Automation: Why It Will Be Different

In the early 1800s, most people were employed in agriculture. Gradually, agriculture went from employing most people to employing ~2% of the population. No mass unemployment resulted. Similarly, advanced nations have seen employment in industry decline as production increases, again with no mass unemployment. Many conclude that even if the large majority of jobs are automated, we’ll find new jobs.

But in what? It’s hard to imagine, but didn’t they face the same problem in 1900? We do have jobs that someone in 1900 could not have imagined such the computer programmer, but most of the jobs people shifted to fall into two categories:

  1. Jobs that existed in 1900 such as restaurant cooks, insurance sailsmen, hotel workers, ect.
  2. Jobs which did not exist in 1900, such as airline pilots, but which serve as a substitute for something which did, as the airplane substitutes for the ship and the train.

Someone in 1900 could have imagined how many new jobs would be required if the working classes were to live like the top 5% in that era. A lot more construction workers would be required to make the houses larger. More cooks would be needed in the new restaurants. More hotels as tourism boomed. Paved roads covering a larger portion of the country. More universities and theaters. More financial institutions to put the newfound wealth.

And that is basically what happened in the 20th century. In some of those things, we got substitutes, TV substituted for theater, airplanes for passenger-transporting trains. There were some totally new products, video games and types of junk food, but the vast majority are working to supply a product or service, such as transportation from point A to point B, which existed in 1900.

Now imagine that same thought experiment today, the working class suddenly starts living like the top 5%. You would see new demand for jobs: larger houses once more, more travel, more universities, ect, but the magnitude of this demand would not be nearly so great. Many of the goods consumed by the top 5% are simply positional goods, no bigger, nor requiring any more labor to produce, than their cheaper alternatives.

So what will the future look like?

In the short term, economies may still see a healthy amount of jobs produced as consumption increases. More people will travel, people will buy larger houses, and more people will go learn useless stuff they will end up forgetting at universities, both creating jobs at universities and taking students out of the labor pool for 3 or 4 years. Government will employ ever more people in jobs of dubious value, enforcing regulations and providing various forms of “help” to the growing underclass. But eventually you will reach the limit of post-scarcity, increased consumption is insufficient to cancel out the effect of automation. At that point, the government will face the choice of employing an unprecedented number of workers at even more useless tasks, restricting automation, restricting work hours, or providing some kind of basic income.

During the transitional period, wages for unskilled people will continue to stagnate. The most marginal among them, teenagers, old people nearing retirement age, and the lazy/criminal/drug addicted will drop out of the workforce. This is already being seen in America with lower labor force participation rates, especially among young men. You will also see regional variation in automation’s effects. Those regions of the country which benefit from increased consumption and/or rent seeking, such as college towns, tourist destinations, centers of government employment, will do well. Those regions which see a high degree of automation, such as those reliant on manufacturing, will become “economically depressed” even as production increases. U.S. manufacturing output is up but employment is down:


Eventually, the whole economy outside of government employment will start to look like that graph.

This entry was posted in Economics, Technology. Bookmark the permalink.