It’s been an interesting day working on the newly released Longitudinal Educational Outcomes (LEO) data, and we are not even close to fully understanding the lessons from this rich and powerful dataset.
To do so will require more time, expertise, and a wider range of data to be released publicly, as well as made available privately to specialist researchers. Still, a day’s worth of insight suggests the following interesting lessons that should help us start asking the right questions about where to look next.
The magic number: £20,800
Call us crude, but the big number that a top-level analysis of this plethora of data will be compared to is probably £20,800. Why? Because according to the Office for National Statistics, this was the median salary for all 25-29 year olds in work in 2014-15.
Put simply, courses producing graduates over and above that figure five years from graduating can just about claim that they still deliver some sort of ‘graduate premium’ above the national average. Again, this comes with a whole host of caveats. But in the cruel, uncompromising view of the Treasury, courses producing graduates on salaries significantly below £20,800 five years after graduation may start getting questions about why the taxpayer’s subsidy (in the form of unpaid student loans) should be directed at them.
There may be extrinsic or intrinsic reasons for this. Courses that take on a higher proportion of women or non-white students are more likely to produce lower earning graduates. So immediate are the pay gaps for these groups that it’s almost impossible to put it down to anything other than outright racism and sexism in the labour market.
But there also clear patterns when it comes to certain sectors, most notably in the Creative Arts. More on that below.
HEDIIP and HECoS
If ever we needed more proof of how essential a new subject classification system is, this is it. Many of the subject comparison tables come with major caveats about what is being compared. For example, the subject grouping ‘Architecture, Building and Planning’ contains within it a wide variety of degree courses that can lead to very different career paths with very different earnings trajectories. Similarly for ‘Mass Communication and Documentation’, and also ‘Creative Arts and Design’. Institutions are known to have form when it comes to manipulating their subject codings in order to maximise metrics advantage.
Though LEO uses the JACS subject classification system, as we have covered on Wonkhe, this system is about to imminently change. A lot will depend on the makeup of the new ‘Common Aggregation Hierarchy’ (CAH) for future iterations.
Specialist courses that directly address local economic needs
Rolling through the results, one can pinpoint a couple of institutions producing high earners in subjects which directly address the very particular needs of local, high productivity industries. The most obvious case here is the University of Aberdeen, which produces the highest earning graduates in Engineering and Technology, no doubt due to its connections with the local oil industry.
No surprises here, but on the whole, the earnings potential is shown to be higher for STEM courses than for arts and humanities. The (very) broad message seems clear: the labour market continues to have greater demand for graduate skills in STEM than it does for arts and humanities. The problem for universities is that these are the most expensive courses to teach, often requiring cross-subsidy from cheaper courses in the arts.
The exception to the above is for Languages courses, not including English, which was separated from the former after the test release in December. The reason becomes clear when comparing the two subjects. While English graduates’ earnings tend to tail off, particularly at mid-to-lower tariff institutions, the returns on Languages courses are much stronger across the sector.
Extra caveats – such as a higher proportion of self-employed earners – apply when considering the earnings outcomes of Creative Arts and Design graduates. But nonetheless, the earnings outcomes for a number of creative courses, five years after graduating, make pretty bleak reading. Thirteen institutions have a median graduate salary at that time of under £17,000, below even the national median for young non-graduates.
STEM subjects – science, technology, engineering and mathematics/medicine – are usually grouped together in a homogenous as a shorthand for ‘sought-after skills’ or ‘well-paying careers’. But LEO shows this is rather too simple.
While Medicine, Engineering, Mathematics and Computer Science all appear to show evidence of high earnings potential and strong labour market demand, earnings outcomes for Physical Sciences and Biological Sciences, particularly at mid-to-lower tariff institutions, are much less strong. Indeed, Biological Sciences earnings outcomes were separated from Psychology in today’s release as the latter was ‘dragging’ it down even further.
The findings on the origins of the gender pay gap in the graduate labour market do not make for pretty reading. The gender pay gap appears as soon as students leave university, and only grows from there. The gap varies between graduates of different subjects, but nonetheless exists in all of them, most remarkably in Nursing, despite over 90% of nursing students being women.
As with so much that can be said about LEO, universities cannot be blamed for discrimination in the labour market. Indeed, universities have been successful at recruiting greater numbers of women than men, partly as a result of gender differences in school attainment, and partly due to careers that appeal more to women (including nursing and teaching) being more likely to require a degree.
A higher number of women entering higher education seems to be a vital first step to going someway to correcting the gender pay gap, but clearly it is not enough. University does not appear to ‘level the playing field’ for the genders. Nor indeed, does it do so for socio-economic classes or ethnicities. This should be seen as a direct challenge for university careers’ services and employability strategies. The problem is not necessarily their fault, but they are well positioned to make positive interventions.