At his school reunion, Andy bumps into his old friend Hannah, who he has not seen since she left for university and he started a job straight after their A levels a decade earlier. Who should buy the drinks?

As a non-graduate aged 21-30, Andy should expect to be earning a median salary of ?27,000, according to the (GLM) statistics ¨C or ?30,500 across his working life, from age 16 to 64. But as a graduate of the same age, Hannah should expect to be making ?34,000, rising to ?42,000 across her working life. So Hannah should, in principle, buy the first pint.

However, although Hannah went to university, she did so on the strength of three C grades at A level and she has struggled in the job market since graduation. In reality, she only earns ?25,000. Meanwhile, Andy finished sixth form with two As and a B at A level and has thrived in the business world, earning ?40,000. This disparate level of success becomes obvious as soon as they launch into the standard ¡°so what are you doing now?¡± questions, so Andy buys the drinks.?

So what is the moral of the story? Is this just the kind of statistical quirk that is inevitable when average figures are compared with particular cases? Or is there more to be said?

̽»¨ÊÓƵ

The UK government¡¯s Office for Statistics Regulation (OSR) watchdog thinks the latter. The problem is that the GLM figures, which are published by the Department for Education (DfE), do not account for the prior educational attainment of university graduates, making it difficult to assess the contribution of their degree to their earning power. Hence, the statistics could be misleading, the watchdog recently warned ¨C so much so that the Department for Education announced that it would cease publication of them.

¡°We have determined that?more robust statistics?on postgraduate and graduate earnings and employment are now produced elsewhere such as the?LEO Graduate and Postgraduate Outcomes?release,¡± the last month.

̽»¨ÊÓƵ

But the LEO (Longitudinal Educational Outcomes) release, too, has now been deemed inadequate. When the latest iteration of the , based on earnings data from HM Revenue and Customs, was published at the end of June, the DfE announced that, next year, it will be renamed ¡°Graduate and Postgraduate Labour Market Outcomes¡± and will be overhauled to ¡°include comparative statistics for non-graduates with breakdowns by prior academic attainment and other student characteristics¡±.

These developments were prompted by a case raised with the OSR by?Paul Wiltshire, a semi-retired accountant and the father of four UK university graduates and current students. Wiltshire has argued stridently that the current graduate premium statistics are flawed?and, in March, he produced his own report?setting out the inadequacy of the existing data ¨C a view that was endorsed by a representative of the Royal Statistical Society. His report also estimated that university expansion in the UK has long since passed the point when ¡°marginal¡± graduates ¨C those with lower prior attainment, who would not have gone to university in previous decades ¨C benefit from an earnings premium.

Wiltshire¡¯s conviction is that most degrees are largely irrelevant to graduates¡¯ employment prospects and that ¡°innate academic ability¡± is the main determiner of salary levels. For those with good A-level results, ¡°university is just a stepping stone on that journey to achieving extra pay¡± ¨C but they are ¡°already on that journey anyway¡±, he told ̽»¨ÊÓƵ.

With a degree in maths and statistics, Wiltshire said he spotted the ¡°fundamental flaw¡± in the available graduate earnings figures quite quickly when he was seeking to test his hypothesis about diminishing marginal returns, and he questioned why statisticians had not held up more of a ¡°magnifying glass¡± to them before.

¡°Possibly people didn¡¯t want to see it,¡± he added. ¡°I think people do like to be positive about higher education, particularly the higher education industry itself because these are now businesses which have to make a profit to survive, so obviously [they] do want to interpret figures in a positive manner.¡±

For instance, Universities UK last year to argue why ¡°university is still worth it¡±, rebutting suggestions that ¡°getting a degree no longer adds to a graduate¡¯s employment prospects¡±. And it is certainly true that the expansion in degree-level study has been justified, at least in part, with reference to the average or median graduate premium. But while Wiltshire could understand the sector¡¯s reluctance to look at the figures too closely, he found it ¡°quite startling¡± that the government had permitted the expansion without having ¡°any reliable statistics¡± on the marginal graduate premium, which can be expected to decline as students with lower academic ability enter the system.

