Definitions:
STEMM: Science, Technology, Engineering, Mathematics (STEM) + Medicine.
STEMM Pipeline Pool: People aged 22-31 who took calculus in high school
Young STEMM Degree Holders: People aged 22-31 with a BA or more in a STEMM subject.
High-Skill: Education through BA and above
Mid-Skill: Education past high school but without BA
Low-Skill: High school completion and below
Young High-Skill STEMM Workers or Workforce: People aged 22-31 working in the STEMM and STEMM-related occupations[1] with education of a BA or more.
Summary of Argument:
By 2034 the US education system will supply between 800 thousand and 4 million fewer Young High-Skill STEMM workers in the US than industry will demand. This will happen because the STEM Pipeline Pool, defined as students who took calculus in high school, started to decrease in 2017, and because the flow of high-skill immigrants to the US is decreasing. What will make this difficult to reverse is the fact that teacher production has dropped so much that the US is certifying a smaller number of mathematics teachers each year than schools are hiring. However we need only a modest course correction to prepare additional mathematics and science teachers and set the US on the path towards an adequate high-skill young STEMM workforce a decade from now. Increasing the number of mathematics teachers quickly by around 2000 and then increasing the number of new mathematics teachers prepared yearly by around 2000 would help set the United States on a better path.
Summary of Workforce Model Used for Projections:
1) The number of Young STEMM Degree Holders is 75% of the STEMM Pipeline Pool. The number of US-born high-skill STEMM workers is 83% of the Young STEMM Degree Holders, and they compose 80% of the Young High-Skill STEMM Workforce. The remaining 20% of the Young High-Skill STEMM Workforce was born abroad.
2) High school calculus enrollment peaked in 2017 and has been declining. A demographic decline is making its way through high schools; there are 9% fewer 8-year-olds than 18-year-olds. Therefore the STEMM Pipeline Pool is about to decrease. The number of international students coming to the US to study STEMM and work here is likely to decline. These factors will create a shortfall of around 800,000 Young High-Skill STEMM Workers by 2034.
3) In 2015 the number of people completing teacher education programs with certificates in mathematics dropped below the number of mathematics teachers hired each year into schools.
4) The impending shortage of Young High-Skill STEMM Workers can be averted by retaining an additional 1000 mathematics teachers and preparing on the order of 2000 more mathematics teachers per year. Opportunities to increase student participation in calculus are greatest in rural and small-town schools, and in schools with concentrated poverty.
A model of the US STEMM Workforce
Near the opening of Rising Above the Gathering Storm, authors of a National Research Council Committee wrote in 2005
Having reviewed trends in the United States and abroad, the committee is deeply concerned that the scientific and technological building blocks critical to our economic leadership are eroding at a time when many other nations are gathering strength. We strongly believe that a worldwide strengthening will benefit the world’s economy—particularly in the creation of jobs in countries that are far less well-off than the United States. But we are worried about the future prosperity of the United States. Although many people assume that the United States will always be a world leader in science and technology, this may not continue to be the case inasmuch as great minds and ideas exist throughout the world. We fear the abruptness with which a lead in science and technology can be lost—and the difficulty of recovering a lead once lost, if indeed it can be regained at all.
They recommended four key actions. The first of them, 10,000 Teachers for 10 Million Minds was to invest $2 Billion a year in scholarships for STEM students to become STEM teachers, with additional funding for university programs and incentives for teachers to enter high-needs schools. Looking back 20 years, it is safe to say we did not follow this recommendation. My STEM teacher program, UTeach, was highlighted in the report, and we received around $30 million from ExxonMobil that enabled us to begin expanding nationally to become a network that currently numbers over 50 universities. The National Science Foundation created the Robert Noyce Scholarship program that started spending around $60 million annually on STEM teacher scholarships. These and other efforts fell far short of the recommendations, and the result was that the number of STEM teachers the country produced rose until around 2012 and then started dropping so that by 2023 it was 40% less than it had been at the 2012 peak.
