A lot of angst has been devoted to discussing the skills gap—the large and growing gap between the skills workers hold and those sought by employers. Reports have been written quantifying the skills gap, how it’s a drag on the economy, or estimating just how much the economy might boom if we manage to close it. One estimate has closing the gap as adding US$11.5 trillion to global GDP by 2028.
Think pieces and TED talks promote “the skills you need”. Training schemes and education incentives have been implemented to reskill workers, providing them with new skills, skills more relevant to a digital age. Despite these efforts, the skills gap continues to grow. It’s now ‘catastrophic’, a ‘critical issue’ for educators, employers and government.
The problem is that the more we’re looked into the skills gap, the less we’re convinced that it’s a problem. This isn’t to say that there aren’t challenges we must overcome as we sail into our digital future—it’s just that the skills gap doesn’t seem to be one of them.
My suspicions were first tweaked during the work for The digital-ready worker,[ref]Evans-Greenwood, P, Patston, T, & Flouch, A 2019, ‘The digital-ready worker: Digital agency and the pursuit of productivity’, Deloitte Insights, <https://www2.deloitte.com/us/en/insights/focus/technology-and-the-future-of-work/learned-helplessness-workforce.html>.[/ref] where one of the key insights was that what we were seeing wasn’t a lack of skills, but workers’ inability to understand when to apply the skills they already had. Workers were suffering from learned helpless.
As it was put in that essay:
The worker has learned, through many interactions with digital tools and technologies, that these tools are only to be used in particular ways to solve particular problems. Experimenting with different ways of using them often leads to unfortunate consequences: confusion, failure, or even a “bricked” device. This reinforces the natural tendency to stick to known, habitual, “safe” tools and methods of use. After accumulating many such experiences, a worker may come to believe themselves incapable of navigating the complexities of a new digital tool, or even the digital workplace in general, without being explicitly taught how to do so—and, consequently, give up even trying.The digital-ready worker, p2
That insight triggered a deeper investigation into the skills gap: how we measure it, what might be causing it, and so on. The more we dug, though, the less it seemed that the skills gap was a thing, and the more it seemed to be a distraction preventing us from seeing the real problem. Like phrenology, theory had come detached from reality and was leading us astray.
There’s four arguments against the skills gap, which we’ll summarise below.
First is an economic argument.[ref]Cappelli, PH 2015, ‘Skill Gaps, Skill Shortages, and Skill Mismatches: Evidence and Arguments for the United States’, ILR Review, vol. 68, no. 2, pp. 251–290, viewed 26 November 2019, <http://journals.sagepub.com/doi/10.1177/0019793914564961>.[/ref] This views the skills gap a measuring an input—a shortage in one of the factors of production—and tries to tally that perceived shortage with evidence of the effects of the shortage. These other effects seem to be missing in action though. While employers report a skills gap, they are not paying more for these in-demand skills (i.e. a skills shortage has not resulted in higher wages for those with in-demand skills), nor are they reporting an inability to fill vacancies due to a lack of skills. This point of view attributes the perceived skills gap to a growing tendency for employers to seek much higher-level credentials than are actually needed for a job. (And over credentialization is a real and growing problem.) This is amplified by the growing tendency for employers to higher from outside the firm, rather than promote from within. A combination of over credentialisation and relative increase in external hiring are driving a perceived growing deficiency in skills.
Next are the sociologists, who have two arguments against the skills gap.
The first argument is against the methodologies used to measure the skills. Typically measurement means analysis of skills data in O*NET to quantify the gap between the skills provide to workers via formal education and training, and the skills that employers are seeking. The problem with this approach is that it is built on shifting sands. The skills captured in O*NET represent fluid concepts whos’ definitions and use has evolved over time.[ref]Payne, J 2017,‘The Changing Meaning of Skill’, in J Buchanan et al. (eds.), The Oxford Handbook of Skills and Training, Oxford University Press, viewed 10 December 2019, <http://oxfordhandbooks.com/view/10.1093/oxfordhb/9780199655366.001.0001/oxfordhb-9780199655366-e-3>.[/ref] Even what we to consider to be (and not be) a skills has evolved, and things that were not previously considered a skill now are. From this point of view, what the skills gap is measuring is concept drift over time.
The second argument is more fundamental.
