The OECD says global steel excess capacity will reach 721 million tonnes by 2027, with additions outrunning closures three to one, as Europe’s melt and pour rule takes effect on October 1 and turns origin documentation into a condition of market access
PARIS, SEPTEMBER 25, 2026
Global steel excess capacity is projected to rise from 601 million metric tonnes in 2024 to 721 million tonnes by 2027, according to an OECD report titled “Navigating the Steel Transition amid Global Excess Capacity,” published this week and reported on Friday by SteelOrbis and EUROMETAL.
The organisation found that over 2025 to 2027, closures of blast furnace and basic oxygen furnace capacity totalling roughly 25.8 million tonnes will be vastly outpaced by 62.1 million tonnes of new capacity additions, concentrated predominantly in China and India. Only 27 percent of identified closures involve corresponding electric arc furnace installations, which the OECD reads as evidence that market conditions rather than environmental priorities are driving plant retirements.
The report warns that overcapacity “depresses steel prices, squeezes profit margins and increases financing costs,” while government subsidies supporting inefficient producers delay the retirement of high emission facilities. As of the second quarter of 2025, announced low emission steelmaking projects totalling 15.5 million tonnes had been suspended or postponed because of excess capacity, elevated energy costs and regulatory uncertainty. The cost premium for low emission steel over conventional steel in Europe rose from 17 percent to 25 percent between 2022 and 2025, as falling conventional prices widened the gap.
The OECD’s policy conclusion is that transition policy must combine investment incentives with mandatory capacity closures and subsidy removal, or decarbonisation will amount to a technological reshuffling of global oversupply rather than a reduction in it.
WHY 721 MILLION TONNES IS THE NUMBER THAT MATTERS
Excess capacity is the quantity of steelmaking capability that exists beyond what the world is buying. At 721 million tonnes it approaches a scale several times the annual output of the entire European Union steel industry, and it sits in a sector where fixed costs are enormous and marginal costs of continued operation are low.
That combination produces a specific and predictable behaviour. A plant that cannot cover its full costs will still run if it can cover its variable costs, because running loses less money than stopping. Output therefore continues past the point where prices justify it, and the surplus goes looking for a market. It finds the markets that are most open, which is why open economies experience overcapacity elsewhere as an import surge at home.
This is the mechanism that generates trade measures. Every major steel producing jurisdiction has responded to the same underlying arithmetic with some combination of safeguards, anti dumping duties, countervailing duties and quotas, and the OECD’s projection says the pressure generating those measures will intensify for at least three more years.
The distributional detail matters too. New capacity concentrated in China and India, against closures concentrated in Europe, describes a structural transfer of production rather than a cyclical downturn. Europe is not waiting out a trough. It is watching capacity leave permanently while capacity is added elsewhere, and its policy response has been designed accordingly.
EUROPE’S ANSWER TAKES EFFECT NEXT WEEK
The report lands six days before the most operationally significant deadline in European steel trade this year.
The EU replaced its longstanding steel safeguard, which expired on June 30, 2026, with a permanent measure under Regulation (EU) 2026/1384, in force from July 1, 2026. The new regime sets annual tariff rate quotas totalling 18,345,922 tonnes, of which roughly 9.15 million tonnes are reserved for free trade agreement partners and the remainder allocated on a most favoured nation basis. Imports beyond quota face an out of quota duty of 50 percent ad valorem, double the previous 25 percent, and that duty stacks on top of any anti dumping or countervailing duties already applying to the same goods.
Country specific quotas go to origins with at least a 5 percent average import share over 2022 to 2024. China holds allocations across 22 sub categories but, unlike most origins, has no fallback access to residual quota pools. Ukraine receives more favourable treatment than other free trade agreement partners.
From October 1, 2026, importers must provide verifiable evidence, such as a mill test certificate, identifying the country where the raw steel was initially produced, which is to say the country of melt and pour rather than the country of last processing. The Commission specified the evidence requirements on August 31, 2026, and must assess by June 30, 2028 whether melt and pour should become the primary basis for quota access.
Country level allocations under the regime were finalised earlier this year. Analysis published by Shanghai Metals Market put South Korea at 101,884 tonnes for stainless cold rolled coil, ahead of Turkey at 69,038 tonnes, Taiwan at 52,985 tonnes and South Africa at 52,607 tonnes, with stainless hot rolled coil allocations led by Indonesia at 35,843 tonnes, India at 26,019 tonnes and South Korea at 20,735 tonnes.
WHAT MELT AND POUR ACTUALLY CHANGES
The melt and pour requirement is the part of the regime that will generate the most disruption next week, and it is worth being precise about why.
Conventional customs origin follows the last substantial transformation. Under that rule, slab melted in one country and rolled into coil in another can acquire the origin of the rolling country, and quota access follows accordingly. Melt and pour ignores the rolling step entirely and asks only where the liquid steel was first cast.
The practical effect is to close the most widely used route around country specific quotas. Trade patterns that developed around processing in intermediate jurisdictions, where material from a constrained origin is rolled in a less constrained one and enters under the latter’s allocation, stop working on October 1.
