Recognition and Its Limits in the CBAM Article 9 Default Carbon Price: A Per-Origin Translation Component

Terawatt Times Institute · EU Transparency Register No. 8269215104555-62 · ISSN 3070-0108

Editor's Note

Drawing from extensive multi-country theoretical and empirical research, this official policy submission by the Terawatt Times Institute directly responds to the European Commission's targeted call regarding the draft implementing act serving as the financial clearinghouse for cross-border carbon accounting—specifically the Commission Implementing Regulation laying down rules for the conversion of the carbon price paid in a third country into a corresponding reduction in the number of CBAM certificates to be surrendered, the evidence of payment of that carbon price, and the qualifications of the independent person (Ref. Ares(2026)4841230). By evaluating the operational mechanics of Article 9, this text proposes an actionable, legally compliant framework to ensure technical recalibration and structural durability.

Abstract

The draft implementing regulation governs how a carbon price paid in a third country becomes a reduction in CBAM certificates, what evidence proves that payment, and how the independent person who certifies it is qualified. Its operative core is the default carbon price that the Commission publishes under Article 9(4) and 9(5) of Regulation (EU) 2023/956. That default is the channel through which most declarants will claim any deduction, because verified facility-level data remains unavailable across most of the exporting base, and the share of declarations resting on default values through the transitional period was large. This submission concerns the default carbon price methodology and the recognition boundary that shapes it.

We treat the mechanism as a system of recognition: the default carbon price decides which carbon costs count and which do not, and the submission examines where that recognition reaches its limits. The first limit is in what the draft recognises. It recognises carbon cost by the form of the instrument that imposes it, a price denominated in currency per tonne under a tax, levy, fee, or emissions trading system, while a substantial population of carbon costs that exporting economies carry through other instruments falls outside recognition even where the effect on emissions is comparable. We call this filter commensuration selectivity, trace its three operating conditions to the regulation's own text, and show that it produces a self-confirming dynamic: by registering only cash-denominated instruments, the mechanism channels exporting-country policy toward those instruments and then reads the resulting landscape as confirmation of its own design. Evidence from thirteen exporting jurisdictions demonstrates that the default methodology misreads real carbon performance in a consistent direction, by margins that frequently exceed the producer's margin on the trade. The second limit is in how the values are assigned. The same evidence supports a finding we term default-setting drift: the gap between an assigned default and physical intensity opens at the moment of assignment and varies without pattern across origins, running from a factor near 1.7 to a factor near 14 and turning negative where a default sits below actual intensity. A distortion of this shape cannot be closed by a better global average or a uniform mark-up, and it calls for a per-origin correction applied within the methodology.

The request is that the Commission, in calculating and publishing default carbon prices under Article 9(4) and 9(5), incorporate an audited translation component converting recognised implicit carbon costs into a carbon-price equivalent, subject to a conservative precision adjustment. This sits within the implementing act's mandate. It carries through the recognition the draft already grants in Recital 15 and Article 8(2), and it leaves the base regulation's definition of carbon price for the legislator and the Article 30 review.

1. The submission and its scope

The Terawatt Times Institute is a nonprofit research institution registered in the EU Transparency Register, working on carbon governance and the intersection of climate policy with trade. Our standpoint is structural and methodological. We represent no exporting country and no industry, and the analysis below holds across the European, British, American, and Canadian border carbon instruments, a point we return to in Section 5.

This submission addresses the draft implementing regulation Ares(2026)4841230 [2], and within it the default carbon price methodology authorised by Article 9(4) and 9(5) of Regulation (EU) 2023/956 [1]. That methodology runs through Article 4 of the draft on default carbon prices for precursors and indirect emissions, Article 6(2) on the conversion of a default carbon price into a certificate reduction, and the attribution rules of Annex I [2]. The definition of a carbon price mechanism in Article 9 of the base regulation belongs to the legislator and to the Article 30 review, and nothing in this submission asks the implementing act to alter it [1].

The request is narrow and specific. The default carbon price that the Commission publishes should be capable of registering a class of carbon cost that the present design renders invisible, through an audited translation component built into the methodology itself. Sections 2 and 3 establish why this is sound in principle and why it falls inside the implementing mandate. Section 4 shows across thirteen jurisdictions why it matters in practice. Section 5 places the European mechanism within the wider family of border carbon instruments to show that the pattern is structural to the instrument type. Section 6 states the request at the level of detail the implementing act can adopt.

