The State of CBAM 2026

1.1 Definitive regime and the regulated actor — adopted law

The definitive CBAM regime began on 1 January 2026.C-001 The annual obligations described in the supporting amendment attach to the authorised CBAM declarant: by 30 September each year, and for the first time in 2027 for calendar-year 2026, that declarant must submit a CBAM declaration for the preceding year and surrender the corresponding number of certificates through the Registry.C-001 The first declaration and surrender therefore cover 2026 imports but fall due on 30 September 2027.C-001

The date sequence matters because commencement, price assignment, certificate sale and annual settlement are not the same legal event. The definitive regime starts in 2026, the price for 2026 embedded emissions is associated with the relevant quarter of importation, sales begin on 1 February 2027, and declaration and surrender occur by 30 September 2027.C-001 C-002 A 2026 price publication therefore establishes a value used in the 2026 import-quarter framework; it does not establish that certificates were available for purchase at that time.C-002 C-004

1.2 The 50-tonne exemption — adopted law

The general mass-based exemption is 50 tonnes of cumulative net mass per importer and calendar year.C-003 The threshold applies to the total net mass of goods under all relevant CN codes aggregated across cement, fertilisers, iron and steel, and aluminium.C-003 “Across four sectors” is a useful summary, but the legal mechanism operates through the aggregation of the covered CN codes rather than through four independent sector-level allowances.C-003

Electricity and hydrogen are expressly excluded from this mass-based exemption.C-003 If an importer exceeds the single threshold during the year, the importer or authorised CBAM declarant becomes subject to the Regulation’s obligations for the emissions embedded in all covered goods imported during that calendar year.C-003 Threshold crossing is thus retrospective across the calendar year covered by the test, rather than only prospective from the shipment that takes cumulative mass above 50 tonnes.C-003

The adopted text also provides for annual reassessment and a delegated-act mechanism if the calculated threshold changes by more than 15 tonnes.C-003 That feature means 50 tonnes is the supported baseline at this research cut-off, while future annual legal updates still require monitoring.C-003

1.3 Certificate price and sale timing — adopted law plus official operations

Member States begin selling CBAM certificates on the common central platform from 1 February 2027.C-002 For embedded emissions declared for 2026, the supporting amendment provides for a certificate price calculated from the quarterly average of EU ETS closing prices for the quarter of importation.C-002 The legal price-assignment period and the later purchase window must therefore be kept distinct.C-002

The Commission published a first-quarter 2026 price of €75.36 and a second-quarter price of €75.28.C-004 Each is a quarterly value per certificate, with one certificate corresponding to one tonne of CO2 equivalent.C-004 These two observed official price publications do not by themselves determine an importer’s certificate quantity, and they do not demonstrate any 2026 certificate sale.C-004

1.4 Free-allocation adjustment — adopted schedule

Under the adopted EU ETS amendment, the CBAM factor is 97.5% in 2026, 95% in 2027, 90% in 2028, 77.5% in 2029, 51.5% in 2030, 39% in 2031, 26.5% in 2032 and 14% in 2033.C-005 From 2034, no CBAM factor applies under that adopted schedule.C-005

The factor describes the legal adjustment associated with the phase-out of free allocation; it is not a blanket percentage that can be multiplied mechanically by every importer’s actual emissions.C-005 Product benchmarks, production routes, relevant precursors and the applicable implementing methodology remain material to the calculation context.C-005 The safe annual-report formulation is therefore a schedule of the adopted factor, not a simplified claim that importers “pay 2.5% of emissions in 2026” without the wider methodology.C-005

1.5 Proposed 2038 path — Commission proposal, not law

COM(2026) 616 proposes replacing part of the adopted schedule with a slower reduction.C-006 Its proposed factor is 97.5% in 2026, 95% in 2027, 91.5% in 2028, 81% in 2029, 59% in 2030, 48% in 2031, 37.5% in 2032, 27% in 2033 and 15% from 2034 through 2037, followed by 0% in 2038.C-006

At the cut-off, the European Parliament procedure record placed the file in the preparatory phase in Parliament after publication of the legislative proposal on 17 July 2026.C-006 The proposal had not displaced the adopted schedule.C-006 Any scenario using a residual 15% factor from 2034 to 2037 must consequently be labelled “proposed”, while the current-law baseline continues to show no factor from 2034.C-005 C-006

Table source: adopted schedule C-005 ; proposed schedule and procedural status C-006 .

