EU Steel Melt-and-Pour Declarations Now Apply, With 50% Duty Outside an 18.35m Tonne Quota

The Analysis

A 20-foot box of cold-rolled stainless sheet leaves Zhangjiagang on a Tuesday and is discharged at Rotterdam roughly three weeks later. Twenty-five tonnes, one bill of lading, one commercial invoice, one certificate of origin, and since 1 October one more question that nobody used to ask: in which country did this steel stop being liquid, and can you prove it with a heat number?

That question is the whole story. Implementing Regulation (EU) 2026/1963, adopted on 31 August 2026 and applied from 1 October, makes the country of melt and pour a declaration element on the customs entry for covered steel products, backed by a mill test certificate that carries the heat number. Its parent, Steel Regulation (EU) 2026/1384, has applied since 1 July and did the damage people noticed first: the annual duty-free quota cut by about 47 per cent down to roughly 18.35 million tonnes across 28 product categories, the out-of-quota duty doubled from 25 to 50 per cent, and allocation switched from annual to quarterly. The transition runs to 30 September 2027, and inside it invoices and other papers may supplement an MTC rather than stand in for it.

I work on the paperwork side, so I start somewhere different from the market commentary. This does not land on a price index. It lands on one field in one declaration, propped up by one certificate that most importers have never had to read past the mechanical properties. If that field is wrong or unsupported, the entry does not go in cleanly; if it goes in without quota cover, the bill is 50 per cent of customs value. What follows is which paper changed, which element changed, where the consignment physically stops, what a stop costs on a real box, and what to chase before this afternoon is out.

Take the numbers in the order an importer meets them, because the order is the point. Roughly 18.35 million tonnes is the new annual duty-free volume, arrived at by cutting the previous allowance by about 47 per cent, which run backwards puts the old figure somewhere near 34.6 million tonnes. It sits across 28 product categories. Above it, duty is 50 per cent where it used to be 25 per cent. Allocation is quarterly. Those are the four figures that belong in your cost model, and the fifth, 30 September 2027, is the day the supplementing-papers tolerance ends.

The quarterly switch deserves more attention than it got. An annual pool spreads the pain and lets a late shipment argue it was always going to be late. Four pools a year produce four separate exhaustion events, and exhaustion does not happen on a schedule you can plan around; it happens on the day the tonnage runs out. If your category's quarter is gone by the second week of November, the remaining ten weeks of that quarter are all at 50 per cent, and the next quarter opens clean whether or not you had a shipment waiting on the quay.

Why now. Europe's steel safeguard had reached the end of the life that WTO rules allow for it, so what replaced it had to be a fresh Regulation rather than another extension. Behind that sat a problem the Commission had been describing in plainer language for a couple of years: steel melted in one country, shipped out as slab or billet, rolled or merely re-rolled in a second, and presented at customs as the second country's product. The non-preferential origin test, built around the last substantial transformation, left plenty of room to argue about where that transformation took place. Melt and pour removes the argument. Steel has exactly one moment when it stops being liquid, and one country where that happened.

Which is also why the two measures arrived as a package instead of one after the other. The quota decides how much steel gets in duty-free; the melt-and-pour declaration decides whose steel it counts as. Turn one knob and the trade routes around it. Turn both and the pattern that used to work, melt in one place, roll in another, declare the second, stops working on paper well before it stops working on the water.

Now to the part that lands on the entry. Three countries appear in a steel file and they are often three different places. There is the country of melt and pour, which is what (EU) 2026/1963 asks for. There is the country of origin under the non-preferential rules, which is what your certificate of origin and most of the rest of your formalities were built on. And there is the country of dispatch, which is simply where the box sailed from. When the three coincide, nothing changes beyond one extra field. When they do not, slab melted in Indonesia and rolled in Italy and shipped from Antwerp, billet melted in China and drawn into wire rod in Turkey and invoiced by a trader in Dubai, you now hold a file in which the new declaration and the old certificate point at different countries, and that is precisely the kind of file that gets pulled.

The certificate behind the new field is where the real work sits. A mill test certificate used to be a quality document: chemical analysis, mechanical properties, a signature, filed against a complaint that might never arrive. It lived with the mill's QC department and with the buyer's technical people. Under (EU) 2026/1963 it becomes a document that supports an entry, and the heat number on it becomes the key that unlocks the declaration. Certificates get issued late, per lot rather than per heat, in the mill's own format and frequently in the mill's own language, and the people who issue them have never once worked to a customs deadline. That mismatch, more than anything written in the Regulation itself, is where I expect boxes to sit.

