Is immigration actually declining?

The New York Times says that immigration has plunged over the past 12 months:

The United States population gained immigrants at the slowest pace in a decade last year, according to an analysis of new census data, a notable slowdown that experts said was quite likely linked to a more restrictive approach by the Trump administration.

The net increase of immigrants in the American population dropped to about 200,000 people in 2018, a decline of more than 70 percent from the year before, according to William Frey, chief demographer at the Brookings Institution, who conducted the analysis.

“It’s remarkable,” said David Bier, an immigration expert at the Cato Institute, of the 2018 numbers. “This is something that really hasn’t happened since the Great Recession. This should be very concerning to the administration that its policies are scaring people away.”

That might be true.  Unlike William Frey and David Bier, this is not my area of expertise.  But I’m not quite convinced, and I’d like to explain my reservations.  Before doing so, you need to understand that “immigration” is a fuzzy concept, and it’s not clear what the word means.  The net increase in people in the USA for reasons other than births and deaths, including tourists and students?  Or excluding tourists but including students?  Or excluding both?  How about illegals?  Does it include people who plan to stay but don’t yet have a green card?  Only those with green cards?  Those who become citizens?

In the past, many of the articles I’ve read on immigration refer to the number that become lawful permanent residents.  Here’s Homeland Security:

Approximately[1] 257,000 foreign nationals obtained lawful permanent resident (LPR) status in the first quarter (Q1) of Fiscal Year (FY) 2019 (Table 1A). Of these admissions, over 140,000 adjusted status from within the United States, and 117,000 entered as new arrivals. Compared to the first quarter of FY 2018, total FY 2019 Q1 admissions decreased by two percent, new arrivals decreased by five percent, and adjustments of status remained unchanged.

These mostly include family related immigrants, but also substantial economic immigrants and a smattering of refugees, asylum seekers, lottery winners, etc. I couldn’t find similar data for FY2018, but the report suggests it didn’t change much between Q1 2018 and Q1 2019. (These are actually fiscal years, so the most recent data is actually Q4 2018.) If you go back two years, the first quarter of FY2017 (late 2016) saw 288,000 immigrants, and legal immigration has been running at an annual rate of a bit over a million for a number of years. That doesn’t seem to be changing very much.

What about illegal immigration? All the evidence suggests that illegal immigration is surging much higher in 2018, with a stronger economy and problems in Central America. So why the big drop in the Census estimate?

AFAIK, the Cato report is based on Census surveys that try to establish the percentage of American who are immigrants. Then they apply that percentage to the total population (also estimated.) Then they take the annual change in that number. That’s certainly logical, but it’s also a method where tiny errors in the levels of immigrant population would translate into large errors in the rate of change.

Here’s a simple thought experiment. Assume the total population is stable in order to simplify the analysis. Suppose the estimated immigrant share in 2016, 2017, and 2018 was 13.1%, then 13.3%, then 13.4%. And let’s suppose the first two figures are exactly right, while the third figure is slightly mis-measured. It should be 13.5%. In that case, the estimated rate of immigration in 2018 would fall in half, even as actual immigration was unchanged. A small error in levels translates into a huge error in rates of change.

If my hypothesis is correct, then there should be other years where year-to-year fluctuations make no sense. Where changes cannot be explained by immigration policy, the economy, or any other known factor. Here’s a Cato graph that shows the big drop in 2018:

This graph is exactly what I’d expect if I thought the survey method was unreliable for rates of change. The declines in 2007 and 2008 do seem related to the housing slump. But the one million decline in 2011 seems weird, as does the surge in 2010. The decline in 2016 also seems weird. And why did immigration nearly double in 2017? It’s clear that the decline in 2018 is not particularly unusual, I don’t think it’s even statistically significant, and hence there’s no reason to link it to immigration policy.

Just to be clear, I believe Trump’s immigration policy has slightly reduced legal immigration, perhaps by 10%. But that’s based on the Homeland Security data. I also suspect that total immigration is increasing, due to the surge in illegal immigration–again, based on government data.

Of course, almost no one pays attention to this blog, so the NYT story will be accepted as the truth. And maybe it is. But I’m not completely convinced.

