Which are the most skill-selected immigrant groups?

Into the United States that is:

Overall, the relationship is strong and positive (r = .56, p < 0.001): immigrant groups that are more skill-selected tend to have higher average incomes. The five most skill-selected groups are: Taiwanese, Nigerians, Swedes, Indians and Swiss. The five least skill-selected groups are: Mexicans, Salvadorans, Hondurans, Portuguese and Cape Verdeans. For example, 82% of Nigerians are high-skilled, while only 4% are low-skilled. By contrast, only 14% of Mexicans are high-skilled, while 57% are low-skilled.

Methodological caveats: I was unable to match a number of the ancestry groups (e.g., ‘Hmong’, ‘Jewish’, ‘Cajun’); the income data are not adjusted for household size or reporting bias.

That is from Noah Carl, here are his other essays.  For the pointer I thank Dan Klein.


I.e., it's hard to be skill-selective among people who can just walk across the border.

We should instead select for Mexicans who have 31 foot ladders

Maybe the wall should have a gate in it that only lets you through if you can demonstrate English proficiency and a high IQ on automated tests.

I think we have a general finding now that what matters isn't the fitness of the source population, but the strength of the selection filter?

If highly skilled migrants leave shithole countries, won't that have an ever increasing negative effect on the long term prospects of the shitholes and those who remain there? In whose interest is it to take the cream? You can't take all the population.

According to the IAB Brain Drain dataset, more than 50% of high-skilled persons born in some of the world’s poorest countries are living in the OECD:


Depends a bit on how much you buy that they're highly skilled because of Greg Clark reasons (similar underlying inequality) and this will persist. Alternate theories may include, say, that Nigeria is a corrupt hope where the people with formal qualifications are mostly the children of those who have cheated the poor out of land and no one much will miss them.

It's perhaps a little like asking if Australia's status as a penal colony long term biased down the prospects of Australians who today have 1/5 ancestry from this group. Possibly not, if this was selection was mostly unjust at root (Victorian rough justice) and didn't really select against the qualities that made Europeans successful and, say, Africans not.

Nigeria is a corrupt hope where the people with formal qualifications are mostly the children of those who have cheated the poor out of land

I don't know much about Nigeria, but in India, people who own land (whether they inherited it or got it through cheating) have little incentive to get formal qualifications. Indians who try to get education and go into the professions are overwhelmingly drawn from the ranks of the landless and resource-less middle classes.

Well. Again it depends on caste.

Landed brahmins or Kayasths more likely to value education despite the landed wealth, while Landed backward castes don't value education anywhere close to the same extent.

Sure, my comment was a generalization. But at least in the south, Brahmins probably have miniscule representation in the landowing class, if my own family history is a guide (from what I've heard, people were "compelled" to sell off their land during the land-reform periods of the '50s.) You are likely to be much better informed than I am on these issues though.

In Punjab / Rajasthan, I like to contrast Khatris and Rajputs. Both with Kshatriya pretensions. Khatris value education a LOT, atleast in modern times. They are very enterprising and eager to fit into the modern economy. While Rajputs don't value education much and tend to be laggards in academics and somewhat downwardly mobile in recent decades.

You're giving me the impression that there's a direct correlation between "education" and developed skills. There ain't. A person with a Phd in art history is nowhere near as skilled as a plumber. A mid-level administrator with a master's degree in the opportunity compliance department of a state university has to call a skilled person when his furnace takes a dump. The normal Mexican hombre is so unskilled that he's forced to do the brakes on his own jalopy. The owner of the lawn service company he works for has to send his new diesel crew cab 4x4 to the dealer for service.

@chuck martel:

I think you may be confusing opportunity cost with lack of ability.

When the MR commentariat mentions "education", they're speaking of formal, institutionalized education. In reality, every normal human receives education every day through life experience. Learning is the acquisition of knowledge, regardless of its source. It's true that a Somali goat herd hasn't had the opportunity to become educated in subjects that might be useful to a hyper-capitalist corporate/consumer society, ergo that might well make him less of a human being, with a smaller "economic potential", than a Danish engineer. His education, and knowledge, is simply in a different sphere than that of the engineer.