On the importer side the liability is clean and it sits at the top. The declarant is the importer, not the mill and not the trader. If the melt-and-pour field is wrong, the exposure is yours, and the certificate you need is in somebody else's hands. That asymmetry is the practical shape of this rule for European buyers: you are answering a question about a furnace you have never seen, on the strength of a document you do not control, against a deadline set by a ship's arrival.

On the supplier side the burden is different, and heavier in a way that is easy to miss. A mill with a proper heat-level traceability system can produce the paper, and the cost is administrative. A mill that sells through stockists, that re-issues certificates per shipment, or that treats the MTC as a sales attachment rather than a traceability record will struggle, and the struggle will show up as delay at your end rather than at theirs. For exporters reading this from Asia: the mills that can tie a coil tag to a heat number to a certificate, on request and within days, are going to take orders away from the ones that cannot, and that sorting has already begun.

Traders and service centres sit in the worst seat of all. A service centre that buys coil, slits it, packs it and sells it is not the melter, and the only certificate it can produce is its own. If its supplier will not pass heat-level data down the chain, the service centre has nothing to declare with, and the goods are not rejected on quality grounds at all, they are simply unproven on origin grounds. Nobody writes a letter saying that. The file just stops moving.

Downstream, the transmission is slow rather than sharp, and it runs on the same mechanism. Stainless is the clearest case. Public figures for January to July 2026 put Indonesian stainless exports to the EU at about 127,900 tonnes, of which roughly 120,000 tonnes, or 93.9 per cent, was flat product, and the established pattern in that trade is European mills buying Indonesian slab and finishing it at home. Once origin is read at the furnace, that pattern reads differently. Wire, tube and automotive grades follow a step behind, not because the rule names them but because their raw material now carries a tracing requirement that has to be answered one level up. I would keep the size of that second-order effect in proportion: the slab logic is direct, the wire and tube numbers are an inference about how strictly the declaration is policed, and an inference is not a forecast.

Timing is where this stops being a compliance story and becomes a cash story. The declaration is made when the entry is accepted for release for free circulation. That is the moment the melt-and-pour field is read, the moment the MTC is either in the file or not, and the moment the tonnage is drawn against the quarter's quota for that category. Not the loading date, not the arrival date, not the contract date. So the practical clock has three hands: whether your certificate exists before the vessel berths, whether your category still has quota on the day the entry goes in, and how many days the file sits if the answer to either one is no.

Put a number on it. Assume one 20-foot container, 25 tonnes of cold-rolled stainless sheet classified in CN 7219.34, customs value 1,850 euros a tonne, so 46,250 euros for the box. Inside the quota, duty is nil. Outside it, at 50 per cent, duty is 23,125 euros, or 925 euros a tonne on material you paid 1,850 euros a tonne for. Before 1 July the same box out of quota would have paid 25 per cent, which is 11,562.50 euros, so the doubling on its own is an extra 11,562.50 euros on a single container. Add Dutch VAT at 21 per cent: 9,712.50 euros on the duty-free value, 14,568.75 euros on the duty-inclusive value, so about 4,856 euros more VAT falls due at the border.

That VAT is recoverable for a taxable importer, but it leaves your account first, so the immediate cash hit on one 25-tonne box is roughly 23,125 euros of duty plus about 4,856 euros of VAT carried for a filing period. Ten boxes a month into the same category, all out of quota, is 231,250 euros a month of duty and about 48,560 euros a month of VAT fronted.

The quota arithmetic is worth doing as well, with the assumption stated plainly. Eighteen point three five million tonnes a year, split four ways, is about 4.59 million tonnes a quarter across all 28 categories; divided equally, which it certainly is not, that is roughly 655,000 tonnes a category a year and about 164,000 tonnes a category a quarter. Your own category is unlikely to be anywhere near average, because the big hot-rolled and long-products lines dwarf the stainless sheet lines, so the average only tells you the order of magnitude and the published per-category quarterly figure is the only one worth planning against. On that basis a single 25-tonne box is a rounding error against a whole quarter, which is exactly why exhaustion is a category-level question rather than a shipment-level one: you are not worried about your box, you are worried about everybody else's boxes landing in the same quarter as yours.

Here is the thing I have not seen in any of the coverage, and it is the one that will cost money in the next ninety days. The change is not that you must now state a country. The change is that the heat number has been promoted into a customs declaration element, and the heat number is produced by a department that has never worked to a customs deadline. Everything upstream of the entry, the quota, the 50 per cent, the quarterly allocation, is now reachable only through a piece of paper generated by a melt shop's quality system and delivered through a commercial chain that has no obligation to deliver it on time.

