2025-09-09 16:17:20
Advanced civilizations sometimes really do throw it all away. I recommend reading the book Lost Enlightenment, which is about how the golden age of Islamic science and scholarship and literature gave way to centuries of religious fundamentalism and backwardness. During the Ming and Qing dynasties, China turned its back on exploration, refused to import Western technologies, and generally de-emphasized science; centuries of economic stagnation and military weakness followed.
These are useful examples, but they’re all from the premodern period, where cutting-edge science and technology didn’t make a huge difference in people’s standard of living; most people were farmers living on the edge of subsistence. But around 200 years ago, technology really started to matter, and living standards began to skyrocket in a way they never had in history. Thanks to technology, humanity is now about 20 times richer on average than we were two centuries ago, and the number keeps going up:
For a civilization to turn its back on algebra in 1100 or on spring clocks in 1800 was certainly foolish. But for a civilization in 2025 to turn its back on the technologies that took our species from indigent peasants to modern standards of living would be just unfathomably insane.
And yet that is exactly what I now see Western civilization doing. In the U.S., the Anglosphere, and Europe, there are simultaneous backlashes against a number of key technologies that have either made the modern world what it is, or promise to make it even wealthier. These include:
vaccines
solar and wind power
batteries
AI
software platforms
anti-obesity drugs
lab-grown meat
self-driving cars
air conditioning
I’m a well-known techno-optimist. Here was a manifesto I wrote about why technology is (usually) a good thing:
And here’s an essay I wrote about why the creation of industrial modernity — which depends crucially on technology — is by far the most important accomplishment of humankind:
In past years, I would write a techno-optimist post at the start of the year, gushing about all the marvels that are coming our way — here are the ones for 2021, 2022, 2023, and 2024. But this year I didn’t write one, because it seemed like America was generally turning away from techno-optimism, toward a dark, reactionary Luddism. I can still write a list of inventions to get excited about, but I feel like I’ll be setting myself up for disappointment when the general reaction is fear, cynicism, and superstition.
So instead, here’s a post about how the West is turning its back on the light of technology — the very thing that made it great in the first place.
The thing that inspired this post — and the header image at the top — is the American right’s turn against vaccines, so let’s start with those.
The chief villain of this story, of course, is RFK Jr., who spent most of his political career as an antivax kook on the left, and was only embraced by Trump in a strategic alliance in 2024. Here’s a video from 2023 in which RFK told a podcast interviewer that the polio vaccine “killed many, many more people” than polio ever did.
That is, of course, complete nonsense. The polio vaccine is incredibly safe; the worst recorded incident of a defective vaccine, in the 1950s, killed a total of 10 people, and the total number of documented deaths ever resulting from the polio vaccine is no more than a couple dozen. In a typical year before the introduction of the vaccine, about 1,000 Americans died of polio every year:
The drop to zero, of course, coincides with the introduction of polio vaccines in 1955, which rapidly defeated the disease.
But this is very far from the kookiest anti-science thing that RFK Jr. has uttered. In his writings, he has expressed doubt about the germ theory of disease itself, and embraced a version of an ancient, discredited theory of disease called “miasma theory”. Ars Technica reports:
[RFK] wrote an entire section on [the germ theory of disease] in his 2021 book vilifying Fauci, titled The Real Anthony Fauci. The section is titled "Miasma vs. Germ Theory," in the chapter "The White Man's Burden."…He writes: "'Miasma theory' emphasizes preventing disease by fortifying the immune system through nutrition and by reducing exposures to environmental toxins and stresses."…
Kennedy contrasts his erroneous take on miasma theory with germ theory, which he derides as a tool of the pharmaceutical industry and pushy scientists to justify selling modern medicines. The abandonment of miasma theory, Kennedy bemoans, realigned health and medical institutions to "the pharmaceutical paradigm that emphasized targeting particular germs with specific drugs rather than fortifying the immune system through healthy living, clean water, and good nutrition."
According to Kennedy, germ theory gained popularity, not because of the undisputed evidence supporting it, but by "mimicking the traditional explanation for disease—demon possession—giving it a leg up over miasma."
Miasma theory? Seriously??? One X user referred to RFK’s ideas as “actual, honest-to-god medieval peasant beliefs”, and he isn’t wrong. The discovery that germs cause disease — and the subsequent development of vaccines, antibiotics, and sanitation to combat germs — was one of the fundamental break points that lifted humanity out of medieval conditions. In 1900, around 0.8% of all Americans were killed every year by infectious disease; today, in a typical year, it’s about a hundred times lower than that:
RFK is already putting his medieval peasant theories into practice, canceling federal support for mRNA research that promises to turn cancer from a death sentence into a manageable condition:
To his credit, Donald Trump has broken with his HHS secretary on the vaccine issue:
"I think you have to be very careful when you say that some people don't have to be vaccinated," Trump said…"They're just, pure and simple — they work," he added. "They're not controversial at all. And I think those vaccines should be used."
But although Trump’s power over the GOP is great, it’s not infinite. Despite Trump promoting the Covid vaccines that his first administration pioneered, the American right was eventually dominated by the antivax movement, leading to a large number of unnecessary deaths among conservatives.
Now, the antivax turn in the GOP is far bigger than RFK. The state of Florida is ending all vaccine mandates for schoolchildren. This won’t only harm kids whose parents choose not to vaccinate them; it will also endanger kids whose parents do vaccinate them. This is because vaccines work via herd immunity; vaccines aren’t 100% effective, but if everyone or almost everyone gets vaccinated, transmission gets halted because vaccines are effective enough that viruses can’t keep spreading.
If just 7% of parents don’t vaccinate their kids for measles, herd immunity against measles will be impossible, and lots of vaccinated kids will get measles too.1 This is why we have vaccine mandates — by going unvaccinated, antivaxers aren’t just placing themselves and their own kids in danger, but everyone else around them as well.
Already, measles is starting to make a comeback in America, as a result of antivax superstition. There have been over 1400 cases this year so far — not a huge number yet, but growing alarmingly. If vaccine mandates are scrapped in red states, those numbers will continue to climb.
Slowly, the American right is turning into a movement against the germ theory of disease, and only bad can come of it.
2025-09-08 17:34:20
“You can go to live in France, but you cannot become a Frenchman. You can go to live in Germany or Turkey or Japan, but you cannot become a German, a Turk, or a Japanese. But anyone, from any corner of the Earth, can come to live in America and become an American.” — Ronald Reagan
There’s a meme going around on right-wing social media, where right-wing influencers ask “What is an American?”. Some don’t give answers, and simply assume that people who read the question will start thinking along the lines they want:
Others merely prompt the reader with evocative images:
But some do give answers, and their answers give a glimpse into how the political right is thinking in America nowadays:
What’s interesting on this chart is that the author — who may or may not himself be American, but who is certainly doing a good job of riling Americans up on social media — doesn’t try to establish a hard and fast cutoff, but instead defines a system of grades, where the longer your ancestors were in the country, the more points you get.