PS. Would Trump administration policies have made people a bit more reluctant to call themselves “immigrants” in Census surveys? Trump plans to deport legal immigrants that get public benefits like food stamps:

In a move that could curb legal immigration into the U.S. and deter millions of immigrants already in the country from accessing public health, housing, and food assistance benefits, the Department of Homeland Security announced a change to the “public charge” rule on Monday.

The government already has measures in place to block people who may become dependent on government services—potential public charges—from entering the country, but DHS’s new rule would make the requirements even more stringent. Immigrants who receive food stamps, Section 8 housing assistance, Supplemental Security Income (SSI), Medicaid, and parts of Medicare could be denied green cards, visa extensions, and immigration status changes.* . . .

 It triggered a chilling effect that repelled immigrant families from public spaces and discouraged them from interacting with public services in the months since.

Why might the Census number be true? Perhaps we are still taking a million immigrants per year, but many non-citizens are leaving, because of a proposed cutoff in benefits. Thus a decline in net immigration. But are that many people actually leaving? I doubt it.



27 Responses to “Is immigration actually declining?”

  1. Gravatar of E. Harding E. Harding
    28. September 2019 at 20:08

    No; immigration is not actually declining.

    Good post here, Sumner (though Massie will survive; he is anti-establishment):

  2. Gravatar of xu xu
    29. September 2019 at 05:45

    The reason immigration is declining is because of Southeast Asia’s rise. It has nothing to do with a merit based immigration policy.

    It does not make sense to work in America, when all the great business opportunities are in Asia. This is simply the natural decline of the west, brought forth by terrible policy positions based on flawed economic theories that Sumner promotes like “comparative advantage”. We can only hope that more businessmen continue to push back against the academic ideologues with common sense solutions to real world economic problems.

  3. Gravatar of E. Harding E. Harding
    29. September 2019 at 08:04

    Also, Sumner, I dispute your contention Trump has been bad for the election of fiscal conservatives. Yes; Flake is out, but Braun is in. Yes; Garrett, Sensenbrenner, Brat, Hensarling, Labrador, and Sanford are out, but Roy, Norman, Burchett, Cline, Steube, and Hern are in.

  4. Gravatar of ssumner ssumner
    29. September 2019 at 08:58

    Xu, Another comical comment. Did you even read the post?

    By the way, Asian immigration to American has been rising as a share of the total, and Latin American has been declining. FYI.

  5. Gravatar of sty.silver sty.silver
    29. September 2019 at 09:58

    Do I understand correctly that all numbers reported here are from the same source? Because the obvious way to test how large of a deal the effect you’re describing (small relative error in large numbers leading to large relative error in their difference) actually is would be to compare the numbers computed in the Cato way by numbers obtained from counting immigrants directly. Even if the definition of what counts as immigrants differs a bit, comparing both graphs could be pretty telling.

  6. Gravatar of Christian List Christian List
    29. September 2019 at 14:11

    Immigration is a winning theme for Trump. The numbers in this area, no matter in which direction, are completely irrelevant.

    I really don’t want to help Democrats with advice, but if they want to win, they should focus on their winning issues. I’m not specifying their winnining issues but I give a little hint: It’s not immigration, it’s not corruption, and it’s not impeachment.

    Do they have a winning issue? Maybe they should run under the theme: “We are all Native American” because we might have had one Indian ancestor 10 generations ago. That’s unifying and brings people back together again.

  7. Gravatar of John Arthur John Arthur
    29. September 2019 at 15:57

    I’m not so sure that the numbers are wrong. The 1965 law puts a cap on 1 million immigrants a year, but not mandated as far as I know. Trump could easily do things to hinder immigrants coming here, and that would probably cause decreases in the total #. Since the process is so hard and easy to fail, it could easily be that certain immigrants get booted out or quit from frustration.
    Also Scott, are you really not worried about immigration at all? Aren’t some cultures better at producing bigger economies than others? See China as an example vs the better run Chile. Don’t you think that constant immigration will reduce per capita growth in the US by making the population geared toward people not entirely conducive to economic growth?

