Immigration, Caplan and Buckets
If you prefer to listen rather than read, this blog is available as a podcast here. Or if you want to listen to just this post:
One of the books I read and reviewed in October was Open Borders: The Science and Ethics of Immigration by Bryan Caplan. In that review I asserted that it was possible completely open borders is a great idea, when you consider the world as a whole. But that you’ll never get people to agree to it because, like Communism, it requires a level of selflessness which most humans simply don’t possess. There are actually lots of reasons for this, but I paid particular attention to the idea that almost no one would vote for unlimited immigration if they thought it was going to reduce what they got paid. And that further, this wasn’t just some irrational fear (although that would probably be sufficient by itself), but that even using Caplan’s own numbers, this was likely to happen.
When I posted that review in various places, including to Caplan’s Twitter account everyone accused me of being an idiot (okay maybe the language wasn’t that strong) and ignoring the Arithmetic Fallacy. In the interest of full disclosure I did understand the Arithmetic Fallacy, but I admit to not fully understanding the totality of their argument. Now that I do, I’d like to revisit things.
To begin with one thing that no one seems to argue about is that the average GDP of the US would drop. Most scenarios have it being cut in half. The Arithmetic Fallacy comes into play when you assume that this means that the average salary of current workers would also be cut, though perhaps not by half. In reality the salary of current workers could go up. Here’s the example Caplan uses in the book:
Average native US income before open borders: $50k
Average foreign income before open borders: $5k
Average US income after open borders: $40k (down from 50)
Average income of original workers after open borders: $60k
Average income of new workers after open borders: $20k
I never questioned this math. I always understand how the fallacy works. But this is a fairly simplistic version of it. For example it assumes an equal number of current workers and immigrant workers, but it could be a lot more or a lot less. Caplan seems to imagine that the more the merrier, because the secret of mass consumption is mass production, but it’s not clear how those numbers affect things in practice, particularly if they increase very rapidly. But that’s a minor quibble, my big issue is with the way that he sticks current workers and new workers into two entirely separate buckets. Because, while the Arithmetic Fallacy illustrates that the incomes can go up for everyone, while having the average go down at the same time, it doesn’t necessarily mean that they will. Caplan argues that we don’t have to worry about that possibility because current workers will always be ahead of the new immigrants. That’s the part I didn’t fully absorb when I read it through the first time.
Let’s take his example of NBA players and preschoolers. He starts the example by asking us to imagine a room full of NBA players with an average height of 6’7”, and then asks us to imagine that a class of preschoolers enter. The average height of the room drops, (say to 4’10) but the pre-schoolers aren’t making the NBA players shorter.
This is obviously true, but if the NBA players represent current US workers, (which they do) and the preschoolers represent immigrant workers (which they also do) and height is a proxy for GDP (again, that seems to be what he’s going for) then while the preschoolers don’t make the NBA players shorter, he is claiming that the presence of the preschoolers metaphorically makes the NBA players taller, and vice versa, just by being in the same room. I understand his argument for saying that this is the case for GDP even if it’s not the case for height, but you can already see where the analogy breaks down pretty quickly. But it is useful because it illustrates one of his central assumptions, that under open borders there will be two buckets of workers: the original workers and the new workers or the NBA players and the preschoolers.
NBA players aren’t threatened by preschoolers, that would be ridiculous. Even if we’re just talking about height and not ability with basketball, the preschoolers are never going to be called on in place of an NBA player, the very thought is risible. And certainly an NBA player is never going to be mistaken for a preschooler. Accordingly it would be just as ridiculous for the current workers to be worried about being interchangeable with the new workers. Just as NBA players will always be in a separate category as far as height, current workers will always be in a separate category as far as income. Current workers will definitely be in the bucket with the average income of $60k, not the bucket with the average income of $20k.