No lack of skills or education or resources makes one "less of a human being" than another, and I never meant to imply any such thing.

You may have misunderstood my earlier comment. I meant to say that the people who outsource their car servicing or plumbing to workers with the requisite skills are not necessarily doing so because they have no ability to fix those things themselves but that there's an opportunity cost for them to do so. Their time is more better spent in doing the work they trained themselves for. That may be "office" work, with results that are not observable the way a plumber's or a car mechanic's are, yet require more training and experience than a plumber's skills and have more market value.

The orthographic conceptualists who make their living by thinking, planning, recording, etc. accomplish nothing until their ideas are translated, often literally, by object manipulators. The ideas of people like Einstein, Hawking, and so on are just so much neural activity until a greasy Neanderthal moves something from point A to point B. It's not a question of opportunity cost. The orthographic conceptualists generally don't have the ability, the skills, to repair their own car, build their own house, or even raise their own vegetables. Certainly, one of them designed the car that they can't repair but nothing was achieved until a collaboration of non-intellectuals put the thing together. For every significant idea that an orthographic conceptualist puts forward, scores more are not only insignificant but would be disasters if adopted. An object manipulator can normally only foul up the things that he actually touches. Orthographic conceptualists can ruin the lives of millions.

I more or less agree with Chuck here.

Besides, let's look at the opportunity cost business from the other side. If the mechanic who fixes the surgeon's car gives the surgeon more time to do critical operations, why shouldn't the mechanic get some of the credit? They're a team, after all. Between them they need to fix the brakes so the car doesn't kill someone, and take out a ruptured appendix so it doesn't kill someone else. So they split the work up sensibly.

I'm equally anecdotal here (little knowledge of the Nigerian class scene), but hard for me to believe that land->wealth->money to achieve formal education isn't a significant thing.

I can't believe that formal education in countries that are poor enough to lack much government provision, isn't mostly dependent on wealth, and that wealth in corrupt countries has a bigger slope on IQ and talent than it has in the West (where wealth and IQ slope is notoriously weak, and income is fairly weak anyway (https://thesocietypages.org/socimages/2008/02/06/correlations-of-iq-with-income-and-wealth/).

Main point being, even if we select for education in poor countries, we're probably not much selecting for IQ or "national prosperity" traits (the sort of thing deep roots); more like some kind of mix of a low level of effective selection on IQ, and a higher degree of selection on a wider range of class climbing traits, including a good dash of arbitrary unfairness. (As an analogy, if we killed off or otherwise removed the entire top 20% of society, that would have an impact on society's overall upper class mentality - the overall composite of psychological traits that make social elites different - far larger than on intelligence. "Draining the swamp"). Brain drain is unlikely to be a serious issue.

I can’t believe that formal education in countries that are poor enough to lack much government provision, isn’t mostly dependent on wealth

You may not be able to believe it, but that just reflects on your lack of knowledge and imagination (sorry to be so blunt.)

It is possible for poor countries to carve out sectors of excellence that run on a meritocratic basis, even if the rest of society sees no improvement whatsoever. The Soviets had a seriously high-quality educational program for smart STEM students. Closer to home (for me), the Indian Institutes of Technologies (IITs), at least until the turn of the century had a completely meritocratic admissions process and the students who got through that rigorous challenge had to pay literally peanuts for the privilege of an IIT degree that was one of the top ways to social mobility. I attended one of these IITs in the late 90s (personally I was from a proper middle-class family by Indian standards), and there were a LOT of students who came from very modest backgrounds but were crazy smart. And almost everyone who goes through that process ends up doing well for themselves in life, whether in India or abroad.

The quality of public secondary education in India is shockingly bad. Shockingly bad - see the PISA. Judged from every formal metric. Wealth and educational divides are fairly vast. Despite all this, measured, IQ difference between different social caste groups are almost non-existent - https://twitter.com/whyvert/status/757792891519369217.

