In his February 6 opinion essay1, Paul Krugman declared “A.I. and automation…won’t ultimately take away jobs, and neither will immigrants.” His reasoning was riddled with mistakes. He couldn’t be more wrong.
Krugman repeats an outdated, false meme that technology creates more jobs than it displaces. This meme is common among people unfamiliar with the nature of technology. Simply put, the nature of technology is to do more with less work. Producing more with less work is the raison d’etre for technology. It does just that. He conflates technology with the expansion of our economy due to the innovators who find new products and services for people to spend their money on and to employ the workers displaced by technology. It seems he infers a causal connection when it is really coincidental.
Ironically, Krugman uses agriculture as an example to bolster his case but botches it. He does acknowledge that the sector now employs a third as many people as it did in 1952, but it’s much bigger than that. From 1870 to 2022, direct farm employment went from about 53% of the labor force to about 1.4%. In 1870, direct on-farm workers numbered 6,790,000, while that shrank to 2,600,000 in 2020, a decline of 62% during a span when the overall population increased 7-1/2 times.2 All this occurred because of technological innovations: John Deere’s tractor, Clarence Birdseye’s food-freezing method, Fritz Haber’s ammonia synthesis, Norman Bourlag’s green revolution, etc. The industry is now bigger than that: Sectors included in agriculture now include food and beverage manufacturing, supermarkets, and food service, eating, and drinking places; most of which did not exist in 1870! In 2020, agriculture and its related industries provided 10.3 percent of U.S. employment. Of this, food service, eating, and drinking places accounted for the most significant share.3
The biggest evidence of job-replacing technology can be found in the labor force participation rate (LFPR). The LFPR has been declining since Q1 2000, a point I call Peak Jobs. There are three principal causes, not just one, as inferred by Krugman: “It’s important to take into account the effects of an aging population, which has caused a long-term downward trend in labor force participation.” A study by NBER stated that, to some degree, the 5.5 million manufacturing jobs lost between 2000 and 2017 were due to “dramatic changes over the past two decades, most notably in automation and the rise of robotics.”4
But this doesn’t take into account something developing in this century that changes everything. Innovators are now working toward a marriage of AI and robotics. Once they put those two together in a package that thinks and moves like a human being, this technology will no longer enhance workers’ productivity, it will become the worker. Another false meme in circulation is the idea that new technologies create jobs because you need humans to build, program, operate, and fix them. Not this technology. Already AI thinks and learns faster than humans, robots build robots, AI programs itself, and self-repairing machines and materials are coming on the market. When all this is put together in one package, tech will replace human workers in large numbers.
Krugman, by citing raw data, misses the big picture, the context of the labor force. He does acknowledge the LFPR decline in this century but fails to mention the other two major forces—automation and globalization— which have as much if not more of an effect: The “lump of labor” he talks about is the size of the labor force within the population. It is growing along with population growth but failing to keep pace with the rate. Consequently, the “lump” is shrinking in proportion.
Labor Force Participation Rate5
A Brookings Institute Hamilton Project report6 looked into a surplus labor force of people who could potentially be in the labor force but chose not to. Pre-Covid, “About 24 million “prime-age” Americans are not in the labor force.” These are people who are neither working nor actively seeking work.
With 6+ million Americans officially unemployed and another 24 million potentially employable, why do we need to import workers?
The claim that immigrants do not take jobs away from native-born workers is also wrong. Of course, they do, to some degree. There’s hard evidence to prove it. Take a look at Christopher Leonard’s “The Meat Racket” and Eric Schlosser’s “Fast Food Nation.” Their excellent investigative reporting shows how the meat processing industry prefers immigrant labor to native-born because they can exploit them. Brilliant investigative reporting by the New York Times has exposed subcontractor labor in the same industry using child immigrant labor to do the dangerous job of cleaning bone-and-meat cutting machines.