This kind of thing is natural in a polyglot country of immigrants like America. In every country you’ll find some form of restrictive nationalism — the idea that no matter what the citizenship laws say, only certain groups of people are truly members of the nation. But in most countries, there’s some kind of hard cutoff you can use — usually, membership in a specific ethnic group or religion. But America is so diverse that any attempt to draw a hard, bright line around which groups are “real Americans” is probably going to fail, because the cutoff will be transparently arbitrary. And so our restrictive nationalists resort to drawing concentric circles, defining a whole spectrum of American-ness based on some combination of family history, race, ethnicity, and religion.
This idea seems to be gaining power on the right. Online, you hear the term “Heritage American” thrown around a lot. C. Jay Engel, who has done a lot to popularize the term, sees “Heritage Americans” as people who value both the racial composition and the racial hierarchy of the U.S. before World War 2:
When I say Heritage American, this is what I mean: those who are ethno-culturally tied to the ethos and spirit of the United States prior to its definitional transformation into a Propositional Nation after World War II. This therefore includes the type of people that came here during the Ellis Island generation, even if that was a significant sociopolitical mistake. We are also the product of our mistakes as a nation.
It includes the blacks of the Old South (like Booker T Washington), though it repudiates any instinct that some of them have to leverage their experience for the purposes of political guilt in our time. It also includes integrated Native Americans with the same stipulation.
It affirms however the domination and pre-eminence of the European derived peoples, their institutions, and their way of life. Heritage America is centered around the experiences and norms of Anglo-Protestants.
This is pretty vague — there’s no test you can really do to tell if someone is or isn’t a “Heritage American” under this definition. In fact, the right can’t agree at all who exactly qualifies. A blogger calling himself “Ragnar Lifthrasir” defines the idea purely in ethnic terms (though he allows for other ethnicities to attain “ally” status):
[U]nderstanding who Heritage Americans are has never been more urgent. These descendants of the Founding Americans—Protestant, English-speaking, Northwestern Europeans who built the nation from Jamestown through the 1870s—represent more than ancestry. They embody a distinct American ethnicity forged through centuries of frontier experience, constitutional development, and cultural synthesis…
[There are also] those who are not Heritage Americans but who unequivocally support Heritage America…I call these citizens, “Ally Americans”. They can be of any ethnicity or faith or immigration vintage. But they advocate for America remaining a predominantly White, Christian nation governed by the spirit and letter of America’s founding documents.
Ben Crenshaw, on the other hand, adds an ideological element to the definition (something that Engel explicitly rejected):
For most nations of the world, to be considered a true German or Italian or Chinese or South African you must be a blood descendent (and thus exhibit common genetic traits), you must live in a geographical location where those people have historically lived, and you must abide by certain customs and ways of life. While America is not a “propositional nation,” neither does she neatly align with how other peoples have historically associated. Heritage America is not reducible to “blood and soil” fearmongering, yet neither are family, kin, and the land unimportant in America’s identity.
What does it mean to be an American? Heritage America is best understood as involving seven inheritances: the English language, Christianity, self-government, Christian government, liberty, equality under the law, and relationship with the physical land.
All these people can agree that “Heritage Americans” are an important group, but none of them can agree on who exactly belongs in that group.
It’s important to recognize that all of these ideas are well outside of the mainstream. YouGov did a poll this summer asking Americans what makes someone an American. Most of the items at the top of the list were behaviors like obeying the law, voting, speaking English, and beliefs like supporting the Constitution and Declaration of Independence, not owing allegiance to other countries, and so on. Ethnicity, race, religion, and family history all scored at the bottom of the list:
YouGov also asked a bunch of head-to-head matchups about which of these traits was more important, and the results came out basically the same. Furthermore, the partisan breakdown didn’t show particularly stark differences between Republicans and Democrats on who belongs:
It’s important to note that this poll came out after Trump’s 2024 victory, after the eruption of popular anger over uncontrolled immigration, and well after the “vibe shift” that saw America take a more conservative turn. It closely matches the results of a similar poll in 2021 by Pew, another similar poll by the VOTER Survey in 2017, and others too. Americans consistently say that race and religion are far less important to them as markers of shared nationhood than belief in American ideals and behavior consistent with that belief.
So why should we care what some online weirdos think, if they’re so out of the mainstream? The answer is that in the age of social media, it’s not necessarily mass opinion that determines policy — at least, not in the short term. Regular Americans may be tuning the “Heritage American” stuff out, but young Republican staffers are absolutely marinating in this online stuff, and they’re the future of the party. As Politico reports, savvy politicians and social media managers courting the young right-wing base are already making reference to the “Heritage Americans” term:
While the specific worldview surrounding “heritage America” may have been incubated online, it is increasingly finding its way into the policies and rhetoric of the Trump administration. In a speech at the conservative think tank the Claremont Institute in July, Vice President JD Vance urged conservatives to reject the view that America is founded exclusively on a common creed, reviving a theme from his acceptance speech at the Republican National Convention last year. “America is not just an idea — we’re a particular place with a particular people and a particular set of beliefs and way of life,” said Vance…If the subtext wasn’t clear enough, he added: “That is our heritage as Americans.”
The iconography of the heritage America movement has surfaced in the Trump administration’s messaging…In early July, the Department of Homeland Security’s official account on X posted a painting of a pioneer couple cradling a baby in the back of a covered wagon, under the caption: “Remember your Homeland’s Heritage.” Later the same month, DHS followed that post up with another one featuring John Gast’s painting American Progress, captioned “A Heritage to be proud of, a Homeland worth Defending.”
And Eric Schmitt, a senator from Missouri, gave a speech at the recent National Conservativsm convention called “What is an American?”, expressing similar sentiments to what you’d hear from online rightists:
At this point, it should be clear that the fact that something is sanctioned by our government does not mean it’s good for our country. That much is obvious with various forms of legal immigration today…
We Americans are the sons and daughters of the Christian pilgrims that poured out from Europe’s shores to baptize a new world in their ancient faith…
My ancestors arrived there from Germany in the 1840s…The first settlers in my state were mostly Scots-Irish—a hard, proud, fiercely independent people, forged in the hills of Ulster and the backwoods of Appalachia, ideally suited to life on the edge of civilization. They were the ancestors—as it just so happens—of my friend and our vice president, JD Vance.
As the historian David McCullough writes, the Scots-Irish families that first settled Missouri “saw themselves as the true Americans”…For some time now, we’ve been taught to be ashamed of these things that defined us[.]
