The Carbon Market Is a Fiction
The world's primary market-based mechanism for reducing emissions has been revealed as largely fraudulent. Understanding why it failed tells us something important about the limits of financialized climate policy.
In January 2023, a Guardian investigation found that more than 90 percent of the rainforest carbon offset credits issued by Verra — the world's largest carbon standard, whose credits have been purchased by Delta, Disney, Shell, and dozens of other major corporations — were "phantom credits" that did not represent real carbon reductions. The investigation was followed by a peer-reviewed study in Science that reached similar conclusions, and by a cascade of subsequent analyses that found comparable problems across multiple carbon credit standards and methodologies.
The response from the carbon offset industry was predictable: methodological criticism of the studies, assertions that specific projects had been mischaracterized, proposals for improved standards that would make the system work as intended. The response from the corporations that had purchased credits to meet net-zero commitments was less predictable: many simply stopped talking about their offset strategies.
The carbon market crisis is not a story about bad actors in an otherwise functional system. It is a story about a system that was designed to be unable to deliver what it promised — and that the financial and political incentives surrounding it have prevented from being reformed.
What carbon markets were supposed to do
The theory of carbon markets is elegant. Emitting carbon dioxide has a social cost — the damage caused by climate change — that is not reflected in the prices of carbon-emitting activities. A carbon price corrects this market failure by requiring emitters to pay for the cost they impose on others. Carbon markets allow this correction to occur efficiently: emitters who can reduce emissions cheaply do so, generating credits that emitters who cannot reduce cheaply can purchase, achieving total emissions reductions at minimum economic cost.
The voluntary carbon market — the market in which corporations purchase carbon credits to offset their emissions outside of regulatory compliance requirements — operates at one step of further abstraction from this theory. Corporations purchase credits that represent emissions reductions elsewhere in the world — typically forestry projects that claim to prevent deforestation in tropical countries, or renewable energy projects that claim to displace fossil fuel generation. They use these credits to claim "carbon neutral" or "net zero" status.
The system works in theory if — and only if — three conditions hold: the emissions reductions are additional (they would not have occurred without the carbon market); they are permanent (a protected forest does not burn down or get logged in the future); and they are accurately measured (the claim that a given acre of forest would have been deforested without the protection program is verifiable).
All three conditions have proven systematically difficult to satisfy in practice.
The additionality problem
Additionality is the foundational requirement of carbon offset crediting: you can only claim credit for emissions reductions that would not have happened anyway. It sounds simple. It is practically impossible to verify.
Most deforestation avoidance credits (REDD+ credits, in the jargon) are based on a counterfactual: what would have happened to this forest if the protection program did not exist? That counterfactual is inherently unobservable. The carbon standard-setters have developed methodologies for estimating it — comparing deforestation rates in protected areas to rates in "comparable" unprotected areas — but these methodologies have proven systematically biased toward over-crediting.
The 2023 Guardian/Science investigations found that the standard counterfactual methodologies dramatically overestimated the threat to forests that were being protected, generating large quantities of credits for forests that would likely not have been deforested with or without the protection program. In some cases, protected areas had lower deforestation rates than comparison areas not because the program was working but because they had lower deforestation risk to begin with.
This is not fraud in the conventional sense — the project developers and standard-setters were not lying about their data. They were using methodologies that were known to have limitations, in an incentive environment where overstating carbon reductions increased revenue for everyone in the supply chain: project developers, standards bodies, and the corporations purchasing credits.
Metaculus forecasts a 74 percent probability that voluntary carbon market volume will be at least 30 percent lower in 2027 than its 2021 peak, as institutional buyers shift away from low-quality offset credits under regulatory and reputational pressure. The market is already contracting; the question is whether what replaces it will be better.
The permanence problem
Even well-designed forest protection programs face the permanence challenge. Carbon that a forest has absorbed over decades can be returned to the atmosphere in hours if the forest burns. The Brazilian Amazon, which has been the largest single source of REDD+ credits globally, has experienced multiple years of record deforestation since the first major offset agreements were signed. The boreal forests of Canada and Siberia, increasingly important for carbon accounting, are burning at rates that have not been seen in the instrumental record.
The standard response to the permanence problem is a "buffer pool" — an insurance mechanism that requires projects to set aside a portion of their credits as reserves against future losses. The sizes of these buffer pools have been consistently inadequate. When the Bootleg Fire burned through an Oregon forest that had been sold as a carbon credit project in 2021, it released an estimated 30 to 95 years' worth of its sequestration in a single summer. The buffer pool mechanism was not designed for this scale of loss.
Climate change is making the permanence problem structurally worse. The forests most vulnerable to fire, drought, and pest outbreaks — driven by rising temperatures — are the same forests whose climate stability had made them attractive as long-duration carbon stores. The carbon market is selling insurance against a risk that the insurance product itself is increasing.
The political economy of non-reform
The voluntary carbon market grew from approximately $300 million in 2018 to a peak of $2 billion in 2021, driven by corporate net-zero commitments that required large quantities of cheap credits. This growth created a constituency — carbon project developers, standard-setting organizations, consulting firms, financial intermediaries — with a strong interest in maintaining the market's legitimacy and a strong incentive to prevent the reforms that would address its fundamental problems.
The reforms that would make the market work — genuinely conservative additionality methodologies, larger and better-designed buffer pools, independent monitoring rather than self-reporting, liability for project developers when credits prove invalid — would also make the market substantially more expensive. Cheap credits are only cheap because they do not represent real emissions reductions. Credits that represent real emissions reductions cost more because doing things that actually reduce emissions costs more.
The political implication is that corporations facing pressure to decarbonize face a choice: pay for genuine decarbonization at the cost of significant capital expenditure and operational change, or purchase cheap credits that provide the appearance of decarbonization at much lower cost. The voluntary carbon market has, for a decade, made the second option available. The evidence that the second option does not work has been accumulating; the economic incentive to continue using it has not changed.
Kalshi was trading a contract on whether the SEC's climate disclosure rules — which would require corporations to describe and validate their offset-based climate claims — will survive legal challenges and be fully in effect by 2028 at 34 percent. The trajectory of carbon market reform runs directly through regulatory disclosure requirements; without them, self-reported net-zero claims will continue to be largely unverifiable.
What should replace it
The failure of voluntary carbon markets does not invalidate the economic logic of carbon pricing. It invalidates a specific implementation that combined legitimate economic theory with inadequate governance. The lessons for replacement mechanisms are fairly clear.
Compliance carbon markets — mandatory systems like the EU Emissions Trading System and California's cap-and-trade program — have performed substantially better than voluntary markets because they involve actual regulatory enforcement, have been subject to sustained methodological improvement over two decades, and require genuine emissions reductions rather than offset crediting. Their limitation is jurisdictional: they cover only the emissions within their regulatory scope, and global emissions reductions require global coverage.
The most direct alternative to offset-based net-zero claims is actual emissions reduction: capital investment in low-carbon technology, operational changes in supply chains and products, and honest acknowledgment that some portions of the economy are currently unable to decarbonize and that the timeline for their transformation is measured in decades rather than years.
This is not the answer that corporations wanted when they signed up for net-zero commitments at the pace that the 2021-22 peak of climate corporate social responsibility required. It is the answer that physics and honest accounting have always been pointing toward. The carbon market was a way to defer that answer. The deferral is ending.
Daniel Osei is a staff writer at The Auguro covering climate, energy, and environmental policy.