Automation and the Consequences of Occupational Decline

Sammanfattning: Essay I. Automation affects workers because it affects the return to their skills when performing different tasks. I propose a general equilibrium model of occupational choice and technological change which takes two important labor market features into account: (i) automation happens to tasks and (ii) workers have bundled skills. Equilibrium skill returns vary across tasks, and the impact of automation on skill returns is task-specific. I find that, to a first-order approximation, skill returns depend only on the relative skill allocation in each task. In equilibrium, automation reduces employment in the task subjected to automation so long as tasks are gross complements. This reduction in employment increases both tasks' intensity in the skill used intensively in the automated task. This increased intensity is coupled with a universal decline in the return to the skill used intensively in the automated task.  Conversely, the return to the other skill increases in both tasks.Essay II (with Per-Anders Edin, Tiernan Evans, Georg Graetz, and Guy Michaels). We assess the career earnings losses that individual Swedish workers suffered when their occupations' employment declined. High-quality data allow us to overcome sorting into declining occupations on various attributes, including cognitive and non-cognitive skills. Our estimates show that occupational decline reduced mean cumulative earnings from 1986-2013 by no more than 2-5 percent. This loss reflects a combination of reduced earnings conditional on employment, reduced years of employment, and increased time spent in unemployment and retraining. While on average workers successfully mitigated their losses, those initially at the bottom of their occupations' earnings distributions lost up to 8-11 percent.Essay III. Does the long-term economic stress of occupational decline cause health problems, or even death? This essay explores this question using Swedish administrative data, and a measure of occupational decline obtained from detailed US data on employment changes over almost 30 years. I investigate whether people who experience occupational decline have higher mortality or hospitalization rates, and in particular if they are more likely to suffer from cardio-vascular disease or deaths of despair: deaths caused by alcohol, drug or suicide. I find that workers who in 1985 worked in occupations that subsequently declined, had a 5-11 percent higher risk of death in the 30 years that followed, compared to same-aged, similar workers in non-declining occupations. For men in declining occupations, the risk of death by cardio-vascular disease was 7-14 percent elevated, while women in declining occupations faced 31-37 percent higher risk of death by despair. The risk was higher for workers who were lowest paid in their occupations.