To these “national conservatives” who now control much of our government, the struggle to make Heritage Americans the country’s dominant group is paramount, taking precedence over economic concerns, and even over defense against foreign enemies. A new National Defense Strategy being considered by the Trump Administration would basically see American power withdraw from the entire world in order to focus on the struggle to deport immigrants from the U.S.:
Pentagon officials are proposing the department prioritize protecting the homeland and Western Hemisphere, a striking reversal from the military’s yearslong mandate to focus on the threat from China…A draft of the newest National Defense Strategy, which landed on Defense Secretary Pete Hegseth’s desk last week, places domestic and regional missions above countering adversaries such as Beijing and Moscow…[I]n many ways, the shift is already occurring. The Pentagon has activated thousands of National Guard troops to support law enforcement in Los Angeles and Washington[.]
To the “natcons”, the paramount threat to America is immigration, not any foreign adversary. JD Vance has already expressed that sentiment with regards to Europe, and there can be little doubt that he and his fellow-travelers — who at this point dominate the Trump administration, and thus the MAGA movement, the Republican Party, and thus the nation as a whole — feel the same about America.
Trump renamed the Department of Defense the Department of War, but it’s clear that the main “war” he thinks he’s fighting is against immigration:
But this idea of immigration as an invasion, and deportation as a war, isn’t shared by the majority of Americans. Gallup has begun to see a major backlash against Trump’s anti-immigrant jihad. The fraction of Americans who wanted to decrease the rate of immigration soared under Biden, but has now plummeted back to where it was in Trump’s first term:
Interestingly, the shift is even sharper among Republicans:
And the percentage of Americans who say immigration benefits the nation is surging:
Again, this shift is happening more among Republicans than among Democrats:
Republicans still generally approve of Trump’s immigration policy. But this shift suggests that many Republicans are satisfied with what Trump has done so far, and don’t want to see it go much farther. They probably do not want to see the nation redefine itself around the blood-and-soil concept of “Heritage Americans”, and reorient the mission of the U.S. Military around mass-expelling people who have less “heritage”.
In fact, I suspect that there’s a very deep and fundamental reason why most Americans — including most Republicans — reject the “Heritage American” concept. Yes, many think of America as a “propositional nation” — an idea defined by the Constitution, the Declaration of Independence, and so on. But I suspect that many also value American culture as a marker of shared nationhood.
When I was growing up in Texas, one of my best friends was born in Shanghai, and didn’t become a U.S. citizen until the age of 18. Culturally, he was a little different than me and the rest of my friends — his mom made dumplings instead of sandwiches, he taught me how to use chopsticks, he didn’t believe in God.
But in all the cultural ways that mattered to us, we were the same. We watched the same TV shows, played the same video games, and listened to the same music. We used the same slang, had the same attitudes toward school, and wanted pretty much the same things for our future. And yes, we believed in the Constitution, and American freedoms, and all of that stuff.
During the 2010s, during our nation’s great national collective freakout over race, I wrote to my friend and asked him if he had ever felt discrimination growing up, or if he had ever felt excluded from the majority. He responded that while once in a great while he faced a little racism from a few jerks, it didn’t dominate his experience. In terms of identity, he told me he just felt very American.
This kind of real, on-the-ground cultural affinity is something too nebulous for YouGov pollsters to ask about, and yet I suspect it’s deeper and more important than most of the more quantifiable markers of American-ness. America is a propositional nation to some extent, but we’re also a cultural nation, bound together by shared habits and attitudes and lifestyles and beliefs. What matters the most isn’t our family’s history in the country, but our own personal history. Shared life experience beats shared heritage in terms of building the bonds of nationhood.
I recently had an extensive online conversation with a dedicated natcon, and it was…weird. He told me that nonwhite Americans can never really be his countrymen. But he also talked a lot about Europe, and about his family’s history in the German Empire. It was clear his notion of “heritage” went way beyond America, and connected to deep ethnic and religious concepts that reach far beyond our nation’s borders.
It felt very foreign to me. When I compare that online guy with someone like my high school friend who was born in Shanghai, there’s no question which one feels more American. My friend can quote the Simpsons and Star Wars with me. We can go to a mall together with his kids, and see a movie and enjoy it together, with shared cultural context. We won’t obsess over his ancestors in China or my ancestors in Europe. What will matter most to us is our shared personal history in the United States of America.
The bond my friend and I share is rooted in the land, and in community. It is tied to the place we grew up — College Station, Texas, and the United States of America — that taught us to be who we are. In contrast, the bond that natcons feel with the German Empire, or with other natcons in Australia and the UK, or with the idea of heritage and whiteness and Christendom and so on, is what I call a vertical community. It’s a notional bond between a bunch of people who find each other online and decide that they have more in common with those distant people than with the people who live next door in physical space.
Vertical communities have always existed, but in the 20th century, place still dominated. Only with the coming of social media did vertical communities begin to make a serious challenge to traditional communities as the organizing social unit of humankind. But the bonds that those online communities create are thin; a human being cannot thrive forever on the sugar high of Twitter flame wars and Facebook likes. The physical world draws us inexorably back to the places we live, and the people who live around us.
Right now, the MAGA movement is in charge of America. It is fundamentally an online creature — a weak bond between a bunch of people whom social media has taught to have the same notional enemies. In our division and our complacency, the American majority, who does not live for online hate memes, has allowed ourselves to put people in power who would tear up our actual heritage in the service of those memes. But eventually, no nation ruled by the Extremely Online can thrive, and I think my countrymen are starting to realize that.
Update: Matt Yglesias has a good post about how National Conservatism is out of step with American history:
2025-09-06 09:15:55
Noahpinion began as a macroeconomics blog. So every once in a while I feel like I ought to report on the macroeconomic situation.
Fortunately, the header image at the top of this post is not a picture of where the U.S. economy stands today; it’s not yet a sinking ship. But with the real economy looking shaky and inflation persistently above target, things are not looking great. The worst news, however, is that the people in charge of the economy don’t look like they have any desire to right the ship; instead, they’re the ones who caused the problems in the first place, and they seem to have every intention of doubling down.
First let’s talk about some of the latest numbers. We basically want our economy to do two things: 1) give everyone a job with a good income, and 2) keep prices stable. In other words, the “macroeconomy” can basically be boiled down to the labor market and inflation. Economic growth is important, but mainly because it makes everyone have a job and makes wages go up — in other words, when we say “the real economy”, we mean growth and the labor market, which are really the same thing.
There’s not much action on the economic growth front — at least, so far. Growth has been kind of slow under Trump, but not in recession territory, and Q2 was actually slightly better than Q1:
But growth numbers only come out once every three months, while lots of labor market data comes out monthly. So if we want a more up-to-the-minute picture of how the economy is doing, we should look at things like job growth and the unemployment rate.
