The ability to attract and retain human capital is critical for regional development. In Colombia, in two regions that are the principal mining royalties receptors, 60 % of people are considered poor. One reason is the misallocation of royalties, caused by the poor design of the project funded by royalties. Consequently, a source of misallocation could be that the royalty receptors lack human capital that could design and supervise royalties-funded development projects. Meanwhile, 25 % of Colombian university graduates struggle to find their first job. One solution to this twofold problem is to promote the migration of recent college graduates. But what motivates the migration from big cities to less developed ones? The intention of migration cannot be studied using data of actual migration and requires data on expectations to include uncertainty in the decision-making process. Using a unique survey on subjective expectations, meaning asking about intentions of migration in a probabilistic form. I study the determinants of the probability of migration intention of 747 future college graduates in Colombia,from the capital city to three mining royalties' receptors cities. I collect probabilistic expectations about migration, availability of amenities in the origin and arrival destinations, and expected income. The results show that the difference between the subjective expectations of expected income and amenities (access to schools, roads and hospitals) at source and at the arrival destination together with a higher level of trust in others, increases the probability of migration to less developed areas in Colombia. In contrast, preference for present returns, being a student with top grades, or from a privileged economic background decreases this probability. Data collection on migration expectations for these groups will contribute to designing public policy interventions to tackle first-graduate employment and promote institutional capabilities in less developed areas.