1.6 Downstream expansion — legislative procedure, not adopted scope

COM(2025) 989 proposed changes concerning downstream goods and anti-circumvention measures.C-049 The Council reached a general approach at the Economic and Financial Affairs Council meeting on 12 June 2026.C-049 Parliament’s ENVI committee then adopted its committee position on 6 July by 56 votes to 11, with 12 abstentions, and the committee report was tabled on 9 July.C-049

Those milestones did not amount to a final adopted regulation. The Parliament procedure record still said “Awaiting Parliament’s position in 1st reading”, with the plenary event indicated for September after the research cut-off.C-049 The cut-off-safe description is therefore “Council general approach plus ENVI committee position”, not “Council and Parliament positions adopted”.C-049 Proposed downstream CN-code changes should not be represented as part of the adopted 2026 product scope.C-049

1.7 Foreign carbon prices — draft implementing detail

By 6 August 2026, the Commission had published detailed draft rules for feedback on recognising carbon prices effectively paid in third countries.C-015 The consultation addressed proof of payment, currency conversion and the eligibility of third-party certifiers.C-015

The Commission initiative tracker distinguished the published draft from the final-act workflow. When checked on 10 August, the draft record showed an internal consultation workflow and no adoption date, while the final implementing-act record was marked upcoming in the adoption workflow and also had no adoption date.C-015 The defensible statement at the cut-off is that detailed draft requirements existed but the final implementing act had not been adopted.C-015 Because this is a volatile status claim, it requires a fresh official check before any publication decision.C-015

The supported operating baseline is deliberately narrower than a complete implementation manual. It establishes the timing of the regime, the actor named in the annual obligation, the annual threshold, the official quarterly prices and a bounded snapshot of authorisation and customs-system activity.C-001 C-003 C-004 C-013 It does not establish a complete customs-to-installation workflow, prove emissions-data completeness or support a general claim about verifier availability.

2.1 A year of operation before the first settlement

The regime was in force from 1 January 2026, but the first annual declaration and surrender for 2026 do not fall due until 30 September 2027.C-001 Certificate sales start on 1 February 2027.C-002 The Q1 and Q2 2026 prices of €75.36 and €75.28 are therefore inputs to the 2026 import-quarter pricing framework rather than cash purchase records from those quarters.C-002 C-004

This sequence defines the minimum supported calendar:

Table sources: commencement and annual deadline C-001 ; sale date C-002 ; quarterly values C-004 .

2.2 Early authorisation and customs processing

The Commission’s launch report said that more than 12,000 economic operators had submitted applications for CBAM authorisation by 7 January 2026.C-013 It also reported that more than 4,100 economic operators had obtained authorised CBAM declarant status before and immediately after 1 January.C-013 The publication does not provide one common exact timestamp for both figures, so they should not be presented as a same-moment conversion funnel.C-013

From 1 through 7 January, 10,483 import customs declarations containing CBAM goods were validated automatically and in real time through integrated customs systems.C-013 That figure is evidence of deployment and system processing in the first week.C-013 It is not evidence that every declaration’s embedded-emissions data, verification evidence or later certificate position was complete or correct.C-013

The early figures consequently answer three narrow questions: applications had been submitted at scale, thousands of authorised declarants existed around launch, and integrated systems were automatically validating CBAM customs declarations.C-013 They do not answer how many applicants were ultimately authorised, how many importers crossed the annual threshold, how many records later required correction, or how ready operators were for the 2027 surrender cycle.

