Research

I use administrative and survey data to study the effect of automation on workers at different stages of the life cycle. Using a difference-in-difference design, I exploit variation across occupations in the share of tasks that can be completed by automatic machines to analyze earnings outcomes by age. I find that the effect of automation on earnings for young workers is four times as large as for older workers. I then develop a life-cycle model with task-based automation that can answer why it is young workers in highly exposed occupations who face the largest decline in earnings. In the model, both wages and amenity values change following an expansion of automatic machines. Amenity declines account for 72 percent of the effect of automation on earnings for young workers, as these declines drive young workers to reallocate to lower-wage occupations. The model points to the importance of accounting for both wage and non-wage implications when analyzing the consequences of automation.

We document new facts about the sources and consequences of sexual harassment in the workplace using administrative and survey data from Denmark. We estimate that following workplace sexual harassment, victims, both men and women, see earnings losses of around 6 percent. Losses are largely driven by workers who transition to new firms following harassment suggesting that the consequences of harassment persist even when workers leave the job where the harassment occurred. Victims who leave their employer after an incident of workplace sexual harassment incident move to firms with a larger share of colleagues who share their own gender.