No abstract

1 With thank Erica Blom for her assistance us with some early data issues, and seminar participants at Yale University for helpful comments. AbstractWe analyze outcomes of college graduates as a function of the economic conditions they graduated into and the skill requirements of their eld of study. To this end, we combine multiple data sets with information on earnings and eld of study for U.S. college graduates graduating between 1976 and 2011. This provides coverage of multiple business cycles and larger sample sizes than the typical cohort-based analyses in this literature. We categorize college majors by indicators of skill in the majors, predominantly the average earnings premium. We then measure earnings, wage, employment, occupation skill level and educational attainment outcomes across graduation cohorts and major skill level. We nd that early careers are disrupted by poor labor market conditions; a large recession at time of graduation reduces earnings and wages by roughly 9% and 13% (respectively) in the rst year, and reduces the probability of full-time employment by 11 percentage points. These e ects are fairly short-lived, fading out over the rst ve years of a career or so. We also nd that the earnings gap across college majors widens in recessions; a typically high-earning major increases his or her earnings advantage by a third when graduating in a bad recession, and this e ect remains large in magnitude for the rst seven years after college graduation. We nd evidence of small, positive educational attainment e ects among low-return majors graduating into a worse economy. We nd no impacts on occupation quality. We also determine that di erential cyclicality of college major cannot account for our ndings. We compare our earnings and wage e ects across recessions and nd that overall earnings losses from poor entry conditions are substantially larger for graduates in the 2001 and 2007-09 recessions, and these earnings losses are more evenly dispersed across college majors.

We study variation in skill demands for professionals across firms and labor markets. We categorize a wide range of keywords found in job ads into ten general skills. There is substantial variation in these skill requirements, even within narrowly defined occupations. Focusing particularly on cognitive and social skills, we find positive correlations between each skill and external measures of pay and firm performance. We also find evidence of a cognitive social-skill complementarity for both outcomes. As a whole, the job skills have explanatory power in pay and firm performance regressions, beyond what is available in widely-used labor market data.