First, close to half of all jobs are at risk of being eliminated by technology that is already in use at commercial locations, while a quarter of all jobs are at risk of trade-related dislocation. To explore this, we mapped data created by researchers at Oxford and Princeton Universities estimating the risk of automation and trade related job losses to the 3,144 US counties. This provided an interesting visualization of where potential job losses may cluster. We then conducted a more formal battery of assessments of the clustering of potential job losses across people and places. The results were eye opening.
It turns out that the higher a county’s risk of losing jobs to automation, the higher the risk in the adjacent counties. This confirms what appears to be true in the maps, and is disconcerting since it means that automation-related disruption likely extends across labor markets. This is true also with trade-related job losses. The other side of this relationship is that counties with lower risk of automation or trade-related job losses are adjacent to other low-risk places. This simple fact implies that regional inequality might be poised for a big increase.
The study also pointed out that as these jobs are lost, other jobs are created, but these new job openings are in different places and require different skills than those that are lost. This prompted us to look at what happens to individual workers and households. There the results were even more startling.
Trade-related job losses due to offshoring or import substitution tend to cut across educational and income levels. There is no correlation between risk of job losses due to trade and either education or earnings. Automation risk couldn’t be more different. There is a strong inverse relationship between educational attainment and risk of automation job losses. The data on wages is even starker. Wages for the lowest risk 10 percent of occupations average about $84,000 per year. Wages for the highest risk 10 percent are about $36,000 per year. Indeed, for workers who have more than a 50/50 chance of automation-related job losses, wages are under $40,000 per year. Those with less than a 50/50 chance of automation-related job losses average almost $70,000 per year.
Taken together, the increasing risk of automation and trade-related job losses will disproportionately impact low-wage, low-skilled workers who live in counties with an abundance of other similar workers. This is a perfect recipe for growing income inequality across regions and households. Beyond the distasteful political manifestations of inequality, we should worry that labor market disruptions falling most heavily on those least prepared to adjust could quickly turn into much broader social and economic problems. Automation, and to a lesser degree trade, impacts the most vulnerable people and places in America. Growing urban migration exacerbates the pain for communities.
The findings of our study (www.bsu.edu/cber/publications) should give readers some concern, but we also want to be clear that technological change, trade and the rise of cities generates untold wealth, opportunity and other secondary benefits that can scarcely be measured. Economic growth is good and the results benefit us more than we imagine. I simply point to the 50 extra years of life that the average American enjoys since the start of the Industrial Revolution.
Still, rapid technological change, particularly as it affects clustered industries or occupations can generate real economic discomfort. Change is not always easy, and not every household finds itself clearly better off in the short run. Our hope in authoring this research is that policymakers can prepare for and embrace these changes. To do so, we should think about policies at the state and local level, especially in education policy, that make vulnerable communities and households more resilient to change.