IN A classroom in Seoul a throng of teenagers sit hunched over their desks. In total silence, they flick through a past exam paper. Stacks of brightly coloured textbooks are close to hand. Study begins at 8am and ends at 4.30pm, but some will not go home until 10pm. Like hundreds of thousands of South Koreans, they are preparing for the suneung, the multiple-choice test that will largely determine whether they go to a good university or a bad one, or to university at all.

Over the course of a single generation in South Korea, degrees have become close to ubiquitous. Seventy per cent of pupils who graduate from the country’s secondary schools now go straight to university, and a similar share of 25- to 34-year-olds hold degrees, up from 37% in 2000. Students scramble to gain admittance to the most prestigious institutions, with exam preparation starting ever younger. Sought-after private nurseries in Seoul have long waiting lists.

South Korea is an extreme case. But other countries, too, have seen a big rise in the share of young people with degrees. In the OECD club of 35 countries, 43% of 25- to 34-year-olds now have degrees. In America the figure is 48%.

Between 1995 and 2014 government spending on higher education in the OECD rose from 0.9% of GDP to 1.1%, while private spending rose from 1.2% to 1.5%. As government subsidies for tuition fees flow through to institutions they have helped inflate costs. Since 1990 fees for American students who do not get scholarships or bursaries have risen twice as fast as overall inflation.

Policymakers regard it as obvious that sending more young people to university will boost economic growth and social mobility. Both notions are intuitively appealing. Better-educated people should surely be more likely to come up with productivity-boosting innovations. As technological change makes new demands of workers, it seems plausible that more will need to be well-educated. And a degree is an obvious way for bright youngsters from poor families to prove their abilities.

But comparisons between countries provide little evidence of these links. Richer countries have more graduates, but that could be because there is more money to spare, and less urgency to start earning. Rich economies grow more slowly, but that is probably because they have fewer easy ways to raise productivity, not because education depresses their growth.

A truth universities acknowledged

The main piece of evidence cited by policymakers is the “graduate premium”—the difference between the average earnings of someone with a degree and someone with no more than a secondary-school education, after accounting for fees and the income forgone while studying. This gap is often expressed as the “return on investment” in higher education, or the annualised boost to lifetime earnings from gaining a degree. Research by the New York Federal Reserve shows that the return on investment in higher education soared between 1980 and 2000 in America, before levelling off at around 15% a year. In other words, an investment equal to the cost of tuition and earnings forgone while studying would have to earn 15% annual interest before it matched the average value over a working life of gaining a degree.

The World Bank has produced estimates of this return for 139 economies. It varies from place to place, but is substantial everywhere. The Economist’s analysis of the data finds that returns are linked to the share of people with degrees, and the range of earnings. Returns in Britain and Germany are similar to those in America. In sub-Saharan Africa, where degrees are scarce and the least-educated workers earn little, they are around 21% a year. In Scandinavia, where wages are less unequal and two-fifths of adults have degrees, they are around 9%.

But as a guide to school-leavers considering going to university—and to policymakers considering expanding access to higher education—the graduate premium is flawed. Even within countries the average conceals wide differences. Most students know that a degree in mathematics or finance is likely to be more lucrative than one in music or social work. What fewer realise is that the graduate premium overstates the financial benefit of embarking on a degree if their school grades barely qualify them for entry, no matter what they study.

In a comparison of the earnings of people with degrees and people without them, those who start university but do not finish are lumped in with those who never started, even though they, too, will have paid fees and missed out on earnings. Their numbers are considerable. In America 40% of college students fail to graduate with four-year degrees within six years of enrolling. Drop-out rates across the developed world average around 30%. It is the students admitted with the lowest grades who are least likely to graduate.

Including dropouts when calculating the returns to going to university makes a big difference. In a new book, “The Case Against Education”, Bryan Caplan of George Mason University argues that the low graduation rates of marginal students, and the fact that, for a given level of qualification, cleverer people tend to earn more, mean that the return on a four-year degree in America ranges from 6.5% for excellent students to just 1% for the weakest ones.

