33 private schools have announced that they are closing permanently at least in part due to the COVID-19 economic downturn. That is up from 26 on our special Monday, May 18 update.

We have adjusted how we collect enrollment data, using a combination of media reports, Private School Review, and direct contact with schools. 5,690 students attend the schools that are closing, up from 5,217 on Monday. Were all of these students to go to public schools, and had none been part of publicly connected school choice such as voucher programs or scholarship tax‐​credits, the new cost to the public purse would be roughly $88,000,000 ($15,424 per student multiplied by 5,690).

As always, the list is expected to grow as schools learn more about the impact of the economic downturn on enrollment and income for the coming school year. We will ordinarily post an update on Cato’s blog every Friday, but if the list reaches 100 schools we may transition to an online, searchable format. You can contact CEF director Neal McCluskey if you need more current numbers, if you know of permanent closures not on the list, or if you believe schools have been listed by mistake. We also welcome suggestions for improving the list.

It’s difficult to outdo the crypto community when it comes to making bold quantitative claims that, stripped out of context, mislead the incautious. But Financial Crimes Enforcement Network (FinCEN) Director Kenneth Blanco recently came close.

In remarks last week to the annual (and, alas, virtual) Consensus conference for crypto professionals and enthusiasts, Blanco declared that, “since 2013, FinCEN has received nearly 70,000 Suspicious Activity Reports (SARs) involving virtual currency exploitation.” That impressive figure was bound to get attention—and it did. The speech is also likely to reinforce the widespread view that cryptocurrency is a hotbed of financial crime. But Blanco omitted to say that, in 2019 alone, financial institutions filed more than 2.3 million SARs regarding all sorts of transactions, and that, according to FinCEN’s own statistics, virtual currency SARs make up just 0.56 percent of all such reports filed since 2014.[1]

Whom to believe—Blanco, or his agency’s numbers? Were FinCEN an obscure or unimportant agency, the answer might not matter very much. But as the U.S. Treasury Department’s illicit finance watchdog, FinCEN is a crucial enforcer of financial regulations—ones which, according to a 2018 survey, community bankers consider the costliest to comply with. Yet, despite FinCEN’s significance, the SAR database (FinCEN’s main resource for law enforcement) is bloated and opaque, and the usefulness of its contents impossible for outsiders to evaluate. Blanco ostensibly believes that crypto is a source of growing mischief. But absent major improvements to FinCEN’s database and reporting, that hunch will remain unverifiable.

To be sure, Blanco’s agency is relatively underresourced despite its leading role in writing and enforcing the Bank Secrecy Act’s many rules. Its 333 employees and $118 million budget look paltry in comparison with another Treasury agency, the Office of the Comptroller of the Currency (3,699 employees and $1.09 billion), and independent financial regulators such as the Consumer Financial Protection Bureau (1,465 and $510 million) and the Securities and Exchange Commission (4,350 and $2.5 billion). Although FinCEN delegates BSA‐​related supervision to the primary regulators of different financial institutions, only it has overall authority for enforcement of the statute.

Perhaps owing to its limited resources, FinCEN has tended to deputize financial institutions to perform the oversight that other regulators undertake directly. For example, BSA regulations require banks and others to collect, verify, and maintain an up‐​to‐​date record of their customers’ personal information, such as their name, address, date of birth, and taxpayer identification number. They must also develop due diligence policies designed to subject risky customers to additional scrutiny. Since May 2018, FinCEN has also required financial firms to collect beneficial ownership information from their corporate accountholders, so that they might include it in their SARs.

Shifting the bulk of the BSA’s burden to the private sector serves to conceal its weight. But the regulations are onerous, whatever view one holds of their impact on financial crime. According to FinCEN’s own conservative estimate, over the next decade financial institutions may spend as much as $1.5 billion just to comply with its May 2018 rule. That figure only reflects direct compliance costs: staff training, longer account opening processes, and so on. The indirect costs of FinCEN’s regulations may, however, be even greater. For instance, their adverse impact on banks’ willingness to take on customers FinCEN might deem risky is substantial. Residents of border states who hold foreign passports, conduct cross‐​border business, and deal in cash are particularly affected. While these residents may fit the BSA’s archetype of a money launderer, most are not criminals. But banks may shun them all because serving them isn’t worth the extra due diligence cost.