¡°These inadequate statistics have allowed the sector to hijack the official figures and mislead the public and Government regarding the benefits of higher education, claiming that ¡®everybody¡¯ will be able to benefit from the supposed average premium,¡± he , in the wake of the announcement of the GLM release¡¯s scrapping. ¡°What is needed is root and branch reform of graduate statistics. I believe it would provide compelling evidence that surpassing around 25-30 per cent HE participation was a monumental mistake, and we certainly should never have let it reach the existing 50 per cent.¡±

That figure was, of course, , announced during his first term as prime minister in 1997. But while that figure has been ¡°over-egged¡±, as he put it to THE, Wiltshire¡¯s worries about the marginal graduate premium also risk being over-egged. For instance, his Telegraph article : ¡°So now it¡¯s official. The ¡®graduate premium¡¯ is a myth¡±. And Wiltshire accused universities of ¡°mass exploitation¡± for ¡°burdening too many of our young adults with morale-sapping student debt for their whole working life, with little or no corresponding improvement in their career prospects¡±.

̽»¨ÊÓƵ

Source:?

Grant Squibb/Getty Images (edited)

Wiltshire and the Telegraph, which has consistently run articles criticising the value of degrees, would no doubt count among the ¡°edu-sceptics¡± that former universities minister David Willetts complained about in a for King¡¯s College London¡¯s Policy Institute and the Resolution Foundation, of which he is chair. In the report, titled Are universities worth it? A review of the evidence and policy options, Willetts condemned the critics¡¯ obsession with the idea that, as Kingsley Amis put it, ¡°more means worse¡± in higher education.

Using both the GLM and LEO ¨C the latter of which he was instrumental as minister in establishing ¨C Willetts concluded that while the graduate premium may have declined in recent years, ¡°the rewards for higher education are still high¡± and are even higher once higher unemployment rates for non-graduates (which do not show up in earnings figures) are taken into account.

¡°If there is a reliable way to boost your earnings, more people should go for it until the benefit is down to zero for the marginal student, which would inevitably bring down average returns,¡± he wrote. In that sense, he welcomed the decline in the average graduate premium as a healthy sign of widening participation but doubted that the UK had yet reached ¡°peak graduate¡±, noting that recent sharp increases in the minimum wage have disproportionately benefited the non-graduates more likely to be in low-paid work. ¡°It is striking how well graduate incomes have held up despite¡­rapid increases in [student] numbers. By the age of 31, graduates are earning 37 per cent more than nongraduates with at least two A-levels ¨C ?30,750 and ?22,500, respectively,¡± he wrote.

However, he conceded that ¡°the prior attainment of students matters¡± for their future earnings and that there is some evidence from the LEO data that some people could have ¡°earned a lot of money anyway if they had just gone straight into work aged 18¡±. But he added that ¡°not many people with good A-levels don¡¯t go to university. So the evidence for what would happen in the alternative scenario is limited.¡±

Since the first LEO release in 2016, political attention has been focused not on the marginal premium but on the widely different average graduate premiums that accrue from different courses and from different universities ¨C in contrast to the flat tuition fee that all domestic students in England, Wales and Northern Ireland pay.

Research by the Institute for Fiscal Studies (IFS) has shown that a substantial lifetime graduate premium of about 20 per cent still exists, even when accounting for graduates¡¯ higher tax and loan repayments. ¡°But there¡¯s quite a lot of quite big differences across subjects and between people, so there¡¯s still within that a portion of people who probably earn less than they would if they hadn¡¯t gone to university,¡± Kate Ogden, senior research economist at the IFS, told ̽»¨ÊÓƵ.

