Zombie Steel Mills: Why State Aid Delays the Inevitable

Executive Summary

Across Europe, a recurring pattern plays out in struggling steel regions: a mill loses money, a government provides support, the mill survives another cycle, and the process repeats. Each intervention is presented as temporary. Few are. The result is a generation of zombie steel capacity — facilities too subsidised to close but too uncompetitive to thrive — which suppresses prices, penalises efficient producers, and transfers enormous costs to taxpayers and downstream industries.

The OECD has consistently found that government support to the steel sector distorts competition, sustains uneconomic capacity, and delays the restructuring that markets require. Its 2025 analysis identified over $100 billion in annual government support to steel industries globally, with a significant share flowing to facilities that would not survive on commercial terms alone.¹

The central argument of this article is not that state aid to steel is always wrong. It is that state aid directed at keeping unviable mills operating is almost always the worst available option — worse for workers, worse for communities, and worse for the steel market. The better use of public funds is managed closure: compensating asset owners, remediating sites, and above all, giving workers the income support, retraining, and time they need to transition to new employment.

Romania's long products sector provides a stark illustration of what happens in the absence of structured support for closure. Eight mills with over five million tonnes of capacity have effectively disappeared over thirty years — not through orderly transition, but through asset stripping, insolvency, and prolonged decline. Workers and communities bore the consequences. The market received no benefit from the delay.

Defining the Zombie Mill

The term "zombie" has a precise economic meaning in this context. A zombie steel mill is a facility that cannot cover its full costs of production — including capital maintenance, environmental compliance, and a market rate of return — without some form of government support. It is not merely loss-making in a cyclical downturn; it is structurally unviable at any realistic point in the market cycle.

Zombie mills share common characteristics. They tend to operate older equipment with high fixed costs and poor energy efficiency, contributing to the persistent overcapacity that depresses profit margins across the entire sector. Their product mix typically concentrates in lower-value commodity grades where competition from lower-cost producers is most intense. They carry accumulated underinvestment, deferred maintenance, and pension obligations that weigh heavily on their cost base. And they operate in regions where closure carries severe political consequences, giving management and owners implicit leverage over governments reluctant to act.

The support these mills receive takes many forms. Direct subsidies and operating grants are the most visible. But zombie mills also benefit from preferential energy pricing, relaxed environmental enforcement, loan guarantees, delayed pension contributions, and repeated restructuring packages that write down debt without addressing underlying cost structures. When one package fails — as it almost always does — the political logic that justified the original intervention typically justifies the next one. The scale of this support across Europe is documented in our European Steel State Aid Database, which covers three decades of government intervention from 1994 to 2024.

The OECD's research adds a particularly important finding to this picture: in economies where production subsidies are most heavily used, support tends to intensify precisely during downturns — switching on when facilities become unviable, and thereby preventing the market adjustment that viability would require. The result is that subsidies do not merely sustain zombie mills; they actively insulate them from the commercial pressure that might otherwise force restructuring.¹

The Political Economy of Closure: Why Governments Always Blink

Understanding why zombie mills persist requires understanding the political economy of steel closure, not simply its economics. The economics of closure are usually clear enough: a facility losing €50–100 million per year in a region with few alternative employment prospects should close. The politics are far more complicated.

Steel employment is concentrated. A mill employing 3,000 workers is typically the largest single employer in its sub-region, with a further multiplier of two to three times in indirect employment in suppliers, contractors, and local services. The announcement of closure is not an abstract economic event; it is a crisis for a specific town or district with a specific political representative. Governments that permit closure face immediate, vocal, and well-organised opposition. The costs of subsidy are dispersed across millions of taxpayers; the costs of closure are concentrated in a single constituency.

This asymmetry systematically biases political decisions towards intervention, regardless of the economic merits. The pattern is consistent across member states and across political systems. A German state government, a French ministry, an Italian regional authority — all face the same calculation, and all tend to reach the same conclusion. Short-term political relief is prioritised over long-term economic rationality.

It is important to note that this does not reflect bad faith on the part of governments. The officials involved frequently understand the economic analysis. What they face is a genuine tension between economic efficiency and social obligation to workers and communities that have built their lives around an industry. The argument here is not that this tension is imaginary, but that prolonging unviable operations is a poor way to resolve it — and that structured closure support is a better answer to the same legitimate concerns.

The absence of meaningful conditions on support compounds the problem. When the UK government intervened at Scunthorpe in April 2025 — passing emergency legislation in a single day to prevent closure of the last two British blast furnaces — the intervention carried no decarbonisation requirements and no commercial viability milestones. The plant was kept running, but on no particular terms. This pattern is more rule than exception: aid is approved under political pressure, conditions are either absent or loosely drawn, and the result is support that delays the next crisis rather than prevents it.