3.1 Dataset boundary

The Commission’s application report uses Transitional Registry data covering Q4 2023 through Q2 2025, with a dataset extraction cut-off of 31 August 2025.C-007 The evidence is declarant-reported registry information that the Commission cleaned for its analysis, not an unrestricted statement of all customs-trade activity.C-007 Electricity is handled separately in MWh, so it does not belong in the physical-tonnage split for the other sectors.C-007 C-008

Two headline totals describe different measures and periods. Around 156 million tonnes refers to reported physical mass of CBAM goods from Q4 2023 through Q2 2025.C-007 The estimate of 167 million tCO2e refers to embedded emissions accumulated during 2024.C-007 The mass total cannot be subtracted from, divided into or otherwise combined directly with the emissions estimate because one is tonnes of goods over seven quarters and the other is tonnes of CO2 equivalent for one calendar year.C-007

3.2 Scale and sector concentration by weight

The Commission reported imports of around 156 million tonnes of CBAM goods between Q4 2023 and Q2 2025.C-007 By weight, iron and steel represented 69%, fertilisers 15%, cement 11% and aluminium 5%.C-008 The four shares sum to the non-electricity weight split reported in the application report and Annex IV.C-008

Table source: C-008 . Electricity is excluded from this weight table and reported in MWh.C-008

The 69% iron-and-steel share establishes concentration by physical weight.C-008 It does not establish that iron and steel represented 69% of embedded emissions, certificate demand, import value, declaration count or importer count.C-008 Likewise, the smaller aluminium share by weight does not demonstrate a proportionately smaller financial or emissions exposure, because the source table is not measuring either of those quantities.C-008

3.3 Estimated embedded emissions for 2024

The Commission estimated 167 million tCO2e of embedded emissions accumulated in 2024 from the Transitional Registry.C-007 This is an official estimate derived from transitional reporting rather than a final verified definitive-regime total.C-007 Its period is calendar-year 2024, while the reported-mass figure spans Q4 2023 through Q2 2025.C-007

The emissions estimate shows that the Registry held information corresponding to about 167 million tCO2e for calendar-year 2024 during the transition.C-007 It does not show the certificates ultimately surrendered for 2024, because 2024 belonged to the transitional period and the first definitive-regime annual surrender covers 2026 imports in 2027.C-001 C-007 It also does not by itself show the effect of free-allocation adjustment, foreign-carbon-price deductions or later corrections.

3.4 Ukraine’s reported concentration

Ukraine was the largest reported country by overall mass in the Commission dataset, accounting for roughly 14% from Q4 2023 through Q2 2025.C-012 The report gives a rounded total of 21.2 million tonnes of CBAM goods, including 18.7 million tonnes of iron and steel and 2.4 million tonnes of cement.C-012 The rounded sector components need not sum exactly to the rounded overall total.C-012

The Commission also described Türkiye as responsible for roughly 12% of overall mass, behind Ukraine’s roughly 14%.C-012 Claims C-043 and C-044 support a bounded Türkiye registry-data assessment of mass, sector mix and actual-value entry shares.C-043C-044 They do not support national readiness, methodological equivalence, verifier capacity or a quantified financial outcome.C-045

The source framing requires care. The report’s figure is labelled “Country of Production”, while surrounding prose uses exporter language.C-012 That wording should not be silently converted into a definitive customs-origin classification.C-012 The 21.2 million tonnes is also physical mass, not import value, embedded emissions or the cost of future certificates.C-012

3.5 What this evidence cannot establish

The official dataset supports claims about reported mass, sector shares by weight, an estimated 2024 emissions total and selected country concentration.C-007 C-008 C-012 It does not on its own establish realised certificate liabilities, verified installation-level intensities, importer profitability, consumer-price effects, GDP effects or carbon leakage. Questions about value-added propagation and leakage require an analytical model with an explicit baseline and scenario, while claims about country vulnerability require a defined exposure metric.C-009 C-010 C-011

The Commission’s Annex IV is not a raw dump. It covers Q4 2023 through Q2 2025, uses an extraction cut-off of 31 August 2025 and was cleaned because reporting across 27 Member States created significant challenges.C-038 Reported tonnes were cross-checked with customs imports so unrealistic high-end outliers could be removed, but customs granularity and procedure information were not uniform in every case, and potential underreporting was not adjusted.C-038 The correct label is therefore “cleaned Transitional Registry data with residual caveats”, not either “raw data” or “fully reconciled customs totals”.