2. The recognition filter and how it operates

Any system that compares things drawn from different settings has to fix what it will count. A university weighing applicants from different schools cannot treat an A from one as identical to an A from another, so it reads each grade against the context that produced it. The sociology of quantification names the act of rendering heterogeneous things comparable through a common metric commensuration, and a long line of that scholarship has established that the metric, in buying comparability, suppresses whatever it does not encode [11][12]. The suppressed information does not vanish. It reappears as the seam along which the system can be gamed, misread, or contested.

The draft makes its commensuration choice with precision. Article 2(3) defines a carbon price mechanism as a carbon tax, carbon fee or carbon levy, or an emissions trading system [2]. Recital 8 binds recognition to schemes taking the form of a tax, levy, or fee, or of allowances under a greenhouse gas emissions trading system that is binding in nature and imposes obligations on all operators in the relevant sector without discrimination [2]. The carbon price that survives conversion into a certificate reduction is therefore a cash sum per tonne, levied through one of a closed set of instrument forms. We call this filter commensuration selectivity: the methodology recognises carbon cost by the form of the instrument that imposes it, before any question of magnitude arises [16].

The filter rests on three conditions, each anchored in the regulation's text rather than in technical necessity [15][16]. The first is temporal commensurability. A recognised cost has to resolve into flows keyed to the annual reconciliation cycle of the EU emissions trading system, rather than into a stock of capital amortised over the life of an asset. The second is instrument fungibility. A recognised cost has to be priced in cash terms that the regulation accepts as a tax, levy, fee, or tradable allowance. The third is verifier authority. A recognised cost has to produce a document that an accredited independent person can attest to under the evidentiary standards of Annex I and the certification architecture of Chapter IV [1][2]. Capital committed to clean electricity, to low-carbon hydrogen, or to electrified process equipment fails all three conditions at once. It is stock-denominated, it generates no cash outflow at the moment its abatement effect accrues, and it yields no instrument an independent person can certify as a carbon-price payment.

Selectivity carries a generative effect that deepens it over time. A government designing industrial climate policy in the shadow of the European, British, and American border mechanisms confronts an asymmetry: a carbon tax or an emissions trading system earns border recognition, while a capital-support programme, a regulatory mandate, or state-owned clean generation earns none. Rational policy under that asymmetry channels scarce domestic bandwidth toward the cash-denominated instruments the border infrastructure can see, whether or not those instruments are the best fit for the country's emission profile. The mechanism thus shapes the population of policies it will later observe.

The two movements compose into a closing loop. Having registered only cash-denominated instruments, the regulator observes an evidentiary landscape in which taxes and trading systems dominate the recognisable space, and reads that distribution as empirical confirmation that the selectivity filter captures the relevant universe of climate action. The filter selects on criteria that then reshape the population those criteria are applied to, so the observed population appears to validate the filter. This dynamic is testable. Policy diversity that is measurable across CBAM-exposed economies today, carbon taxation in some, trading infrastructure in others, substantial capital programmes elsewhere, is predicted to narrow through 2030 toward cash-denominated instruments unless a recognition channel is opened for the alternatives. The methodology set by this implementing act is one place such a channel can be opened.

3. Paid price and invested capital carry the same carbon cost

Set two producers side by side. They face the same emissions and the same abatement opportunity. The first pays a carbon price of a given amount per tonne and keeps emitting. The second commits capital to electrification, to clean hydrogen, or to process redesign, and removes the tonne. In their effect on the atmosphere the second has done at least as much as the first. In the economics of the firm the capital outlay that eliminates the emission is equivalent to the price that would have been paid to keep it. The behavioural signal a carbon price is built to send has already been received and acted on by the second producer, in advance and at the producer's own expense.

The draft accepts this equivalence in one place, under conditions drawn tightly. Recital 15 and Article 8(2) provide that where an installation benefits from subsidies financed by the revenues of a carbon price mechanism and reinvested in the installation's decarbonisation, the value of those subsidies is not treated as a compensation that reduces the carbon price effectively paid [2]. The conditions are that every installation under the mechanism is eligible for the support, that the granting authority's decision is public, and that the purpose of the support is to cut the beneficiary's emissions [2]. The drafters, in other words, have looked at money flowing toward abatement and declined to net it against the recognised carbon price, on the reasoning that funding decarbonisation differs in kind from refunding a charge. That reasoning concedes the economic substance this submission rests on.

