From self-driving cars to self-checkout machines in grocery stores, automation is popping up everywhere. In a recent Economic Synopses, Economist Sungki Hong and Senior Research Associate Hannah Shell took a look at how changes brought on by automation and computerization might affect employment and income inequality.
In the essay, Hong and Shell used estimates by Carl Frey and Michael Osborne to help determine how likely automation is to affect certain occupations. Frey and Osbourne identified the tasks of each occupation that may become automated—perception and manipulation tasks, creative intelligence tasks and social intelligence tasks—and then used a machine learning algorithm to calculate the probabilities that a computer could perform those tasks.
Hong and Shell then merged those estimates with occupational employment and wages data from the Bureau of Labor Statistics’ 2017 Occupational Employment Statistics dataset. “Occupations with large employment and low income have a higher automation probability,” they found. This includes office and food service jobs.
The authors then studied the effect of automation on income inequality in hypothetical scenarios. To do this, they calculated the Gini coefficient, which measures income inequality in a population. (A Gini coefficient of 1 represents a perfectly unequal distribution, while a coefficient of zero represents equal incomes.)
Hong and Shell also calculated the 90-50 ratio, or the ratio of the income of the 90th percentile to the 50th percentile, and the 50-10 ratio, which is the 50th to the 10th percentiles. Higher ratios indicate higher income inequality.
Hong and Shell calculated Gini coefficients and percentile ratios assuming that if an occupation has a 60 percent chance of automation, 60 percent of the employees would get one of the following three hypothetical, alternative labor market outcomes after automation:
“Automation increases inequality in every scenario because it tends to displace the lowest-paid workers,” Hong and Shell wrote.
In the first scenario, when affected employees became unemployed, the Gini coefficient more than doubled, going from 0.31 to 0.70, Hong and Shell found. However, this is an extreme example, they said. “It is used here only as an exercise to understand where automation will most impact the income distribution.”
The second and third hypotheses—that affected employees would earn the minimum wage or take a 20 percent pay cut—were more likely. On the minimum wage, the Gini coefficient was 0.43 with a 50-10 ratio of 1, meaning the bottom half of the distribution would earn minimum wage.
The scenario closest to reality, though, is that the affected employees would take a 20 percent wage cut, Hong and Shell wrote. The Gini coefficient increased to 0.35, and the 90-50 ratio increased more than the 50-10 ratio, “indicating that income becomes more concentrated at the top of the distribution,” they wrote.
The authors said some caution is needed in interpreting these numbers, pointing to other factors that influence the impact of automation:
