Let me start with a summary of some headlines and marquee events from the past couple of years that you have likely encountered if you follow the carbon market.
- In July 2025, India’s Ministry of Environment, Forest and Climate Change published a list of activities eligible under Article 6.2 of the Paris Agreement.
- In August, India signed its first bilateral cooperation agreement under this framework with Japan.
- Indonesia lifted its four-year pause on international carbon trading through a Presidential Regulation.
- Kenya mandated that 40 percent of net earnings from land-based carbon projects (on public or community land) flow to community beneficiaries.
- At COP30, Brazil launched an Open Coalition on Carbon Markets backed by 18 countries.
- KOKO Networks, once positioned as a poster child for clean cooking financed by carbon credits, went into financial distress amid a broader weakening of the voluntary market and questions about cookstove methodology.
- Microsoft’s shift toward durable removals has added to the uncertainty around nature-based voluntary credits.
If you read these stories, you will have noticed how casually the terms carbon market and carbon credit move across them, as if they referred to a single, coherent thing. They do not. India’s Article 6.2 list, Kenya’s benefit-sharing rule, KOKO’s cookstove credits and Microsoft’s procurement strategy sit under different mechanisms, with different rules, different accountability and different consequences for the climate. The commentary was often thoughtful. But little of it did justice to the distinctions that any serious discussion of pricing, trade or environmental outcomes actually requires.
Let me expand on those headlines to give some more detail. India’s list concerns Article 6.2, the sovereign accounting route under the Paris Agreement. Indonesia’s regulation authorises both Article 6 and voluntary activity. Kenya’s mandate applies across all land-based projects credits irrespective of the market mechanism. Brazil’s coalition is a political coordination platform. KOKO’s crisis was about voluntary cookstove credits under Verra methodologies. Microsoft’s shift is a corporate procurement decision in the voluntary market. These are different things, but the vocabulary makes them all look like the same thing.
The carbon market, or carbon credit, nomenclature works as if one name had been given to five individuals from the same family, connected by shared lineage. They share an origin, but they are not the same person. And our habit of using one name for all of them is not a harmless simplification. It makes the market harder to navigate, harder to regulate, and harder to use for the two things it exists to do: price a carbon externality and move finance toward climate action.
One word, five instruments
Let me expand on what the missing vocabulary, or the over-simplification, does in practice. Take the phrase carbon market itself. The compliance market, which includes the EU ETS (European Union Emissions Trading System), California’s programme and India’s forthcoming compliance mechanism under the Carbon Credit Trading Scheme, among others, now covers roughly 29 percent of global greenhouse gas emissions, with about 87 pricing instruments in operation and more than 107 billion dollars in government revenue in 2025, according to the World Bank’s State and Trends of Carbon Pricing 2026.
The voluntary market is a different order of thing entirely. It recorded transaction values of around 535 million dollars in 2024, with volumes down about a quarter on the year, as buyers moved toward what they regarded as higher-quality credits. That figure comes from Ecosystem Marketplace’s State of the Voluntary Carbon Market 2025.
The prices tell the same story. EU allowances traded in 2025 in a band of roughly 60 to 75 dollars per tonne. Voluntary credits averaged closer to 6 to 7 dollars, and even that average conceals a wide spread of activities, with removal credits carrying a premium of nearly 400 percent over reduction credits. These are not different prices for the same product. They are different products with the same label.
The instruments underneath are genuinely distinct. A compliance-based emission allowance is a legal right to emit under a cap, granted by a government. A voluntary credit is a documentary claim that a reduction happened somewhere else, governed by an independent standard. An Article 6.2 transfer is a movement between two national ledgers, with a corresponding adjustment to prevent double counting. An Article 6.4 credit sits under UNFCCC supervision. A contribution claim, in the VCMI (Voluntary Carbon Markets Integrity Initiative) sense, deliberately avoids saying that one tonne here cancels one tonne there.
Each carries its own obligations, its own verification, its own line of accountability. So, when a Kenyan official, an Indian regulator and an Indonesian developer all say carbon market, there is no guarantee they mean the same thing. When a Swiss buyer, a Singaporean intermediary and a Brazilian developer all say high-integrity carbon credit, they may not be pointing at the same object at all.
Is vocabulary really the problem?
