The day carbon became a number the auditor has to sign

For most of its short life, the corporate carbon credit lived outside the financial statements. It was a sustainability line item, a marketing budget with a retirement certificate stapled to it, signed off by a team that reported to the chief sustainability officer rather than the chief financial officer. On 19 May 2026 that ended. The Financial Accounting Standards Board issued ASU 2026-02, creating a new ASC Topic 818 for environmental credits, and in doing so it pulled the carbon book onto the balance sheet, under the eye of the controller, the audit committee and, decisively, the external auditor.

This is not a disclosure tweak. It is a change in who is accountable for the number. A credit that used to be expensed and forgotten is now an asset that must be carried, tested and, where it has lost value, written down on a fixed reporting cadence. The people who must sign that number have, until now, never had to form a defensible view on what a tonne of avoided or removed carbon is actually worth. That is the quiet revolution inside a dry accounting update.

What the standard actually does

01 · Two books, two regimes

ASU 2026-02 draws a hard line between two kinds of credit. Compliance credits, the allowances a company is legally required to surrender under a regulatory scheme, are carried at cost and are not impairment-tested. They behave like a prepaid obligation. Noncompliance credits, which is to say the entire voluntary market, are treated very differently. They are carried at cost less impairment, tested at each reporting date, with an optional fair-value election that runs changes in value through earnings.

In plain terms, the credits a company bought to make a "net zero" or "carbon neutral" claim are now the credits exposed to write-downs. The discretionary book, the one nobody on the finance side priced when it was acquired, is the book that now sits in the impairment machine.

02 · Impairment is a one-way door

The standard makes impairment irreversible. If a vintage is written down because a methodology fell out of favour, a project under-delivered, or a ratings gate moved against it, that loss is locked in. It cannot be reversed if sentiment recovers. This is the same asymmetry that governs goodwill and other indefinite-lived intangibles, and finance teams know what it means: an impairment is a permanent dent in earnings, not a temporary mark that washes out next quarter.

The credits a company bought to make a claim are now the credits exposed to write-downs.

That asymmetry changes the psychology of the carbon book entirely. A purchasing decision made in good faith two years ago can surface as a charge against this year's profit, with no path back. The asset can only ever be worth less than the company paid, never more, unless the fair-value election is taken, and that election carries its own volatility into the income statement.

03 · Gross, not net

ASU 2026-02 requires gross presentation. The credit asset is not netted against the obligation it was bought to satisfy. Both sides appear in full. The standard introduces an Environmental Credit Obligation, the ECO liability, measured on the carrying amount of owned credits where the obligation is funded, plus the fair value of the unfunded excess where it is not. The asset and the liability stand separately on the face of the statements.

Gross presentation matters because it removes a hiding place. A company can no longer assume the credits it holds quietly offset the obligation it owes and present a tidy net figure. The market sees the full size of the position and the full size of the commitment. For a business carrying a large legacy book against a large forward neutralisation pledge, both numbers are now visible, and both are auditable.

The reporting-date problem

04 · A market with no clock now has one

The hardest part of the standard is the cadence. Impairment is tested at each reporting date. The voluntary carbon market has never operated on a quarterly clock. It is opaque, thinly traded, bilaterally priced and split across dozens of methodologies and vintages, with weighted average spot prices around 6.10 US dollars in 2025 but a quality spread that runs from roughly 3 US dollars at the weakest tier to north of 26 US dollars for high-rated afforestation, reforestation and revegetation credits by late 2025. There is no clean screen price to lift.

Yet a controller closing the books in March, June, September and December must now produce a defensible carrying value and a defensible impairment view for a book that may span ten vintages, six methodologies and four geographies. The question "what is this worth today" cannot be deferred to an annual sustainability report written six months in arrears. It has to be answered on the close calendar, to the same evidentiary standard as every other line on the balance sheet.

The granularity is the trap. Impairment is not a single book-level judgement; it is a series of judgements at the level where value actually moves. A reforestation vintage from one geography can be repriced by a methodology downgrade while a removal credit from another holds firm. A CORSIA eligibility change can strand one cohort and leave its neighbour untouched. A controller who tests the book as a single lump will either over-impair, taking an unnecessary charge against earnings, or under-impair, leaving an overstated asset for the auditor to find. Neither outcome survives scrutiny, and both are the predictable result of treating an opaque, heterogeneous book as if it were one fungible holding.

