Why Labels No Longer Convince Anyone

In the first article of this series, we saw with George Akerlof why a market where the buyer can’t verify quality ends up driving out the good producers: when everything looks alike, everything gets paid at the average price, and those who do better either leave — or give up doing better.

Akerlof posed the problem. Michael Spence, who shared the 2001 Nobel Prize with him, formalized the first answer. And his answer explains, fifty years on, why our shelves overflow with labels that no one believes anymore.

The Diploma Is Useless — Which Is Why It Works

Spence’s idea is baffling the first time you encounter it. His founding example is the labor market. An employer cannot directly observe a candidate’s value. So the candidate emits a signal: a diploma.

And here is Spence’s provocation: in his model, the diploma works even if it teaches nothing at all. Its value doesn’t come from its content. It comes from its cost. Earning a demanding diploma requires years of effort — effort the least capable candidates cannot supply, or not at the same price. The diploma sorts candidates not by what it teaches, but by what it costs to obtain.

That is the definition of a credible signal: a statement that only those telling the truth can afford to make.

Hold on to that sentence. It is the test that every label, every certification, every commercial promise should pass. And it is precisely the test the food world has stopped passing.

The Heat Death of the Label

Let’s apply Spence’s criterion to a supermarket shelf.

“Made in France.” “Traditional recipe.” “Carefully selected.” “Committed to the planet.” How much does it cost to emit these signals? The price of a line on a package. Nothing distinguishes the one telling the truth from the one who’s roughly telling it, since the cost of emission is identical for both: next to nothing.

A signal that everyone can emit at the same price no longer signals anything. That’s mathematical, and it’s measurable: studies on food distrust show that roughly two out of three consumers no longer believe the claims brands make. This isn’t irrationality. On the contrary, it’s a perfectly rational reaction to signals that have become free.

Even serious labels, the ones built on real specifications, suffer a second-order erosion: their own proliferation. When dozens of logos share the same package, the consumer can no longer assess what each one cost to obtain. The costly signal drowns in the noise of free ones — and the whole shelf falls back into Akerlof’s world, where everything is worth the same for want of being told apart.

Spence wrote as much back in 1973: a signaling equilibrium collapses as soon as the signal stops differentiating by its cost. We are there.

The Cost Europe Imposes — and Doesn’t Charge For

Yet Europe does have a costly signal at continental scale. Reread the European standard through Spence’s lens: sanitary, environmental, social, it imposes a considerable cost of entry. Our producers pay it for real — in compliance work, in inspections, in production constraints. In theory, it is the credible signal par excellence: only those who genuinely do the work can afford to comply with it.

Except this signal is broken twice over.

Broken on the inside, first: a mandatory cost no longer differentiates anyone. When everyone has to comply, complying pays nothing. The European producer pays the cost of the signal without ever collecting the signal’s premium — the worst of both Spencian worlds.

Broken at the border, second: this cost is only partially required at entry. The imported product reaches the same shelf, at the same price, without bearing an equivalent verification of its practices. This is the whole debate over mirror clauses: demanding rules for those who produce here, a far lighter burden of proof for what comes in. An opening without symmetry of proof turns the standard into its opposite: a tax for some, a sieve for others.

Let’s be clear about what follows, because this is where the debate usually derails: the answer is neither fewer standards nor closed borders. The answer is proof that holds up against everyone, regardless of origin. The Andalusian or Moroccan producer who proves their practices deserves the same access to the shelf as the market gardener outside Paris — and the same price for the same truth. That the scale of imports raises real questions elsewhere — about the balance of supply chains, about food sovereignty — is a separate debate, and we’ll come back to it. But that debate can’t even be held calmly until the first problem is solved: without proof that holds up against everyone, you’re not comparing practices, you’re comparing prices. Opacity forecloses even an honest debate about imports.

What Spence Hadn’t Foreseen

One limit remains in Spence’s model — and this is where our era changes the game.

In Spence’s model, the credible signal is necessarily costly for whoever emits it. That is even its definition. And in 1973, it couldn’t have been otherwise: producing proof was expensive, and verifying it was more expensive still — human audits, physical inspections, mountains of documents no one cross-checks. Transparency was structurally a tax on virtue. That is exactly how food certification still works today: whoever does things right pays to prove it, while whoever does the bare minimum pays nothing. Many good producers give up, not for lack of quality, but for lack of cash flow.

But the cost structure underpinning this reasoning has just collapsed. A temperature sensor costs a few euros. Every producer carries in their pocket a device capable of timestamping and geolocating a production event. And artificial intelligence can now read, cross-reference, and check the consistency of this data at a marginal cost close to zero — what used to require an auditor now happens continuously, on every batch, with no site visit needed. It isn’t Spence who’s been refuted. It’s the economics of proof that have changed centuries.

There is therefore now a second path, unthinkable in 1973: making the signal inimitable not because it is expensive, but because it is factual. A diploma, a label, remain declarative — a paper attesting something. Proof recorded as production happens — this batch, this plot, this date, this transport temperature, logged the moment the facts occur and tamper-proof from then on — doesn’t need to be costly to be credible. It’s credible because a cheat cannot retroactively produce facts that never took place.

This shift changes who pays. If proof can be produced at near-zero cost by whoever is already doing the work, nothing forces the producer to pay in order to stand out. The logic can be reversed: make those who consume the value of transparency bear its cost, and turn proof into an asset for whoever generates it — a recognition, not a tax. And if proof becomes almost free, then we can finally measure — and one day reward — what the market has never known how to see: the positive effects of work done well. But that is another series of articles.

That is the conviction shaping what we’re building with VeraTrace: a registry of those who prove, where the producer who traces doesn’t buy a logo — they build, action after action, a product passport that no one can copy. Spence’s signal, stripped of its tax. And demandable of everyone, from the border to the plate.

Akerlof showed why opaque markets punish the best performers. Spence showed under what condition a signal can save them. But one question remains open: when the seller doesn’t signal, what does the buyer do? And what happens when entire economies open up to one another without ever asking that question? It so happens that a third man shared that 2001 Nobel — and spent his life on precisely these two problems. Meet Joseph Stiglitz.