The error evidence is concrete. Different conventions for decimal and thousands separators produced numeric-input problems; an earlier choice between kilograms and tonnes produced unit errors; broader-than-required or incorrect CN codes entered the Registry; and variants such as TR and TC represented the same country inconsistently.C-039 The Commission linked reductions to removing redundant fields, standardising formats, adding warning labels, imposing rigid validation and centralising operator and installation information.C-039 This supports a design principle: control units, code lists, allowed formats and identities as early as possible, then preserve the correction trail.

The narrative says incorrect CN-code entries fell sharply by Q2 2024 and stabilised at residual levels, while incorrect country codes affected about 0.5% of declarations in early periods and declined.C-039 The chart does not print exact quarter-by-quarter values, so the report should not digitise apparently precise percentages from the line. “Virtually disappeared” is the Commission’s qualitative description, not proof of zero error.

The Operators Portal supplies another bounded implementation example. The Commission processed 573 access requests between December 2024 and August 2025, reported high early rejection volumes, and said clearer guidance reduced eligibility errors and duplicate applications.C-048 It did not publish an exact rejection count or rate.C-048 Portal-access rejection is a different population and event from an emissions-reporting error, a failed annual declaration or a verification problem.

Actual-value reporting changed materially during the transition. Overall declared use rose from 8% in Q4 2023 to 53% in Q2 2025, and entries above 1,000 tonnes rose from 25% to 93%.C-042 The report’s terminology moves among importers, imports, entries and declarations; the safest denominator for chart percentages is “import entries”.C-042 The residual “Other” category includes non-actual cases such as “actual values not available”, so it should not be described simply as default values.C-042

Most importantly, transitional emissions were not verified. The report estimated them by multiplying reported tonnes by transitional default values, with one global average default per CN code for non-electricity goods and no country or production-method nuance.C-041 That methodology supports an aggregate estimate within its stated boundaries. It does not rank plant performance, validate a supplier’s actual value or provide a definitive-regime verified inventory.

The mass distribution argues against an equal-weight tour of every sector. Iron and steel represented 69% of reported tonnes, a much larger physical share than fertilisers at 15%, cement at 11% or aluminium at 5%.C-008 That is a prioritisation signal for evidence operations, not a financial ranking. A tonne of one product cannot be assumed to create the same embedded emissions, benchmark adjustment, carbon-price deduction or certificate exposure as a tonne of another.

Ukraine led the country concentration in the Commission data at roughly 14% or 21.2 million tonnes in Q4 2023–Q2 2025, including 18.7 million tonnes of iron and steel and 2.4 million tonnes of cement.C-012 Its significance is therefore simultaneously sectoral and geographic. The World Bank’s separate 0.5% Economic Exposure Index for Ukraine adds an analytical lens, but it uses older data, a USD 100 carbon price and a GDP denominator rather than the Registry’s physical-mass measure.C-011 The two figures should sit beside each other only with those differences stated.

Electricity and hydrogen need a unit guardrail rather than token sector chapters. Electricity is reported in MWh and excluded from the four-sector weight split, while both electricity and hydrogen are excluded from the 50-tonne mass exemption.C-003 C-008 Their exclusion from one table or threshold does not mean exclusion from CBAM itself. It means the mass comparison is not methodologically valid for those goods.