Banks’ strong desire to avoid considerable penalties and reputational damage makes them eager to avoid falling foul of BSA regulations. But this precautionary zeal comes at the cost, not only of lost business, but of resources that might be employed much more productively elsewhere. One example of waste is so‐​called “defensive reporting”: among the 2.3 million SARs filed in 2019, more than 11 percent (262,987) bore the tag “Other Other Suspicious Activity,” hinting that the reports were filed only as a precaution.

Such precaution may be warranted in certain cases. Perhaps a few defensive SARs have even helped to bring financial criminals to justice in the past. It’s difficult to know because FinCEN doesn’t make such information publicly available. Still, many experts point out the alarming rate of “false positive” reports, a finding that should concern advocates of greater financial inclusion because SARs are a decisive factor in banks’ decision whether to close customer accounts. And while some of the SARs with imprecise tags like “Other Other” include additional, more specific classifiers, many don’t. Again, one can’t know the exact proportion of each because FinCEN’s public database doesn’t list information on individual reports.

In forums public and private, FinCEN officials often state that criminal financial activity is on the rise, that they need all the information they can get from financial institutions, and that every single report counts. But most of the time, FinCEN’s word is all the supporting evidence outsiders can hope for. That must change. If, as Director Blanco has repeatedly suggested, financial malefactors are warming up to cryptocurrency, FinCEN should be the first to take this trend seriously by listing “virtual currency” (the agency’s preferred term) as a discrete SAR category. FinCEN should also take a leaf out of the CFPB’s book and release regular reports about suspicious activity trends across the financial system, as the Bureau does for consumer finance trends. Despite mounting SARs, FinCEN has published very little in the last 15 years on the growth of new payments instruments. Yet greater provision of information and data may help not only to alert market participants to rising threats, but to clear up the cloud of suspicion that presently hovers over the crypto industry.

If Blanco is serious about demonstrating his agency’s efficiency and value, he must be transparent with policymakers and the public. Any other approach is likely to hobble beneficial financial activity, without deterring those who seek to undermine our security.

*******

[1] The SARs database on FinCEN’s website yields 12,533,814 reports since 2014. Data for 2013 are unavailable.

[Cross‐​posted from Alt​-​M​.org]

I've been reading, writing, and tweeting about the Michael Flynn prosecution quite a bit lately, and I've been getting significant feedback from people who strongly dislike Flynn and think he's about to get away with committing a serious crime due to sinister political machinations behind the scenes. As explained in this post, I take no strong position on Flynn's character or whether he did or did not commit any crimes—most Americans have, so he'd be in good company.  But I think many in the "lock him up" crowd are making two fundamental errors that are in urgent need of correction. 

Those errors, which turn out to be inextricably intertwined, are: (1) Flynn is plainly guilty of lying to FBI agents, so the attorney general's motives in dropping the case against him must necessarily be suspect; and (2) given the character of the defendant and the alleged crime, the Flynn case must necessarily be a poor vehicle for spotlighting the pernicious role of coercive plea bargaining in our criminal justice system—as Pulitzer-Prize-winning columnist George Will did yesterday. As explained below, these errors are momentous, and they have been embraced by many influential bloggers, law professors, and other opinion leaders who help shape public perceptions about the legal system. I hope some of them will see this post and read it in the spirit of good will with which it is offered. 

The first point—that the decision to drop the Flynn prosecution must necessarily have been made for crass political reasons—appears to be based on an incomplete understanding of the evolving fact record together with uncharacteristic confidence in the integrity of a coerced plea. The essential facts are these:

Michael Flynn was charged with making false statements to FBI agents during a January 24, 2017 interview at his White House office regarding conversations Flynn had had in the preceding weeks with Russian Ambassador Sergey Kislyak. (Flynn was also charged with making false statements in connection with certain filings under the Foreign Agents Registration Act, but I'll put that to one side for now; among other things, Flynn never pleaded guilty to those crimes, and DOJ's attempt to prosecute another high-profile figure, Gregory Craig, for similar violations failed miserably and would likely have failed against Flynn as well.)

Represented by lawyers from the large D.C. law firm Covington & Burling (who it now appears had a serious conflict of interest), Flynn initially asserted his innocence. Crucially, Flynn's attorneys pressed the prosecutors from the Special Counsel's Office to turn over the memorandum of interview or "302" prepared by the agents who questioned him regarding his communications with Kislyak. Throughout November 2017, SCO prosecutors repeatedly rebuffed those requests even as they ratcheted up the pressure on Flynn to plead guilty in exchange for a recommendation of no jail time. But Flynn continued to maintain his innocence, and his attorneys continued pressing for production of the 302 and other discovery—which the government continued to withhold.

