DHS Goes Nuclear on High-Skilled Immigration with $100k fees on STEM OPT

Executive Summary

This past week, the Department of Homeland Security (DHS) published a Notice of Proposed Rule Making (NPRM) proposing charging universities $70,000 for each international student they recommend for Optional Practical Training (OPT), the program that lets F-1 students stay in the US to work in their field during or after completing their studies. Each subsequent recommendation of a candidate, including the 24-month STEM extension, would cost another $30,000, so a STEM graduate who uses all three years of OPT would cost their school $100,000.

DHS argues these fees will reduce fraud. Our analysis finds that they will more likely devastate the most important pipeline for global talent that the US has, compromising the American economy and jeopardizing our leadership in science and innovation. And, the fees ignore targeted fixes for fraud. DHS fails to analyze these risks or less burdensome alternatives.

This analysis reviews the best available evidence on how the fee would affect high-skilled immigration.

Key findings:

• The fee would sharply reduce STEM OPT and international student enrollment, which DHS assumes will not change.
  • A recent economics paper that estimates the elasticity of supply and demand for OPT suggests that the proposed fee would reduce OPT (and STEM OPT) participation by 23-99%.
  • In a recent survey of current and recently graduated F-1 students conducted by IFP, STEM students claimed they would be less likely to stay in the US as a result of fees like the one proposed. According to the survey, students are 40–80% less likely to try to stay under the proposed fees. Among graduates in STEM OPT-eligible fields only, the reduction is nearly identical, at 40–82%.
• The lost STEM talent will cost about $362 billion in annual GDP after ten years. That is the average estimate of a number of scenarios. After a decade, the high-skill STEM workforce will shrink by 6.3% and GDP will be 1.2% smaller. This would cost $2,680 per household per year on average.
• The fee will hurt the American students and workers DHS says it is protecting. International students’ tuition helps fund seats for domestic students. IFP estimates the fee would mean about 46,000 fewer Americans enrolling in bachelor’s and master’s programs each year.

Introduction

Since September 2025, the US government has attempted three times to institute $100,000 fees on the H-1B program.¹ Now, the push for $100,000 fees on high-skilled immigrants comes to OPT, threatening the on-ramp for a talent pool critical to American economic leadership.² DHS suggests that OPT cannot matter much to students because “only about 20 percent of F-1 nonimmigrants participated in OPT in 2024.” That figure is the result of dividing one year’s OPT participants by every international student enrolled, including those who are years from graduating and are not eligible for OPT. Many, if not most, of the students will go on to use OPT after they graduate. Inflating the denominator conceals the agency’s own statistics — according to SEVIS data, of F-1 graduates, 54% use OPT, including 64% of STEM graduates and 67% of PhD grads, showing that OPT is a central facet of international education in America.

DHS claims that the OPT fees will “combat fraud, strengthen the integrity of the immigration system, and protect U.S. workers,” citing an investigation that found more than 10,000 OPT participants working for “highly suspect employers,” including recently-formed companies and employers listing residential addresses as worksites for hundreds of students.³ While DHS has not concluded that these cases are all actual instances of fraud, there is nothing suggests this 3% of OPT participants couldn’t simply be barred from OPT by new regulatory text restricting suspect employers.

DHS’s analysis does not demonstrate that deterring fraud requires fees totalling $100,000. Instead, DHS says it chose the amount because it is “comparable to the fee for H-1B visas,” and that the magnitude will “closely align with the H-1B fees to reduce students and employers from using OPT to circumvent the H-1B visa program fees.”

DHS acknowledges another goal for the fee: reducing the OPT program. The expectation is that schools would “only recommend the most promising students,” leaving “a smaller pool of participants.” As DHS states in its press release and social media postings, the OPT tax is needed because “Optional Practical Training was never meant to be a back door into the American workforce.”⁴

In fact, OPT is not a back door into the American workforce, but the front door. DHS is treating OPT and the H-1B as substitutes, when in reality, they work in sequence. Eighty-two percent of international STEM graduates who stay to work in the United States secure OPT before getting an H-1B.⁵

In misconstruing the programs as substitutes, DHS suggests students “pursue appropriate channels to obtain employment in the United States other than the OPT, such as through the H-1B visa following graduation.” This is unworkable and ignores the H-1B calendar and how retention works in practice. The H-1B lottery is conducted in March, and new H-1B workers cannot start until October 1. To start work in October, a May graduate must have been registered by an employer that March and selected in the lottery. Otherwise, the student must wait for the following year’s lottery and start work the following October, 17 months after graduation. Even a student who lands an offer early, is registered, wins the lottery, and has an approved H-1B petition cannot work between graduation and October without OPT. For students graduating in the fall rather than the spring, the interruption in status would be even longer, and graduates may be less likely to consider returning to the United States. To prevent an interruption in status requiring leaving the country, a graduate would need to get OPT before their H-1B, increasing the total cost far above the proposed fee for either OPT or H-1B on its own.

If both this and DHS’s proposed H-1B fee go into effect, the cumulative fees will amount to $173,265 for 12-month OPT followed by an H-1B, or $203,265 for STEM OPT followed by an H-1B (the typical OPT path under DHS’s assumptions).

Elsewhere, DHS claims that the fees are “on par with the educational costs that foreign students incur for one or two years of education.” DHS offers no data to justify this comparison, and publicly available administrative data from the US government’s Integrated Postsecondary Education Data System (IPEDS) suggests DHS overestimates the tuition paid by international students who go on to use OPT.⁶

All of the benefits DHS is hoping to achieve by this rule, including fewer applicants for investigators to monitor and “job opportunities to U.S. workers who would have lost job opportunities to OPT students” depend on the rule reducing the OPT pool. In contradiction, DHS’s economic analysis assumes that schools will not stop or reduce recommending OPT and STEM OPT for graduates eligible for the STEM extension. Therefore, DHS’s own analysis suggests that the proposed fee won’t make the STEM population any easier to monitor, and will neither deter fraud among STEM graduates, nor help US STEM workers by reducing the number of international STEM graduates competing for jobs. Since DHS’s analysis assumes that STEM graduates will not be affected by the new fee, the fee’s success depends only on the effect on the non-STEM population. But DHS never considers the obvious implication of their logic: there is simply no need to charge the fee for individuals eligible for the STEM extension.

In any case, our analysis finds that both enrollment and OPT participation, STEM and otherwise, would fall under this rule. The next sections explain why DHS is wrong to assume there will be no change in STEM participation and demonstrate the implications for the American economy: hundreds of billions in lost GDP annually within a decade, and fewer American students able to afford a college education each year.

DHS’s analysis is deeply flawed

DHS’s proposed fee would substantially reduce OPT participation in two ways. First, fewer international students would get OPT because schools, students, and employers would not be willing and able to pay the fee.⁷ Second, given less opportunity to enter the US workforce, fewer international students would enroll in US schools in the first place.⁸ DHS fails to model the latter at all and fails to model the former for STEM graduates, asserting that any decline in OPT participation would be limited to non-STEM graduates. The best available evidence suggests participation would be reduced across the board.

DHS concedes it “cannot confidently predict the extent of any reduction in participation,” but its analysis assumes a 0% reduction among STEM graduates even in the scenario where non-STEM participation is maximally reduced. In this “low participation scenario,” schools stop recommending OPT for non-STEM graduates altogether but keep paying $100,000 for every student eligible for the STEM extension. In its “high scenario,” nothing changes from the status quo and OPT participation is unchanged. In the primary scenario, DHS simply takes the average of both scenarios. The only thing that changes across these scenarios is the number of non-STEM graduates participating in OPT; across all scenarios, DHS assumes the fee will cause no change to STEM graduates. DHS acknowledges that schools “could also experience declines in F-1 nonimmigrant student enrollment,” but does not account for it in its model of OPT participation.

DHS does report a lower number of STEM OPT participants in its low scenario than in its high scenario, but crucially, this reflects a different baseline (2024 participation levels vs. a three-year average with a fixed rate of extensions) and is unrelated to the effect of the fee, which it assumes will not affect STEM participation. DHS says it offers these two scenarios because the program has generally seen “fluctuations” in size. In other words, the lower STEM OPT number in the low scenario reflects a more conservative projection about the future of OPT, with or without a fee. In no scenario does DHS think that the fee will cause any decline in OPT participation by STEM graduates.

The argument DHS supplies is simply that $100,000 for three years is better value per year at roughly $33,000 per year than $70,000 for one year, so schools “may be more willing to pay” for STEM students. Immigration economist Michael Clemens estimates that universities profit about $13,700 per year on each international master’s student, and about $23,000 on each bachelor’s student.⁹ An international student pursuing a two-year STEM master’s degree therefore earns a school roughly $27,000. In Clemens’s words, $100,000 is “several times the margin a typical master’s program earns on an international student.”¹⁰ DHS assumes schools will pay nearly four times what each student they enroll brings.

DHS notes that schools may pass the fee “onto F-1 nonimmigrant students, all students, or employers.” We already have evidence for how employers respond to a $100,000 fee — after the September 2025 proclamation imposed $100,000 on new H-1B petitions for workers abroad, initial consular H-1B petitions fell 91%, by DHS’s own count.¹¹ And that was to secure a visa that could be renewed for at least six years, twice OPT’s maximum of three, providing a longer period over which to recoup the cost.

The best estimates of what this proposed fee would actually do come from a paper by immigration economist Michael Clemens, “Taxes on International Students: Demand and Supply Response for Post-Graduation Foreign Workers on Optional Practical Training.”¹² Clemens investigates how much OPT participation and international enrollment would decline depending on how much schools can pass on the costs to students and/or their employers. His elasticities are estimated across all fields and apply to STEM students.¹³ If students bear the cost, the share of new international graduates going onto OPT (i.e., the OPT participation rate) falls by 23% for bachelor’s graduates and 42% for master’s graduates, the lower bound of his analysis. If universities bear it, OPT participation declines by 46% for bachelor’s graduates and 99% for master’s graduates; if employers bear it, both bachelor’s and master’s graduate participation in OPT falls by 84%.

Survey data also projects fee-driven reductions in STEM OPT participation, and explicitly tests for any difference between STEM and non-STEM reductions. Between August 14 and October 5, 2026, IFP surveyed 734 current and former international students. Among STEM OPT extension-eligible students, the share who report they would be likely to try to stay in the United States after graduation falls from 80% to 15% when there are no restrictions on who can pay the fee. The size of that effect depends on how the question is asked but is robustly large. When we asked a differently worded question that also stated only schools can pay large OPT fees, the share reporting they would likely try to stay in the US falls to 48% instead. Respondents may have assumed that schools would be obligated to pay the fee or that if only one party is allowed to pay the fee, there would not be free riding. In short, this data suggests a 40–82% decline in the share of STEM students who predict they would try to stay in the US after graduating.

In addition to failing to consider an effect of fees on STEM OPT participation, DHS fails to consider an effect of fees on STEM enrollment. While it acknowledges schools “could also experience declines in F-1 nonimmigrant student enrollment,” DHS does not quantify this effect and only considers the decline in OPT participation of non-STEM students. But in addition to that decline, a fee that shrinks OPT would also be likely to reduce international student enrollment in the first place. Clemens estimates that fewer OPT opportunities would cut international enrollment by 3–31% at the bachelor’s level and 5–58% at the master’s level, depending on who bears the fee, on top of declines to participation from the enrolled population. Students in our survey suggested sharper declines: 70% of STEM-eligible respondents reported that they would not have enrolled in a US program had the proposed fees been in place. This reduced enrollment in American universities would further shrink the STEM workforce.

The proposed fee would compromise American innovation and economic growth

DHS knows declines in U.S. university enrollment by international students would be costly. Finalizing its duration-of-status rule in July, DHS wrote that it “acknowledges the economic contribution of nonimmigrant students and exchange visitors through tuition and regional spending, as well as contributions to research, innovation, and teaching,” adding that, “any reduction in enrollment has the potential to reduce the economic benefits that nonimmigrant students and exchange visitors offer.”¹⁴ In that rule, DHS argued that enrollment would not fall significantly. In this rule, DHS explicitly aims to achieve “a smaller pool of participants.” Yet its analysis fails to quantify the fee’s cost to innovation and the economy.

Those costs could be substantial. CSET studied the 550 recipients of four leading STEM awards: the Nobel Prizes in physics, chemistry, and medicine, the Lasker Award, the Turing Award, and MIT Technology Review’s Innovators Under 35 from 2014 to 2024. Of those who lived, worked, or studied in the United States, 47% were foreign-born. CSET traced how 184 of the foreign-born awardees first arrived in the US, finding that two-thirds entered as students. Based on this data, CSET warns that restricting OPT “could weaken the pathway through which most foreign-born U.S. awardees arrived.”¹⁵

OPT supplies much of the STEM workforce. One paper published by the National Academies of Sciences, Engineering, and Medicine using OPT administrative data estimates that foreign-born workers with US degrees make up 19% of America’s high-skill STEM workforce, 28% of those with master’s degrees, and 35% of those with doctorates and that OPT is a major contributor to that workforce.¹⁶ The paper models how changes in OPT or other policies affecting international students would affect the US STEM workforce, productivity growth, and therefore economic growth and provides quantitative estimates for the economic effects of declines in OPT.

Applying the model from that paper, we find that the magnitude of decline in STEM OPT participation described above, if sustained, would cost at least $110 billion and as much as $1.18 trillion to GDP each year after 10 years, or about $820 to $8,730 per household per year.¹⁷ This is a large range, but demonstrates substantial consequences to the American economy even on the conservative end, and reflects uncertainty in the precise scale of enrollment and participation effects. We consider four scenarios: the high and low end of Clemens’s paper on the elasticity of OPT and enrollment to new taxes and the high and low end of the survey results on OPT participation. Using Clemens’s estimates, we combine his findings on the drop in enrollment with his findings on the drop in OPT. It relies on the conservative assumption of the National Academies paper that half of those who no longer participate in OPT due to the fee will still find another route to a work visa. We also draw two scenarios from our survey data on the decline in STEM students who report they would stay. To be conservative, we do not consider the enrollment effect suggested by the survey at all. Within each of these four scenarios, we get a range of GDP effects from the range of TFP estimates that come from the National Academies paper. Across all these scenarios, the proposed fee would mean that annual GDP would be .36%-3.87% smaller. If that occurred today, that would be a cost of $110 billion to $1.18 trillion, or about $820 to $8,730 per household per year. Clemens’s elasticity estimates, which we assume to be more reliable than the survey, suggest a GDP reduction of at least $110 billion to $744 billion per year, for a central estimate of $362 billion per year — $2,680 per household.

The proposed fee would harm US students and workers

Beyond its effects on the US economy in aggregate, the fee would also impose more direct costs on American workers and students, failing to protect domestic workers and students as DHS intends. Evidence suggests that contrary to DHS’s assumptions:

1. The fee would not increase wages or employment for American STEM workers.
2. The fee would reduce the number of Americans able to enroll at colleges or universities.
3. The fee on OPT would make the H-1B program more likely to undercut US workers.

