Will Hawaii Be the Next Puerto Rico?
No, But Things Could Get Rough
The most recent Census population estimates showed something interesting for Hawaii: a population decline.
So 2017 shows a decline, but the 2017 estimates were actually an upward revision in most back-years vs. 2016. So in terms of revisions, 2017 was a good year for Hawaii’s population, while in terms of trend, it was bad.
Here’s what Hawaii’s population growth looks like in the long run.
I did not show population before 1830 because it is hotly debated. However, the population decline from 1830 to the 1870s is widely accepted and attested to in records from when Hawaii was independent.
In the postwar period, there are several instances of “bad years” of population growth: the early 1950s during demobilization and the Korean War deployments, 1962–63 for reasons unclear to me (possibly delayed tsunami response?), the late 1990s/early 2000s, and today. Maybe this year will be a one-off blip!
Let’s look at what caused the changes observed. I have birth and death data for Hawaii going back to the 1920s, and we can use that data plus the Census population estimates to impute a residual-based migration figure. What does it show us?
First of all, the birth rate has been declining. This is largely due to the aging of the population, with fewer reproductive-age women, but is also at least partly due to declining age-controlled fertility rates, a topic I’ll come to later. The death rate, meanwhile, is rising, though I think the sharper rise near the end should not be taken too seriously as there’s a discrepancy between CDC and Census death data showing up that I’m not sure exactly how to interpret. As, interesting trivia, you can see Pearl Harbor quite clearly in the data, as well as the delayed-impact of the 1919 influenza pandemic, hitting Hawaii in 1920 instead.
In terms of migration, we can see mostly periods of gains, but some periods of losses too. This includes both domestic and international migration, of course. To break out sources of migration, we need data specifically on domestic migration. We can only get that from survey sources and the Census. Here are estimates of net domestic migration into Hawaii from various sources:
As you can see, these estimates can be volatile and often disagree on individual years, but broadly they show a shared record that Hawaii’s domestic migration declined slowly from the 1950s to the 1970s, then was stable or even rose a bit, then fell sharply into the 1990s, then rose even more sharply… then churned at around -2,500 to -7,500 for a while, and has since declined fairly sharply. The black line is what I will use for core population estimates from here on out.
So if that’s domestic migration, what must international migration have been?
While there have been bumps, overall international migration has been somewhat more stable than domestic migration, and, of course, rather strongly positive at least since the 1970s. The early 1960s population trend remains befuddling to me… but oh well.
Converting these to rates, we get:
So here you see Hawaii’s problem. It is not experiencing record-low domestic migration rates; they were lower as recently as the late 1990s. But the late 1990s also saw high immigration rates… while immigration rates today are at best stable, possibly falling. Meanwhile, over the course of the 1990s, Hawaii averaged a 1% annual rate of natural increase. Since 2010, it has averaged 0.6%, and falling quickly. So while domestic migration may be leading Hawaii’s plunge downwards, it is helped along by weaker immigration and weakening rates of natural increase.
Why Is Natural Increase Falling?
Let’s start with the simplest component to assess: natural increase. Why is Hawaii’s rate of natural population growth slowing?
Well, simply, birth rates are falling and death rates are rising.
But, why are birth rates falling?
Here we can decompose Hawaii’s fertility into three components:
- Age-compositional change
- Marital-compositional change
- Age-and-marital-chained change
So let’s look at those components. We can start with our probably most reliable data, CDC birth data. Here’s a graph of how many births would occur in Hawaii if we chained the age-and-sex composition to 2007, if we chained the age-specific birth rates to 2007, and if we chained age-and-sex composition and birth rates.
Let’s start with age-chaining. The total number of births in Hawaii would actually be lower in 2015 if it were not for a change in age composition. Basically, Hawaii’s population is more tilted towards prime-childbearing-age-women in 2015 than in 2007, according to CDC.
Next up is the rate-chained line. And that’s the moneymaker: if the rate of childbearing for each age-group had remained constant, the number of children born in Hawaii would have steadily risen instead of falling. And if we add in the age-chain as well, it gets a bit bigger.
By 2015, declining age-specific fertility rates can account for over 8,500 “missing babies” in Hawaii.
Unfortunately, CDC data does not give us estimates of the number of married women by age, so cannot be used to compute that. But we can look at the American Community Survey for that. Now, the ACS shows way more volatile births than the CDC; they’re not directly comparable for a lot of reasons. But we can at least see what effect controlling for marriage has on the ACS, and guess that it might be similar for Hawaii.
