Impacts from Automation Diffuse Locally

Workers that become automated may transfer productivity gains to their co-workers or make it easier to automate their jobs too. In this paper, I empirically investigate how automatable jobs have diffused impacts to neighbouring jobs in North American cities between 2007 and 2016. Results indicate that jobs that share similarities with neighbouring high-risk jobs grew less, even when controlling for their own technical risk of automation. Conversely, jobs that share complementarities with neighbouring high-risk jobs grew faster, possibly indicating productivity gains from working with recently automated jobs. In addition to the analysis in this paper, I provide an adjusted index of job automation risk that accounts for local diffusion of impacts (negative and positive) in US cities.