A STEM Workforce Debate

As a labor economist and director of a research institute, I am often asked to make forecasts about economic conditions. To be honest, I often demur because to forecast the economy means that nine times out of ten you will be wrong. Forecasting occupation demand is even more fraught because a dynamic economy will end up creating and destroying jobs at such a rapid pace as to be highly unpredictable. With that background, I read Ron Hira’s “Is There Really a STEM Workforce Shortage?” (Issues, Summer 2022) with great interest. While I mostly agree with Hira’s approach, there are aspects of this question that deserve a more nuanced discussion.

I wholeheartedly agree with Hira’s critique of Bureau of Labor Statistics Employment projections, and his conclusion that “technological disruptions, and their effects on employment, are notoriously difficult to predict.” Exhibit A is the impact of COVID-19 on the labor market. In February 2020, the United States had 152.5 million people employed, and by April 2020 over 20 million were out of work. It was only in August 2022 that employment finally exceeded February 2020 levels. These kind of employment shocks are not anticipated and are difficult to incorporate in models. Like Hira, I recommend that people take employment projections with a grain of salt.

That said, the aftermath of the COVID-19 pandemic has created an unprecedented labor shortage. According to the Bureau of Labor Statistics, as of July 2022 there are two job openings per every unemployed worker. Recent data from the Indeed Hiring Lab suggest that the shortage of STEM workers is more acute than in other fields. Using Indeed’s dataset, which calculates the percentage change in job openings by selected occupations since the labor market peak of February 2020, job postings in STEM fields were up as of July 26, 2022, including for software engineers (93.7%), medical technicians (78.8%), and electrical engineers (81.1%). In contrast, jobs in business fields were up half as much—in insurance (60.9%), marketing (46.9%), and management (42.3%).

Finally, while the current shortage (or lack thereof) of STEM workers may be debatable, the lack of diversity in STEM occupations is not. According to data from the National Center for Science and Engineering Statistics, in 2019 only 29% of employed scientists and engineers were female and 15% were from historically underrepresented groups. The National Science Board’s Vision 2030 plan rightly focuses on the lack of diversity in STEM education and employment.

In the long run, unless all children receive access to a high-quality K-12 education, including sufficient coursework in mathematics and science that will prepare them to participate in STEM careers, demographic trends suggest that there will be fewer STEM workers. This lack of diversity may lead to a lack of discovery. A new study in the journal PNAS shows that gender diversity in research teams generates higher impact and more novel scientific discoveries, and the same has been found for ethnic diversity. STEM education is an investment in the nation’s economic future and should be available to all students regardless of race and gender.

Donna K. Ginther

Ron Hira’s article is an exercise in the giving of good advice. He is correct to advise us to demand better data that paint a fuller picture of the STEM labor market’s nuanced realities. He is also correct that we should make more honest and responsible use of the data we already have. But the advice that strikes me most strongly is the exhortation Hira leaves unwritten. Like most good advice, it can be summarized succinctly: follow the money.

Demanding better data about the STEM workforce raises the question of why we don’t have better data already. Aspiring to more truthful STEM workforce debates leads one to ask who has an interest in keeping the debates mired exactly where they are. Hira argues that we are not suffering the national STEM worker shortage our national discourse assumes; the bulk of his text is a point-by-point dismantling of the misuses of data that sustain this mistaken view. But the question of who benefits from the prevailing view is the heart of his argument, the moral undercurrent supporting his data deep-dive.

Hira’s own answer to that question is clear. He suggests that official statistics on how many STEM jobs are offshored every year would be useful, for example—and then reminds us that both the National Academy of Engineering and Congress have sought exactly this data from federal agencies, only to be thwarted by business interests. Hira recounts Microsoft president Brad Smith’s misuse of unemployment data to suggest a worker shortage in computer-related occupations, when there was in fact a surplus. He unpacks wage data to show that contrary to the higher wages a true shortage would prompt, STEM wages have been largely stagnant for years as employers increasingly meet their STEM labor needs through lower-paid contractors and the abuse of guest-worker programs, rather than through higher pay, a more diverse talent pool, and better professional development.

