What Airport Traffic Tells Us About Island Migration
Lots in Puerto Rico… Maybe Not Much in Hawaii
Short post today. I’ve been inundated with responses to my previous piece in which I engaged in wild, non-empirical, borderline apocalyptic speculation about urbanization and greenfield development. Apparently what The People want is not well-researched arguments with interesting policy conclusions, but exciting speculations about why we need to set fire to the cities (no, we don’t. seriously.). But enough about that, and back to the data today.
Salim Furth from the Heritage Foundation was good enough to raise an interesting question to me:
Okay, so what is the best way to measure migration in “real time”? This is an interesting question. Most migration sources are delayed between 1 and 10 years. But there have been attempts at using other data sources, like this interesting attempt by Badu to measure movement for the Chinese New Year. Or consider a recent study that used Twitter and Fouresquare geotagging to look at residence and mobility. Very neat. Then, of course, there’s the ubiquitous “moving van” data.
However, Salim and I got chatting about maybe using aviation data. I have some background in this and knew it was available with great precision and recent vintage, but was not sure about access. As it turns out, Salim found a way to get monthly aviation statistics by airport for any airport in the United States. Pretty cool. The data is here. It’s a bit cumbersome, but it works.
Now, this data is little use for estimating migration within the continental U.S. since huge shares of migrants move by car, and some by train, etc. The gaps are just too big.
But this data could be very useful for predicting migration for areas that receive almost exclusively airborne migrants.
What areas could that be? Well, given that comparatively few people migrate by boat these days, it’s gotta be islands. Luckily, the United States does have two island areas: Puerto Rico and Hawaii.
Air Traffic in Puerto Rico and Hawaii
Apparently, There’s a Tourist Season
The chart below shows net domestic and international passenger originations and destinations for Puerto Rico and Hawaiian airports. I exclude any airport with under 200,000 total annual passenger volume. My sample captures more than 95% of total passenger volume to and from Puerto Rico and Hawaii.
But holy cow that data is noisy. And the reason why is obvious: Puerto Rico and Hawaii are both major tourist destinations, thus have major seasonal swings in arrivals and departures. Hawaii, for example, has peak net arrivals in December, with a lower peak in June. Peak net departures in August, with a lower peak in April. Puerto Rico meanwhile has a huge peak in December for net arrivals, with a much lower June peak. Net departures peak in May and August about equally.
So we need a better chart. So let’s instead take the rolling 12-month totals.
Puerto Rico
A few things stand out. Let’s do Puerto Rico first. Puerto Rico’s international net balance of air traffic is quite small compared to the domestic balance, and crosses the zero axis several times, which means that almost Puerto Rico’s entire air traffic imbalance is with the mainland United States. That’s good news from a data quality point of view, because we have lots of PR-US migration data, and because it means we aren’t getting any kind of weird data quirk involving how international flights get classified, continuing flights to the mainland, etc.
Also as can be seen, Puerto Rico’s air traffic balance worsened from near-balance in 2004 and 2005 to substantial losses during 2006, 2007, and 2008. Early in the recession, around 2009 and early 2010, Puerto Rico seems to have greatly improved its air passenger balance. But that changed by mid-2010, which passenger balances returning to low 2006–2008 levels. But then in 2014, things got even worse, with Puerto Rico’s air passenger balance falling to a loss of nearly 100,000 people over a 12-month period. That is intense. It’s out of step with Puerto Rico’s recent trends, is a sharp decline, and does not seem to show any strong evidence yet of abating.
Hawaii
Hawaii could hardly be more different. Hawaii’s net balance is large and varied for both domestic and international passengers. This is a problem. International balances are likely to mess up the data as passengers and flights change status from “international” to “domestic,” and international net flows are also a major weak spot in the available migration data. So with big international net flows here, we’re likely to get some chaos in comparison with migration data.
As can be seen, Hawaii maintained strong domestic net balances until early 2006, when they began declining. While international balances had been negative, they fwll even more around that time. But in 2007, international balances rose, then fell dramatically, even as domestic balances stayed low. The result was total net balances of nearly -90,000 during 2008. Then passenger balances rose sharply, first for international passengers, then in early 2010 for domestic ones. Even after domestic balances fell again in 2011, international balances rose enough to compensate. Although total balances have been slowly declining since 2012, they remain at moderate levels today.