It has been reported, credibly in my judgment, that the stalemate was brought to a head when the SCO leaked to certain reporters that a guilty plea from Flynn would ensure that Flynn's son, who was under investigation as Flynn senior's business partner (and also happened to be the father of Flynn senior's four-month-old grandchild) would not be prosecuted. This is the sort of despicable tactic one associates with tyrants and dictators; but to our infinite discredit, it appears to have become a routine feature of American prosecutions as well.

In any event, we do know that something caused Flynn to suddenly change his mind in late November of 2017 and agree to plead guilty to a single charge of lying to FBI agents. He signed a "statement of the offense" to that effect (along with the alleged FARA violations) on November 30, and appeared in court to enter his guilty plea—on the charge of lying to FBI agents only—the next day, December 1, 2017.

Notably, despite having pleaded guilty more than two years ago, Flynn has not yet been formally convicted of that crime because the conviction does not technically happen until the sentence is pronounced. And for various reasons, that has not happened yet.

In a post last month, we raised concerns about the unforeseen and underappreciated costs of expanding export controls on U.S. technology. Either those concerns fell on deaf ears or the administration did its due diligence and determined that the expected benefits outweigh the expected costs because—earlier this week—the Commerce Department published new rules further restricting Huawei’s access to U.S. technology.

U.S. exports to Huawei have been tightly controlled since Huawei and its affiliates were placed on the Entity List in 2019 for national security reasons. However, because of the design of U.S. export regulations and the nature of technology supply chains, Huawei and its affiliates were still able to import semiconductors from foreign producers that use U.S. chipmaking equipment and software. The new rules are intended to close this loophole and completely cut off Huawei from U.S. technology.

Explaining the purpose of those new rules, Commerce Secretary Wilbur Ross—betraying naïve expectations that Huawei would have just thrown in the towel and shut down its operations after last year’s U.S. sanctions—offered:

Despite the Entity List actions the Department took last year, Huawei and its foreign affiliates have stepped‐​up efforts to undermine these national security‐​based restrictions through an indigenization effort. However, that effort is still dependent on U.S. technologies. This is not how a responsible global corporate citizen behaves. We must amend our rules exploited by Huawei and HiSilicon and prevent U.S. technologies from enabling malign activities contrary to U.S. national security and foreign policy interests.

Although we have been skeptical from the start that this is the right way to proceed with China, the die most definitely has been cast and the technology trade war is moving ahead at full speed. Of course, the U.S. government (many in the Trump administration and many in the Congress) has its reasons (some factual; some presumptive; some political) for this course of action. So, instead of rehashing concerns already raised, we offer (in convenient bullet point fashion) the most relevant facts and assumptions culminating in the current policy, as well as the expected benefits and likely costs of that policy. Unfortunately, the list of likely costs is long.

Facts

• The U.S. government sees the Chinese government as a bad actor.
• The U.S. government sees Huawei as an adjunct of the Chinese government.
• The U.S. government sees Huawei as the leader in 5G technology.
• The U.S. government sees Huawei’s leadership in 5G technology as a threat to U.S. national security.
• The U.S. government sees a vulnerability to Huawei’s 5G leadership in Huawei’s dependence on U.S. semiconductors and semiconductor technology.
• The U.S. government seeks to exploit that vulnerability by depriving Huawei of the technology it needs to continue to dominate 5G.

Assumptions

• Targeting Huawei with export controls and entity list restrictions to deprive it of needed inputs will slow or stop Huawei’s progress.
• U.S. sanctions on Huawei from the supply side will compliment U.S. efforts to compel other governments to forego purchasing Huawei gear on the demand side.
• Slowing or stopping Huawei’s progress will enhance U.S. national security.
• U.S. national security will be enhanced because U.S. or U.S.-backed 5G companies will emerge and fill the void as standard‐​setters and dominant suppliers of 5G network gear and consumer products.
• Leadership in 5G begets leadership in the next generation of communications technology and other technologies; followership consigns to more followership.
• The expected benefits of the U.S. government’s approach outweigh its expected costs.

Benefits (if the assumptions are accurate)

• The Chinese government’s ability to control or have disproportionate influence over global information and communications networks (and whatever other currently unforeseen powers that control or influence would bestow upon Beijing) will be reduced.
• Reducing Beijing’s power is—in this context and with certain caveats—akin to enhancing U.S. national security.
• Impeding Huawei’s success (albeit, through compulsion of other governments and laws restricting private companies from engaging in commerce or research and development with Huawei) could buy time for U.S. companies or U.S.-backed companies to emerge and take leadership in 5G and 6G technology space, providing U.S. economic and security benefits that might not otherwise manifest.

