There is a lot of question, concern and doubt surrounding the idea of technological unemployment. So much discussion that there is essentially a backlash of writing on how the issue is overhyped. The arguments are pretty standard, robots and AI aren’t THAT far along yet. Still plenty to do. And the most common, people have been afraid of technology before and it has always created new jobs and new work. People, like taxi drivers and truck drivers will simply have to get retrained. But the problem with the discussion is not that one side is wrong, rather it is that we are having two different conversations. The miscommunication lies in timeframe. All of the arguments for robs NOT disappearing are short run economic arguments. My argument is a Long Run argument. I have argued here:
The summary, is that in the Long Run, there are precious few skills that AI and Automation will not do better than a human. As a result, the only jobs for humans that will remain are those where the human controls the capital/labor decision or where the job values a human for their actual humanness. In that scenario, unemployment is well over 80% and society is dramatically changed. So, if you accept my long run argument, you would be right to wonder how we get from here to there. That is what this post is about. I believe that by highlighting this long run process, that it will help calm some short term fear and help people to better understand technological unemployment.
- Companies will put off hiring to absorb displaced workers due to automation
- We will NOT see significant layoffs related to Automation and this will be fuel of the technological unemployment nay-sayers, who inaccurately use ATMs as their argument for how technology creates jobs
- Job growth and hiring will continue to slow over the coming decade (decreased by more than 50%, while population growth has remained fairly steady)
- New graduates will continue to see a good job market as they are the most likely to have current skills
- 40 and 50-somethings will become the unemployed, as the technological skills pass them by
- 40 and 50 something will become the rise of the “part-time” worker, consultant, jack-of-all-trades and according to the BLS marginally attached U6 unemployed or completely unmeasured
- Economic shocks will result in higher levels of layoffs, similar to the 2009 Financial crisis and in contrast to the previous 40 years of market shocks.
- U6 (BLS — total unemployed and marginally attached workers) numbers become the key to understanding true unemployment. This number currently stands at 8.6% and hit a high of 17.1% during the 2008-09 financial crisis.
Let’s use Walmart as our example to understand a few of these points, ignoring the innovation that may occur in the near future, let’s start the clock with Walmart moving to autonomous truck drivers. here’s the likely process.
Step 1 — autonomous driving, but still with a driver. In a few test cases. This is standard practice for any company, especially when big changes are made. Small scale tests for a fixed period of time to determine scalability, feasability and roll out procedures. No one gets unemployed
Step 2 — Step 1 was a perfect success and the plan is to replace all truck drivers with full autonomous tractor trailers. More than 7,000 people out of work, right? No, you’d be wrong. Companies hate to fire people. It is bad press, it can damage their brand. It makes them look profit-centric, instead of customer oriented. So instead, Walmart will announce the move, with big publicity to highlight the savings and at the same time they will also announce that all 7000 employs will remain and be retrained.
How does this work? Essentially the company will go on a semi-hiring freeze. They are now overstaffed by 7k people. But a company like Walmart is turning over about 50k employees per year. So if they only hire 43k in a given year, that is something that people will hardly notice. In fact they can put out a press release highlighting that they hired 43k people in this particular year. The problem though, is that they hired 7k LESS than they would have or should have.
So that’s how it starts. Technological Unemployment begins with a reduction in hiring and we’ve been experiencing it for a decade already. When that happens at company after company, then hiring will stagnate first. However it will be difficult to see. New graduates will continue to be hired, especially if the education system is able to tailor its programs to match demands of a dynamic marketplace. The Bureau of Labor Statistics data backs this up.
The rate of job creation has slowed pretty dramatically. From 1970–2000, the US was creating about 2 mil jobs per year, consistently. Since 2000, that rate has fallen to less than 1 mil per year. While population growth has held steady at 2.55 mil per year. This is the hidden acceleration of automation. To grow profits and grow businesses, companies need less and less employees to be successful.
The unemployment rate for new college graduates. Even during the financial crisis of 2009, where we experienced peak near term unemployment, the unemployment rate for college graduates was never over 5%. As long as Universities continue to provide current, in-demand skills, graduates will find work. This also makes intuitive sense, because students, by their definition, have time to learn, whereas, under our current structure, education is not built into our day-to-day jobs. This makes workforce obsolescence a real issue. (See my blog post on Lifetime Learning)
So that sets things up to become generational. It will be the 40-somethings and 50-somethings who will increasingly find it difficult to find new work as the technological skills may pass them by. More and more people in the middle of their careers will be forced into entrepreneurial endeavors and part-time work. Often, jobs won’t become eligible for unemployment benefits, like self-employment and many IRS 1099 filers. These people, unemployed or underemployed are not counted in any of the surveys, essentially they fall out of the work force. This number will increase over the coming years. I actually expect the Bureau of Labor Statistics to look to refine their labor participation measurement to try to capture these displaced workers.
This plays out in a different way. Over the past 40 years we have experienced market meltdowns and crises of various magnitudes. But the financial crisis in 2009 was probably the most poignant. The U6 participation rate jumped from 7.5% to 17.1% as the effects of the crisis enabled companies to actually eliminate people, the PR be-damned. That rise is unprecedented in the post-depression employment era. I believe that future crises will be used the same way as a significant form of layoffs and I expect this kind of jump in the U6 to be more of the norm than an outlier. It has taken the US economy nearly a decade of growth just to return to the mean U6 level of the previous 20 years. You can expect this U6 number to be extremely sensitive to market shocks and I’d expect a general upward trend in U6 over the coming 2 decades. This is where unemployment shows itself. I expect by the 2030’s, unless they change the calculation, U6 unemployment will be consistently around 20%
In the end, over the coming generation, there will be less job. Less full-time work and increasing replacement of human labor with automation, but it will be an insidious creep towards higher functional unemployment. Hopefully it will not be so subtle as to be neglected by the appropriate policy response or even misunderstood by technological unemployment nay-sayers.