The concession defines both the opening and its limit. The opening is that the draft already registers the substance of decarbonisation investment when the funding originates inside the carbon price mechanism, and the default carbon price methodology is the instrument through which the same substance can be registered when abatement reaches the producer through another channel [1][2]. The limit is equally clear, and stating it plainly is what keeps the request credible. We do not propose that the notional carbon value of avoided emissions be deducted directly as an Article 9 payment. Consider a Gulf aluminium smelter that cuts its footprint through solar power: at a carbon price near €85 per tonne the avoided emissions carry a notional value in the hundreds of euros per tonne of metal, yet Article 9 operates on cash actually paid, the producer's jurisdiction levies no carbon price to pay, and the aluminium accounting boundary excludes the indirect emissions where the solar advantage sits [1]. Treating that notional figure as a direct deduction would require rewriting the cash-outlay basis of Article 9, the Annex II emission boundary, and the free-allowance architecture of the EU ETS in a single move, none of which an implementing act can do [16].

The space between the concession and the limit is exactly where this implementing act works. A direct deduction of notional avoided-emission value lies outside the mandate. The recognition of an evidenced, instrument-backed implicit carbon cost, translated into a carbon-price equivalent and entered into the published default carbon price, lies inside it. The remainder of this submission develops that translation, after establishing through the country evidence why the gap it addresses is consequential.

4. Thirteen jurisdictions, one pattern

Default values carry the methodology. When a declarant cannot present verified facility-level emissions, both the embedded emissions and the recognisable carbon price fall back to figures the Commission publishes, and through the transitional period the large majority of declarations rested on those figures rather than on measured data [7]. Whatever the default carbon price methodology recognises or omits is therefore what most of the trade actually meets at the border. The Institute's country-intelligence programme has reconstructed, at facility and CN-code level and against the published regulatory parameters, how the methodology performs across the exporting base. Each of the thirteen cases below was developed independently and traces every regulatory figure to its source in Regulations (EU) 2023/956, 2025/2547, 2025/2620, and 2025/2621 [1][3][4][5].

Three terms recur. Carbon intensity is given in tonnes of CO₂-equivalent per tonne of product, written tCO₂e/t, the amount of greenhouse gas embedded in each tonne shipped. A default value is the intensity the Commission assigns when measured data is absent, raised by a mark-up that the default-value regulation sets at 10 percent in 2026, 20 percent in 2027, and 30 percent from 2028 for industrial goods, with fertilisers held to a uniform 1 percent [5]. A benchmark, written B_EU, is the intensity of the most efficient EU installations producing the same good, and it fixes how much of an importer's emissions are shielded from charge while free allocation is withdrawn on a schedule that lifts the phase-in factor from 2.5 percent in 2026 toward 100 percent by 2034 [1][4][6].

Table 1. Default carbon accounting against measured intensity, thirteen exporting jurisdictions (2026 parameters unless stated).

Note. The cost figures in the fifth column are each computed at the carbon price used in the underlying jurisdiction analysis, which ranges from about €70 to about €92 per tonne of CO₂ across the cases and reflects the price assumption appropriate to each. Cross-country comparison of the drift documented in Section 4.5 rests on the ratio of the assigned default to measured intensity, columns three and four, which is independent of the carbon price and therefore comparable across all thirteen rows regardless of the figure used in any single case.

The cases sort into three behaviours. Each speaks to a different aspect of the methodology this implementing act sets, and together they describe a single phenomenon expressed in thirteen institutional settings.

4.1 Default values that diverge from measured intensity

The largest group shows the published default departing from measured carbon intensity by margins wide enough to invert the incentive the mechanism exists to create.

Mozambique offers the cleanest illustration, because its physics is unambiguous. The country's sole aluminium smelter runs on hydroelectric power from the Cahora Bassa complex, yet it carries the highest country-specific default for unwrought aluminium found anywhere, 3.198 tCO₂e/t, against a direct-emission intensity that four independent reconstructions place between 1.76 and 1.90 [5]. The high figure traces to a single drought year, when failing reservoir levels forced the smelter onto South African coal-fired backup supply and its operator reported a combined intensity of 11.4 tCO₂e/t, a crisis state that the data hierarchy then captured and fixed into a permanent value [5][14]. A modern pre-baked-anode smelter cannot reach 3.198 as a direct-emission figure under any normal operation, and the cost of the resulting gap is €137.53 per tonne. The methodology here lacks any mechanism to distinguish a transient energy emergency from a structural carbon profile, and the absence is invisible to a producer with limited monitoring capacity until the value is already set.

Egypt and Algeria show the same divergence arising from a different cause, the absence of the producer's actual technology from the default matrix. Egypt's steelmakers run direct reduced iron with electric arc furnaces at roughly 0.84 tCO₂e/t, but the default table for hot-rolled coil offers only the blast-furnace route and assigns 4.454, which places the default-path certificate cost near €249 per tonne against the actual-data figure and leaves a difference of €213 per tonne that exceeds the leading producer's entire operating margin [3][5]. Algeria's export steel is made entirely through direct reduction, the default tags it as blast-furnace at 3.300 against an intensity near 0.70, and because the misclassification moves both the emission figure and the benchmark deduction in the same adverse direction, the gap reaches €138.64 per tonne in 2026 and €256 by 2034 [4][5]. A control case inside Algeria sharpens the point: its ammonia exports, where no route misassignment occurs, show a constant gap of about €16 per tonne, an eight-to-sixteen-fold smaller distortion that confirms the steel gap originates in classification rather than in carbon [5].

