Without power when I took this picture. How many will have occupants at all in 40 years?

How Low Will Puerto Rico’s Population Go?

Yesterday, I was in Puerto Rico, speaking at a listening session hosted by the Fiscal Control Board, which I will hereafter call Promesa, after the law that established it. They had me attend as a speaker on Puerto Rico’s demographic future since, after all, I have written about the topic quite a lot.

But since what I said is probably of some general interest, I wanted to show here what I said there.

Basically, I wanted to look at plausible forecasts for Puerto Rico’s population. So I built a population model for Puerto Rico out to 2060. I benchmarked it to Puerto Rico’s 2011–2016 demographic experience in terms of fertility, mortality, and migration rates. All benchmarks are individual age-and-sex-specific inflow/outflow/fertility/mortality rates. I also subdivide inflows between international inflows, inflows of Puerto Ricans from the mainland Puerto Rican population, and other inflows from the mainland. I also enable separate parameters to create separate shocks for different age/sex/variable categories (so for example I can easily input a shock for “Mortality for men under the age of 35” or “International inflows of women over age 22”, etc). I also model the Puerto Rican population on the mainland using reported ancestry data and an assumption about identity-drift over time.

For my baseline forecast, I assume that the 5-year cumulative effect of Hurricane Maria is 220,000 worse net migration via higher outflows and reduced inflows. Here’s my baseline:

As you can see, Puerto Rico’s population rose steadily for most of the 20th century… but in the 21st, it has had severe struggles. My baseline assumption assumes that by about 2050, net migration balances for Puerto Rico, it assumes fertility rises from its current lowest-low levels of about 1.4 births per woman, and it assumes that age-specific mortality declines too. So I’m not being horribly pessimistic here. I’m actually assuming things get better!

So let’s run some scenarios. First off, let’s simulate Hurricane Maria being much worse than I expect: net losses of 500k. And then let’s also simulate it being much better than I expect: net losses of 50k. And let’s also simulate the effect of a second Hurricane Maria hitting Puerto Rico in the 2030s. It’s important to simulate future shocks because most experts believe that hurricane intensity in the Atlantic basin is increasing, and some believe hurricane frequency is also increasing, and thus any responsible forecast needs to account for higher odds of hurricane disasters in the future.

As you can see, even if the impact of Maria over 5 years is the absolute smallest that anyone would responsibly forecast, Puerto Rico’s decline remains severe. A much weaker Maria raises my forecast of 2020 census results 6.7%, and raises my 2060 census results 9.1%, so it’s not nothing, but the point is decline remains “baked in.”

A strong Maria effect, meanwhile, is pretty significant. Puerto Rico loses 9.1% of its population vs. baseline as of 2020, and is 20% below baseline in 2060. And worse even than this is if a second major hurricane strikes. When Hurricane Lymanita strikes Puerto Rico in 2030, population takes another shift down, and Puerto Rico ends up a solid 25% below baseline.

I also want to note two other lines there. The dotted gray line is what my model forecasts if I remove all Maria-specific shocks. And the blue line is the official U.S. census bureau population forecast for Puerto Rico released in August 2017. As you can see, Census and I have calibrated our models *very* similarly. I’m not using some fantastical parameterization that I just made up on the fly; I’m building parameters in roughly the same way Census does. And for the record, I did not look up Census’ most recent estimates until *after* I had fully built my model and all of the forecasts shown here, so I didn’t cheat and try to hit Census’ forecast. I independently confirmed it for the pre-Maria population conditions. But post-Maria is different.

When we compare our hurricane effects to my pre-Maria baseline, we can see that hurricanes Maria + (potential) Lymanita reduce Puerto Rico’s 2060 population vs. pre-Maria baseline by somewhere between 9.5% and 38.2%. That is a very big effect. Hurricane Maria is a game-changer. Worth noting as well that my estimates compare favorably with benchmark forecasts of long-run GDP output loss vs. pre-Hurricane benchmarks. Those experts linked say Puerto Rico can expect to underperform GDP per capita trend by ~21% by 2032. I say Puerto Rico can expect to underperform in terms of total population vs. baseline by 9.5% to 38.2% in 2060, or 2.3–16.1% by 2032 . Add those together with a 10 percentage point error band around the GDP per capita estimate, and you get…

Estimates of Puerto Rico’s total GDP in 2032 should be revised down by between 12% and 41% vs. pre-Maria baselines.

