Building an Evergreen $1 Billion Fund for Science and Technology Career Advancement
The H-1B visa for “specialty occupation” workers has become a significant element of the U.S. employment-based immigration system. Less well-known is that employers of H-1B workers annually pay hundreds of millions of dollars for domestic education and training programs in science, technology, engineering, and mathematics (STEM), administered by the Department of Labor (DOL) and the National Science Foundation (NSF). This fee-based funding stream was created in the late 1990s and has not been meaningfully updated by Congress in the succeeding decades. It is mandatory funding, tied to a continuous flow of H-1B filing fees rather than the annual congressional appropriations process. Both the Obama and Trump administrations seized on this unique pot of money for advancing education and training priorities for Americans without new legislation or appropriations.
The Biden administration can take even greater advantage of this funding to launch innovative programs that advance U.S. economic competitiveness and diversify the STEM talent pipeline—two mutually reinforcing goals. Specifically, in this paper we recommend:
- Reestablishing the TechHire Initiative to rapidly train U.S. workers for in-demand technology jobs
- Establishing a new Advanced Research Projects Agency—Labor (ARPA-L) to conduct high-impact R&D programs that create breakthroughs to meet America’s workforce challenges
- Significantly increasing the number of graduate research fellowships dedicated to students in STEM fields who completed their undergraduate education at non-R1 universities
- Significantly increasing the number of faculty training grants in STEM fields where a dearth of professors has created a bottleneck for graduate education (e.g. artificial intelligence)
In addition, Congress should increase the fees paid by H-1B employers to reflect (a) the increase in inflation over the past two decades, as well as (b) the ability of major corporations, which are often the most prolific sponsors of H-1B workers, to pay more than small businesses.
Background
A Brief History of the ACWIA Fee Account for STEM training
In the 1990s, the technology sector was growing rapidly and the demand for high-skill workers was quickly outpacing supply. The H-1B visa, first enacted at the beginning of the decade, was becoming a popular option to bring in such skilled workers. In 1997, the number of applications for H-1Bs exceeded the established cap of 65,000 visas for the first time. The next year, demand was so high that the cap was reached within days of the opening date for new filings—and this has been the case nearly every year since.
Congress considered legislation to increase the H-1B cap, but faced strong bipartisan opposition, as well as pushback from labor unions and professional associations. As a compromise, along with raising the caps for three years, Congress established a special fee for sponsors (“petitioners”) of H-1B workers which would be deposited into a new fund called the “H-1B Nonimmigrant Petitioner Account.” Because these provisions were included in the American Competitiveness and Workforce Improvement Act of 1998, immigration practitioners often refer to “ACWIA fees” and the “ACWIA fund.”
Originally, the ACWIA fee amounted to $500 per qualifying petition. The fee was increased twice since its creation: once in 2000 (to $1,000) and once in 2004 (to its current level of $1,500). The 2004 adjustment also specified for the first time that employers with 25 employees or fewer would pay a lower rate of $750 per qualifying petition. Congress mandated that the funds be distributed primarily to DOL and NSF to support domestic STEM education and technical skills training programs. The current distribution of funds is as follows (see the Appendix for full statutory details):
- 50 percent to the Secretary of Labor for job training programs;
- 30 percent to the Director of NSF for scholarships for low-income students enrolled in STEM programs;
- 10 percent to the Director of NSF for a direct matching grant program that funds public-private partnerships in K-12 education;
- 5 percent to the Secretary of Homeland Security to improve processing employment-based temporary visa and green card petitions; and
- 5 percent to the Secretary of Labor to decrease the processing time for labor certification applications.
With this money, DOL has relatively wide latitude to make competitive grants to businesses, business related nonprofit organizations, education and training providers (such as community colleges), “entities involved in administering the workforce development system,” and economic development agencies. These grants are intended to support job training programs that help both unemployed and employed workers learn new skills to obtain a job or promotion, especially in industries experiencing significant growth. To determine these in-demand industries, the Secretary of Labor must consult with state workforce investment boards and take into account sectors that are “projected to add substantial numbers of new jobs”; “are being transformed by technology and innovation requiring new skill sets for workers”; “are new and emerging businesses that are projected to grow”; or “have a significant impact on the economy overall or on the growth of other industries and economic sectors.”
