EU Scraps Solar Glass Duties of Up to 75.4% From Oct 22, Refunds Back to July 2025

The Analysis

Here is how this shipment actually moves, because the news coverage describes taxes and this decision changes paperwork. A container of solar glass leaves a kiln in Anhui, forty tonnes to the box, and lands at Rotterdam. A Dutch importer enters it for release for free circulation. The declaration line carries CN code ex 7007 19 80 plus one of four TARIC codes, 7007 19 80 12, 18, 80 or 85, and the duty calculator behind that line has been doing something brutal to this importer's cash flow for fifteen months: an anti-dumping duty of 67.1 per cent of customs value for anyone without an individual rate, or 17.5 to 75.4 per cent for the producers that had one, with a countervailing duty of 3.2 to 17.1 per cent stacked on top. At the residual rate with the highest countervailing rate, that is 84.2 per cent. Nobody negotiates a number like that. It just sits in the cost line and gets paid.

From 22 October 2026 it stops. Implementing Regulation (EU) 2026/2314 was published in the Official Journal on 21 October and takes effect the following day, repealing the anti-dumping duty on solar glass from China, and Implementing Regulation (EU) 2026/2310 does the same to the parallel countervailing duty. It also reaches backwards. Duty collected on goods already cleared for free circulation since 22 July 2025, the day the review was initiated, can be asked back. No interest comes with it, and the request goes to the customs administration of the member state that did the clearing, under that administration's own procedure.

Which leaves two different jobs on one desk today. One is forward-looking and mostly a systems question: does the declaration software stop applying the measure on the right date, and does the customs debt stop being secured against your guarantee. The other is retrospective and mostly a file hunt with real money at the end of it. Do the second one before you call your supplier about price, because the party customs pays back is usually not the party whose margin absorbed the duty.

Start with what precisely got repealed, because two small words in the scope carry most of the practical risk. The commodity line is CN code ex 7007 19 80, and that ex matters: the measure never covered the whole of subheading 7007, only the part identified by the ten-digit TARIC codes 7007 19 80 12, 7007 19 80 18, 7007 19 80 80 and 7007 19 80 85. If your declarations sat inside those four codes, the duty applied and, since 22 July 2025, you probably paid it. If your product is solar glass in commercial terms but was declared somewhere else in the nomenclature, there is no refund waiting, because no solar glass duty was ever due. The common case here is the glass that arrives already assembled into a finished photovoltaic module classified under heading 8541. Those importers are not part of this story at all, and the sooner their teams establish that, the sooner they stop generating internal emails about it.

Inside the four covered codes there were two populations, and they are worth separating before anybody estimates a number. The producers that cooperated and received an individual anti-dumping rate were paying between 17.5 and 75.4 per cent. Everybody else was paying the 67.1 per cent residual rate, which is the default position for exporters that were not individually examined. On top of either of those sat a countervailing duty of between 3.2 and 17.1 per cent, aimed at the subsidy rather than the dumping. Duties in the European Union stack rather than merge, so the realistic combinations ran from roughly 20.7 per cent at the very bottom, where a 17.5 per cent individual rate was paired with a 3.2 per cent countervailing rate, up to 92.5 per cent, where a 75.4 per cent individual rate was paired with a 17.1 per cent countervailing rate. The residual payer with the top countervailing rate was handing over 84.2 per cent of the declared customs value before any transport inland had even been invoiced.

The reason this ended now is a set of findings, not a change of mood in Brussels. The review had to ask whether removing the measures would lead to a continuation or recurrence of dumping and injury. It concluded it would not, and the evidence underneath is blunt: the Union industry had stopped producing the product under investigation during the review period. No injury can continue or recur if nobody in the Union is making the thing that would be hurt, and an absent producer also removes the constituency that normally turns up to defend a measure. Keeping it in place was additionally found not to be in the Union interest, the limb where users and importers get counted. European module assemblers buy this glass or they buy nothing, and they paid a tariff for years on an input their own industry no longer makes.

The share number explains why the politics were easy. Chinese material held 30.5 per cent of the Union market when the measures went in. It sits at about 3 per cent now. A duty that coincides with its target falling to a rounding error is not protecting an industry, it is taxing a memory. That also hints at durability: every condition justifying this repeal is a statement about who produces what, and production decisions reverse faster than a Commission file can be opened.

So who is touched and by how much? Anyone who imported glass under those four TARIC codes between 22 July 2025 and 21 October 2026 carries two separate numbers: cash saved on future entries from today, and a reclaimable amount sitting in a national customs ledger. The module and installer layer further up the chain never saw the duty at the border but paid it inside every panel mounted on every roof or tracker, and that group collects nothing from customs and everything from its own purchasing department. Then there is the buyer who switched away from Chinese glass for reasons that stopped existing this week, whose sourcing decision belongs back on a landed cost comparison this quarter.

