As with most supposed panaceas, the narrative surrounding science, technology, engineering and math (STEM) subjects suggests that these fields, more than others, provide the critical skills necessary to ensure a ready supply of high-quality workers for the future. Due to the rapidly accelerating rate of technological change, so the thinking goes, STEM subjects must be the emphasis of educational efforts to train a globally competitive workforce moving into the middle of the 21st century.
Although workers in the future will interact with technology more than at present, the notion that STEM subjects should hold primacy in education is overstated. Furthermore, allocating the lion’s share of financial, educational and labor-training resources to STEM occupations is too simplistic an answer to the complex, ever-changing problems of labor-skill accumulation in an increasingly technological job market.
STEM Majors and the Tech Economy
STEM and Skill Turnover
One issue with emphasizing STEM skills relative to others is the high rate of skill turnover in most STEM fields. Proper education requires that the technological skills students learn not change for long enough that they can actually learn and then use them in practice.
As David Deming and Kadeem Noray found, however, the rate of skill turnover in STEM job listings was relatively higher than in more traditional fields. From 2007 to 2019, at least 29 percent of vacancies posted by the same firms for the same STEM-focused occupations contained at least one new skill requirement. Many listings required more. Moreover, computer- and math-intensive occupations’ skill-turnover rates were 47 percent compared to skill-turnover rates of 20 percent for job listings in education, law, and community and social services.
High rates of skill turnover require workers to start from scratch any time new tools take the place of old systems and technologies, whether they be programming languages, software, database architectures or communications media. The pace of technological change can quickly render existing skills outdated and previously acquired specializations effectively meaningless. In fact, the high rate of STEM skill obsolescence in technical roles has led workers in tech-intensive occupations to flee their fields.
High-ability workers in STEM fields who can more quickly adjust to a rapidly changing work environment face a dilemma. There are higher returns for accumulated experience in jobs that increasingly require abstract decision making. As a result, workers who can effectively adapt to constantly changing STEM skill requirements give up better pay by not accepting jobs in non-STEM fields that experience less skill turnover. Highly adaptable STEM workers can accumulate static skills more quickly than workers in these same fields who aren’t as adaptable.
Thus, the nimble-minded workers that STEM fields require are incentivized to take jobs in non-STEM occupations. Rather than facing the effort of constantly starting from scratch when STEM skills become obsolete, adaptable workers who are able to rapidly acquire new skills can quickly build up more on-the-job knowledge than their peers within occupations where skills do not change as rapidly. They stand to earn more money while also facing a less arduous career path.
This disparity is why, for instance, less than 50 percent of engineering graduates work as engineers. Conversely, STEM skills acquired during workers’ educational years provide an advantage when first entering the STEM job market, but the rate of skill turnover diminishes this advantage over time.
An Evolving Job Market
Another trend in the tech labor market that has exacerbated the flight of skilled workers, and particularly younger workers, away from STEM occupations further undermines the case for STEM skill primacy. Since the 2000s, the rising demand for skilled labor has been met with little or no employment growth in high-paying, skill-intensive jobs. This trend has been accompanied by decreasing returns to cognitive test scores in terms of employment and wage growth relative to the onset of the computing revolution 35 years ago. Moreover, these declines are almost exclusively concentrated in STEM-specific jobs.
These recent trends indicate that the types of tasks the labor market values from STEM graduates has changed. It is no longer the case that four years of a STEM major leads to a job that matches technical, cognitive skills to problems that they can directly address.
What jobs, then, experienced major boons? From 1980 to 2015, occupations requiring both cognitive skills and large numbers of social interactions experienced the largest gains in wage and employment growth. The increasing amount of workplace automation, which flattened “middle-skill,” heavily routinized forms of labor has played a large part in this shift. Positions that formerly consisted of abstract-task jobs like acquiring or collecting information, performing low-level analyses, and assembling it into coherent forms have all been automated. Occupations like research assistants and paralegals are recent examples of this trend.
Now, however, automation has shifted business labor needs to prioritize those positions that focus on the more abstract tasks of interpreting and applying insights that information provides. As will become clear, the problems that these types of positions encounter in the present day are often best addressed by teams of specialists rather than single workers.
Automation is brittle, however — it cannot adjust quickly and requires explicitly definable rules and processes. For this reason, machines can be either a substitute for or a complement to workplace tasks that meet those criteria. Machines handle sensorimotor tasks and complex problem-solving that combines different skills poorly though. For instance, software employed at barbershops can quickly coordinate online referrals, schedule appointments, collect information on what clients want from their barber, and charge the client when the appointment is complete. But machines cannot give someone a haircut.
