I'm not an academic (but I am a new paid subscriber!) so I don't read many academic papers. It seems to me that it would require mind reading to come up with an accurate answer to this question. By that I mean:
When a factory automates (or moves overseas) the purpose of doing so is to produce the same or better quality output at a lower unit cost. If a company that takes these steps is alone in its industry in doing so while its competitors operate as before, then, assuming it gets the better quality and lower costs as planned, it will benefit with higher profit margins. However, if competitors also do the same thing with the same results, it is only a matter of time before the now higher margins cause a price war to break out, as smaller players benefit more from gaining market share through discounts than they do by keeping margins high. The price war continues until profit margins for everyone have shrunk to the point where the ROI on new investment is no longer that attractive.
The workers at the automated factory are now more productive and can demand more pay, while the workers whose jobs were shipped overseas are at least temporarily unemployed. But looking at the factories with more robots to count the number and pay of the human employees is missing the point. The fact that the industry after the price war is now selling better products at a lower price than before the automation/offshoring means that its customers can get what they want at a lower price and have money left over.
What do customers do with that money left over? They spend it on something, and it would take mind reading to know what that was. But whatever it is, that will require more employees to provide.
The example I've used is the fact that entry of Chinese manufacturing into world markets in a big way starting in roughly 1990 +/- allowed Americans to save lots of money on what they bought. At the same time there began rapid growth in the number of health clubs, yoga and Pilates studios, etc. Maybe it was a coincidence, but I think the savings on cheap products gave consumers more to spend on personal improvement services. That created big growth in the number of employed personal trainers, yoga instructors, etc.
I'm not saying that a newly unemployed 60 y.o. factory worker in Georgia could move to Denver to teach yoga, just that there was a big jump in the number of employed yoga instructors. I'm not sure how their pay stacked up versus working in a dirty, noisy textile factory, but working conditions were certainly better if nothing else.
But I'm not a mind reader. I can't prove that that is how people spent their money, but they got it from somewhere, and lower consumer prices is a good candidate. Unless these academic studies try to take that into account, they are missing an important part of the effects on employment of automation and offshoring.
All the while - factories, even automated factories have productive capacity. Like in the 1940s, those car factories were reconfigured to produce tanks and planes, and after 1945 went back to producing new cars. While yoga instruction is a service, and not a particularly hi-tech at that.
I also think there is a classification problem - the IT concerns like Google, Microsoft, Facebook, Amazon or Paypal are all industry, not services.
Very interesting. I'm curious if you could drive a bit deeper into the displacement though. When low skill labor is getting replaced by higher skill labor, how quickly is this seen happening relative to the adoption of increased automation and are the studies tracking whether these are the same people (as in the Japanese model of upskilling existing workers) or something more along the lines of the typical objection that native workers are losing their jobs and then being replaced with immigrants on work visas?
The experience of the Rust Belt with offshoring certainly suggests that traditional economic assumptions that workers will naturally and quickly find other work after being displaced are not necessarily correct. Likewise, the "learn to code" meme (implying that it's easy to transition from lost jobs into new tech jobs) has aged badly, showing both that upskilling is often more difficult than assumed due to mismatches between the qualities demanded by the previous positions and the new positions and the more recent significant cuts in both wages and total employment in many tech positions showing that the pace of change is making it increasingly difficult to even determine what skills displaced workers 'should' be retraining into.
There are concerns that even when the aggregate is neutral or mildly positive, there are severe concentrated losses in certain places and sectors, so it is quite logical for the workers in those to oppose changes that will drastically harm them in return for diffuse mild benefits to a larger number elsewhere.
Similarly, I'm interested in transition choke points regarding the pace of change and numbers displaced. Upskilling has relatively limited pipelines in many cases. How do we measure the available capacity to retrain, identify the productive areas to retrain into, and give workers the signals and resources needed to align themselves into new careers without creating the equivalent of traffic jams in our retraining pipelines?
It is notable that the decline of the Rust Belt is associated not so much with being wracked by trade, but by insufficiently free trade: https://www.nber.org/papers/w20538
The problem is that automation requires elite blue collar - hard to get that without moving education back to where it was in the 1950s. Plus introducing selective schools, plus technical training, plus training for discipline, impulse control and delayed gratification.
When manufacturing is less than 20% of GDP and even less of labor, I don’t think even if robots replaced most of those workers the unemployment numbers would not produce a groundswell for UBI nor the tax revenue to support it. AI may impact more jobs, but right now it’s an enhancement not a replacement for existing workers.
It won’t just be human labor in manufacturing being automated but many different industries from service to financial services, including people with post-secondary degrees. UBI is inevitable.
I work a job currently threatened by automation. I’m not going to make a fool of myself welcoming my own replacement! Automate a large portion of good American jobs and what kinds of jobs do you expect to replace them? BS email jobs, driving Uber, sex work on Onlyfans, and the like. No thanks!
The thing about the gig economy is it allows for both low unemployment AND the existence of a precariat serf class with no benefits who can be made docile with UBI and the hope of employer-provided health insurance. More automation may allow the US to remain “competitive” on a macro level but at the cost of further inequality and social unrest/alienation, as already happened with the automation and offshoring of manufacturing jobs.