It is worth being fair to our need for simplification. Every emerging field leans on rough language early on, and carbon was no exception. A shared, imprecise vocabulary is often what lets very different actors cooperate before they could agree on precise definitions. It is arguably what got environment ministries, finance ministries, corporates and NGOs into the same room at all.
Simplification did something more fundamental too. It produced a single tradable unit. Donald MacKenzie, in his study of how carbon markets were built, shows that turning a tonne of one gas and a tonne of another into one interchangeable unit is a political and technical achievement, not a fact of nature, and that supporters and critics alike underestimate how much institutional work it took. His point is cautionary rather than celebratory: the same act of making things the same flattens real differences in how, where and when emissions are avoided or removed. But without that simplification, a global market would not have been conceivable at all.
But here is the issue that simplification creates. The compacting of many distinct things into one, the very move that made this market possible, is now hindering its growth. A market has to compare things, and comparison needs the unit to hold still. So long as the unit is only a label in a negotiation, its vagueness is harmless, even useful at times. But the moment a price is attached to it, that same vagueness stops being a convenience and becomes an obstacle, because buyers can no longer tell what quality they are paying for or how to price it at all.
This is not an abstract argument. It has been studied in a neighbouring sector, one that sits inside the same broad sustainability domain: sustainable finance. When Berg, Kolbel and Rigobon examined ESG ratings from six major agencies, they found that ratings of the same companies correlated only between 0.38 and 0.71. Conventional credit ratings, by contrast, sit near 0.99. Most of that divergence came not from disagreement about what to value, but from measurement, from each agency turning the same word into a different yardstick.
Two consequences followed, and both carry straight over to carbon. Buyers could no longer tell leaders from laggards. And, more damaging, companies lost the reason to improve, because the market kept telling them different things about what good even meant. The parallel is exact. ESG had six referees who could not agree on the score. The carbon market has the same condition, one word doing the work of five, and it produces the same result: a price that cannot locate the quality beneath it.
So, the problem is not that we simplified too early. We simplified correctly, and it worked. The problem is that we are now in the later stage, where the same undifferentiated language that once enabled coordination has started to inhibit scale, frustrate regulation and let finance flow toward the appearance of climate action rather than the action itself.
A closer look at the most frequently uttered word in carbon credit transactions: integrity
Integrity is one of the central quality claims of the carbon credit industry. Yet it means several different and unrelated things, even though we have institutions dedicated to defining principles, guidelines and ratings for it.
The Integrity Council issues a pass-or-fail label against its ten Core Carbon Principles. The independent rating agencies issue graded scores instead, and they do not even score the same dimensions. BeZero builds its rating on additionality, carbon accounting and permanence. Sylvera scores carbon, additionality and permanence, but deliberately keeps co-benefits out of the headline grade. Calyx refuses to blend the dimensions at all, and reports greenhouse-gas integrity, development impact and social risk as separate scores.
The predictable result is that the same credit receives different verdicts. A project can be rated BBB by one agency and C by another, and the agencies themselves now advise buyers to treat such divergence as a warning and to cross-check across providers. Carbon Market Watch’s comparison of the rating agencies documents this divergence in detail. When the referees tell you to consult several referees because they disagree, integrity has stopped describing the credit and started describing the hope of the person selling it.
And that is only the narrow case, where every rater is at least looking at the same object, the credit. Widen the lens and the same word is stretched across quite different objects. VCMI uses it to judge a company’s public claim. SBTi uses it to judge a company’s target. The Greenhouse Gas Protocol governs how a company accounts for its emissions in the first place. CORSIA (Carbon Offsetting and Reduction Scheme for International Aviation) decides whether a credit is eligible for one specific compliance use in aviation. Article 6 decides whether a government has deducted a transferred credit from its own books. Five different questions, one borrowed word. A credit can satisfy any one of these and fail the rest, and very few people using the phrase in a headline could tell you which gate they mean.
And what is additionality here?
Additionality is the most critical dimension of any carbon credit framework. If a project is not additional, there is no basis for the credit at all. But additionality is not a single measurement.
It is at least four. There is financial additionality, which asks whether the project would be viable without carbon revenue. There is regulatory additionality, which asks whether the reductions go beyond what the law already requires. There is common-practice additionality, which asks whether the activity is already widespread in the region. And there is barrier additionality, which asks whether the project overcomes some technological or institutional obstacle. These are four different questions about four different things: the balance sheet, the statute book, the regional adoption curve and the barrier landscape.