Indeed, the IFS recently called for the Office for Students to introduce an earnings metric as a further way of measuring English universities¡¯ graduate outcomes for quality assurance purposes, ¡°crucially¡± controlling for ¡°prior attainment and various demographic characteristics as well as subject studied¡±.

The thinktank said that looking at the salaries of graduates up to five years post-study would provide insight into how well degrees prepare students for the workforce that is being missed by the regulator¡¯s current focus on continuation, completion and progression. However, a recent report on the future of higher education by Edward Peck, the outgoing vice-chancellor of Nottingham Trent University who is soon to become?chair of the OfS, made no reference to graduate salaries, suggesting that taking up such a suggestion may not be his priority.

Willetts himself conceded in his report that there are some courses, such as creative arts, whose graduates ¡°appear to end up earning less over their lifetimes than if they had not gone to university in the first place¡±. And the Conservative Party went into last year¡¯s general election further student recruitment to ¡°rip-off¡± degrees with ¡°poor¡± job prospects. Culinary arts management, wildlife media and concept and comic art were among seven ¡°Mickey Mouse¡± degrees mentioned by Conservative sources to , as the party argued that one in five graduates would have been better off not going to university.

The party¡¯s manifesto said the measure would ¡°protect students from being missold [sic] and the taxpayer from having to pay where the graduate can¡¯t¡± ¨C reflecting ¡°a critique of universities widespread across the media¡±, as Willetts put it.

¡°I guess all sides sometimes pick and choose [statistics],¡± said Ogden. ¡°I think you get that whenever any politician or [the] media use statistics because they haven¡¯t got the space to put in all the caveats that people who produce them would ideally see.¡±

̽»¨ÊÓƵ

Golo Henseke, an associate professor in applied economics at UCL, said the origin of the debate around graduate salaries can be traced back to the tripling of tuition fees in 2012, when students became ¡°consumers¡±.

¡°It is easier for the government and the media to criticise higher education providers by claiming to be protecting consumer rights,¡± he said ¨C though cracking down on ¡°rip-off¡± degrees also has an added benefit for the government in lowering the amount of public money that is never repaid by graduates through the loan system, he added.

Equally, universities¡¯ own reliance on graduate premium data in their promotional materials is a natural consequence of operating in an expanded and ¡°marketised¡± system, Henseke said. ¡°They need to get students in and they use any signal they can to prospective students assuring [them] that their course is relevant, whether that¡¯s decent pay at the other end or¡­high rankings. It¡¯s a game they have to play in the current system.¡±

Costas Milas, professor of finance at the University of Liverpool, agreed. He said that what used to be a ¡°considerable¡± premium in the 1970s, 1980s and even 1990s has indeed declined, with key factors being the expansion of higher education and grade inflation, which have lowered the value of a degree in employers¡¯ eyes.

However, Milas added that the importance of the graduate premium in student recruitment events, such as presentations to sixth-formers, meant that universities would do ¡°everything they can in their power¡± to ¡°push it back up¡± if the graduate premium fell too low, such as updating programmes ¡°with new ideas and new modules that are going to be very attractive to prospective candidates¡±.

Source:?

Leila Cutler/Alamy (edited)

Some worry that in playing the graduate premium game, universities are selling themselves short. David Veevers, a lecturer in early modern history at Bangor University, said this narrative had brought the sector to a ¡°tipping point¡± by narrowing the terms of the debate on the value of degrees.

¡°Universities have been swept up now in the government narrative¡± about the importance of considering graduate earnings, he said, adding that this is ¡°the end result of reducing the university system to a financial commodity. I think universities have been so busy defending themselves and not really getting ahead and getting out the positive messages about what [wider benefits] universities contribute.¡±

He noted that graduate earnings were ¡°one of a number of different ways you can measure the success of degree programmes at universities, and for a long time this country valued education for education. If you reduce that simply to what will these students earn, I think that¡¯s a slightly catastrophic way to approach a public good like higher education.¡±

Willetts¡¯ report agreed, noting that the benefits of higher education also include local and national economic impacts (such as higher tax revenue), and non-economic returns to graduates, such as the fact that graduates are ¡°70 to 80 per cent more likely to report excellent health¡±.