The unemployment rate just came out today, and it’s slowly rising. It’s still low in historical terms, at less than 5%, but it’s creeping steadily up:
A better measure of the overall health of the labor market is the prime-age employment rate (also called the “employment-population ratio”), which doesn’t depend on who says they’re looking for a job. As of August it was still at very high levels, just a little bit lower than it was in 2024:
So that’s not too worrying, yet. But we have another source of data about jobs, which is a survey measuring how many jobs employers add each month — the “jobs numbers” that you always see reported in the media. The August numbers just came in, and they don’t look too hot:
The Bureau of Labor Statistics (BLS) reported that the U.S. added 22,000 jobs in August, well below expectations for 75,000. Employment data for June was revised to show a net loss for that month, the BLS said, though July’s figures were revised slightly higher…Taken together, the U.S. has added 598,000 jobs so far this year, compared with 1,144,000 for the first eight months of 2024.
In fact, job growth has been weak since Trump started announcing big tariffs on “Liberation Day” back in April:
Now, one possible reason we could have slow job growth with stable employment levels is that Trump could be forcing a bunch of illegal immigrants out of the country. If you don’t care about jobs for illegal immigrants, then maybe that wouldn’t bother you. But the native-born unemployment rate has been creeping up as well:
And jobs for the native-born, which had been rising a bit, took a tumble in August as well.
Also, when we look at which industries are seeing weak job numbers, a lot of them don’t look like the type of industries you’d expect to employ a lot of illegal immigrants:
We can also look at some other various numbers to get an overall picture of the health of the labor market. ADP, the private payroll processing company, showed a slowdown in hiring last month:
U.S. private-sector hiring rose less than expected in August, data released Thursday shows, offering the latest indication of trouble in the labor market…Private payrolls increased by just 54,000 in August, according to data from processing firm ADP published Thursday morning. That’s below the consensus forecast of 75,000 from economists polled by Dow Jones, and marks a significant slowdown from the revised gain of 106,000 seen in the prior month.
Jobless claims are rising too:
Applications for US unemployment benefits rose to the highest since June, adding to evidence that the labor market is cooling…Initial claims increased by 8,000 to 237,000 in the week ended Aug. 30.
US job openings fell in July to the lowest in 10 months, adding to other data that show a gradually diminishing appetite for workers amid heightened policy uncertainty…Available positions decreased to 7.18 million from a downwardly revised 7.36 million in June, according to Bureau of Labor Statistics data published Wednesday. The median estimate in a Bloomberg survey of economists called for 7.38 million openings.
And a private-sector job placement firm reports that companies are reducing hiring plans and announcing more job cuts:
Hiring plans fell to the weakest level for any August on record and intended job cuts mounted amid broader economic uncertainty, according to outplacement firm Challenger, Gray & Christmas…US-based companies announced in August plans to add 1,494 jobs, the fewest for the month in data going back to 2009. Of the 30 industries tracked by Challenger, hiring plans were concentrated in aerospace and defense, industrial goods and retail…Announced job cuts jumped from a year ago to almost 85,980 and marked the largest August total since 2020. Excluding the impact of the pandemic, the number was the highest for any August since the Great Recession in 2008.
So basically, the labor market doesn’t look great. It’s not a catastrophe yet, but things are steadily weakening.
How will the Trump administration respond? One thing they’ll probably try to do is to cook the books, by firing government employees who report accurate economic numbers, and bringing in apparatchiks who massage the numbers to make the news look better. After the previous month of bad jobs numbers, Trump fired the head of the Bureau of Labor Statistics and replaced her with — to put it quite bluntly — a partisan operative who appears to know little about economics.
This will be harder than simply putting Trump’s loyal apparatchiks in charge, though. There are a lot of different economic numbers, and the U.S. government is very transparent about how it collects them. If you mess with them, everyone will know the numbers are now fake, and economic confidence will plummet. Furthermore, political operatives who don’t actually understand economic numbers in the first place will naturally have trouble figuring out how to cook the books in a consistent way.
Another thing Trump will try to do is to blame the Fed for the weak labor market. This will gain a lot more traction — even the New York Times says that the latest jobs numbers strengthen the case for rate cuts. So Trump’s next move will be to try to get the Fed to bail him out. And this will probably work; Fed Chair Jerome Powell says that the time has come for rate cuts.
But there’s a reason why this won’t fix everything that’s wrong with the macroeconomy. Rate cuts work by boosting aggregate demand. When aggregate demand is too low, inflation should be low as well (because people are paying less for things, so prices aren’t being pushed up). But inflation is higher than the 2% target, and looks like it’s creeping back up:
And markets are expecting inflation to stay above target for the next 5 years.
This suggests that the economy’s gradual slowdown is being caused not by high rates, but by a shortage of aggregate supply. When there’s a negative supply shock, you can still boost the labor market by cutting rates, but this comes at a steeper price — you end up exacerbating inflation. Many economists argue that this is what the Fed did in the 1970s, when it cut rates to support the job market during an oil shock.
The current supply shock isn’t as bad as the 1970s, obviously. 2.5% inflation is not that bad. But the problem here is that unlike in the 1970s, the supply shortage has an obvious cause: Trump’s tariffs, which are making it harder for manufacturers to produce things by cutting them off from their supply chains.
Jason Furman notes that Trump’s immigration crackdown is yet another negative supply shock. Trump’s ICE recently raided a Hyundai factory in Georgia, arresting hundreds of factory workers.
Plenty of data show U.S. manufacturing in a parlous state. The Institute for Supply Management is showing a contraction in the sector, and everyone agrees that tariffs are to blame:
US factory activity shrank in August for a sixth straight month, driven by a pullback in production that shows manufacturing remains bogged down by higher import duties…The group’s index of factory output sank 3.6 points to 47.8, moving back into contraction territory for the first time in three months.
Iconic American companies like John Deere are suffering from the higher prices they now have to pay for imported materials like steel and aluminum.
Manufacturing employment isn’t doing great either. It was shrinking in 2024 (possibly due to rate hikes), but is shrinking much faster since Trump took office:
America has now lost about 78,000 manufacturing jobs in 2025, on net. Manufacturing, of course, is uniquely exposed to tariffs because of its reliance on supply chains.
In other words, the sky is not yet falling, and the ship is not yet sinking, but the Trump administration is doing everything it can to make things worse. The U.S. economy is incredibly resilient, but if you hit it with enough bad policies, even the most resilient economy will sour. We’re now seeing a slow souring of our economy — a new normal that the government just expects Americans to accept, in service of ideological goals that most of the population doesn’t share.
