Will the continued rise of automation cause widespread job losses?

Evidence seems to indicate that the net effect of automation on employment is actually negligible, although its structural effect on the job market is very real and should be addressed

It is clear that the jobs most likely to fall victim to automation are the low-skilled ones, especially the ones that involve tasks that can be mimicked by a machine 

As labor-saving technologies advance, concerns regarding job security will likely rise in tandem. According to a study by PwC, 37 percent of the 10,000 people surveyed across five countries were worried automation would put jobs at risk. The bulk of available evidence seems to suggest there is little reason to worry, as the net effect on employment will likely be small and may, in fact, result in gains. But dealing with the fallout from a large structural shift in labor will present challenges.

At highest risk of being automated are low-skilled jobs, particularly those with repetitive tasks that can easily be replicated by a machine

The jobs at highest risk of being automated are low-skilled ones, particularly those comprising predictable and repetitive tasks that can easily be replicated by a machine. Meanwhile, occupations requiring more abstract skills – those that are harder to codify and copy, and that cannot easily be mimicked in the absence of advanced artificial intelligence – will face a lower risk of being automated.

“One pattern that becomes clear is that there is more erosion at the low end of the skill distribution as opposed to the top of the distribution,” said Grace Lordan, Associate Professor of Behavioral Science at the London School of Economics.

“We do see some erosion at the top end of the distribution, but at the bottom of the distribution, there is going to be much more hollowing out.” According to a study by McKinsey Global Institute, 30 percent of work tasks across 60 percent of occupations could be automated, while between 400 and 800 million people may need to find new work due to automation by 2030.

A report by PwC suggests machines will encroach further on manufacturing jobs as they become more adept at completing physical tasks that require dexterity. The report estimates the resulting job losses could be as high as 44 percent in many countries.

The sky isn’t falling
despite the high figures in their reports; both McKinsey and PwC maintain the losses will likely be made up for elsewhere in the economy. “[The high estimates of ‘at risk’ jobs] just reflect technological potential, so they tell us how much theoretically could be done by machines,” said Ulrich Zierahn, Senior Researcher at the Centre for European Economic Research. “But the actual employment effect is likely to be very different.”

Factors like the high cost of implementation for firms and labor laws are likely to slow the roll out of automated technologies. Indeed, Zierahn’s research suggests that a task-based approach to predicting the impact of automation, rather than a traditional occupation-based one, could be of greater use.

“Usually researchers say that if the [proportion of automatize tasks] is above 70 percent then you would regard the job [as] at risk,” Zierahn said. “However, being at risk just [refers to] the level of exposure to technological change, whereas the potential employment effect of that exposure is another question.”

Additionally, some research shows that even if automation drives down labor demand in manufacturing, mass redundancies wouldn’t necessarily follow. “We do not find any systematic evidence that manufacturing firms really fire workers,” said Wolfgang Dauth, Assistant Professor of Empirical Regional and International Economics at the University of Würzburg. “What we do see is that they reduce the number of labor market entrants that they would otherwise have hired, so younger people then go into the service sector rather than the manufacturing sector.”

In the past, disruptive events like the Industrial Revolution have often been referenced as instances where naysayers overreacted. Dauth explained: “Throughout economic history there has always been a new wave of automation that made [a] certain kind of work… redundant, but overall we don’t really feel that this time might be different.”

However, the increasing rate of technological advancement could potentially set this era apart from previous ones. “In honesty, my feeling is that it is different,” said Lordan, pointing to the cratering housing market that precipitated the 2008 financial crisis as an example of an unprecedented event that everyone thought impossible. “Things can happen for the first time, and you probably want to encourage policymakers to think about it.”

Smoothing the transition
Net employment effect notwithstanding, public policy will need to reflect the needs of the new job market. Even in a country like Germany, which has a low risk of net job loss and is well equipped to deal with these labor shifts due to its tradition of investing heavily in retraining its workers, there is a gap between those who benefit from automation and those who do not.

“We do see that there are distributional effects of automation,” Dauth said. “[This] means that we will see differences in that higher skilled people mostly benefit from automation, whereas lower skilled people see reduced employment prospects, but on the net there is no overall negative effect of automation.”

According to Zierahn, the structural effects that automation will have are important – specifically, how the shifts between occupations and industries present significant challenges. “We have to support workers in adjusting to the change by getting the right training to get the right skills,” he explained. “It seems that inequality is rising due to technological change because it is particularly the low-skilled workers who are most exposed and who suffer most from the change.”

The economic benefits technology will bring are real, but governments will have to pull off a balancing act between remaining competitive and minimizing the impact machines will have on large portions of their workforce.