity and heterogeneity of migration makes the interpretation of migration indicators as much art as science. There’s not a consistent rule for whether we should view a migration record as “good” or “bad,” or whether we should think about natural disasters or energy policy or university recruitment or sectoral shifts or land use rules as primary determining factors for migration. All the same, by looking at more and more detailed migration information, policymakers can get a better sense of where they are, and where they might be headed.
See my previous post, about Europe’s refugee crisis.
Start my series on migration from the beginning.
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.
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.