Yet from 2010 through 2024 as shown in Figure 1, the total number of US workers in STEMM occupations was growing, with the fastest growth for the high-skilled. Salary for workers born in the US was almost flat: a reasonably strong indicator that no worker shortage existed. There is no sign that a dramatic decrease in the number of new US STEM teachers, despite the recommendation of the Gathering Storm report to invest billions a year to increase them, resulted in dire consequences[2].
Figure 1: Growth in number and income of all U.S. STEMM workers. They are broken into six categories depending on whether they are born in the US or born abroad, and their education level. Data from American Community Survey through IPUMS (University of Minnesota).
I want to gain some perspective on what happened. Were those of us who said the US needed more and better STEM teachers for the purpose of competitiveness crying wolf? Where has the US STEM workforce come from? Has it mainly come from abroad? If the failure to increase or even maintain US STEM teacher production has not affected us until now, will it in the future? If we have been wrong in predictions before, why will this time be different?
Building a workforce takes time. There is a long interval between finishing high school and beginning to work and variation of up to ten years in the time of entry for students who complete high school at the same time. This makes it easy to underestimate the extent to which we benefit or suffer from decisions made decades ago. To be specific, let us think about the young high-skill STEM workforce in 2024, the most recent year for which the American Community Survey data are available. Young people with this educational level begin entering the workforce in large numbers at age 22 and have mainly completed the process of entering by age 31. These oldest entrants were completing high school at 18 years old in 2011 and entered school at age 6 in 1999. This means they went through school during the peak period of No Child Left Behind. Going back four annual cohorts we have 27-year-olds in the workforce who completed high school in 2015, which is the year that the Every Student Succeeds Act (ESSA) replaced No Child Left Behind (NCLB). But of course their own schooling was completely determined by the testing and highly qualified teacher policies of NCLB. We have to push back another four years to 23-year-olds in the workforce in 2024 to find students whose high school experience was under ESSA, and they had elementary and middle school experiences governed by NCLB.
Thus today’s Young High-Skill STEMM Workforce grew up almost entirely in schools governed by NCLB. Even those who disliked the emphasis in NCLB on standardized tests and highly qualified teachers may acknowledge that it created pressure on schools to broaden and improve STEM offerings . We are just beginning the post-NCLB era.
There is a pool of students coming out of US high schools who will be the pool for the STEMM workforce. The longest data series I can find to serve as a proxy for this is in Civil Rights Data Collection, and it is the number of students in high school taking AP Math (up to 2012) and taking Calculus (including dual-enrollment students, after 2012). In 2012 both these values are available and they are almost identical[3].
Calculus is only a proxy for a STEMM readiness in high school, but it is not unreasonable. There are three main reasons for this.
1) The chances of admission to competitive undergraduate science and engineering programs go up for students who have successfully completed calculus. The admissions requirement is usually calculus readiness, not calculus itself, but nothing establishes readiness better than successful completion.
2) Calculus is a degree requirement in many universities and for many STEM degree majors. Successful completion of calculus in high school, even if it must be repeated in college, increases the chance of college completion, and therefore the chance of STEM degree completion.
3) Taking Calculus in high school is part of developing an identity as a future STEM professional, signaling an aspiration to enter this professional area or at least to leave the option open.
There is a considerable research literature on these topics. I pull information from just three. Figure 2 displays in graphical form a model for becoming a US STEM major where taking calculus in high school is one of the factors. Only 16% of high school students who do not obtain Calculus credit in high school become STEM majors. Chang et al[4] conclude that “calculus course completion is the most predictive factor among 102 examined variables, including individual, family, and school factors.” Giani et al[5] investigated whether calculus is preferable to statistics. They concluded that if one measures earnings the two cannot be distinguished, and they offer a thoughtful analysis of the benefits of statistics, but the “rise of statistics should not be interpreted as a reason for schools to shift away from offering calculus. The data corroborate what has long been understood: calculus retains unique value for high school students eyeing STEM majors in college. AP Calculus AB students are significantly more likely than AP Statistics students to attend selective colleges, major in STEM, and pursue careers in engineering, computer science, and physical sciences.” Finally, any discussion of calculus needs to acknowledge that typical course sequencing in the US makes it difficult to reach high school calculus unless students have Algebra I in 8th grade[6]. Thus any story of high school calculus is really a story of a sequence of mathematics opportunities and challenges beginning in middle school. This needs to be kept in mind, even when mainly focusing on calculus itself.