The fee would not increase wages or employment for American STEM workers

DHS’s proposed rule would not even benefit native-born US STEM workers on average, let alone native-born US workers writ large. Studies have explored how the increase in OPT participation following the creation of STEM OPT in 2008, and its expansion in 2016, have impacted US workers. Intuitively, one might expect that even if these expansions of OPT would help Americans on average, they could harm American STEM workers who compete most closely with OPT participants. Instead, the best available evidence suggests that increased OPT participation spurs firm creation, which raises demand for native-born workers with similar skillsets, and raises wages for native-born STEM workers. That means that even US workers with similar skills to foreign talent that would otherwise receive OPT would lose opportunities under the proposal.

Specifically, a recent study by Kwon, Lee, and Monras investigated the 2016 STEM OPT extension that increased the supply of OPT participants and found that demand for American workers with similar skills to OPT participants increased in the places where those OPT participants settled.¹⁸ The increased supply of high-skilled immigrants caused demand for native-born workers with similar skills to increase as the result of increased firm creation.

An older study by Demirci looking at the 2008 extension found that a larger number of international graduates raises wages for native-born STEM workers on average, despite reducing employment among recent native-born master’s graduates in the same field as the OPT participants.¹⁹ This suggests the best possible outcome of the DHS rule could be for one small subset of native-born workers most similar to OPT participants to benefit at the expense of most American workers, including most American STEM workers.

More likely, the Kwon, Lee, and Monras result suggests that the proposal by DHS to place large fees on OPT with the intent to reduce participation in the program will reduce firm creation, decrease demand for native-born workers with skillsets similar to those of OPT participants, and not even help the American workers most similar to OPT participants on net.

The fee will reduce the number of Americans able to enroll at colleges or universities

The proposal stands to reduce, not expand, opportunities for Americans to get a college education. International students pay higher net tuition, and that revenue helps fund seats for domestic students. One analysis shows that at the graduate level, every 10 additional international students bring 8 more American students.²⁰ DHS allows schools to pass the cost of the fee on to “all students,” which would exacerbate the effect.

To estimate how much the rule would reduce opportunities for American students, we developed a model that treats the fee as a wedge between what international students pay for a degree with OPT and what schools keep. How much the burden of the fee is split between the students and the school depends on how sensitive students and schools are to price. Either way, Americans lose. If schools pass the fee on to students, fewer international students enroll, leaving less tuition to subsidize seats for domestic students. If schools pay the fee, they collect less from each international student who still comes, shrinking that subsidy. Under the best available estimates of the elasticities, schools absorb a little over one-quarter of the fee and pass the rest on.²¹ This translates into approximately 46,000 fewer American students enrolling in bachelor’s and master’s programs each year under DHS’s proposal. Appendix A describes the model in more detail.

The fee will make the H-1B program more likely to undercut US workers

The fee would make the H-1B program more likely to undercut US workers, advantage IT outsourcing companies at the expense of employers key to critical fields, and allow an increased share of H-1Bs to go to workers sponsored directly from abroad and who are paid notably less than Americans with similar levels of experience and education.

A $70,000-100,000 fee on graduates of US schools would shift H-1B demand toward employers not recruiting from US schools. Those employers tend to hire lower-paid workers, who are more likely to displace Americans. In fiscal year 2024, workers on F-1 student status received 45% of new cap-subject H-1Bs, and another 43% went to people living abroad with no US status.²² The former earned 10% more than the latter.²³

Immigration economist Chad Sparber estimates that if the OPT pipeline into H-1B were restricted so H-1B visas that go to F-1 students instead went proportionally to other applicants, total H-1B wages would fall 1.8-4.5%.²⁴ He warns that “efforts to curtail the F-1 OPT pipeline into the H-1B would reduce employment of some lower wage workers, but they would also increase the share of H-1Bs going to very low wage Indian employees.”²⁵ The Economic Innovation Group finds a similar pattern, concluding that “the OPT fee further rigs the H-1B lottery to favor outsourcers.”²⁶ This tilt toward companies bringing in workers from abroad is particularly concerning given that H-1B workers at IT outsourcing firms earn $9,000 less a year than H-1Bs who entered the US as international students, despite the outsourced workers being six years older on average.²⁷ Moreover, those workers are far more likely to undercut the wages of similar Americans.²⁸

In short, raising the cost of OPT would shift H-1Bs toward the lower-paid hiring the administration says it wants to curb.

DHS does not adequately consider whether less burdensome alternatives could accomplish the agency’s anti-fraud goals

DHS is required to consider less burdensome alternatives when engaging in notice-and-comment rulemaking and explain why it decided against adopting them. The agency’s proposal did not seriously engage with potential alternatives that would be less burdensome to the American economy, employers, universities, or workers while still accomplishing the proposal’s goals of rooting out fraud in the OPT program and deterring employers from substituting lower-paid foreign workers for native workers.

Alternative fee structures

The agency could have proposed a number of alternative fee structures that would impose fewer costs on OPT participation while still effectively targeting suspect employers and incentivizing greater vigilance by universities recommending students for OPT.

