The Impact of Fear of Automation

In this paper, we establish a causal effect of workers' perceived probability of losing one's job due to automation on worker's policy preferences and workplace intentions. In a representative sample of the US workforce, we elicit the perceived fear of losing one's job to robots or artificial intelligence. We document a strong relationship between fear of automation and intentions to join a union, retrain and switch occupations, preferences for higher taxation, higher government handouts, populist attitudes, and voting intentions. We then show a causal effect of providing information about occupation-specific job loss probabilities on preferred levels of taxation and handouts. In contrast, the information treatment does not affect workers' intentions to self-insure by retraining or switching occupations, but it increases workers' self-reported likelihood of joining a union to seek more job protection. The treatment effects are mostly driven by workers who are informed about larger job loss probabilities than they perceived.