The claim that nearly half of all US jobs are at risk of being automated has been repeated many times. But is it that straightforward? Gavin Kelly offers some facts that moderate the claim. He concludes that although the impact of evolving technology on jobs is currently impossible to predict, there are other, very real issues we should fret about: decreasing productivity levels, stagnant wages, and poor skills.
We are said to live in an increasingly ‘post-fact’ world but it remains the case that an eye-catching statistic can gain its own momentum and shape public debate regardless of its providence or even its author’s original intent.
So it is on the question of the rise of the robots and the future of work. It has become a mainstay of public discourse that the world of work is on the cusp of being radically transformed by a wave of job-destroying robotic technology. The killer fact that distils this threat is the oft-repeated finding that 47% of jobs in the US economy are at risk of being automated over the next decade or two. Related studies find that across the OECD 57% of roles are at risk, more still in emerging economies. Hug your job tight.
Politicians, business leaders, and some of our most respected economic institutions have in different ways repeated or replicated these gloomy-doomy conclusions. Our media laps up these findings having a near insatiable appetite for features on how humans are destined for the scrap-heap as they lose out in the race against the machine. Future-gazing pieces speculate about how we will all find meaning when the work runs out. Seminars on designing a welfare state fit for a ‘post-work world’ are all the rage.
Yet seasoned observers of the evolution of jobs markets tend to raise an eyebrow about grand claims about the disappearance of work from large swathes of the economy. And they are more sceptical still when it comes to what sound like spuriously precise estimates of how many jobs are likely to be automated some decades in the future.
That scepticism has now been given concrete form by an important, if predictably underreported, study by the OECD. It replicates much of the original (and, it should be said, thought-provoking) report by Carl Frey and Michael Osborne which produced the ‘47% of jobs at risk’ finding — but it moderates a few vital assumptions.
Rather than assessing the risk that a whole occupation will be replaced by smart-machines the OECD authors instead take a more granular look at the underlying bundle of tasks that make up different jobs, considering how automatable each of them are. The point being that even within occupations that appear ripe for technological upheaval — say, bookkeeping or retail sales — there is much non-routine work that will prove very hard to mechanise in the near future.
This more forensic approach results in a collapse in the estimate of the proportion of US jobs at risk from 47% to just 9% (9% is also the OECD average). Still significant, but a massive down-grade. What’s more, the authors suggest their figures are likely to be an over-estimate of actual job losses. In part because many workplaces are sluggish in utilising new labour-saving technologies and workers performing tasks under threat often find ways of adapting their role so as to complement smart machines. But also because there will be some offsetting employment gains: new jobs get created due to the demand for innovative technology, and tech-induced price reductions mean consumers have more disposable income available to spend on other things.
The correct response to all this isn’t to release a sigh of relief as we switch from believing that only one in ten jobs are at risk rather than one in two. Other studies will surely come along with new estimates. No, the lesson is that we should be highly circumspect about all such claims — especially those with shock and awe headlines — and interrogate the underlying assumptions closely.
None of this is to say that the onset of intelligent machines won’t disrupt all manner of working patterns, including in previously protected white-collar occupations. It would be very surprising if they didn’t: every phase of capitalist development is associated with new technology that triggers the rise and fall of different roles.
The question is whether this time will be so very different? Will this era of shockingly low productivity growth be a precursor to great technological advances that overshadow everything we’ve ever seen before in terms of their jobs-impact? Perhaps. None of us can know. But, as the OECD shows, it’s hard to arrive at that sort of conclusion based on the available evidence.
For now the UK remains a jobs-rich nation. Employment is at a record high and worklessness is less of an issue than it has been for a generation. Yet there are plenty of issues right in front our eyes to fret about: low investment, flat-lining productivity, stagnant wages, poor skills to name a few. Of course, none of these catch the eye like a story on how the march of the machines is destined to make humans obsolete. But, for now at least, they are more real.
Note: this was originally published on Gavin’s personal blog and is reposted here with thanks.