Serbia and Bosnia and Herzegovina add the verification dimension. Serbia's cleanest electric-arc producer measures 0.166 tCO₂e/t and would owe €7.62 per tonne on verified data, yet absent an EU-accredited verifier it is assigned the blast-furnace default of 2.651 and owes €99.06, a thirteen-fold swing that turns on the single binary of whether certification is available, and that widens toward eighteen-fold by 2030 as mark-ups and free-allocation withdrawal compound [4][5]. Bosnia's aluminium producer dismantled its primary smelter in 2019 and now remelts scrap at roughly 0.35 tCO₂e/t, while the default table carries only the primary route at 1.700, so the facility is charged 4.9 times its actual intensity precisely for having completed the transition the mechanism is meant to reward, with the annual cost of the gap growing toward €28 million against a monitoring system that costs some €50,000 to €60,000 a year to run [5].

Taiwan and Mexico complete the group from the two ends of the accounting. Taiwan's stainless steel carries a default of 8.670 tCO₂e/t, fourteen times the terminal furnace's own intensity, because the attribution formula loads the high-carbon nickel precursor onto the finished product and the terminal producer's own performance disappears beneath it [3][5]. Mexico shows the benchmark side of the same logic: its scrap-based electric-arc route is shielded on 8 percent of its emissions where a blast-furnace operator would be shielded on 74 percent, so the firm with the lowest emissions pays the most, with the cleanest producer facing €67.9 per tonne [4]. Vietnam rounds out the pattern by isolating the variable: a mill verified at 1.91 tCO₂e/t against a 2.35 default holds a €40.8 per tonne advantage that exists only if it can document the figure, which makes carbon-data capability, rather than carbon performance, the determinant of its border cost [5].

4.2 Paid carbon cost that evaporates in translation

A second group shows carbon cost that producers have genuinely borne dissolving as it crosses into recognition.

China presents the starkest case because the scale of the underlying effort is large. Its national emissions market requires participants to pay for allowances, and its steelmakers have invested heavily in efficiency, yet the carbon price that survives Article 9 translation amounts to roughly €0.30 per tonne, a deduction under two percent of the obligation [1][5]. A Chinese fastener producer on the default path faces €529 per tonne against €119 on verified data, a gap driven by a country-specific default of 6.375 tCO₂e/t that rises to 8.288 with the 2028 mark-up, where the producer's own intensity sits below the EU average [5]. The gap between real and recognisable cost persists because the instruments through which China constrains carbon, differential electricity pricing and mandatory capacity replacement among them, take forms the filter does not register [5]. Serbia's carbon tax, structured to fall only on emissions above a baseline, yields a recognised Article 9 value somewhere between zero and €1.87 per tonne [1]. Turkey shows the cost on both sides of the ledger. Its primary aluminium carries a default grid intensity that overstates the sector's actual electricity emissions by roughly 1.5 tCO₂e/t, worth about €120 per tonne for a producer that cannot document its own power sourcing, and its Special Consumption Tax holds an activatable Article 9 deduction pathway that has gone structurally unused across the responsible ministries [1]. In each case a domestic carbon position exists, and the methodology registers little or none of it.

4.3 Recognition that turns on access, scope, or data integrity

A third group shows the recognition question hinging on something other than price.

Nigeria's most efficient urea plant operates near global best practice at about 1.05 tCO₂e/t and would owe €12 per tonne on verified data against €42 on the default, but the binding constraint is institutional distance from EU-accredited verification rather than the charge, since the fertiliser mark-up is held to 1 percent and the financial penalty alone is modest. Nigerian fertiliser exports to the EU fell 92 percent between 2022 and 2024, from €168 million to about €13 million, as supply chains restructured ahead of the mechanism, a collapse driven by the compliance channel operating independently of the financial one [5]. Ghana's single aluminium smelter sits at roughly 1.75 tCO₂e/t against a 1.700 default, a near-match that holds only while indirect emissions stay outside the calculation; a revision of Annex II to bring them in would move the per-tonne cost from about €39 toward €420, so the recognition state of an entire national export rests on one pending scope decision [1][5]. Morocco shows the inverse of the penalty cases and exposes a data-integrity cost: its fertiliser default of 0.550 tCO₂e/t sits below the producer's estimated 0.76, which makes withholding actual data the cost-minimising strategy and steadily corrodes the quality of the evidence base the methodology depends on [5].

