By applying a regression discontinuity design to national data of students at four-year colleges, this study identifies the substitution effects of exogenously received increases of grant aid on hours of paid labor, earnings, and borrowing while in college. Results confirm students substitute grant aid for both paid labor and borrowing. An average increase of $1,100 in grant aid reduces weekly job hours by 1.5-2 hours per week for women, corresponding to a decline in annual earnings of $850, and reduces borrowing by an average of $300-$400 dollars among all students. We find limited evidence of grant aid's impact on academic outcomes.