The agency failed to consider using its statutory fee-setting authority

ICE regularly engages in rulemaking to set fees in accordance with its authorities as granted by Congress through both 8 U.S.C. 1372 for fees charged per student and exchange visitor and 8 U.S.C. 1356(m) for fees charged per school. The former, enacted in 1996, required the creation of the Student and Exchange Visitor Program (SEVP) and the latter, enacted in 1988, required adjudication fees to be paid into a self-sustaining Immigration Examinations Fee Account funded by user fees, including fees to cover certification processes for higher education institutions seeking authority to enroll international students. These statutes enable DHS to revise fees charged per student and exchange visitor for what is now an I-901 filing (1372(g)(2)) and fees paid by SEVP-certified higher education institutions for I-17 filing and related site visits (1356(m)) to fully recover the costs of administering SEVP. DHS (and the legacy Immigration and Naturalization Service (INS)) finalized such fee rules in 2001, 2002, 2004, 2008, and 2019 to adjust fees in accordance with the costs of running the program. Nothing is stopping the agency from doing the same through this rule, yet it has inexplicably chosen not to. The agency has historically used Activity-Based Costing (ABC) to calculate the funding needs of the SEVP program, setting fees accordingly to recover those costs, another practice it abandons in this proposal.

In 2019, during the first Trump administration, DHS finalized an SEVP fee rule that increased or established eight fees to fund the program’s growing administrative costs. DHS explicitly cited growing funding needs to combat fraud and ensure program integrity as the basis for its rule:

“… the new fees are based on estimates of funding needed to maintain and enhance SEVP’s capability to achieve programmatic goals associated with its statutory mandate, including supporting national security and countering immigration fraud through the continued development and implementation of critical system and programmatic enhancements.”²⁹

If the agency believes that the resources needed to prevent fraud and abuse of the OPT program have grown non-linearly with the program’s size, it should again be able to show its work through its ABC methodology. By statute, the agency can raise as much revenue as it needs to faithfully and effectively administer the program through fees if it can document those costs. It has not attempted to provide that documentation in the proposed rule.

The agency failed to identify even obvious alternatives to its preferred $100,000 total in fees, suggesting the rulemaking might have been undertaken with a foregone conclusion

The justification DHS provides for the specific level of its proposed fees is extraordinarily thin. While the agency notes that it considered higher fees (such as also applying the $70,000 fee requirement to students participating in OPT a second time after completing an additional degree), it does not document considering any other fee structures, which would likely yield lower fee levels. There are two alternative, less burdensome fee structures in particular that, while they would still impose unacceptably high costs on the American economy and the country’s ability to attract and retain international talent, DHS failed to consider, despite these alternatives appearing to align with the agency’s stated goals and reasoning.

The first alternative would be a fee equal to the value of the so-called “FICA exemption.” DHS’s proposal noted that many OPT participants are not subject to Federal Insurance Contributions Act (FICA) taxes, a longstanding problem in the OPT program.³⁰ DHS argues that the discount provided to employers of OPT workers by the FICA exemption creates a financial incentive to hire international students over native graduates. Along with fraud and abuse, the alleged negative labor market effects of OPT, in part due to the FICA exemption, were one of DHS’s primary justifications for the rule (even though many OPT participants are not eligible for the FICA exemption in the first instance).³¹

Yet the administration failed to consider a less burdensome alternative: an OPT fee equal to the average foregone tax revenue from the FICA exemption. The agency could set an OPT fee equal to 15.3% of the typical participant’s earnings, equivalent to what the employee and employer would owe in FICA taxes if the exemption did not exist. If the agency believes that its proposed fees create an incentive for universities to be more scrupulous in recommending students for participation in OPT, the same should be true of a fee equivalent to the FICA exemption. Setting fees this way would require some exemptions since not all F-1 students on OPT and a STEM OPT extension are eligible for the exemption from FICA payments.³² Such a fee, where it applies, would be substantially lower than what DHS has proposed, putting less of a financial burden on program participants and the broader economy.

Second, the agency failed to consider a fee in line with the per-year cost of its proposed H-1B fee. The agency claims that its proposed OPT fees “closely align with the H-1B fees” from last year’s presidential proclamation that created a $100,000 charge on H-1B petitions on behalf of beneficiaries abroad. However, this comparison is illogical. The presidential proclamation fee, as well as DHS’s proposed $103,265 fee on cap-subject H-1B petitions, apply only to initial petitions, not renewals. Under the statute, H-1B status lasts up to three years and can be renewed for a total of six, and only more after that if the beneficiary has an approved green card petition. Students can participate in the OPT program for a maximum of three years. This makes the per-year cost of the proposed OPT fee twice the per-year cost incurred by students sponsored for an H-1B. Halving the proposed OPT fees to actually be in line with the per-year cost of DHS’s proposed H-1B fee is clearly a less burdensome (though still quite costly and harmful) alternative, but one that the agency did not consider.

