By Brian Santo, contributing writer
Far fewer jobs are at high risk of being eliminated than the truly alarming estimate from a 2013 paper that calculated that nearly half of all U.S. jobs are susceptible, according to a new study. But even at the significantly lower new estimate, automation is still highly likely to wipe out a lot of jobs.
The 2013 study projected that 47% of all jobs in the U.S. alone are highly likely to be threatened by automation. A new report, “Automation, skills use and training,” widens the scope to 32 industrialized countries and pegs the number at 14% on average, differing by a few points (plus or minus) from one country to the next.
The new report considers more data and provides a more thorough analysis. But precision in the calculation of the risk creates a distinction that might be considered less comforting than the lower number should. Yes, only 14% of jobs are highly likely to be automated, defined as having a probability of over 70%. Another 32% of all jobs, however, have “only” a 50% to 70% probability of being affected by some automation.
Et voila! We’re back at 46%…
The new report was published by the Organisation for Economic Co-operation and Development. The OECD is a policy organization concerned with economic and social well-being. Its members include the North American countries, many European countries, and a few each from Asia and South America. The new study includes data from 32 of 35 members.
The 14% figure translates into tens of millions of jobs around the world that are highly likely to be lost. The 32% figure translates into hundreds of millions more at significant risk.
This is a problem for which solutions are needed, but, the new study notes, there’s been little investigation into what those solutions might be.
It’s a truism that automation threatens jobs. One question that economists would like a precise answer to is which jobs.
The old distinction between blue-collar and white collar is a misleading oversimplification. To get a handle on the employment trends and the effects of automation, researchers break down jobs into the types of tasks that they’re required to accomplish:
- Manual routine
- Manual non-routine
- Cognitive routine
- Cognitive non-routine
- Interactive non-routine
Automation threatens any job where the tasks are repetitive, or routine. The jobs that prove more immune to automation are the ones that require creativity, flexibility, problem-solving, or people skills, which is to say, jobs that are non-routine.
Incidentally, the analysis reveals that younger workers are far more susceptible to the threat of automation than any other age group and that women are somewhat more susceptible to the threat of automation.
What is changing, and changing very rapidly, is what qualifies as “routine.”
The authors of the new study referred to a paper from 2003 that enumerated a set of tasks likely to remain immune to automation. All of the examples from that paper have either been automated or are on the verge of being automated.
Recent technological innovations, the new study noted, expand the list of potentially automatable jobs to a far wider range of occupations, including, but not limited to, drivers, translators, tax analysts, medical diagnostics, legal assistants, security guards, law enforcement officers, teachers, HR workers, financial analysts, and even software programmers.
“In fact, the scope of what digital technologies can do expanded so much that more recent work on job automation found it easier to ask ‘what is [it] that computers cannot do’ than to keep asking ‘what is [it] that computers can do,’” says the new report.
In short, being a lawyer, an engineer, or a biologist does not necessarily inoculate you from the dangers of automation. You can have a higher degree, but if you aren’t engaged in something non-routine or interactive, your job is at risk.
Automation is consequently leading to the “hollowing out of the middle-paid, middle-skilled jobs in developed countries.” Eroding the middle class leaves high-paid high-skill non-routine jobs and low-paying low-skill jobs. The study calls that “job polarization.” The phrase “income inequality” is not used in the new report.
Future studies sponsored by the OECD will focus on the effects of automation on wages.
The latest study reiterates the long-held assumption that new technologies lead to the creation of new jobs.
The study glosses over a notion that an increasing number of people are contemplating — that as more and more tasks continue to be rendered routine by advanced technology, automation will eliminate more jobs than technology creates.
Granted, this worry is largely conjecture based on limited evidence, but the authors of the new report apparently have no data to address this line of thought directly either. They cite other studies, most of which assume that technology still creates jobs, all of them theorizing that any evidence to the contrary can be explained by other factors. Lacking clear evidence, the authors are obliged to punt on that aspect of the issue.
What they do address directly is the need for solutions to job dislocation and the disturbing lack thereof.
The authors state, “Given the gravity of the technological transformation we are undergoing, there is astonishingly little research effort in understanding the subsequent response through skill adjustment.”
The solution reflexively offered for the problem is retraining.
The positive example that the study cites is banks. After banks introduced automated teller machines (ATMs), tellers no longer needed to be able to handle cash transactions. Banks retrained tellers to help customers with more sophisticated financial services.
But that example raises an interesting distinction. Employers have some incentive to retrain the employees when only a portion of some occupation’s tasks are automated. If too many of an occupation’s tasks are automated, employers have little incentive to sponsor retraining.
Government retraining programs, meanwhile, are not always available, nor are private-sector retraining programs. When such programs exist, they can be inadequate. The authors cite one positive example, Germany, where retraining programs are more widely available and are successful, albeit with some qualifications.
Of course, workers can invest in their own retraining — presumably. The jobs most likely to be automatable are also linked to lower wages, which means that many workers who have lost their jobs might not be able to afford retraining.
Also, “the risk of automation declines with educational attainment, suggesting that workers who sort themselves in more automatable jobs tend to invest less in human capital to start with,” noted the study. “As a result, they might also be less motivated to learn new things.”
The study concludes that those most susceptible to losing their jobs to automation receive very little retraining, whether from their own employers, government agencies, or private-sector organizations.
Not addressed in the study are some of the assumptions that economists tend to make about employment that aren’t necessarily true.
One assumption is that if a job exists, any worker with the qualifications should be able to fulfill it. Not taken into account is the unemployment environment. There are hundreds, if not thousands, of people vying for many jobs, and no matter what kind of retraining a job-seeker might have, many of those jobs demand years of previous practical experience.
Another assumption is that all workers are fully mobile. The assumption ignores any number of preferences and obligations that make it hard, if not impossible, for some individuals to be able to consider jobs on the other side of a state, let alone the other side of a country.
Another thing not considered in the recent report is the timing of all of this. “When?” is a question that will be investigated in subsequent OECD-sponsored studies.
Advertisement
