Employer Learning, Statistical Discrimination and University Prestige

Abstract This paper investigates whether employers use university prestige as a signal of workers’ unobservable productivity. Our test is based on employer learning-statistical discrimination models, which suggest that if employers use university reputation to predict a worker’s unobservable quality, then college prestige should become less important for earnings as a worker gains labor market experience. In this framework, we use a regression discontinuity design to estimate a 13% wage premium for college graduates in their first year of the labor market who were barely accepted by one of the two most prestigious universities in Chile compared with those barely rejected by these two schools. However, we find that this premium decreases to 4% for workers with 6 or more years of labor market experience. This result suggests that college prestige becomes less important for employers as workers reveal their quality throughout their careers.