Embedded emissions minus the benchmark times the free allocation still retained, times the certificate price, year by year, and why the 2.5 per cent shortcut understates every year before 2034.
The CBAM cost per tonne of imported steel is embedded emissions minus the free allocation adjustment (the product benchmark times the share of free allocation EU producers still retain), times the certificate price. At an illustrative 2.0 tonnes of CO2 per tonne, a benchmark of 1.37 and 80 euros a certificate, it is 53.14 euros a tonne in 2026, 103.56 in 2030 and 160.00 in 2034. The intensity, benchmark and price stay the same throughout; only the share retained moves, and it triples the 2026 cost by 2034.
Worked in full in The CBAM Compliance Handbook by Julian R. Sterling, with every figure reproduced in a free workbook.See the book on Amazon →
The definitive period of the Carbon Border Adjustment Mechanism, Regulation (EU) 2023/956, began on 1 January 2026. Importers pay for the embedded emissions of covered goods, including iron and steel, by surrendering CBAM certificates, whose price follows the EU ETS auction price. Certificates for 2026 imports cannot be bought before 1 February 2027, and the 2026 obligation is surrendered by 30 September 2027. The importer does not pay for all the embedded emissions: CBAM deducts the free allocation an EU producer of the same goods still receives, and that deduction shrinks from 2026 to 2034 as free allocation under the EU ETS is phased out.
| Input | Value |
|---|---|
| Imports a year, tonnes of steel | 12,000 |
| Supplier's verified embedded emissions, tCO2 per tonne | 2.0 |
| Embedded emissions a year, tCO2 | 24,000 |
| Product benchmark, tCO2 per tonne (illustrative) | 1.37 |
| Certificate price, euros, held flat (illustrative) | 80.00 |
| Steel price, euros a tonne (illustrative) | 600.00 |
The benchmark is illustrative. It is the same 1.37 used in how the CBAM free allocation adjustment is calculated, so the two articles reconcile. The Commission sets benchmark values by product and, where relevant, by production route in implementing acts: look up the value for your CN code before budgeting. A carbon price effectively paid in the country of origin can also be deducted, which is not modelled here.
Certificates per tonne = intensity − benchmark × share of free allocation retained
Cost per tonne = certificates per tonne × certificate price
2026: (2.0 − 1.37 × 97.5%) × 80 = 0.66425 × 80 = 53.14 euros. At the full rate, 2.0 × 80 = 160.00.
In Excel, with intensity in B2, benchmark in B3, price in B4 and the year's share retained in a row from D5: =($B$2-$B$3*D5)*$B$4
The 2026 figure splits into two pieces with very different prices. The 0.63 tonnes above the benchmark carry no free allocation at all and are paid in full: 50.40 euros. The 1.37 tonnes up to the benchmark are paid only on the phased-in 2.5 per cent, 0.03425 certificates or 2.74 euros. In 2026, 94.8 per cent of the cost is the excess over the benchmark.
The 2.5 per cent shortcut. The book's worked example, and many first budgets, multiply all embedded emissions by the phased-in share: 2.0 × 80 × 2.5% = 4.00 euros a tonne in 2026. That is a simplification that holds only for a supplier sitting exactly on the benchmark: at 1.37 tCO2 a tonne both methods give 2.74 euros. Above the benchmark, where most imported steel sits, the shortcut understates the cost, here by a factor of 13.3 in 2026. On terms: the ETS Directive calls the share retained (97.5 per cent in 2026) the CBAM factor; the book uses the name for its complement, the phased-in 2.5 per cent.
| Year | Free allocation retained | Certificates | Cost per tonne, € | Annual cost, € | % of steel price | 2.5% shortcut, €/t |
|---|---|---|---|---|---|---|
| 2026 | 97.5% | 7,971 | 53.14 | 637,680 | 8.9% | 4.00 |
| 2027 | 95.0% | 8,382 | 55.88 | 670,560 | 9.3% | 8.00 |
| 2028 | 90.0% | 9,204 | 61.36 | 736,320 | 10.2% | 16.00 |
| 2029 | 77.5% | 11,259 | 75.06 | 900,720 | 12.5% | 36.00 |
| 2030 | 51.5% | 15,533 | 103.56 | 1,242,672 | 17.3% | 77.60 |
| 2031 | 39.0% | 17,588 | 117.26 | 1,407,072 | 19.5% | 97.60 |
| 2032 | 26.5% | 19,643 | 130.96 | 1,571,472 | 21.8% | 117.60 |
| 2033 | 14.0% | 21,698 | 144.66 | 1,735,872 | 24.1% | 137.60 |
| 2034 | 0.0% | 24,000 | 160.00 | 1,920,000 | 26.7% | 160.00 |
| Nine years | 10,822,368 |
The cost per tonne rises 1.9 times by 2030 and 3.0 times by 2034. Of the 10,822,368 euros paid over the nine years, 72.8 per cent falls from 2030 onwards. The two methods meet only in 2034, when free allocation ends: over the nine years the shortcut would budget 7,852,800 euros and miss 2,969,568. 2026 is still the cheapest year the mechanism will have, but at 8.9 per cent of the steel price it is not a rounding error.