Americans are not exactly happy about this, giving Trump lower ratings on economic issues than on cultural issues like immigration:
But so far, the economy isn’t bad enough, nor the anger severe enough, to pose a serious threat to Trump.
America is getting the macroeconomy it voted for, and it’s grumblingly accepting it.
2025-09-04 10:53:37
Sorry for the slow publishing this week…my rabbit had some health issues! But he’s fine now, thankfully.
Anyway, I have a fun video debate for you today! It’s me versus Oren Cass, on the topic of tariffs!
I also have an episode of Econ 102, where Erik and I discuss the question of who will ultimately profit from the AI boom (something I plan to write more about in the coming months):
Anyway, on to this week’s list of interesting things! This week’s theme is reality interfering with the best-laid plans of policy wonks and populists alike…
For years, I’ve been a big advocate of the “just give people cash” theory of the welfare state. I’ve written a bunch of articles in support of the idea. My reasons are:
Cash benefits are much easier to administer than other kinds of benefits.
A lot of economic research found that unconditional cash benefits don’t do much harm to the labor market — i.e., people generally don’t stop working when they start getting checks in the mail. The research also found that people don’t tend to fritter away the cash on drugs and alcohol, as many conservatives fear.
Unconditional cash benefits can reach a lot of people who fall through the cracks of other welfare programs — for example, people who can’t earn any income (and so can’t get the EITC) or don’t have kids (and so can’t get the Child Tax Credit).
In the late 1990s, America mostly canceled a traditional welfare program (AFDC/TANF) and replaced it with some cash-type programs (EITC and CTC), and the experiment went very well.
However, in recent years, some new research has come out that tempered my enthusiasm for the cash benefit revolution. First, a basic income trial in Denver failed to decrease homelessness, which is one thing you’d really like to see basic income do. Then, an even bigger basic income trial in Texas and Illinois found that just $1000 a month caused 2% of people to stop working — a very big disemployment effect, contradicting the results of earlier studies. Worryingly, this study is much more believable than any of the more optimistic studies, since it’s a very large randomized controlled trial. (Of course, it’s just one study; the papers showing little effect are still more numerous, even if no single one is as reliable.)
Meanwhile, a lot of these studies are finding that cash benefits aren’t really doing much to improve quality of life for the people who get the cash. You can measure various things we think curing poverty ought to improve, like health, education, employment, housing, etc. And unfortunately, these recent studies show that cash benefits aren’t making those indicators look much better. Kelsey Piper recently wrote an article in The Argument summing up some of this evidence. Some key excerpts:
The team behind Baby’s First Years…gave their control group $20 per month and their experimental group $333 per month. They recruited low-income mothers…The experimental group had more money (as you would expect) and worked a little less (but it’s probably good if mothers of young babies can cut back on hours slightly). They also spent slightly more money on stuff for their kids.
But there was not much else. The cash transfers did not improve maternal health outcomes or child health outcomes. They had no effect on stress, depression, body mass index, how often children got sick or the children’s overall health. They did not improve mothers’ self-reported relationship quality or measures of psychological distress. There was no effect on child development…
A Compton, California, RCT tried $500 per month for two years. They found “no significant effects of the transfers on labor supply; assets; psychological well-being; financial security; or food insecurity.” The biggest effect they found was on other income sources: The groups receiving transfers worked fewer hours or got paid less than people in the control group. The lost nontransfer income averaged $333 a month…
The OpenResearch unconditional income study tried $1,000 per month for three years, while the control group got $50 per month. They found that participants worked less — but nothing else improved. Not their health, not their sleep, not their jobs, not their education, and not even time spent with their children. They did experience a reduction in stress at the start of the study, but it quickly went away…
Direct cash payments not making people better off is a surprising finding — and while Baby’s First Years and Compton are relatively small sums of money, OpenResearch is larger.
This is all extremely disappointing. In these well-designed real-world experiments, cash benefits are making people produce less in the market, while not really making their lives much better. And as Piper goes on to explain, nobody can really figure out why. People aren’t using the cash to buy drugs or gamble or whatever; they’re working a bit less, but mostly they’re just spending the cash on things they need. It just doesn’t make their lives look less like the lives of poor people.
Matt Bruenig, a prominent socialist advocate of cash benefits, attempted to write a rebuttal of Piper’s piece, but it wasn’t up to his usual standard of quality. Other than pointing out a study from Finland that shows more optimistic results, he didn’t really rebut the evidence that Piper brought to bear. Bruenig talks about the historical importance of Social Security and the success of Nordic countries. But these arguments miss the mark, because they don’t address the question of what additional cash benefits can do for Americans in the present day.
A more valid counterargument — and one that Bruenig touches on, but could have been a lot more explicit about — is that poor people having more cash is simply a good thing in and of itself, whether or not their kids become healthier or they get a better education or they report less depression. Being able to afford more food, more transportation, more housing, etc. makes your life better, even if it doesn’t make you lead a healthier lifestyle.
This is why I still support cash benefits. (Though the recent results on disemployment do give me pause, because economies do need to produce things in order to have things to redistribute.)
But I think the recent evidence points at a deeper truth, which is that many of the social ills we believe result from low income — drug use, poor health, broken families, violence, and so on — might not be solvable by giving people more TVs, sofas, cars, and cell phones. Material goods are important, but they aren’t everything, and we should be looking for additional ways to help poor people lead happier lives.
One of my long-running theses about the U.S. economy is that the higher education sector has gotten too bloated and is due for a shakeout. The most important reason is that enrollment rates are falling. That’s partly because fewer young people are choosing to go to college, but a lot of the story is just declining birth rates, leading to a dearth in the number of young people in America:
Many universities will try to fight this trend by letting in more international students. That would be a great move — and good for the country as a whole — except that the Trump administration and the Republican Party more generally are trying to push foreign students out of the country en masse. And they are succeeding — international enrollment is dropping, putting the very survival of many colleges in danger.
However, we shouldn’t let this obscure the fact that college itself is becoming a less valuable proposition for the marginal student. A new paper by Bleemer and Quincy finds that a university education isn’t nearly as good of an escalator into the middle class as it used to be. There are several reasons, but one is that students from higher-income backgrounds have been choosing more lucrative STEM majors, while students from lower-income backgrounds have been crowding into lower-value humanities majors:
We show that two recent trends in college major choice – the ‘death’ of humanities enrollment…and rising computer science enrollment – have both been driven by higher-income students…[T]he ordering across disciplines’ average wages has stayed remarkably constant over time, with humanities at the bottom and engineering and business majors earning the highest wages.