Other less burdensome restrictions besides fee structures

In lieu of massive fee increases, DHS could have considered changes to OPT eligibility to exclude employer types or employment arrangements the agency believes are historically associated with fraud or the undercutting of American workers. For example, DHS did not consider restrictions on OPT participation for employment at IT outsourcing or staffing firms, H-1B-dependent employers, or employers who place program participants at third-party worksites to work side-by-side with employees of client firms. Further, using the agency’s internal data on fraud cases, DHS could have developed fraud pattern indicators and proposed barring employers from participation if they meet such criteria. The agency considered none of these alternatives, each of which would have better targeted the abuse motivating the rulemaking effort without imposing massive costs on rule-following participants. Transparent eligibility criteria would minimize the enforcement burden on the agency and immediately make it clear to universities, employers, and students who is eligible to participate in OPT and who is not. Further, to the extent relevant to what DHS has discovered in its internal abuse analysis, DHS could consider making changes to OPT eligibility to bar graduates of unaccredited schools. Under current regulations, STEM OPT extensions can only be granted if the issuing school is accredited³³ but no such requirement attaches to either CPT, pre-completion OPT, or 12-month OPT. Moreover, both the One Big Beautiful Bill Act (2025) and the Secure America Act (2026) appropriated tens of billions of dollars to ICE that could be used to scale up staffing levels in the SEVP program or bolster anti-fraud and national security efforts by the agency. Between the two bills, Congress appropriated roughly $60 billion to ICE, including for staffing expansion and infrastructure upgrades. The agency has not explained why such resources could not be used for fraud detection and deterrence in the SEVP program, or why such appropriations are not sufficient to perform these functions.

Furthermore, DHS could address fraud concerns within the OPT program by applying the safeguards that already govern the STEM OPT extension process to all OPT employment.³⁴ STEM extension employers are currently required to:

• Be enrolled in E-Verify and possess a valid Employer Identification Number (EIN) to help screen out illegitimate employers; and
• Provide detailed and regularly updated employment and training information via the submission of Form I-983 Training Plans to the sponsoring university’s DSO. This includes detailed information concerning (but not limited to) physical working location, salary, employment duties/responsibilities, employer oversight, and performance evaluation process.

In the NPRM, DHS concludes that applying the STEM OPT requirements to all OPT “would [not] be sufficient as an initial response to the overarching OPT program integrity concerns,” given their position that such procedural and documentation requirements “do not eliminate the risk of fraud, abuse, or noncompliance and may require substantial time and resources to implement, monitor, and enforce effectively.”

Elsewhere in the NPRM, DHS assesses existing STEM OPT safeguards as insufficient because “the employers for STEM OPT are not required to submit any business documents or supporting documents to DHS, such as tax records or proof of business. Furthermore, DHS does not generally verify the employer attestations unless and until there is a site visit.”

It may be that, standing alone, imposing the STEM OPT requirements on all of OPT is not sufficient. However, it does not follow that the available regulatory safeguards should be ignored, especially if they are accompanied by new requirements. DHS could propose binding regulatory text that provides affirmative obligations and restrictions on employers hiring OPT participants. By way of example, the following types of revisions could be considered as new proactive guardrails for integrity in OPT and STEM OPT, none of which create new bureaucracy or operational concerns and instead unequivocally communicate new, easy-to-understand standards to the regulated community:

• There could be an employer qualification regulation that applies equally to OPT and STEM OPT extensions.
• Establish in regulation that all OPT and STEM OPT employers must also have a DUNS number from Dun & Bradstreet. Using this universal numbering system would facilitate both SEVP and USCIS (through the agency’s existing Validation Instrument for Business Enterprises (VIBE) program) engagement in quality assurance, with the understanding that if problems occur, individual employers will be given an opportunity to provide further information.
• Establish in regulation that sole-proprietor self-employment is not permitted, while allowing entrepreneurial activities through recognized accelerators and incubators that are themselves well-established employers.
• Establish in regulation that no employer founded less than one year prior to signing the Form I-983 may hire an OPT participant, while allowing entrepreneurial activities through recognized accelerators and incubators.
• Establish in regulation that no employer whose place of business is a private residence may hire an OPT participant.

By not actively considering such measures, ICE either completely misunderstands or ignores the role of agency rulemaking in enforcement and compliance, perhaps in part because the Student and Exchange Visitor Program resides in an agency whose primary mission is law enforcement and not policymaking. One of the key purposes of rulemaking in federal agencies is to adopt general principles that can be applied without reexamination and are binding and enforceable on the regulated community.³⁵ Applying the concept of universal employer qualifications for both the STEM extension and OPT generally can proactively enhance standards and mandate compliant behaviors, such as with the type of exemplars provided above – and those integrity improvements avoid steps that cannot be operationalized to scale, like submission and review of tax records and proof of business documents or comprehensive site visits.

Conclusion

The best available evidence suggests that the proposed fee would cut both OPT and enrollment, cost hundreds of billions of dollars a year, and leave tens of thousands of American students with less access to quality higher education. DHS must evaluate all of these costs against the benefits it claims, and establish why it can’t accomplish its goals in a less destructive way.

