How has technology disrupted the job market?
Jobs are usually classified as white collar/blue collar; routine or non-routine. Routine means codifiable; we can write precise rules that describe what the human being does, and as long as the task doesn’t require extreme agility, we can tell a machine how to do this job, and the machine can do it. This takes place across both the white collar and blue collar job spectrum, so for example, we have computer systems that can keep track of the books and such, and robots that can do an increasing array of physical tasks.
So the short answer is that digital technologies (among their many impacts) are reducing routine white and blue collar jobs. Looking to the future, the range of tasks that machines can do looks to be expanding rapidly. From and economic point of view, this process forces a restructuring of the economy in which people shift to non-routine jobs.
One result is that today we are seeing a phenomenon called job polarization. This includes a decline in the middle income range, and an increase in either the upper 20 percent or the lower 20 per cent of this scale. It’s a well-documented, 20- to 30-year trend, coupled with rising inequality and a downward trend in routine jobs. Routine jobs are actually disappearing—and the inflection point is in the year 2000, when the routine job decline accelerated.
When I went to work in academia, each department had almost as many assistants or non-teaching staff as we did faculty. Now, we have two members of non-teaching staff. We don’t need the rest. We all type, we all use the internet. All of those lost jobs are middle income jobs and on the corporate or business side, you’ll find functions like the general ledgers were done by hand. Those jobs are middle-income jobs and those are gone too.
What happens when these jobs disappear?
We don’t know because we can’t track the job losses and link them directly with individuals, but it looks like we took away the routine jobs, so that these workers moved into non-routine areas. Some of them may have ended up in the upper end of the distribution, others ended up in the lower end of the distribution. But what we can say is that most of that job loss is attributed to technology.
If you go back to the first Industrial Revolution, the British one, where you had steam engines and where machines began to do manual work, you will find that that event also took out jobs in the lower end of the income spectrum. Still and all, people, after a difficult period, were able to make changes to their human capital, and they made it so that cities and began to work. The process took decades. But this round of technological disruption, hit mainly the middle income group, not the bottom, and that’s part of what creates the challenge for us, in socio-political and economic terms, because middle classes can act as a feeder in an upwardly mobile society.
Has globalization had something to do with this?
Globalization has definitely had an effect on the structure of our economies and the nature of work. Sandile Hlatshwayo and I carried out a study* in the U.S. which took all the sectors in the U.S. economy and divided them up according to goods and services that traded internationally, and another set that didn’t. The study found that the goods and services in the tradable part of the economy generated no net additional employment, even as the American economy grew by 27 million net new jobs. None of these jobs were generated in the tradable side; all of the job growth was generated in industries that were not subject to international competition. Manufacturing jobs disappeared. So it looks as though a combination of globalization and labor-saving (automation) pushed workers out of middle-class jobs and pushed them one way or the other on the income spectrum. And we are now left with income polarization and unhappy people.
Where do you see the end of this?
Technologists say we’re just at the beginning of the capacities of machines and digital technologies to do things- think of self-driving cars. Certainly recent breakthroughs in what machines can do are stunning.
Previously we couldn’t explain to machines how humans do many things: like how we listen to a language and translate, or how we can recognize a chair, especially since there are infinite versions of a chair. These were not codifiable. But now, machine learning works differently – more like the way a child learns. By pattern recognition, imitation, examples. High speed access to huge digital data bases accelerate the process
There are questions about what machines are capable of doing and some of the talk is wondering what the limit to this is and how to deal with the people that have been displaced by all this. What if it just keeps going and what if we have just seen the start?
One final point. Once a digital technology passes a cost threshold and becomes competitive, there is no return: the costs just keep going down. With that dynamic, at some point, labor intensive jobs may disappear as well. And there is a reasonable chance that these technological changes and disruptions will not only define this generation, but also the next.
*The Evolving Structure of the American Economy and the Employment Challenge, which is referenced by Dr Spence in this interview, was first published by the Council on Foreign Relations.