The skills gap is based on the concept of skills upgrading: new technology is introduced into the workplace to automate simpler, lower-level skills, forcing workers to upgrade and focus on more complex higher-level skills. There are good examples of this in the literature, the best being telephone operators where automation made worked its way from simple local exchanges up to more complex regional and then international ones.
Skill upgrading, however, is only one possible pathway by which technology can shape work, and it’s not the dominant pathway.[ref]Spenner, KI 1983, ‘Deciphering Prometheus: Temporal Change in the Skill Level of Work’, American Sociological Review, vol. 48, no. 6, p. 824, <http://www.jstor.org/stable/2095328>.[/ref] Another obvious pathway is skill downgrading, where automation replaces the higher-level skills and delegates the lower ones to workers. Think autonomous cars and their safety drivers, or pick-n-pack workers in a warehouse working under the instruction of a inventory management system. We called this “working for a computer” in The new division of labour.[ref]Evans-Greenwood, P, Hillard, R, & Marshall, A 2019, ‘The new division of labor: On our evolving relationship with technology’, Deloitte Insights, <https://www2.deloitte.com/insights/us/en/focus/technology-and-the-future-of-work/the-new-division-of-labor.html>.[/ref] The dominant pathway in the qualitative sociological evidence, though, is muddling through where technology changes some tasks while replacing others (via both upgrading and downgrading), and the worker adapts. A great example is how digital vinyl printing transformed the sign writing profession during the 90s to 2000s without displacing any workers.
Unfortunately, only skill upgrading is visible in skills data sets, such as O*NET, due to their nature, and so only part of the effect of technology on work is visible in the measured skills gap. While skills upgrading may (or may not) be a problem, it will be dwarfed by other effects (which also may or may not also be problems) and so tells us little.
Finally, the last argument is from research psychologists who trying to make sense of skill and human performance. Their point is that while skill is an intuitive term, one that makes common sense, our common sense is misleading us. Consider transfer learning, which swings in and out of fashion, where a worker can learn a skill at one task and then transfer it to a similar task. This is usually not the case. We don’t learn behaviours, skills, but behaviours in context,[ref]Serrien, B et al. 2017, ‘Changes in balance coordination and transfer to an unlearned balance task after slackline training: a self-organizing map analysis’, Experimental Brain Research, vol. 235, no. 11, pp. 3427–3436, <http://link.springer.com/10.1007/s00221-017-5072-7>.[/ref] and skills can rarely transfer between contexts. There’s two reasons for this:
- Motor abundance[ref]Latash, ML 2012, ‘The bliss (not the problem) of motor abundance (not redundancy)’, Experimental Brain Research, vol. 217, no. 1, pp. 1–5, viewed 10 June 2021, <http://link.springer.com/10.1007/s00221-012-3000-4>.[/ref] there is no correct way to perform a skill, we have many ways to achieve the same thing.
- The extended mind[ref]Clarke, A & Chalmers, D 1998, ‘The Extended Mind’, Anaylsis, vol. 58, no. 1, pp. 7–19, <https://era.ed.ac.uk/bitstream/handle/1842/1312/TheExtendedMind.pdf>.[/ref] means that we use the environment around us to think and deliberate, rather than computing answers in our head.
It’s the interaction between the abilities of the worker and the environment they find themselves in, rather than their ‘skills’, that matters. This nicely echoes what we found both on The digital-ready worker and Unshackling the creative business,[ref]Evans-Greenwood, P et al. 2021, ‘Unshackling the creative business: Breaking the tradeoff between creativity and efficiency’, Deloitte Insights, <https://www2.deloitte.com/us/en/insights/topics/innovation/unshackling-creativity-in-business.html>.[/ref] where context matters as much, if not more than the workers attributes, particularly in a networked world knowledge is often no more than a Google away.
Given these four arguments we concluded that measuring the skills gap is a bit like phrenology—our theory about human performance does not connect with reality, and while we can measure a gap it doesn’t seem to have any measurable effects. Consequently, efforts to close the skills gap—via STEM and education programmes—will not deliver the desired outcomes.
Where to from here? The obvious approach focus has been on teasing apart the interactions between worker, coworkers, workplace and work. If we can find a more appropriate way to frame the problem that’s niggling us, then we’re more likely to make progress addressing it. There’s a lot of interest research into areas like creativity that we can leverage, work that takes a more system-based approach to understand not just the worker, but also the interaction between worker, work, and workplace and how these all contribute to performance.