Digitization–Based Automation and Occupational Dynamics

We examine the relationship between occupational automation probabilities and employment dynamics over nearly two decades. We show that employment and wage shares of occupations with a higher automation risk have declined in Sweden over the period 1996-2013. This has occurred both at the aggregate private business sector but also within firms, where the wage share changes have been larger than the employment share changes. Combining the automation risk in workers’ occupations with individual worker characteristics, we find substantial heterogeneity. This includes that education dampens the automation risk of workers, as the average automation probability of low-skilled workers is almost twice as high as of university graduates. Employment shares in highrisk occupations have moreover declined across all wage levels, and most so in high-wage occupations. Our results indicate that it is not necessarily the level of risk that matter, but rather the heterogeneity between occupations.