The European Commission occupies an inherently difficult position: it is both the guardian of competition rules and the body to which member states apply for exceptions under Article 107 of the TFEU. This tension is structural rather than a failure of will — but it does mean that a framework formally prohibiting subsidy has in practice approved significant volumes of support to steel facilities under various permitted exemptions.

Case Study: Ilva/Taranto — Europe's Most Expensive Steel Problem

The Ilva steelworks at Taranto in southern Italy has a nameplate capacity of approximately 8–10 million tonnes per year, making it the largest integrated steel plant in Europe by installed capacity. It is also the most instructive example of how zombie mill dynamics play out at scale over decades. In practice, the plant produced less than two million tonnes in 2024 — a fifth of its nameplate capacity — whilst continuing to absorb public funds.

Ilva entered state ownership in the 1990s as part of Italy's broader steel nationalisation, and was privatised to the Riva Group in 1995. [For full chronology, see our Riva history page]. It subsequently became the subject of a criminal investigation into environmental violations, leading to its seizure by Italian authorities in 2012. Since that point, the facility has passed through a succession of administrators, restructuring plans, ownership changes, and government rescue packages, culminating in ArcelorMittal's acquisition in 2018 and its subsequent exit from the management agreement in 2019.

The Italian government has committed an estimated €4–5 billion to the facility since 2012 through a combination of administrator funding, environmental remediation commitments, and restructuring support. The plant now operates under the ownership of a state-controlled entity, Acciaierie d'Italia, whilst a sale process to a new private operator is under way. Meaningful conditionality — requiring the incoming operator to close coal-fired capacity and install electric furnaces — has been attached to the most recent support package. That this conditionality arrived after more than a decade of largely unconditional intervention illustrates the broader structural problem.

The environmental situation at Taranto compounds the economic analysis. The plant sits adjacent to a city of 200,000 people and has been associated with elevated rates of certain health conditions in the surrounding population. Environmental remediation costs are estimated in the billions of euros. The community bears severe costs from the mill's continued operation; it would also bear severe costs from abrupt closure without transition support. This is precisely the situation that structured, adequately-funded managed closure is designed to address. Yet it has never been resolved, because successive governments have preferred to keep the mill nominally operational rather than confront the full cost of closure.

Taranto is an extreme case, but it illustrates the general pattern: each intervention defers rather than resolves the underlying problem, and the accumulated cost of deferral typically exceeds what managed closure would have cost at an earlier stage.

The Romanian Long Products Sector: A Case for Closure Support That Never Came

Taranto illustrates what happens when governments over-intervene to keep unviable mills alive. Eastern Europe has a different but equally instructive story — of mills that received no structured support at all, and paid a heavy price for that absence. Romania's long products sector illustrates this opposite failure: governments that under-intervened, providing no structured closure support, and leaving workers and communities to bear the consequences alone.

Thirty years ago, Romania operated eight carbon steel long product plants with a combined capacity of over five million tonnes per year. Today, only one remains operational. The story of how the other seven disappeared is not a story of orderly market adjustment. It is a story of asset stripping, insolvency proceedings, ownership changes that changed nothing, and prolonged decline that left workers, creditors, and local economies worse off than an early managed closure would have done.

Five of the eight plants — Ductil Steel Buzău, Ductil Steel Oțelu Roșu, Industria Sârmei Câmpia Turzii, Laminorul Brăila, and the Târgoviște Special Steel Plant — passed under the ownership of Russian steelmaker Mechel. Mechel subsequently sold the group to Invest Nikarom for a nominal sum in early 2013. This transaction is characteristic of the inappropriate ownership changes discussed elsewhere: a distressed portfolio sold for essentially nothing to a buyer without the capital, expertise, or commitment to run the assets. All five entered insolvency.

The exception — and an instructive one — is the Târgoviște plant, which was subsequently acquired from insolvency proceedings by AFV Beltrame Group in March 2022 for approximately €38 million. Beltrame, one of Europe's leading producers of steel bars and special steels, committed to investing around €100 million in modernisation of the steelworks and rolling mills, with a further €300 million planned for a new eco-friendly rebar and wire rod facility in Romania.

The Beltrame operation in Romania has attracted commercial financing, including at the group's Donalam rolling mill in Călărași. This is precisely the model that the Romanian long products sector needed more of: commercially grounded acquisition by an operator with relevant expertise, supported by financing conditioned on genuine investment commitments — not the nominal ownership changes that characterised most of the sector's decline.

The other plants followed different paths to the same destination. Laminorul Focșani, with 800,000 tonnes of rebar and rod capacity, permanently ceased operations as early as 1997, with the loss of over 1,000 jobs. Lamdro Turnu-Severin, with 550,000 tonnes of rebar capacity, closed in 2020 under Max Aicher ownership. ArcelorMittal Hunedoara, with 1.4 million tonnes of light long product capacity, suspended operations on multiple occasions over the past decade before closing indefinitely in September 2025.