The larger story, then, is about commercial interests “controlling the narrative” on the meaning and role of labor, to the detriment of workers. The STEM workforce debate is just one instance of this systemic problem. For decades, the US policymaking apparatus has given itself over to an economic orthodoxy that treats labor as merely one factor of production among many, which capital is free to reshuffle, discard, downsize, lay off, or underpay as may be required to juice the bottom line. My organization, American Compass, argues that we would do better to recognize that workers are cocreators of economic value rather than merely commodities to be purchased. Hira’s argument points in a similar direction, and reminds us that more informed and constructive STEM workforce discussions will require honesty about whose interests are being served.

Chris Griswold

Long out of fashion, industrial policy has come back into vogue, amid bipartisan concerns over economic and military vulnerabilities in an intensifying sphere of global competition. Underlying much of this discussion is the fear that America lacks sufficient STEM talent to carry forward its legacy of technological innovation and to maintain its lead over China. In his article, Ron Hira raises important questions about whether such concerns are supported by the facts. 

Hira acknowledges that the lack of availability of detailed data represents a constraint to more effective analysis of imbalances between the supply and demand of STEM talent. As he points out, traditional public data only allow for analysis at the aggregate level, and typically only through a sectoral lens. Just as in any field, STEM roles differ in the skills they require and, correspondingly, in the availability of needed talent, as illustrated in Hira’s article by the contrast between life scientists and software engineers. In the same way that a sectoral lens is insufficient to analyze labor shortages for specific STEM roles, looking only at categories of STEM roles is insufficient to analyze the availability of specific skills in demand in the market. There is no single “skills gap” in the market, but rather different gaps for different skills. Overall, conferred degrees and employer demand in what the Bureau of Labor Statistics refers to as “computing and mathematics” occupations may be in balance, but the pipeline for specific talent can still be severely anemic at the level of specific roles.

This is even more the case when we consider the question of whether existing programs of study are aligned to industry demand at the skill level. For example, while universities may be conferring more than enough STEM degrees to meet demand at the categorical level, these university programs may not be teaching enough of the specific skills that are required by industry, whether those be technical skills such as cloud architecture or soft skills like teamwork and collaboration. Significant gaps between skills taught and skills sought can be as problematic as broader imbalances—but less perceptible.

The assertion that supply and demand are in balance (or that the market is possibly even glutted) also depends on the notion that supply follows demand and not the other way around. There is an argument to be made that jobs follow talent in the knowledge economy. Rather than simply filling demand for STEM roles by entering the workforce, STEM graduates can also launch enterprises, create new products, or drive innovations that ultimately create greater demand for STEM skills. Although demand is never infinitely elastic, growing the strength of the STEM talent base is likely to stimulate demand correspondingly. Simply put, if America can reassert itself as a STEM talent hub, its innovation economy will grow, spurring further demand growth.

STEM is also a field with particularly high attrition—a phenomenon the economists David J. Deming and Kadeem L. Noray study in a recent analysis on “STEM Careers and the Changing Skill Requirements of Work.” According to their article, upon graduation, applied science majors enjoy a salary premium of 44% over their non-STEM peers. Ten years out, that shrinks to 14%. Because of the speed of skill replacement in STEM, STEM workers are less likely to enjoy an experience premium. By the time they have acquired significant on-the-job experience, many of the skills they acquired during their education are no longer seen as relevant. This high rate of skill replacement leads to a loss of the skill premium evident immediately after graduation. Accordingly, many ultimately leave STEM roles in order to continue their career progression. Given these defections, a straight demand-graduate analysis could understate gaps in the market, as assumptions about the number of new graduates needed to meet market demand must consider higher attrition of existing workers and not only new jobs created. 

Hira is correct that there is a need to revisit old assumptions. New, more granular, more timely data sources will afford decisionmakers a more precise awareness of the nature of current and emerging talent gaps and provide a more effective basis for action.

Matt Sigelman

Ron Hira asks whether there is really a STEM workforce shortage and, while noting differences by field, largely answers no.

I largely disagree, but also think that “shortage” is the wrong way to think about whether the United States has enough scientists and engineers. Markets tend to clear. There is neither a fixed number of positions for scientists and engineers in the labor force nor a fixed number of ways to use people with that training.

