The sorting machine: how a vintage gets repriced overnight, with no watchlist

A credit ratings agency downgrades a bond, and the market has usually seen it coming. There is an outlook change, a negative watch, a quarter of deteriorating metrics. The carbon market does not work this way. When the Integrity Council for the Voluntary Carbon Market assesses a methodology against its Core Carbon Principles, or when the CORSIA programme defines what is eligible for a compliance period, the outcome is binary. A category of credits is labelled, or it is not. And when the line moves, it can move an entire cohort at once.

This is the structural feature that the holders of large carbon books most consistently underestimate. They think of integrity as a property of an individual credit, something that was checked at the point of purchase and is now settled. In reality, integrity is conferred at the level of the methodology and the vintage, by institutions that periodically redraw the boundary of what counts. The Core Carbon Principles label is assessed at the methodology and vintage level, not credit by credit. CORSIA eligibility is defined by programme and by period. When either assessment changes, every credit that sits inside the affected category is repriced together, regardless of any individual project's merits.

Binary, cohort-wide, and without a watchlist

Consider what this means for a treasury function that thinks in market terms. In conventional credit markets, repricing is graduated and signalled. An issuer drifts from A to BBB over time, and the spread widens to match. A portfolio manager can act on the outlook before the downgrade lands. The voluntary carbon market has no equivalent machinery. There is no published watchlist of methodologies under review with a probability attached. There is an assessment process, and then there is an outcome. The day before a methodology is judged not to meet the Core Carbon Principles, the credits trade on the assumption that it will. The day after, the assumption is gone, and so is the premium that depended on it.

The same binary logic governs CORSIA. The Carbon Offsetting and Reduction Scheme for International Aviation defines eligible units for each compliance phase. A unit that qualifies for one period may not qualify for the next, and the eligibility decision is categorical. For an airline holding units against a CORSIA obligation, the question is not whether a particular credit is good. It is whether the category the credit belongs to remains inside the line the programme has drawn. That line is redrawn by the programme, on the programme's timetable, not the holder's.

The absence of a watchlist is not a minor procedural gap. It is the difference between a market a treasurer can manage and one that can surprise the book by an order of magnitude in a single decision. A holder cannot hedge a reclassification it has no instrument to anticipate. The only protection is to hold a continuously updated read on where the integrity boundary sits and how exposed the book is to it, rather than a certificate that records where the boundary sat on the day of purchase.

01 · The cohort moves together

The cohort-repricing argument is the heart of the risk, and it deserves to be stated plainly. When an institution like the Integrity Council assesses a methodology, the unit of judgement is the methodology, not the project. A favourable assessment lifts every compliant project that uses that methodology. An unfavourable one weighs on all of them at once. The same is true of a vintage cut-off: credits issued before a certain date may be treated differently from those issued after, and the entire pre-cut-off vintage moves as a block. The holder of a diversified book is not protected by diversification across projects if those projects share a methodology or a vintage that the sorting machine reclassifies together.

This is why a book built on the logic of holding good individual certificates can fail suddenly even though every credit in it was carefully chosen. The careful choice was made against an integrity boundary that has since moved. The credits did not change. The classification did. And because the classification is conferred at the cohort level, the loss is correlated across the whole holding rather than diversified away. A treasurer who would never hold a bond portfolio with that kind of hidden correlation often holds a carbon book with exactly it, because the correlation is invisible until the sorting machine acts.

Diversifying across projects does not protect you when the methodology or the vintage is the thing being reclassified. The cohort moves as one.

02 · The integrity premium is already large, and it is widening

If the sorting machine only shuffled labels, it would be an administrative curiosity. It does much more than that, because the label is now priced. The market has developed a substantial and widening integrity premium, and the gap between the strongest and the weakest credits is no longer marginal. By late 2025 the strongest credits commanded roughly three times the price of the weakest. Afforestation, reforestation and revegetation credits rated at the upper end, around BBB-plus, averaged more than twenty-six dollars, while lower-rated equivalents sat near fourteen dollars. The weakest tier of credits traded close to three dollars.

The direction of travel matters as much as the level. Across 2025 the price of high-quality afforestation and reforestation credits rose from roughly fourteen dollars to roughly twenty-six dollars, while weaker credits did not follow. The overall market told the same story from a different angle: end-user spend was roughly steady at a little over one billion dollars in 2025, even as the volume of retirements fell. Value was migrating toward quality rather than toward volume. The integrity premium is not a theoretical construct that might emerge. It is the dominant feature of current pricing, and it is the precise quantity that a reclassification moves.

Put the two facts together and the exposure is stark. The premium between strong and weak credits is roughly three to one. The mechanism that decides which side of that line a cohort sits on is binary and unsignalled. A reclassification does not nudge a holding down by a few per cent. It can move a cohort from the premium tier to the discount tier, which is to say it can take a meaningful fraction of the value of an entire category in a single decision the holder did not see coming.

03 · Why a certificate finds out last

This brings the argument to its central distinction, the one that separates a book that can survive the sorting machine from one that cannot. A book that holds certificates holds a record of a past decision. The certificate says what was true at issuance and at purchase. It does not update when the methodology is reassessed or when the eligibility boundary moves. The holder of certificates therefore learns about a reclassification the way everyone else does, after it has happened, when the price has already moved and the disclosure question is already live.

A book that holds a live read on risk is a different thing entirely. It carries, for each holding, a current view of where the integrity boundary sits, which methodologies and vintages are exposed to reassessment, and how much of the book is concentrated in cohorts that could move together. It treats integrity not as a verdict delivered once at purchase but as a status that has to be maintained continuously, because the institutions that confer it act continuously and without warning. The difference is the difference between finding out last and being positioned in advance.

The independence of that read is not optional. A book monitored by the broker who assembled it, or marked by a party that also sells into the same market, carries the conflict into the very function that is supposed to detect the risk. The party that wants the cohort to stay in the premium tier cannot be trusted to tell the treasurer when it is about to leave. What a finance function needs is an independent, continuously maintained read on cohort exposure, held by a party that takes no position in the credits and earns nothing from the classification staying favourable.