4.4 The pattern

Read together, the thirteen cases describe one behaviour expressed thirteen ways. Across jurisdictions that differ in industry, in instrument, and in institutional capacity, the default carbon accounting misreads real performance, in a consistent direction, by margins that frequently exceed the producer's margin on the trade. The misreading is not random noise around a correct central value. It runs predictably against the producer wherever the producer's actual technology, energy source, or policy instrument sits outside the narrow forms the filter encodes. That predictable directionality is the empirical signature of commensuration selectivity operating as architecture, and the methodology this implementing act sets is what determines whether the signature persists past 2026.

4.5 Default-setting drift

The thirteen cases support a proposition about the default values themselves, distinct from any single country's circumstance. The default values drift from physical intensity at the point of assignment, and they drift by different amounts and in different directions for each origin. The direction of the error is broadly systematic, since the values are set conservatively and tend to sit above true intensity. The size of the error follows no consistent rule. It runs near a factor of 1.7 for Mozambique, near 14 for Taiwan, and below actual intensity for Morocco, where the assigned value sits under the producer's real footprint. We term this default-setting drift: the gap between an assigned default and the physical intensity it stands in for, opening at the moment of assignment and varying without pattern across origins.

The drift originates in the assignment rule rather than in measurement error after the fact. The default-value regulation sets each value from the most reliable available data, ordered by the dominant production route of the origin country, and falls back to the average of the ten highest-intensity exporters where reliable country data is absent [5]. The rule fixes a number. It tests neither the direction nor the magnitude of the deviation between that number and the physical reality it represents, and it carries no step that checks whether the deviation it produces is consistent from one origin to the next. The country evidence supplies that missing check. The deviations are not consistent: a crisis-year reading locked into permanence for Mozambique, a production route absent from the matrix for Egypt and Algeria, a precursor loaded onto a terminal product for Taiwan, a low-carbon line read as its high-carbon default for Serbia and Bosnia. One rule, applied across heterogeneous technologies and data environments, produces drift that differs in size and sign for each origin it touches.

The distinction carries weight because it determines what kind of correction can work. The economic literature has established that a border adjustment resting on default intensities distorts the abatement incentive that carbon pricing is meant to create, and has proposed letting producers demonstrate an intensity below the default as the remedy [20]. Recent sectoral work on chemicals has found that consistent reporting incentives call for product- and country-specific mark-ups, and has noted the legal and diplomatic exposure that high fallback values carry when they read as a presumption of guilt against producers without verified data [18]. A separate strand has modelled country-average regulation and concluded that the coarser approach performs about as well as plant-level targeting, on the reasoning that under-penalising the high-intensity producers offsets the loss from imprecise targeting [19]. The drift documented here speaks directly to that last conclusion. If the deviation between default and reality were random noise around a correct central value, a country average would indeed suffice, and a better global figure or a uniform mark-up would close the gap. The deviation is not noise. It is a structured product of the assignment rule, different in size and direction for each origin, and no single global value or uniform mark-up reaches a distortion that takes a different shape in every country. What the evidence calls for is a per-origin, evidence-strength-adjusted correction applied within the methodology, which is the instrument Section 6 sets out. The drift is a flaw in how this mechanism assigns its values; Section 5 shows that the recognition boundary which frames it is a property of border carbon instruments as a class.

5. The pattern is a property of the instrument, not of one regulation

The selectivity documented above is reproduced beyond the European mechanism, which matters for how the Commission should read this submission. Default-setting drift, in Section 4.5, is a structural flaw in how one mechanism assigns its values. The pattern named here operates one level up, across mechanisms. It is not a drafting weakness particular to CBAM. It is the equilibrium that border carbon instruments settle into when they are built on a mature domestic carbon price.

The United States Clean Competition Act sets charges on imported high-carbon goods against a domestic industry benchmark and computes credits against domestic carbon payments rather than against clean-production investment, reproducing the cash-outlay logic of Article 9 in statutory form. The United Kingdom's CBAM, announced in 2024 for operation from 2027, follows the European design closely, including the direct-emission boundary and the carbon price deduction. Canadian discussions of a border adjustment have assumed the same architecture. The convergence does not come from drafters copying one another. It comes from three constraints that any such instrument faces at once: national-treatment discipline under the General Agreement on Tariffs and Trade, which limits how imports and domestic goods may be treated differently [10]; coherence with the home emissions trading system, which cannot recognise across the border activities its internal accounting does not register equivalently; and administrative enforceability, which favours cash payments a verifier can check over stock-to-flow conversions a verifier would have to contest case by case. The intersection of those three has a single stable solution in flow-denominated selectivity, and independent designs gravitate to it.