And you want me to believe that with all of this graduates from the IITs have a purely meritocratic "crazy smart" natural intelligence advantage, and are generally from striving, poor, modest backgrounds? Perhaps if you get into the school the tuition is low, but India is not a society with high education based social mobility, and the thresholds are significant. Sorry, but, I don't believe you have any "natural merit" - I'm sure you're well educated, but you are unlikely to represent some inherent human capital advantage that will be lost to your nation if you emigrate. There's too much noise. To be blunt, your reply simply seems nothing more than self satisfied onanism.

Further, even if you did have some "merit", how representative of the Indian educational elite are the IITs? They probably account for a very low share of both the higher education system overall and of highly skilled Indian migrants to the West.

I very specifically stated in my earlier comment that institutions like IITs were exceptional, and not representative of the norm. You wrote 3 paragraphs criticizing the exact opposite of what I said. Your reading comprehension skills leave a lot to be desired.

Sorry, but, I don’t believe you have any “natural merit” – I’m sure you’re well educated, but you are unlikely to represent some inherent human capital advantage that will be lost to your nation if you emigrate.

LOL. So it's cool if I emigrate, right? You'll welcome me with open arms?

To be blunt, your reply simply seems nothing more than self satisfied onanism.

I'm not sure exactly what you are criticizing here. I just stated a particular very well-known fact, that the IITs are flagship universities in India that have a rigorous and meritocratic entry process. If you want to challenge that fact, you'll have to do better than produce half-baked rants based on whatever caricature you hold about third-world societies. (I did not express any self-satisfaction; that was purely your inference. I did praise a number of my fellow students, and I stand by that opinion.)

The quality of public secondary education in India is shockingly bad. Shockingly bad – see the PISA. Judged from every formal metric. Wealth and educational divides are fairly vast. Despite all this, measured, IQ difference between different social caste groups are almost non-existent – https://twitter.com/whyvert/status/757792891519369217.

Ahh.....IQ and Twitter, the last refuges of the alt-righter. Immigration is so threatening to you that you must completely destroy those immigrants' natures, their characters, and their home countries. Now if you actually had even basic scientific training, you would understand that two data points (which is what the PISA scores provide) don't translate into a theory. But that's too much to hope for, when even your reading comprehension is sorely lacking.

The nature of these "high" skills matter. If a *hole country produces a lot of programmers which it doesn't need, then their exodus will be beneficial to them and to their home countries (through remittances) without causing an harm to those countries. Someone trained as a programmer will likely suck at (and be miserable at having to) building functioning political institutions, roads, sewers, etc., which would actually make those countries better places to live.

On the flipside, the potential for emigration to the USA can create a really perverse set of incentives in these countries. The smartest kids may train in jobs for which there is no local market because they hope to get a leg up in the green card line. Then there will be few people existing (or smart enough) to help build the kinds of things that will produce real benefits to their societies.

An odd "if". Its seems unlikely that a country would produce a lot programmers which it doesn't need. But, then again, US universities have been manufacturing battalions of social workers and criminologists and special education teachers whose contribution to society is dubious at best. More mysterious is why there is a lack of university programs in the prison guard field. Those skills should be taught in high school, like metal shop once was.

Its seems unlikely that a country would produce a lot programmers which it doesn’t need.

I was talking about India. At any given time, it seems most of our programmers (definitely our best ones, despite what the disgruntled folks on the Internet believe) work either in America or for American clients.

Technology and automation isn't ubiquitous in India itself (nowhere to the extent that it is in the US), so this is not surprising.

Shithole Switzerland?

"The five least skill-selected groups are: Mexicans, Salvadorans, Hondurans, Portuguese and Cape Verdeans."
Yet, Cape Verdeans (actually, the correct name is Cabo Verdeans) are morally superior to us, Americans.

Would this be just legal immigrants?

One reason for the high Indian household income is because there is arguably close to zero illegal immigration from India. Indians are a conservative lot and dare not break the law to immigrate. In contrast to Pakistanis / Bangladeshis who are more agreeable to the idea of flouting rules.