Part of that difference is because the weakest students attend the worst universities, where drop-out rates are highest. When they make it into better institutions, the returns may be higher. In a study published in 2014 Seth Zimmerman of the University of Chicago compared the earnings of school-leavers in Florida whose grades were close to the minimum for admission to a good state university. Those just above the cut-off were much more likely than those just below to start courses in good institutions. They graduated at a rate similar to that of the broader student population. They went on to earn considerably more than those just below the cut-off, and their return on investment was substantial.

Overstating the graduate premium is not the only reason policymakers overestimate the wider benefits of increasing the share of young people who go to university. The usual way to calculate the social returns of higher education is to sum up all the graduate premiums and subtract any public subsidies. But degrees are in part a way to access a “positional good” that benefits one person at the expense of another. Part of the premium comes from gaining an advantage over others in the competition for a good job, rather than the acquisition of productivity-boosting skills and knowledge. A complete calculation would include not just gains to graduates, but losses to non-graduates.

Degrees are also signalling devices. The premium includes the income-boosting effects of personal characteristics that are more likely to be held by those with degrees, not because they acquired them at university, but because they possessed them on admission.

As degrees have become more common, their importance as signalling devices is rising. Recruiters, who pay none of the cost of jobseekers’ higher education, are increasingly able to demand degrees in order to screen out the least motivated or competent. A recent study by Joseph Fuller and Manjari Raman of Harvard Business School found that companies routinely require applicants to have degrees, even though only a minority of those already working in the role have them. This increases the graduate premium—but by punishing non-graduates rather than boosting the absolute returns to degrees.

Analysis by The Economist of American census data finds that between 1970 and 2015 the share of workers aged 25-64 with at least a bachelor’s degree increased in 256 out of 265 occupations. Some of these are intellectually demanding jobs that changed a lot over that period, such as aerospace engineer or statistician. Others are non-graduate jobs such as waiting tables. Sixteen percent of waiters now have degrees—presumably, in most cases, because they could not find a graduate job. But other jobs that are mostly done by graduates, such as journalism, nursing and teaching in primary schools, used to require only shorter training, often received while working. Today, having a degree is usually an entry requirement.

The Economist has produced a measure of over-education by defining a graduate job as one which was staffed mostly by degree-holders in 1970. We find that just 35% of graduates work in such occupations today, down from 51% 45 years ago. Judging by job titles alone, 26.5m workers in America—two-thirds of those with degrees—are doing work that was mostly done by non-graduates a half-century ago.

That calculation exaggerates the trend. Advances in technology have doubtless made some of these jobs more demanding. But not all of them, at least judging by pay. We find only a weak link between higher shares of graduates in an occupation and higher salaries. For around half of the occupations that employ higher shares of graduates now than a half-century ago, real wages have fallen.

Andreas Schleicher, the head of education research at the OECD, reckons that “countries have skills shortages, not degree shortages”. The way universities have come to monopolise higher education, he says, is a problem in part because universities do not suit all kinds of learners. And university dropouts tend to see little in the way of financial benefit from the part of their course that they have finished.

One promising development is that of “micro-credentials” or “nano-degrees”—short vocational courses, often in computing and IT. Udacity, an online education company, offers a variety, including one in self-driving cars approved by Uber and Mercedes-Benz, and another on digital marketing approved by Facebook and Google. EdX, a collaboration between MIT, Harvard and other leading universities, offers similar courses free. Students can take exams to prove their mastery of the material for a few hundred dollars.

Boot campus

For now, such courses are mostly add-ons to degrees, rather than replacements. Three-quarters of edX’s students already had a bachelor’s degree upon enrolling. But the collaboration with sought-after employers makes it more plausible that they could eventually become established as a stand-alone testament to a job applicant’s worth.

In the meantime the decision not to go to university remains risky, even though many graduates will end up doing work that used to be done by non-graduates—or struggle to find a job at all. Around half of unemployed South Koreans now have degrees. For them, the very concept of a “graduate premium” may seem a mockery. Kim Hyang Suk, a recruiter in South Korea, says that half the applicants for customer-service jobs at her firm are graduates, even though only a secondary-school education is specified.

She would prefer school-leavers with experience, says Ms Kim, to inexperienced graduates whom she will have to train. She is not looking for swots, but people who are “engaging, good on the phone”. But when few employers are this open-minded, most young people will want a degree. It may not boost their earnings as much as they had hoped, but without one, they will probably fare even worse.