This assumption of there being two buckets is what I have a problem with, because it seems entirely too neat and clean. “The average salary of the US is going to go down, but don’t worry because as a current worker you’re always going to be above average.” But why would this necessarily be so? Certainly it’s not a law of nature. Caplan offers a few reasons for why this would be. Current workers are more skilled, have a better grasp of the modern world, and above all they are native English speakers. To this I could also see adding a better network, and more beginning assets. But however great these advantages are it doesn’t feel like they’re unassailable or permanent. It would appear to me that the buckets are more permeable than Caplan lets on.
One reason for thinking this is the book itself. While from an income perspective Caplan seems to regard the divisions as relatively unchanging and immutable. From every other perspective his argument seems to be the opposite, that immigrants aren’t that much different from the people who are already here. That within one generation their English is almost as good as the people who were born here, that culture takes centuries to “spread by persuasion, but only one generation to spread by immersion”, and that “the average immigrant is [only] microscopically more liberal than the average native.” So if they’re so similar on most of these metrics, why would the be so dissimilar in the salaries they can command? Why would the immigrant average be ⅓ that of the native worker? And how is that none of the natives, no matter how low-skilled, never end up in the lower paid bucket? Actually, I’m not suggesting that this is Caplan’s claim, but before I get to that discussion let me propose a different metaphor.
Imagine that instead of talking about NBA players, preschoolers and their differing heights, that instead we use the example of adults and teenagers and their relative earning potential. Say we have a company that only hires people over 18, and after a policy change they’ll hire anyone 14 and older. And that after this change, the company can use teenagers to double its workforce.
This would appear to be a much better analogy than Caplan’s. To begin with I imagine that there are two very clearly defined buckets when you’re talking about teenager salary vs. adult salary. And it’s also possible to imagine that mostly Caplan is correct, that the teenagers make more by working in an adult workplace, and the adults are more productive if they have teenagers around to offload stuff to. But the gap between the two is far more permeable than the gap between NBA Players and preschoolers. You can imagine right out of the gate that at the lower end of the adult skill range that you might have lots of teenagers that also have that level of skill. That 17 year olds are pretty similar to people between 18 and 21. But, perhaps most importantly, teenagers grow up and with each passing year the difference between the teenager bucket and the adult bucket narrows.
I suspect there’s also issues of supply and demand at the lower level of skill. That if you have a janitor making 30k a year, that adding a bunch of teenagers, all fighting for the same low-skilled jobs, lowers that to maybe 15k a year, even if the teenagers increase the salary of the engineers at this company by quite a bit.
I also suspect that it increases the salary issues at the upper end as well. If you have some high paid developer making twice as much as a younger developer, but who’s actually less productive (say more skills, but less willingness to work 60 hour weeks) the fact that you’ve now got a deluge of young people all looking to be developers has to factor into that.
Perhaps I’m wrong about the last two points, Caplan would seem to be arguing that I am, fair enough, but what he does argue, unmistakably, is that the average salary of the company will be cut in half. Is he also saying that despite this massive reduction in average salary that none of the adults currently working at the company will end up in the same bucket with the teenagers? Will end up with one of those salaries that caused the average to be cut in half? Even if we look ahead five years? Or ten?
(One mechanism whereby this could happen is ageism. The parallel for immigration would be racism, which I assume Caplan is against.)
I assume that this is not Caplan’s claim, if it is, well then I guess the math works out, but I think he’s wrong. If it’s not his claim, then the question becomes how many adults end up in the “teenager bucket”, or rather how many current workers end up in the immigrant bucket. Let’s review his numbers.
Average native US income before open borders: $50k
Average foreign income before open borders: $5k
Average US income after open borders: $40k (down from 50)
Average income of original workers after open borders: $60k
Average income of new workers after open borders: $20k
Is there a cohort of low-skilled American workers who were in the $50k bucket, but, after immigration, end up in the $20k bucket? Is there a cohort of current workers whose salary is going to drop in an open borders world? Caplan talks about how current workers are “normally going to be managing and training the new arrivals.” How many adults do you know that aren’t fit to train or manage anybody? Could that be the cohort I’m talking about? If we’re agreed that there are some people who are, unfortunately, going to be below average, and when the average salary drops their salary drops as well, then the next question is how many?
