Image: Container ships on Saigon River 3 by Syced, licensed under CC0 1.0.
Your audit checked the emissions surcharge on an Asia to North Europe invoice, found the same figure on the carrier's published tariff, and passed the line. All that check established is that the carrier can copy its own number onto a bill.
Index-linked charges are not prices. They are the output of an arithmetic expression whose inputs sit in public indices, in a contract clause, and in the vessel's actual routing. Comparing the output to a tariff sheet tests none of those. It is the one audit step that cannot fail for the reason you care about.
A quoted surcharge is a result, not a rate
Two carriers have published enough of their method to show the shape of the problem. MSC's ETS surcharge explainer sets the charge as a carbon coefficient multiplied by the EUA index, taken as a three month average and scaled by the directive's phase-in percentage. The coefficient is defined as the in-scope CO2 emissions for a direction divided by the TEUs carried in that direction, with reefer cargo multiplied by 1.5 for the electricity it draws. Alphaliner noted at the time that MSC was one of very few lines to publish its methodology at all.
Maersk's bunker formula has the same structure: fuel price multiplied by a trade factor. The fuel price is an average across key supply ports over a window the carrier describes as typically three months, read off Platts' index for 0.5% sulphur fuel oil. The trade factor is consumption in tonnes per forty foot equivalent on that lane, adjusted for headhaul and backhaul imbalance. Maersk's December 2025 notice put the fuel input at $461.54 per ton for the 26 August to 25 November review period, down from $509.82 in the previous one, with the resulting tariffs effective 1 January 2026.
Every term on the right of those equations is observable. Two are public, one is a date range the carrier states in advance, and one belongs to the carrier. An audit that stops at the left side never touches any of them.
Which reading applies is a date question
The tariff Maersk put into force on 1 January 2026 was built from an index window that had already closed on 25 November. So the question is never what fuel was worth on the day the vessel sailed. It is which review period governed that sailing, and what the index averaged across it.
Teams that reach for a spot quote on the sail date get a disagreement and conclude they have found something. Usually they have found the wrong window. The reverse failure is quieter: the charge looks plausible against today's market, so nobody notices it was computed from a period that closed two quarters ago.
Movement inside a window is real. EU carbon permits closed at €88.08 per tonne on 13 October 2026 according to Trading Economics, inside a four week range of €84.02 to €88.35. Averaging flattens that, but only if the average covers the span the contract names.
Contract type decides the formula too. CMA CGM's quarterly floating BAF applies to contracts and quotations with at least three months of validity, while spot and shorter bookings take EBS instead. A floating bunker charge billed against a spot booking is not an inflated figure. It is a published figure applied under the wrong clause, and no rate comparison will ever surface that.
The coefficient is where the money sits
The index is public and identical for every carrier buying allowances. The multiplier in front of it is not, and that asymmetry shows up directly in what shippers are billed. Searoutes' figures for the first quarter of 2026 put the ETS charge per dry TEU from East Asia to North Europe at $84 for Maersk, $82 for Hapag-Lloyd, $84 for CMA CGM and $93 for ONE. On intra North Europe moves the same table reads $16, $49, $55 and $83.
One regulation, one allowance price, and a fivefold spread on the short lane. The carriers are not disagreeing about carbon. They are disagreeing about emissions per box, which turns on vessel utilisation, trade balance and how a line allocates a voyage across the containers aboard.
That term has been contested at scale. Transport & Environment's March 2024 study examined 565 voyages across 20 ships operated by Maersk, MSC, CMA CGM and Hapag-Lloyd, and concluded that surcharges exceeded actual compliance cost on roughly 90% of them, working from an assumed carbon price of €90 per tonne. Maersk rejected the finding on the grounds that no fixed €90 price had been announced. Notice what the two sides were arguing about: an input. Neither party disputed the headline figure on the invoice, because the headline figure was never the point.
Scope is decided by the vessel, not the contract
From this year the EU scheme covers 100% of emissions on voyages between two EU ports, up from 70% in 2025, and 50% on voyages between an EU port and a non-EU one, up from 35%. Methane and nitrous oxide entered scope at the same time.
Which of those two percentages applies turns on where the ship last called before entering the EU. Searoutes illustrated the gap with services arriving at the same region: one whose final pre-EU call is Tanjung Pelepas carries 89% of its distance in scope and lands in a $30 to $45 per TEU band, while one calling Felixstowe last has 3% in scope and bills $3 to $10. Your contract text is identical in both cases.
Reroutings therefore change the correct charge without changing a single negotiated term. Chase holds the port call sequence that settles which percentage was owed, and on current Red Sea routings that sequence is not stable from one sailing to the next.
Hapag-Lloyd told customers to expect roughly a 45% increase from the 2026 regulatory step, and its Far East Asia to North Europe tariff rose $54 in a quarter to $168 per forty foot dry container. A variance rule tuned to flag that jump flags a correct charge, and one loosened to permit it waves through whatever follows.
What recomputation requires
| Input | Rate match | Formula audit |
|---|---|---|
| Reference checked | Carrier's own tariff sheet | The figure the contract formula produces |
| Index | Not used | The named series, averaged over the stated window |
| Period | Not used | Resolved from sail date against published effective dates |
| Multiplier | Not used | Trade factor or carbon coefficient, held per lane and direction |
| Scope | Not used | 50% or 100%, derived from the actual pre-EU call |
| Equipment | Not used | Reefer uplift where the tariff carries one |
| What it catches | A mistyped tariff | An input that moved while the charge did not |
How Penny handles a formula line
Penny stores the expression rather than the answer. For each index-linked line she resolves the tariff period that governed the sailing, pulls the index readings across that window, applies the coefficient the contract names, and derives what the charge should have been before looking at what it was.
Differences then get attributed to a term instead of escalated as a blanket dispute. The window was wrong, or the coefficient does not match the contracted one, or the scope percentage assumed an EU to EU leg the vessel did not sail. A dispute that names the input is one the carrier can settle without a meeting.
When the contract names no coefficient at all, that absence is the finding, and it belongs upstream. Miles can pin the term at renewal far more cheaply than Penny can argue it after payment, which is the same point our post on why road audit logic collapses on containers makes about evidence.
Where to start this quarter
Pick one trade lane and one quarter, then write down four things from the contract: the index it names, the review window it specifies, the multiplier it fixes, and the scope rule it assumes. Rebuild three invoices from those alone and compare.
If the contract does not supply all four, stop there. You have the result already. A surcharge you cannot reproduce from its own formula is a surcharge you can only accept, and accepting it is what your audit has been doing all along.
Book a walkthrough to see Penny rebuild index-linked surcharges from your own contracts and the published indices behind them.