Bounded Türkiye evidence box

The official chain starts with a defined customs spine. Customs authorities communicate importer identification, the CBAM account number where applicable, CN code, quantity, country of origin, declaration date and customs procedure to the Registry.C-016 Those fields identify the import event and regulated account context. They do not identify the producing installation, production route, precursor chain, actual emissions, verifier or final certificate quantity.C-016

Customs data then participates in review rather than standing alone. Authorities cross-check customs information against each declarant’s Registry data, and declarations and verification reports may be reviewed against customs information and other evidence.C-017 This creates a reconciliation requirement: quantities, codes, origins, dates and actors should remain linkable to the later emissions and certificate record. Cross-checking does not mean every field reconciles automatically, and accepted customs processing is not a conclusion about the quality of installation evidence.C-017

The Registry provides records and exchange pathways among declarations, importers, operators, emissions reports, verifiers, verification reports and supporting documents.C-018 The Operators Portal facilitates exchanges between operator, verifier, independent person and authorised declarant.C-018 That is an official exchange architecture, but it is not a prescribed enterprise data model or proof that every organisation’s ERP, procurement and sustainability systems are integrated.

The annual declaration is the declarant-level aggregation point. It contains total quantity by goods type, total embedded emissions, the number of certificates to surrender and, where applicable, copies of accredited-verifier reports.C-019 The difference between customs inputs and declaration outputs explains the need for lineage: a business must be able to connect imported quantities to the installation and method evidence that produces the embedded-emissions result, then connect that result to adjustments and certificates. That description is CBAM Pulse analysis of the supported handoffs, not a statement that regulators require a particular software schema.

Registered third-country operators provide a potential installation-side anchor. Registration records operator identity and the location of each installation, including full address and geographical coordinates; an operator may disclose verified embedded-emissions and third-country carbon-price information to an authorised declarant.C-020 Registration and disclosure are voluntary in the terms described, and the declarant’s legal responsibility is not transferred away.C-020

A useful preparation record therefore has at least six linked layers:

1. Customs event: importer/account, CN code, quantity, origin, date and procedure, tied to the source declaration.C-016
2. Operator and installation: the legal entity and physical installation associated with production, with identifiers and evidence references.C-020
3. Method state: actual value, adopted default or prescribed mixed treatment for each relevant production and precursor component.C-021 C-024
4. Verification: operator report, summary, verifier identity, scope and opinion where actual emissions are used.C-022 C-026
5. Adjustments and declaration: the aggregation of quantity, embedded emissions, certificate number, verification reports and any supportable third-country carbon-price reduction.C-019 C-036
6. Certificate and retention events: purchase, holding, surrender, repurchase/cancellation and actor-specific record clocks.C-032 C-034 C-035

This model is intended to expose missing joins and unclear ownership. It does not file, calculate, verify or certify anything, and it cannot establish compliance.

For non-electricity goods, adopted law permits embedded emissions to be determined using actual emissions or applicable default values.C-021 Actual emissions trigger verification by an accredited verifier; defaults are an adopted method with prescribed values and conditions, not a supplier estimate invented to fill a spreadsheet cell.C-021 Electricity and indirect-emissions rules differ, so this distinction should not be generalised without regard to goods and method.

The actual-value handoff starts with the operator. The operator prepares a detailed emissions report and a summary, transmits the package to the verifier—through the Registry when registered—and the verification package carries the operator summary into the declarant’s evidence chain.C-022 Commercially sensitive detail may remain in the full report while the declarant receives a summary.C-022 A request email or supplier spreadsheet is not equivalent to this complete chain.

Complex goods deepen the lineage requirement. For externally sourced precursors, the adopted calculation method can require country of origin, source installation, production route or relevant parameters, reporting period, quantity and specific embedded emissions.C-023 When periods or installations are mixed, prescribed weighted-average rules may apply.C-023 Country alone does not replace installation identity when an actual value is claimed.

Incomplete or inconclusive precursor evidence does not authorise an invented supplier-specific value. The adopted rules direct the relevant quantity to applicable default values and allow actual emissions for a production process to coexist with defaults for one or more precursors.C-024 The operating consequence is to preserve method state component by component. A record should show where actual evidence stops, where a default begins, which default applies and why.

