EU Sets Final 39.8-48.1% Anti-Dumping Duty on Chinese PET Spunbond From Oct 31 for Five Years

The Analysis

The notice went up on 30 October and the duty started on 31 October. One day between them. For anyone with a container on the water, that day is the whole story.

The European Commission set a final anti-dumping duty on Chinese PET spunbond at 39.8% to 48.1%, in force from 31 October 2026 for five years. The rates split by company. Hubei Unibon takes 39.8%. Changde Tiandingfeng and every other Chinese producer take 48.1%. The goods in scope are needle-punched polyester filament sheet above 70 grams per square metre and between 0.5 and 1.8 millimetres thick, under CN ex 5603 13 90 and 5603 14 20/80. China shipped 20,800 to 25,400 tonnes into the EU in 2024, about 29% of the market.

I work on entry documents, so let me tell you where this actually lands. Not on the price list. On the declaration.

Begin with the two letters sitting inside the tariff code, because that is where this will be fought. CN ex 5603 13 90 and ex 5603 14 20/80 are not product codes. They are cupboards. The ex means the ten-digit line holds more than one product and the duty attaches to a slice of what is inside. Which slice is fixed by words, not by the number: needle-punched, polyester, filament, over 70 g/m², 0.5 to 1.8 mm.

So the question at the desk is never which code. It is whether the roll in front of you passes those tests, and the tests are physical ones. A spunlace line, a thermal-bonded line, a staple-fibre line can all sit inside the same ten-digit number and none of them pay this duty. That is good news for about half the people reading this, and it is only good news if the file backs it up. The entry needs a technical basis behind it: grammage, thickness, bonding method, polymer. Not the sales brochure. The mill's own quality sheet, with the test method printed on it.

Here is what catches people on their first shipment after a ruling like this. Your own data sheet is evidence. If the specification in your system says 72 g/m² and the customs lab measures 68, the lab number does not rescue you, because the entry you signed declared a product inside the scope, and the sheet that put it there came from your supplier and was filed by you. The document that would have kept you outside the scope was already in the folder. It said the wrong thing.

And when the goods are inside the scope, the declaration carries more than a rate. It carries the additional code that tells the system which slice of the ex line you are entering, the producer's identity, and the duty amount in a separate field. Three fields, one of which an inspector reads against the invoice and one against the specification. Get the third one wrong and the first two will not save the entry.

One structural detail worth noticing, because it tells you how the scope was drawn. The two codes sit in different weight bands of the same heading, and the ruling reaches into both of them. That is not how a tariff works. That is how a product description works. Whoever drafted this defined the goods first and then went looking for the boxes they lived in, which is why the ex matters and why a friendly customs broker who says "5603 13 90, no duty on that" is answering a question nobody asked. The code is where the duty is collected, not where it is defined.

Read the rates backwards and the procedure shows itself. One company at 39.8, the whole field at 48.1. An 8.3 point gap between a single mill and everybody else is not a negotiation. It is the shape of a sampling exercise. The Commission examined a handful of exporters, built a rate on each one's verified costs, and left the rest of the field on a residual. If you were never in the investigation, the residual is your rate, and it does not matter that your costs are lower than the mill next door. That number is administrative. It has nothing to do with your economics.

Why this product, and why now. China put 20,800 to 25,400 tonnes into the EU in 2024, which was 29% of the market, so the market runs somewhere between 72,000 and 88,000 tonnes and domestic producers held about 71% of it. Seventy-one percent share, and a case was still brought. A complaint does not get filed over volume at that level. It gets filed over price at the bottom of the range, in the commodity grammages where one roll is much like another and the only real variable is cents per kilo.

The five-year term tells you something about how the case was built. Five years is the standard period, not a judgement about how long the injury lasts, and the duty does not quietly end on 31 October 2031. It ends when an expiry review closes, and those reviews run 12 to 15 months, so the practical horizon for planning is nearer 2033. If you are writing a supply agreement with a term that crosses 2031, the duty clause needs to say what happens during the review, not just what happens on expiry.

Three populations, three different problems, and they do not share a timeline.