As you can see, chaining for marital status and age has a big effect. More than 75% of this is the independent effect of marital status. If we assume that CDC would show a fairly similar effect of marital status if marital status data existed, we could probably, then, expect to get a slightly higher birth number again. My guess would be missing births are somewhere substantially north of 9,000 kids from 2007 to 2015, and, to the present day, it’s almost certainly at least 13,000 or 14,000 missing births.
Given that Hawaii had over 2,000 missing births in 2015 alone, and that fertility has almost certainly fallen since then, it’s worth noting that Hawaii’s population decline was just 1,145 people.
So what might Hawaii’s population trend have looked like with stable fertility from 2007?
The difference is substantial. By 2017, there’s about a 16,000 person population difference, even assuming higher births increase mortality and worsen net migration flows. While population growth still slows, it doesn’t go negative, though it’s likely it would become negative within a year or two anyways.
So births are a big part of the story. What about deaths?
The above chart shows two different ways of thinking about mortality. The yellow one shows the average index value of the crude death rate across 13 age groups CDC tracks. As you can see, most age groups have seen stable or falling mortality.
The second measure, the “total mortality rate,” is just the sum of the annual mortality risk over time, excluding over age 85 mortality. As you can see, it’s falling.
So Hawaii’s death rate may have some kind of increased mortality story going on (opioids) somewhere, but, on the whole its headline mortality trend isn’t driven by a wave of death. It’s driven, of course, by aging. We don’t have age-specific populations for vintage 2017 estimates from Census yet, but the 2016 ones shows that the over-60 share of Hawaii’s civilian population rose from 21% in 2010 to over 24% in 2016. We can expect it to keep rising. And as it rises, so will deaths.
Why is Migration Falling?
The bulk of the change in Hawaii’s migration is in domestic migration, so I will focus on that. The main question we want to answer about Hawaii is the “who” question: who’s coming, who’s going?
Normally, this is fairly easy to answer: we would use the ACS. The trouble is that the ACS does not perform extremely well for Hawaii vs. the Census population estimates. Here’s the data:
ACS is extremely volatile, bouncing wildly from year to year. So just keep in mind as I discuss it that much of the volatility should probably be seen as statistical noise worth smoothing out… but there’s not a straightforward way to do that and preserve underlying demographic detail. So, alas.
So let’s look at age groups.
The teal and magenta lines are thicker so you can see them because they’re the ones I want to discuss. Retirees and college students are the two categories where Hawaii has been most hammered in recent years, though the 2016 drop was largely prime age people from 23–55. But for retirees and students, there’s a real “trend” in recent years, an actual multi-year observable decline. The only group to get a real improvement, meanwhile, is 23 to 30 year olds, though in 2016, they barely broke even.
What might drive away college students and retirees? Well, curiously enough, the amenities research on both suggests Hawaii should be wonderful for them: both groups are thought to be drawn by beaches, tourist amenities, warmth, sunshine, scenery, etc. When we look at college students-age migrants, however, it’s not just that the net balance is falling: gross inflows are falling! From an average of about 11,000 18–22 year olds arriving each year between 2006 and 2011, inflows have now fallen to about 7,000. Outflows have actually been stable, at around 9,000.
The story is different for retirees: inflows of retirees have risen, from about 4,000 or so per year from 2001 to 2013, to about 6,000 per year more recently. But retiree outflows have risen more: from about 5,000 per year from 2001–2012, to about 8,000 per year since then.
And where are all these folks headed? Well, I can’t give you a comprehensive map for every year, but we can at least look at 2016.
As you can see, Hawaii has big losses to to many of the western states, as well as other states around the union, while its gains are concentrated in the eastern U.S., especially the southeast and Illinois. I don’t know what drives these trends, especially the long-distance ones. But The People Demand Maps.
Location, Location, Location
One possible explanation for Hawaii’s recent poor performance is costs. Here’s Hawaii’s CPI over recent years vs. the U.S. average:
Since the early 2000s, prices in Hawaii have risen much faster than on the mainland.
We can look at Hawaii statewide on a shorter timeframe as well. BEA Regional Price Parities consistently show Hawaii around 17–19% higher priced than the rest of the U.S. So it’s an expensive place, and it’s getting more expensive relative to the mainland U.S. over time.
But that’s not a big deal if incomes rise. Lots of expensive places have positive net migration!
What has happened to economic activity?
It’s grown, but not by any means at stellar rates. Since 2006, the best that can be said is that Hawaii did about as well as the rest of the U.S. So with real economic output not outpacing the mainland and local prices high, is there any reason for workers to stick around?