The compliance burden is real and it falls on the importer. A mill test certificate identifying the melt location must be available at entry, which means it must be obtained from the mill, travel with the material through however many intermediaries handle it, and match the goods presented. For material bought on the spot market, through traders, or from stockholders holding mixed origin inventory, that chain of documentation frequently does not exist today.
The exposure is not a penalty in the usual sense. An importer who cannot evidence melt and pour origin loses access to the quota that the shipment was priced against, and the alternative is the 50 percent out of quota duty. On a container of coil, that is not a compliance inconvenience. It is the difference between a profitable trade and a substantial loss.
THE CAPACITY ARITHMETIC IN DETAIL
The headline figure conceals a more revealing structure, and the structure is where the policy implications live.
Consider the closure side first. The OECD identifies roughly 25.8 million tonnes of blast furnace and basic oxygen furnace capacity closing between 2025 and 2027. Only 27 percent of those closures are paired with corresponding electric arc furnace installations. In other words, for every four integrated plants that shut, roughly one is being replaced with a lower emission alternative and three are simply going away.
That ratio is the quiet finding in the report. A closure paired with an electric arc furnace is a transition. A closure with nothing behind it is a retreat. The OECD’s reading, that market conditions rather than environmental priorities are driving retirements, follows directly from that ratio: firms are not choosing to replace old capacity with clean capacity, they are being forced out and not returning.
Now the addition side. Some 62.1 million tonnes of new capacity arrives over the same period, concentrated in China and India. Net, the world adds roughly 36 million tonnes of capacity over three years into a market the OECD expects to grow more slowly than that.
The second order effect concerns the type of capacity being added. New capacity built today is more efficient than the capacity it competes against, which means it has lower marginal costs, which means it runs harder in a downturn and pushes older capacity out faster. Overcapacity is therefore self reinforcing in the short run: the new tonnes displace old tonnes rather than meeting new demand, and the displaced producers keep running as long as they cover variable costs.
Against that, the 15.5 million tonnes of suspended or postponed low emission projects represents deferred clean capacity, and the green premium widening from 17 percent to 25 percent in Europe between 2022 and 2025 explains why. Clean steel competes against conventional steel priced at the marginal cost of a plant that is losing money and running anyway. That is not a competition clean steel wins on price, and subsidy is the only mechanism that closes the gap.
WHY THE GLOBAL FORUM HAS NOT SOLVED THIS
The institutional response to steel overcapacity has existed since 2016 in the form of the Global Forum on Steel Excess Capacity, and the persistence of the problem a decade later is worth explaining rather than simply lamenting.
The forum’s difficulty is structural. Capacity closure is politically expensive and locally concentrated, while the benefits of closure are diffuse and accrue mostly to producers in other countries. A government that closes a plant absorbs the entire cost in jobs and regional decline, and shares the benefit of firmer prices with every competitor worldwide. That is a textbook collective action problem, and voluntary forums do not solve those.
The second difficulty is measurement. Capacity is easier to announce than to verify, and a plant that is idled rather than dismantled can restart when prices recover. Commitments to close capacity have repeatedly translated into idled capacity that reappeared, which corrodes trust in the next round of commitments.
The consequence is that jurisdictions default to unilateral defence. Quotas, safeguards and duties do not require anyone else’s cooperation, and they deliver domestic benefit quickly. They also displace the problem rather than resolving it, which is the pattern the OECD report describes and the pattern importers must plan around.
THE CARBON LAYER ON TOP
Trade defence is no longer the only administered cost attaching to imported steel in Europe, and firms modelling landed cost need both layers.
The EU’s carbon border adjustment mechanism applies to steel among other covered goods, and its compliance phase began in 2026. It operates on a different logic from a tariff: the charge reflects embedded emissions in the imported good, adjusted for any carbon price already paid in the country of production, rather than the price or origin of the product.
For an importer, however, the practical effect is arithmetically similar to a duty, and it stacks with the quota regime rather than replacing it. A shipment can face a most favoured nation tariff, a quota constraint with a 50 percent out of quota rate, an anti dumping or countervailing duty where one applies to the origin, and a carbon charge reflecting its emissions intensity. Landed cost models that capture only the first of these will be wrong by margins that matter.
The interaction with the OECD’s findings is direct. Producers with high emissions intensity, frequently the same subsidised integrated capacity the report identifies as the source of the surplus, face the largest carbon charge on entry to Europe. That is the intended design. Whether it accelerates cleaner production abroad or simply redirects dirty tonnes to markets without a carbon border charge is the open question, and the answer so far has been more of the latter than the former.
STAKEHOLDER REACTIONS
European producers have argued consistently that overcapacity abroad, rather than any failure of competitiveness at home, explains the sector’s condition, and the OECD’s figures support that reading in broad terms.
The industry mood at this week’s Kallanish Global Flat Steel 2026 conference in Istanbul was described in similar terms. Metin Tayfun Iseri, chairman of the Turkish Flat Steel Import, Export and Industry Association and a board member of Colakoglu Metalurji, told the opening keynote on Wednesday that worldwide excess production capacity has left steelmakers unable to defend their margins, with protectionist policies and geopolitical instability compounding the difficulty, according to EUROMETAL.
