Twenty years of institution-level research has found no evidence that international students displacedomestic students in U.S. higher education. Using Integrated Postsecondary Education Data System(IPEDS) enrollment data for degree-seeking students at 1,626 in-person four-year institutions withcompetitive admissions, dynamically ranked into selectivity deciles from 2003 through 2024, this studyidentifies system-level displacement that is large, statistically robust, and invisible at the level ofanalysis every prior study employed. The principal findings are: Between 2013 and 2024, international enrollment growth at competitive institutions exceededtotal capacity expansion by more than 200,000 seats. The displacement of domestic studentsover this period is not a statistical inference. It is arithmetic. Year-over-year international enrollment growth at the most selective decile is negativelycorrelated with domestic enrollment at every downstream selectivity tier, with multiple tiersreaching statistical significance under Pearson, Spearman, and Newey-West corrected methods. The displacement survives multivariate regression controlling for capacity expansion and nettuition revenue, directly testing and rejecting the cross-subsidization hypothesis thatinternational enrollment benefits domestic students. A natural experiment exploiting three years of exogenous international enrollment declineconfirms the direction of the effect at p = 0.001. The displacement is not evenly distributed. It concentrates sharply on specific domesticsubpopulations identifiable by race and gender. Replication of the prior literature's institution-level framework reproduces the null result,confirming that the displacement mechanism operates through hierarchical redistribution acrossinstitutions — a process structurally undetectable at the unit of analysis previously employed. There are 975,000 fewer white men and women at competitive institutions today than in 2003, despitetotal enrollment expanding by 1.4 million seats. The prior consensus that international enrollment isbenign to domestic students is unsupported by the data when examined at the appropriate level ofaggregation.
Ryan Grant (Mon,) studied this question.