NSF, on the other hand, has more statutory restrictions on how it can use its allocated ACWIA fees. Scholarships for low-income individuals pursuing associate, undergraduate, or graduate STEM degrees cannot exceed $10,000 per year for up to four years, although up to 50% of this funding stream (15% of the total ACWIA fund) may be used for “undergraduate programs for curriculum development, professional and workforce development, and to advance technological education.”
NSF’s K-12 STEM education grants (10% of the total H-1B fund) must be awarded to public-private partnerships that serve one or more of the following purposes specified by Congress:
- Support the development and implementation of standards-based instructional materials, models, and related student assessments that enable K–12 students to acquire an understanding of STEM, and to develop critical thinking skills;
- Provide systemic improvement in training K–12 teachers and education for students in STEM;
- Support the professional development of K–12 STEM teachers in the use of technology in the classroom;
- Stimulate system-wide K–12 reform of STEM in rural, economically disadvantaged regions;
- Provide externships and other opportunities for students to increase their appreciation and understanding of STEM, including summer institutes sponsored by an institution of higher education for students in grades 7–12;
- Foster partnerships of industry, educational institutions, and community organizations to address the educational needs of disadvantaged communities;
- Provide college preparatory support to expose and prepare students for careers in STEM; and
- Provide for carrying out certain NSF systemic reform activities.
Programs currently funded by ACWIA fees
Both NSF and DOL provide publicly-available data on the ACWIA fees that are spent on the agencies’ programs. Table 1 includes the total amount of funding received by NSF and DOL from fiscal years (FY) 2010 to 2021 as noted in the agencies’ annual budget requests.
NSF currently uses its money from ACWIA fees to fund two programs: Scholarships in Science, Technology, Engineering, and Mathematics (S-STEM) and Innovative Technology Experiences for Students and Teachers (ITEST). By the end of FY 2018, the agency had received almost $2 billion in cumulative ACWIA fees to support scholarships, as well as K-12 students and teachers.
NSF must allocate three-quarters of its ACWIA receipts (30 percent of the total account) to scholarships for lower-income students pursuing associate’s, bachelor’s, and advanced STEM degrees. Through the S STEM program, NSF makes grants to higher education institutions (about 90 in FY 2019) which then award scholarships of $10,000 per year for up to 4 years. Between FY 1999 and 2018, the S-STEM program resulted in 87,890 scholarships for U.S. students (including both citizens and permanent residents).
Although the nature and amount of these scholarships are fixed in statute, Congress does provide the NSF Director wide discretion to spend up to 50 percent of the current S-STEM funds “for undergraduate programs for curriculum development, professional and workforce development, and to advance technological education,” all of which “may be used for purposes other than scholarships.” This means that an annual amount of around $50 million is available for such supporting programs.
Department of Labor
Over the past decade, DOL has used its ACWIA fee receipts to fund a series of job training initiatives, usually tied to a presidential priority. DOL has cumulatively received about $2.5 billion in ACWIA fees to train professionals in the United States. The Secretary of Labor has wide discretion to designate “high growth industries and economic sectors” as targets for this funding, based on the following factors:
- The sectors are projected to add substantial numbers of new jobs to the economy;
- The sectors are being transformed by technology and innovation requiring new skill sets for workers;
- The sectors include new and emerging businesses that are projected to grow; or
- The sectors in question have a significant impact on the economy overall or on the growth of other industries and economic sectors.
Using ACWIA fees, the Obama administration issued funding opportunity announcements for programs to support job training for the long-term unemployed (“Ready to Work”), coding bootcamps (“TechHire”), and apprenticeship programs, among other priorities. The Trump administration also used these funds to support its efforts to expand apprenticeship programs (“Closing the Skills Gap”).
The Ready to Work program (RTW) was launched in 2014 as a response to those who lost their jobs during the Great Recession and remained under- or unemployed as the economy recovered. DOL is in the middle of evaluating the success of this program and is expected to complete its study by May 2022. In 2017, the agency released an interim report that examined the first year of grantees’ operations in Maryland, California, New York, and Washington. The programs provided specialized, one-on-one counseling to the participants and coordinated with local occupational training programs and employers in relevant sectors.
