[jsmith12673]

Economics noob here: the paper concludes that past automation is predicted by 2 key factors: pay and employment. What does that mean exactly?

Are they saying that higher a job pays, the more likely it is to get automated? Or the other way around?

Also what does 'employment' refer to exactly? Is measuring how well the demand for labour is being met in a certain job?

[JumpCrisscross]

> Are they saying that higher a job pays, the more likely it is to get automated?

Yes. "Simple economic theory predicts that, all else equal, employers are more eager to automate jobs with higher pay and more workers. So these two factors should predict job automation. And we do in fact see such effects in Table 3, though more consistently for pay than employment."

> what does 'employment' refer to exactly?

Number of people employed.

[Nextgrid]

For automation to make sense from a business point of view the cost of developing & running the automation should be less than the cost of doing the work manually over an equivalent timeframe.

This is why for example there are a lot of bullshit office jobs that can be automated fairly easily but will not be because there's plenty of cheap labour to do it manually (for cheaper than what developing & maintaining the automation would cost) and it also earns them PR points about "creating" jobs.

[foogazi]

No one will automate for automations sake - businesses choose how to deploy capital to improve efficiency, which can be achieved through scale or going after the largest cost

[082349872349872]

In the mid-twentieth century (IIRC) there was a "paradox" observed that the US was capital-rich but its exports were labour-intensive. Is this situation still true?

[PeterisP]

"Are they saying that higher a job pays, the more likely it is to get automated?"

Yes. One aspect that I've seen in e.g. computer vision and robotics for industrial automation is that there are a lot of processes for which we have the technology to automate production, but it's not worth because it's currently cheaper to get the job done manually by someone in a poor country getting paid very, very little.

However, the obvious implication is that if/when the worker conditions in the world improve and these workers start requiring reasonable wages then these jobs will be automated pretty much immediately. If the shop owner has to choose between buying a robot or providing a handful of rice to a team of near-slaves, the people are cheaper; If the shop owner has to pay something comparable to the minimum wages in the first world, then the best worker gets to supervise and clean the imported robots and all the others are out on the street.

If the first world would stop offshoring these jobs to poor countries, then that would not mean large numbers of blue collar jobs coming back, that would mean shifting to automation.