Appendix

This appendix sets up a simple model in which prospective international students demand a bundle composed of a U.S. degree and a period of OPT. The OPT fee is treated as a tax wedge between what students pay and what universities effectively retain, with the burden split between the demand and supply sides according to their elasticities. Since international students subsidize domestic students, domestic enrollment is affected in two ways. First, a higher wedge reduces the number of international students who enroll, which directly lowers the revenue that can be used to support domestic students. Second, among the students who still enroll, universities absorb part of the wedge, which indirectly reduces the revenue available to cross-subsidize American students. In that sense, the fee crowds out Americans by reducing universities’ revenue from international students along both the extensive and intensive margins.

Model

This section provides an overview of the toy model used to highlight the dynamics created by the fee. While the model is simple, it makes clear how equilibrium quantities and prices depend on elasticities. Because Americans are crowded out through both extensive and intensive margins, the number of domestic students falls meaningfully under a wide range of elasticity assumptions. The model uses a constant-elasticity approach rather than a local linear approximation because the change being studied is large relative to the baseline yearly tuition levels faced by many programs.

Let Q₀ denote baseline enrollment, P₀ the baseline price of the degree–OPT bundle, P^c₁ the post-policy price paid by students, and P^u₁ the post-policy net price retained by universities. The policy wedge is τ, so

$$P^c_1 – P^u_1 = \tau.$$

Demand and supply are written in calibrated constant-elasticity form:

$$Q^D = Q_0 \left( \frac{P^c}{P_0} \right)^{-\eta_D}, \qquad Q^S = Q_0 \left( \frac{P^u}{P_0} \right)^{\epsilon_S},$$

where η_D > 0 is the absolute value of the demand elasticity and ϵ_S > 0 is the supply elasticity.

If we define the normalized equilibrium quantity q ≡ Q₁/Q₀, then the exact post-policy equilibrium must satisfy

$$q^{-1/\eta_D} – q^{1/\epsilon_S} = \frac{\tau}{P_0}.$$

That equation is the core of the exercise. Once q is known, the equilibrium quantity and both prices follow immediately. “Derivation of the Equilibrium Condition” below shows where this condition comes from, and “Numerical Solution and Degree-Level Aggregation” explains how it is solved numerically in the degree-level exercise. The resulting decline in equilibrium international enrollment can then be translated into the number of American students crowded out using the estimated revenue loss faced by universities.

Results

Table A1 reports the degree-level outcomes under the demand and supply elasticities estimated by Clemens (2026) and the fee schedule in the NPRM. Baseline quantity is the annualized pre-policy OPT count in that degree group, which can be read as average yearly entry. Baseline price is the average yearly tuition in that group. Yearly wedge is the annualized fee used in the calculation. Equilibrium quantity and quantity change show how enrollment by international students responds after the fee is imposed. Equilibrium price is the new international student-facing price, while paid by university, P^u, is the amount of the wedge absorbed by the institution rather than passed through to the student.

The demand elasticities are Clemens’s size-weighted instrumental-variables estimates of how the nationwide count of international students responds to the price of the degree: −0.60 for master’s and −0.49 for bachelor’s (Clemens 2026, Table 1, column 3). The supply elasticities are his estimates of how universities’ supply of seats for OPT graduates responds to the price they receive: 0.56 for master’s and 0.63 for bachelor’s (Table 5, panel b). The NPRM charges $70,000 for initial OPT and $30,000 for any subsequent OPT, including the 24-month STEM extension, so the total wedge is $100,000 for STEM programs and $70,000 for all others.

The crowd-out rows split the effect on Americans into two pieces. Ext. captures the extensive margin: the number of American students crowded out due to the revenue lost from lower international student enrollment. Int. captures the intensive margin, meaning the number of American students crowded out due to the revenue lost when universities absorb part of the wedge for international students who still enroll. Total adds those two margins for that degree column, and the Total column sums the degree columns. The intersection of the Total row and the Total column gives the number of American students estimated to be crowded out due to the wedge.

Table A1: Degree-level outcomes under Clemens (2026) elasticities and the NPRM fee schedule

	Master’s	Bachelor’s	Total
Demand elasticity	−0.60	−0.49
Supply elasticity	0.56	0.63
Baseline Q	73,095.6	35,027.8	108,123.4
Baseline P	32,166	31,451
Yr. wedge	45,710	21,308
Eq. Q	46,569.4	28,509.4	75,078.9
ΔQ	26,526.2	6,518.4	33,044.5
Eq. P	65,347	46,436
Paid by univ.	12,530	6,323
Ext.	21,220.9	5,214.7	26,435.6
Int.	14,512.1	4,585.2	19,097.3
Total	35,733.0	9,799.9	45,532.9

Note: Yr. wedge averages the annualized $100,000 STEM and $70,000 non-STEM fees, weighted by equilibrium quantity. Totals may not sum because of rounding.

One useful column for intuition is the master’s column. Baseline annual quantity of international students is about 73,096, the baseline yearly tuition is about $32,166, and the annualized wedge averages $45,710 across STEM and non-STEM programs. After the fee, equilibrium quantity of international students falls to about 46,569 and the student-facing price rises to roughly $65,347. Universities still absorb about $12,530 of the wedge per remaining student. In that column, the extensive-margin crowd-out effect is about 21,221, while the intensive-margin effect is about 14,512. The two channels are of comparable size: fewer students arrive, and the students who remain generate less net revenue for the university.