This carries a constructive implication rather than a counsel of despair. A translation component that converts an evidenced implicit carbon cost into a carbon-price equivalent, disciplined by audit and by a conservative treatment of uncertainty, is the move that keeps the regulator's evidence base plural while respecting all three constraints. It does not ask the mechanism to abandon cash-denominated verification. It gives the mechanism a documented, certifiable way to register the carbon cost that flows through other instruments, so the filter stops narrowing the policy landscape it observes. A jurisdiction that builds this component first sets the template the others are likely to follow, given how closely the instruments already track one another.

6. The request: a translation component in the default carbon price

We ask the Commission to build into the default carbon price published under Article 9(4) and 9(5) an audited component that converts recognised implicit carbon costs into a carbon-price equivalent [1][2]. The component operates entirely within the default carbon price methodology of Article 4, Article 6(2), and Annex I of the draft, and it leaves the Article 2(3) definition and the evidence rules for explicit prices as the draft has them [2]. Three properties keep it disciplined, administrable, and defensible.

6.1 Tiered and auditable eligibility

An implicit carbon cost qualifies for translation only on conditions that mirror the rigour the draft already applies to explicit prices. The cost must rest on a public legal instrument. It must apply without discrimination across the operators in its sector, the same non-discrimination test that Recital 8 sets for recognised carbon price mechanisms [2]. And it must be evidenced to the independent person on documentary footing equivalent to what Annex I requires for a tax record or an allowance surrender, so that the certification architecture of Chapter IV carries the new component without modification [2]. Tiering follows the strength of the evidence. An instrument with a published statutory rate and auditable coverage data enters at full translation; one resting on looser documentation enters at a discounted weight; one that cannot be evidenced to the independent person does not enter at all. This keeps the door open to genuine, instrument-backed carbon costs and closed to unverifiable or discretionary claims, which protects the integrity the whole regulation depends on.

6.2 A conservative precision adjustment

The translated value entering the default carbon price should be set so that uncertainty is paid for by the party claiming recognition. We propose the form B = μ + k·σ [17]. Here μ is the central estimate of the implicit cost expressed as a carbon-price equivalent in euro per tonne of CO₂. The term σ is the standard error of that estimate, a statistic that measures how tightly the available evidence pins the figure down: a small σ means the estimate is well supported, a large σ means it is loosely supported. The factor k is a penalty coefficient set above zero. Because the formula adds k·σ rather than subtracting it, a larger uncertainty pushes the recognised figure further from the claimant's interest, so a precise estimate is recognised close to its face value while a loosely supported one is discounted toward conservatism. An auditor follows the same instinct when adding a margin of safety to a number that cannot be confirmed exactly. The coefficient k can be calibrated through the bilateral pilots that Article 2(12) of the base regulation already permits, which lets the Commission tune the conservatism of the instrument against real cases before generalising it [1]. The effect is a published default carbon price that can admit translation without inviting overstatement, because the structure of the formula makes overstatement self-penalising.

A worked illustration shows the structure in operation, using the Mozambique aluminium case from Section 4. The recognition problem there runs through the embedded-emissions default rather than through a paid price, and the same precision logic applies to the intensity the methodology assigns. Measured direct intensity converges on a central estimate of about 1.83 tCO₂e/t, with the four independent reconstructions spread across roughly 1.76 to 1.90, which puts the standard error near 0.06 [5][14]. A conservative penalty factor of k = 2, applied as μ + k·σ, yields a recognised intensity close to 1.95 tCO₂e/t. That figure sits at the cautious edge of the measured range and remains far below the assigned default of 3.198, so it captures the bulk of the gap worth €137.53 per tonne while leaving a deliberate margin in the Commission's favour [5]. A claimant presenting weaker evidence, and therefore a larger σ, would see the recognised figure pushed further toward the default, which is the discipline the formula is built to impose. The same arithmetic governs an implicit-cost claim expressed directly as a carbon-price equivalent.

6.3 The definition of carbon price is preserved

The component changes what the default carbon price can reflect. It does not change what counts as a carbon price under Article 9 of the base regulation. The legislator's instrument list stands untouched, and the explicit-price evidence rules operate exactly as drafted. The implicit-cost translation enters only the Commission's own computation of the default carbon price, the figure the mechanism already publishes and already permits declarants to use in place of certified actual data [1][2]. In this sense the request asks the methodology to do more of what it already does, drawing on a wider but still fully evidenced set of inputs, in the same spirit that Recital 15 already directs the methodology to treat recycled-revenue decarbonisation as carbon cost rather than as compensation [2].

