The Number That Made Researchers Do a Double Take
Picture this: You are a factory worker in your 40s. Your job involves loading parts into a machine, pressing a button, and watching a robot arm do the rest. Your employer just bought a new automation system. Your coworkers whisper about layoffs. You update your resume.
Now imagine your paycheck arriving three years later. It is 7 percent fatter than it would have been if that robot arm had never shown up.
That is not a hypothetical. That is a finding from a 2022 working paper by economists at MIT, Boston University, and the National Bureau of Economic Research. They tracked what happened to wages at firms that adopted industrial robots between 1990 and 2016. The conventional story says automation crushes low skilled workers. The data told a different story.
The researchers found that when a company installed robots, wages for workers without a college degree actually rose. Not just a little. They rose more than wages for college graduates at the same firms. The robot arm did not replace the worker. It made the worker more productive. And when workers become more productive, their pay tends to follow.
This is not what most people expect. The public conversation around automation has been dominated by fear. The robot is coming for your job. The algorithms will replace you. But a growing body of research suggests the reality is more complicated and more interesting. Automation does not just raise wages for the programmers who build the machines. It can raise wages for the people who work alongside them.
The German Data That Changed the Story
The MIT paper is not an outlier. It confirms a pattern that has been hiding in plain sight in the data for years.
In 2017, a team of economists led by David Autor at MIT published a landmark study of German manufacturing firms. Germany is a useful case study because it adopted robots earlier and more aggressively than almost any other country. If automation destroys low skilled jobs, Germany should be a disaster zone.
Autor and his colleagues found that each additional robot per 1,000 workers reduced the probability of being laid off for the average worker. Not increased it. Reduced it. The robots were not replacing workers. They were allowing firms to produce more, compete better, and survive.
The wage effects were even more striking. Workers at firms that adopted robots saw their wages grow faster than workers at firms that did not. The effect was largest for workers in the middle of the skill distribution: machinists, assemblers, repair technicians. These were not tech workers. These were people who ran machines for a living.
The key mechanism was productivity. When a factory adds a robot, the output per worker goes up. The firm can either keep the same number of workers and produce more, or keep the same output and reduce workers. In Germany, most firms chose the first path. They kept their workers, sold more goods, and shared some of the gains through higher wages.
Why the United States Looks Different
Here is where it gets tricky. The German story is not the American story. At least not entirely.
In the United States, the relationship between automation and wages is more mixed. A 2020 paper by economists at the University of Zurich and the London School of Economics examined the impact of industrial robots on local labor markets across the United States. They found that robot adoption reduced employment and wages in the regions most exposed to automation.
But that was the aggregate effect. When you zoom in to the firm level, the picture changes. The same MIT team that studied Germany also looked at American firms. They found that, just like in Germany, workers at firms that adopted robots saw their wages rise. The difference was that fewer American workers were at those firms. The benefits were concentrated, not shared.
Why? The researchers point to two factors. First, American firms are more likely to lay off workers when they automate. German firms have stronger labor protections and works councils that give workers a voice in the process. Second, American workers are more likely to quit their jobs and move to other firms. When they do, they often end up at firms that have not automated, where wages are lower.
The lesson is not that automation is bad for low skilled workers. The lesson is that the distribution of the benefits depends on institutions. In Germany, the gains are shared broadly. In the United States, they are captured by a smaller group.
The Counterintuitive Case of the Call Center
Manufacturing is not the only place where automation raises wages for low skilled workers. The service sector tells a similar story.
Consider the call center. It is the kind of job that people imagine will be the first to go. The AI can handle the simple questions. The chatbot can direct the rest. Why would you pay a human to answer the phone?
A 2021 paper by economists at Stanford, Harvard, and the University of Toronto studied what happened when a large technology company introduced an AI assistant for its customer service representatives. The AI did not replace the workers. It gave them a tool. The tool suggested responses to common questions, flagged relevant information, and handled the most routine interactions.
The result? The workers became more efficient. They handled more calls per hour. They resolved issues faster. And their wages went up.
The researchers found that the AI assistant increased productivity by 14 percent on average. The least experienced workers benefited the most. Their productivity jumped by 35 percent. The AI was not making the best workers better. It was making the weakest workers competent.
This is a pattern that repeats across industries. Automation does not just replace people. It augments them. It turns a worker who could only handle the simple cases into a worker who can handle the hard ones. And that makes them more valuable.
The Skill That Nobody Talks About
There is a subtle point here that most discussions of automation miss. The wage gains do not come from the automation itself. They come from the reorganization of work that automation enables.
When a factory installs a robot, the robot does not just do the same job faster. It changes the entire workflow. The worker who used to spend eight hours loading parts now has time to inspect the final product, troubleshoot problems, and coordinate with other teams. These are higher value tasks. They require judgment, experience, and communication skills.
A 2019 paper by economists at Princeton and the University of Chicago studied how firms reorganize work around automation. They found that the most successful firms did not simply replace workers with machines. They redesigned jobs so that humans and machines complemented each other. The human handled the parts that required flexibility, problem solving, and social interaction. The machine handled the repetitive, physically demanding, or dangerous parts.
This is why the wage gains go to low skilled workers. The automation does not make their existing skills obsolete. It creates demand for skills they already have: the ability to read a situation, make a judgment call, and communicate with a colleague. These skills were always valuable. The automation just made them more visible.
The Limits of the Research
The evidence is real but it has boundaries. The studies I have cited focus on specific types of automation: industrial robots in manufacturing and AI assistants in call centers. They do not cover every kind of automation.
The researchers also note that the wage gains are not automatic. They depend on how firms choose to use the technology. A firm that views automation as a way to cut costs and reduce headcount will produce very different outcomes than a firm that views it as a way to expand production and invest in workers.
There is also the question of timing. The studies cover periods when automation was adopted gradually. What happens when automation accelerates? What happens when AI starts doing tasks that require judgment and communication? The answer is not yet clear.
Finally, the research does not address the workers who are displaced. Even in the best case scenario, some workers lose their jobs to automation. The wage gains for the workers who stay are real. But they are cold comfort to the worker who has been laid off.
What This Actually Means
- ▸Stop assuming automation is a zero sum game. The data shows that firms can adopt robots and raise wages for low skilled workers at the same time. The outcome depends on how the technology is deployed, not just whether it is deployed.
- ▸Pay attention to how firms reorganize work, not just which jobs they eliminate. The wage gains come from redesigning jobs so that humans and machines complement each other. The worker who used to load parts now inspects the product. That is the transformation that matters.
- ▸The institutions matter more than the technology. German workers saw bigger gains than American workers because German firms were less likely to lay people off and more likely to share productivity gains. The same robot produces different outcomes in different systems.
- ▸The least experienced workers benefit the most from augmentation tools. The call center study showed that AI helped the weakest workers the most. Automation can be a leveling force if it is designed to support people, not replace them.
- ▸The real risk is not automation itself. It is how we choose to distribute the gains. The technology creates more value. The question is whether that value ends up in the pockets of shareholders or in the paychecks of workers. The research shows both outcomes are possible. The choice is ours.