Pure-Chance Jobs vs. a Labor Market: The Impact on Careers of a Random Serial Dictatorship for First Job Seekers

Does a worker's first job affect her long-run career? Do any such "first job effects" vary across workers of different types? If so, can policy improve upon a "free" labor market by altering initial matches with employers? We begin to study the impact of market design on the performance of entry-level labor markets, by comparing 20 years when Norway assigned doctors to their first job---residencies---through a Random Serial Dictatorship, with the post-2013 era when the RSD mechanism was replaced with decentralized job-finding. We first estimate the consequences for long-run earnings of different employers for male and female workers. We do so by exploiting RSD-generated random, individual level variation in workers' initial choice set over employers. We then decompose preferences over employers into a component that is due to first job effects and another that is due to the "amenity value" workers of a given type associate with employers of a given type. Finally, we show how realized first job effects, amenity values, and overall worker welfare differ, for each group and in total, in a decentralized labor market compared to a randomized-choice-sets system, by describing how worker-employer matches changed after 2013.