Market structure
The auditor paradox: why the chain over-credits.
It was not a rogue project. It was the assurance chain working as designed. When everyone in the chain is paid to move the credit forward, the bias is structural, and methodology alone cannot fix it.
Over-crediting in the voluntary carbon market is usually discussed as a scandal, a story of bad actors and worthless paper. That framing is comforting because it implies the problem is exceptional. It is not. Over-crediting is the predictable output of an assurance chain in which every link is paid to move the credit forward, and in which the body that verifies a project is retained and paid by the developer it is verifying. You cannot engineer your way out of that with a better methodology. A methodology is a ruler. It does not change who is holding it, or who is paying the person holding it.
The evidence is not anecdotal
The case against forest carbon is no longer a matter of opinion. In 2020, a team led by Thales West and Andreas Kontoleon applied synthetic control methods to twelve voluntary REDD+ projects in the Brazilian Amazon and found their claimed reductions in deforestation were systematically overstated relative to credible counterfactuals. In 2023, the same line of work scaled up: a Science paper combining six independent ex post evaluations across REDD+ projects concluded the projects had claimed roughly 10.7 times more avoided deforestation than the independent estimates justified, with the gap driven not by the choice of satellite data but by selection bias in how baselines and control areas were drawn.
In January 2023, a nine-month investigation by the Guardian, Die Zeit and the non-profit SourceMaterial reported that more than 90 per cent of a major standard's rainforest offset credits were likely to be "phantom credits" that did not represent genuine reductions, with the threat to forests overstated by about 400 per cent on average across sampled projects. And the critique has held up under refinement rather than collapsing. A 2025 Science study of 52 REDD+ projects across twelve countries found that only 19 per cent met their reported emissions targets and roughly 13 per cent of issued credits were supported by counterfactual analysis. That last paper is, if anything, the more damning one, because it is sympathetic. It accepts there are real partial gains and still finds over-crediting to be systematic.
Why the chain over-credits
Trace the money. A project developer designs a project and stands to earn revenue for every tonne it can claim. The developer then selects, retains and pays a validation and verification body, a VVB, to confirm those claims. The registry books the resulting credits and earns issuance fees. A broker moves them. Each party in that sequence is paid more when more credits exist and when they exist sooner. Not one of them is paid to find that the number is too high.
The conflict sits squarely at the verification step. Across the major standards, fees for validation are paid by the project developer, which means the auditor's client is the audited party. The standards are aware of this and require conflict-of-interest declarations from their VVBs. But a declaration does not change the cash flow. The verifier still competes for repeat business from the same developers it assesses, and the developer still chooses which verifier to hire. A baseline that runs hot generates more credits and a happier client. A conservative baseline generates fewer credits and a developer who shops elsewhere next cycle. The structural pressure points one way.
A methodology is a ruler. It does not change who is holding it, or who is paying the person holding it.
We have run this experiment before
The carbon assurance chain is a near-perfect replay of the credit-rating model that failed in 2008. There, the agency that graded a structured security was paid by the bank that issued it, and the agency competed for that issuance fee against rivals willing to be more accommodating. The Financial Crisis Inquiry Commission concluded that the rating agencies were "essential cogs in the wheel of financial destruction" and that their ratings had been shaped by "the pressure from financial firms that paid for the ratings" and "the relentless drive for market share". The same Commission noted that of the mortgage securities Moody's rated triple-A in 2006, the overwhelming majority were later downgraded to junk.
The mechanism is identical. Issuer-pays grading, fee competition on the side that wants the high number, and a downstream buyer who treats the grade as independent because it is wearing the costume of independence. The instruments differ, mortgages then, tonnes of avoided carbon now, but the incentive geometry is the same. When the party that benefits from a favourable opinion is the party who hires and pays for that opinion, you should expect the opinion to drift favourable. It is not corruption. It is gravity.
The fix is already visible at the edge of the market
The most instructive development is not another methodology revision. It is the arrival of carbon insurers who put their own balance sheet behind a credit's validity. Underwriters such as Kita, Oka and CFC now write cover for invalidation, reversal and non-delivery, paying out, by some market accounts, premiums in the region of 2 to 10 per cent of credit value depending on risk.
The detail that matters is who they are assessing and why. An insurer does not rate the buyer of the policy as a favour to the buyer. It rates the project and the probability that the credit fails to deliver, because the insurer is the one who writes the cheque when it does. That is the precise inversion of seller-paid advice. The verifier paid by the developer has every reason to accept the developer's number. The underwriter exposed to the loss has every reason to test it. The market is, in effect, growing its own independent second opinion, and it is doing so on the only foundation that has ever produced honest opinions, which is capital at risk.
Independence has to be bought by the loser
The carbon market has spent a decade trying to fix a payment problem with a documentation solution. Better baselines, stricter methodologies, more granular satellite data, all of it useful, none of it sufficient, because none of it touches the question of who pays the person doing the reading. As long as the read is bought by the party that profits from a high number, the read will trend high. The insurers stumbled onto the answer by accident, because their own money was on the table. The answer is to make the read serve the capital that loses when the carbon turns out fake.
Over-crediting is a payment problem wearing a methodology costume. The read should be paid for by the capital that carries the risk, not by the credit moving forward. Kyroq reads, prices and monitors the asset for the party that takes the loss, which is the only position from which an honest number has ever been produced.