People who try estimate the numbers of illegal immigrants point out that the numbers of people who overstay their visas (obtained legally) quite possibly outnumber those who crossed the border illegally. And in their estimates, Indians are heavily represented in the former category. There are even so-called Dreamers whose parents used to be on H1B visas, got laid off, and never bothered to leave. I don't know how reliable the numerical estimates are, but the number of Indians living illegally in the US is likely to be non-trivial, despite the fact that none of them used the services of the likes of El Chapo.

Sure. But I am talking in a relative sense. Across both the categories you mention, my hunch is that the % of Indians in US who are staying illegally is a small fraction of the number for Pakistan or Bangladesh.

This survey estimates that about 500k of the 2 million foreign born Indians in the U.S are illegal immigrants... the report does not include coutrues that had less than 100k illegal immigrants. Since the US has about 300k residents born in Pakistan, one could posit that the illegal immigration rate is not significantly higher than that of Indians.
http://www.pewhispanic.org/2016/09/20/1-birth-regions-and-nations/ .

Also, Pakistanis face much more scrutiny than Indians, as far as immigration is concerned, which means that I would bet the opposite - the Indian rate of illegal immigration would be significantly higher.

Immigration from India has increased rapidly in recent years, to the point that, according to the Migration Policy Institute "Indian immigrants represent the second-largest origin group in the United States, accounting for 4.7 percent of the total foreign-born population." https://www.migrationpolicy.org/article/indian-immigrants-united-states

Well that's an embarrassing quote, given how many more Chinese there are in the country

far from zero.

"Although the vast majority of Indian immigrants in the United States are legally present, there were approximately 267,000 unauthorized Indian immigrants in the 2010-14 period, according to Migration Policy Institute (MPI) estimates, comprising less than 3 percent of the 11 million unauthorized immigrants in the United States.

MPI also estimated that, in 2016, approximately 15,000 Indian youth were immediately eligible for the 2012 Deferred Action for Childhood Arrivals (DACA) program. However, as of March 31, 2017 (the latest data available from the U.S. government), just 3,741 Indian youth had applied for the DACA program, and 3,182 had been approved."

Because of the county based quotas for green cards, Indians are far far behind other groups, so there are incentives to go illegal.

Rate of visa overstay is the key metric on visa waiver rules, and yes, Indians overstay their visas.

What nobody else want to say is because of the need for financial support and interviews, it is basically impossible for someone under 40 who is Indian to get a tourist visa. I've had multiple cousins denied entry for family visits -- the suspicious is they are looking for a job or spouse.

it is basically impossible for someone under 40 who is Indian to get a tourist visa

I don't know if it's impossible, but there's a high rate of denial, and the selection of people to deny seems to be highly random and subjective (if the visa officer thinks he's been too lenient with the previous 3 people in the line, he/she can just deny the next applicant on a whim.)

When I used to work in the US, my parents and brothers had no problem in applying for and getting 10-year multiple entry visitor visas. But recently, when I applied for a short business trip visa, I had to wait a month before the visa was approved. And I have zero history of visa violations, went to a highly-ranked STEM program, work for one of the biggest US multinationals, etc.

As I mentioned on a different thread, if you are lucky enough to be a citizen of a visa waiver country, you get free entry into the US for anything short of study or work regardless of what your individual characteristics are. And if you happen to be a citizen of a country outside that club, you get treated like a leprous criminal.

Talking to FSO who do the interviews it is not random.

The screen is "is this person going to overstay their visa". The smallest tell will set off that flag for anyone under 40. Again the default position is to deny the visa.

Parents, older people are less likely to do that although parents are some of the most likely to overstay.

You can see the same rules do not apply to Chinese.

The smallest tell will set off that flag for anyone under 40.

I appreciate your response, but without knowing what these "tells" are, it is hard to evaluate whether the offers are doing a good job. As you know, they have zero obligation to tell the applicants anything at all. I'm just basing my opinion on what I've seen and heard standing in line for a visa, and also my most recent experience which I recounted above (the only possible "tell" could have come from my face or demeanor, which is inherently subjective. If my extensive uneventful history of travel to, and stay in, the US, doesn't count for anything while making a decision, then I can't but believe they are doing something wrong. It's like a TSA charade in some ways.)

