The automation index captures an occupation’s risk of being affected by automation
using four measures:
- % of time spent on high-risk work
- % of time spent on low-risk work
- Number of high-risk jobs in compatible occupations
- Overall industry automation risk
This methodology starts with the underlying work on task content. We use estimated
task time shares, derived from O*NET work activities, and regress them for each
occupation based on Frey and Osborne’s published “computerization probabilities”
(2013). This helps us identify which tasks are positively and negatively correlated
with automation risk.
This classification is then linked with the task time shares to identify the share of each
occupation’s time spent in high- and low-risk work, from an automation perspective.
Then we look at the place of an occupation in the broader context of labor
market automation risk. Using occupation compatibility scores, we look at all similar
roles (defined as having an O*NET compatibility score over 75) and find the percentage
of jobs in those similar roles that are at risk of automation.
Finally, using staffing pattern data, we multiply the share of an occupation’s jobs in 3-
digit NAICS industries by that industry’s share of at-risk jobs to calculate the overall
industry automation risk.
We then standardize all these measures and scale the index so that 100 = the “average
worker,” defined as the average index across all occupations, weighted by job numbers
in 2018. The index has a standard deviation of 15. Note that the share of time spent on
low-risk work is a negative contributor to an occupation’s index score (making the index
score lower) while the other three measures are positive contributors (making
the index score higher).