Follow that through to the reconciliation problem, because this is where files actually break. The quota is counted in tonnes. The MTC is issued per heat. The entry is made per goods item on a commercial invoice, and the box in front of you may hold four heats from two casts. So a declaration that can be checked needs a heat register: heat number, tonnes in that heat, certificate number, coil or bundle tags it covers, invoice line it belongs to, entry number it was declared under. Six columns, per shipment, and the mill is the only party able to fill in the first three.

If your certificate lists a range of heats for a lot, or one certificate covering twelve heats with no per-heat weight split, or a heat number that does not match the tag welded or wired onto the coil, you cannot allocate tonnes to a country with any confidence and you cannot draw quota against them. That is the failure mode. It is not an officer disagreeing with your legal theory; it is a spreadsheet that does not add up.

The transition date is being read as breathing room, and I would read it the other way round. Yes, until 30 September 2027 other papers, invoices, certificates, correspondence, can supplement an MTC. That is a tolerance for imperfect evidence, not a suspension of the quota and not a suspension of the 50 per cent. The quota is already burning quarterly, and every quarter between now and September 2027 is a quarter in which your out-of-quota exposure is 50 per cent instead of 25. Waiting for the paperwork to be perfected costs you the duty, not the deadline. Anyone planning to see how the transition goes is planning to pay double duty on the shipments that went in unproven.

Which brings me to the conditions that would flip this, and three are worth writing down. A big one is supplier capability: if your mill can produce heat-level MTCs within days and the melt country is one you are comfortable declaring, the whole rule is an administrative cost and the 50 per cent never touches you. Another is the shape of your quota line, because a country-specific allocation and a residual global pool behave completely differently, they exhaust on different dates and the competition for the residual pool is far fiercer, so find out which one you are in before you model anything.

And then there is practice, which is not uniform. How hard an entry is looked at, what language the certificate has to be in, and whether a supplement is accepted in the first weeks of the transition all depend on the port. One file at Rotterdam and another at Piraeus can get two different answers in the same month, and neither answer is a precedent for the other.

There is also a boundary that the headlines are blurring, and it is worth being precise about. Dr. Ingrid Voss has made the point more than once in these pages: a melt-and-pour declaration tells you where the steel solidified, but it does not by itself settle which origin rule applies to some other purpose, whether preferential treatment is available to you, or how an existing anti-dumping or countervailing duty interacts with the new field. Those are separate tests with separate conditions, and a declaration that satisfies one can still fail another. Where exactly that boundary falls in your product's case is a question for your legal counsel and for the text of the measures, not for a forwarder and not for me. What I can tell you is operational: assume the field has to be right on its own terms, and do not assume that getting it right buys you anything elsewhere in the entry.

On alternatives, the honest answer is that there are fewer than usual and every one of them costs something. Moving sourcing to a melt source you can document is a supplier qualification exercise, and for most buyers that is a six to twelve month job rather than a November job. Moving the timing of the declaration, pulling entries forward into a quarter that still has quota, or holding goods in a customs warehouse until a fresh quarter opens, is a genuine lever, and warehousing is the cleanest of them, because it separates arrival from entry and lets you choose when the tonnage is drawn; it costs storage and it changes your cash cycle, but it is deliberate and worth pricing against 50 per cent.

Moving the market altogether, redirecting volume to South-East Asia, the Middle East or Africa, is what the commentary keeps suggesting, and it is a pricing decision rather than a paperwork one: those markets absorb volume at thinner margins, and everybody squeezed out of Europe arrives there in the same quarter. The traps are duller and more expensive. A certificate of origin that quietly contradicts the melt-and-pour field is the one I expect to see most. A lot certificate carrying a range of heats is the second. The third is the service centre that re-issues papers under its own name and drops the heat number on the way through.

So, concretely, before this afternoon is out. Begin by emailing the mill or the supplier rather than the forwarder, and ask for the MTC on every EU-bound shipment already on the water: heat number, per-heat tonnage, certificate number, and the standard it was issued to. Next, read that certificate against the coil tags and the commercial invoice and see whether the weights add up to the shipped tonnage, because if they do not, that is your problem today rather than at the port. With that done, call whoever files your entries and ask two narrow questions, which field carries the melt-and-pour country and what supporting document code they want the MTC attached under.

After that, find the published quarterly quota position for your specific category and country line and write the number down, since you cannot manage the 50 per cent risk without it. Then build the heat register, six columns, and turn heat-level MTC delivery into a condition of the next purchase order instead of a request made after shipment. Papers that do not reconcile cause trouble the whole way down the line, and under this rule the certificate is the whole thing.

↑ Back to the key points

— By Derek Xu