Market structure
The empty chair in carbon risk.
The analytics layer is consolidating into underwriting at speed. Every pairing joins a view formed before the fact to the capital carrying the risk. None of them is an outcome record, because the asset they would pair with has never been built.
Something unusual is happening in carbon risk analytics, and it is happening quickly enough that most descriptions of this market are already out of date. Through 2025 and 2026 the specialist underwriters writing carbon risk have been pairing with external data providers: rating houses feeding assessment data straight into underwriting platforms, market-data firms co-authoring supply forecasts with managing general agents, monitoring technology supplied into broker facilities. Capacity is being syndicated, strategic investors are taking stakes, and the analytics layer is consolidating into the underwriting stack at speed.
Every one of those arrangements does the same thing. It joins a view formed before the fact to the capital that carries the risk. Not one of them supplies a record of what happened after.
That is not an oversight by any of the parties involved. It is because the asset they would need to pair with has never been built.
Two questions, and only one is being answered
Underwriting a carbon position requires answers to two questions that sound similar and are not.
What is true now? Is this project real, is the boundary defensible, is the methodology sound, is the forest standing this month, has anything changed since verification. Monitoring, ratings and satellite analytics answer this, and the honest assessment is that they answer it well and are improving fast.
What happened before, and how often were we wrong? When credits of this type, this vintage, this geography and this risk profile were written previously, how often did they reverse, get invalidated, lose eligibility or fail to deliver, and by how much. Nothing in the current stack answers this.
The first question is a monitoring problem and this market is solving it. The second is a record-keeping problem and this market has not started. The distinction matters because the second question is the one that gets asked when a book leaves the building.
The moment independence becomes worth paying for
Independence in carbon is usually discussed as an integrity virtue, which badly undersells it. Its commercial value is concentrated at one specific point in the chain: the moment a book is sold on.
An underwriter's internal model is the right tool for deciding what to write and at what price. It is the wrong tool, structurally rather than qualitatively, for one job: proving that book to somebody else. A reinsurer asked to take a share, a capital provider funding the vehicle, a rating committee assessing it or an auditor signing it cannot validate a book using the model that wrote it.
This is not a criticism of anyone's modelling. It is a property of who built it.
A book is not a market record. It is a sample selected by the party being assessed, priced by the party being assessed, in the years that party chose to write.
Suppose a carbon underwriter has been writing since 2022 and has three full years of claims experience by 2028. That experience is real and valuable, and it is still not a loss table, for three reasons that cannot be engineered away.
It covers only the risks that carrier chose to write. Selection is baked into every observation, which is exactly what a reinsurer needs to see through and cannot.
It cannot say how the risks that were declined performed, which is the counterfactual that distinguishes skill from luck.
And it cannot be published without exposing pricing. An internal book is commercially confidential by construction, which means it can never become the shared reference the market prices against.
What the absence of a record actually costs
The best evidence for what happens to a market without a shared outcome record does not come from carbon. It comes from the adjacent field of physical climate risk, where thirteen specialist providers were asked to assess the same one hundred properties. Their damage estimates correlated as weakly as 0.2.
The instinctive response is that some of those vendors must simply be better. That may well be true, and it is unfalsifiable, which is the entire problem. Without a record of what actually happened to those buildings, there is no way to establish which estimate was closer. The academic literature makes the same point from another direction: a climate risk premium identified using one vendor's data disappears when the analysis is repeated using another's. If the answer changes with the supplier, the answer was never about the asset.
Carbon has the identical structure, one step earlier. Several rating providers, different methodologies, genuine and public disagreement, and no mechanism whatsoever for settling it. No rating product in this market has ever been publicly backtested against realised outcomes, because the outcomes have never been collected.
Why the seat stays empty
If the gap is this visible, the obvious question is why a well-capitalised incumbent has not simply filled it. Four structural reasons, and each one is worth understanding because each is also a defence.
| Party | Why the outcome record is not theirs to build |
|---|---|
| Rating providers | An outcome record is the instrument that grades their own product. Publishing it invites the comparison. The incentive is not absent, it is inverted. |
| Underwriters and MGAs | Their record is their book, and their book is their pricing. Publishing it means publishing their edge. |
| Brokers | Paid on placement. A determination party whose economics depend on transactions happening carries the same structural problem as a verifier paid by the developer. |
| Registries and standards | Publish their own events, which is the job. Assembling a cross-scheme record that compares programmes against each other is a different job with obvious institutional friction. |
| Data aggregators | Mirror registry transactions extremely well. But issuance and retirement are not losses, and no aggregator currently distinguishes a wildfire buffer draw from a clerical correction. |
The seat is empty because everyone who could occupy it is disqualified by their own economics. That is the same reason every physical commodity market ended up with buyer-paid or exchange-paid inspection: seller-paid assurance is not credible at size, and the parties with the data are the parties with the incentive not to publish it.
The rotation makes it worse, not better
The market's shift from avoidance toward removals is usually described as a quality upgrade. In outcome-data terms it is the opposite.
Avoidance credits have twenty years of programme history, several regulator-published reversal records and a substantial academic reassessment literature. It is a thin evidence base by the standards of any mature asset class, and it is far richer than what exists on the other side.
Durable removals have a handful of operating facilities, contracts mostly not yet due, at least one significant supplier insolvency, and a delivery record that almost nobody has assembled with the censoring handled properly. The rotation moves capital toward the pathway where the outcome record is thinnest and the failure mode is most severe: a removal that fails to deliver is not a partial loss, it is a contract that did not happen, against a counterparty that may not survive the decade.
The asset class is changing. The question, what happens when it does not perform, did not.
What filling the chair actually requires
Three things, in order, and none of them requires displacing anybody's internal model.
Publish the outcome record. Every realised loss, dated, typed and resolved to the document that decided it, free to read and cite. It is the test set the whole class currently lacks, and it has to be free, because a record people pay to see cannot become the standard, and the standard is the point.
Run in parallel rather than in replacement. The useful proposition to an underwriter is a benchmark: your model and an external read on the same submissions, compared over twenty risks. Nobody has to lose an argument for the comparison to be worth having, and after twenty submissions the comparison itself is the document.
Answer accumulation at the book level. Reversals are physically correlated by fire weather, drought and regional policy. A portfolio of forest projects in one basin is not diversified in any useful sense. That is a question about the whole book rather than any single risk, and it is the first question a capital provider asks.
The uncomfortable sentence
If you write carbon risk today, your analytics are probably better than the market gives you credit for. Several of the stacks assembled in the last two years are genuinely impressive pieces of engineering.
The problem is not quality. It is that the person who has to take your book cannot check it with your tools, and there is currently nothing else for them to check it with.
That gap will not be closed by anyone selling credits, and it cannot be closed by a carrier's own model. It requires a record that belongs to no one in the trade. Every comparable market worked this out eventually, usually after an event that made the absence expensive. Carbon still has the option of working it out first.