I think I've been hearing about "skills shortages" for as long as I've been in the computer industry, yet at various times heard anecdotes from highly-qualified and experienced software developers (including me) that finding work takes many months. Cries of "skills shortage" have grown even louder since international borders were closed due to the COVID-19 pandemic; yet even now computing students sometimes tell me how difficult finding work can be and the Australian Computer Society recently published its own collection of anecdotes from migrants supposedly brought into Australia to combat skills shortages yet struggle to find work.
Peter Cappelli, a management professor at the University of Pennsylvania, investigated this phenomenon in the wake of the Global Financial Crisis in Why Good People Can't Get Jobs (2012). He dismisses the explanations frequently offered in popular media: there aren't enough skilled workers (many university graduates work in positions that do not use their skills); universities are teaching the wrong skills (employers predominantly complain about lack of experience not lack of academic credentials); would-be employees want too much money (that's a market economy for you); and would-be employees are not willing to move to where the jobs are (relocating a family comes at enormous cost with uncertain rewards). Other academic investigations into popular claims of a shortage of science and technology skills have concluded that the STEM crisis is a myth (Charette, 2013) and humanities graduates earn more than those who study science and maths (Hurley, 2020)—though, with regard to the last, medicine and engineering graduates are far ahead of both humanities and science graduates.
Shopping for skills
Cappelli describes a number of phenomena that mean the labour market doesn't work like the standard model of a market, and uses the computing industry as a particularly egregious example. In what Cappelli calls the "Home Depot" (an Australian might say "Bunnings") view of filling job vacancies, employers have a shopping list of skills that they require to fill a space in their organisation; they look amongst the available workers for one with the matching combination of skills; and they slot this worker into the space. But the number of skilled workers cannot be scaled up and down according to demand in the way that mass-produced items of hardware can be.
The computing industry in particular prides itself on how quickly technology changes, with programming frameworks and hardware platforms changing from year to year, and exciting new paradigms appearing every five years or so. In this environment, he says, expecting to browse the shelves for exactly the right part (worker) is like designing a car whose requirements for an engine change unpredictably from year to year and expecting that a suitable engine will appear on the market just in time to drive the car away. Yet Silicon Valley (he says) has championed the "free agent" model of employment in which computing workers are thought to appear in exactly this way.
Instead, he says, employing organisations should pay attention to how the necessary skills will be developed, just as a car manufacturer with particular needs for an engine would take care that its suppliers could produce one. This might be by forming partnerships with universities or other training centres, working with government or industry associations to develop apprenticeships, and/or training existing employees rather than expecting the outside world to produce skills at need.
Of course training and related activities cost significant amounts of time and money, so organisations tend to prefer finding someone ready to go while workers tend to prefer earning money to paying to study. So each side stares off the other, hoping the other (or the government) will pay for the necessary training.
Hiring by the numbers
Cappelli's other gripe is with the hiring process. On one hand, Internet job boards have allowed employers to reach large numbers of would-be employees; but this means sorting through very large numbers of applications. To sort the deluge, recruiters resort to automated matching systems and hiring managers resort to specifying long lists of skills intended to whittle down the candidates to those with the exact skills required for the vacancy. But these practices tend to a hunt for "purple squirrels" or "unicorns", combining a long list of skills that no one person is likely to have.
In my own experience, the computing industry is again an egregious example. The typical advertisement for a software development position is a long list of technology products that the employing organisation happens to use, making searching for work rather like a game of buzzword bingo. Only someone already working in the position is likely to have that particular combination of technology, but the whole point of recruitment is to find someone who isn't in the organisation already.
Several academic authors have proposed systems for describing the skills required of software developers in a more generic way (Furtmueller and colleagues 2011; Moustrafas et al. 2015), but these approaches don't appear to have caught on. In the meantime, we come back to the problem of training: maybe a blue squirrel with experience in Java, say, can readily acquire a purple hue by a short course in C#—but who's going to pay for the course?
Finding or creating skills
From the news articles I've read, what I've heard from my students, and what I've seen myself on job boards, "skills shortages" remain more or less as Cappelli diagnosed them. Cappelli points to a few US organisations who filled their skills gaps with training programmes, mentions the strong apprenticeship programmes that exist in countries like Germany and Switzerland, and alludes to union-sponsored training programmes that existed before unions were largely driven out of the private sector in English-speaking countries. But noises from the computing industry, such as from the newly-formed Tech Council of Australia, remain largely about skills appearing from government, from migrants, or anywhere but the computing industry; and looking for work remains a game of buzzword bingo.
References
Peter Cappelli. Why Good People Can't Get Jobs, Wharton School Press, 2012.
Robert Charette. The STEM crisis is a myth, IEEE Spectrum 50(9), September 2013, pages 44-59.
Elfie Furtmueller, Celeste Wilderom and Mary Tate. Managing recruitment and selection in the Digital Age: e-HRM and resumes, Human Systems Management 30, 2011, pages 243-259.
Peter Hurley. Humanities graduates earn more than those who study science and maths, The Conversation, 19 June 2020.
Evangelos Moustroufas, Ioannis Stamelos and Lefteris Angelis. Competency profiling for software engineers: literature review and a new model, Proceedings of the 19th Panhellenic Conference on Informatics, October 2015, pages 235–240.