Here’s real compensation per worker for Hawaii and selected other states from 2008 to 2015:
But, hold on, that’s nothing new, right? Well, the ratio of Hawaii’s real employment returns to other Pacific states has fallen over time.
So what drives Hawaii’s cost differential? Well, one is population growth in a place with very limited land. But we shouldn’t take that too seriously: Hawaii is the 19th densest state in the union, less dense than other island territories like Puerto Rico, American Samoa, or the Northern Mariana Islands.
Of course, Hawaii has protected much of its land, and it is very mountainous. It’s possible it is at the border of developable density, as remaining land is mountainous or legally protected, though, again, Puerto Rico is quite mountainous and yet far, far denser.
Thus, to the extent Hawaii is “running out of land,” that is almost certainly due to political choices about land use and protected areas, not a physical shortage of space to build on. And indeed, if you look at a topographical map of Hawaii, you’ll see large, relatively flat plains at a safe distance from volcanoes in Maui, Molokai, and a fair amount on the big island of Hawaii as well. More likely, the issue is land use rules, of which Hawaii has extremely strict ones. That drives up the cost of housing and commercial real estate. Likewise, the booming tourism industry in Hawaii may be creating a wedge between the price of real estate for vacation-usage and for residential-usage, pushing residents out in favor of vacationers, as suggested in this pretty good WaPo article. Another factor is the Jones Act, which restricts who can ship goods to Hawaii, thus driving up costs there.
But this is a bit of a non-explanation. Costs have been high in Hawaii for a long time. Why now would they be biting?
Well, here’s ratio between new housing units permitted in the last 5 years in Hawaii, and population change over the last 5 years, going back to 1980–1984:
That’s a big change! Whereas normally Hawaii permitted about 1 housing unit for each 1.5 to 2.5 new people, corresponding to typical household size, that fell sharply to 2000, meaning there was an oversupply of houses being permitted. But then it recovered until about 2006 or 2007. But then, suddenly, housing permitting dropped off a cliff even though population growth was chugging along nicely… and so only 1 house was added for every 3, then 3.5, then 4, then nearly 5 people! The housing shortage allegedly created extremely high homelessness for Hawaii, though that was as of 2007, which is a bit early for this spike.
So that being the case, there must be a really tight market for real estate in Hawaii, right? Here’s Hawaii’s vacancy rate (i.e. the share of housing units with no resident occupying them) compared to the nation on the whole:
Hawaii’s vacancy rate is rising, and above the national average. That is, from vacancy rates, if anything, we would have to conclude that Hawaii actually has a very loose housing market: it’s got a lot of vacant units available! This is not what we see in rapid-cost-escalation places like DC and California, though we do see it to some extent in New York. On the other hand, Hawaii’s real estate is not very much like Puerto Rico’s, another tropical island paradise with declining population.
So what’s going on?
We can look at the characteristics of these vacant units and see the reason for vacancy. Here are the reasons given for each vacancy type:
If what’s really going on is that AirBnB is replacing residential units, it’s not totally clear what category the unit should fall under. Would it be listed as “seasonal use”? Many AirBnB owners use a property for themselves part of the year, and list it on AirBnB the rest. Or would it be listed as “For Rent,” since it is, after all, listed for rent? Or would it be “Other”?
My guess is, there’s no single classification for AirBnB. Indeed, AirBnB is screwing with the data in a weird way: it turns residential property commercial without it ever showing up in housing or construction data as such. I can say this though: Hawaii actually does not have an unusually large share of its housing units listed as vacant due to seasonal or recreational occupancy.
But here’s the remarkable thing: other data confirms the story of a loose Hawaiian real estate market! We can look at the amount of inventory listed for sale (as reported by Zillow) over the last 12 months and compare it to population. Here’s Hawaii vs. selected states and the nation.
As you can see, Hawaii’s housing inventory is actually stable or growing as a share of population and, while it used to have tighter inventories than the nation on the whole, it now actually has bigger inventories!
All of the cities shown here have below-national-average average time on the market… but some are much farther below average than others, and Honolulu, along with Atlanta, is getting much closer to the national average.
So this is odd. The vacancy rate is rising, the relative time a listing sites on the market is rising, the amount of inventory on the market is flat or rising… and yet local prices are rising.
Is AirBnB to blame for this? Well, AirBnB usage could explain a market with high vacancy and high prices, but it seems like smart AirBnB business owners would buy up more houses and list them too. But maybe most AirBnBs in Hawaii are operated by folks who don’t want to manage a large number of properties. Maybe.
