In this paper, the authors divide actual years of schooling into effective years, repeated years, skipped years, inefficient routing years, and dropout years. Estimation of earnings functions reveals that this topology is statistically superior to the usual concepts of either actual or effective years of schooling. Based on the authors' distinction, they formulate tests that discriminate between the human capital theory and the screening hypothesis. The results give strong support to the human capital theory and refute the screening hypothesis.