There is also a tail risk the close calendar exposes. Because impairment is irreversible and tested every period, the first reporting date after an adverse event, a registry suspension, a ratings downgrade, a buffer-pool shortfall, is the moment the loss crystallises. The market may take months to settle on a new clearing price, but the company cannot wait for that consensus. It must form a view now, document it, and defend it. The cadence converts slow-moving market sentiment into a hard quarterly decision with a permanent earnings consequence.

05 · The advisory firms own the interpretation, not the answer

Every major firm has rushed to publish. Deloitte, PwC, EY and KPMG, alongside BDO and Crowe, have all issued guidance on ASU 2026-02. They own the interpretation of the standard, how to classify a credit, when to test, how to present the ECO. What none of them sells is the continuous, defensible input: the independent, reporting-date carrying value and impairment read for a specific book of credits. Interpretation tells you the rules of the game. It does not tell you what your particular position is worth on 31 December.

Why a seller's mark is not an audit-grade input

06 · The conflict baked into the obvious source

The instinctive place to get a price is the platform that sold you the credit, or a data provider that serves both buyers and sellers. This is precisely where an external auditor will baulk. A fair-value or impairment input is only as good as the independence of its source. A mark produced by a party that originated, brokered or holds an interest in the same credits carries an obvious conflict, and an obvious incentive to support the carrying value rather than challenge it.

A mark from the people who sold you the credit is not an independent input. It is a counterparty's opinion of its own inventory.

Auditors are trained to discount counterparty-sourced valuations for exactly this reason. A market mark from a sell-side or two-sided source is, in substance, a counterparty's opinion of its own inventory. It may be a perfectly reasonable trading price. It is not an independent, audit-grade valuation input, and presenting it as one invites the question every audit committee fears: who checked the checker?

07 · The standard finance teams will demand

The discipline finance will reach for is the discipline it already applies to every other hard-to-value asset: an independent valuation, methodologically transparent, refreshed on the reporting cadence, defensible to a sceptical auditor, and produced by a party with no position in the underlying credit. The input must be classifiable, by vintage, method and geography, so that impairment can be tested at the right level of granularity. It must reflect not only price but integrity status, because a methodology downgrade or a CORSIA eligibility change can strand a whole vintage regardless of where it last traded.

This is the work that ASU 2026-02 has made compulsory and that the market is not yet built to deliver. The standard creates the demand for an independent fair-value and impairment read. It does not create the supply. That gap is now a finance problem on a statutory clock, mandatory for fiscal years beginning after 15 December 2027 and open to early adoption today, not a sustainability aspiration. The method every controller will be measured against is being set now.

The wider consequence

Once carbon is a number the auditor signs, every adjacent decision changes. Procurement can no longer treat a credit purchase as a soft cost; it is acquiring an impairing asset. Treasury has to consider how the book behaves through earnings if the fair-value election is taken. The audit committee has to satisfy itself that the carrying value rests on defensible, independent evidence. The general counsel has to recognise that an impairment is also a paper trail, a documented admission that an asset bought to support a public claim has lost value.

There is a strategic question hiding behind the mechanics, and it belongs to the chief financial officer rather than the sustainability team. If a credit can only be carried at cost less an irreversible impairment, and if the use cases that gave it value are narrowing under SBTi and the EU consumer rules at the same time, the rational finance response might be to impair, stop buying and walk away. That is the central commercial tension the standard introduces. But walking away does not extinguish the forward commitments already signed, nor the public claims already made, nor the obligation the ECO liability now puts on the balance sheet. The book has to be valued and managed whether or not the company buys another tonne.

The transition window is short. ASU 2026-02 was issued in May 2026, and finance teams are already inside the first reporting cycles where the classification and measurement choices bite. The companies that treat this as a checkbox will discover, at the worst possible moment, that their carbon book is an unvalued liability dressed as an asset. The companies that treat it as a genuine valuation problem will be looking for an independent source they can stand behind.