One last thought. One way we could look at this would be sub-state variation. Where is population growth slowest in Hawaii? We don’t have county data yet for 2017, but I’m going to take a guess at it based on the 2017 statewide estimates and the method used for back-year revisions, and guess that the population growth for each county from 2010–2017 looked like this:
The weakness is in Honolulu. The outlying islands are indeed having a growth-slowdown, but not as bad as Honolulu. A sidenote: this is a really fascinating case where the core city has great climate, high density, beaches, a progressive government… and is shrinking, while the outlying islands, which can’t properly be considered “suburban” in any sense but are properly rural or micropolitan, are actually growing. Kind of weird. But there’s good reason to think tourism may have agglomerative dis-economies.
Also worth noting: housing costs in the outlying islands do seem a bit lower, according to Trulia’s maps of sale price per square foot.
Let’s turn to a different story in greater detail: college-age migrants.
Kids These Days
The most durable long-run worsening of net migration in Hawaii appeared to be in the 18–22 crowd. Using the Census 2016 population estimates, here is the 18–22 population over time in Hawaii:
That’s a sharp decline to 2016. We don’t have 2017 age data yet, but when we do, it seems likely it will show a continuation of this fall. Likewise, recent years have seen declines in the population between ages 40 and 60, with gains among those aged 23–40, and 60+.
Some of this is just a cohort effect, but much of it isn’t. Here’s the change, controlling for cohort size, of 17–20 year olds, so basically college-freshman-age:
It was growing at a good clip via migration, then the pace of growth began to fall, then it went negative, and since 2012 it has been quite negative. This, of course, includes international migration as well; you can imagine domestic migration being much more severe.
So what’s happening? Well, the first place we should look is in college recruitment data. And thankfully, we can get full-sample recruitment data from all the colleges in Hawaii about where their incoming freshmen lived before coming to college. Here’s smoothed inflows into Hawaii of college freshmen from other states, total smoothed inflows of 17–19 year olds from other states, and the difference between those two.
So inflows of 17–19 year olds fell even as college admissions from the mainland rose. So these are lower inflows of non-college-student 17–19 year olds.
Who could this be?
Well, what sorts of things do 17–19 year olds do?
They work at restaurants (not bars). They work in manual labor jobs throughout the hospitality sector. They work in agriculture. They sit on the beach and surf. These are all basically lower-wage jobs, or hobbies that don’t take much money beyond rent (which is rising!).
So what happened to Hawaii’s low-wage sector? Well, one thing that happened is the minimum wage went up, starting in 2015, though recent research suggests effects from the minimum wage actually materialize as soon as the bill is passed, not when the wages are implemented. In many states or cities with higher minimum wages, it is easy for workers displaced by lost work to find jobs across a jurisdictional line, or the economy is very diversified. But Hawaii has an undiversified economy heavy on low-wage hospitality jobs, and workers absolutely cannot shop across jurisdictions unless they are mer-people. If there’s anywhere in the country that a minimum wage could create a population response, it would be in Hawaii.
Conveniently, we can look at the population of college-age people broken out by school and work status!
The blue lines show employed 17–19 year olds, the red lines non-employed. The dotted lines show school-enrolled 17–19 year olds, the solid lines show non-school-enrolled.
The key to note here is that since 2010 or so, our period of greatest interest, there’s basically no difference in trend between the school-enrolled and the non-school enrolled, when you control for employment states. But for the employed young, population is stable. For the non-employed young, population falls. For some reason, starting around 2011 or 2012, it became harder for young Hawaiians to get a job.
But that’s too early for the minimum wage increase to be the direct cause, though it could be part of the steepening after 2014.
Maybe it’s something sectoral? Well, the first thing to do is to look at the macro-indicators: does Hawaii have lots of jobs generally? It’s unemployment rate is very low, among the lowest in the nation, but that can disguise a lot, so I’ll look at employment-to-population.
So Hawaii is having some very real employment strength since 2009, with the share of the population that had a job rising much faster than in the mainland, nearly reaching pre-recession levels.
But is that because population growth is slowing down, or because employment growth is speeding up? Well, we can look at employment and population growth rates in Hawaii and the US.
A few things are worth noting here. First of all, Hawaiian employment has been more volatile but, broadly, has basically tracked with the US on the whole. Sure it contracted in 2012, but it had seen substantially above-average growth in 2010 and 2011, and again in 2014 to 2016. On the whole, Hawaiian employment growth has done about as well as the U.S. generally.
Rising employment-to-population, and falling unemployment, then, is not about a booming demand for workers. The demand for workers in Hawaii over the last decade has been quite similar to on the mainland. “Improving” labor market conditions reflect a growing number of people leaving the island.