Another criticism of a focus on calculus is that the shortage of a Mid-Skills STEM Workforce poses an equally, or more serious problem. A prominent example comes from semiconductor manufacturing, where the Semiconductor Industry Association has predicted a shortage of 26,000 technicians[7].
Figure 2: Reproduced from Chang et al [Note 4] showing how the probability of becoming a STEM major depends upon various factors including calculus enrollment and completion in high school. The chance of becoming a STEM major without calculus is only 16%, although other factors are also important. The full caption to the figure reads: STEM pipelines — predicting STEM college major choice in high school (2009–2016). The water drops represent the probability of declaring a college major in STEM. PR stands for percentile rank. The CART results illustrate how these four variables influenced the 2009 cohorts’ college major choice in 2016. In summary, if high school students do not earn any calculus credits, the likelihood of majoring in STEM disciplines will be only 16%. Furthermore, even if students earn calculus credit(s), their chance of pursuing a postsecondary STEM degree will still be low (22%) if they do not exhibit a high science identity in the 11th grade (PR <74). On the other hand, if students earn calculus credit(s) and have a high level of science identity in the 11th grade (PR ≥74), the likelihood of enrolling in a STEM college major will increase substantially (from 16 to 37%). Interestingly, the probability of students majoring in STEM will be boosted to 68% if students earn calculus credit(s), have a high level of science identity in the 11th grade, and either earn at least 9.8 credits in STEM-related courses or have high math achievement in the 11th grade (PR ≥97)
The SIA does not identify shortfalls in K-12 education as a component of the problem; the educational programs of the Chips and Science Act were narrowly focused training in community colleges. This particular example seems to be driven by the challenge of hiring a manufacturing workforce in the US at wages that can compete with those paid in Asia. It is certainly true that if Algebra I were taught better in high schools, more students would be able to enter community college certification programs without remediation in mathematics, and as a matter of justice we should aim for this outcome. However I have not yet found evidence that educational deficits rather than salary are preventing STEM industries from obtaining enough mid-skilled people.
A more comprehensive study comes The Georgetown University Center on Education and the Workforce[8]. This report concludes that “Nationwide, there is an annual shortage of nearly 712,000 certificates and associate’s degrees aligned with high-paying middle-skills occupations in four occupational groups: bluecollar (360,800), management and professional office (253,000), STEM (87,500), and protective services (10,600) occupations. These shortages are projected to persist at least through 2032. [p. 13]”
This conclusion appears at odds with the middle panel of Figure 1. The difference in methodology is that the Georgetown University group has added up all the certificates and degrees awarded that lead logically to the jobs (using the Integrated Postsecondary Education Data System) and they find fewer certificates and degrees than jobs. The problem with this line of argument is that if one examines the makeup of the Mid-Skills Workforce, it breaks down with 50% having an Associate’s degree (this did not change from 2010 through 2024), and most of the remainder have one or more years of college credit but no degree. The average age of these workers is 45; this indicates they are career professionals. Salaries adjusted for inflation have been flat. Thus, whether or not these workers acquired credentials that logically seem to prepare them for their jobs, they are holding the jobs and being paid for them. In 2010 the average income for Mid-Skills STEMM workers with an Associate’s Degree was $72K (2024 dollars) and for those with some college no degree was $63K. In 2024 the average salary for those with an Associate’s degree fell slightly to $69K and did not change for those with some college no degree.