  8. Gravatar of John Arthur John Arthur
    29. September 2019 at 16:01

    I hope I am not assuming too much of your points, this is inferred from your other posts on this issue.
    I know you are a utilitarian, but do you think America is richer on a per-capita basis from immigration today than it would have been without the 1965 act? The median household for Whites is already higher than the per-capita GDP. Probably, without immigration, the countries per capita GDP would be dramatically higher.
    Do you think, from a nationalistic, self interested perspective, that immigration is truly making America better than if it didn’t had any immigrants since 1965?

  9. Gravatar of John Arthur John Arthur
    29. September 2019 at 16:04

    Note, I do think IQ difference between groups are largely overrated. The dumbest people are around 85 IQ, and the smartest are at 104ish IQ.
    I am more referring to culture, like the culture in Latin America vs the more European Capitalist culture in the US.

  10. Gravatar of Benjamin Cole Benjamin Cole
    29. September 2019 at 16:28

    It is lugubrious that a major developed nation cannot even approximate the real number of immigrants, legal and illegal, in any given year or decade.

    How many illegal immigrants are in the United States? Pew says 11 million, MIT says 20 million, maybe some other guy would say 30 million.

    This same nation has suffocated net new production of housing to about 1 million units a year, and cannot effectively build infrastructure anymore. Scott Sumner says the United States should not even try to build infrastructure.

    Why is immigration sacralized among elites? If the employee class is not paid enough to reproduce and replace itself, then elites need to import labor.

  11. Gravatar of Bob Bob
    29. September 2019 at 18:28

    Looking at lawful permanent residents is an extremely imperfect way to measure immigration: It doesn’t just fail to look at illegal immigration, but it is not a very sensible way of looking at legal immigrants that don’t follow a family route.

    Quite a few immigrants come through work visas, and even when one goal is to stay here permanently, the road to a green card is often so onerous, immigrants live here, legally, for well over a decade before they get permanent residency. In general, the lists for permanent residency rarely get shorter: More people come here temporarily and meet al the requisites for permanent legal residency than there are legal resident work visas. I know some people that have been waiting for their green card since 2002, and they’ve been living in the US legally all along.

    And if you ask any corporate immigration lawyer, getting any of those visas keeps getting harder: Even though quotas haven’t been cut, a variety of procedural tricks occur that make it far harder to, in practice, come in and stay in the country.

    So while I have no strong analysis that would indicate whether immigration has gone up or down, Looking at green cards just isn’t going to get you good results at all.

  12. Gravatar of ssumner ssumner
    29. September 2019 at 19:51

    Silver, No, they come from different sources.

    John, It’s possible for immigration to make every single person in the world better off, and still reduce US per capita GDP. Composition effects. But I doubt that immigration is currently reducing our GDP/person, as we are getting lots of highly talented immigrants. I wish we got even more of the talented ones.

    You mention China. In recent years Asia has overtaken Mexico as a source of immigrants.

    The goal should be to make people better off, not focus on GDP/person.

    BTW, I live in Orange County, which has been greatly improved by immigration. And then consider Silicon Valley, where immigrants have been a huge boon to the US. It’s why we dominate high tech at the global level.

    Ben, You said:

    “If the employee class is not paid enough to reproduce and replace itself, ”

    LOL, take a deep breath Ben. Mali has the world’s highest birth rate. What’s its wage level? What is the wage level in low birthrate Europe?

    Bob, I actually agree, as I indicated in this post. There’s no perfect way to count immigrants, which is why I don’t necessarily believe the study cited in the NYT.

  13. Gravatar of Benjamin Cole Benjamin Cole
    29. September 2019 at 20:05

    Scott Sumner–

    Yes, poor nations do have higher birth rates….yet the empirical observation is that developed nations have below-replacement-level birth rates.

    If we assume “economic man” or “rational actors,” then the employee-class of developed nations has made the rational and valid decision that it is too expensive to raise kids, and also save for retirement, pay for college, pay the rent, meet medical expenses, etc.

