In August, the Department of Homeland Security published a notice to amend the regulation regarding foreign students’ visas and proposed capping all international student visas at four years, with extensions required for longer stays. While the proposal has not yet been formally approved, a decision is expected very soon. One stated goal is protecting American students from displacement. Capping visas wouldn’t do that—a look at the basic economic incentives at play explains why—but it would strike a blow to America’s innovation engine at a moment when competition for consequential breakthroughs, such as achieving artificial general intelligence, is intensifying across the globe.
Vice President J. D. Vance has explained the administration’s concern about admissions anxiety this way: “It’s bad for the American dream for a lot of kids who want to go to a nice university and can’t because their spot was taken by a foreign student.” That indeed would be bad if it were happening, except it’s not. International students represent fewer than 6 percent of undergraduates, and they typically finish in four years anyway. A four-year visa cap won’t affect them.
It’s at the graduate level—where America trains the researchers who drive technological progress—that a cap would matter. International students do make up a large fraction of STEM graduate students in the United States. But that’s because they apply for these spots at far higher rates than Americans do. In 2021, 80 percent of engineering graduate applications came from international students; in math and computer science, the figure was 73 percent.
That clearly isn’t a case of displacement. Rather, it reflects rational choices being made by American students. Few apply to these programs because the economics don’t work for them. Consider the numbers: the median annual salary of someone with a bachelor’s degree working in computer science is $110,000. A master’s degree bumps that up to $133,000. Compare that to the real cost of the investment needed to secure a $23,000 gain: two years of forgone salary ($220,000), plus tuition and expenses ($100,000). It takes at least fourteen years to recover that $320,000.
For PhDs, the investment pays out even less. Doctoral programs in sciences and engineering take a median of 5.6 years. Even with a stipend, these students earn less than half what they could command working in the private sector, plus they are sidelined as their peers climb corporate ladders. Once graduates enter the job market, they find that industry positions often pay little more than what a bachelor’s holder with equivalent experience can command. Meanwhile, academic jobs are scarce.
But the calculus is very different for the most talented international students. For them, an American PhD opens access to the international labor market often paying four to six times their home-country salaries. The disparity in economic incentives might seem like a problem to be solved, but we should welcome it, because it attracts worldwide talent that directly benefits American workers and the US economy.

For decades, graduate programs have functioned as America’s primary talent-acquisition system. More than 300,000 international students currently pursue graduate science and engineering degrees here, and many of these students become the researchers and entrepreneurs who drive American innovation. Immigrants are responsible for one quarter of US patents. About one‑quarter of the country’s billion‑dollar startups—companies that employ hundreds of thousands of American workers—have at least one founder who came as an international student.
This system works through meritocratic competition. The best students worldwide choose US universities for their superior research facilities, faculty, and access to the world’s most dynamic economy. A four-year cap will undermine that advantage.
The rule change won’t stop international students from coming. But it will shift who comes. With four-year visa caps, many of the most talented students with the most options—precisely the ones America most wants to attract—will opt for countries that provide greater stability.
Nearly every PhD candidate hits the four-year mark before defending their dissertation, so international doctoral applicants choosing between Stanford, Imperial College London, and ETH Zurich will have to factor in a new risk to their education choice.
At Stanford, they will need a bureaucratic extension in year four—documentation, fees, months of adjudication and an uncertain outcome. At Imperial or ETH, they simply continue their work.
Research by economist Britta Glennon shows how visa restrictions play out: When America diminishes access, firms shift operations abroad. For every H-1B rejection, companies hired 0.4 employees in other countries. Microsoft’s 2007 decision to open a Vancouver development center, citing US immigration difficulties, was a preview. A four-year student visa cap would surely supercharge this shift.
When talented researchers choose universities abroad, they build networks, conduct research and launch careers there. The breakthroughs happen there. The patents get filed there. The economic value accrues there.
DHS has identified genuine fraud concerns in student visas: some students have maintained their visa status for fifteen to twenty-five years through serial enrollment. But a four-year cap would create the most burden precisely where fraud is least likely: in doctoral programs at top research universities. Meanwhile, the rule change would not increase the number of American students in PhD programs around the country: American students still won’t be applying, because the economics still won’t work for them.
What will change is America’s competitive position in the global race for talent and innovation—by pushing many next-generation breakthroughs from American labs to foreign ones.
Paola Sapienza is the J-P Conte Family Senior Fellow at the Hoover Institution, where she co-directs the J-P Conte Initiative on Immigration. She is a founding member of the Hoover Program on the Foundations for Economic Prosperity.