This dovetails with another trend: rampant grade inflation. Rose Horowitch recently wrote a good article about this for The Atlantic, and there’s research supporting the notion that grade inflation has been increasing steadily at American universities for some time now. Matt Yglesias argues that the problem is worse in the humanities:
An uncomfortable possibility is that many (most?) American universities have transformed themselves to act as high-value training centers for rich kids and low-value diploma mills for working-class kids. This means that the marginal students — the people who really need college the most — aren’t really getting the education they deserve. A crappy humanities diploma that employers know was essentially handed out for free will not increase your wages much in the job market, even if it has a prestigious name stamped on it.
That’s not really a service worth paying for. No wonder college tuition is falling.
The big story of China’s economy over the last few years is that the leadership tried to replace a real estate boom with a manufacturing boom. The problem with this idea was that just producing a ton of stuff doesn’t actually mean you’re adding real economic value. So China has been paying its companies to compete each other’s profits to zero:
The Chinese word for this is “involution”.
This results in at least three problems:
You end up with a lot of junk that nobody really wants. This actually reduces productivity growth, because you’re producing a lot of stuff but it isn’t worth much to anyone.
Your companies become unprofitable, which might not bother the central planners, but certainly makes a lot of regular Chinese businesspeople very mad.
At the macroeconomic level you can get deflation from the vicious price wars, which then exacerbates the debt problem from the real estate bust.
At this point, you can do one of three things:
You can find some new underdeveloped sector of the economy to direct resources to.
You can curb production, which will raise profits and lead to more sustainable growth, but will slow growth in the short-term.
You can just keep doing what you’re doing, turning all the metals in the Earth’s crust into solar panels and electric cars and humanoid robots, like that AI that just wants to turn everything into paper clips.
Given Xi Jinping’s track record, you might assume that he’d go for option (3). And so far there hasn’t been an all-out campaign to curb involution or to redirect resources toward the service sector (the only sector of the Chinese economy that’s still underdeveloped). But you can see the Chinese government sort of pivoting in the direction of an anti-involution policy:
And so, it seems like a pretty big deal to see China trying to crack down on its overcapacity issues in recent months. Last year, Beijing announced an “anti-involution” drive, promising to tackle gluts in new ways. These efforts seem to be gaining some additional momentum as policymakers continue to grapple with deflation and what really does look like a balance sheet recession…[J]ust today, the Ministry of Industry and Information Technology vowed to “curb disorderly, low-price competition” and “encourage the orderly exit of outdated production capacity through market-based and legal approaches” in the solar industry…Beijing isn’t just trying to clamp down on disorderly expansion and unproductive competition; it’s also nudging its industries toward profit discipline, capacity rationalization, and market-driven consolidation.
No one seems to know exactly what China is doing in order to crack down on involution — they just assume the Chinese state gets what it wants from companies. But everyone seems to agree it’s real. In fact, you can see a big drop in the manufacturing investment numbers:
Of course, it’s not clear whether this is a government-directed slowdown or a market reaction to the unprofitability of the lending boom of 2022-24. But it’s notable that solar and EVs, which seem to be seeing some of the most vicious involution — with even China’s standout national champion, BYD, under intense pressure from heavily subsidized rivals — are seeing particularly large seasonal drops in production:
Whatever is causing China’s manufacturing investment to slow down, it’s definitely contributing to a slowdown in economic growth.
It turns out that Stein’s Law applies even to China’s vaunted industrial might.
America’s right wing is kind of insane about energy. Much of the rest of the world switches to cheap solar and batteries, and China is using solar to blow past the U.S. in electricity generation:
And yet America’s right wing is waging a protracted, desperate, all-out campaign to destroy the solar industry, with Trump declaring that America will run on coal instead. In an age of cheap solar and batteries, that’s just going to contribute to America’s spiraling electricity costs, and hurt the economies of red states like Texas that have been leading the green energy boom.
So why is the right attacking cheap energy?
The answer is “culture wars”, but that’s not very satisfying. Right-wingers didn’t simply decide that solar and batteries were hippy-dippy bullshit because they didn’t like the look of solar panels and windmills! OK, a little bit of it was that, but mostly it’s because the right thinks of renewable energy as being all about climate change, rather than about securing cheap reliable supplies of electric power.
In one sense, you can’t blame the right for thinking like this — after all, climate change activists on the left pretty much only talk about solar and wind in terms of the carbon emissions they prevent (or at least, they did until recently). That sets conservatives’ alarm bells ringing, because they think of climate change primarily as something that leftists play up in order to bring down capitalism and soak the rich. The right sees green energy not as a form of abundance, but as a degrowth policy, designed to force Americans to live with less.
This is, of course, ridiculous. But it wasn’t always! There was a time, not too long ago, when solar power was much much much more expensive than it is now (and wind and batteries, too). If we had forcibly converted to solar in the 1980s, it would indeed have ruined our economy. This has only changed in the last decade, but like most people, conservatives are slow to update when the world changes.
In fact, to see how a mismanaged and misguided push for green energy can hurt a country’s economy, you have to look no further than the UK. Over the past 20 years, the UK’s energy use per capita has fallen steadily, putting it well behind China and other rich countries:
This happened because of a bunch of British government policies to restrict the use of fossil fuels and nuclear,1 in the name of promoting green energy. Coal was phased out completely, and gas and nuclear severely curbed, in order to make way for renewables that were very slow to ramp up:
But the UK, unlike most poor countries, is not very sunny. And it doesn’t have a lot of land for wind farms. The UK was one of those small special cases where it really did make sense to keep using fossil fuels and nuclear. The UK is a relatively small nation, so this wouldn’t have made much of a difference to climate change. But it’s also a nation dominated by environmentalism and degrowth-adjacent ideology, and so the decision was made to reduce the nation’s energy generation.
As The Economist reports, this has caused energy prices in the UK to skyrocket, immiserating the population:
British electricity prices have become expensive. For decades, those prices rarely strayed far from the rest of Europe’s. Now household bills are 20% above the average for the major European economies, and industrial bills 90% higher...The gap with America is starker yet…
Britain’s clean-power push has started to land on bills. Paying for the new pylons and wires that will stitch together a more complex grid where power is generated in pockets all over the country is driving up network costs. So are “balancing costs”, whereby generators are paid to smooth out the volatility of renewables…Since 2019 green subsidies and network costs…have contributed about two-thirds as much to the real-terms rise in bills as the wholesale price of electricity has…Both are set to keep rising…
High bills have squeezed Britons’ finances….Output of energy-guzzling goods like chemicals, plastics and metals has fallen by more than 20% since gas prices first spiked…Energy flows into everything, even the services that dominate Britain’s economy. References to energy costs in corporate reports and earnings calls have more than tripled since 2019, driven by the consumer, tech and financial sectors.
