A random sample of 54 research universities comprised of 23,394 scientists and a mathematical model show how some scientists license their inferior discoveries through their technology licensing offices (TLOs) while privately taking their finest discoveries directly to market. Misappropriation increases with heightened entrepreneurial activities on campuses. Autonomous TLOs and royalty sharing with departments are associated with reduced misappropriation. Agency theory should better account for conditions in which principals have information to sanction agents but instead choose to absorb costs stemming from agents' opportunistic behavior.