Whether there is a Mid-Skills STEM Worker shortage depends on the necessity of credentials. The country is debating whether the BA represents needed education or a credential unrelated to skill[9]. There is less of a debate in the Mid-Skill arena. Half the workers do not have and have not had a degree. The Georgetown shortage estimate comes from adding in those who did not acquire a certificate: 87,000 people a year are entering Mid-Skill STEMM occupations without degree or certificate. To complete the argument, one has to establish that this is a problem: that employers are unhappy, that the workers do not stay, that they perform poorly, or that access to jobs is unfair. This is missing. And since the highest-paid Mid-Skill STEMM workers are STEM Managers, only 40% of whom have Associate’s degrees, requiring a credential is probably not viable.
Figure 3: STEMM Pipeline Pool for the Young High-Skill STEMM Workforce. The orange line shows a projection that assumes no change other than a decline due to a drop in the US school-age population working its way through the system. Civil Rights Data Collection.
Therefore I return to Calculus, which I believe prefigures a genuine shortage the US will find difficult to remedy. As shown in Figure 3, the number of AP Math/Calculus students rose sharply until 2016 and then started declining in 2018[10]. Before focusing on the decline after 2018, it is worth briefly pausing to celebrate the rise, which lasted over a quarter of century. The most recent data are from 2022 which show a drop from a peak of nearly 700,000 in 2016 to around 600,000 in 2022. In order to compose projections for the next ten years, we need an estimate of the how the high school STEMM pool will evolve. Note that a demographic decline is working its way through the public schools; absent immigration the Census Bureau estimates a 9% drop in 18-year-olds over the next decade, and a 4% drop in a low-immigration scenario[11]. For the Advanced STEMM Pool I estimated that the decline in Calculus enrollment will stabilize at the 2022 level except for a further 5% drop over the next decade due to the demographic decline.
Figure 4: STEMM Pipeline Pool, Young STEMM Degree Holders, and Young High-Skill STEMM Workers born in the US and abroad through 2024.
Figure 5: Models for Young STEMM Degree Holders and Young High-Skill STEM Workers born in US compared with historical data.
Figure 6: Illustration of the way that 10 lagged cohorts are used to find the size of the STEMM pipeline pool coming from high school. Back in 2010 the workforce was given by 50% of the pool. By around 2018 this percentage had risen to about 62% and persisted.
Figure 4 shows the data that leads naturally to projections for the next decade. The top graph shows the STEMM Pipeline Pool defined by having had AP Math (prior to 2012) or Calculus (2012 and after). For each year this is done by summing up the math enrollment for the prior 4 to 13 years, as shown in Figure 6. For example, for 2024, the high school STEM pool is made up of those who took Calculus or AP Math between 2011 and 2020. The next curve shows the Young STEMM Degree Holders. The ratio of the Young STEMM Degree Holders to those in the STEMM Pipeline Pool increases between 2010 and 2024, as shown in Figure 5. In 2010 it is 63%, while by 2020 it has risen to 78% and then hovers around this value until 2024. The final curve shows the Young High-Skill STEMM Workers born in US. The ratio of Workers to Degree-Holders held remarkably constant at around 83% from 2010 to 2024, also shown in Figure 5.
This leads to a simple rule. Find the STEMM Pipeline Pool. Multiply this by 78% to get the number of US-born Young STEMM Degree Holders four years later. Multiply again by 83% to find the number of US-born Young High-Skill STEMM workers. These ratios underly the projections in Figure 7. The high school STEMM pool is projected by assuming that the drop in Figure 3 stabilizes apart from the slow drop expected due to a decline in numbers of US youth. The US-born Workforce projection is obtained by multiplying the Pipeline Pool projection by 62%.
Figure 7: Projections of STEMM Pipeline Pool, and Young High-Skill STEMM Workforce, born in US, born abroad, and Total.