Ouch.

Okay, let’s look at other scenarios.

Let’s start from the bottom. What if after Hurricane Maria, things spiral out of control? What if relief funds are misused, Puerto Ricans just keep fleeing, and broadly speaking living conditions on the island enter a death spiral, falling to levels seen in other nearby Caribbean countries? What happens then? Well, in this “migration never stabilizes or gets worse” scenario, with long-run net migration running at ~-2%, population falls to 700–800k by 2060. By 2060, Puerto Rico would have fewer people than it had in 1899. That’s insane. This kind of scenario is extremely, extremely unlikely. Then again, mass depopulation events have occurred in human history. It’s worth noting here that if Puerto Rico had Hawaii’s population density, it would only have about 780,000 people. In other words, there’s a case to be made that the stable Jones Act-compliant islander population really might be vastly lower than currently expected. Hawaii is further from the mainland or other population hubs, which should reduce its stable population, but it also has a huge military presence, which should increase it.

Next up, what if migration stabilizes at about -1%? Well, population ultimately falls below 1.5 million. Again, this is worse than all but the very worst hurricane scenarios.

But what about optimistic scenarios? I assume migration balances by 2050 or 2055 in my baseline. What if it balances in 2035, then remains around 0.5% afterwards? Well, population still declines, but not nearly as fast.

And finally, what if Puerto Rico got hundreds of thousands of immigrants, and then had a permanently higher baseline immigration rate?

Well… population still declines. This one blew my mind when I modeled it yesterday after my speech. A big immigrant boom of several hundred thousand immigrants could stabilize population out to 2030… but then decline sets in again! More on why that is later on.

Let’s consider some more modest, plausible, policy-relevant scenarios.

Let’s start from the bottom again. The first scenario is a dark one: what if the opioid epidemic hits Puerto Rico as hard as it hit Appalachia, raising mortality, lowering live births (largely via higher infant mortality), depressing inflows, raising outflows? Well, it turns out a plausible forecast of a severe opioid epidemic impact is that it could be nearly as bad as a major hurricane. Now I will admit, I’m aggressively and personally concerned with the opioid issue, so I’m probably overstating the impact. But you can consider it to be less a central estimate of “drugs get worse” and more a reminder that “there are so many other bad things that can happen.” Puerto Rico has seismic activity! Tax reform might be bad for investment in Puerto Rico! Spain could attempt reconquest! Okay I’m joking obviously on that last one but the point is it’s a big, scary world out there. Always gotta keep an eye on downside risk events.

So what about upside events? Well, what if Puerto Ricans on the mainland became twice as enthusiastic about returning to Puerto Rico (i.e. the rate of US->PR moves as a percentage of mainland Puerto Rico ancestry individuals doubles)? Well, population is… a little bit higher. A little bit. The point is, encouraging return is not likely to greatly alter Puerto Rico’s population trend at plausible levels of return. It would take a really seismic, structural shift to enable returnees to boost population.

What about a baby boom? If total fertility eventually rose to 2.7 kids per women, making Puerto Rico the highest-fertility place in the US (vs. today, when it is the lowest fertility place), population would indeed be higher. But again, not a lot higher.

Next up, what if Puerto Rico got a special immigration program rolling that enabled them just to match mainland-US rates of international inflow? What happens then? Well, population is higher… but still trending down.

And what if we get a returnee boom, a baby boom, and an immigrant boom, all at once? Significant improvement in trend! But… still down.


Why is Puerto Rico Declining?

Everybody talks about emigration. But it’s actually not emigration that’s killing Puerto Rico. Or, rather, it’s not just emigration.

Here’s a map of the 50 states showing the difference between their total fertility rate and Puerto Rico’s.