The exhibit below summarises the decline:

The critical point about the Romanian case is not that these closures were wrong. Most were economically justified, and the European long products market did not need this capacity. The critical point is how they happened.

There was no structured programme of worker transition. There was no site remediation funding. There was no regional economic development support. In most cases, workers received minimal or no severance. Communities that had built their economic and social life around these plants were left to manage the consequences with little external assistance.

Asset stripping was a recurring problem. Once a facility entered financial difficulty, equipment was often sold or removed rather than maintained, destroying any residual value and making restart progressively less viable. By the time formal insolvency proceedings concluded, little of value remained.

Some of these plants persisted in a state of intermittent or suspended production for over 25 years — not because they were viable, but because the legal and financial mechanisms for orderly closure were absent or dysfunctional. This prolonged twilight served nobody. Workers could not plan their futures. Creditors could not recover assets. Communities could not attract alternative investment to sites that remained nominally operational but economically inert.

This is the negative case for structured closure support. Not that governments should have subsidised these mills to keep them running — that would have been the wrong use of public money. But that genuine transition funding, directed at worker support, site remediation, and regional development, could have produced substantially better outcomes than the prolonged decay that actually occurred.

There is, however, a critical legal constraint that is rarely acknowledged in policy discussions. Under current EU state aid rules, direct closure aid to steel companies — compensation to owners for permanently ceasing production — is prohibited. When the ECSC Treaty expired in 2002, the European Commission confirmed that rescue aid, restructuring aid, and direct closure aid to steel firms are all incompatible with the internal market.³

This is a fundamental difference from the coal sector, where a separate regulatory framework explicitly permitted closure compensation — the mechanism that made the German coal phase-out financially viable for mining companies.⁴ For steel, only social payments to redundant workers are permitted, subject to strict conditions. Site remediation and asset write-down compensation to companies are not.

This gap in the regulatory framework is a direct cause of the disorderly closures seen across Romania and elsewhere. Without the ability to fund structured exit, both governments and companies face a stark choice between continued operation and disorderly collapse.

The Better Use of State Aid: Funding Closure, Not Continuation

Where State Aid to Steel Is — and Is Not — Permitted

The argument here is not that all state support to steel is illegitimate. There are circumstances where public funds are appropriately deployed in the sector:

• Funding genuine decarbonisation investment where market returns are insufficient — supported through mechanisms including IPCEI and, since June 2025, the CISAF, which replaced the earlier Temporary Crisis and Transition Framework
• Supporting worker transition during plant closure
• Remediating contaminated sites that private owners cannot afford to clean
• Preserving the small number of genuinely strategic speciality capabilities that the market alone would not sustain

The transition towards scrap-based EAF production offers a more viable long-term path for many European producers than prolonged subsidy of ageing integrated capacity — and the OECD's position reflects this. Its analysis is more precise than a general opposition to state support.

Its Steel Outlook 2025 identifies three specific targets for reform: market-distorting production subsidies, below-market financing that allows uneconomic facilities to borrow at artificially low rates, and capacity investment driven by government support rather than commercial demand. Its research shows that in economies where these instruments are most heavily used, subsidies during downturns display counter-cyclical characteristics — intensifying precisely when market pressure would otherwise force restructuring. This is the mechanism that produces zombie mills: support that switches on when facilities become unviable, preventing the market adjustment that viability requires.

At the same time, the OECD explicitly supports government investment in decarbonisation, provided it does not add to excess capacity or confer undue competitive advantage. The distinction the OECD draws — between support that sustains production and support that transforms it — is the same distinction that should govern European state aid policy.¹

The European Commission's own restructuring aid rules require that any support to a firm in difficulty must lead to genuine long-term viability — not simply defer the next crisis. A facility that cannot demonstrate a credible path to commercial sustainability should not qualify for aid. In practice, however, the viability requirement has often been interpreted loosely, and the compensatory measures required to offset market distortion have not always been sufficient to address underlying overcapacity. The Commission is currently revising its Rescue and Restructuring Aid Guidelines — a process that offers an opportunity to tighten these requirements and bring the steel sector's treatment into closer alignment with the principles the guidelines are meant to enforce.

Yet the framework as it stands rewards the wrong behaviour. Reforming state aid policy requires addressing the structural incentive problem at its root. Governments intervene because the political costs of closure are immediate and concentrated, whilst the economic costs of subsidy are diffuse and deferred. Making those costs more visible would require:

• Rigorous sunset clauses that require explicit renewal rather than allowing support to roll forward indefinitely
• Mandatory reporting of total support per tonne of production
• Systematic evaluation of whether supported facilities have achieved the viability milestones attached to aid approvals.