The final major contribution to the High-Skill Young STEMM workforce comes from people born abroad. They have been around 20% of the total. It would have been natural to project this growth to continue except that the United States is adopting stricter policies around high-skilled immigration. Employers will have to pay a $100K fee for H-1B visas, F-1 and J-1 visas may be limited to 4 years, and in Texas higher education has been forbidden use of H-1B visas. Based on these trends, I project that between 2024 and 2034 the number of STEMM workers aged 22-31 not born in the US will drop to half its 2024 value[12]. As a result, as shown in Figure 7, the number of available STEMM workers aged 22—31 will drop by about half a million people from 2024 to 2034. This is due to the projected drop in high-skilled immigrants, and due to the projected slow turnover in the pool of STEMM ready students coming from high school.
Figure 8: Projection of available Young High-Skill STEMM Workers compared with projected need from the Bureau of Labor Statistics.
The problem with this scenario, as shown in Figure 8 is that the need for STEMM workers is anticipated to rise. The Bureau of Labor Statistics (BLS) anticipates an overall rise in STEM and health occupations of around 8%[13] from 2024 to 2034. This seems conservative for the highly educated population considered here, as an 8% rise over the next decade falls well below the trend of the last 15 years, which would lead to 50% growth over the decade. Even so, this model predicts shortages will begin to emerge in 2028, and by 2034 the US will be short around 800,000 Young High-Skill STEMM workers, unable to fill the shortages even if 100% of Young STEMM Degree Holders enter STEMM occupations.
Because the trajectory of STEMM jobs has been so consistent, it is worth asking whether it is more plausible that the trajectory will continue, or that the lower BLS prediction will come to pass. The BLS uses a complex set of interlinked models to project job growth[14]. These are much more sophisticated than a naïve projection. But it seems fair to ask how well their methods have worked in the past in this domain.
The BLS projected STEM job growth in 2014; in fact they prepared a substantial report intended for broad dissemination[15]. A summary of their projections for various sections appears in a chart[16]. They predicted that from 2014 to 2024 STEM jobs would grow by 9% and healthcare practitioners & technical medical workers would grow by 16.4%, for an aggregate growth rate of around 12%, or 1.2% per year. Their projection concerned the full range of workers in the occupations, not limiting them to an age range and education level as I have done. This partly explains the discrepancy between the growth rate they projected and the actual growth in jobs for young high-skill STEMM workers. Thus naïve extrapolation in 2014 would have produced a much more accurate picture of the growth in this sector than using the BLS projections, as shown in Figure 9. This therefore leads me to propose the projection of a gap between young high-skill STEMM worker supply and demand shown in Figure 10. The BLS projection can be viewed as a lower bound.
Figure 9: Comparison of 2014 BLS projected growth in STEM and medical professions with employment data.
Figure 10: Projection of demand for young high-skill STEM, medicine, and health professions workers through 2034 compared with domestic supply estimated from high school calculus enrollments. The future contribution of young high-skill international workers, historically 25% of the workforce, is left as a question mark.
It is possible that AI will so reduce the need for STEMM workers that the growth trajectory radically alters. Or that high school calculus enrollment will stop predicting the college STEMM population as it has for the last 15 years. However it seems prudent to ask if the ascending trajectory of STEMM-ready high school students the US achieved through 2016 can be recovered, and if accomplishing this goal requires renewed attention to preparing and supporting secondary mathematics and science teachers.
Figure 11: Data on STEMM occupations, again restricted to those with a BA and above, but now without regard to age, showing aggregate income received by people in different occupation groups.
Before leaving the subject of workforce, I want to cast the topic in a somewhat different light. Instead of focusing on what industry needs, let us focus on what people receive. This is shown in Figure 11. The point I want to make is that the STEMM occupations have provided a means for millions of people to earn trillions in income, and this source of distributed wealth has been growing rapidly. This is one way to quantify how encouraging ambitious young people to go to college has been paying off. What happens if these lines stop going upward?
