Apocalyptic visions of robots stealing workers’ jobs are not only misguided but have diverted attention from more significant trends.
There has been a lively debate over the last decade on the ‘future of work’—the implications for employment of recent technical change. With few exceptions, it has had a predominantly negative tone. Recent waves of change—computerisation, robotisation and, more recently, artificial intelligence—are often portrayed as impersonal, pervasive, ineluctable forces, which can bring enormous benefits (mostly for consumers) but also major challenges (mostly for workers).
Projected employment losses over the near future are often counted in the millions and, the argument goes, the outlook is particularly bleak for low- and medium-skilled workers, who will be easily replaced by machines because of a lack of complementary skills. Potential policy responses are generally framed in a defensive, almost fatalistic way—at most to try to mitigate these inescapable tendencies. Most frequently mentioned are opportunities for re- and up-skilling of the forthcoming masses of displaced workers or—presuming there will anyway be too few jobs for all—some form of unconditional income support.
Evidence thin
But the urgency and scale of this debate contrasts with the thinness of the evidence supporting these prognoses. Many of the numbers fuelling it do not come from observation of employment trends but from forecasts, building on expert assessments and/or econometric modelling, which introduce non-linearity in the projections on the premise everything is going to change soon—which, of course, results in the expectation of sweeping employment losses.
Comparing past projections with actual numbers however illustrates how far off these can be. A recent study by the Organisation for Economic Co-operation and Development found net employment growth between 2012 and 2019 in the occupations and countries it had considered at highest risk from automation in 2012—albeit at a slower pace than the rest.
Other widely-quoted evidence concerns the alleged polarising effect on employment of recent technological change. But the actual shifts in employment in question are gradual, slow and diverse and the association with technical change is circumstantial and increasingly contested—including when a gender lens is applied.
Most recently, evidence from the empirical literature on the impact of robotisation on employment has been hotly debated, with some influential papers claiming a robust causal effect of robots on jobs. But even then the effect is so small it is hard to determine whether it is positive or negative. Certainly it is not big enough to justify the coverage of these findings in the popular press and the associated public fears: according to a 2017 Eurobarometer survey, 72 per cent of Europeans believe that ‘robots and artificial intelligence steal people’s jobs’.
Different light
Indeed, the discussion about the employment effects of robots epitomises the hyperbole which has characterised the future-of-work debate. In a recent paper, we present some basic descriptive evidence on industrial robots in Europe, which on its own puts the whole discussion in a rather different, and considerably less dramatic, light.
First, the vast majority of robots in use today are deployed in manufacturing. Elsewhere, robots have very few economic uses—mostly domestic appliances (robotic vacuum cleaners) and toys. Considering that manufacturing only employs around 15 per cent of all workers in advanced economies, this already considerably qualifies robots’ potential impact.
Moreover, within manufacturing, three-quarters of robots are deployed in just three sub-sectors: automobile, rubber and plastic, and metal products. This is compounded by geographical concentration, especially in Germany (see bar chart). Thus in 2017 27 per cent of the entire stock of robots in Europe were in the German vehicle industry, which accounts for less than 1 per cent of all European employment.
Distribution of industrial robots by sector and country (2016)
Secondly, current industrial robots essentially perform physical tasks that involve the moving and precise manipulation of objects within standardised industrial processes. They are not at all anthropomorphic—if anything, most look like mechanical arms, ending in an effector which may or may not resemble a human hand and which typically performs the precise moving or manipulation. Although they generally have axes that enable them to move in several directions, they remain within a predefined and rather limited space. And even though they can be reprogrammed to change the specific tasks they do, mostly they remain physically constrained to perform a very particular application.
Essentially, these robots are the most recent (and advanced) iteration of automation technologies used in manufacturing for many decades. They are marginally better because they are more flexible and precise but they do not herald the disruption or discontinuity often depicted. In European manufacturing they are more likely to displace other (less sophisticated) robots than workers.
Even in manufacturing, the amount of labour that can be replaced by robots is already quite small, partly because of previous waves of automation. Most human labour in advanced economies is now dedicated to tasks that involve social interaction, processing of complex information or problem-solving. The tasks that robots can do—repetitive manual tasks involving strength and/or dexterity—are already rather marginal, even within manufacturing. So the potential for labour replacement is limited a fortiori.
Misleading and damaging
How is it possible then that the idea of ‘robots stealing our jobs’ has become so prominent in public debates on the future of work? The idea of technology as an inexorable force which threatens our way of life is very appealing to public imagination and a recurrent theme in the history of socio-economic ideas. It seems to emerge in periods when an economic crisis with a negative impact on employment coincides with visible diffusion and acceleration of technical change. There was indeed such a coincidence in the 2010s, with the aftermath of the financial crisis and the consolidation of the digital revolution.
The narrative is however misleading and can be damaging to public debate. Exaggerated ideas of impending ‘robocalypse’ have likely exacerbated economic anxiety and may have skewed the debate towards inappropriate policy responses. Automation is often blamed for troubling developments in European labour markets, such as rising wage inequality or the polarisation of job opportunities. Yet this distracts public attention from more proximate causes: labour-market deregulation, the weakening of collective bargaining, falling public spending, the decline of publicly-owned capital (such as social housing), offshoring of manufacturing or tax avoidance by the super-rich.
We are not denying that robots, or more generally recent automation technologies, have had any impact on employment in Europe. But this is not of the disruptive or accelerated nature often implied—rather, a gradual and small effect following prior secular trends in manufacturing.
Nor do we preclude the possibility that other technologies associated with the digital revolution have an important effect on employment. In particular, the growing digitisation of services, strongly accelerated by the need for social distancing during the pandemic, may lead to employment losses in the medium term. But these, again, should not be blamed on automation: in some cases (as with online banking or self-service checkouts), the customer is increasingly providing the labour; in others (as in retail), the shift online implies massive market concentration and employment losses are the result of economies of scale, increased logistic efficiency and perhaps deteriorating working conditions.
Finally, while the digital revolution has had disruptive effects on labour markets, these are not primarily employment losses. The digitalisation of economic activity and platformisation of work—both greatly accelerated by the pandemic—are, among other effects, transforming work organisation, intensifying market concentration and eroding the effectiveness of labour regulation and collective-bargaining institutions. A shift in the debate on the future of work towards these trends, likely more consequential, would make the discussion more productive.
The views expressed in this column are those of the authors and cannot be taken as representing the official position of the European Commission.
Enrique Fernández-Macías is a Researcher working in the Joint Research Centre (JRC) of the European Commission in Sevilla. He holds a PhD in Economic Sociology by the University of Salamanca, and his main research interests are job quality, technology & occupational change and the division of labour. Before joining the JRC, he worked in the University of Salamanca and Eurofound (Dublin).
David Klenert is an economist in the European Commission, having been a postdoctoral researcher at the Mercator Research Institute on Global Commons and Climate Change.
José-Ignacio Antón is an associate professor at the University of Salamanca, principally interested in labour and health economics.
