Association for Public Policy Analysis & Management

Recent labor market trends make it increasingly likely that U.S. workers will need to adapt to changing skill requirements during their careers. For some workers, adaptation may require changing occupations as labor market polarization arising from automation, outsourcing, or skill-biased technological change reduces the demand for routine manual and cognitive skills. Yet little is known about the career trajectories of different types of workers due to a lack of longitudinal data with sufficiently large sample sizes. Using a novel dataset of 23 million resumes collected from U.S. recruitment, staffing, workforce agencies and job boards by Burning Glass Technologies, we study occupational transitions over the course of individual careers. Exploratory analyses indicate that roughly 9 percent of individuals transition across narrowly-defined six-digit occupation categories each year during our period of study (2002-2018). Of those who do transition across six-digit occupations, three-quarters also transition across more broadly defined two-digit occupations suggesting a high degree of labor market fluidity. In comparison, annual occupational mobility estimates from the literature range from 4 percent using the Current Population Survey to measure six-digit transitions in the 2000s (Molloy et al., 2017) to 21 percent using the Panel Study of Income Dynamics to measure three-digit transitions (Kambourov and Manovskii, 2009) in the 1990s. We go beyond calculating annual transition rates to measure occupational mobility paths over longer periods based on the longitudinal listing of jobs on each resume to assess the degree to which individuals move across occupations during their careers. Additional information parsed from the text of the resumes can be used to determine the degree to which these occupational transitions occur in conjunction with obtaining additional training or education (e.g., certificate, associate degree, bachelor’s degree, or advanced degree).Using data on workers who entered the labor market at different points in time and reside in different locations, we can also assess whether the role of additional training and education has changed over time, particularly as distance learning has expanded opportunities for those in more rural areas. Finally, we test whether the frequency and nature of occupational transitions vary across age cohort, gender, region, initial occupation and industry, level of education when entering the labor market, and business cycles. Understanding the various factors that influence occupational transitions may help guide workforce development policy aimed at re-training workers displaced by structural changes in the labor market. This is particularly important as new technologies, such as artificial intelligence, replace many of the routine cognitive tasks performed by humans. Understanding how individuals can apply their existing skills to new occupations or acquire new skills that are in demand can improve the efficiency of the labor market while ensuring a more inclusive strategy for workforce development.