Teacher Supply
In estimating the future supply of highly educated STEMM workers, high school calculus emerges as a key indicator. It gains young people admission to college STEM degrees, prepares them to succeed, and correlates well over the past 15 years with STEMM degrees issued and STEMM workforce participation.
On its current trajectory, the US will see a slow decline in the number of high school students completing calculus. This is decline is not inevitable, but it will be difficult to reverse because of teacher shortages.
Here is the big picture of how the numbers work out. Over the next decade we will be short 800,000 STEMM workers with a BA or above. To supply these, we need to increase by 1.25 million the number of students who take calculus. This is over a decade, so it is only 125,000 a year. On the one hand this is more than a 20% increase over those who took it in 2022. On the other hand it is not really a very large number, nor is it much above the 2016 peak, and with class sizes of 20 and 6 sections a day it can be accomplished by 1100 full-time equivalent teachers. Preparing this many new teachers is a task the US could accomplish in a year if we were determined to do so.
The US is far from having exhausted kids able to master calculus and proceed to advanced STEM training Figure 12 shows that the rate at which high-schoolers take calculus depends very strongly on school poverty concentration and school location. Students in urban and suburban public schools with no families eligible for free and reduced lunch are more than three times as likely to take calculus as students in schools with 80% or more free and reduced lunch students. Even when poverty concentration is low there is a 10% gap between suburban/urban and rural/small town students. It is impossible to believe that these huge disparities result from fundamental differences in student capacity that will not respond to skilled instruction.
Figure 12: Percentage of high school students taking calculus as a function of school poverty concentration (percentage of students eligible for free and reduced lunch) and district locale. Data from Civil Rights Data Collection 2021-2022.
Will it be difficult to offer additional mathematics classes to boost the number of Calculus students in rural and low-income schools? Yes, it will be difficult because of mathematics teacher shortages, and the shortages will be growing.
To understand why, let us look at the years of experience distribution of middle and high school mathematics teachers in the US. This is shown in Figure 13. The average number of years of experience is 12.5.
Figure 13: Distribution by years of school experience of middle and high school mathematics teachers. Ribbon gives 95% confidence interval. Data from National Teacher and Principal Survey 2020-2021
We can use Figure 13 to calculate the number of new teachers needed to keep the teacher population stable. The number of mathematics teachers entering the profession is given by those with one year of experience. This comes out to 8900 teachers with a margin of error (95% confidence) of about plus or minus 900.
Compare now with Figure 14. The US prepared more than 8000 mathematics teachers per year until 2014-2015, and then the number slipped below that. In 2022-2023, the number of mathematics certificates awarded (6746) was well below the lower bound (8800) of teachers entering the secondary mathematics teaching profession. We are no longer preparing as many mathematics teachers as are starting to teach.
Figure 14: Number of mathematics certificates issued per year in the United States, 2010-2011 through 2022-2023 and the number of all teachers listing mathematics as their primary content area in their first year for three selected years. Data from Title II data collection and from the National Teacher and Principal Survey.
Figure 15: Distribution by age of beginning middle and high school mathematics teachers, those with one to five years of experience. National Teacher and Principal Survey , 2020-2021.
How can this be? One possibility is that we are drawing on the past. From Figure 15 we see that 30% of new mathematics teachers enter teaching at age 32 or higher, 10 years or more after they finished college. Back in 2010-2011 there were thousands more people who obtained mathematics teaching certificates than have been entering the profession. We can still draw on that pool, which reached a peak in 2015 of 12,000 excess mathematics teachers certified since 2011. If this picture is right, by 2023 the pool was drawn down to 6000, and by 2030 it will be gone.
What about high school science? The picture is a little harder to draw because there is not a single teaching certification for science that is as clean as certification in mathematics. Since physics is the science subject districts often say is the hardest to staff, I compared new certificates issued for physics and general science (which usually includes physics) and new teachers identifying their assignments as physics or general science. As shown in Figure 16, as with mathematics, it appears that school were hiring more teachers in these areas by 2016 than educator preparation programs were certifying. Since physics is optional (opt-in) for most high school graduation plans, teacher shortages can be masked by reductions in course offerings.