Summed across the two degrees, about 45,500 fewer American students enroll each year. Universities absorb about 28 percent of the wedge and pass the rest through to students. Table A2 reports the same total for other elasticity pairs.

Table A2: Total Americans out across bachelor’s and master’s programs, by elasticity pair

Demand elasticity	Supply 0.50	Supply 0.80	Supply 0.99	Clemens supply
−0.30	34,007.3	28,972.4	27,066.9	32,391.2
Clemens	47,004.7	42,328.1	40,431.7	45,532.9

Note: Clemens’s demand elasticities are −0.60 for master’s and −0.49 for bachelor’s; his supply elasticities are 0.56 and 0.63.

Across the table, the number of Americans crowded out varies materially with the elasticity assumptions: it rises as demand becomes more elastic and falls as supply becomes more elastic. University revenue is hit on two fronts at once. When supply is very inelastic, universities absorb a large share of the wedge on the students who remain. When supply is more elastic, the burden shifts more toward students and the quantity response grows. Those two adjustments move in opposite directions, and over this range the intensive margin shrinks faster than the extensive margin grows.

Table A2 does not include supply elasticities below 0.5. At 0.10 and 0.01, universities would absorb about $24,000 and $28,000 a year on each remaining master’s student, well above the $13,735 a year that Clemens (2026, footnote 23) estimates universities net on an international master’s student, and Clemens argues that a university would not pay a fee larger than that margin. Across the remaining pairs, the total ranges from about 27,000 to 47,000.

Doctorate students are treated differently in the crowd-out accounting. Following Clemens (2026), they are excluded, because international doctoral education is financed more by universities and research grants than by tuition and so provides little cross-subsidy to domestic students.

Derivation of the Equilibrium Condition

The derivation is short. Start from

$$Q^D = Q_0 \left( \frac{P^c}{P_0} \right)^{-\eta_D}, \qquad Q^S = Q_0 \left( \frac{P^u}{P_0} \right)^{\epsilon_S}, \qquad P^c – P^u = \tau.$$

At equilibrium, Q^D = Q^S = Q₁. Define q ≡ Q₁/Q₀. The demand equation implies

$$P^c_1 = P_0 q^{-1/\eta_D},$$

and the supply equation implies

$$P^u_1 = P_0 q^{1/\epsilon_S}.$$

Substituting both expressions into the wedge condition gives

$$P_0 q^{-1/\eta_D} – P_0 q^{1/\epsilon_S} = \tau.$$

Dividing through by P₀ yields

$$q^{-1/\eta_D} – q^{1/\epsilon_S} = \frac{\tau}{P_0}.$$

That expression shows directly why the wedge can be split between students and universities. If demand is very inelastic, the student-facing price rises sharply. If supply is very inelastic, the university absorbs more of the wedge through a lower retained net price.

Numerical Solution and Degree-Level Aggregation

The numerical exercise applies the model within degree-specific tuition bins rather than to one pooled market. The program-level data are SEVIS records of OPT participants in fiscal years 2018 through 2022, aggregated by school, degree level, and STEM designation, with each cell’s count of OPT records and average annual tuition and fees.³⁶ The steps are as follows.

First, tuition is binned in $5,000 increments separately for master’s and bachelor’s programs, and separately for STEM and non-STEM programs because the fee differs by STEM status. Within each bin, tuition is averaged using the observed program counts as weights. The observed counts are pooled over multiple fiscal years, so the bin-level totals are divided by five before solving the equilibrium in order to convert them into annualized baseline quantities. The wedge is annualized as well: the calculation divides the total wedge, $100,000 for STEM programs and $70,000 for others, by two for master’s programs and by four for bachelor’s programs.

Second, for each degree–tuition bin and each elasticity pair, the model solves for the normalized equilibrium quantity q. Because the wedge is large relative to many baseline tuition values, the calculation uses the exact nonlinear condition rather than a first-order pass-through approximation. Numerically, the solution is found on the interval bounded above by q = 1 and below by the full-pass-through quantity ratio.

Third, the bin-level solutions are aggregated back to the degree level. Equilibrium quantities are summed across bins. Equilibrium prices are averaged across bins using equilibrium quantities as weights. The resulting degree columns therefore combine within-bin price heterogeneity with a common elasticity pair.

Finally, the crowd-out metrics are built from the degree-level outputs. The extensive-margin component scales the drop in equilibrium quantity. The intensive-margin component scales the revenue loss on the students who still enroll when universities absorb part of the wedge. Both are scaled by 0.8 American students per international student, the ratio Shih (2017) finds at the graduate level, with the revenue loss converted to international-student equivalents at baseline tuition:

$$\text{Ext.} = 0.8\,(Q_0 – Q_1), \qquad \text{Int.} = 0.8\,Q_1\,\frac{P_0 – P^u_1}{P_0}.$$