Using discontinuities from the Danish college enrollment system, we find that students who are marginally accepted into their preferred program in a broad field that is different from their next-best choice (e.g., business rather than science) experience significant and long-lasting rewards as a result. In contrast, students whose preferred and next-best programs lie within the same broad field do not. Exploiting data from online job postings, we find that the estimated effects on skill usage similarly vary according to the degree of similarity between preferred and next-best choices.