Career Consequences of Firm Heterogeneity for Young Workers: First Job and Firm Size
[Job Market Paper]
I study the link between firm heterogeneity and young workers' long-term career outcomes. Using administrative (social security) data from Spain, which include workers' labor market histories and education, I investigate the long-term effects of landing a first job at a large firm versus a small one. Size could be a relevant employer attribute for inexperienced young workers because large firms are associated with greater training, higher wages, and enhanced productivity. The key empirical challenge is selection into larger firms—for instance, more able people may land jobs at these firms. To overcome this challenge, I develop an instrumental-variables approach that leverages large firms' year-to-year idiosyncratic hiring shocks within each region. These large-firm hiring episodes, in turn, generate variation in the composition of regional labor demand. I find that starting at a larger firm leads to substantially better career outcomes such as lifetime income. To shed light on the mechanisms driving this result, I test whether the effect is (i) due to workers staying with their first employer; (ii) driven by job search mechanisms favoring large-firm workers; (iii) present for those who lose their first job; (iv) explained by experience acquired at large firms being more valuable. These tests find support for two complementary channels. The first is a job search channel by which a larger first employer leads to subsequent jobs at other large firms. The second is a human capital channel by which on-the-job skills developed in formative years are more valuable if they are acquired at larger firms.
Displacement, Diversity, and Mobility: Career Impacts of Japanese American Internment
One of the largest population displacement episodes in the U.S. took place in 1942, when over 110,000 persons of Japanese origin living on the West Coast were forcibly sent away to ten internment camps for one to three years. Having lost jobs and assets, after internment they had to reassess labor market and location choices. This paper studies the long-run career consequences of this episode for those affected. Combining information from Census data, camp records, and survey data I develop a predictor of a person's future or past internment status based on Census observables. Using a difference-in-differences framework I find that internment had a positive average effect on earnings in the long run. This effect is robust to different control groups of non-interned Japanese and Chinese Americans. The evidence is consistent with information and skills exchange, possibly enabled by the camps' economic diversity, followed by increased occupational and geographic mobility as likely mechanisms. I find no evidence of other potential drivers such as increased labor supply, or changes in cultural preferences. These findings provide evidence of labor market frictions preventing people from accessing their most productive occupations and locations, and shed light on the resilience of internees who overcame a very adverse initial shock.
It's Not the Student's Major, It's the Student that Matters: Towards Plausibly Causal Estimates of the Returns to Majors (with Ian Hoffman and Caroline M. Hoxby)
Raw differences across college majors in students' earnings on their early jobs are often publicized as "returns to majors" in the U.S. Although these raw differences receive a great deal of attention and appear to influence students' choices, they are not plausibly causal returns. That is, they do not represent the change in lifetime earnings that students would experience if they were to change majors. This is because students who differ on aptitude and other dimensions (i) choose postsecondary institutions that differ not just in selectivity but also in which majors preponderate, (ii) choose different majors within any given postsecondary institutions, (iii) attend postsecondary schools for different numbers of years, and (iv) pay different amounts for their postsecondary education. In addition, early job earnings are a poor guide to lifetime earnings differences because some majors, particularly those that teach technical skills that depreciate, are associated with high initial earnings but unusually flat earnings growth over the lifetime. In this paper, we make the most ambitious attempt to date (for the U.S.) to estimate returns to majors that plausibly approach the causal returns. We do this by using comprehensive, administrative data; controlling flexibly for a student's pre-college aptitude and achievement; controlling flexibility for the selectivity of a student's college (even employing college fixed effects); controlling flexibly for students' pre-college family background and location; accounting for differences in the costs of students' education; using longitudinal data to account for differences in years of postsecondary education; and computing lifetime earnings as opposed to early job earnings. We find that the much publicized raw differences in early-job earnings enormously exaggerate differences in plausibly causal lifetime returns. Raw earnings most overstate plausibly causal returns for majors that (i) are almost never picked by students with modest test scores; (ii) are prevalent at very selective institutions but unavailable at non-selective ones; (iii) teach skills that generate high initial earnings but unusually flat earnings growth; (iv) are male-dominated. Engineering is an example. Even after producing the most credible estimates we can, we are left with earnings differences that probably still reflect differences in the preferences of students who choose different majors. For instance, some remaining earnings differences appear to represent preferences for altruism versus materialism.