Do we need a labour theory of degree value?

A number of recent policy developments have brought me to question how we measure the value of a degree. Whether it is TEF, LEO, or the latest HEPI reports with the HEA and Unite Students, the question of value is at their core, and it is understood largely in terms of students’ expectations and outcomes, particularly those which relate to careers and salaries.

But there is a rather large problem with how we conceptualise value, particularly economic value, from this information. This problem is exacerbated by the growing uncertainty over the future shape, size, and requirements of the graduate workforce as a result of rapid technological change.

The unpredictability of value

The market model of choice and competition upon which our sector is now based asks students to consider what the value of their degree is long before the bulk of its value has materialised. We ask applicants to make value judgements about degrees they haven’t yet started based on the experiences of those who went before them, experiences which haven’t yet led anywhere.

This problem is made most clear by the ‘lag effect’ of Longitudinal Educational Outcomes data. LEO is claimed to offer applicants an understanding of the economic returns on a degree over time, whilst it gives us wonks a quantifiable value to a degree for a cohort of graduates as they find their way in the labour market. But LEO is making assumptions about the value of a degree to a future cohort of graduates based on another cohort who graduated some years before. The labour market can change considerably within the three years of a degree – there was a global financial meltdown at the end of mine – let alone in the time it takes to acquire decent longitudinal data.

Take medicine, right at the top of the pile of graduate earnings. The average starting salary of medicine graduates in graduate-level jobs in 2010 was 16% higher in real terms than it was in 2015. In dentistry, another high performer, the average starting salary had fallen 13% over the same period. In contrast, the average starting salaries for building and construction graduates rose 21% in cash terms and 5% in real terms between 2010 and 2015.

What is important to note about these figures is that when graduates in 2015 were choosing their courses back around 2011-12, the league tables would have been using the DLHE data for that 2010 cohort. How helpful was that information for students?

Information on the value of a degree is rather like information on investments. Whenever you decide to take out a stocks and shares ISA or change your pension you will see a familiar disclaimer: past performance is not an indicator of future success. Perhaps we should add this to Unistats and UCAS information for university applicants?

Degree value in an automated future

The labour market is changing rapidly. Technology-rich environments will increase demand for high-level skills in new areas, some of which we may well have not even thought of yet. This will likely mean that many young people currently starting out in school could end up in jobs which do not currently exist. It will also mean that current professions may have a very different career trajectory in the future.

Microsoft recently got leading technologists to explore what new jobs would be available within the next 10 years. Imagine graduates in 2025 entering work as a “virtual habitat designer” or a “digital cultural commentator”.

I doubt many university careers advice centres have much to say about the jobs of the future, although I should mention, with praise, the University of Kent’s careers service, which has a wealth of online information on this topic. Largely, however, the list of career possibilities in a course prospectus or in an open day talk will be those which graduates are currently entering. We’re stuck in the present when we should be considering what jobs will be there in the future.

Yet, on the other hand, many of the jobs that do get mentioned may no longer exist by the time the student celebrates their thirtieth birthday, or the tasks within them will have been altered dramatically by machine learning and artificial intelligence. One study has estimated that 35% of UK jobs are at high risk of automation over the next 10 to 20 years. A large portion of these jobs are low-skilled jobs, where the routine tasks can easily be taken up by machines. But as computers become smarter, they can take on more complex tasks which may affect many graduate jobs, from metallurgists to accountants, pharmacists to legal researchers.

Automation is not a new phenomenon. The invention of the seed drill in 1701 and the flying shuttle in 1733 both had a profound effect on agricultural labour and the textiles industry respectively, although, as with the current technological advances, these were not felt immediately due to financial, political and cultural factors that inhibited change.

But the new revolution is different, because it involves not the substitution of human physical labour for machines but the substitution of human mental labour for artificial intelligence. The Royal Society’s recent report on machine learning revealed an almost even split amongst experts as to whether the impact of artificial intelligence and automation would have a positive or negative impact on employment.

