Cosmetic produce standards

Why do sound, edible vegetables get discarded before they ever reach a shop shelf?

A forked parsnip, a bent carrot, a cauliflower two centimetres under size: sound, nutritious, indistinguishable once chopped, and tipped back onto the field. The usual verdict is that supermarkets are fussy and shoppers vain. Both are partly true and neither explains the pattern, because the discard mostly happens before the supermarket, at the grower's expense, on produce a buyer never sees and never pays for. So where does the cut get made, by whom, and against what?

Take the grading rule seriously as reasoning rather than vanity. Appearance is a proxy — cheap, instant, verifiable by eye at the pace of a moving belt. Ask what it is a proxy for, and the answer is mostly not "taste". Uniform size predicts that a box packs to weight, that a case fills to a stated count, that an automated line will not jam. Colour predicts maturity, and so remaining shelf life. A blemish sometimes marks a fungal entry point that will progress in transit. The proxy carries genuine information — but about logistics and transit risk, not edibility, and that mismatch is where the waste is generated.

Now the part that sets the cut-off. Price the two mistakes. Reject a good item: you lose its margin, call it pennies. Accept a bad one: it travels, it fails, and the failure surfaces at the far end of a chain — a pallet rejected on arrival, a load short against contracted volume, a penalty clause, in the worst case delisting and the loss of a season. Orders of magnitude larger, and it lands on the packer, who is also the one setting the cut-off.

With errors that asymmetric, work out where a rational threshold sits. Tighten until the marginal expected cost of rejecting one more borderline item equals the marginal expected cost of passing it. If passing costs a hundred times more, the threshold moves until borderline items are rejected roughly a hundred times more readily than passed. It ends up deep inside the edible boundary — not by error but by construction. Nobody had to be fussy; the arithmetic put it there.

And it repeats at every hand-off. The retailer writes a tolerance and a penalty into a contract; the packer, whose inspection is noisy, adds a margin so as not to breach it; the grower adds another rather than have a load turned away at the packhouse gate. Margins compound, and the produce meets the tightest.

Two further channels, without which this would be too tidy. Contracts specify volume with penalties for shortfall and yields are uncertain, so growers over-plant as insurance; in a good season that surplus is never rejected for looks because it is never offered to anyone. And once a crop is off contract, lifting it costs more in labour and haulage than a secondary buyer will pay, so the field is left.

Three common explanations, each insufficient rather than simply wrong.

"Supermarkets are wasteful." The discard largely precedes the supermarket and is absorbed by growers and packers. Retail-stage loss is real but a different, smaller stream; blaming the visible end of the chain misplaces the decision point.

"Consumers irrationally demand perfection." Something to this — offered identical produce at identical price, shoppers do pick the unblemished item, which is why imperfect lines sell only at a discount. But that cannot explain specifications written for case fill and machine handling, nor loss on produce never offered for sale.

"It is the law." Mostly not any more. The EU withdrew marketing standards for most fruit and vegetable types in 2009, keeping a short list (recalled). What replaced them were private buyer specifications — frequently tighter, because a private spec answers to one buyer's logistics rather than a compromise among many.

That last fact is the account's best test, close to a natural experiment. If cosmetic fussiness were the driver, deregulation should have removed most of the loss. It did not, because the asymmetric-cost logic, the compounding margins, the surplus planting and the harvest-cost floor were untouched.

Now the number I will not give. It is tempting to state what share of a crop is lost this way, and I decline, because the figures in circulation do not mean what they appear to: definitions differ on whether unharvested surplus counts, whether diversion to animal feed counts, whether processing trim counts. The famous claim that around a third of food produced is lost or wasted traces to a 2011 study that mixed measured and modelled estimates across very unlike food systems and was never a farm-level measurement. Directionally the loss matters; a share for cosmetic rejection alone does not exist cleanly.

Falsifying my own account: what I hold most loosely is how the three channels divide. Whether the largest stream is grading rejection, insurance over-planting, or crop left below haulage cost varies by crop, region and contract form, and I doubt it has a general answer. What survives is the structural claim: the cut-off is placed by whoever bears the cost of being wrong, and that is not the party bearing the cost of throwing food away.

• Why paying for grading outcomes rather than penalising rejections would move the threshold, and why buyers resist it.

Every term the collection defines is gathered in the glossary.