A word means nothing when everyone has an interest in saying it. The good worker and the bad both claim to be diligent; the sound firm and the shaky both call their product reliable; and because the claim costs nothing and serves everyone, it persuades no one. What the listener trusts instead is the thing that would be too expensive to fake — the action only the genuine article would find worth taking. This is a signal: not a statement of one's quality but a demonstration of it, made credible precisely by its cost. The diploma, the warranty, the unhedged guarantee, the willingness to be held to one's word — each works not by what it says but by what only the real thing could afford to do.
Recall the impasse at which the market for lemons left us. The seller of a good car wishes nothing more than to reveal its quality, for which a buyer would gladly pay; but he has no way to do so, since whatever he says in its favor the seller of a lemon will say in favor of his, and the buyer, knowing this, discounts both equally. The honest assurance and the dishonest are, on their face, identical — words, costless to utter and therefore worthless as evidence, what the theorists of communication call cheap talk. The good seller's problem is not that he lacks something to say but that saying it changes nothing; his quality is real, and invisible, and his protestations of it are drowned in the indistinguishable chorus of everyone making the same claim. What he needs is not a better assertion but a different kind of act — one that the lemon-seller, for all his willingness to say anything, would find it against his interest to copy.
Michael Spence, in work of the early 1970s that completed the edifice Akerlof had begun and earned the two a shared Nobel, identified precisely what such an act must be. A signal can separate the good from the bad only if it is costly, and — this is the crux — differentially costly, cheaper for the high type to send than for the low. Then, and only then, will the high type find it worthwhile and the low type not, so that the mere fact of the signal's being sent reveals the sender's quality, as words never could. Spence's illustration, which has shadowed the economics of education ever since, was the academic degree. Suppose schooling teaches nothing of use to an employer, but that the abler student finds it less arduous, less costly in effort and in risk of failure, to obtain a given credential. Then the able will acquire the degree and the less able will not — not because it has made them productive but because it is cheaper for the productive to get; and an employer, unable to observe ability directly, can read it off the diploma. The signal works without conveying a single fact about its own content, by the sheer logic of who can afford to send it.
Once the principle is grasped, signals are seen to be everywhere, quietly doing the work that words cannot. The manufacturer who offers a long and unconditional warranty signals the soundness of his product, for a warranty is cheap to give on goods that rarely fail and ruinous on goods that fail often, so that only a confident maker will extend it. The firm that spends lavishly on advertising may be signaling quality by the spending itself, as the economist Phillip Nelson argued — not by what the advertisement says, which is cheap talk, but by the conspicuous outlay, which a fly-by-night seller expecting no repeat custom would never recoup and so would not undertake. The shopkeeper who builds an expensive storefront, the professional who frames his credentials, the borrower who pledges collateral, the seller who stakes a costly reputation on every sale: each spends, visibly, in a way that would not pay unless his quality were real. The expense is not incidental to the message. The expense is the message.
This is the feature that makes signaling so counterintuitive, and so revealing. In the ordinary way of thinking, the cost of a signal is waste — money or effort that buys nothing of substance, that a more efficient world would dispense with. But the cost is not a regrettable accident of the signal; it is the whole of what makes it work. A signal that were cheap would at once be imitated by every pretender, and so would cease to separate anyone from anyone, conveying nothing; only the costly signal, too dear for the impostor, can be believed. Biology discovered the same logic, in what is called the handicap principle: the peacock's preposterous tail, useless and metabolically ruinous, is credible as a sign of fitness precisely because only a truly fit bird could afford to grow and drag it. Veblen had seen it in the human world long before, in the conspicuous consumption by which the rich signal their standing through expenditure whose very point is to be visibly beyond the reach of others. To signal is, in a sense, to burn money where it can be seen — and the fire is the proof.
