How Jobs Actually Disappear
In a Nutshell
Jobs disappear mainly through gradual task automation rather than mass layoffs, as companies stop replacing departing workers and never advertise the positions again. This creates a slow, silent erosion of opportunities that the unemployment rate and layoff statistics fail to capture. The result is lasting harm for displaced workers, who lose job-specific skills and earnings potential, while new jobs emerge too slowly or require different abilities to offset the losses.
These notes were generated by AI and may contain inaccuracies.
When people imagine automation taking over jobs, they tend to envision a single dramatic moment. A factory gate closes, a headline announces the layoff of 10,000 workers, or a robot enters as a queue of workers leaves. This is already happening, but it's not the way most jobs disappear. Most jobs disappear quietly through such a gradual process that no one can pinpoint the exact day it happened.
The quiet type of displacement is the one that should be most concerning precisely because it is so easy to ignore.
A job is not a single thing. It's a set of tasks. A bank employee counts money, processes deposits, answers questions, detects fraud, and sells products. The legal assistant reviews documents, drafts letters, organizes files, and checks citations.
Automation rarely takes over an entire job in one fell swoop. Instead, it takes on tasks one at a time.
Economists Daron Acemoglu and Pascual Restrepo have built an influential framework around this idea. When machines take over tasks that humans used to perform, they call it the displacement effect. When the economy creates new jobs in which humans have an advantage, they call that the redemption effect.
In the four decades following World War II, these two powers were roughly balanced. Technology has destroyed some types of work and created new ones, and overall employment has continued to grow. But their research indicates that in recent decades in the United States, displacement has been occurring at a faster pace than restoration.
Here's what this looks like in reality. Think of an automated teller machine, or ATM. ATMs proliferated across the United States in the 1980s and 1990s, and you might have expected bank employees to disappear. But they did not disappear. At least, not at the beginning. Because ATMs made running each branch less expensive, banks opened more branches, and continued to employ tellers for years.
Then smartphones appeared. Mobile deposits, online transfers, and app-based banking have reduced the reasons for visiting a branch, and since then, the employment of tellers has been in steady decline. There was not a single day when that job was fully automated. Instead, there were 30 years of small discounts.
Or think of the telephone operator. In the early twentieth century, making a call required someone to sit in front of a keyboard and plug the cables into the correct sockets, and that person was usually a young woman. Bell System employed female workers in huge numbers.
Then, city after city, the company switched to mechanical communication. Economists James Feigenbaum and Daniel Gross studied what happened next, and their findings point to two opposing trends.
Women who worked as telephone operators were affected. After 10 years, the likelihood of them continuing to work as operators was much lower, and many of those who remained moved on to lower-paying jobs.
The next generation performed better than you might expect, because the increased demand for secretaries and restaurant workers absorbed most of the young women who would have otherwise worked in dial-up jobs.
But the youngest entrants to the labor market, aged between 16 and 18, often ended up in low-paying jobs. And so the local economy adapted, but part of the cost of this adaptation fell on people who never knew which door had been closed, because you can't see a job that was never offered.
This is the pattern to notice. The most common way automation reduces employment is not through layoffs. It is the job vacancy that is never advertised.
Let's do the math. Imagine a company with 1000 employees. Each year, about 10% of them leave due to retirement, relocation, or a change of career path. Normally, the company replaces them.
Now, suppose it adopted a program that takes on an increasing part of its routine work, and decided to stop replacing people who leave. No one is being fired, and there are no newspaper headlines. After one year, the company had 900 employees. Two years later, it had 810 employees. After 3 years, it had 729 employees. This is a 27% reduction in the number of employees, which is more than a quarter of the workforce without laying off a single employee.
The press release talks about efficiency and natural decline, but the local newspaper does not cover the matter. Hundreds of people who would have been employed go to look for work elsewhere, often without ever knowing that a door has been closed.
Now, apply this to the entire economy. When people lose their jobs due to structural change, the evidence is alarming. Studies of displaced workers date back to the early 1990s and have found that the earnings of these workers often remained much lower for many years afterward, sometimes permanently. They are not just losing a salary. They also lose the accumulated value of skills and relationships that were specific to the job that disappeared.
This is important because the quiet version of displacement defeats the usual early warning systems. We are monitoring the unemployment rate, but the unemployment rate can remain low while the quality of opportunities erodes. We are monitoring mass layoffs, but the bigger story is the hiring that never happens. We tell ourselves that new jobs are always appearing, and often that actually happens. But no one can promise in advance which jobs will appear, or where they will appear, or who they will be for. It may require different skills and pay different wages than the jobs it replaces.
If we want to see automation clearly, we need to look at different numbers. Monitor the labor force participation rate, especially among people in their prime working years. Monitor the share of national income that goes to wages. Monitor the hiring of young people for entry-level positions. Observe whether new types of work are created at the same rate that old types disappear.
There is a reason why this is important that goes beyond just economics. The slow erosion of opportunities rarely leads to a single moment of public decision. It is not voted on by a legislative body, it is not announced by an executive official, and there is no protest with a clear objective. Statistics always arrive late, so by the time a pattern emerges, the choices that led to it have long since been made.
None of this should cause panic, but it certainly calls for attention. The forces involved are slow, cumulative, and easily justifiable at each individual step, and that is precisely what makes them so powerful.
Jobs don't usually disappear with a lot of fuss. Instead, they disappear through thousands of small deletions. By the time the total becomes clear, it is very difficult to recover those jobs.
Keep David Shapiro in your library
Save the videos and channels worth coming back to, and find them again in one place.