I'm not an academic (but I am a new paid subscriber!) so I don't read many academic papers. It seems to me that it would require mind reading to come up with an accurate answer to this question. By that I mean:
When a factory automates (or moves overseas) the purpose of doing so is to produce the same or better quality output at a lower unit cost. If a company that takes these steps is alone in its industry in doing so while its competitors operate as before, then, assuming it gets the better quality and lower costs as planned, it will benefit with higher profit margins. However, if competitors also do the same thing with the same results, it is only a matter of time before the now higher margins cause a price war to break out, as smaller players benefit more from gaining market share through discounts than they do by keeping margins high. The price war continues until profit margins for everyone have shrunk to the point where the ROI on new investment is no longer that attractive.
The workers at the automated factory are now more productive and can demand more pay, while the workers whose jobs were shipped overseas are at least temporarily unemployed. But looking at the factories with more robots to count the number and pay of the human employees is missing the point. The fact that the industry after the price war is now selling better products at a lower price than before the automation/offshoring means that its customers can get what they want at a lower price and have money left over.
What do customers do with that money left over? They spend it on something, and it would take mind reading to know what that was. But whatever it is, that will require more employees to provide.
The example I've used is the fact that entry of Chinese manufacturing into world markets in a big way starting in roughly 1990 +/- allowed Americans to save lots of money on what they bought. At the same time there began rapid growth in the number of health clubs, yoga and Pilates studios, etc. Maybe it was a coincidence, but I think the savings on cheap products gave consumers more to spend on personal improvement services. That created big growth in the number of employed personal trainers, yoga instructors, etc.
I'm not saying that a newly unemployed 60 y.o. factory worker in Georgia could move to Denver to teach yoga, just that there was a big jump in the number of employed yoga instructors. I'm not sure how their pay stacked up versus working in a dirty, noisy textile factory, but working conditions were certainly better if nothing else.
But I'm not a mind reader. I can't prove that that is how people spent their money, but they got it from somewhere, and lower consumer prices is a good candidate. Unless these academic studies try to take that into account, they are missing an important part of the effects on employment of automation and offshoring.
All the while - factories, even automated factories have productive capacity. Like in the 1940s, those car factories were reconfigured to produce tanks and planes, and after 1945 went back to producing new cars. While yoga instruction is a service, and not a particularly hi-tech at that.
I also think there is a classification problem - the IT concerns like Google, Microsoft, Facebook, Amazon or Paypal are all industry, not services.
Very interesting. I'm curious if you could drive a bit deeper into the displacement though. When low skill labor is getting replaced by higher skill labor, how quickly is this seen happening relative to the adoption of increased automation and are the studies tracking whether these are the same people (as in the Japanese model of upskilling existing workers) or something more along the lines of the typical objection that native workers are losing their jobs and then being replaced with immigrants on work visas?
The experience of the Rust Belt with offshoring certainly suggests that traditional economic assumptions that workers will naturally and quickly find other work after being displaced are not necessarily correct. Likewise, the "learn to code" meme (implying that it's easy to transition from lost jobs into new tech jobs) has aged badly, showing both that upskilling is often more difficult than assumed due to mismatches between the qualities demanded by the previous positions and the new positions and the more recent significant cuts in both wages and total employment in many tech positions showing that the pace of change is making it increasingly difficult to even determine what skills displaced workers 'should' be retraining into.
There are concerns that even when the aggregate is neutral or mildly positive, there are severe concentrated losses in certain places and sectors, so it is quite logical for the workers in those to oppose changes that will drastically harm them in return for diffuse mild benefits to a larger number elsewhere.
Similarly, I'm interested in transition choke points regarding the pace of change and numbers displaced. Upskilling has relatively limited pipelines in many cases. How do we measure the available capacity to retrain, identify the productive areas to retrain into, and give workers the signals and resources needed to align themselves into new careers without creating the equivalent of traffic jams in our retraining pipelines?
It is notable that the decline of the Rust Belt is associated not so much with being wracked by trade, but by insufficiently free trade: https://www.nber.org/papers/w20538
The problem is that automation requires elite blue collar - hard to get that without moving education back to where it was in the 1950s. Plus introducing selective schools, plus technical training, plus training for discipline, impulse control and delayed gratification.
UBI on some level is all but inevitable.
When manufacturing is less than 20% of GDP and even less of labor, I don’t think even if robots replaced most of those workers the unemployment numbers would not produce a groundswell for UBI nor the tax revenue to support it. AI may impact more jobs, but right now it’s an enhancement not a replacement for existing workers.
It won’t just be human labor in manufacturing being automated but many different industries from service to financial services, including people with post-secondary degrees. UBI is inevitable.
Yep
Great work, I always knew automation is a net positive for people
I work a job currently threatened by automation. I’m not going to make a fool of myself welcoming my own replacement! Automate a large portion of good American jobs and what kinds of jobs do you expect to replace them? BS email jobs, driving Uber, sex work on Onlyfans, and the like. No thanks!
The thing about the gig economy is it allows for both low unemployment AND the existence of a precariat serf class with no benefits who can be made docile with UBI and the hope of employer-provided health insurance. More automation may allow the US to remain “competitive” on a macro level but at the cost of further inequality and social unrest/alienation, as already happened with the automation and offshoring of manufacturing jobs.