Verification reports preserve that state. They identify precursors for which defaults were used and, for actual-value precursors, carry information about origin, period, installation and the verifier of those actual values.C-025 This is more than a final emissions total: it is a provenance structure that can be reviewed and reconciled.

The assurance standard is bounded. Actual emissions require a CBAM-accredited verifier applying a risk-based reasonable-assurance standard and issuing the prescribed report.C-026 Reasonable assurance is not absolute assurance, a warranty of the declarant’s entire filing or a product compliance guarantee.C-026 The materiality provisions should not be turned into a blanket permission for an error percentage.

Physical site visits are the baseline in the adopted verification discipline, with virtual visits or waivers available only under specified risk-based conditions and a separate route for extraordinary circumstances.C-027 This is not a universal claim that every installation must receive an annual physical visit; nor does the existence of flexibility quantify cost, travel demand or market capacity.C-027

Accreditation is scope-specific by activity group, open to qualifying legal persons outside the EU through a Member State national accreditation body, and subject to surveillance, reassessment and possible scope reduction, suspension or withdrawal.C-028 Accreditation and Registry registration are distinct steps; an accredited verifier requests registration, after which its scope, dates and certificate information are held in the Registry.C-029 These facts establish governance architecture. They do not quantify provider supply, throughput or queues.

Certificate sales begin on 1 February 2027 through the common central platform, and purchases are recorded in the authorised declarant’s Registry account with identifier, price and sale date.C-031 For 2026 embedded emissions, the price basis is associated with the quarter of importation, which is why the Commission could publish 2026 prices before sales opened.C-002 C-004 A treasury view should therefore distinguish estimated certificate quantity, assigned price, purchase date and cash event.

From 2027, holding, surrender, repurchase and cancellation are separate legal events. The account is subject to a quarterly minimum holding rule; annual surrender removes the required certificates; eligible excess may be repurchased on request; and older residual certificates can be cancelled without compensation.C-032 The holding rule is not simply 50% of final verified liability, because its calculation uses statutory alternatives, adjustments and timing rules.C-032 Certificates are not indefinitely bankable.

The first cycle is unusually compressed. Certificates bought in 2027 for 2026 embedded emissions may be repurchased only in 2027, and qualifying remainders are cancelled without compensation later in 2027.C-033 That treatment should not be generalised to later vintages, and repurchase remains request-driven and conditional.C-033

A defensible treasury cycle therefore uses dated states rather than one liability number: import-quarter exposure, current method state, expected adjustment, price basis, expected purchase window, minimum holding position, surrender requirement and residual-certificate treatment. This is a planning framework, not financial advice or a guarantee of final liability. It should carry ranges and unresolved evidence flags rather than a false point estimate where supplier, verifier or foreign-price information remains open.

Record retention also requires more than a generic four-year label. Declarants keep calculation records and verifier reports through the end of the fourth year after the year in which the declaration was or should have been submitted.C-034 Carbon-price evidence follows a declaration-based fourth-year formulation, while registered operators retain verification and calculation records for four years after verification, and operator carbon-price documents have their own trigger.C-035 These are minimum CBAM windows for specified records, not universal deletion dates.

Foreign carbon-price evidence

Adopted Article 9 allows a certificate reduction only for a carbon price effectively paid in a third country, net of rebates or other compensation, with records of the actual payment and independent certification.C-036 A nominal carbon-price rate is not automatically the amount deductible.C-036 The evidence chain should preserve the jurisdiction and instrument, covered emissions, payment event, payer, currency, period, rebates or compensation, independent certification and the connection to the goods and installation.

At the cut-off, detailed implementing rules on proof of payment, currency conversion and third-party certifier eligibility were still draft/upcoming rather than finally adopted.C-037 The publication-safe stance is to describe Article 9’s adopted core while marking detailed processing fields and eligibility conclusions as update-sensitive. No workflow should hard-code the consultation draft as final law.