The answer to that is unclear, but I’m willing to bet that the number would be significant, and more importantly, even if the number isn’t significant, that a huge number of people are worried about this very thing, and probably understandably so. If you doubt this latter assertion I direct your attention to the 2016 election.
Pulling all of this together, if someone is opposed to immigration because they think they might make less money, then it’s only reasonable to call them irrational if that never (or very rarely) happens. Otherwise I think it’s a rational fear. If a reduction in salary happens to some people, but is unlikely to happen to him you may call him a pessimist. But once again, he is not being irrational. In his own way, he’s hedging against the small chance of a very large harm. Particularly in light of the fact that we still live under capitalism where money is pretty central to everything.
None of this is to suggest that open borders isn’t a terrific, world changing idea for the vast majority of people. Or that it isn’t a fantastic moral good which all men aspiring to any degree righteousness should support. But I am suggesting that it would not be unreasonable for some people to conclude that they would be voting against their own interests by supporting it.
This is where the selfishness I talked about in my original review comes into play. And the reason I connected it to communism. In both cases people are promised a world much better than the one they currently live in, if they can just abandon their baser emotions of greed and selfishness. Now it’s reasonable to ask whether communism failed because of these baser emotions or whether it failed because it lacked the means to effectively produce the right goods and services, or whether it failed because Stalin and Mao were particularly ruthless tyrants. But the fact that it expected people to abandon their baser emotions certainly didn’t help.
It’s also reasonable to ask what degree of abandonment is required. In theory you might argue that communism requires a complete and total abandonment of the baser emotions of greed and selfishness, while open borders only requires a small abandonment of these emotions, which are in any case irrational. And perhaps this difference will be enough for Open Borders to succeed where communism didn’t, but I suspect that it won’t be, and I suspect that Caplan’s keyhole solutions (restricting benefits to citizens, delaying citizenship, charging people to enter) have a much better chance of changing people’s mind that telling them that they’ve fallen prey to the Arithmetic fallacy. But given that this is the territory we’re fighting over it is worth trying to get to the bottom of the question: how many people are legitimately entitled to that fear, and how many people are truly being irrational?
As far as I can tell from the book Caplan is arguing that all or very nearly all of these people are being irrational. That if everyone in the US knew the facts that they would embrace open borders, both on humanitarian grounds but also because it would add trillions of dollars to the economy and raise everybody's wages.
My argument has been laid out in this post. Are there other people arguing that Caplan is wrong as well? As it turns out there are.
Garett Jones, Caplan’s colleague at GMU, published a working paper arguing more or less the same thing I am.
How would Open Borders—a policy of unlimited immigration—change the wages of current residents of the United States? To answer this question, I begin by running the same quantitative experiment that Caplan runs on page 131 of his graphic novel Open Borders: The Science and Ethics of Immigration. This experiment presumes that the only two drivers of national income per capita are national average IQ and an unexplained productivity residual.The unexplained productivity residual plays a key role in the case for Open Borders, and I critique that residual. I use the same constant returns to scale framework as Caplan, in which the migration of every human being to the United States would increase global output per capita by about 80%. I then estimate that in the benchmark model, where IQ's social return is much larger than its private return, the per-capita income of current U.S. residents would permanently fall by about 40%. This is not an arithmetic fallacy: this is the average causal effect of Open Borders on the incomes of ex-ante Americans. This income decline occurs because cognitive skills matter mostly through externalities:because your nation's IQ matters so much more than your own, as I claim in 2015's Hive Mind. Therefore a decline in a nation's set of average cognitive skills will tend to reduce the productivity of the nation's ex-ante citizens.