Now the part that no headline mentioned, and it is the part that decides whether any money actually moves. Nothing about this repeal is automatic, and nothing about it is refunded to whoever suffered the cost. The refund goes to the person who paid the duty at the customs counter, and that means the claimant is the declarant named on the entry, or whoever was debited when the duty became a customs debt. In a lot of these programs, though, the Chinese exporter priced DDP and quietly absorbed part or all of the duty to keep the European customer, or the importer pushed the cost into the module price it charged onward.

In either case the cash coming back lands in one legal entity's bank account while another party was the one who actually financed it. Someone has to sort out who gets what, and if that is not settled before the claim is filed, it turns into the worst kind of argument: one where both sides can point at the same money. Dr. Ingrid Voss would start every question like this from the same place, which is who the rule was written for and against whom it runs, and here that habit is worth adopting, because the rule runs against the payer and not against the loser.

Add three further constraints wrapped inside the same sentence of the Regulation, each of which quietly reduces what people will actually collect. The first is interest: there is none. The Union took these amounts and holds them without compensation, and the new Regulation does not change that. The second is geography: the claim goes to the member state that collected, under its own procedure, and twenty-seven procedures are not the same procedure. A group that cleared in Rotterdam, Antwerp and Hamburg is running three separate processes with three different documentary appetites, and the slowest one sets the cash date for the whole exercise. The third is evidence: you do not get repaid against a spreadsheet that says we imported a lot of glass. You get repaid against entry numbers, and specifically against the movement reference numbers and duty demand notes attached to the entries that actually carried the measure.

Every good piece of trade news arrives with a condition attached. The obvious flip is production: if European producers start making this glass again, or buy assets and restart lines, the industry limb reappears and a fresh complaint becomes thinkable. It would take time, since a new investigation has to be initiated and concluded before any provisional duty could even theoretically apply, and nothing it produces reaches backwards past its own initiation. Chinese pricing over the next two quarters decides how likely that becomes. If share climbs back toward double digits on prices that drop by most of the duty saving, the political arithmetic looks different by the middle of next year. There is a flip closer to home as well: importers who collect windfall sums and pass nothing to installers and developers should not be surprised to find their own pricing behaviour quoted in a future complaint.

Do the arithmetic, with the assumptions written down where everyone can argue with them. Assume a European importer cleared Chinese solar glass with total customs value of one million euro between 22 July 2025 and 21 October 2026, declared under one of the four covered TARIC codes, and assume it was not granted an individual rate, so the residual 67.1 per cent applied. That is 671,000 euro of anti-dumping duty. Assume further that the countervailing duty applied to those same entries was the upper end of the range, 17.1 per cent, giving another 171,000 euro. Total duty paid, 842,000 euro, on a million euro of declared value, and in principle all of it is within the look-back window.

Run the same base at the other corners of the ranges and you get a sense of how much this varies between companies that think they are in identical positions. That same million euro cleared by a company holding the lowest individual rate of 17.5 per cent, paired with a 3.2 per cent countervailing rate, paid 207,000 euro. Cleared by a company carrying the top individual rate of 75.4 per cent with the top countervailing rate, it paid 925,000 euro. The distance between those two is more than four times the lower figure, so any internal estimate built on a headline percentage is going to be wrong for somebody in the room, and probably wrong by a lot.

Interest is where finance will start asking questions, and they are right to. Take the residual case above, 842,000 euro, assume the payments were spread evenly so the average euro left seven and a half months ago, and assume a cost of funds of five per cent a year: roughly 26,000 euro of carrying cost is gone, which nobody sends you, and more if the cash was funded on expensive working capital lines.

The recurring saving usually beats the one-off recovery. Assume volumes hold at a hundred thousand euro of customs value a month through the coming year. At the residual rate plus the top countervailing rate, stopping the duty saves 84,200 euro a month, a million euro over twelve months, with no filing attached. It hits January immediately and needs nobody's permission. It also creates a question procurement has to answer before the saving erodes on its own: how much of that monthly amount do you keep when the next purchase order is negotiated?

Structure for that negotiation comes from numbers already in this Regulation. Assume the glass arrived with duty fully loaded into the price, so one euro of pre-duty glass cost the importer 1.842 euro landed. Remove the duty and it costs 1.000 euro: the duty was about 45.7 per cent of what was actually being paid. Assume glass accounts for eight per cent of the landed cost of a finished module, an assumption, not an industry statistic. Pass the whole relief through to the module buyer and module landed cost falls by about 3.7 per cent; split it evenly and it falls by roughly 1.8 per cent. Those 1.8 percentage points are what the next two negotiating rounds are really about, and over a year they beat most freight savings anyone will bring you this quarter.