Anyone got an explanation for Portugal being on this list?

Its not nearly as developed as you might think, given its Western European location.

Until recently highly trained Portuguese did not, in general, speak English and this would be unlikely to migrate to the states. Ditto for other countries on the list and vice versa for all the high skill selectivity countries.

Are there shithole regions of the United States, where the inhabitants are unskilled, do not move, are on opioids, and marry their next of kin.

I would bet there is less internal migration in the US from or to that region, even though leaving the area to a low unemployment area might make one better off.

Trump Country.

Sorry, bill. you tried pretty hard, but this still isn't as stupid as rayward calling Lebanon an exceptionally stable country


Kris, Thanks for the link about Appalachia. We were there for a few weeks a year ago. It's not the issue of coal, but rather the level of human capital and human capital investment.

What was interesting to me about the post was that the focus was on country of origin, and not skill levels. So, if it is easy to enter as an illegal immigrant (Mexico) with low skill levels, what does that say about a Nigerian who is unlikely to enter crossing the border illegally. It might mean that since the cost of entry is higher for a Nigerian, its more likely the entrant is skilled, and may even be sponsored or a graduate of a US university.

Merit based immigration makes sense, and is independent of country of origin, except in the mind of our President.

I agree that merit-based immigration is the most sensible policy for any country (that does not have a mass of virgin territory to conquer) to adopt.

The President just seems to be channeling the theories of the alt-right, the smartest of whom (like Steve Sailer, who often comments on this blog) believe that the state of a country is a reflection of all of the inhabitants of that country: not just the average person but even their "best and brightest". And if that country is racially homogeneous, it reflects on the genetic makeup of those people. The alt-righters abuse (in my opinion) the statistical notion of regression to the mean to hypothesize that smartness in such people is an outlier characteristic, and their descendants (assuming no racial mixing) will inevitably descend into mediocrity. So if one believes all of this, it is rational to associate countries rather than individuals with smartness or other positive qualities, and come to a conclusion that immigrants ought to be selected purely on the basis of nationality.

It is sad to see the Trump Derangement Syndrome hitting otherwise perfectly clever and reasonable people such as you Kris. If we believe the version of the discussion given by Democrats, then the Democrats in the meeting was proposing to give Haitians, Salvadorian, and a few other countries a large quota of immigration taken from the dismantling of the current lottery visa. Then Trump asked why one should take people from these (s..thole) countries instead of Norway for example.

So the people proposing a highly non-meritocratic system were the Democrats here. Trump didn't propose anything at this point (otherwise he would have said so publicly) but asked a rhetorical question: if you want to take people just because they are from a specific country, why not at least choose a country where the people will likely be more educated?

Let us all leave all bad faith indignation to career politicians and activists. Trump has proposed a meritocratic system similar to Canada's, and this is rare among top-level politicians in this country. If you are for a meritocratic system, then you should support him on immigration policy. That doesn't mean
you have to agree with him on everything immigration... (For instance, I thought the lottery system was quite clever in meritocratic terms, and I wish it could continue somehow).

Thanks for the compliment, Joël.

For what it's worth, I think something like the RAISE act would be a good idea. Or at least, it would be a good place to start negotiations. The Democrats shouldn't dismiss it out of hand, and try to negotiate up (e.g., converge onto some point between 500k and 1 million immigrants, maybe, and concede the principle of merit-based immigration.)

That said, I think you are being disingenuous when you say that Trump was indirectly referring to a merit-based system through his comments. He may have been, but he's been exhibiting all the subtlety of a mace over the past two years, so count me unconvinced. Also, my understanding of what the Democrats (and Flake and Graham) demanded was not to give special preferences to El Salvador and Haiti, but to do a one-time allocation of the unused diversity (and maybe chain migration) visas to facilitate the stay of people from those countries who have already been living in the US for almost 2 decades. It wasn't like they were suggesting a long-term country-based preference, one that would favor s***hole countries in favor of, well, others. But since, as you point out, none of us knows exactly what transpired in that private conversation, all we can do is speculate.

