Job Market Paper

College Major Specificity, Earnings Growth and Job Changing

Earnings growth is a key feature of the early career and mean growth rates vary substantially between college majors. I find that the specificity of a college major's skills is a key driver: general majors initially earn 18% less than specific majors but close the gap to 6% within thirteen years. To study the relationship between job changing and earnings growth, I link survey data to administrative earnings records. General majors switch occupations, employers and industries 20-30% more often. A decomposition shows differences in job changing rates account for 30-50% of the explainable portion of the earnings-growth difference between majors.

Working Papers

The Skill Content of College Majors: Evidence from the Universe of Online Job Ads

with Steven W. Hemelt, Brad J. Hershbein, and Kevin M. Stange (Submitted), NBER WP #29605

We use the near universe of U.S. online job ads to document four new facts about the skills employers demand from college majors. First, some skills––social and organizational––are demanded from all majors whereas others - financial and customer service - are demanded from only particular majors. Second, some majors have skill demand profiles that mirror overall college graduate demand, such as Business and General Engineering, while other majors, such as Nursing and Education, have relatively rare skill profiles. Third, cross-major differences in skill profiles explain considerable wage variation. Fourth, although major-specific skill demand varies across place, this variation plays little role in explaining wage variation. College majors can thus reasonably be conceptualized as portable bundles of skills.

Grads on the Go: Defining College-Specific Labor Markets for Graduates

with Johnathan G. Conzelmann, Steven W. Hemelt, Brad J. Hershbein, Andrew Simon, Kevin M. Stange (Under Review)

This paper introduces a new measure of the labor markets served by colleges across the United States. The geographic dispersion of alumni is more than twice as great for highly selective 4-year institutions as for 2-year institutions. However, more than one-quarter of 2-year institutions disperse alumni more diversely than the average public 4-year institution. In an application of these data, we find that the average strength of the labor market to which a college sends its graduates predicts college-specific intergenerational economic mobility. We also use the data to quantify the extent of state-level “brain drain” and to identify institutional peers. 

Works in Progress

Employer Learning and College Selectivity

I investigate whether employer learning about worker productivity contributes to differences in the labor market outcomes between college graduates from selective and unselective institutions. At labor market entry, employers can’t fully measure a worker’s productivity, and instead use observable proxies like the selectivity of a worker’s post-secondary institution to make inferences. As there are large difference across U.S. institutions in mean standardized test scores (a correlate of worker productivity), institutional selectivity is likely an informative signal. However, the variation in test scores is much larger at unselective institutions, and consequently, mean test scores are relatively less informative signals of the productivity of graduates from these institutions. I capture this distinction by embedding institutional selectivity in a model of employer learning and statistical discrimination. Analysis using data from the Baccalaureate & Beyond 1993/2003 reveals two key findings that are aligned with the model predictions. First, at labor market entry, employers are able to more accurately predict the ability of graduates from selective institutions: among graduates from selective institutions, workers with high test scores (ACT or SAT), earn 3% more than those with low test scores at labor market entry, but among graduates from unselective institutions the difference is not statistically different from zero. Second, employers only slowly learn about the productivity of graduates from unselective institutions: among graduates from unselective institutions, the gap between high and low scoring graduates grows to 10% after ten years in the labor market indicating that employers can increasingly observe and reward differences in productivity.

Decomposing Earnings Differences Across College Majors

I investigate the mechanisms driving earnings inequality between college majors. Using data from the 2003-2017 waves of the National Survey of College Graduates (NSCG), I find that at baseline, majors earning one standard deviation above the average major earn roughly 18.5% more a year. Earnings differences between majors, as measured by the coefficients on major fixed effects in a sparse log earnings equation, capture both differential sorting of college majors across careers and the average premium paid to skill for that major within a career. Accounting for differences in careers – a combination of occupation, employer attributes, job tasks and job levels – between college majors reduces the earnings advantage a high-paying major over the average major by 56% to roughly 7.6% a year. Controlling for sorting across careers compresses the distribution of major-specific earnings, but on average the relative ranking of majors in terms of earnings is unchanged. To quantify the relative importance of different aspects of the career in reducing earnings differences, I employ the Gelbach (2016) decomposition. Results indicate that one-third of the total reduction in earnings differences is due to occupation, one-third is due to employer attributes, and an additional 20% is due to differences in job tasks and levels.