Wiltshire agrees that universities are essential for the country, but he thinks that the public interest argument for higher education ¡°only works if, as a society, we decide that we pay for it¡­through general taxation, just like we do for people doing A levels¡±.

He approved of that financing mechanism, backed up by means-tested maintenance grants, ¡°but only if we reduced participation right back down to 10 to 15 per cent. It would be costly, but far cheaper than writing off all the unpaid loans for the number of wasteful degrees we have now,¡± he said. ¡°And the government would just have to do its utmost with the data available to ensure that the reduced number of degrees were in the ¡®best¡¯ and most reasonable range of subjects.¡±

He is ¡°quite happy for a certain amount of degrees to be purely academic and have no relation to careers. This is good for society. But it has to be restricted to those individuals who are going to benefit most ¨C and in turn benefit society most ¨C and this is those who are ¡°clever¡± academically. It is great that we will have politicians, journalists, judges, senior civil servants who may have studied history or philosophy or English literature and really gone deep into the subject and expanded their mind.?

The other way of measuring the success of university courses that is prominent in the media is the proportion of graduates in work. But the by the Higher Education Statistics Agency (Hesa) was designed to capture a more nuanced understanding, asking graduates whether their work is meaningful, their career is on track, and if they are using graduate skills. Last year¡¯s release showed that over 80 per cent of UK-domiciled undergraduates found their current activity meaningful 15 months after graduation.

And that points to another defence of universities: the fact that graduates value their student experience and what it led them to. Willetts references showing that 87 per cent of graduates would choose to do an undergraduate degree again ¨C even if nearly 40 per cent of that cohort would do a different degree.

That is why Willetts warns against fashioning higher education policy around LEO data by banning courses that offer no graduate premium. Rather, he believes that earnings data should be ¡°available to prospective students but then ultimately it is a personal decision for them¡±. The LEO is a ¡°fantastic tool to enhance student choice: it does not need to be used instead to suppress it. Well-informed students should see the risks they are running.¡±

Of course, next year¡¯s revamped LEO may make it even clearer that some would-be students on some courses at some universities might be better off, in financial terms, if they were to think again. But, for his part, Willetts welcomed the development. ¡°It is right to broaden the range of this data and break it down by prior attainment and other characteristics,¡± he told ̽»¨ÊÓƵ. ¡°But it is important that the raw data is still available¡±.

He also cautioned the DfE against using the data to make ¡°speculative counter-factuals¡±, adding that these ¡°can be left to outside researchers, who may have different views on how best to construct them¡±.

He also urged the DfE to expand the LEO to ¡°cover much longer periods. Limiting data to outcomes after five years focuses attention on short-term gains rather than longer-term careers, which may have a different earnings trajectory.¡±

Wiltshire, too, offered cautious praise of the LEO¡¯s reform, saying it was ¡°a significant step in the right direction that next year¡¯s LEO data will for the first time enable a direct comparison between the pay outcomes of graduates and non-graduates with the same A-level results, as I have requested. This will enable a far better assessment of the extent of the value added by gaining a degree, if any.¡±

But he also called for data to be gathered on ¡°the extent to which the jobs that graduates end up in bear any relation to the degree taken. Because it hardly seems right to credit the pay of a graduate for their job as, say, an NHS administrator solely on the fact that they have completed an unrelated archaeology degree.¡±

But whether any of that information would significantly alter enrolment on to particular courses remains moot. Veevers said his conversations with his students suggest that, for many, graduate earnings expectations may not be the overriding factor in course selection that politicians and commentators assume it to be.

̽»¨ÊÓƵ

¡°They go to university to learn about a subject they are passionate about,¡± Veevers said. ¡°They are less concerned with how wealthy they¡¯ll be as a result.¡±