TechHire was established in 2015 and has aimed to build talent pipelines in technology sectors throughout the country. Initial funding for the program amounted to $100 million in grants to support partnerships that train young adults and other disadvantaged groups, such as people with disabilities, individuals with limited proficiency in English, and those with criminal records. A full evaluation on the benefits of the program is expected from DOL in September 2021.
Closing the Skills Gap awarded grants to 28 public-private partnerships in early 2020 that amounted to almost $100 million. The program aims to achieve “large-scale expansions of apprenticeships in industries including advanced manufacturing, healthcare, and information technology.” Likely because the Closing the Skills Gap program is still so new, there are no studies announced to evaluate its impact yet.
Plan of Action
Recommendations for High-impact STEM Education and Training Programs
As currently authorized by Congress, the ACWIA fees yield an approximately $350 million annual fund for STEM education and training that is essentially on autopilot, funded by employers rather than taxpayers. The Biden administration has an opportunity to focus DOL and NSF on using these funds to advance its top priorities of economic recovery and racial equity.
Specifically, DOL can ramp up the TechHire initiative for in-demand technology jobs and establish a new Advanced Research Projects Agency—Labor (ARPA-L) to conduct high-impact R&D programs that create breakthroughs to meet America’s workforce challenges. NSF can significantly increase both the number of graduate STEM research fellowships dedicated to underserved students as well as the number of faculty training grants in fields where a dearth of professors has created a bottleneck for graduate education (e.g., artificial intelligence).
Reestablish the TechHire Initiative
The TechHire initiative, described in more detail above, has already demonstrated the value of involving technology companies in rapid STEM training programs. One of the first TechHire grants was awarded to LaGuardia Community College and helped them form a partnership with state and federal agencies, along with software development and training companies. The goal was to provide intensive training in tech skills to low-income young adults and as of 2019, over 80 percent of students in the bootcamp graduated. Retention was over 90 percent. This is just one of the 39 partnerships established by the program, which serves communities in 25 states.
No further DOL funds have been awarded to the TechHire initiative since its inception in 2015, however. Especially as our country embraces an increasingly tech-focused work environment, further tech skills training will be essential. We recommend allocating $50 million per year to the TechHire initiative to sustain it and establish new public-private partnerships across the country. To encourage high-impact outcomes, the revitalized TechHire initiative could make grants above a certain award amount (e.g., $2 million) contingent on demonstration of wage gains following training, and could allow non-profits (not only workforce boards) to serve as the lead applicant.
Establish a new Advanced Research Projects Agency—Labor (ARPA-L)
With the nature of work changing rapidly, one federal initiative that could significantly boost the United States’ long-term competitiveness in high-impact industries would be the development of an Advanced Research Projects Agency for the Department of Labor (ARPA-L). According to a Day One proposal developed by former Defense Advanced Research Projects Agency (DARPA) Director Arati Prabhakar and Coursera executive Jeff Kaplan, ARPA-L would drive innovation in workforce training and labor market outcomes, where major research efforts are currently lacking. By weaving research advances together with lessons from the real world, ARPA-L aims to catalyze high-impact R&D focused on creating powerful, scalable approaches to pressing workforce issues including unemployment and market disruption. With the support of Congress and the White House, this new organization should be housed within the Department of Labor in order to best deliver bold advances that ultimately change what’s possible for America’s workers.
The ARPA model is known for its success in creating radically better approaches to hard problems by conducting solutions-oriented R&D. DOD’s DARPA, now in its seventh decade, conducted the pivotal R&D for new military capabilities such as stealth and precision strike and, more broadly, for new information technologies from the internet to artificial intelligence. DARPA’s track record inspired the establishment of the Department of Energy’s ARPA-E and the Office of the Director of National Intelligence’s IARPA. Both of these ARPAs are well underway, with robust portfolios of R&D programs and encouraging results. They demonstrate that it is possible to adapt the DARPA model for different public purposes.
Though this ARPA model has been highly successful for national security and energy research, it has not yet been implemented for the improvement of workforce training and education programs. ARPA-L would be an innovative addition to DOL, particularly because the agency’s current budget does not include any funding for workforce training research and development. Some potential research and development areas to close the skills gap include:
- Develop and validate diagnostic systems to improve skills assessment and enable workers to better understand how their skill set matches the needs of employers.
