This is a condensed version of the original article.
The authors present a simple, two-period model of human capital accumulation on the job and through college attainment. They use a calibrated version of the model to explain the observed flattening of the life-cycle earnings profiles of two cohorts of workers. The model accounts for more than 55 percent of the observed flattening for high school-educated and for college-educated workers. Two channels generate the flattening in the model: selection (or higher college attainment) and a higher skill price for the more recent cohort. Absent selection, the model would have accounted for no flattening for high school-educated workers and about 23 percent of the observed flattening for college-educated workers.