The mill at 39.8% holds the only workable rate in the country and it knows it. Its EU allocation just turned into scarce capacity. Watch what happens to price: it can move most of that 8.3 point gap into its own selling price and still be the cheapest Chinese origin in the market. Buyers with no volume commitment will find the allocation gone by the first quarter. If you have a supply agreement with that plant, read the volume clause this week, not next month.

Everybody at 48.1% is out of commodity grades and stays only where the specification is hard to move. Smaller market than it sounds, stickier than people expect.

The EU converter is the one who actually eats this. A converter buys roll goods by the tonne and sells wipes, filtration media, geotextile, automotive trim. Its customer qualified that material over six to twelve months, often against a filtration or medical standard that cannot be swapped by changing a purchase order. So the converter pays the duty at the border, passes on what the market allows, and absorbs the rest for at least two quarters. That margin compression is the real transmission channel, and there is no line for it in the trade statistics.

Then the fourth group, which nobody writes about. Turkish, Indian and Vietnamese needle-punch capacity gets a gift from this ruling, and the gift is smaller than the headline suggests. That capacity is deep in commodity geotextile weights and thin in the 70 to 120 g/m² filtration and medical grades, at short notice, with the certifications. A 48.1% duty on a grade nobody else can make on time is a 48.1% duty your customer ends up paying.

The clock is the part I would pin to the wall. Anti-dumping duty attaches when goods are released for free circulation. Release means the entry. Not the bill of lading date, not the contract date, not the day the container was loaded in Ningbo. Published on 30 October, effective on 31 October, so every entry accepted from the 31st carries the duty.

Asia to North Europe runs 30 to 40 days port to port plus a week at each end. A booking loaded in the third week of September and discharged in the last week of October is in scope. The price on that contract was agreed in August, when the number did not exist. There is no grandfathering for contracts already in force and no transition period written into the notice.

This is where the Incoterm decides who is angry. On CIF or FOB, your buyer is the importer of record and your buyer pays. On DAP or DDP, the seller is the importer of record and the seller pays, and that seller priced the deal with no duty inside it. Two things happen next, and I have watched both this year. The seller renegotiates, or the seller stops answering. A DDP contract renegotiated in the first week of November costs you a price increase and a fortnight. A DDP seller who goes quiet costs you the goods.

One more timing note that saves people money and is almost never used. Customs warehousing does not avoid the duty, but it defers it, and deferral is not nothing. If a shipment lands in the last week of October and the customer is not ready to take delivery, put it under customs warehousing and let the entry date fall where you choose. That buys you the time to sort out which producer declared what, and it stops a sorting problem from becoming a tax problem at the counter.

Now the thing the coverage left out, and it is the one that decides your margin.

The rate follows the producer. Not the exporter, not the brand, not whoever issues the invoice. Anti-dumping duty is assessed against the manufacturer, and the manufacturer is a fact you prove at the border, not a name you get to choose. Which turns the 8.3 point gap from a market observation into a document question. You either have the paper or you do not.

Run the chain. You buy from a trading company in Hong Kong or Shanghai. The invoice comes from the trader. The goods were needled at a plant in Hubei, or Changde, or somewhere in Jiangsu. If the entry names the trader as producer, or names no producer at all, you take the residual: 48.1%. If it names a producer that is not on the list, same result. The only route to 39.8% is a producer declaration naming the plant, matching the invoice line for line, and able to survive a verification visit. On the arithmetic below, that single page is worth €2,191 a container. It is the cheapest document in the file by a wide margin, and it is the one nobody asks for until the entry has already been lodged.

This is the rule premise Ingrid Voss keeps putting at the top of her notes, and she is right about it: in an anti-dumping case the dutiable unit is the producer, so the commercial chain in front of the plant is invisible to the duty and the paperwork behind it is everything. A trader can add margin. A trader cannot add a rate.

The second thing missing from the headline. Duty is charged on customs value, and customs value includes freight and insurance to the border. Your CIF number is the base, not your ex-works number. Teams that model this off the factory price understate the hit by exactly the freight component, and on a container of roll goods that is not a rounding error.