Regardless of the balance of new job creation over automation, what matters is that our education system must develop the skills necessary to ensure resilience to complex and rapid change. The Royal Society report states:

“While not necessarily replacing jobs or functions outright, machine learning will force us to think about our occupations, and the skills necessary to function in a world where these systems are ubiquitous.”

And yet, for the most part, our traditional degree courses are barely able to cope with understanding and delivering the right skills for employment today, let alone in the growingly unpredictable future. What is more, LEO has confirmed just how much of what we value about degrees has been based on the institution a graduate attended.  But, according to the latest CBI Education and Skills Survey, 87% of employers see the attitudes and aptitudes for work as the most important information when recruiting graduates, compared to just 13% who saw the university a graduate attended as important. Does this mean that we will see the variance in outcomes by institution begin to narrow in the future, or is there something more intrinsic about the value of a degree from a certain institution than reputation alone?

In the future, employers will be less likely to take for granted the capabilities of a graduate on the basis of their degree certificate alone. University name, subject studied, and classification are fairly opaque heuristic devices. It will be necessary for graduates to demonstrate far more clearly that they have the broad range of skills and experience that can help them to creatively adapt to technology-rich environments.

Are some degrees future-proof?

The answer is no; although no doubt some degrees will offer better protection than others.

If we match the graduate outcomes of particular degree subjects to the probability of those jobs being automated, we see subjects like economics and finance potentially hit, particularly in that long tail of more modestly paid graduate jobs in data analysis and accountancy. Law and politics are also more vulnerable because of many service jobs becoming automated. Even physics and maths have a high number of graduate jobs at high risk of automation.

On the other side, we find both subjects with currently favourable graduate outcomes, such as medicine and biological sciences, and subjects with fairly poor graduate outcomes in the creative arts, sports science, and English. What these subjects have in common is they tend to lead to jobs that require a level of creativity, cognitive ability, or manual dexterity that would be very difficult for computers to mimic.

But as I’ve already said, current outcomes are not a predictor of future outcomes. None of these degrees are future-proof. Even in medicine, tasks currently perforqmed by medical professionals across triage, diagnosis and even some basic non-invasive surgery, may well be assigned to computers and robots in the next decades.

Garbage in, garbage out

‘Garbage in, garbage out’ is commonly used in computer science to explain how the outputs of computers are limited by the quality and internal consistency of the data and code we input into them. Similarly, in the future, the quality and consistency of the holistic package of skills learnt in a degree will matter more to graduate outcomes than it currently does, as will the breadth and depth of skills that the student has already acquired. This poses the question of whether the narrow, outdated A-level model of post-16 education can survive for much longer in its current form.

What will matter for graduates of the future is whether their degrees offer them a broad enough range of skills that are difficult to computerise. These skills will come from across the disciplines, which will mean degrees must be more interdisciplinary, or at the very least adopt flexible pedagogies and learning environments which can allow a broader range of skills to develop within the context of discipline specific knowledge and experience.

Higher education ultimately cannot deliver all of this if we still perceive the value of a degree on the basis of the information we have about previous and current student cohorts. We need a more transcendental view of value, which, rather than attempting a rationalisation of what a degree has offered immediately, considers the conditions of possibility for degree value to come. We must derive value less from how the world is, and more from how it might be.

This doesn’t mean to say all this quality and outcomes data is useless. The wise investor doesn’t pick stocks and shares with a crystal ball, they still assess risk and reward from what they have available, but always with the caveat that we can never predict the future, only plan for it.

The Royal Society’s futures projects attempt to understand how we can best make use and benefit from scientific and technological advancement. This isn’t about making predictions; it’s about changing expectations and creating a process for thinking about uncertainty and planning for how to respond to complex systems. Such thinking is required for understanding what value a degree might offer in the future.

In developing a positive image of how we respond to future uncertainty through our a priori knowledge, we can understand what we really ought to value in higher education and begin working on methods to achieve it.