Three Essays on Connecting to Work

This is a three essay dissertation on empirical labor economics, with a focus on connecting to work and human capital. The first chapter investigates the occupational choice behavior of science and engineering graduates. Many science, technology, engineering, and mathematics(STEM) degree holders work in jobs that are unrelated to their field of study. This paper uses an endogenous switching regression model to explore the possibility that mismatch is the result of rational sorting behavior, whereby individuals who expect a higher salary in unrelated occupations select into those fields. Analysis of the 2013 National Survey of College Graduates (NSCG) suggests that workers that are better suited to working in unrelated jobs do tend to select into those jobs, although the costs of mismatch identified in the prior literature persist despite sorting behavior. This suggests that rational sorting is an important component of an explanation of educational mismatch in the STEM workforce, but it does not entirely alleviate concerns about mismatch in this population. The second chapter studies how registered apprenticeship sponsors respond to the business cycle. Evidence from other countries suggests that apprentice employment is relatively robust to the business cycle, but we know little about how apprentice employment changes with the business cycle in the United States. This paper uses administrative data on registered apprentices to study the cyclicality ofapprentice employment at the sponsor (i.e., employer) level. Much like their European counterparts, employment of American apprentices declines only modestly when unemployment increases. Most of the adjustment of apprentice employment levels occurs through adjustments in hiring. The third chapter focuses on Georgia's Jobs Tax Credit. The Jobs Tax Credit others increasingly generous benefits to establishments that increase their employment levels in targeted high-poverty counties. This study estimates the causal effect of the credit on employment and earnings growth using a regression discontinuity design (RDD). The RDD exploits the fact that benefit levels jump discontinuously at nonmanipulable points in a pre-determined ranking of Georgia's 159 counties. The RDD tests suggest that the credit does not increase employment or earnings in targeted counties.