    In poor nations, children are often a form of social insurance, but the cost of raising kids is very low. In fact, in rural areas, children can actually be productive, and do farm work. I see some variations of this in rural Thailand, where living costs are low, relative to Western urbanized regions, and families still replace themselves.

    How many people are raising large families in NYC, SF, Tokyo, Hong Kong, etc. I see Hon Kong is down to 1.2 babies per female. They have to import labor.

    OT but in the ballpark:

    “Deutsche Bank said in a recent report that it believes “helicopter money,” a term that refers to extraordinary monetary policies that can spur spending (including giving cash directly to households), “could be highly effective if properly deployed.” The bank points out that when interest rates are low but savings rates are high, fiscal policy — or monetary policy that blurs the line with fiscal policy — can be more effective.”—CNN.

    I never thought I would see such a rapid change in financial industry attitudes towards money-financed fiscal programs. The financial industry, including BlackRock, Pimco and Ray Dalio, are calling for variations of helicopter drops, (fig leaf here) “monetary-fiscal coordination,” in which the central bank buys a lot of bonds.

    Central bankers ar already pleading impotence in next recession. Interesting times.

  14. Gravatar of Mark Mark
    30. September 2019 at 02:32

    Yes, I had the same skeptical reaction to the article as well. We do know for a fact that the number of new legal immigrants is down by about 5-10% over 2016 (although 2016 was itself a high point), and there hasn’t been a huge spike in deportations of illegal immigrants either (another number that we know for a fact). So a 70% fall in net immigration seemed pretty implausible.

    Ben, richer people in developed nations also have lower birthrates than poorer people. Low birthrates are caused by the fact that people are getting married later. Later marriage is not caused by economic constraints (in fact, for most people, marriage saves money through shared living expenses and tax benefits). The median age for first marriage for college-educated women in the US is now almost 30; at that point you don’t have time for a large family. The reason for this is that, with less economic and cultural pressure to settle down, people are becoming much choosier about which partners they pick. I dated far more people than my parents did, and as a result was almost 10 years older when I did get married. I expected more from my spouse because I knew I could have an okay life on my own, and potential spouses expected more from me for that same reason. This is a good thing for long-run compatibility (as seen by the dramatic fall in the divorce rate), although not so good for incels. If you want to boost birthrates in developed countries, I think the best policy might be to increase the dating pool through more immigration of young unmarried people, especially women (as there are more unmarried young men than women in the US) and especially those with higher education/skill levels (who are more likely to assimilate, and whose corresponding natives have the lowest fertility levels). Also, the one developed country with above-replacement fertility is Israel, a country with high economic inequality, high immigration, and a high cost-of-living.

  15. Gravatar of Benjamin Cole Benjamin Cole
    30. September 2019 at 03:24

    “I think the best policy might be to increase the dating pool through more immigration of young unmarried people, especially women”—Mark.

    Especially if they are good looking! Yahoo!

    To be sure, lots has changed in the last two generations. Along the West Coast, it takes two incomes (at least) to buy a house the 9-to-5 dad bought in the 1960s, with a stay-at-home wife. The WSJ just reported a family of four takes $20k a year in medical insurance.

    Moody’s just released a report that said in part,

    “Census Bureau released the results of the 2018 American Community Survey (ACS), which revealed that income inequality increased to its highest level in more than 50 years. Without a clear policy response to reduce widening inequality, we expect this trend to persist, ultimately contributing to higher fiscal costs and potentially more elevated political risk in the US….Should labor incomes remain stagnant in real terms for large segments of the population, intensified polarization in incomes is likely to boost popular support for redistribution. Therefore, rising inequality is likely to impede an effective policy response [to rising federal debt] because political pressure will grow from lower-income households for increased government support,” added the credit-rating agency.”

    Egads, and that is a Wall Street credit-rating agency talking, not some Berkeley pointy-heads.

    Well, this is not an argument that can be settled.

    I do not mind ZPG, and in fact declining populations might lead to lower living costs and higher living standards (see rents in Sapparo, Japan, or some Midwest cities).

    Hong Kong is an example of a region with high rents. 1.2 babies per woman and still falling. The free-market citadel, the city on the hill—except the native population does not replace itself.