Costs

• Cutting off Huawei from U.S. semiconductors, semiconductor equipment, and software will expedite China’s development of indigenous semiconductor production capabilities and, ultimately, put the world’s largest market for semiconductors out of reach of U.S. producers within a few years.
• Cutting off Huawei from semiconductors made with U.S equipment in third countries will compel chipmakers in those third countries to purchase non-U.S. equipment, ultimately drying up current U.S. export markets.
• Cutting off Huawei will inject even more uncertainty into global information and communication technology (ICT) markets, which likelywill slow the process of standards setting, which likely will retard product development schedules, which likely will deter investment in new technologies, and which likely will be resolved only by bifurcation or even greater splintering of global technology standards.
• Bifurcation or splintering of technology standards would significantly limit scope for economies of scale in production, as firms all along the ICT supply chain would be producing for fewer customers or producing in separate production runs for customers that follow different sets of standards.
• U.S. supply chain warfare could prove contagious, encouraging Chinese restrictions on exports of rare earth minerals or other inputs and Chinese retaliation against U.S. technology companies, while opening the door to all countries to treat trade as a strategic weapon rather than as a tool of cooperation and economic betterment.
• Technology decoupling will inspire a cold‐​war style competition between the United States and China to win the hearts and minds of third countries through the offering of carrots and the threats of sticks.

President Trump is reportedly considering a ban on H-2B temporary workers that a few Republican senators proposed this month. Companies hire H-2B workers in seasonal jobs outside of traditional farms, such as landscaping, meat packing, and seafood processing. The plan is supposed to “save” jobs for U.S. workers. Yet this justification holds no water: unemployed Americans already reject 100 percent of jobs for which H‐​2Bs are hired.

H-2B regulations require that employers recruit U.S. workers before they can hire any foreign workers. Employers must do the following:

• Not lay off any qualified U.S. worker for 120 days before the job start date;
• Post a notice of the job’s availability for all current U.S. employees;
• Contact all its former U.S. workers in the prior year and solicit their return to the job;
• Have the State Workforce Agency post the job on the state’s electronic job registry for unemployed U.S. workers for about 60 days; and
• Recruit and accept all qualified U.S. workers interested in the position until three weeks prior to the job start date.

In other words, the H-2B program already has a built‐​in unemployment adjustment. When unemployment increases, H-2B approvals automatically fall, and not just because of a government mandate, but also because employers won’t hire anyone for jobs that aren’t in demand right now. There aren’t too many carnivals in operation right now, so that industry will not be requesting workers this year. Figure 1 shows how during the last spike in unemployment, the number of H-2B requests and approvals fell quite significantly.

But Figure 1 shows that unemployed Americans won’t accept certain H-2B positions even when unemployment is high. Landscaping, forestry, meat and poultry processing, and certain construction jobs are so physically demanding that Americans would simply prefer unemployment to taking them. In many cases—such as in forestry—the jobs require constant time apart from family and friends. It is not reasonable to expect that temporarily furloughed Americans will ditch inflated unemployment benefits for dangerous and physically taxing jobs away from their families.

These recruitment and acceptance requirements come on top of onerous wage and hour regulations that require employers to pay inflated wages and benefits to both the new hire and all existing similar employees. The H-2B minimum wage requirements exceeded every state’s minimum wage requirement by an average of 60 percent last year.

During the economic recovery, the goal should be to help businesses get back to the pre‐​pandemic status quo as quickly as possible. Forcibly keeping jobs open hinders that goal and will delay the recovery, hurting U.S. workers elsewhere in the economy. The government should stop trying to micromanage the labor force and focus on addressing the actual cause of the unemployment crisis: the COVID-19 outbreak.

Last week in this space I noted that many businesses are faced with puzzling dilemmas as they try to reopen with social distancing without running afoul of the Americans with Disabilities Act. One issue I didn’t mention: if they require the wearing of face masks as a condition of entering the premises, they may run into some customers who claim to have non‐​obvious disabilities which entitle them, as an accommodation under the ADA, not to have to wear a mask. Even if they strongly suspect such a customer of pulling a fast one, it may seem the less risky legal course just to back off, given that the law confers on business no right to demand medical documentation.

