Who Came, and Who Stayed

Two lines, two different questions. The first chart counts arrivals and is close to complete: the Institute of International Education has run an annual census since 1948, and the numbers are a headcount. The second chart is an estimate of departures, and it is much narrower, because the United States has never systematically tracked whether international students leave. What exists is the National Science Foundation's Survey of Doctorate Recipients, which follows people who earned science and engineering PhDs. So the stay rate here describes doctorate holders, roughly five percent of the international student population, and is almost certainly higher than the rate for the whole group. Undergraduates and taught masters students leave at far greater rates. Treat the second chart as the ceiling, not the average.

The policy markers are where the two charts meet. Nearly every inflection in the enrollment line sits next to a legal change rather than an economic one. The 1965 Hart-Celler Act abolished national-origin quotas and Asian enrollment climbed for three decades afterwards. The 1990 Immigration Act created the H-1B visa and with it a defined path from a student visa to a job, which is the mechanism that turns arrival into settlement. After September 2001 the SEVIS tracking system and slower consular processing flattened enrollment for four years. The 2008 and 2016 extensions of STEM Optional Practical Training, to seventeen and then twenty-four months beyond graduation, are visible directly in the shaded band: OPT went from 67,804 in 2010 to 294,253 today, and now accounts for one in four international students in the country.

The scarcinality reading. Nothing about American universities became physically scarcer or more abundant at any of those moments. The classrooms, the laboratories, the faculty, and the willing students all existed continuously across every one of those turns. What changed each time was a legal permission, and the framework's point is that a permission is a claim, not a resource. This is a case where the ordering is unusually clean: the binding constraint on how many talented people work in American science has almost never been the supply of talented people, and it has almost never been the physical capacity to teach them. It has been the number of pieces of paper the state chose to issue. When that number went up, in 1965 and 1990 and 2008 and 2016, the flow rose to meet it. When it went down, it fell.

That makes the current moment legible in the framework's own terms. New enrollment fell 17 percent in autumn 2025 and visa issuances fell 36 percent over the preceding summer, and the proposals now pending would replace duration-of-status admission with fixed end dates and restrict or eliminate Optional Practical Training. Read alongside the research funding cuts, the same mechanism appears twice: capacity that took decades to build being reduced by decisions whose consequences arrive long after the decision-makers have gone. The stay rate is the part worth watching, because it is the point at which an educated person becomes a permanent addition to the country rather than a temporary visitor, and unlike enrollment it is measured with a lag of five to ten years. By the time it moves, whatever moved it will be a decade old.

Sources: Institute of International Education, Open Doors census of international student enrollment, 1948/49 to 2024/25, downloaded directly from the published series. Optional Practical Training counts are reported within that census from 1990. Stay rates from the Center for Security and Emerging Technology analysis of the NSF Survey of Doctorate Recipients, covering graduating classes 2000 to 2015 measured as of February 2017, and from NSF NCSES for the 2017 to 2019 cohort measured at five years. The 2026 enrollment point is an estimate reflecting the reported 17 percent fall in new enrollment and is drawn dashed.