Robots not taking jobs: economist busts automation myth

Article3 min11 November 2016By David Marin-Guzman - originally published by The Australian Financial Review on 11 November 2016.

If you were somehow looking forward to a jobless future, what with robots taking over the workplace, you may be waiting a long, long time.

Scientists, businesses and media pundits can't stop talking about the "unprecedented" pace of technological change and the risk that up to 40 per cent of jobs could be automated.

This week Elon Musk, himself at the cutting edge of developing driverless cars, added to the hype by predicting robots would replace so many jobs countries would be forced to implement a universal basic income.

For University of Melbourne professor Jeff Borland, the rise of the robots makes for "a sexy story". The problem is there is absolutely no evidence to support it.

The widely respected labour economist and self-confessed contrarian has caused a stir with his study Are Robots Taking Our Jobs?, which revealed automation's effect on the workplace was "vastly overstated".

The paper prompted inquiries from Department of Employment analysts and senior union leaders skeptical of the robot hype and has highlighted a long-standing divide between technologists and economists.

Borland, along with fellow Melbourne uni economist, Michael Coelli, rigorously mapped out the shifts in demand over time for jobs with routine skills – the kind of work technology is replacing.

They discovered that the trend away from routinised jobs today was no different than was going on back in the 1970s and the 1980s, well before the introduction of computers into the workplace in the 1990s.

Moreover, the pace of job turnovers was no greater today than it was in the late 1960s. Borland theorises that, just as has been the case with waves of automation since the industrial revolution, technology is creating jobs as fast as it is eradicating them.

New digital technologies have increased demand for software designers, programmers and managers, while lower costs of production from automation have resulted in higher real incomes that in turn lead to more demand and more jobs.

Professor Jeff Borland - University of Melbourne

This time it's different

But there is one difference. This time, big data and advances in computerisation mean technology is no longer replacing low-skilled blue collar jobs. It's threatening cognitive-skilled white collar work – the bookkeeper, the lawyer, the doctor, the market analyst.

Some analysts argue these technological developments are happening so fast they could eradicate more jobs than they create.

Robotics expert and Queensland University of Technology professor Peter Corke is a case in point. He recalls how he used to make "sanguine statements" in lectures about how technology created jobs as much as it eradicated them. "I don't feel comfortable saying that now. I don't convince myself anymore."

But for Borland, technologists like Corke are confusing changes in the automation of certain types of jobs with changes in the total quantity of work.

No reduction in jobs

 "The rate of jobs being routinised may be changing, but that doesn't mean that technology in aggregate is having a larger disruptive effect on the labour market than it did previously."

He says the total amount of work available has actually remained stable since the mid-1960s, with the main variations due to recessions and changes in the business cycle.

If computers were causing major disruptions to work, Borland said you might expect to see jobs lasting for shorter periods. In fact, the opposite was happening.

Data from 1982 to 2013 showed there had been no increase in females working in jobs for less than a year. In fact, there had been a notable increase in females staying in jobs for 10 years or more.

Job tenure data for males showed similar trends, with no increase in the proportion of short-duration jobs.

Even the idea that technology's routinising of cognitive skills is a new development is open to question, Borland says.

"Automation of mid-skilled jobs may be something that's a little bit different, but it's not totally different.

"The weavers during the industrial revolution were seen as skilled workers, trained in the craft, earning higher than agricultural workers at the time when the cotton spinner replaced them."

Don't believe the hype

It's not the first time pundits have sought to downplay the historical tendency of automation to create jobs and predicted their age's wave of innovation would be the one to wipe out work.

In 1930, the economist John Maynard Keynes famously predicted the 19th century had unleased such technological change that "our grandchildren" would work only three hours a day.

A 1983 report to the National Science Foundation entitled The Next Generation of Robots predicted "today's robotic developments … will progressively reduce the labour used in industrial production, ultimately to something rather close to zero".

Borland notes humans have a tendency to exaggerate the current state of change for their own purposes. There's a natural bias to believe we live in significant times.

The visible rise of the digital economy over the past 20 years has no doubt contributed to that perception.

"You can't deny Amazon is changing the way people buy books, you can't deny Airbnb is changing the way we live – you can't deny new technologies are changing the way the economy operates.

"Clearly that's happening. But what I say is it's always been happening.

"In the start of the 20th century, when Ford and the big new US companies created huge mass production lines, that had a huge effect on the way goods were produced and economic activity happened.

"When you go back to the industrial revolution, when you introduced new ways of spinning cotton and new ways of weaving, that had a huge effect on the way economic activity was done.

"You can accept the technology is changing the way economic activity happens, but that does not follow that we're heading to a workless future."

'World without work' unlikely

For Borland, predictions of a jobless future are ultimately more utopian than real.

"There may be other reasons for a guaranteed minimum income but one reason is not because there is going to be a future where there is no work," he said.

"I think it will take a lot of reorientation for work to disappear. The thing is we forget people want more stuff ... We'd have to lose our inquisitiveness and drive to consume and have more."

That was not to say automation didn't raise policy issues.

The maritime industry is one of the main industries most affected by automation in the last few years.

Steve Lee was one of 160 workers who lost their jobs last year through Patrick's push to fully automate its Port Botany terminal.

At the time, Lee had worked at the port for more than two decades and was ranked the second-best-performing crane operator on site.

He never expected a machine would replace him, leaving him to cope with a $1.26 million mortgage he'd signed up to the year before.

After four months, Lee got rehired at the port as a contractor. But he said his experience shows automation can happen to anyone.

"It was a big reality check for me after 26 years to be told you don't have a job because the robots are coming in," he said.

"Especially when you don't give the company any reason to dismiss you."

Borland says policymakers still have to deal with adjustment problems around automation, including assisting those adversely affected and ensuring a flexible institutional environment that can take advantage of new technologies.

But that has always been the issue with technology, he says.

Outside of the data, the problem of robots taking over our working futures is merely speculation.

"And I'd rather speculate based on the way history's been for the last 250 years than any other way."

The article presents the views of the author only and not those of Dexus.

World without work is unlikely

You can accept the technology is changing the way economic activity happens, but that does not follow that we're heading to a workless future

Read on for more insights

Prism Mailing List
Back to top