A project can pass one and fail another without any contradiction. A soil-carbon project may be genuinely unviable without carbon money, and so financially additional, while failing common practice because the same farming technique is already normal in that district. Different standards then bundle the tests differently. Verra and the American Carbon Registry lean on a hybrid of three. The old CDM stacked several conditions together and waved some project types through as automatically additional. So even the number of hurdles a credit must clear is not constant across the market.
The raters diverge again, and here additionality rhymes with integrity. BeZero explicitly rejects the yes-or-no test and scores additionality as a probability on an eight-point scale. The CDM treated it as a gate you either cleared or did not. So, a credit can be additional as a verdict and moderately likely additional as a score. Those are not the same statement, and they are not even the same kind of statement.
Underneath all of this sits a problem no methodology is able to address comprehensively. Every additionality test is an attempt to establish what would have happened in a world without the project, a world that never existed and cannot be inspected. Even the Carbon Offset Guide, a source broadly sympathetic to offsets, concedes that these tests require subjective judgement and rest on assumptions about the future. And the party best placed to model that absent world is very often the party that profits when the model returns the answer additional.
Things get more complicated when we start looking at the interplay of these words with others. Permanence (another quality parameter for carbon outcomes) asks whether the carbon stays out of the atmosphere and for how long. Co-benefits capture the health, biodiversity or livelihood value alongside the carbon. These three properties trade off against each other, and the vocabulary does do justice to this trade-off. The project types’ richest in co-benefits, such as smallholder cookstoves and community forestry, are often the hardest to prove additional and the hardest to guarantee permanence. The project types easiest to prove permanent, such as engineered geological storage, are often thin on co-benefits.
Integrity and additionality are only two examples. Permanence, baseline, leakage and removal each carry the same problem: a single word compacting several distinct ideas, making the market easier to talk about and harder to run.
We need a more effective and precise vocabulary
This article is not a critique of carbon markets, nor an argument against them. It is an argument about the language we use to run them, and about who pays when that language stays loose. That loose language has a cost, and it does not fall evenly. Project developers and investors in the global South, who most often meet these definitions for the first time when their own credits are being assessed, are the ones most exposed when a category shifts or a methodology is downgraded.
So, if the argument holds, the remedy is not more generalisation but more distinction. This does not require new institutions or a grand redesign. It requires the discipline to say which of the five family members we are talking about, every time, and to resist the convenience of the shared surname. Some of the machinery for this already exists. The registries are beginning to label Article 6 status separately. The rating agencies already score the dimensions apart, even if the headlines blur them again. What is missing is the same discipline in policy language and in public discussion.
This matters most for the countries now writing their rules. India, Indonesia, Kenya, Brazil and others are deciding, in this narrow window, what their carbon markets will mean and how their credits will be described. If they inherit the loose vocabulary wholesale, they inherit its confusions and its asymmetries with it. If they insist on the distinctions, they give themselves a market that can actually be navigated, regulated and priced.
The point of a carbon market was never the credit itself. It was to put a price on a harm that markets had ignored, and to send money toward repairing it. A vocabulary that cannot tell its own instruments apart cannot do that job. Getting the words right is not a semantic indulgence. It is a precondition for the market doing the thing we built it to do.
Further reading
- Donald MacKenzie, “Making things the same: Gases, emission rights and the politics of carbon markets” (Accounting, Organizations and Society, 2009). The clearest account of why trading emissions requires an act of equivalence-making rather than a natural measurement. If you read one thing here, read this. Larry Lohmann’s companion piece in the same issue, “Toward a different debate in environmental accounting,” is the critical counterpart.
- Berg, Kolbel and Rigobon, “Aggregate Confusion: The Divergence of ESG Ratings” (Review of Finance, 2022). Six agencies rating the same companies agree only loosely, and most of the disagreement comes from measuring the same word differently. The closest thing we have to a controlled experiment on the cost of vague terminology.
- Fankhauser et al., “The meaning of net zero and how to get it right” (Nature Climate Change, 2022). Leading scientists arguing that net zero left physical science without its meaning being pinned down.