In promoting the government’s international women’s day STEM announcement, science minister Karen Andrews is reported as stressing that ’75 per cent of the jobs in the fastest-growing industries need STEM skills’. This is a variant on a common claim. In the same article, it also appears as 75 per cent of all occupations, as it has here, here, here, here, here and in many other places. The industries version is less common but still a regular claim, for example here, here, and here.

The original source of this 75 per cent claim has always been hard to find. Some government documents cite this 2015 PwC report. But it does not have any data supporting this number, giving this 2011 article in the Journal of STEM Education as its source. Unfortunately the article does not substantiate the claim either, other than by pointing to this 2007 US Department of Education report. But the report does not mention the 75 per cent statistic at all, and so the trail goes cold.*

Even without checking the claim’s provenance it sounds suspect. How do we define fastest? Top 10, top 20, top half? It would be easy to manipulate the list to produce the desired result. And then there is the distinction between fastest growing (as a percentage of previous levels) and greatest growth (absolute numbers of jobs). Occupations can easily have very high percentage growth rates if they start from a low base, while in absolute terms adding many fewer jobs than larger occupations growing by a smaller annual percentage.

When the Department of Employment last year tried to put some numbers on STEM trends (without mentioning the 75 per cent; a dangerous step when dealing with actual data) they highlighted the problem. On their analysis, by 2023

STEM occupations will grow by 10.8 per cent (271,300 people), whereas all non-STEM jobs are projected to grow by 6.1 per cent (614,900 people) over the same period.

In other words, 70 per cent of the new jobs will be in non-STEM occupations, which puts a different perspective on the need to boost STEM education.

A link to the list of STEM occupations used in this analysis raise further issues. It includes occupations that few people would think of as STEM, even if they do usually require IT literacy and some mathematical competence: accountants, economists, policy analysts, librarians and archivists.

It also includes all the major health occupations. While I think this is defensible in itself – there is a lot of applied science in the health professions – it is not how STEM is normally defined. Including health has a big effect on the estimates of STEM employment, since health care is the largest industry for employment.  Due to population growth and ageing, there is a high degree of certainty that it will continue growing. It is not speculative ‘jobs of the future’.

A chapter on STEM in Mapping Australian higher education 2016, which I wrote with Beni Cakitaki, called for scepticism and nuance when discussing STEM study and employment. It includes distinct fields and occupations on quite different trajectories, making STEM a not-very-helpful analytical grouping. The chapter identified a major over-supply of biological science graduates, along with issues in IT (the ‘T’ in STEM) and engineering (the ‘E’) markets.

The 2019 three-year out graduate survey confirms that there is no evidence of shortages of graduates in the core STEM fields.  More than a quarter of science and mathematics graduates reported not fully utilising their education and skills, and a fifth of IT and engineering graduates. By contrast in health and education, less than 10 per cent of graduates reported not using their skills and education.

Health and education are highly-feminised fields. Fortunately for their future employment prospects, more women are choosing courses that lead reliably to jobs than are following factoid claims about future STEM employment.

*  Update, 9/3/20: Reader Louise Hargreaves has identified a possible source for the 75 per cent claim. In February 2003 the US Department of Education launched a mathematics and science initiative. This included a concept paper, which on page 8 says ‘of the 20 fastest-growing occupations projected by the U.S. Bureau of Labor Statistics (BLS) through 2010, 15 of them require substantial mathematics or science preparation’. 15/20=75 per cent.

Reference checking this claim takes me to these estimates. It does not label its occupational growth estimates by their maths or science levels, but counting both IT occupations and higher-level health assistants I can get to 15/ top 20. It is not an unbroken trace back from contemporary 75 per cent claims, but it meets four of the criteria set by the clues we do have: it is about relative growth in occupations, it is based on a list, it is from the United States, and it is from before 2011.

Another table in the same BLS report shows that the list has the issue I identified above. If we rank occupations but their absolute rather than relative growth, STEM skills look less important.