One issue I did mention last week is that the ADA creates legal risks should a business screen those who enter the premises for fever using some method such as a contact‐​free temperature gun. (Amazon announced last month that it was checking more than 100,000 employees a day this way, and checks at store entrances are familiar in some Asian countries.) Two new articles make it clear that this is one of those situations where you can look forward to being sued if you do and sued if you don’t.

Consider first a Slate article arguing against legislation to protect businesses from lawsuits related to COVID-19. The gist of its argument is that we already have a litigation system under which “liability is not likely to present a huge problem” or pose “burdensome difficulties” for businesses that “take reasonable action to keep their customers safe.” Welcome news, and so simple too!

But now that you think of it, how do we identify in hindsight a business whose safety efforts have fallen short of what is reasonable? The Slate article has some ideas on that:

A business that might be considered to have properly reopened can still find itself liable to customers for failure at the level of implementation of whatever safety protocols are required. …. what if a reasonable business would do more, such as taking the temperature of each customer as they enter the business? (Even though that’s not being widely done in the United States right now, it’s an easy precaution to take — and can at least bring down the rate of transmission.)

In sum, then, if someone sues you claiming to have contracted the virus at your business establishment, and their lawyer’s main theory is that you should have been doing front‐​door temperature checks, Slate is going to nod and say the system is working as it should.

Meanwhile, the New York Times reports that the venerable American Civil Liberties Union has issued a new report that is strenuously critical of temperature‐​sensing technologies as a screen against contagion in public places, even when done by businesses on private property. (You may wonder how the ACLU came to see pushing back against private business practices as part of its mission — especially when it lets slide so much rights‐​mangling activity by governments — but that’s another, and sadder, story.)

The ACLU report makes much of various facts that hardly anyone disputes — temperature sensors are far from ideally accurate, some people return “hot” results who do not have COVID-19 while others who do have it are not running a fever, and so forth. The argument for sensors has never been that they are perfect, but that by detecting at least some potentially contagious arrivals, they shift the odds and thus reduce overall spread of the disease in conjunction with de‐​crowding, mask use, and other measures. The report concludes that temperature sensing should go forward only if public health authorities affirmatively call for its use, and it flags possible theories, from data privacy to disparate racial impact, by which lawyers might trip up unwary businesses that go forward with it absent such a mandate.

In its wisdom, the American legal system does not give you a way to avoid legal exposure, with all its costs and miseries — but at least it gives you some choice as to which set of lawyers you will have to face off against.

The Trump administration reportedly is considering a plan this week to restrict or ban foreign students from working after graduation from U.S. universities. Under Optional Practical Training (OPT), foreign students can study for at least one year after receiving their degrees. Foreign students in Science, Technology, Engineering, or Math fields can extend OPT for another two years.

The plan is supposedly to benefit unemployed Americans—almost none of whom work in the relevant fields—after states reopen for business. The program is the main on‐​ramp for American companies to recruit and retain foreign talent in the United States. Gutting it would harm U.S. workers by undermining the very companies that would employ them.

What is the OPT?

The U.S. government created the Optional Practical Training (OPT) in 1947 to allow foreign student to seek jobs “where employment for practical training is required or recommended by the school” for “a six‐​month period subject to extension for not over two additional six‐​month periods.” In 1991, it split OPT into pre‐​completion and post‐​completion authorization and permitted only a single year of post‐​completion OPT “directly related to his or her area of studies.”

What is STEM OPT?

In 2008, the Department of Homeland Security (DHS) permitted extensions of OPT for Science, Technology, Engineering, or Math (STEM) graduates of U.S. universities for up to 17 months because the “ability of U.S. high‐​tech employers to retain skilled technical workers, rather than losing such workers to foreign business, is an important economic interest for the United States.” In 2016, DHS extended the period to 24 months, allowing for a 3‐​year period of total post‐​graduation employment. The 2016 rule requires employers hiring a worker on a STEM OPT extension to attest that the worker is not replacing a U.S. worker and will receive similar pay to the employer’s similarly situated U.S. workers. DHS can conduct an on‐​site audit to review employer records to confirm the accuracy of these attestations.

How many foreign students apply for OPT?

In fiscal year 2019, U.S. Citizenship and Immigration Services (USCIS) approved 223,284 requests for Employment Authorization Documents (EADs) for all types of OPT—the highest number ever. Nearly all approvals were for post‐​graduate OPT—either regular OPT (152,029) or STEM (69,353). Figure 1 shows that the numbers of approved EADs has nearly tripled from less than 80,000 in 2003 to more than 223,000 in 2019—nearly all of this growth occurred during the Obama administration from FY 2009 to partway through FY 2017. STEM OPT accounts for almost half the growth.