As the article notes, expensive energy is causing a political backlash in the UK. And why shouldn’t it? The UK chose to deprive its citizens and companies in order to produce a totally performative emissions reduction, setting an example that almost no other countries followed:
U.S. conservatives look across the pond at that failure, and they think “We can’t let that happen here.” They don’t understand that America is much sunnier than the UK and has a lot more land, and that green energy is therefore potentially much cheaper here than there. All they see is the clash of political movements and ideologies.
When Trump fired Fed governor Lisa Cook, it raised a lot of concerns about central bank independence. Although Stephen Miran, Trump’s nominee to replace Cook, says he will respect the Fed’s independence, this is only the latest in a long line of Trump’s attacks on the Fed. During a recent interview, JD Vance declared that the President, not the Fed, should have the power to determine monetary policy.
This is very bad, but it’s not always clear to regular folks why it’s bad. I wrote a post outlining some of the negative effects that the end of Fed independence could have on the U.S. economy. But an even simpler way to get the gist is to look at what happened to Turkey over the last few years.
Hakan Kara and Alp Simsek have a new paper explaining what happened when Turkey’s President Erdogan decided to take over monetary policy and force interest rates down. Simsek has a good thread telling the basic story. Here are a few excerpts from that thread:
I’ve long thought that of any modern leader, Trump seems most similar to Erdogan. We should pay attention to Turkey’s failed experiment, and avoid following a similar path.
In my debate with Oren Cass at the top of this post, Oren attacked the field of economics, declared that it isn’t a real science, and scoffed at predictions that tariffs would hurt the U.S. economy. Meanwhile, the economist Brian Albrecht wrote a post explaining some simple econ models showing the downsides of tariffs:
Here are some excerpts from Albrecht’s post:
Now let’s add the most important realistic feature, which I think is the most important…The economy at Home has “Climbers” who gather coconuts, but the economy in [the] Foreign [country] has “Weavers” who are skilled at making sturdy ladders. Ladders…are a productive input that allows a Climber to gather far more coconuts. Home imports these ladders from Foreign, paying for them with coconuts…
The government at Home, needing revenue, decides to place a tariff on imported ladders…The result harms the economy’s productivity. The tariff makes ladders (a key tool for production) more expensive. Faced with this higher cost, some Climbers decide it’s no longer worth it to buy a ladder and go back to climbing with their bare hands. Their productivity plummets. The entire production system at Home becomes less efficient…
Let’s extend our ladder example, but now think about what happens over time. Instead of each Climber just buying one ladder when they need it, imagine that ladders are durable and Climbers can accumulate them. The more ladders a Climber owns, the more productive they become…Home imports these ladders from Foreign and, over time, builds up a stock of them…
Now, suppose the government places a tariff on imported ladders. This doesn’t just affect the Climbers who want to buy a ladder today. It affects the entire process of capital accumulation in the economy…With the tariff making ladders more expensive, fewer new ladders get imported each year. The economy’s stock of ladders grows more slowly. Some old ladders break down or wear out, and they’re not replaced as quickly because of the higher cost. Over time, the entire economy becomes less productive as the capital stock shrinks relative to what it would have been.
They harm the economy’s ability to become more productive over time. A tariff on capital goods is like a tax on economic growth itself…The effects compound. With fewer ladders, workers are less productive. With lower productivity, the economy generates less income. With less income, there’s less money available to buy new ladders, even if the tariff were removed. The economy gets stuck in a lower-productivity equilibrium.
Who should we believe? Oren Cass’s bombastic pronouncements that economics isn’t a real science? Or Albrecht’s conceptual explanations of how supply chains and capital accumulation ought to work in a global economy?
Well, I don’t know. But here’s a news story about U.S. manufacturing activity:
US factory activity shrank in August for a sixth straight month, driven by a pullback in production that shows manufacturing remains bogged down by higher import duties…
“We continue to have weak demand overall, still due to tariff uncertainty,” Susan Spence, chair of the ISM’s Manufacturing Business Survey Committee, said on a call with reporters. [emphasis mine]
Hmm, perhaps a discipline doesn’t have to be a “science” in order to give you useful information about how the real world works…
Environmental movement types don’t think of as “green” even though it produces no carbon. This is, of course, very foolish, and it comes from ancient tribal political conflicts that almost no one remembers anymore.
2025-09-01 17:52:57
This Substack newsletter of mine has gotten pretty popular over the years, but it didn’t start out that way. While writers like Matt Yglesias often made a big splash when they left their mainstream publications to become substackers, and got lots of subscriptions right away, I started out with very few — only 70 paid subscribers in the first week, compared to thousands for Matt.
I’ve often wondered why I didn’t make a big splash when I started my Substack. One big reason was probably that unlike Matt and most others, I didn’t quit my mainstream media job right away. I started this newsletter near the end of 2020, and I didn’t leave my corporate writing job at Bloomberg Opinion until almost a year later. Substack started out as a side gig for me, meaning that my regular audience could keep reading my stuff in the regular place.
Not immediately quitting Bloomberg also meant that the inception of my Substack didn’t become a news story in and of itself. When he left Vox, Matt talked a lot about how Vox had been suppressing his viewpoint. He even went on Rogan! That generated a lot of buzz and controversy about whether the mainstream media was censoring centrist views. In contrast, “Bloomberg writer moves his personal blog from Blogger to Substack” just doesn’t make a very compelling headline.
But in fact, I do have a story about why I eventually left Bloomberg to write for my Substack full time in late 2021, and it’s kind of an interesting one. I kept it under wraps for years, but now I finally feel comfortable sharing that story — at least, as I remember it now.
In fact, there were three reasons I left, two of which are rather mundane. The first was that in summer 2021 I developed chronic vestibular migraines, which made it hard to look at a computer screen for more than four hours a day; this forced me to choose between my hobby and my job, and I chose the former.1
The second reason was that Bloomberg Opinion, the division of Bloomberg I worked for, had some management changes that resulted in more layers of bureaucracy being added to the writing process; pitches had to go through more layers of approval, making it much harder to maintain my normal level of output. This made Bloomberg Opinion a less attractive place to work.
Neither of those reasons are very interesting or unusual. But there was a third reason, which is that in May of 2021, I became involved in an incident between Bloomberg and the Chinese Communist Party.
2025-08-30 17:09:30
The debate over whether AI is taking people’s jobs may or may not last forever. If AI takes a lot of people’s jobs, the debate will end because one side will have clearly won. But if AI doesn’t take a lot of people’s jobs, then the debate will never be resolved, because there will be a bunch of people who will still go around saying that it’s about to take everyone’s job. Sometimes those people will find some subset of workers whose employment prospects are looking weaker than others, and claim that this is the beginning of the great AI job destruction wave. And who will be able to prove them wrong?
