Low-wage workers face a double blow from automation, a new Oxford University study has found; they are both more likely to lose their jobs due to new technologies and less likely to have the skills required to switch to newly created jobs.

The COVID crisis has accelerated workplace automation and, while technological change is likely to create as well as destroy jobs, today’s study shows automation can increase unemployment even if the total number of jobs available in the economy stays the same. This is due to mismatches between the skills of unemployed workers and those required by job vacancies: new jobs in data science are of little help to an unemployed taxi driver, for example.

The study, published in the Journal of the Royal Society Interface, is the first to show how differences in workers’ opportunities to switch occupations is key to determining who is most vulnerable to automation-related unemployment.

The researchers found that low wage workers are likely to face a doubly difficult time. Because they are more at risk of their roles being automated in the first place, displaced low wage workers then enter a job market where many of the alternative roles they might have transitioned into have also been automated away.

For example, if food preparation workers lose their jobs due to automation, many of them are likely to look for work as cooks and cashiers, but these roles are themselves at high risk of automation. In contrast, higher wage workers displaced by automation – say, statistical technicians – will enter a job market with far more options still open to them.

The study also revealed that unemployment risks are not limited to those directly displaced by automation. Childcare workers are at low risk of automation, for example, but are likely to face a much tougher job market due to other displaced workers trying to enter their industry.

By revealing which workers are most at risk of longer spells of unemployment, the researchers’ model can improve the targeting and effectiveness of worker support and retraining schemes.

Economist Maria del Rio-Chanona says, ‘If we carry on as we are, those with least to fall back on will bear the brunt of automation-related unemployment. Low wage workers face a double hit: more likely to be displaced by technology, they are also more likely to see their pool of alternative options drain away.

‘But there is nothing inevitable about this outcome. Our model can help policymakers design targeted retraining and other support to properly equip low wage workers for the changing economy. With COVID fast-forwarding automation in many industries, the need for such support is crucial.’

Rita Maria del Rio Chanona
‘If we carry on as we are, those with least to fall back on will bear the brunt of automation-related unemployment. Low wage workers face a double hit: more likely to be displaced by technology, they are also more likely to see their pool of alternative options drain away.' R. Maria del Rio-Chanona, INET Oxford economist

Professor J. Doyne Farmer, who co-authored the study with del Rio-Chanona, says, ‘We know we need policies to support fossil-fuel workers transition to new roles in order to build a green economy; building an increasingly automated economy is no different. We need policies to support those workers most at risk so that they can prepare and adapt.’

‘We’ve seen economic and social divisions grow over the past decades and without proper action, automation could cause further deep distress. But with the right policy frameworks in place, including well-targeted support for low wage workers, it could power a better economy for all.’

J. Doyne Farmer
‘Without proper action, automation could cause further deep distress. But with the right policy frameworks in place, including well-targeted support for low wage workers, it could power a better economy for all.’ Professor J. Doyne Farmer, INET Oxford's Director of Complexity Economics

Notes to editors:

For further information please contact Jessica Kaplan, Communications Manager at the Institute for New Economic Thinking at the Oxford Martin School, University of Oxford (INET Oxford): Jessica.Kaplan@inet.ox.ac.uk / +44 7956 641 829.

  • The study was authored by R. Maria del Rio-Chanona, Dr Penny Mealy, Dr Mariano Beguerisse-Díaz, Dr François Lafond, and Professor J. Doyne Farmer, and published in the latest issue of the Journal of the Royal Society Interface.
  • R. Maria del Rio-Chanona is an INET Oxford economist who specialises in developing a data-driven network model of the labour market to understand the impact of automation on employment. Dr Penny Mealy is a Research Fellow at SoDA Labroratories, Monash Business School, Monash University, and an Associate at INET Oxford. Dr Mariano Beguerisse-Díaz is a Senior Research Fellow at Oxford-Emirates Data Science Lab and the Mathematical Institute, University of Oxford. Dr François Lafond is Deputy Director of the Complexity Economics programme at INET Oxford, a Senior Research Officer at the Oxford Martin School Programme on Technological and Economic Change and the Smith School for Enterprise and the Environment, and an associate member of Nuffield college. Professor J. Doyne Farmer is Director of the Complexity Economics programme at INET Oxford, Baillie Gifford Professor in the Mathematical Institute at the University of Oxford, and an External Professor at the Santa Fe Institute.
  • Economists Jose Maria Barrero, Nicholas Bloom & Steven J. Davis estimate that 42% of U.S. jobs lost to the pandemic will not return. A recent World Economic Forum report found that automation had been hugely accelerated by the pandemic and predicted that by 2025 it would disrupt 85 million jobs globally. In a recent report of the Commission on Workers and Technology, the UK think tank the Fabian Society also found that the COVID crisis had accelerated workplace automation.
  • Study authors built a complex network model of occupation transition and unemployment, drawing on a large pool of data that included: automation probability estimates for each occupation; census data that tracked how people transitioned between occupations in the U.S.A. between 2010 and 2017; and the accompanying employment and unemployment rates for this period.