So that’s the air traffic data. Now let’s look at the migration data.
Air Traffic in Puerto Rico and Hawaii
Lots in Puerto Rico… Maybe Not Much in Hawaii
So let’s compare that passenger balance data to annual migration data. As before, we’ll start with Puerto Rico.
Puerto Rico
As is often the case, I find that the best fit comes from lagging the migration data by a year. I’ve explained many times before why the temporality of migration data can be inconsistent and highly sensitive to very specific monthly patterns — recall that Puerto Rico and Hawaii both have extreme seasonality in net passenger balances. Ergo, yeah, I resituate the migration data to get a better fit. Below, I provide the annual (Jan-Dec) domestic passenger balance for Puerto Rico, versus Census Population Estimates data for all migration (PR migration data is not broken out by PEP), and ACS data on domestic net migration.
Huzzah for good data! Look at that! We’ve got a pretty good match on volume of net outflows, and even some agreement on trends. It’s far from perfect, but it’s enough that we can say that net air passenger balances for Puerto Rico are probably a reasonable advance indicator of what later-released official migration data will say. Not perfect, but in the ballpark. Assuming ACS and PEP are correct, we would attribute the errors to arrivees who die after arrival (boosting net balances too high), migrants via ship (directionally ambiguous), minor airports excluded (directionally ambiguous), and individuals who are part of the net balance but not counted as migrants by PEP or ACS (potentially some individuals enrolled in schools or other institutional housing, although ACS should capture these people).
So if you want a good “real time” (or closer to it) indicator of migration in Puerto Rico, net air passenger balances are probably a good tool.
Hawaii
But then we get to Hawaii. The graphs below show Hawaii data. The left chart shows international migration, the right domestic. I have not bothered to lag the Hawaii data because, as you’ll see, there’s not really a good correlation almost no matter how you cut the data.
I’ll admit, I don’t quite know what to make of this. If you believe the domestic air passenger data, then Hawaii has way more erratic migration than either PEP or ACS would lead you to believe. Oddly enough though, for the full period of complete data running 2005–2014, the three sources nearly agree on cumulative net migration. Passenger balances say -38,000, PEP says -37,000, ACS says -47,000. There could be several things going on here. It could be that the seasonality of Hawaii’s air traffic is causing the big swings. But that’s weird, because it’s not like each year corrects the previous one: there are years of delay. It’s also possible that we’re seeing some weird military migration/deployment situation that is captured in air traffic, but not in PEP and ACS data. This shouldn’t be the case as PEP and ACS allegedly include military migration, but alas, no data is perfect.
The international data is even worse. If you believe the passenger balance data, Hawaii was actually losing people to international migration from 2002–2009. That would be remarkable. I suspect again that we are seeing some heavy military influence on this data, but, again, PEP should actually capture that.
The gap in these sources is huge. Air passenger data suggests that from 2002–2015, Hawaii has lost 89,000 people. But PEP data suggests Hawaii has gained 83,000 people. These cannot both be true. Now, I’m 100% ready to believe that the PEP data undercounts emigration, but this level of undercounting is astonishing. A gap of 172,000 people amounts to over 12% of Hawaii’s total population. It doesn’t seem even remotely possible that the Census Bureau’s population estimates missed by so wide a margin, which means our international passenger balance must be horribly wrong somehow, or Hawaii receives a large number of immigrants via boat.
Conclusion
So while the air passenger data seems like a good predictor of Puerto Rico’s migration trends, it seems like a very poor predictor of Hawaii’s. It’s not immediately clear why this is, but seems likely that a combination of factors, ranging from the role of the military, to the prevalence of international migration, to the structure and classification of the local aviation sector, all have a role to play in determining the usefulness of the data. Puerto Rico watchers may benefit from following the air passenger balance data, but migration-watchers generally may not derive as much benefit. And if somebody can explain to me why Hawaii isn’t looking as pretty as Puerto Rico, I’d be much obliged.
Check out my new 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.