In other words, the good scenario for the labor market is that we continue to exist in a perpetual state of anxiety about whether or not we’re all about to be made obsolete by the next generation of robots and chatbots.
The most recent debate about AI and jobs centers around recent college graduates. Derek Thompson wrote a post suggesting that a slowdown in job-finding for recent college grads could be the first sign of the job-pocalypse. A number of news articles ran with this story and treated AI job destruction as a proven fact, but some pundits pushed back on the narrative, citing various data sources. I wrote about the whole controversy in this post:
Then, Sarah Eckhardt and Nathan Goldschlag of the Economic Innovation Group, a think tank, came out with some research that found no detectable effect of AI on recent employment trends. (I covered this research in my last roundup post.)
Eckhardt and Goldschlag looked at several measures of which jobs are more “exposed to” AI. They found that for three of the five exposure measures they looked at — including their preferred measure, from Felten (2021) — there was no detectable difference in unemployment between the more exposed and the less exposed workers. But for two of the measures, there was a small difference, on the order of 0.2 or 0.3 percentage points:
The EIG researchers conclude that AI probably isn’t taking jobs yet, and if it is, the effect is still very small at this point.
Eckhardt and Goldschlag were wise to title their research note “AI and Jobs: The Final Word (Until the Next One)”. Indeed, the next word on the topic came out almost immediately, in the form of a paper by Brynjolfsson, Chandar, and Chen, entitled “Canaries in the Coal Mine? Six Facts about the Recent Employment Effects of Artificial Intelligence”.
Brynjolfsson et al. do something very similar to Eckhardt and Goldschlag — they use two measures of how exposed a job is to AI, and then they compare recent employment trends for more and less exposed workers. Their finding is startlingly different than that of the EIG team:
Our first key finding is…substantial declines in employment for early-career workers (ages 22-25) in occupations most exposed to AI, such as software developers and customer service representatives. In contrast, employment trends for more experienced workers in the same occupations, and workers of all ages in less-exposed occupations such as nursing aides, have remained stable or continued to grow.
Our second key fact is that overall employment continues to grow robustly, but employment growth for young workers in particular has been stagnant since late 2022. In jobs less exposed to AI young workers have experienced comparable employment growth to older workers. In contrast, workers aged 22 to 25 have experienced a 6% decline in employment from late 2022 to July 2025 in the most AI-exposed occupations, compared to a 6-9% increase for older workers. These results suggest that declining employment AI-exposed jobs is driving tepid overall employment growth for 22- to 25- year-olds as employment for older workers continues to grow.
Bharat Chandar, one of the authors, has written a blog post explaining the paper’s findings:
Now, I know Erik Brynjolfsson, and I know that he is a very good and careful economist. But I’m suspicious of this particular result, for one big reason: I can’t see any reason why the employment effect of AI should fall only on recent college graduates.
Brynjolfsson et al. find that since late 2022, employment has grown robustly among most segments of the workforce. Only very young workers who are also highly exposed to AI have seen their employment numbers fall:
Notice that the workers in their 30s, 40s, and 50s who are judged to be most heavily exposed to AI have seen robust employment growth since late 2022.
How can we square this fact with a story about AI destroying jobs? Sure, maybe companies are reluctant to fire their long-standing workers, so that when AI causes them to need less labor, they respond by hiring less instead of by conducting mass firings. But that can’t possibly explain why companies would be rushing to hire new 40-year-old workers in those AI-exposed occupations!
Think about it. Suppose you’re a manager at a software company, and you realize that the coming of AI coding tools means that you don’t need as many software engineers. Yes, you would probably decide to hire fewer 22-year-old engineers. But would you run out and hire a ton of new 40-year-old engineers? Probably not, no! And yet Brynjolfsson et al.’s data says that this is exactly what’s been happening since 2022. Empirically speaking, it’s a great time to be a middle-aged software engineer or customer service rep!
Now, maybe we can quickly come up with a story to explain this. Maybe older workers gain human management skills that complement AI, while younger workers are valued primarily for their technical abilities, putting them in more direct competition with AI, or something like that. But come on — before an AI critic saw this result, would they have predicted that the coming of AI would lead to a hiring bonanza for middle-aged customer service reps? Probably not.
Unless we can come up with a compelling story for why AI should only replace the young, this finding smacks a bit of “specification search”. During any 3-year period, there will probably be some group of workers who do a little bit worse than the rest. We can’t just keep pointing at those groups and yelling “It’s AI! It’s AI!”. Yes, any such group might be “the canary in the coal mine”, but we should have some reason to believe ex ante that they’re a canary instead of a pigeon or a sparrow.
It’s also a bit fishy that Brynjolfsson et al. find zero slowdown in wages since late 2022, even for the most exposed subgroups:
This just doesn’t seem to fit the story that AI is causing a large drop in labor demand. As long as labor supply curves slope up, reducing headcount should also reduce wages. The fact that it doesn’t suggests something is fishy.
There’s also the question of whether Brynjolfsson et al. chose the right measure of AI exposure. Their main measure is one of the ones from Eloundou (2024). Eckhardt and Goldschlag also look at that same measure; it’s one of the two that do show a slight increase in unemployment among AI-exposed workers. So that’s reassuring — we know there’s some consistency between the two papers.
But EIG’s finding means that it’s important to check all available (and credible) measures of AI exposure. Brynjolfsson et al. should probably check the measures from Felten et al. (2021), Webb (2022), and others.
They do use one other measure of exposure — the Anthropic Economic Index. This is a measure of how often people ask Anthropic’s LLM, Claude, about a particular topic. Of course, this could also just be measuring how much people use AI to complement their own skills. So Brynjolfsson et al. basically just asked the LLM whether it thought the people writing in to ask Claude about each task were trying to avoid doing that task, or make themselves better at doing it. This is how they decide which jobs are in danger of being replaced, versus which jobs will simply see productivity boosts from AI.
Honestly, I don’t put a lot of stock in this measure of AI exposure. We need to wait and see if it correctly predicts which types of people lose their jobs in the AI age, and who simply level up their own productiveness. Until we get that external validation, we should probably take the Anthropic Economic Index with some grains of salt.
So while Brynjolfsson et al. (2025) is an interesting and noteworthy finding, it doesn’t leave me much more convinced that AI is an existential threat to human labor. Once again, we just have to wait and see. Unfortunately, the waiting never ends.
Update: Josh Gans has a good post about why we might see companies hiring older workers in AI-exposed occupations:
Basically, the idea is that experience is a complement to what AI can do, while formal education (which is all young workers have) is a substitute. This would essentially just mean that AI is making on-the-job training a lot more important. Which is good news if we can solve the traditional problem associated with on-the-job training, which is “Who pays for it?”. More on this in an upcoming post.