This is also part of what gives farm-states super-low unemployment. If you don’t have a job in rural South Dakota, you don’t stick around to look for one. A reasonably intelligent person in South Dakota can look at their skills and look around them and make a decent forecast of how many job openings there will be in their field for several years. So non-employed people leave South Dakota, giving it and other farm-states very low unemployment (plus, farm labor draws people in).
But if there’s employment growth and a falling supply of workers, wages should be rising, right? Well, here’s average hourly wages for non-supervisory workers in retail and hospitality industries (likely industries for low-skilled workers) in Hawaii, as a ratio of hourly wages for the same on the mainland.
Wages are indeed rising! Hawaii’s sector-specific wages have risen sharply vs. the rest of the nation since the early 2000s.
But what does that look like in real terms? Well, using BEA’s RPPs since 2008, here’s the difference in real wages (before 2008, I back-cast RPP the above-mentioned CPI ratio, calibrated to the 2008–2015 RPP):
You see it turns out, as fast as wages are rising in Hawaii (much faster than on the mainland!), prices just spiral upwards even faster. Real wages in the leisure and hospitality sector have been falling since 2013, and, while retail wages have risen, retail workers in Hawaii still make about 5% less than on the mainland, in real terms. And in 2017, even retail wages fell.
So this is a case where we’re back to housing costs. We tried to escape local cost of living by pivoting to college-age decline… but it wasn’t among college students! It was among non-enrolled young people! So we had to look at their work, and, lo and behold, their employment growth has been more-or-less average, and Hawaii’s rapidly-rising cost of living has gobbled up any wage increases. Indeed, it’s possible that legal wage increases have even accelerated local price increases.
Which Retirees?
Before I conclude, a note on older-age migration. Josh Zumbrun at the Wall Street Journal suggests:
I discussed the college-age group… what about these other groups?
Here’s migration for four groups (smoothed): 60–70 year olds (early retirees), 71+ (very aged), early career (25–35), and late career (36–59).
So it’s hit-or-miss. Migration did fall noticeably among earlier retirees, especially in 2012 and later. So as people plan for retirement and figure out where they money will allow them to live once labor income ceases, they are increasingly selecting away from Hawaii. Very-aged migration became negative for a while, but had a strong year around 2014, which popped it back up, but it has since declined again.
But early-career movers (the teal line) have seen increasingly strong migration since 2008. People in their 20s and 30s are moving to Hawaii more and more. This may be due to job growth, or it may be that, as more people delay marriage and childbearing, their preference are shifting more towards individualistic amenities like scenery and climate, instead of proximity to family, and they may have little need for a large living space. Late-career migration was about stable.
Conclusion
Hawaii’s population is declining. This has happened before, and it always bounced back rapidly, and it may again. However, a rapid recovery is becoming less likely because, this time, Hawaii faces higher mortality and lower fertility than during previous episodes of negative migration. To maintain the same growth, Hawaii needs more and more inflows, because its birth rate is falling. In 2017, Hawaii had its lowest rate of natural increase on record. Furthermore, there is little prospect for international migration to suddenly boom and give Hawaii a boost either.
The proximate cause of Hawaii’s worsening migration is the escalating cost of living. The timing of Hawaii’s outflows is closely associated with worsening real wages and a sudden worsening in the rate of housing permitting compared to population growth. Employment growth rates are now declining, and have converged with the national average. In previous episodes of population decline (late 1990s/early 200s), Hawaii’s job growth rate rose during the years of steepest decline. Today, Hawaii’s job employment growth rate is falling in tandem with population (and indeed, the decline in population growth rates preceded the decline in employment growth rates).
Furthermore, it is difficult to reverse high housing costs. Even if factors driving up prices are alleviated, it may take a long time for mainland costs to equilibrate with Hawaii. Alternatively, a sudden price crash could reset costs, but would probably create a negative local demand shock via wealth effects, only incompletely compensated for by induced demand from the mainland.
As such, this decline is probably not going to be followed in 2018 by a strong bounce-back. I don’t think this is the last year of growth Hawaii is going to see; that’s far too pessimistic for a state that still has a positive balance for natural increase. But I do think it’s likely that the 2020 population will essentially be at or below the 2016 state population level.
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 also a Research Fellow at the Institute for Family Studies, a Senior Contributor at The Federalist, and an Advisor at Demographic Intelligence, the nation’s leading producer of rigorous national- and regional birth and marriage forecasts. I’m married to a kickass Kentucky woman named Ruth.
DISCLAIMER: 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.