4.1 Model contract

The OECD study uses an enhanced static Leontief input-output price model with substitution, elasticities, quantity changes and supply-chain propagation.C-009 Its baseline represents the 2019 global economy, an EU ETS price of around €25 per tonne, no CBAM and the 2019 free-allocation endowment.C-009 The Full Policy Mix applies an ETS price of €80 per tonne, removes relevant free allowances and introduces CBAM.C-009

This is a fully implemented-policy experiment applied to a 2019 economic structure, not a dated forecast of the phased economy in 2030 or 2034.C-009 It captures immediate, first-order short-run effects and excludes longer-run adaptation, backfilling and new foreign climate policies.C-009 The aggregate value-added result cited below is before revenue recycling.C-009

4.2 Value-added propagation beyond covered sectors

In the OECD model, sectors outside direct CBAM coverage account for 83% of the total European value-added loss associated with removing free allowances and introducing CBAM.C-009 The result reflects supply-chain propagation into the much larger group of non-covered sectors rather than limiting the analysis to directly covered industries.C-009

Under the Full Policy Mix, aggregate EU value added is 0.29% below the 2019 baseline.C-009 The OECD also reports that the aggregate macro effect across all EU industries remains −0.29% with or without CBAM when relevant free allowances have been removed in the scenario comparison.C-009 That does not make CBAM economically irrelevant: the model can redistribute sectoral and geographic effects even where the EU-wide aggregate remains the same.C-009

The 83% and −0.29% figures should be kept adjacent to their own denominators. The 83% is a sectoral decomposition of European value-added loss associated with the policy change, whereas −0.29% is an aggregate EU result relative to the 2019 baseline under the Full Policy Mix.C-009 Combining them into “83% of the −0.29% is caused by CBAM” would erase the study’s comparison structure.C-009

4.3 Leakage and global emissions

At the all-industries level, the OECD model estimates a 7.4% leakage rate in its initial higher-carbon-price scenario, rising to 19.2% when relevant free allowances are removed without CBAM.C-010 The model associates the latter scenario with a 175-million-tonne reduction in EU emissions that is partially offset by a 34-million-tonne increase in partner-country emissions.C-010 The 19.2% figure is the all-industries result; the narrower result for CBAM industries is 18.8%.C-010

Adding CBAM to the higher-ETS-price, free-allocation-removal policy mix increases global emissions reductions by 36% relative to that same scenario without CBAM.C-010 In the model, the all-industries leakage rate moves to −12.4% with CBAM.C-010 The 36% uplift is therefore a comparison between two policy scenarios that both contain the higher ETS price and removal of relevant free allowances; it is not a comparison with a world containing no climate policy.C-010

None of those figures is an observed 2026 measurement.C-010 They are results generated by a static short-run model using a 2019 economy and fully implemented policy settings.C-009 C-010 They should not be described as a measured leakage rate since CBAM entered the definitive regime, nor as a forecast that fixes the outcome for a specific future year.C-010

4.4 OECD “do not infer” box

5.1 Metric definition

The World Bank’s Economic Exposure Index measures aggregate estimated excess carbon payments across covered products as a share of a country’s GDP.C-011 “Excess” refers to estimated carbon payments relative to average EU producers under the note’s methodology.C-011 The index is not a computable general equilibrium model and is not a forecast of GDP contraction.C-011

The World Bank reports an Economic Exposure Index of about 0.6% of GDP for Mozambique, 0.5% for Ukraine and 0.2% for Egypt.C-011 These values indicate the scale of estimated excess carbon payments relative to the denominator of national GDP under the index assumptions.C-011 They do not mean that CBAM will reduce Mozambique’s GDP by 0.6%, Ukraine’s by 0.5% or Egypt’s by 0.2%.C-011

Table source: C-011 . These are exposure-index values, not projected GDP losses.C-011

5.2 Inputs and assumptions

The index uses 2017 GTAP11-CE emissions intensities, 2022 WITS mirror trade data and 2022 GDP.C-011 It applies a notional carbon price of USD 100 per tCO2, and the index values scale with that price assumption.C-011 Electricity and hydrogen are excluded from the analysis.C-011

