How much can late schooling investments close racial and ethnic skill gaps? We investigate this question by exploiting the large differences in completed schooling that arise among teenagers with birthdays near school-entry cutoff dates. We estimate that an additional year of high school raises the Armed Forces Qualifying Test (AFQT) scores of minorities in the NLSY 79 by 0.31 to 0.32 standard deviations. These estimates imply that closing existing racial and ethnic gaps in schooling could close skill gaps by between 25 and 50 percent. Our approach also uncovers a significant direct effect of season of birth on test scores, suggesting that previous estimates using season of birth as an instrument for schooling are biased.