    This is success?

  16. Gravatar of Mark Mark
    30. September 2019 at 04:30

    Ben, here is some actual data: In 1960, American households spent about 29.5% of their income on housing. By 2002-03, the number had only increased slightly, to 32.8%. During that same time, the average size of a new home doubled, so it’s quite likely that costs per square foot have fallen relative to incomes. Meanwhile, the overall share of disposable income has increased from 36% to 49% because of significant falls in the cost of food and clothing relative to incomes. So it’s pretty clear that living standards are higher in the US now than in the past.

    Parts of the West Coast have experienced faster housing price growth, but there’s nothing stopping people from the West Coast moving to the Midwest. Costs on the West Coast are high because the amenities there are better than the amenities in the Midwest so there is more demand of people wanting to live there. And I say this as someone who lives in the Midwest. Also note that fertility in the Midwest, where housing is cheap, is not that different from fertility in California, where it’s expensive: Pennsylvania has even lower fertility than California.

    Success is creating a system that delivers the highest possible standard of living (including things not included in GDP like leisure time) given its resource constraints. Hong Kong is a success by this measure (compare it to cities in Mainland China). Success is not maximizing fertility—otherwise we would want to copy Mali (or among developed countries, Israel).

  17. Gravatar of Michael Sandifer Michael Sandifer
    30. September 2019 at 05:32


    You say you want the US to get an even greater number of highly talented immigrants, but my thinking is it’d be better to let the market determine the talent composition of US immigration. After all, comparative advantage applies, right?

    Also, is it necessarily true that bringing in immigrants who are less productive on average would lead to lower GDP/capita? To use a simple example, an attorney who’s a great typist can improve productivity/capita by hiring a typist, even if the typist is much less skilled at typing than the attorney.

    Perhaps I’m just being too simplistic, with an undergrad econ understanding.

  18. Gravatar of John Arthur John Arthur
    30. September 2019 at 08:42

    I agree with you on Hispanic and Asian immigration. I was more referring to immigration like Somalian and Dominican immigration, which in my opinion has not been good for America.
    Just like Apple can fire bad employees, we need a system to fire “bad immigrants”. By bad I mean high crime and low income

  19. Gravatar of John Arthur John Arthur
    30. September 2019 at 08:48

    I think the biggest problem isn’t that some immigrants are “low-skilled”, but that they bring crime and other social problems with them.
    Think about how easy it has been for East Asian countries to develop, in contrast to India and Africa. Its not like Africans and Indians aren’t working hard-they are but they have significantly lagged behind. Why should we change our demographics to the point where it becomes harder for us to grow?

  20. Gravatar of John Arthur John Arthur
    30. September 2019 at 08:55

    You often talked about how Donald Trump was a risky move that would more likely backfire. As you put it, on one hand we could become like Singapore or Switzerland, or we could fail and become a banana republic. Yet you have the same view on immigration. Making America’s demographics more like the Third World may make us Bermuda…or could make us like South Africa. Is that fair for American children?

  21. Gravatar of ssumner ssumner
    30. September 2019 at 10:10

    Ben, LOL. Upper middle class people in rich countries have one kid because they can’t afford two, and Malians have six because . . . ?

    I had one kid; do you think I couldn’t afford two? Please, be serious.

    Michael, I’m not opposed to low skilled immigration. But high skilled immigration is also good.

    John, You referred to Somali immigrants. Consider this:

    Between 2004 and 2018, in fact, the community established more than a dozen small businesses — grocery stores, restaurants, cafes and money transfer shops — along with two main nonprofit organizations. One of those is Somali American Faribault Education, which provides refugees with ESL programs and connects them to jobs; another is the Somali Community Resettlement Services, which connects refugees to employment, housing and county services.

    “Faribault has become our city,” said Hassan Mursal, a businessman who’s lived in the town since 2002. “Our children are succeeding. The community is thriving. We play a role in the security and safety in our neighborhoods.”

    The growing acceptance of the Somali community is partly the result of longevity. Mursal said the Somali community’s business success and cultural staying power has earned the respect of some white residents who used to look at them with scornful eyes when they first resettled in Faribault.