The third is smaller, and it is the one that produces penalties rather than tax bills. The declaration and the technical file have to agree. Weight, thickness, bonding method, polymer. If the entry says needle-punched filament at 75 g/m² and the sheet in your own system says 68, you have handed an auditor a contradiction you built yourself. Post-clearance recovery plus a penalty is a worse day than the duty itself, because the duty was always going to be paid by somebody and the penalty is paid by you.

Here is where the conclusion turns over. All of the above assumes Chinese origin. Needle-punching is a substantial transformation, so material needled in Vietnam from Chinese filament is Vietnamese goods and outside this duty. That door is real and it is open.

It is not free. The Commission can open an anti-circumvention investigation and, if it finds the pattern, extend the duty to the flow that was redirected. That is a slow instrument and a blunt one, but it exists, and a plan whose entire margin rests on a single origin pivot has a single point of failure that one official decision can remove.

The honest reading is that origin switching pays at one end of the range and not at the other. On commodity geotextile, where the specification is a weight and a tensile figure, a new mill can be qualified inside a quarter and the duty is avoided at a landed cost increase of maybe 8 to 12%. On a filtration or medical grade with a registered specification, requalification runs six to twelve months and your customer's change-control process is the real gate, not your sourcing team. There, paying 48.1% and passing it through is cheaper than moving. Same duty, opposite answers, and the variable is the specification. Never the price.

Which also means the industry's next move is predictable and slower than the market expects. Capacity does not appear in filtration grades in a year. It appears in geotextile grades in nine months, and that is where the price collapse will be, not where the shortage is.

Put numbers on it. These are my assumptions, labelled as such, not figures from the ruling.

One 40-foot high cube of 100 g/m² needle-punched roll goods, 12 tonnes net, CIF Rotterdam at €2.20 a kilo. Customs value €26,400. At 48.1% the duty is €12,698. At 39.8% it is €10,507. The gap is €2,191 on a single container, which is 8.3% of customs value, because that is what a rate gap is once you stop calling it a percentage.

Twenty containers a quarter is €43,820 of difference. For a converter turning over €8m to €12m a year, that is the gross margin on the quarter moved from one column to another by one signed page.

Import VAT sits on top and makes the cash position worse before it makes it better. VAT is charged on customs value plus duty, so at 19% you finance €7,428 instead of €7,012 on that box. The €416 comes back, so it is cash flow and not cost, but you carry it for 30 to 45 days, and on twenty containers a quarter it is real working capital parked in a VAT account.

Then look at what has to happen to your selling price. To hold margin, the roll has to leave at €2.20 plus €1.06 of duty, so €3.26 a kilo. At €3.26, Turkish material at €2.40 and Indian at €2.55 win every line where the specification allows them. They lose where it does not. Which brings the whole thing back to a single question, and it is not a customs question: is your grade a commodity or a specification?

What I would do, in this order, this week.

Begin by sorting the open book by entry date. Released before 31 October, arriving before 31 October, arriving after. The first pile is closed. The second pile is the one that gets missed: cargo discharged on the 29th and entered on the 3rd pays, and there is nothing to appeal, because the charge attaches at release and not at discharge.

Next, get a producer declaration from every supplier with an open order. One page, on the mill's letterhead, naming the legal entity that needle-punched the goods, its address, and the invoice numbers covered. Not an email saying our factory. A signed page, obtained this week, before the cargo sails, because a declaration collected after the entry is a declaration collected after the rate was fixed.

After that, pull every DDP and DAP contract with a Chinese seller and decide which ones you want to keep. A pass-through clause signed in November is worth more than an argument in January, and a seller facing 48.1% will take a renegotiation over a cancellation every time.

Then, for goods already in scope, do the customer-side arithmetic before the internal one. If the grade is registered, quote the increase and defend the volume. If it is commodity, start qualifying two non-Chinese mills now, with both qualified inside two quarters, and book the 8 to 12% landed increase on transition volume as a switching cost rather than a margin loss. Take the loss in one line of the P&L where everyone can see it, because the alternative is taking it quietly for two years.

One more thing to keep straight. A binding tariff information decision settles the code. It does not settle the scope. Do not buy a BTI and think you have bought a shield: the scope lives in the product description in the regulation, and the entry has to match the product, not the ruling.

↑ Back to the key points

— By Derek Xu