Kenya has capped what it will sell and withheld land carbon from trading altogether. Indonesia has sent grants to three hundred community groups instead. Two opposite instruments in the same week, and both are answers to the same problem: an attribute nobody can cheaply measure cannot be made into a right.
On 6 October Kenya launched its Guide for Strategic Engagement in Carbon Markets 2026, which sets a cumulative ceiling of 10 million tonnes of carbon dioxide equivalent on international transfers between 2025 and 2030 and rests on four stated principles: predictability, uncompromising integrity, people before carbon, and value beyond carbon. Officials framed the ceiling as protection for the country's Nationally Determined Contribution under the Paris Agreement. "Kenya is open to responsible and high-integrity carbon investment," said Environment Cabinet Secretary Deborah Barasa. The Principal Secretary, Festus Ng'eno, put it more bluntly: "For years, carbon markets were discussed as a future opportunity. That future is now here."
Most coverage will stop at the ceiling. The more consequential provision is quieter. As reported when the framework was first described, the annual allocation is capped at about 1.67 million tonnes, a conditional priority list covers renewable energy, transport and waste, and forestry and other land-use projects are excluded for the time being, until baselines and data are strong enough to manage reversal risk.
A country whose carbon-market reputation was built largely on rangeland and forest projects has declined, for now, to let them into the instrument it has just designed.
In the same week, Indonesia moved in the opposite direction with the same asset class. On 6 October it opened the fifth phase of its FOLU Net Sink 2030 Community Fund, allocating IDR 20 billion to roughly 300 community groups, financed through results-based climate cooperation with Norway. The earlier rounds give the scale: phases one to three disbursed IDR 19.31 billion to 561 community groups, reaching 31,512 recipients across 36 provinces, and the fourth phase in February 2026 put IDR 7 billion behind about 200 groups. On those figures the average grant roughly doubles in the fifth phase, from something near IDR 35 million per group to something near IDR 67 million. That arithmetic is ours, not the ministry's.
Two governments, one week, one asset class, opposite instruments. Kenya is withholding a tradable right in land carbon. Indonesia is handing out money instead of creating one.
Both are doing the same thing.
The idea that carbon markets rest on is usually credited to Coase's 1960 analysis of the problem of social cost, and usually only the first half of it. That half says that if a right is clearly assigned, the parties will bargain to an efficient outcome whoever holds it. It is the half that produced three decades of emissions trading.
The half Coase actually spent the paper on is the other one. Once the costs of discovering prices, negotiating, contracting and enforcing are positive rather than zero, the initial assignment of the entitlement stops being neutral and starts determining the result. More awkwardly for market design, Coase's conclusion was comparative rather than enthusiastic: the market is one institution for handling an externality, and firms, regulation and simply leaving the harm uncorrected are others, each with its own costs. Which one is best is an empirical question about transaction costs, not a matter of principle.
Carbon markets inherited the slogan and skipped the comparison. Kenya has just run it.
The sharper instrument for reading both decisions is Barzel's 1982 account of measurement cost and the organisation of markets. Barzel's point is that commodities are bundles of valuable attributes, that measuring those attributes is costly, and that the way a market is organised is largely a response to that cost. Where an attribute is expensive to measure, parties do not simply price it badly. They reorganise around the problem: they bundle, they standardise, they warranty, they integrate, they ration, or they refuse to trade the attribute at all.
A tonne of avoided land-use emissions is an unusually bad commodity by this standard. Its valuable attribute is a counterfactual, what would have happened in the absence of the project, and its durability is conditional on a fire season, a change of government, a commodity price or a land dispute twenty years out. Kenya's stated reason for excluding forestry, that baselines and data are not yet strong enough to manage reversal risk, is a measurement-cost statement wearing environmental clothing. The attribute cannot be cheaply measured, therefore the right is withheld.
The 10 million tonne ceiling is the same logic applied to the state's own position. A government that could precisely measure its mitigation pipeline would not need a cumulative cap; it would know, year by year, what it could safely sell. Rationing is what you do when you cannot yet price.