How many OPT recipients are employed?

Not all OPT recipients find jobs, but a much greater share are finding them than in the recent past. Unfortunately, an exact comparison between the share with and without jobs is not available, but Figure 2 compares OPT approvals to OPT participants with jobs. In 2007, there were more than 3 approvals for each participant with a job. By 2018, the ratio was nearly 1:1. Overall, 200,162 OPT participants had jobs in 2018.

What companies employ OPT participants?

Table 1 displays the top 100 companies for OPT hires in 2018 and 2017 as well as the entire period from 2003 to 2018. The largest OPT employer with 2,911 OPT employees was Amazon, which makes most of its hires under the STEM OPT program. Other technology companies—Integra, Intel, Google, and Microsoft—fill out the top 5 with more than 1,000 employees each. The Top 100 companies make 18 percent of OPT hires. Under the H-1B program, the Top 100 employers make 35 percent of H-1B hires. This disparity is compatible with previous research concluding that OPT is more common among startups than the more expensive, complex, and bureaucratic H-1B program.

What countries do OPT participants come from?

In 2018, two thirds of OPT participants came from India and China, according to the Institute of International Education (IIE). IIE—with the sponsorship of the U.S. Department of State—conducts a survey of international students including those with OPT. It shows that an astounding 42 percent of Indian international students are currently enrolled in OPT. That share is well over double the rest of the world. The difference can be partly attributed to the difficulties that Indians have obtaining employment‐​based green cards compared to the rest of the world. They end up using OPT to maintain status in the United States, while other nationalities can simply adjust to permanent residence.

IIE doesn’t report data specifically on STEM OPT, but Immigration and Customs Enforcement data show that again, India represents 59 percent of all STEM OPT participants, and 40 percent of Indian STEM students are involved in OPT—by far the highest share of any nationality with more than 2 participants. It also shows that the United States risks losing nearly half a million STEM students if the government rescinded OPT.

How has COVID-19 affected OPT occupations?

OPT participants can work in any field directly related to their major, but as the list of employers makes clear, they overwhelmingly work in technology and computer professions. As Stuart Anderson first noticed, computer occupations have almost entirely remained unscathed by the recent economic downturn. As of mid‐​April, the unemployment rate had actually declined slightly since January. Anderson comments that this positive employment situation “makes citing unemployment in these occupations as a reason for new restrictions on H-1B visas and international students on OPT more dubious.”

What does research say about OPT?

Business Roundtable and the Interindustry Forecasting Project at the University of Maryland found that “scaling back OPT would cause the unemployment rate to rise 0.15 percentage points by 2028.” In 2019, economist Madeline Zavodny found that “A larger number of foreign students approved for OPT, relative to the number of U.S. workers, is associated with a lower unemployment rate among those U.S. workers.” In March 2019, Jeremy Neufeld used OPT data to conclude that “higher levels of OPT participants in a region lead to increased innovation in that region, as measured by the number of patents, higher average earnings among the college educated. In addition, it finds no evidence of adverse effects on average earnings, unemployment, or labor force participation.”

Stuart Anderson has also found that 39 percent of Nobel Prizes in chemistry, medicine, and physics since 2000 have been immigrants to the United States. International students dominate the STEM majors. NFAP has also found that in 2015, more than three quarters of the graduate students in electrical engineering, petroleum engineering, computer science, and industrial engineering were international students. A majority of the graduate students in pharmaceutical sciences and chemical engineering—two majors of particular interest today—were also foreign‐​born. In 2018, NFAP found that nearly a quarter of American startups valued at $1 billion or more had a founder who entered the Untied States as an international student, which had created 1,200 jobs per company. Michelle Zatlyn, co‐​founder of Cloudflare, and Ashifi Gogo, founder of Sproxil, both used OPT to launch their businesses.

In 2008, the Department of Homeland Security (DHS) when it launched STEM OPT concluded that the expansion was a national security necessity. “With their large and growing populations of STEM‐​graduate scientists, high‐​tech industries in [Russia, China, and India] and others in the OECD now compete much more effectively against the U.S. high technology industry.” It notes that China and India had 31 percent of global R&D staff up from 3.2 percent in 1990. It would be strange for the department to reverse itself completely a decade later, even as the underlying trends continue.