At 1.43, Puerto Rico’s total fertility is insanely low. It’s what we call “lowest-low” fertility, and it has enormous negative population growth effects. In fact, declining fertility is a huge reason for Puerto Rico’s severe population inflection: as of 2000, the island had similar or higher fertility than the mainland US. Not so anymore. Today, Puerto Rico even has lower fertility than communist Cuba: and when you’re behind your Soviet-space neighbors, you’ve really gone low.

That’s not all. We can look at Puerto Rico’s median age vs. other states too.

Pro-tip: population forecasts for Maine, Vermont, New Hampshire, and West Virginia are not super impressive either! When you’ve got lots of old people and low fertility, the result is that, yes, your population forecast slopes downward.

And what about migration?

Well, here’s estimated total net migration difference between Puerto Rico and other states in 2016, a very bad year for Puerto Rico.

Nobody has net migration even close to as bad as Puerto Rico.

But what’s driving this? According to all the news reports, you’d think this was driven by high out-migration. So. Is Puerto Rico’s out-migration higher than for other states?

Turns out Puerto Rico’s out-migration rates are quite normal. Most states have higher outflow rates than Puerto Rico does, and the only ones with lower ones are very big, populous states. Puerto Rico’s outflows are lower even than those of the other geographically separated areas like Hawaii and Alaska!

Now, take this with some salt. I have to derive outflow rates from nets and inflows, and there may be some errors. Plus, I think the 2016 net estimate for Puerto Rico is too favorable. Probably outflow rates are a bit higher than the figure I use here. But even if outflow rates in 2016 were lots higher than implied in the data, Puerto Rico’s outflow rates still would not be unusual.

So why is net migration so low?

It’s all about inflows.

MAP

Puerto Rico’s gross inflow rate is under 1%. That’s insanely low. It’s immigration rate is under 0.1%. When I first cut the data, I actually ran back and looked for errors in my spreadsheet because I simply did not believe inflows could be that low. But they are. Puerto Rico’s inflows are spreadsheet-bustingly-low.

This makes sense given a comparison of Puerto Rico to other U.S. states: it’s poorer and has fewer job prospects. And for all that “climate” amenities can be nice, “hurricane” disamenities are very much not nice.

So when you think about Puerto Rico’s population decline, outflows, the thing everybody wants to talk about and that make for sexy news stories, should be secondary. Fertility, aging/mortality, and lowest-low inflows are the real story here.


Some Comparisons

Regular readers know I like to find exotic comparisons for population scenarios. For Puerto Rico, I compared its experience to Ireland and Corsica. Both are islands that experienced severe population collapses and had shifting political relationships with a sovereign nation over them, so they make really neat case studies for Puerto Rico. Here’s how the benchmarking plays out:

Plausible estimates of Puerto Rico’s population future suggest it will be worse than what Ireland experienced in the century after the potato famine in the 1840s. We ran through a lot of scenarios for Puerto Rico and while many did better than the Irish case, nont came close to matching the Corsican experience… why? Well, Corsica was resettled by a wave of tens of thousands of relocated French North Africans after decolonization. It received a migrant shock equal to 10–20% of its population, and they totally reshaped the island; plus France has higher fertility than Puerto Rico. Puerto Rico has a faster-aging population, so to get a Corsican-type experience, it would need at least half a million, maybe nearer a million, immigrants within a few years to get that V-shape. That ain’t gonna happen.


Conclusion

Puerto Rico’s population forecast is fairly dire. I bring no good news. The only optimistic notes I can offer are these: Japan’s experience suggests that population decline can be managed with a degree of decorum if a society works together, invests heavily in productivity-enhancing measures, and opens itself to immigrants as needed; but of course, Puerto Rico’s decline is far more severe than Japan’s. Corsica’s case shows that bounce-backs are possible; they do happen sometimes. I could also have shown the Faroe Islands for a good example. But the bigger the region, the harder it is to bring about a population snapback, and Puerto Rico is very policy-constrained on that front since they cannot control their immigration policy. Overall, I think this forecasting exercise made me more pessimistic about Puerto Rico’s future because, although decline-reversal or graceful-decline are possible, the conditions for them seem unlikely to be met in Puerto Rico.

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.

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.