    I’m not saying there are no problems, but I think the problems are vastly overstated. Crime varies among immigrant groups just as it varies among native born groups. But overall it’s not high for immigrants.

    You also mentioned unsuccessful Indians. They are the single most successful group in America.

  22. Gravatar of John Arthur John Arthur
    30. September 2019 at 12:56

    Scott, thanks for the link. I agree that my fears might be overblown. I think that in the West, we take it for granted how easy it is to have economic growth. Indeed, that is probably why we were so shocked about China’s growth, we were so used to the Developing world being mediocre in terms of growth.
    Somalians are a mixed bag however
    I think they have the lowest income and highest crime in America, but it may be improving.
    Indian Americans are really successful though, I wonder why India is doing so badly then. If it was self-selection, wouldn’t there be regression to the mean? They also do really good in India, where many of the Indians originated from refugees from the British African terrority, no self selection there.
    Do you think India will be a success story in the future?
    I really hope we do become Bermuda, but no one in the media is willing to talk about this issue, even when the facts are on their side(low Hispanic Crime), so people are left to come up with their way to the truth.

  23. Gravatar of John Arthur John Arthur
    30. September 2019 at 13:35

    I mean really good in Britain, not India

  24. Gravatar of Benjamin Cole Benjamin Cole
    30. September 2019 at 16:08

    Scott Sumner: typically, orthodox macroeconomisst say that economies and culture respond to economic incentives. If you have a vast welfare state you will create a welfare class. Lower taxes on investment will lead to more investment.

    So why does the employee class not reproduce itself in developed nations?

    Does the low price of immigrant labor alter a tipping-point? You say no.

    You say you had only one child but could have afforded more, but you also have indicated you have a working wife. Perhaps you did not need a working wife, but many people today say they do need a two-income family.

    How does a two-income family have more than two kids? Answer: By mistake.

    How can we say that people respond to economic incentives and that people also have very small families today, but they are not responding to economic incentives?

  25. Gravatar of ssumner ssumner
    1. October 2019 at 07:44

    John, You said, “Indian Americans are really successful though, I wonder why India is doing so badly then.”

    Well, the first step toward enlightenment is questioning one’s views. It’s not just culture—look at the two Koreas.

    As far as the Somalis, I doubt their problems are any worse than the Irish in the 1850s. They had high crime rates when they first came over.

    To be clear, I’m not saying that culture doesn’t matter at all; some immigrant groups are obviously more primed for success that others. But almost all do well relative to their home country. Even sub-Saharan Africa immigrants have pretty decent average income levels in America.

    Now if we suddenly took in 5 billion immigrants, things would quickly deteriorate here. But there’s no chance of that happening in our political system. Reasonable increases in immigration can be easily handled.

    I do expect India to be at least a modest success story. Don’t forget that moving from low income to middle income is itself a huge boon to human welfare. They don’t have to become Switzerland to be a success.

  26. Gravatar of John Arthur John Arthur
    1. October 2019 at 10:48

    Scott: Hmm, the analysis of the Somalians and the Irish isn’t very rigorous, so I am still unconvinced.
    After all, either the immigrants will converge to the dysfunctional Descent of Slaves’s here in the US, or the Decent of Slaves will converge to the immigrants. I am very skeptical that Black immigration will be a boon for the US.
    However, since most of our immigrants are Latino and East and SouthEast Asian, I feel more assured. I think immigration should stop and we should have a rapid reform of the US economy, since immigration is clearly not helping Europe very much.

  27. Gravatar of John Arthur John Arthur
    1. October 2019 at 10:50

    After all, the Decent of Slaves and the Black immigrants are genetically the same. I think you are removed from this reality, since there are not alot of Africans in Orange County and Boston. But in Houston, where I live, they are responsible for 70% of all crime. Latinos actually have dramatically lower crime rates than Whites here!
    Asians of course, are ultra low in their crime.
    I think Africans commit homicide at a rate of 800% higher than Whites and Hispanics.
    Bringing more is probably not a good idea.

Leave a Reply