A five-year supply contract signed for 2026 to 2030 carries 4,187,952 euros of CBAM cost on these volumes, an average of 69.80 euros a tonne. A budget that holds the 2026 figure, 637,680 euros a year, flat for five years would carry 3,188,400, short by 999,552. A budget built on the 2.5 per cent shortcut would carry 1,699,200, short by 2,488,752. The question for the contract is who bears the shrinking free allocation in 2029 and 2030, and whether the price clause refers to the supplier's verified intensity, the benchmark for the CN code and production route, the certificate price, or all three. How to calculate a CBAM surcharge per tonne writes that clause as a formula with a true-up.
| Certificate price | 1.6 tCO2/t | 2.0 tCO2/t | 2.4 tCO2/t |
|---|---|---|---|
| 60 euros | 53.67 | 77.67 | 101.67 |
| 80 euros | 71.56 | 103.56 | 135.56 |
| 100 euros | 89.44 | 129.44 | 169.44 |
Moving the certificate price from 80 to 100 euros adds 25.89 euros a tonne in 2030; moving from a supplier at 2.0 to one at 2.4 adds 32.00, the full price on every extra tonne of CO2. Because the deduction depends on the benchmark and not on the supplier, the gap between a 1.6 and a 2.4 supplier is 64.00 euros a tonne in every year at 80 euros, already in 2026 (21.14 against 85.14). The intensity is the input the importer controls, through supplier choice and the quality of the verified data, which is why the data request belongs in the contract. Default values, used when verified data is missing, are set to be unfavourable.
The common mistake is to apply the phased-in share to all embedded emissions: 4.00 euros a tonne in 2026, which looks negligible. It is right only for a supplier on the benchmark. Here the true figure is 53.14, because every tonne above the benchmark is paid at the full certificate price from the first year. The opposite error, pricing at the full rate of 160.00 euros a tonne, is right only from 2034. The second mistake is to treat the deduction as a fixed parameter of the product: it is a calendar, and the same tonne from the same mill costs 3.0 times more to import in 2034 than in 2026.
The same phase-in, run on the book's Northbeck imports with your own quarters, intensity and prices, is in the free workbook for this case. What to hold above the obligation, and what that buffer costs, is in what a CBAM certificate buffer costs to hold.
Only for a supplier sitting exactly on the product benchmark. The deduction is the benchmark times the 97.5 per cent of free allocation EU producers retain, so emissions above the benchmark are paid in full. At 2.0 tCO2 a tonne, an illustrative 1.37 benchmark and 80 euros, steel costs 53.14 euros a tonne in 2026, not the 4.00 the 2.5 per cent shortcut gives.
The ETS Directive uses the term for the share of free allocation retained by EU producers: 97.5 per cent in 2026, 51.5 in 2030 and zero in 2034. The book uses it for the complement, the phased-in share, 2.5 per cent in 2026. Either way the deduction per tonne is the benchmark times the share retained: 1.33575 tCO2 at an illustrative 1.37 benchmark in 2026.
No. The deduction depends on the benchmark, not the supplier, so every extra tonne of CO2 is paid at the full certificate price from 2026. At 80 euros and an illustrative 1.37 benchmark, a supplier at 1.6 tCO2 a tonne costs 21.14 euros a tonne in 2026 and one at 2.4 costs 85.14, a gap of 64.00. Default values, used without verified data, are set to be unfavourable.
This article is one calculation from The CBAM Compliance Handbook. The book takes the same case from first principles to the decision, chapter by chapter, and every figure it prints is a live formula in the free companion workbooks.
Get the book on Amazon →Free companion files
Also on Amazon UK · Amazon Germany · Amazon France · Amazon Canada
Reading guide: regulation, compliance and banking → · All 1706 articles →
If this book helped, or didn’t, a few lines on Amazon are worth more than they look: they are what the next reader goes on. Write a review. The workbook stays free either way.