- Demonstrate accelerated skill development that uses information about an individual (e.g., age, initial skill level, and prior experience) to devise an individualized training regimen for interest and expertise.
- Explore and evaluate new ways to personalize and accelerate the training process by building on advances in learning science and neuroscience.
- Advance and test the effectiveness of emerging innovations like human-computer interaction and mixed reality for training for complex tasks.
- Experiment with and assess alternative certifications and micro-credentialing programs to train and upskill youth and displaced adult workers around the country while connecting participants directly to employers.
In addition, ARPA-L would support timely labor market data collection and analysis to evaluate the research and training programs. Conducting labor market analysis with ARPA-L would help with the development of innovative training programs, as well as allowing employers, employees, and the federal government to respond to economic changes. Some examples of useful analyses include:
- Experimenting with and testing data tools for each worker, combining information on local job postings, wages, and training requirements with information about credentialing and training services to offer workers the most meaningful and actionable information for their career goals in real time.
- Assembling data from public and private sources on regional labor trends and testing their effectiveness in enabling employers to make more impactful, targeted, and timely investments in local workforce development opportunities.
- Collecting and analyzing diverse datasets to identify targeted, effective leverage points for innovative labor policy interventions.
Allocating $100 million per year from the ACWIA fund to kickstart ARPA-L would put the United States on a much better path to supporting U.S. workers and sustained wage growth in our changing national and global economy. This can be accomplished administratively in the immediate term, with Congress authorizing and appropriating a larger program after a strong track record has been established.
Optimize STEM graduate fellowships for students from emerging research universities
Higher education R&D funding is scarce, and is not distributed equitably. The American Physical Society found that in 2018, out of more than 600 colleges and universities that received federal science funding, 22 percent received over 90 percent of the funds. These institutions serve only 43 percent of all students and only 34 percent of underrepresented minority students in the United States. This distribution of funds means that two thirds of underrepresented minority students and almost 70 percent of students who receive Pell grant funding have significantly fewer opportunities to engage in cutting-edge scientific research.
Without undergraduate research experiences afforded by federal R&D funding, students at emerging research universities are then less competitive for future NSF-funded opportunities at any university, such as graduate fellowships. “Emerging research institution” (i.e., non-R1) is a category that includes geographically diverse state schools and nearly all minority-serving institutions.
NSF already uses the ACWIA fund to address this problem in part, through the S-STEM program described above. Colleges and universities apply for competitive grants to “increase the number of low-income students who graduate and contribute to the American innovation economy with their STEM knowledge,” for example through innovative curricula. While these institution-level awards have merit, they create a patchwork of programs for which the lion’s share of low-income STEM students are ineligible at any given point in time.
In contrast, consider the prestigious NSF Graduate Research Fellowship program, where individual students directly apply for three years of financial support, with an annual stipend of $34,000 plus $12,000 to the university where they pursue their graduate-level STEM education. Based on an increase in appropriations, Congress doubled the total number of such fellowships over the past decade (from around 1,000 to 2,000).
To lower barriers to graduate STEM education for outstanding students of all backgrounds, NSF should consider allocating $50 million of its ACWIA funds to an individual-level scholarship program—like the NSF Graduate Research Fellowships—open to students who obtained their undergraduate degree from an emerging research institution. To be clear, these fellows could pursue their graduate degree at any research university, whether R1 or emerging.
For its part, Congress should lift the statutory cap of $10,000 for such scholarships, which gets smaller in real terms with each passing year.
Increase the number of faculty training grants in critical STEM fields
The demand for faculty in cutting-edge fields, such as AI, is rising rapidly. According to a report by the Center for Security and Emerging Technology (CSET), the number of bachelor’s degrees in computer science and engineering almost tripled between 2009 and 2017. In addition, the enrollment for introductory courses in AI in 2017 was three to five times higher than in 2012. The flow of faculty moving from institutions of higher education to industry has also increased dramatically, so it has become quite difficult to properly support the rising number of U.S. students interested in an education in AI.
This dearth of qualified professors represents a major long-term constraint on AI education in the United States, and will no doubt constrain U.S. competitiveness in other advanced fields as they develop in unexpected directions in the future.
