This article investigates the hypothesis that when measures of specific human capital (such as job tenure) are included in earnings functions, there may be a sample selection bias because of job-matching effectsbecause workers with high unobserved match quality receive and accept high wage...