Numerous media resources have exposed the widespread abuse of the H1B visa program. The largest intentional abuse is to hire skilled Indian workers from temp agencies. Then, they force current workers to train their replacements before being laid off. Pulitzer-Prize-winning reporter Hedrick Smith, in his book “Who Stole the American Dream?” lists many cases of this happening in corporate America. He estimates that onshoring has cost over one million jobs lost to foreigners.7 And that doesn’t include the offshoring of services, such as call centers, to Asia. Offshoring is the third major cause of the declining LFPR.
Krugman himself is one of the “people who imagine that they’re sophisticated and forward-thinking when they’re actually recapitulating old fallacies.” Of course, technology and immigrants replace native-born workers; solid evidence proves it.
Notes:
“Trump, Immigration and the Lump of Labor Fallacy,” Feb. 6, 2024, Paul Krugman, https://www.nytimes.com/2024/02/06/opinion/trump-immigration-lump-of-labor.html
1870 figures come from NBER publication Labor Force and Employment, 1800–1960, Stanley Lebergott, Chapter URL: http://www.nber.org/chapters/c1567
“Ag and Food Sectors and the Economy”, USDA, https://www.ers.usda.gov/data-products/ag-and-food-statistics-charting-the-essentials/ag-and-food-sectors-and-the-economy/ , accessed 6.22.22
“The Transformation of Manufacturing and The Decline In U.S. Employment,” Charles et al, Working Paper 24468, http://www.nber.org/papers/w24468 .
U.S. Bureau of Labor Statistics, Labor Force Participation Rate [CIVPART], retrieved from FRED, Federal Reserve Bank of St. Louis; https://fred.stlouisfed.org/series/CIVPART, February 9, 2024.
“For Labor Day, facts about US jobs and workers”, Brennan Hoban, Friday, September 1, 2017, Brookings online, https://www.brookings.edu/blog/brookings-now/2017/09/01/for-labor-day-facts-about-u-s-jobs-and-workers/?utm_campaign=Brookings%20Brief&utm_source=hs_email&utm_medium=email&utm_content=55909542 accessed 9.10.19\
Smith, Hedrick, “The Skills Gap Myth”, p. 290, Random House, 2012
“Trump, Immigration and the Lump of Labor Fallacy,” Feb. 6, 2024, Paul Krugman, https://www.nytimes.com/2024/02/06/opinion/trump-immigration-lump-of-labor.html
What did Kurt Vonnegut, the novelist, and François Mitterrand, the socialist president of France from 1981 to 1995, have in common with Donald Trump? Both, at some point, believed in what economists call the lump of labor fallacy. This is the view that there is a fixed amount of work to be done and that if someone or something — some group of workers or some kind of machine — is doing some of that work, that means fewer jobs for everyone else.
And Trump clearly shares that belief. As I noted in my most recent column, it underlies his hostility to immigration — well, that and his belief that immigrants are “poisoning the blood of our country.” It also underlies his protectionism.
So this seems like a good time to talk about the lump of labor fallacy, how we know it’s a fallacy and why it’s a zombie — an idea that refuses to die and instead keeps shambling along, eating people’s brains.
First, about Vonnegut and Mitterrand.
Vonnegut’s first novel, “Player Piano,” published in 1952, envisaged a grim future in which automation has led to mass unemployment: The machines can do everything, so there’s no need for human workers.
Mitterrand, coming to power in a nation that had experienced a large rise in unemployment since the early 1970s, reduced France’s retirement age from 65 to 60, in part because he and his advisers believed that encouraging older French citizens to leave the work force would free up jobs for younger workers. Mitterrand’s successors have spent decades trying to undo the damage.
Why is there always a substantial group of people — the lumpencommentariat? — who believe that there’s a limited amount of work to be done, so machines that increase productivity or immigrants entering the work force take away jobs? Many of these people probably haven’t even tried to think their views through. But it’s also true that something like the lump of labor story does make sense if you think about an individual industry in isolation.