The framework uses country-sector average emissions intensities rather than the actual emissions of each exporting firm.C-011 This may overstate exposure where export-oriented producers are cleaner than their country-sector average.C-011 Eligible domestic carbon-price credits are not deducted.C-011

The index does not model adaptation, cost pass-through, changes in output or welfare effects.C-011 It therefore ranks and compares a defined form of exposure under static assumptions; it does not simulate how producers, importers, consumers, governments or exchange rates respond.C-011

5.3 Relationship to the official Ukraine mass figure

The Commission’s Transitional Registry data and the World Bank index place Ukraine near the centre of two different evidence frames. Ukraine was the largest reported country by physical mass in the Commission data, at roughly 14% and 21.2 million tonnes through Q2 2025.C-012 The World Bank’s Economic Exposure Index for Ukraine is approximately 0.5% of GDP under its static assumptions.C-011

Those figures should not be merged into a single exposure calculation. The 21.2 million tonnes is a cleaned registry-derived mass total over Q4 2023 through Q2 2025, while 0.5% is estimated excess carbon payments divided by GDP using older emissions-intensity inputs, 2022 trade/GDP data and a USD 100 carbon price.C-011 C-012 Their common value is triangulation: one shows observed reported concentration by mass, and the other shows analytical exposure under a defined index.C-011 C-012

5.4 World Bank “do not infer” box

The purpose of an evidence-led annual report is not to repeat the Regulation in narrative form. It is to show which evidence changes a real decision, which uncertainty needs an owner and which attractive conclusion remains unavailable.

For importer and customs teams

Maintain cumulative annual mass by importer and the relevant covered CN codes, because threshold crossing has whole-year consequences under the adopted rule.C-003 Preserve the customs source record and later correction history for identification, code, quantity, origin, date and procedure fields, because those fields enter the official Registry chain and can be cross-checked against declarant data.C-016 C-017 This is preparation guidance, not a customs classification opinion.

For procurement and supplier teams

Ask for installation identity and method evidence early enough to distinguish actual values, defaults and mixed precursor treatment. Actual-value use depends on an operator report and verification chain, while complex goods can require precursor origin, installation, period, quantity and emissions lineage.C-022 C-023 Where evidence is incomplete, record the applicable default-method state rather than inserting an unsupported supplier estimate.C-024

For sustainability and verification owners

Treat “actual” as the beginning of an evidence question, not its conclusion. The definitive-regime chain requires an accredited verifier and reasonable assurance, and verification reports preserve method and precursor lineage.C-025 C-026 Separate verifier identity, accredited scope, Registry status, reporting period, installation access and outstanding evidence. Do not promise an opinion or infer service availability from architecture alone.

For finance and treasury teams

Maintain scenario ranges that separate certificate quantity from certificate price and both from purchase timing. Published 2026 quarterly prices are not purchases, while the first-cycle sale, holding, surrender, repurchase and cancellation events occur under distinct rules in 2027.C-004 C-031 C-033 Preserve current-law and proposal scenarios separately, and treat third-country carbon-price reductions as evidence-dependent rather than automatic.C-005 C-006 C-036

For leadership

Use the official concentration evidence to allocate attention without converting mass into cost. Iron and steel dominate reported tonnes, and Ukraine and Türkiye are material country cases, but each financial conclusion still depends on product, installation, emissions method, adjustments and price timing.C-008 C-012 C-043 Use the OECD and World Bank studies as scenario and exposure lenses, not as forecasts of the company’s outcome.C-009 C-011

A bounded readiness test

A preparation workspace can ask whether each in-scope import population has: a named owner; a traceable customs source; cumulative threshold visibility; operator and installation identity; explicit method state; precursor lineage where relevant; verifier evidence for actual values; documented foreign-price evidence where claimed; a dated certificate scenario; and actor-specific retention references. A gap in one field is a preparation signal, not proof of non-compliance. A completed checklist is likewise not proof of compliance.