The better policy issues are whether we would benefit from more scientists and engineers and whether people receiving that training have rewarding careers. With a few exceptions, there is overwhelming evidence that the answer is yes to both. Drawing on data from the National Survey of College Graduates, the National Science Foundation Science and Engineering Indicators, and the US Bureau of Labor Statistics, we find:

• 85% of recent science and engineering (S&E) graduates say their jobs are related to their degrees: 80% at the BS level and 97% at the PhD level, measured one to five years after degree.
• Employment in S&E occupations by individuals with bachelor’s degree and above grew by 39% between 2010 and 2019, more than five times the growth rate of the labor force.
• Degree production in S&E fields grew slightly slower than growth in S&E occupational employment—by 38% at the bachelor’s degree level and 30% at the PhD level.
• Unemployment rates in S&E are low. Average unemployment in 2021 was 2.4% for computer and mathematical occupations; 3.3% for architectural and engineering occupations; and 2.2% for life, physical, and social science occupations.
• Pay is high for recent graduates in most S&E fields—and rising. For bachelor’s degree recipients one to five years after their degree, average salary in private industry was $61,242 in 2019. This ranged from $44,910 for physical science to $76,368 for computer and mathematical science. For recent PhD holders, average salary was $115,000.

Differences in our conclusions come from different treatment of occupation data. Counts of jobs in STEM occupations should not be compared with headcounts of degree. Many people with bachelor’s-level STEM degrees pursue careers in other fields, such as law and medicine. Many new PhDs have student visas and may not want or be able to stay. Also, many S&E graduates who report that they are doing work related to their degree are not in formal S&E occupations.

I also disagree about the meaning of changes in occupational wages as an indicator of the labor market value of skills. If the average wages for PhD geoscientists in industry were to fall from its $184,000 average, would that mean the skill is not in high demand? Would society be better off with fewer people with that skill?

Changes in occupational wages are not even a good measure of changes in the demand for skills—fast-growing occupations grow fast by bringing in people with less direct training, less education, and less experience. For this reason, the average wage rate often falls in fast-growing occupations.

There are many career-path issues worthy of policy concern, such as whether researchers in a particular field are too old when they receive their first independent grant, or whether older programmers have problems finding new jobs? But limiting the supply of talent, either by immigration rules or education policy, is a blunt policy tool that may have little effect on such issues. And it is probably not a good deal for the scientists and engineers who do remain. Rather than enhancing careers by reducing “competition,” much R&D activity would leave the United States or not take place at all.

Mark Regets

Ron Hira’s article presents a valid challenge to the long-standing argument that there is a STEM workforce shortage in the United States and causes us to reconsider the premise of decades of STEM education policies and initiatives that are based on the “shortage” argument. Hira proposes that this argument has not only been unsubstantiated, but is based on often flawed, incomplete, and misinterpreted data.

As an African American female chemist and social scientist who comes from a low-income and first-generation background, the critical importance of broadening participation in STEM is paramount. In order for the United States to remain competitive in a global science and technology driven economy, we must engage all of our human capital—particularly those like myself who have been historically disenfranchised and discouraged from scientific pursuits. However, as a STEM policy adviser, I am also keenly aware that STEM policy is shaped by not only by data, but by public sentiment, perception, and stakeholder voices who are the loudest. As Hira posits, voices such as those from students who are the target of these policies, and workers who are the end product of these efforts, are often excluded.

Hira lays out several factors that have influenced this notion of a STEM workforce shortage and how these factors have been based on limited data and the exclusion of dynamic processes and situational caveats. He correctly asserts that employment projections, which are based on current trends that are extrapolated out, may be applicable to occupations with stable trends, such as the legal field, but this method is inadequate for occupations that do not have stable trends, such as the computer sciences. Moreover, as Hira notes, a seemingly low unemployment rate in STEM fields, which evidence shows is actually high due to miscalculations based on comparisons with the national employment rate—a composite across all labor markets—has created inaccurate estimations. Errors of this type can lead to regressive impacts on progressive efforts related to inclusive outreach, recruiting, and hiring policies as organizations rely on these rates, projections, and forecasts to formulate staffing budgets.

Overall, Hira presents a sound, documented argument that the decades-long perception of a STEM workforce shortage in the United States is based on unsubstantiated evidence and flawed data and is often driven by stakeholders who do not necessarily advocate for current or future US STEM workers. Hira lays the foundation to seed discourse on a real and transformative conversation not only about the validity of a STEM workforce shortage, but more importantly about the implications for policy and for current and prospective US STEM professionals. Examining the layered, multifaceted, and cross-sectional mitigating factors at play, informed by analyzing disaggregated data on the persistent unemployment, underemployment, wage disparity, and barriers that impact minoritized groups such as BIPOC and persons with disabilities, who are an untapped source of US STEM talent, would greatly enhance this new conversation that is needed.

Iris R. Wagstaff