Even if Trump really believes in going back to country-based preferences (as in the 1924 Act), I don't think it's that big a deal because that proposal is likely to be quickly shot down when a public debate i Congress ensues. So, overall, I blame the leakers (maybe it was the Democrats) for the breakdown in the recent talks; they fanned the flames and made it hard for everyone involved to budge from their positions.

Yes, agreed that was a very interesting article. I was surprised at the low crime rate.

I'm impressed by the p-factor of 0.001, which is 99.9% significance, a welcome chance from the usual 95% in social studies.

I think the "Gravity Equation" plays a role here, as noted upstream: if a candidate is from the far side of the world, chances are they are qualified to get into the USA, since they cannot just walk across the border illegally. Having said that, I did notice a large number of ethnic looking Chinese when I visited Mexico. And there's even lots of Koreans in the Philippines.

"the income data are not adjusted for ...": were they at least adjusted for where in the country the immigrants lived?

Immigrants from Morocco do surprisingly better than Immigrants from Australia on restoration hardware projects.

So that's why Stewie didn't want any Portuguese.


Wish we could measure entrepreneurship capacity as easily as skill level. A highly skilled programmer may not create many jobs for other people. If we had a system in California that didn't discriminate against small business (relative to well connected big business) there'd be less of a problem.

I wonder if 'skills' is what we should be selecting for. What about character, patriotism and goodness?

My money is on the Lebanese but people don't consider their entrepreneurial activities to be skilled so they are overlooked.

Since Noah and I use the same IAB dataset, the results obtained are similar. The IAB dataset is aggregated from the census of various countries and thus thus the data exclude illegal immigrants. It should also be noted that the Noah's results are specific to the USA situation and cannot be generalized to other countries. For example in 2010 the 'legal' Mexican immigrants to USA only 14.2% were graduates, but that for UK it was 83.4%.

Despite all the noise about the competition from the Chinese immigrants, though the Taiwanese immigrants were at 83% graduate, those from Chinese mainland were at 54%. At absolute numbers those from mainland China were 5X those from Taiwan. Mainland China did not sent their best. Comparatively those from India were at 82% graduate and Nigeria at 82.4%. Those from Nigeria for Canada were at 90.7% graduate.

I am a bit puzzle about the independent variable "Percentage high-skilled migrants minus percentage low-skilled migrants". The two components are from two disjointed sets, there is another independent variable medium-skilled migrants though the sum of all three should give unity. As such it seems not to be to attach reasonable physical meaning to it.

In addition to this valuable research, we need to look more carefully at how immigrants' grand children perform. It is this descension or convergence that is relevant for the largest and most difficult group -- Hispanics -- but to a lesser extent any group of high selected immigrants who then bring in weaker or even weflare using relatives due to US law. Note that the high level of African immigrants does not stop them from using high relative levels of welfare compared to whites and Asians.


Yes, there's so disturbing data suggesting convergence to local ethnic means over generations. This should be checked as a matter of urgency, as the value of such immigrants would need to be properly discounted over the longer span.

Should we be looking at high-skill immigrants, or high-surplus immigrants?

If the latter then it seems to me to make more sense to have immigrants who will work in relatively low-paying jobs, because these are the jobs where the market is most likely to be competitive, so that consumer surplus is greatest.

That's not necessarily the case, of course. A small share of a big surplus might be better, but the idea ought to influence our thinking, so that unlike some (Tyler?) we may feel that expected income is not the best measure of contribution.

Interesting, but wrong.

We're ambivalent over consumer or producer surplus for this problem. So long as it is captured nationally.

Avoid the Low-Paying jobs! Total value created is slight, and the welfare costs large. If an American on $30k is a net welfare consumer (and burden on all sorts of externalities), then an immigrant on the same salary may also be.

Comments for this post are closed