Therefore, NSF should consider allocating another $25 million from its ACWIA funding stream to incentivize universities to create new faculty positions in STEM fields where there is a teaching bottleneck. To that end, NSF could expand and adapt its Faculty Early Career Development (CAREER) Program, which provides awards of up to $400,000 over five years to promising faculty members.
Recommendations for Congress: Growing the Pie
As described above, the ACWIA fund can significantly advance STEM education and training priorities in the United States, without any further action by Congress, through optimal use of the existing $350 million annual funding flow.
But the size of that flow is somewhat arbitrary, and ought to grow. This is especially important now that experts are warning that China has the resources to surpass the United States in AI and other STEM fields over the next few years. Congress should therefore increase the size of the pie by raising H-1B fees in an equitable way.
Currently, the ACWIA fee structure has two tiers based on the size of the employer filing the petition. Congress set the fees at $750 for employers with at most 25 U.S. employees and $1,500 for employers with more than 25 U.S. employees. However, this fee structure has not changed since 2004—during which time inflation has increased by over 30 percent—and it also does not take into account the financial resources of major corporations that hire the great majority of H-1B workers.
Congress should update the fee structure so that (a) the two current fee tiers are increased 30% to account for past inflation; (b) a new fee tier is added for companies larger than the Small Business Administration’s 500-employee threshold for a “small business”; and c) all fees are automatically indexed to inflation in the future.
Higher fees for large companies were recommended by Microsoft in 2012, when it published a proposal for Congress to allocate additional 20,000 H-1B visas for professionals in STEM fields and to require large companies to pay a fee of $10,000 for each petition. Microsoft also proposed recapturing unused green cards and allocating 20,000 of them annually for STEM professionals. Sponsors for these green cards would pay $15,000. These new funds, which would amount to about $500 million per year, would then be dedicated to domestic STEM education programs.
It is important to note that H-1B petitions in certain circumstances are exempt from ACWIA fees. These exemptions include petitions from:
- Institutions of higher education;
- Nonprofit entities related to or affiliated with institutions of higher education;
- Nonprofit research organizations or governmental research organizations;
- Primary or secondary educational institutions; or
- Nonprofit entities that engage in “an established curriculum-related clinical training program for students.”
Fees are also not required for most H-1B extensions under any kind of employer.
With these details in mind, we calculated the estimated revenue that would be generated by the modernized fee structure proposed above. We referred to USCIS data on current H-1B employers, annual rates of the submission of petitions, as well as USCIS’s analysis of petitions from small entities (at most 500 employees) and “non-small” entities (above 500 employees).
We estimate that a modernized ACWIA fee structure could bring in around $1 billion per year, or about triple the current revenue level. The data and our estimates can be found in Table 5 and Table 6 below.
Conclusion
As China and other countries ramp up spending to boost their own domestic research and development capabilities, the United States must act to maintain its global scientific and technological leadership.
Since its creation two decades ago, the ACWIA fund has been a valuable and reliable resource to support STEM workforce training and education programs at DOL and NSF. Congress should grow this annual funding stream to $1 billion—at no cost to taxpayers—by modernizing the ACWIA fee structure to keep up with inflation and reflect the size of the large corporations petitioning for most H-1B professionals.
Even before Congress takes these overdue actions, the administration should allocate the existing annual flow of ACWIA funds to expand the TechHire initiative, institutionalize a new ARPA–L, support a new generation of underserved STEM graduate students, and eliminate faculty bottlenecks in critical STEM fields.
The time is ripe to seize this opportunity to harness America’s home-grown STEM talent to accelerate innovation and power the nation’s inclusive economic growth.
The authors would like to thank Amy Nice, Ryan Burke, Remco Zwetsloot, Diana Gehlhaus, and Mark Elsesser for their insightful recommendations during the drafting of this report.
The education R&D ecosystem must be a learning-oriented network committed to the principles of innovation that the system itself strives to promote across best practices in education and learning.
Across the country in small towns and large cities, rural communities and the suburbs, millions of young people are missing school at astounding rates.
CHIPS is poised to ramp up demand for STEM graduates, but the nation’s education system is unprepared to produce them.
An analysis of the President’s FY25 budget proposal by the Alliance for Learning Innovation found a lot to like.