The consolidation continues on the ground. HKM permanently shut down blast furnace A in Duisburg on September 24 after 53 years of operation, according to EUROMETAL, and German steel sales fell 18 percent month on month in August on persistently weak demand. ArcelorMittal said its Ukrainian steel plant cannot restart safely. Each of these is a local decision, and together they are the European half of the OECD’s global picture.
Exporting countries read the same data differently. For India, where capacity is being added to serve domestic demand that is growing faster than anywhere else of comparable size, being named as a source of global excess alongside China is a characterisation its industry disputes. Capacity built to serve a growing home market is not, on its own, an export threat, though the OECD’s point is that capacity built anywhere can become an export when domestic demand disappoints.
Turkish producers sit in the most exposed position of all, holding significant European quota, operating a rolling sector that historically used imported slab, and facing a melt and pour rule that puts the sourcing of that slab under direct scrutiny.
ECONOMIC IMPACT
Three effects follow from the combination of persistent overcapacity and tightening trade barriers.
The first is price divergence by jurisdiction. When a large market imposes hard quotas with a punitive out of quota rate, prices inside that market decouple from world prices. European domestic hot rolled coil prices saw a slight increase this week with producers remaining bullish despite limited demand, according to EUROMETAL, which is behaviour that makes sense only in a market where supply is administratively constrained rather than demand driven.
The second is displacement. Tonnes that cannot enter Europe do not evaporate. They move to markets with weaker defences, which then experience their own import surges and impose their own measures. This is the cascade that has produced a steady rise in trade remedy actions in steel worldwide, and the OECD’s capacity projection implies the cascade continues.
The third is the decarbonisation cost the OECD highlights directly. Low carbon steel projects totalling 15.5 million tonnes have been suspended or postponed, and the green premium in Europe widened from 17 percent to 25 percent between 2022 and 2025. Cheap conventional steel makes expensive clean steel harder to justify, and every deferred project pushes the sector’s emissions trajectory further from its targets. Trade measures that support domestic prices are, in this respect, decarbonisation measures, which is a large part of why Europe is willing to defend them.
IMPLICATIONS FOR IMPORTERS, EXPORTERS AND SUPPLY CHAINS
For anyone buying, selling or moving steel into the European Union, the next seven days matter more than the next seven months.
Verify melt and pour documentation for every shipment arriving on or after October 1. That means obtaining mill test certificates that identify the country of initial casting, confirming they accompany the material through every intermediary, and checking that inventory already in transit is covered. Material bought from stockholders or traders is the highest risk category, because mixed origin inventory frequently cannot be traced to a single melt location.
Reprice contracts that assumed quota access through a processing origin. If the commercial logic of a supply arrangement depended on rolling in a jurisdiction with available quota, that logic expires next week. The out of quota duty of 50 percent, stacked on any existing anti dumping or countervailing duty, is severe enough that the trade simply does not work.
Monitor quota utilisation actively rather than at quarter end. Country specific allocations exhaust at different rates, and an importer who discovers exhaustion at the border discovers it too late. Quota consumption data is published, and building it into purchasing decisions is the difference between planning and reacting.
Understand the stacking. The 50 percent out of quota rate is additional to trade defence duties already in force on the same product and origin. For goods subject to existing anti dumping measures, the combined effective rate can exceed any plausible margin on the trade.
Watch the 2028 review. The Commission must assess by June 30, 2028 whether melt and pour becomes the primary basis for quota access rather than a documentation requirement layered on conventional origin. If it does, the supply chain restructuring that firms are doing tactically this autumn becomes permanent.
For exporters outside the EU, quota allocation is now a strategic asset. Origins with allocation and the documentation systems to prove melt and pour will capture demand that others cannot serve. Origins without either will find their European business migrating regardless of price.
For buyers in markets that do not currently maintain steel safeguards, prepare for displaced volume. The material that cannot enter Europe will arrive somewhere, at prices that will look attractive and will eventually attract a trade remedy petition from domestic producers. Purchasing strategies built on those prices have a limited life.
OUTLOOK
The OECD’s projection through 2027 describes a problem that trade measures manage rather than solve. Quotas and duties redistribute where surplus steel goes. They do not close the plants producing it, and the report is explicit that only mandatory capacity closures combined with subsidy removal would do that.
That is a coordination problem of the kind the international system currently handles poorly. The Global Forum on Steel Excess Capacity exists precisely for it, and has not delivered closures at the scale the arithmetic requires. In its absence, each jurisdiction defends its own market, the surplus rotates, and the measures multiply.
For companies, the planning assumption should be that steel trade becomes more administered, not less, through at least 2028. Quota systems, origin verification, carbon border charges and trade defence duties are converging into a single compliance regime in which the ability to document where material came from and how it was made determines whether it can be sold at all.
The firms that treat that as a procurement problem will pay for it. The firms that treat it as a documentation and systems problem, and solve it early, will find themselves holding capacity that their competitors cannot legally supply.