Carl Frey: Worker-replacing and labour-augmenting technological change

Profile: Dr Carl Benedikt Frey, Oxford Martin Citi Fellow, Oxford Martin School, University of Oxford, England, UK.

Dr Carl Frey highlights three key points to understanding the impact of automation on work - and where solutions may lie in transitioning.

"How labour will fare from future automation depends on whether it will be predominantly labour replacing or enabling. How workers fare depends on the race between task displacement and new task creation, and how easily workers can transition into emerging jobs."

Economic growth is in large part the result of labour-saving technologies that allow us to produce more with fewer workers. The wealth of nations can be seen as a steady flow of labour-saving inventions over the centuries. The share of the wealth being captured by labour, however, depends on the characteristics of those inventions. A helpful model has recently been put forward by Daron Acemoglu and Pascual Restrepo, conceptualizing technical progress as either enabling or labour displacing. The notion of machines being capable of taking over work from humans in some tasks is important, because it means that technology can reduce the labour demand, wages and employment, unless it is counterbalanced by other economic forces. Even though growing productivity still raises total income, offsetting the displacement effect in part, it may not fully counterbalance the negative effects of technological displacement. The creation of new tasks in which labour holds the comparative advantage, as Acemoglu and Restrepo point out, is essential raise the demand for labour and the labour share of income. How workers fare, in other words, depends on the race between task displacement and new task creation, and how easily workers can transition into emerging jobs.

How labour will fare from future automation depends on whether it will be predominantly labour replacing or enabling. In 2013, Michael Osborne and I, published a paper entitled “The Future of Employment: How Susceptible Are Jobs to Computerisation?”, estimating that 47% of U.S. jobs are exposed to automation for the next decades. The intuition underpinning our analysis was that the future of work can be inferred by observing what computers do. As is well-known, back in 2003, David Autor, Frank Levy, and Richard Murnane documented that jobs which were intensive in routine tasks began to disappear around 1980. What’s less well known is that their findings were entirely predictable. In his 1960 essay, Herbert Simon predicted the demise of routine employment, suggesting that computers hold the comparative advantage in “routine” rule-based activities. Similarly, in several case studies conducted in the 1960s, the U.S. Bureau of Labor Statistics found that:

“A little over 80% of the employees affected by the change were in routine jobs involving posting, checking, and maintaining records; filing; computing; or tabulating, keypunch, and related machine operations.”

If the next generation of AI technologies will turn out as brilliant as some of us think, there are reasons to be optimistic. Brilliant technologies are preferable to mediocre once even if they replace a lot of workers. Larger productivity gains mean growing incomes that will better offset any displacement. But realizing those productivity gains as well as the share of income accruing to labour depends on workers being able to shift into new jobs and tasks. It is therefore of great concern that labour markets have become less fluid in recent decades. Reduced labour mobility may be a contributing factor to lower wages, increased inequality and lower productivity growth. Abolishing barriers to job switching should thus be a policy priority. The structure of labour markets is based on rules and institutions created by governments, and some of these may require some oversight. Labour market institutions like occupational licenses and non-compete clauses, for example, have seemingly served to reduce occupational mobility.

Second, like reallocation, geographic relocation is a critical feature of a dynamic labour market. Migration provides a mechanism by which regions adjust to trade and technology shocks. Throughout history, migration from places where employment has contracted to areas with more plentiful job opportunities and higher productivity has served to equalize incomes across space. Computer technology has played a pivotal role in shaping the geography of work. My own research with Thor Berger suggests that the Computer Revolution shifted economic activity between places as new jobs began to cluster in cities with a greater abundance of college-educated workers. At the same time, however, jobs are being automated away in once prosperous manufacturing towns.

Why then are people not moving to where new jobs are being created? One reason is seemingly financial. Even if other places provide more and better job opportunities, moving is an investment which requires liquidity upfront. Because relocation is an investment, where a worker spends money upfront for the opportunity to get a job, and many unskilled workers face financial constraints, there is a case for subsidizing their relocation.

Third, if workers cannot acquire new skills to take advantage of new technologies, productivity and wages will suffer. Looking forward, my own research with Michael Osborne suggests that workers in jobs that typically require a college degree are less susceptible to automation. As discussed, workers with a college degree have been better able to adjust to automation. But most citizens will not be able to go back to college later in life. The hard part will be to reskill those who lose their jobs late in their careers. Overall, the evidence shows that the effectiveness of reskilling initiatives are highly context and worker specific. Thus, education and training may not be appropriate in response in every case for every worker. Instead, for displaced workers for whom reskilling is not a realistic or appealing option, compensation is a more sensible policy option. For situations when workers suffer financial downgrading, there are already a variety of public wage subsidies in place which could be expanded.

For more: Automation and the future of work – understanding the numbers


Pernille Gronkjaer: The Human Quest Unfolds


Andrew Shreve: Biomaterials Science: technology development and education