The coronavirus pandemic has devastated people’s livelihoods across the world, but many are clinging to the hope that the economy will recover. Two new studies suggest that even in the event of a full recovery, permanent shifts toward automation could bring about significant social unrest.
Alex Chernoff from the Bank of Canada and Casey Warman at Dalhousie University in Canada published a paper examining the impact of COVID-19 on labor-saving technology, alerting scholars to job losses that are potentially on the horizon. The authors classify jobs according to two criteria.
The first criterion is a job’s susceptibility to automation. The jobs that scored highly on this scale can be codified in a straightforward manner, especially with the assistance of the advanced artificial intelligence technologies that have been recently developed. At present, jobs such as reservation ticket agents, telephone operators, air traffic controllers and payroll clerks are on the verge of disappearing as humans are replaced by computers. In contrast, manicurists and upholsterers can feel relatively secure about their future job prospects.
The second criterion is a job’s viral infection risk, which is based on the level of physical proximity, face-to-face discussions, and exposure to disease or infections. Unsurprisingly, dentists, dental hygienists, and respiratory therapists score highly here, whereas writers, logging workers, and satellite dish installers represented jobs with a very low infection risk.
Chernoff and Warman then make the logical argument that jobs that score highly on both dimensions are the ones that are most likely to be lost to automation over the coming months, and permanently so, as businesses look to make their operations both efficient and pandemic-proof.
The jobs that fall into this unfortunate quadrant in the automation/infection-risk matrix included customer service representatives, nurses, medical assistants, correctional officers, and pharmacy technicians. Jobs such as elementary school teachers and accountants were medium risk, while the low-risk jobs included janitors and truck drivers. Worryingly for policymakers, looking to close the female-male wage gap, the authors also found that women were significantly over-represented in the jobs that had the highest risk of being lost to COVID-19 induced automation.
While the thought of an AI nurse or prison warden may seem far-fetched, the transition is not expected to be one robot for one human. Instead, consider that medical professionals and correctional officers perform a wide range of tasks, some of which can be assigned to robots overseen by humans: taking blood pressure, checking that inmates are in their cells, delivering medicine, and searching cells for illegal contraband can in principle be performed on a large scale by one highly-trained individual, supervising a fleet of high-intelligence robots. The more complex tasks, such as comforting a patient or maintaining discipline in the prison, will still be performed by humans, but the total number of humans needed will fall.
University of Zurich professors Bruno Caprettini and Hans-Joachim Voth in Switzerland recently published a paper indicating that this type of “technological unemployment” can generate significant amounts of social unrest. They analyzed detailed data relating to series of disturbances that occurred in Britain in the 1830s, known as the “Captain Swing” riots. From 1830 to 1832, over 2,000 riots broke out in 21 different counties, and most of the rioters were rural workers led by craftsmen and artisan. The government eventually responded with a hard line, ordering soldiers and militiamen to attack rioters.
The damage caused by the riots – which included 492 agricultural machines – led to a parliamentary inquiry, and it inspired a large amount of academic interest in the riots’ underlying causes. These included a poor harvest, unmet demands for electoral reforms, inspiration from the French Revolution, declining incomes and income-support in rural areas, and – most relevant to today – the progressive mechanization of agriculture that contributed to decreased living standards for agricultural workers.
Professors Caprettini and Voth hypothesized that technological unemployment was an underemphasized cause. To investigate this, they gathered data on 66 local newspapers to see the specifications of farms listed for sale. Some of the farms included threshing equipment, which was the new technology that undermined the livelihoods of rural manual laborers. They also gathered data on soil suitability, and found that the adoption rates of these threshing machines was highest where one expected it to be – in the areas most suitable for wheat production.
Finally, they compared these data to those on the incidence of the riots, and they found a very strong relationship between the adoption of threshing machines and the occurrence of civil unrest. Thus, the observed destruction of agricultural machines was more than generic violence – it was an expression of frustration toward the perceived cause of the rural workers’ distress, as well as a potentially strategic, though ultimately futile, attempt to rebalance the economic equation in workers’ favor.
The nearly 300 years that have passed since the industrial revolution have witnessed many bouts of technological unemployment, and most have not led to social unrest, usually because the jobs that are lost are replaced by new ones. After all, on the eve of the COVID-19 pandemic, US unemployment was below 4 percent, and there was no evidence of a sustained increase in global unemployment caused by automation or any other technology.
However, as with many events in 2020, there is a sense that on this occasion, we are witnessing a departure from conventional trends. Economic hardship in advanced and emerging economies has increased sharply during the last few months, and permanent automation may well be the straw that breaks the camel’s back.