6.4 What the component resolves

A translation component of this form addresses the thirteen cases at the level where they arise. Route misclassification of the kind documented for Algeria, Egypt, and Serbia is corrected within the methodology, rather than left to an independent person who has no matching default to certify against [3][5]. Genuine carbon costs of the kind China, Serbia, and Turkey carry become partially recognisable instead of evaporating, which also slows the policy-narrowing dynamic described in Section 2 [1][5]. The data-withholding incentive visible in Morocco weakens once accurate declaration can capture recognition the bare default denies [5]. The default-setting drift identified in Section 4.5 is what the precision adjustment in 6.2 is built to absorb: because the drift differs in size and direction for each origin, a per-origin term carrying its own standard error fits it where a single global figure cannot, and the penalty factor holds the correction conservative wherever the evidence is thin. The mandatory revision of the default-value regulation foreseen in its own Article 2 is the near-term vehicle for the route corrections, and this conversion methodology is the right place to introduce the translation structure that the route corrections will need [5].

7. Relationship to the Article 9 mandate and the review horizon

The recognition boundary itself, the closed list of instrument forms, belongs to the legislator and to the Article 30 review of Regulation (EU) 2023/956 [1]. We raise the diagnosis of commensuration selectivity with that review in view, because the choice of what to recognise will shape the mechanism's standing with its trading partners over a longer horizon than this implementing act spans.

The comparison that frames the stake is with the Clean Development Mechanism of the Kyoto period. That instrument earned its authority from broad multilateral participation, registered thousands of projects, and channelled a large volume of investment toward mitigation in developing economies. Its decline came from a different direction than its strength. Consensus governance left it unable to correct methodological problems once they surfaced, confidence in the environmental integrity of its credits eroded, and the mechanism slid into dormancy while its supervisory machinery still functioned. The present mechanism presents the mirror image of that profile. Its capacity to adapt is strong, its review architecture is built in, and the strain it accumulates comes instead from the parties who carry its costs without a seat at its design. Both instruments rest on commensuration, the rendering of heterogeneous climate effort into a single comparable metric, and both inherit the selectivity that the act of commensuration carries [15]. Recognising this lineage is what lets the present mechanism avoid the specific failure that overtook its predecessor.

The practical reading is straightforward. A border instrument that prices effectively while recognising narrowly accumulates exactly the kind of legitimacy pressure that an otherwise capable design can avoid, and it accumulates that pressure among the trading partners whose cooperation the mechanism needs to function. Opening recognition channels within the methodology, inside the mandate as it stands, is the constructive response available now, and the window for it is bounded. As free allocation is withdrawn after 2030 and the financial stakes of the mechanism compound, both the political difficulty of reform and the cost of continued misrecognition rise together [1][6].

8. Conclusion

The draft implementing regulation does what its mandate asks, and does it with care. The methodology at its centre, the default carbon price under Article 9(4) and 9(5), can do more than it currently does, and the change sits within the act's own authority. Capital committed to clean production carries a carbon cost economically equivalent to a carbon price, the draft already accepts that substance in Recital 15, and an audited translation component in the default carbon price would let the methodology reflect decarbonisation wherever it occurs and through whatever instrument delivers it. The evidence across thirteen exporting jurisdictions measures the cost of leaving the methodology as drafted, and the reproduction of the same pattern in the British, American, and Canadian instruments shows that the question is structural to border carbon adjustment as a type. We submit this analysis in support of a mechanism that is both effective and durable, and we welcome the opportunity to provide the underlying country-level modelling to the Commission services on request.

References

[1] European Parliament and Council of the European Union. Regulation (EU) 2023/956 of 10 May 2023 establishing a carbon border adjustment mechanism. Official Journal of the European Union, L 130/52, 16 May 2023. (Article 9 on the carbon price effectively paid; Article 9(4) and 9(5) on default carbon prices and implementing powers; Article 2(12) on bilateral agreements; Article 30 on Commission review; Annex II on direct-emission goods and the free-allocation phase-out schedule.)

[2] European Commission. Draft Commission Implementing Regulation laying down rules for the application of Regulation (EU) 2023/956 as regards the conversion of the carbon price paid in a third country into a corresponding reduction in the number of CBAM certificates to be surrendered, the evidence of payment of that carbon price, and the qualifications of the independent person. Ref. Ares(2026)4841230, 13 May 2026. (Article 2(3) and Recital 8 on the definition of a carbon price mechanism; Recital 15 and Article 8(2) on recycled-revenue subsidies; Article 4 on default carbon prices for precursors and indirect emissions; Article 6(2) on conversion into a certificate reduction; Annex I on the attribution methodology; Chapter IV on the independent person.)