When rules are implicit or hazy and problems are ill-defined, they cannot be solved by machines, which require definable, repeatable processes. Effective solutions to these wicked problems generally require breaking them down into component parts. Individual pieces of these variegated problems can then be met with a wide variety of different skills. Automation has freed human labor resources in high-skill industries to address these new and difficult types of problems.
As a result, a combination of cognitive skill and, increasingly, the ability of workers to engage with each other to “skill-trade” to address constituent parts of complex issues has increased the productivity of teamwork in the workplace. This is especially true of teams with a wider distribution of background training and skills. A high amount of sociability acts as a lubricant. It lowers the costs of task coordination and facilitates the advancement of diverse viewpoints that are critically important in unstable, idiosyncratic environments of the very sort in which machines fail spectacularly.
Stemming the Tide
How should education and recruitment for high-skilled occupations in increasingly tech-heavy workplaces look if STEM subjects do not provide the returns they once did? The first solution, as Deming and Noray suggest, is to recognize the trade-off between investments in STEM (or any specialized job-training skills) and general education. As their research shows, vocational preparation that STEM supplies can ensure a smoother transition from education into work.
Unfortunately, rapid technological change, especially in STEM fields, can quickly render these hard-earned skills null and void. Applicants to technical positions who don’t possess degrees with applied STEM majors or have non-traditional backgrounds but who do have a variety of experience or training (e.g., double majors in totally different subjects, various jobs worked or experience across different industries) could eventually outperform applicants from traditional STEM fields. More traditional candidates could prove either not flexible enough to adapt to changing workplace tasks over time, or they could leave for occupations with less job-task turnover. Interviews or job listings should target better candidates by seeking a wider variety of relevant traits as opposed to focusing solely on the years of experience with or number of current technological skills candidates possess.
Another potential solution is for employers to look at jobs in terms of the tasks comprising them and hire accordingly based on what workers are likely to do when those tasks change. For instance, many jobs in website development used to require experience with Adobe Flash. After Steve Jobs announced that Apple products would no longer support Flash in 2010, though, the demand for workers skilled in Flash plummeted.
Consequently, many younger workers or workers with wider skill sets quickly moved away from jobs requiring Flash. This shift led employers to have to pay premiums to workers with longer career horizons and those with wider skill sets not to leave Flash-intensive jobs. Older workers with shorter career horizons or who had deep knowledge of Adobe at the expense of other, more recent technical skills ended up not facing wage decreases even though they were unable to transition away from Flash jobs, however.
This example is important because it presented a potential arbitrage opportunity for labor-skill allocation. Businesses had an opportunity to maintain older workers in Flash-intensive jobs at the same rates of pay while transitioning younger workers who required premia to maintain Flash jobs into roles emphasizing new, emerging technologies.
Thus, targeting workers with different career horizons and skill breadths depending on how they react to changes in technology could make sense. For instance, older workers with shorter career horizons have little incentive to acquire new skills, whereas younger workers with longer careers ahead of them have the incentive to learn a variety of new skills but not to go too deeply into any one of them. Different mixes of skill depths and breadths can benefit teams by enhancing their abilities to focus on different tasks. This could mean hiring older workers on contract or 1099 status, maintaining workers at lower hours or on emeritus status for project work, or mixing together differently skilled workers in transition-focused teams to quickly adapt workplaces from old to new technologies.
Finally, employers should emphasize job candidates’ experiences with cooperative endeavors. Identifying attributes of workers that illustrate their collegiality, like participation in sports, clubs, community organizations or civic institutions, are excellent predictors of success on working teams. Similarly, the variety of group-project work accomplished in previous roles should be considered excellent resume boosters for applicants to high-skilled tech positions.
Potential solutions to the dual problems of hiring and maintaining an effective workforce are endlessly more complicated than simply subsidizing more students and young workers to go into STEM fields. While some STEM training can help lessen the frictions between recent graduates and the world of high-skilled labor, the rapidly changing technological landscape calls for adaptability and teamwork rather than routine-intensive and highly automatable technical skills. Employers will gain hiring and productivity advantages now and in the future by remembering the full version of the famous quote: “A jack of all trades and a master of none is oftentimes better than a master of one.”