Kenya's Climate Change (Carbon Markets) Regulations, 2024, in force since 17 May 2024, require that a land-based project on public or community land make an annual social contribution to the community of at least forty per cent of the previous year's aggregate earnings, and at least twenty-five per cent for a non-land-based project on the same categories of land. It is among the most generous statutory benefit-sharing floors anywhere.
It is also a percentage of a number that almost nobody can see. Carbon Market Watch's work with AlliedOffsets found that about nine in ten intermediaries do not disclose their fees or profit margins, and the authors state plainly that no public research exists to establish how much voluntary carbon market finance actually reaches mitigation projects. Where the gross is unobservable and the chain between buyer and project contains undisclosed margins, a mandated share of "aggregate earnings" inherits every ambiguity in the definition of earnings.
This is Barzel's problem in its purest form. The community's entitlement is an attribute of a transaction it cannot measure. Raising the percentage does not fix that; it raises the return to mismeasurement. A right that cannot be audited is not a right. It is a hope with a number attached, and the number is doing reputational work rather than distributive work.
Indonesia's instrument avoids the problem by refusing the asset. A grant of roughly IDR 67 million is verifiable with a bank statement, a receipt and a visit. What it does not do is create something the group owns, or any claim on the value of the carbon outcome if that value rises. The trade is explicit once you name it: legibility in exchange for ownership. Kenya has chosen ownership on paper with weak legibility. Indonesia has chosen legibility with no ownership. Neither has both, and it is not obvious that anyone currently can.
When a valuable attribute cannot be measured cheaply, a government has four instruments. Each one parks the residual risk somewhere different, and the choice of where is the policy.
Withhold the right. Kenya's exclusion of forestry and land use from Article 6 transfers. The residual risk sits with project developers and the communities who were counting on that revenue, who get certainty in exchange for nothing.
Ration the quantity. The 10 million tonne cumulative ceiling, with roughly 1.67 million tonnes a year. The residual risk sits with buyers, who face a queue, and with later projects, which may find the budget spent.
Mandate a share of a base you cannot see. The forty and twenty-five per cent floors. The residual risk sits with communities, who hold a claim whose denominator is set by the counterparty.
Bypass the market. Indonesia's community fund. The residual risk sits with the public purse and, in the long run, with the communities, who receive cash flow rather than an asset and must reapply each phase.
The useful question for a funder is not which of these is right. It is which one a given programme is quietly assuming. A "carbon market readiness" grant usually buys registry software, methodology training and transaction support, which is to say it buys the third option, the one whose residual risk lands on the community. Funding the measurement capability that would make the first option unnecessary is slower, less visible and considerably rarer.
The documents describe intentions. Three things would tell you whether the entitlements are real.
Ask communities what they believe they are owed, and on what basis. The gap between the statutory percentage and the understood percentage is itself a finding, and it is obtainable in a week of interviews. Our note on what numbers can and cannot establish sets out why this cannot be done with a survey instrument alone.
Reconstruct one disbursement chain end to end, in both countries. For Kenya, from credit sale to community account. For Indonesia, from the Norwegian results-based payment through the environmental fund management agency to the group's bank account. Compare the number of hands and the recoverable share.
Follow the Indonesian groups from a previous phase. With 561 groups funded in the first three phases and roughly 200 more in the fourth, there is now a cohort old enough to ask the only question that matters about a grant: what is still running, and what stopped when the money did.
For governments, the conclusion is uncomfortable but clean: a benefit-sharing percentage is only as strong as the audit right attached to it, and an audit right is cheaper to legislate than a higher percentage. For funders, it is that measurement capability is the asset that makes every other instrument available, and it is the one least likely to appear in a results framework.
If this reads close to a decision you are weighing, we would like to hear about it. A short note is enough to start.
The Lab is an independent research team, not an advocacy organisation. We do not campaign, endorse products, or accept commissions where a finding is expected in advance. Where a partner's own theory of change is what needs testing, that is a legitimate engagement, and the Lab tests it honestly.