I was going to essentially end with this but I just barely saw Caplan’s rebuttal. And here’s where it gets a little bit tricky. Wading into any long standing argument, let alone an argument between two colleagues, runs the risk of missing all manner of important points which by this point are part of the assumed and unspoken foundation of the argument. Nevertheless despite this risk I’m going to dip my toe into things, mostly because I think it reveals some very interesting things about Caplan’s argument and my problems with it, but first Jones’ argument.
As was mentioned in the quote I included, Jones wrote a book in 2015 called Hive Mind: How Your Nation’s IQ Matters So Much More Than Your Own. (If you’re curious, here's a review.POST) And as you can tell from the title of his book, his argument concerns the IQ of the entire nation, which, Caplan conceded, would probably be lowered under Open Borders. And since, according to Jones, the income of individuals depends almost entirely on the average IQ of the nation as a whole, this leads to his claim that the “per-capita income of current U.S. residents would permanently fall by about 40%”. I’m sure I’m vastly over-simplifying his argument, but since I’m mostly interested in Caplan’s response to that argument, I don’t feel there’s much point in getting deeper into the complexities.
Caplan’s has an interesting response. First he points out that Jones largely agrees with him about the overall increase to global output, pegging it at 80%, but that in order for this to happen, while still decreasing the income of current residents by 40%:
Garett argues that more than 100% of the gains will go to immigrants! So even though open borders nearly doubles the production of mankind, it reduces living standards of the current inhabitants of rich countries by a massive 40%. [Emphasis original]
First off, and maybe I’m missing something. Is there any possibility that the gains and losses are distributed unequally? That, as I argued above, some of the current workers will drop from 50k to 20k, a reduction of 60%, which is more than Jones’ 40%. While many or maybe even most will see their salaries actually go up? This seems like another example of Caplan having two very well defined buckets. With gains being distributed equally to each bucket. But gains could be unequally distributed to the buckets and within the buckets. But beyond all that, here the debate takes a surprising turn. Caplan continues:
How is this possible? Drawing on earlier work, Garett insists that the personal payoff for IQ is modest. 1 IQ point raises earnings by about 1%. Since current U.S. IQ is about 11 points above the world average, the current citizens of rich countries will end up earning roughly 60% of what they now earn. In other words, Garett’s concern is that under open borders, income will be too equal for current residents of high-IQ countries to maintain their standard of living.
The surprising bit, is that as a retort to Jones, Caplan starts arguing for the importance of individual IQ and that gains from it are probably significantly higher than 1%. If you think that the natural consequence of this is greater inequality across the board, then you’re not alone, that was also the thought that occurred to me and Caplan agrees:
Would IQ have a big effect on personal economic success under open borders? Would there be high inequality under open borders? If you answered Yes to both questions, you should be on my side. [emphasis original]
I would answer yes to both of those questions, and as hard as it may be to believe I am largely on Caplan’s side, but having answered yes, I then have to wonder if there’s large amounts of inequality and IQ has a large effect on success then why does it not follow that some of the current workers with below average IQ would end up in the below average/immigrant/20k salary bucket rather than the 60k average salary bucket?
I realize that it’s really way too late to bring in the idea of a normal distribution, and that also salary isn’t a normal distribution. (Though it’s more of one at the low end than the high end.) But to put it another way, if someone is three standard deviations below the mean, and the mean drops, it’s conceivable that their salary is going to drop as well.
To conclude what has ended up being a much longer and more rambling post than I initially intended. My argument was never with Caplan’s moral or economic claims, and I am personally in favor of revising immigration policy around the keyhole solutions Caplan advocates. It is probably politically impossible even so (and definitely impossible without such measures) and one of the things that makes it difficult is that, similar to communism, it’s advocates expect that people will be able to shed a host of baser emotions like greed and selfishness. The argument that such baser emotions are irrational, is going to be largely ineffective even if it were true, but I’m not sure it is. There’s certainly ample room for uncertainty.
Perhaps you’re also uncertain about whether to support this blog? Well may I suggest a keyhole solution? Start with a dollar a month. You know donation is the humanitarian thing to do.