Figure 16: Number of physics and general science certificates issued per year in the United States, 2010-2011 through 2022-2023 and the number of all teachers listing physics or general science as their primary content area in their first year for three selected years. Data from Title II data collection and from the National Teacher and Principal Survey.
Looking back and forward
Figure 17: Total new teaching certificates for US from 1999-2000 onward and total new STEM teaching certificates issued from 2010-2011 onward. Data from Title II Data Collection. STEM teacher numbers not available prior to 2010-2011.
Wind time back to 2005, when Rising Above the Gathering Story advocated preparing 10,000 STEM teachers for 10 million minds. This was aspirational. The country did not fully rise to the challenge. Yet there were many positive developments. The National Science Foundation created Noyce Scholarships, UTeach began expanding nationally, the American Physical Society and American Association of Physics Teachers created PhysTEC, and numerous additional organizations including National Math and Science Initiative, the Science and Mathematics Teacher Imperative, and Change the Equation advocated increasing the numbers of science and mathematics teachers. Interest reached its peak in 2010 with the release of the second edition of the Gathering Storm report, the formulation of a 10-year goal to prepare 100,000 new STEM teachers, and the founding of 100K in 10 to accomplish this goal. This energy and interest certainly played a role in leading the number of students enrolled in calculus to the peak Figure 3 shows was achieved in 2016.
Yet in retrospect we also see from Figure 17 that the peak of interest in preparing STEM teachers was also a peak in the numbers prepared. Over the decade starting with 2010 the number of STEM teachers prepared per year in the US dropped by more than half. We did prepare more than 100K teachers total, and thus met the goals of 100K in 10, but this was accomplished while production fell. Disaster did not strike, certainly not immediately. One explanation for this is that national data were incomplete; only in 2010-2011 did a national tally of new STEM certificates issued become available. Thus it was not really possible in 2005 or 2010 to know quite how many teachers we needed, and how precisely they would affect the future workforce. In retrospect the country was on a fairly good trajectory, doubling the pool of STEM-ready high school students between 2000 and 2016, and more than doubling the number of STEMM degrees and early career STEMM workers between 2010 and 2024.
We have fallen off this trajectory, but not by far. We need a course correction, not a revolution. Increasing preparation of mathematics teachers by 1000 to 2000 per year should suffice at least to maintain the status quo, and an additional 1000 to 2000 teachers, most easily obtained by retention measures to increase by around 10% the median years of teacher experience, would support an emphasis on calculus and STEMM professions sufficient to ward off growing workforce shortages.
Figure 16 shows that a similar course correction in teacher preparation and support is needed elsewhere in STEM, particularly in physics, chemistry, and computer science. We know more than we did 20 years ago about preparing teachers and have much better data systems to monitor progress towards specific goals. Goals no longer need to be aspirational, unrealistic, celebrated and ignored. We can set realistic, achievable, goals for teacher preparation and support. Let us monitor the workforce needs of technology and health sectors, and prepare students accordingly. Over the longer run, let us plan for all US students to have a high school education that provides them STEMM opportunities if they choose to take them, with or without going to college, and determine how to prepare teachers to meet these goals.
UTeach started the movement to revitalize commitment of public universities to STEM teacher preparation back in 1997, and we have continued to work at this project, learning as we go, to the present day. The country’s need for what we uniquely know how to do is now clear.
More than 50 universities are in the UTeach network. We are collectively graduating about 500 STEM teachers a year, but every program should have the capacity to prepare at least 30, and reaching this practical goal would triple our numbers. If half of this increase were mathematics teachers, the UTeach network alone would be supplying half the national target to restore the High-Skill Young STEMM Workforce.
Allowing the United States to remain competitive in Science, Technology, Engineering, Mathematics, and Medicine is completely achievable, but we have to prepare and retain teachers to enable this and we must do so now.