And now the prices that signaling commands array themselves in the familiar disguise. A price is a verdict on a hidden contest, mistaken for a fact of nature — and the wage a degree commands, the premium a trusted brand earns, the higher sum a warranted good fetches, present themselves as the market's reward to genuine quality, as if the credentialed worker were simply worth more and the price were nature's recognition of the fact. But signaling tells a subtler and more disquieting story. The reward attaches not to quality itself, which remains as invisible as ever, but to the capacity to signal it — to having sent the costly sign — and that capacity is the product of a signaling equilibrium, a contingent arrangement of who can afford to prove what, and not a fact of nature at all. The graduate's premium is, in part, the return not to anything the schooling instilled but to the diploma's power to separate; the brand's markup is the return to the costly history that makes the brand believed. What looks like the price of quality is the price of provable quality once again — but now we see that even the proof is bought, and that its cost is folded into every figure the market quotes.
The objection presses itself with unusual force here, and from two directions at once. The first is waste: if the signal conveys quality without creating it — if the degree, in the pure model, teaches nothing — then the vast resources poured into sending it are a dead loss, an arms race in which each credentials harder to stay visible above the rest, all running faster only to keep their places, while the credential required for a given job ratchets ever upward in the inflation pure signaling breeds. The second is justice: a signal sorts by the cost of sending it, and if that cost is lower for the already advantaged — who can afford the long schooling, the unpaid internship, the polished accent and manner — then the signal may track privilege as faithfully as merit, and dress an inherited head start in the colors of demonstrated worth. Faced with this, one may ask whether signaling is a solution at all, or a wasteful and regressive sham. The reply does not deny the charge; it absorbs it. For where signaling is wasteful, or where it launders advantage into apparent merit, that is no refutation of the thesis but its sharpest confirmation: it shows beyond doubt that the credential's reward is the artifact of an institution — a signaling game with its own costs and its own beneficiaries — and not the wage that nature pays to worth. To see the waste and the injustice is precisely to see the price for the contingent, contestable thing it is.
So the signal completes what the lemon began. The first plate showed that in a market of hidden quality the good cannot reveal itself, and the trade decays; the second shows that it can reveal itself after all, but only by spending — by burning resource to prove what a costless word could never establish — and that the price which results reflects the signaling arrangement as much as the quality beneath it. Between them the two plates redraw the meaning of quality in a market: it is never simply present, exerting a natural pull on price, but must be made visible through costly signs, so that what the market pays for is always quality-as-signaled, the thing wrapped in the expensive proof of itself. The lemon is the wound and the signal the first suture; but the suture, we have seen, leaves a scar of its own — the cost of the signal, the inflation of the credential, the advantage it grants those best placed to send it. The market does not get its information for free, and the bill appears, disguised, in every price it sets.
For here is where the second of the volume's contests comes to rest. A price is a verdict on a hidden contest, mistaken for a fact of nature; and in the signal the contest is one of revelation — over who can credibly show what they are, and at what cost. The verdict it returns, the credentialed wage and the branded premium, is read as the market's measure of quality, as though quality declared itself and the price merely recorded the declaration; but quality declares nothing, and the figure is in truth the toll of the signaling game — a game with its own price of entry, its own waste, and its own winners, which a different arrangement of screening or certification or law could redraw entirely. The diploma's wage is not what the able are worth; it is what it costs to prove that one is able, in a world that has settled on the diploma as the proof. Change the proof, and the price changes, though no one's quality has moved an inch. The seller can now be heard — but only because he has paid to speak, and the price he commands is the price of the speaking.
A signal is credible only because it is costly — and so the market, which could not hear quality speak for free, hears it speak at last, but charges, in every price, for the proof.
With the signal the volume's opening pair is complete. The lemon and the signal — the failure and its first answer — together set the frame for all that follows: that in a market of unequal knowledge, both the breakdown and the remedy are made things, institutions and not facts of nature, and the price between buyer and seller records neither quality plainly nor its plain absence, but the state of the long contest to reveal it. What remains of the volume is further answers, each with its own ingenuity and its own cost — the screen by which the uninformed make the types sort themselves, the seal by which a trusted third party vouches, the mandate by which the law drags the truth into the open. Each is another way of letting quality be known where it could not be; each leaves its mark on the price. The good can be told from the bad, in the end — but never for nothing, and never quite for certain. The argument remains open.