Now to the documents, because this is where files get lost. Start at classification. The ten-digit code on your entry is the only thing that decides whether your shipment is in or out of this repeal, so pull the entry register before doing anything else. What changes on 22 October is not the code itself. Your goods classify exactly where they classified yesterday morning. What changes is the measures layer that hangs off that code inside customs IT: the TARIC record for those four codes stops returning an anti-dumping and a countervailing measure, and the duty calculation that used to fire off it stops firing.

The fields your declarant fills in downstream of it change as a consequence, notably the additional information block that descends from the old single administrative document and is still known in most member states as box 44, where the codes and authorisations evidencing a measure were being entered. Those inputs become unnecessary, and if your software autocodes them, someone should switch that automation off rather than leaving it to produce entries with orphan references.

Which brings up the practical failure mode of day one. National platforms do not all take TARIC updates at the same speed, and the declarant's own rate tables do not update themselves. It is entirely possible for entries lodged on 22 or 23 October to still calculate a duty that no longer exists in law, and to debit it from a company account. So do two things today. Ask your declarant, in writing, to confirm which version of the TARIC measure table their system is running and when it was last refreshed. Then, for the entries you are lodging this week, verify the duty line on each entry before it is accepted, not after it is debited. If you do catch a wrong debit, do not let it go through quietly while someone sorts it out later; later claims are always harder than contemporaneous notes, and the MRN with a written reservation on file is worth more than an email thread three months from now.

While these duties were live they formed part of the customs debt, and for many importers that debt was covered by a comprehensive guarantee or deferment rather than paid entry by entry. A large duty inflates the guarantee reference amount you were asked to put up, and that guarantee consumes either a bank line or cash collateral. With the duty gone, the reference amount is stale and it will not adjust itself. Someone in treasury should be asking your customs authority to reassess it against the lower expected duty profile, citing this Regulation explicitly. Whether the figure comes down depends on your own import volumes, but leaving a guarantee sized for an 84.2 per cent duty in place while you finance a quarter of inventory is paying for nothing.

Then build the reclaim file as if the officer reading it has never met you. For each entry in scope you want the movement reference number, the accepted declaration, the duty demand note or national equivalent, and proof the amount was settled, matched line by line to that demand note. On top you want the commercial documents: invoice and packing list tying goods to the declared description, whatever evidence of Chinese origin you hold, and the customs value declaration supporting the base used. If entries were made under indirect representation, or where the payer was not the importer of record, document that chain before filing, because the first question any authority asks is whether you are the right applicant.

Two edge cases belong on the checklist before resources are committed. Similar goods declared under codes near 7007 but outside the four listed lines generate no refund, and if you suspect past entries were miscoded, that is a separate matter with its own exposure to a post-clearance check, worth raising with your adviser. Goods entered under a suspensive procedure, inward processing included, may never have carried the duty at all, so there is nothing to reclaim on those lines even though the cargo was identical. Sorting in-scope lines from out-of-scope lines is dull work, and it decides whether a claim takes three weeks or six months.

One further item sits in another department and gets missed almost every time: import VAT is charged on a base that includes customs duty. If that duty is now being repaid, part of the VAT collected on it may have been over-collected, and correcting it is a national rather than a Union question, often dependent on whether that import VAT was deducted at all. Put it in the same file today, because nobody will remember it in a week.

Before this afternoon there is a realistic list. Pull the entry register covering 22 July 2025 to 21 October 2026 and filter for the four TARIC codes, which most companies can do from their broker portal exports and their declarant can certainly do. Compute duty actually paid per entry from demand notes rather than from an estimated percentage, because estimates built on the residual rate miss by the distance between 20.7 and 92.5 per cent. Establish who the payer was in each period, including any change of entity, restructuring or change of declarant inside the window, since reorganisations are the most common reason a claim stalls. Then send one written note to whoever owns the customs guarantee and national VAT filings, and one to whoever owns supplier pricing, telling them to concede nothing on the next purchase order until the duty position is clear.

The traps are conventional and people still fall into them. Do not open a price conversation with a Chinese supplier before you know whether the duty sat inside the price you were paying, because a DDP arrangement may leave nothing to trade away. Do not rush a large replacement order; non-Chinese sources will argue their premium deserves reopening, and whether they get that chance is your call. Expect the same inspection rate, because risk profiles are not rebuilt overnight and identical containers keep getting pulled. Nothing here is permanent: the findings were about who was producing what, and production changes without consulting your supply chain.

↑ Back to the key points

— By Derek Xu