CBAM benchmarks: production routes and precursor calculation trace

Choose a calculation path before looking up a number

Implementing Regulation 2025/2620 Article 2 sets the actual-data adjustment path and Article 3 sets the default path when default specific embedded emissions are used in the declaration. Annex point 5 distinguishes Column A process-related benchmarks from Column B default benchmarks. An actual-data pathway still uses regulatory benchmarks; actual does not mean replacing the benchmark with the plant’s emissions intensity.

The default-values method decision belongs to the dedicated method guide. This page follows the benchmark calculation record after that decision. Record goods identity, code, reporting period, actual/default basis, applicable column and evidence reference before copying any table value. A code is a lookup input, not evidence that a shipment has been classified correctly.

The existing phase-in guide explains the annual CBAM factor. Here the question is different: which process and precursor terms the factor belongs with, and whether those terms already include it. A reviewer needs that distinction before accepting a spreadsheet labelled benchmark adjustment.

Keep process, precursor and final-good quantities separate

Under Regulation 2025/2620 Annex points 2–3, the explanatory formula is FAA = imported mass × SEFA. Process-level specific free allocation is CBAM factor × cross-sectoral correction factor (CSCF) × Column A benchmark. For a complex good, add the specific mass of each relevant precursor multiplied by its SEFA, repeating the precursor calculation as required by Annex point 3.3.

Do not add precursor contributions to a whole-chain Column B value as if it were only the final process. Do not apply the annual factors twice to precursor SEFA already adjusted for its applicable period. The precursor boundary and evidence controls explain which inputs belong in the calculation; this trace shows where the resulting free-allocation inputs enter.

Regulation 2025/2547 Article 15 permits actual final-process emissions with default values for one or more precursors. Record that mixed basis component by component. It is neither a blanket prohibition on mixed inputs nor permission to ignore the verification dependencies of an actual-emissions route.

Reproduce the Commission’s steel-fastener illustration

Commission Guidance No. 5d §4.3 describes an example installation making screws and nuts from purchased carbon and high-alloy steel. Table 4-14 uses CN 73181290 for the carbon-steel example and a process benchmark of 0.038 t CO2/t. The legal table in Regulation 2025/2620 Annex point 5.3 lists the same Column A value for CN 7318 12 90. This is a study of the published example, not classification advice.

The guidance’s precursor value comes from its earlier example installation, not from a universal carbon-steel emissions factor. Table 4-14 follows the simplified presentation in the preceding examples: these displayed benchmark-equivalent values still need annual-factor treatment. Page 44 explicitly illustrates that presentation with an assumed 2027 factor of 0.95 and CSCF of 1.00. Keep that pedagogical convention distinct from an already year-adjusted operational precursor SEFA.

Check the arithmetic and make rounding visible

Using the displayed inputs, the precursor contribution is 1.176 × 1.242 = 1.460592 t CO2/t product. Adding the final-process benchmark gives 0.038 + 1.460592 = 1.498592 t CO2/t product, which rounds to the guidance’s 1.499. This is the reproducible displayed-input calculation, not an assertion of unrounded underlying operator data.

The source quantities give 20,000 ÷ 17,000 = 1.176470588… t precursor/t product, rather than exactly 1.176. Replacing only the displayed consumption with that ratio gives 1.499176470… before annual factors. Both round to 1.499, but intermediate results differ. A real calculation needs a documented precision policy and consistent source precision, not a mixture of rounded and unrounded inputs chosen to obtain a desired result.

For an explicitly illustrative extension, use the guidance’s 2027 presentation assumptions of CBAM factor 0.95 and CSCF 1.00 throughout this simplified chain. The displayed-input result becomes 0.95 × 1.00 × 1.498592 = 1.4236624 t CO2/t product. For an invented import mass of 100 t, FAA = 100 × 1.4236624 = 142.36624 t CO2. The 100 t scale is our arithmetic illustration; it is not a shipment in the guidance.

This result is only a free-allocation adjustment under those stated assumptions. It is not gross embedded emissions, a payable amount or a certificate quotation. For operational use, re-establish each reporting period and factor under the applicable law; Regulation 2025/2547 Article 7 fixes 2026 as the reporting period for goods imported during 2026. Do not use this 2027 illustration for those imports.

Use the Column B contrast as a method check, not a choice

For the same CN row, Regulation 2025/2620 Annex point 5.3 lists Column B values of 1.364 (C), 0.475 (D) and 0.066 (E), reflecting the specified route distinctions. Table 4-14 contrasts its carbon-steel example result with 1.364. The difference is not an invitation to select whichever number gives a preferred liability.

On a default path, Annex point 5.1 requires the route indicated for the relevant country of origin in the default-values act. Regulation 2026/1740 replaced relevant annexes of Regulation 2025/2621 and applies from 1 January 2026; its recital 2 expressly identifies missing or incorrect route indicators affecting the adjustment. Keep the corrected source version in the decision record. Do not infer a route from a generic supplier description or a broad HS heading alone.

Leave a calculation trace that another reviewer can replay

For multiple precursor installations, Regulation 2025/2620 Article 4 requires the relevant weighted-average treatment unless sufficient evidence supports the permitted installation/subset approach. Keep the emissions and free-allocation input populations consistent. A high-quality report from one plant does not justify using its benchmark contribution for unrelated material.

Retain the source row and column, code, route/year indicators, precursor quantities, periods, factor sources, precision policy and intermediate operations. Link them through the customs-to-installation lineage tests so that a changed package or route indicator marks dependent outputs for review.

Use a stop condition for unresolved inputs: no guessed route, missing-precursor zero or undocumented lower value. The output is a calculation trace for review, not a verification opinion or a complete liability calculation. Benchmarks, gross emissions and any eligible third-country carbon-price deduction remain separate parts of the eventual assessment.