It also requires making managed closure genuinely available as an option — and here the current framework presents a specific obstacle. The German coal phase-out deployed €40 billion over three decades, including direct closure compensation to mining companies, worker transition payments of €5 billion, and €30 billion in regional development funding.² This was possible because coal benefited from a dedicated regulatory framework permitting closure aid. Steel does not. Under the rules established after the ECSC Treaty expired in 2002, direct closure aid to steel companies is prohibited.³ The German model cannot simply be replicated for steel without a change in the EU state aid framework.

What is currently permitted is narrower but still significant: social payments to redundant workers, environmental remediation funded through regional or cohesion instruments, and transition support through the EGF^e and ESF+^f. These instruments have been underused in steel closure situations — Romania being the clearest example. For steel, an effective transition programme must work within these constraints. The closure aid prohibition now represents a significant anomaly: coal companies could receive structured exit compensation; steel companies facing identical social and environmental challenges cannot.

Conclusions

The zombie mill problem in European steel is not a mystery. Its causes are well understood: high exit barriers, concentrated employment, political short-termism, and a state aid framework that bends under member state pressure. Its consequences are equally clear: sustained overcapacity, depressed prices, penalised efficient producers, wasted public funds, and an industry structure that remains far more fragmented and competitive than commercial logic would produce — making rational pricing and long-term investment harder for everyone in the market.

The Romanian long products sector provides a particularly instructive lesson, precisely because it represents the failure of the alternative to subsidy dependency. These mills did not receive the sustained production support that has kept Taranto nominally alive. They collapsed — but they collapsed badly, over decades, without the worker transition support, site remediation funding, or regional development investment that would have made the process manageable. The result was neither the market efficiency of clean closure nor the social protection of managed transition. It was prolonged decay — a lingering death that served neither the market nor the communities it left behind.

The core conclusion is straightforward: state aid to steel should be redirected from sustaining production to funding closure. This is not an argument for abandoning workers or communities — it is an argument for helping them more effectively. Transition support, retraining, early retirement bridges, and regional economic development produce better outcomes than keeping an unviable facility running for a few more years until the next crisis comes along.

This conclusion is not, in fact, controversial at the analytical level — the OECD's own research, and the Commission's own state aid guidelines, point in the same direction. The obstacle is not a lack of understanding. It is a structural incentive problem that rewards short-term intervention over long-term reform.

The Taranto and Scunthorpe cases point to a second reform of equal importance: the conditionality attached to any continuing production support. The Commission's €390 million rescue loan to AdI in February 2026 represented a partial step forward — the new operator at Taranto is now required to commit to shutting down coal-fired operations and building electric furnace capacity as a condition of support. But this conditionality arrived after an estimated €4–5 billion had already been spent over more than a decade with no such requirements attached. In the United Kingdom, the emergency intervention at Scunthorpe under the Steel Industry (Special Measures) Act 2025 carried no meaningful decarbonisation conditions at all; it was crisis management, not industrial strategy.⁵

The lesson is not that conditionality is impossible — it is that it must be established at the outset, not retrofitted after the money has been committed. Aid to facilities that are not yet at the point of closure should be conditional on a credible, independently verified pathway to commercial viability: specific capacity reduction milestones, binding decarbonisation commitments with timelines, and sunset clauses that trigger automatic review if milestones are missed. Aid that carries no such conditions is not restructuring support — it is deferred closure at public expense.

There is one specific policy reform without which even the best-intentioned transition programme will struggle. The EU state aid framework should be amended to permit direct closure aid to steel companies, equivalent to the framework that enabled structured coal sector exit. The current prohibition — under which steel companies cannot receive compensation for permanently ceasing production, whilst coal companies could — creates a perverse incentive to continue operating rather than close.

Removing this anomaly would not require abandoning the general principle that production subsidies to steel are prohibited. It would require only recognising one distinction: that aid to facilitate exit from production is categorically different from aid to sustain it. The social and environmental case for funding orderly closure is at least as strong in steel as it ever was in coal.

A legitimate concern about any closure aid framework is that it might generate windfall payments to owners for closures that would have happened anyway on commercial grounds. The German coal phase-out addressed this through verified capacity destruction requirements — mandatory physical decommissioning of equipment, confirmed before payments are released — which prevent facilities from receiving closure compensation whilst retaining the option to restart. A steel closure aid framework designed on the same basis would be resistant to this risk.

Reforming the EU state aid framework to permit closure aid requires a Commission proposal and member state agreement — a medium-term legislative ambition rather than a quick fix. But the case for reform is analytically strong, the coal precedent exists, and the Commission's current revision of its Rescue and Restructuring Aid Guidelines provides a natural moment to begin that conversation.