[1] The precise occupation codes I used are the ones highlighted in green and yellow as STEM and STEM-related in https://www2.census.gov/programs-surveys/demo/guidance/industry-occupation/2018-census-stem-related-and-non-stem-occupation-code-list.xlsx
[2] For a detailed presentation of evidence that worries about workforce shortages are overblown, see Peter H. Cappelli, “Skill Gaps, Skill Shortages, and Skill Mismatches: Evidence and Arguments for the United States,” Industrial and Labor Relations Review 68, no. 2 (2015), https://doi.org/10.1177/0019793914564961.
[3] I know even from personal experience that this definition of the STEMM pool is not perfect. My high school did not offer AP math and I did not have calculus in high school and I became a physicist . I will use this proxy anyway.
[4] Chi-Ning Chang et al., “Predicting STEM Major Choice: A Machine Learning Classification and Regression Tree Approach,” Journal for STEM Education Research 6, no. 2 (2023): 358–74, https://doi.org/10.1007/s41979-023-00099-5.
[5] “Calculus or Statistics: Does It Matter?,” The Thomas B. Fordham Institute, accessed March 7, 2026, https://fordhaminstitute.org/national/research/calculus-or-statistics-does-it-matter.
[6] Janet Johnson, “The Algebra Gatekeepers,” February 10, 2025, https://www.educationprogress.org/p/the-algebra-gatekeepers.
[7] Semiconductor Industry Association, “Chipping Away: Assessing and Addressing the Labor Market Gap Facing the U.S. Semiconductor Industry,” Semiconductor Industry Association, July 20, 2023, https://www.semiconductors.org/chipping-away-assessing-and-addressing-the-labor-market-gap-facing-the-u-s-semiconductor-industry/.
[8] Emma Nyhof McLeod et al., Bridging the Middle-Skills Gap: Connecting a Diverse Workforce to Economic Opportunity Through Certificates and Associate’s Degrees (Georgetown University Center on Education and the Workforce, 2025), https://cew.georgetown.edu/wp-content/uploads/cew-bridging_the_middle-skills_gap-fr.pdf.
[9] “Tear The Paper Ceiling,” accessed March 11, 2026, https://www.tearthepaperceiling.org/. argues against credentials. I favor them at least for teachers and doctors . Since Texas tried a large-scale experiment removing the need for teaching certificates for teachers we have specific evidence of the negative consequences of dropping that required credential. See M. Marder et al., “Beyond the Tipping Point: The Rise of the Uncertified Teachers in Texas: Report and Recommendations from the University of Texas at Austin.,” September 9, 2024, https://bit.ly/uncert-teachers-report.
[10] According to https://crdc.communities.ed.gov/sites/default/files/2024-12/Courses%20and%20Classes%20%28COUR%29%20Module_1.pdf schools are supposed to report dual-enrollment calculus students both in the calculus table and the dual-enrollment table. Therefore if reporting is done properly, the drop does not reflect siphoning off students for dual enrollment.
[11] https://www.census.gov/data/datasets/2023/demo/popproj/2023-popproj.html
[12] Despite alarming news stories there is not yet evidence of much of a drop in international student enrollment in the US in 25-25. Overall total international student enrollment is down only 1%. https://iie.widen.net/s/hp5vgzgbpp/iie_fall-2025-snapshot_key-findings Graduate student enrollment is however down 12%, and these are people most likely to be seeking STEMM jobs.
[13] https://www.bls.gov/emp/tables/stem-employment.htm. https://www.bls.gov/emp/tables/emp-by-major-occupational-group.htm
[14] https://www.bls.gov/emp/documentation/projections-methods.htm
[15] https://www.bls.gov/spotlight/2017/science-technology-engineering-and-mathematics-stem-occupations-past-present-and-future/
[16] https://www.bls.gov/opub/ted/2015/employment-projections-for-occupational-groups-2014-to-2024.htm