The Diploma

That schooling pays is not in doubt. The high-school graduate out-earns the dropout, the college graduate out-earns the high-school graduate, and the gap is wide — by one careful count the typical college graduate earns on the order of a quarter to a half more than the worker who stopped before, a premium that holds, remarkably, across almost every occupation there is.¹1The college wage premium is large — one tabulation finds a median of about 28 per cent across occupations, and more in raw full-time terms — and it appears in some 90 per cent of occupations, including many with no academic content at all. The fact is not in dispute; its cause is. It holds even in trades where nothing taught in a lecture hall could plausibly matter — the credentialed bartender, the degree-holding cashier, out-earning their uncredentialed fellows at the very same tasks. So the fact stands firm: education and the wage rise together, and the diploma marks the rise. The dispute is not whether the educated earn more. It is why.

The first and oldest answer is the natural one: school makes people better at work. This is the theory of human capital, given its modern form by Gary Becker — that knowledge and skill are themselves a kind of capital, built up by schooling and training as surely as a machine is built up by investment, and that the educated earn more because they can produce more.²2Human-capital theory (G. Becker, Human Capital, 1964; J. Mincer; T. Schultz): education and training build productive capacity, so the educated earn more because they are more productive. A degree is an investment — its costs (tuition, and earnings forgone while studying) paid for a higher wage across a working life — and the measured return is substantial. On this view a degree is an investment, undertaken like any other: the student pays its costs — the tuition, and the years of earnings forgone while studying — in exchange for a higher wage across a working life. The surgeon earns her premium because medical school genuinely taught her medicine; the engineer earns his because he can do what the untrained cannot. There is nothing mysterious here, and a great deal of truth in it: much of what schooling teaches is real and useful, the professions plainly require it, and the workers who hold it can, in fact, do things the rest of us cannot. Were this the whole story, the wage would simply be paying, exactly as it ought, for a skill that did not exist before and now does.

But there is a second answer, stranger and more unsettling, first set out by Michael Spence in 1973. Suppose, Spence said, that school taught nothing useful whatever — and watch how the premium could survive even so. People differ in ability, and that ability is hard for an employer to see across a hiring desk. But it is easier for the able to get through school than for the unable: they find the work less costly, less punishing, less likely to defeat them. So the able go further in school — not because it improves them, but because they can; and employers, knowing this, pay for the diploma as a proxy for the ability that reaching it reveals. The schooling sorts rather than builds. On this account the premium is just as real, but its cause is not what it appears: the educated earn more not because school made them productive, but because arriving at the diploma proved them able to begin with.³3Signaling theory (M. Spence, ‘Job Market Signaling,’ 1973; Nobel 2001; cf. K. Arrow’s ‘filter,’ 1973, and J. Stiglitz’s ‘screening,’ 1975). Since ability is costly to observe and the able find schooling less costly, education sorts on pre-existing ability and the credential signals it — so the premium can survive even if school builds no skill at all. Here is the whole quarrel in a single line: in the first theory, school creates the gap between workers; in the second, it merely discloses a gap that was already there.

Does school create the difference, or only reveal it?

How could one ever tell the two apart, when both predict the very same thing — that the educated earn more? The deciding evidence, such as it is, lies in a quirk the human-capital story strains to explain. If each year of school simply added skill, then each year ought to add roughly equally to the wage. It does not. The year that grants the credential pays out of all proportion to the rest: by careful estimate the twelfth grade — the one that yields the high-school diploma — lifts earnings more than the ninth, tenth, and eleventh combined, and the final year of college more than the first three together.⁴4The sheepskin effect (diplomas were once written on sheepskin): the credential year pays disproportionately — twelfth grade more than grades nine through eleven combined, the last college year more than the first three — and the gap persists after controlling for ability. It strains pure human capital, since the learning is spread evenly while the wage leaps at the credential. Its reading is itself debated — a last semester is no costly signal in Spence’s strict sense — but the paper plainly earns beyond the years. The leap is not in the learning, which is spread evenly across the years; it is in the paper, which comes only at the end. Economists call it the sheepskin effect, after the hide on which diplomas were once written, and it is the signal theory’s strongest exhibit: the identical coursework, lacking the credential, pays a fraction of what it pays with it — as though the market valued not what was learned but the proof of having finished. Offered a Princeton education with no diploma, or a Princeton diploma with no education, the worker who wanted the higher wage should take the paper.

If much of what the diploma earns is signal rather than skill, an uncomfortable consequence follows, and Bryan Caplan has pressed it hardest: that a large share of our vast outlay on education buys no new productivity at all, only a reshuffling of who stands where in the queue. By Caplan’s reckoning — contested, but seriously argued — as much as four-fifths of the wage premium may be signaling.⁵5B. Caplan, The Case Against Education (2018), estimates roughly 80 per cent of the premium is signaling, making schooling largely a wasteful ‘arms race’ of credential inflation. The figure is much disputed: critics put pre-existing ability alone at a quarter to a half of the premium, and the mainstream view is a hybrid in which human capital is substantial. How much is skill and how much signal is genuinely unsettled. If so, education becomes in part an arms race: each person rationally seeks more of it to stand out, but as all do, the credential demanded for any given job ratchets upward — yesterday’s diploma becomes today’s bare minimum — while the work itself grows no more skilled. What is privately essential turns out to be socially close to wasteful, effort spent only to hold one’s place. And yet the case must not be overstated, for the other hand is heavy too. Education plainly does build real skills, nowhere more than in the technical and professional fields that keep the modern world running; the able are not merely sorted but genuinely trained; and the careful attempts to weigh how much of the premium is signal arrive nowhere near agreement — some put the pre-existing-ability share at a quarter to a half, leaving real room for skill that schooling truly makes. The honest verdict is a mixture whose proportions no one has fixed: the diploma is part investment and part ornament, part skill and part sorting, and which part rules remains among the most consequential open questions in the economics of work.

Part skill, part sorting — and the wage pays for both.

So the piece of paper does at least three things at once, and the wage pays for all of them without troubling to say which. It certifies a skill that schooling truly built — the human capital that lets the surgeon operate. It signals an ability the holder brought to the gate and merely proved by passing through — the sorting that would lift her wage even had school taught her nothing. And it marks, quietly, an advantage she may have carried all along: the talent or fortune or family that made the diploma reachable for her and not for others, so that some of what looks like a return to education is really the return to whatever made the education possible.⁶6The premium braids three strands — human capital (skill built), signaling (ability revealed), and the bias of pre-existing advantage (those who would have earned more in any case being likeliest to get the schooling). They are tangled so tightly in the data that no one has cleanly separated them, which is why the value of education stays among economics’ most consequential unsettled questions. These three strands are braided so tightly through every educated wage that no one has cleanly pulled them apart, and perhaps no one can. The diploma arrives as a single thing, and is paid for as one.

And so the folio’s anatomy comes nearly to its close. We have weighed the wage’s justice and its hours, its trust and its compulsion, its amount and its boundary, its rigidity and its measure; the authority it buys, the risk it sheds, the machine that rivals it, the meaning it cannot hold, the demand it sustains, the leisure it might buy back. Here we have asked why one wage towers over another, and found the answer folded inside a credential whose worth we cannot fully account for — part the skill it builds, part the rank it confers. The wage rewards the educated richly, and certainly. What exactly it is rewarding, it has never had to know.