[3] European Commission. Commission Implementing Regulation (EU) 2025/2547 of 10 December 2025 laying down rules for the application of Regulation (EU) 2023/956 as regards the methods for the calculation of emissions embedded in goods. Official Journal of the European Union, 22 December 2025. (Production-route system boundaries; precursor attribution formula; sectoral accounting provisions.)

[4] European Commission. Commission Implementing Regulation (EU) 2025/2620 of 16 December 2025 laying down rules for the establishment of benchmarks and the free allocation adjustment for the purposes of the carbon border adjustment mechanism. Official Journal of the European Union, 22 December 2025. (Route-specific benchmarks B_EU; the Column A and Column B free-allocation adjustment values; the electricity free-allocation adjustment.)

[5] European Commission. Commission Implementing Regulation (EU) 2025/2621 of 16 December 2025 laying down rules for the establishment of default values for the calculation of embedded emissions. Official Journal of the European Union, 22 December 2025. (Country-specific and fallback default values by CN code; production-route assignment; the default-value mark-up schedule of 10, 20, and 30 percent for industrial goods and 1 percent for fertilisers; the data-reliability provisions of Recitals 1 and 2; the mandatory revision foreseen in Article 2.)

[6] European Parliament and Council of the European Union. Regulation (EU) 2025/2083 of 8 October 2025 amending Regulation (EU) 2023/956 as regards simplifying and strengthening the carbon border adjustment mechanism. Official Journal of the European Union, 20 October 2025.

[7] European Commission. Report from the Commission on the functioning of the Carbon Border Adjustment Mechanism. COM(2025) 783 final, 2025. (Transitional-period reliance on default emission values.)

[8] European Commission. Proposal for a Regulation of the European Parliament and of the Council establishing a carbon border adjustment mechanism, Explanatory Memorandum. COM(2021) 564 final, 14 July 2021.

[9] Council of the European Union. Council Decision (EU) 2017/2240 of 10 November 2017 on the signing of the Agreement between the European Union and the Swiss Confederation on the linking of their greenhouse gas emissions trading systems. Official Journal of the European Union, L 322, 7 December 2017.

[10] General Agreement on Tariffs and Trade 1994, Articles I (Most-Favoured-Nation Treatment), III (National Treatment on Internal Taxation and Regulation), and XX (General Exceptions). World Trade Organization.

[11] Espeland, W. N., and Stevens, M. L. "Commensuration as a Social Process." Annual Review of Sociology, vol. 24, 1998, pp. 313–343.

[12] MacKenzie, D. "Making things the same: Gases, emission rights and the politics of carbon markets." Accounting, Organizations and Society, vol. 34, no. 3–4, 2009, pp. 440–455.

[13] Organisation for Economic Co-operation and Development. Effective Carbon Rates 2024: Pricing Greenhouse Gas Emissions through Taxes and Emissions Trading. OECD Publishing, Paris, 2024.

[14] South32 Limited. Climate Change Action Plan 2025. (Reported combined Scope 1 and Scope 2 intensity for the Mozal aluminium smelter under drought-driven backup power supply.)

[15] Liu, A. Y. A First-Principles Framework for Climate Governance: Core Functions from Cybernetic Theory. Terawatt Times Institute, 2026 (ISSN 3070-0108).

[16] Liu, A. Y. The Clean Production Paradox: Capital Investment, Accounting Boundaries, and the Architecture of Recognition. Terawatt Times Institute, 2026 (ISSN 3070-0108).

[17] Liu, A. Y. The Untranslated Billions: Implicit Instruments, Translation Protocols, and the Path Beyond Commensuration. Terawatt Times Institute, 2026 (ISSN 3070-0108).

[18] Embodied emissions of chemicals within the EU Carbon Border Adjustment Mechanism. Nature Sustainability, 2025. DOI 10.1038/s41893-025-01618-5. (Finding that product- and country-specific mark-ups are needed for consistent reporting incentives, and that high fallback default values raise "guilty until proven innocent" concerns.)

[19] Clausing, K. A., and co-authors. The Global Effects of Carbon Border Adjustment Mechanisms. Working paper, 2023. (Plant-level emissions modelling; finding that country-average regulation performs comparably to plant-specific regulation.)

[20] Mehling, M. A., and Ritz, R. A. Going Beyond Default Intensities in an EU Carbon Border Adjustment Mechanism. MIT CEEPR Working Paper 2020-019, October 2020; published as Oxford Review of Economic Policy 39(1), 2023, pp. 123–133, DOI 10.1093/oxrep/grac043.