For example, long ago one of my uncles operated a factory using plastic injection molding to produce lawn ornaments — basically, he was supplying the then-burgeoning suburbs of New York with pink flamingos. Since there was, presumably, a limited demand for pink flamingos, machines that allowed production of pink flamingos with fewer workers would reduce employment in the industry, while entry of new producers would take away jobs from existing pink-flamingo workers.
Or to take a less whimsical example, there are limits to the amount of food people want to consume, so rising productivity in agriculture leads to reduced need for farmers. America has about twice as many people now as it did when Vonnegut published “Player Piano” but employs only around a third as many people in agriculture.
But while there’s limited demand for pink flamingos or wheat, there’s no evidence that there’s limited demand for stuff in general. When incomes rise, people will find something to spend their money on, creating jobs for workers displaced by technology or newcomers to the work force. Machines do, in fact, perform many tasks that used to require people; output per worker is more than four times what it was when Vonnegut wrote, so we could produce 1952’s level of output with only a quarter as many workers. In fact, however, employment has tripled.
Just to be clear, I’m not saying that technological progress can never hurt workers or some groups of workers. Technology that makes a traditional occupation largely disappear — for example, the way freight containerization more or less eliminated the need for longshoremen — can be devastating for those displaced. And what’s referred to as biased technological change, which reduces demand for some inputs while increasing it for others, can reduce the real incomes of large groups. Many economists believe that skill-biased technological change, which raises the demand for highly educated workers as an input and decreases it for less skilled labor, has been a factor in rising inequality, although many others, myself included, are skeptical. It’s at least arguable that capital-biased technological change caused real wages to stagnate during the early stages of the Industrial Revolution.
But the crude argument that technological progress causes mass unemployment because workers are no longer needed is just wrong.
What about competition from new workers? If you’re worried about immigrants taking jobs away from native-born Americans, consider the effect of a truly huge influx to the labor market: the mass movement of American women into paid work from the mid-1960s to around 2000. Did working women take jobs away from men? I’m sure many men thought they would. But they didn’t. Here are rates of employment during prime working years, 25 to 54, for men and women over time:
The big rise in women’s employment didn’t come at men’s expense. True, there has been a small decline in male employment over the past six decades, perhaps reflecting the decline of manufacturing and the emergence of left-behind regions in the heartland. But the millions of women entering the paid work force clearly didn’t displace male workers.
Which brings me to current concerns about immigration. As I noted in my column, Trump and those around him clearly believe that immigrants take jobs away from native-born Americans. And I also noted that all of the increase in employment since the eve of the Covid-19 pandemic has involved foreign-born workers. So did this rise in immigrant employment come at the expense of native-born workers?
When looking at numbers here, it’s important to take into account the effects of an aging population, which has caused a long-term downward trend in labor force participation. So I asked Arindrajit Dube of the University of Massachusetts, Amherst, one of America’s top labor economists — and someone who knows his way around Bureau of Labor Statistics data much better than I do — to calculate employment rates among prime-age native-born Americans. Here’s what he found:
Even though immigrants as a group are responsible for all recent employment growth, they haven’t been taking jobs from the native-born, who are more likely to be employed in their prime working years than they were before the pandemic.
By the way, I mentioned earlier that Trump’s protectionism involves the same kind of lump-sum thinking that pervades his views on immigration. Trump and those around him, like Peter Navarro, his top trade adviser — currently facing prison time for contempt of Congress — are obsessed with trade deficits. If you read what they’ve had to say on the subject, it’s clear that they imagine that there’s a fixed amount of demand in the world and that any business that goes to foreigners is business lost to America. It’s lump-sum all the way.
Now, of course I don’t think that this evidence — or, for that matter, any evidence on any subject — will change Trump’s thinking on this or anything else. But there are some people who imagine that they’re being sophisticated and forward-thinking when they’re actually recapitulating old fallacies. No, A.I. and automation, for all the changes they may bring, won’t ultimately take away jobs, and neither will immigrants. Don’t join the lumpencommentariat.