SteelOnTheNet Insights Podcast

Introduction (00:00–03:03)

Welcome to Steelonthenet dot com Podcasts. I'm Dr. Andrzej Kotas, and today I want to look at something the steel industry talks about constantly, and almost never explains.

We all know the industry has too much capacity. It has had too much capacity for most of my working life. Every downturn produces the same commentary: rationalisation is overdue, the marginal producers must exit, the market will clear. Ministers say it. Analysts say it. Boards say it about their competitors.

And yet the closures don't come. Or rather — they come far more slowly, and far less predictably, than the economics would suggest. Mills that have not covered their cost of capital in a decade are still rolling steel. Furnaces that lose money in every year but the peak of the cycle are still lit.

So the question I want to put to you today is not "how much capacity needs to close?" We know that answer, roughly, and have known it for years. The better question is why it doesn't. Once you understand why loss-making mills stay open, a great deal else falls into place. You understand why capacity pledges keep failing, why supply forecasts keep over-predicting rationalisation — and, if you own one of these assets, something rather uncomfortable about your own position.

There are three reasons, and I'll take them in turn. The first is that closing costs more than continuing — in cash, immediately. The second is that in much of the world, the owner isn't carrying the full loss anyway. And the third is that waiting has genuine value in a cyclical industry — but it is value that decays, and most owners are not paying to maintain it.

Section One: It Isn't Sentiment (03:03–06:07)

The popular explanation is that steel is emotional. Steelmaking towns, generations of families, national pride, the last blast furnace in the country.

There is something in that. I have sat in rooms where the emotional weight of a closure decision was entirely visible. But as an explanation for the industry's behaviour over forty years, it is thin. Steel executives make hard decisions routinely. They cut headcount, they mothball lines, they walk away from markets. They are not sentimentalists.

The real explanation is duller and much more powerful. Closing costs more than continuing — not in some abstract, long-run sense, but in cash, in the first year. A mill can be losing money on every tonne it produces and still be cheaper to run than to shut. That is not irrationality. That is arithmetic.

The reason is that operating a steel plant defers an enormous quantity of cost, and exit crystallises all of it at once. Think about what a running mill is postponing.

• Pensions, funded gradually.
• Redundancy, unpaid while people are still employed.
• Decommissioning of coke ovens and gas systems.
• Remediation of slag banks, lagoons, contaminated ground.
• Permit surrender conditions.
• Take-or-pay on gas, power, rail and port access.
• Long-term ore and coal contracts with years to run.

None of that appears in your operating margin. All of it appears the moment you decide to leave.

So when a board looks at a mill losing perhaps thirty million a year, and at an exit that might crystallise several hundred million of liabilities, the cash-preserving decision is to keep going. And again next year. And the year after.

That is the heart of it. What looks like denial is usually just cash arithmetic — and it holds for a very long time after the mill has stopped earning its capital.

Section Two: The Liability Wall (06:07–09:48)

Let me put some structure on that liability stack, because in my experience it is the part boards understand least well until they are forced to.

Pensions. In older integrated plants this is frequently the single largest item, and it can be larger than the enterprise value of the business it sits behind.

The British Steel Pension Scheme is the clearest illustration I know. After its sponsor announced it had lost two billion pounds in five years and could not continue covering those losses, the scheme — with around one hundred and thirty thousand members — was restructured in twenty seventeen through a Regulated Apportionment Arrangement. That mechanism lets a company end its responsibility for a pension scheme, with the Pensions Regulator's approval, where continuing to support it would mean the company inevitably became insolvent.

Sit with that for a moment. The pension was not a background item on a balance sheet. It was the thing that would have closed the business, and the steelmaking could only continue once it had been formally detached.

Then come the rest. Redundancy, where statutory minimums are the floor and enhanced terms negotiated over decades fall due as cash immediately. Decommissioning — coke batteries above all, which you cannot simply switch off and lock the gate on. Remediation, against standards that have tightened steadily, so the liability grows even when nothing happens on site. And the contracts: take-or-pay on industrial gases and power catches people out, because the oxygen plant was built on an agreement that does not care whether your furnace is cold.

Here is the point to hold on to. Every year of continued operation defers all of this — and quietly makes it worse, as standards tighten, the reline clock runs down, and deferred capital accumulates.

So the mill is not just loss-making. It is becoming more expensive to leave.

Section Three: Someone Else Pays (09:48–12:51)

The second reason mills don't close is that in many jurisdictions, the losses aren't entirely borne by the owner.

I looked at this in Episode Eight, on Chinese overcapacity, where provincial governments have strong incentives to keep local mills running regardless of the national position. But it would be a mistake to treat that as a Chinese phenomenon. Credit, energy relief, employment measures, direct public ownership — all have the same effect. They break the link between a mill losing money and a mill closing. If a market's exit mechanism is that unprofitable producers leave, support of this kind switches that mechanism off.

And it cuts both ways.

Look at Port Talbot. Tata ended ironmaking there, closing Blast Furnace Four, the sinter plant and primary steelmaking, after the earlier loss of the Morfa coke ovens and Blast Furnace Five. The company's stated position was that the heavy end had reached the end of its operational life. Steelmaking is planned to resume in twenty twenty-seven to twenty twenty-eight through one and a quarter billion pounds of electric arc furnace investment — of which five hundred million is a Grant Funding Agreement with the UK Government.

Read that carefully. Our own assessment is that public money did not prevent the closure here — it enabled it, conditioned on replacement rather than exit. That is an interpretation, not a company statement, and I put it to you as such.

That is the pattern worth noticing. Capacity that leaves this industry is very often not leaving at all. It is being substituted. On the eighteenth of January, twenty twenty-six, Algoma permanently ceased production at its blast furnace and coke batteries, ending one hundred and twenty-five years of coal-based integrated steelmaking and completing its transition to electric arc furnace steel. The blast furnace closed. The steelworks did not.

If you are modelling supply, that distinction is everything.

Section Four: The Option to Wait (12:51–15:55)

The third reason is the most defensible of the three, and the one I have the most sympathy with. Steel is cyclical. A mill that cannot cover its costs at today's prices may cover them handsomely at the top of the next cycle. Holding an asset through a trough therefore has genuine value. And mothballing gives a management team a way to say "we are waiting" rather than "we have failed."

The problem is that what you are holding decays, in ways that appear nowhere in the accounts. Care and maintenance is not free — security, refractory preservation, insurance, a skeleton crew. The campaign clock does not pause. Skilled operators disperse, and they do not come back. Deferred capital accumulates and permits drift toward renewal.

So a great many mothballed assets are options nobody is paying the premium to maintain. They look like flexibility. They are increasingly just delay.

You can see the ratchet in practice. Consider Redcar, before the events most people remember. The blast furnace, with Lackenby steelmaking and the South Bank coke ovens, was mothballed at the end of January, twenty ten, after an international consortium walked away from a long-term offtake agreement. That looked like the end. It wasn't: the site was sold and the furnace relit in twenty twelve. Hold that thought — we'll come back to how it ended.

The sequence you see repeatedly is idled, then indefinitely idled, then closed, with "temporary" doing an enormous amount of work at every stage. Sometimes the restart comes. More often the years pass, the workforce disperses, the reline falls due, and the decision gets made by default.

But waiting does sometimes pay, and it can pay for a long time.

The B blast furnace at Granite City in Illinois had been idle since the autumn of twenty twenty-three; the A furnace since twenty twenty. In December, twenty twenty-five, after what the company described as several months of analysing customer demand, U.S. Steel began the process of restarting the B furnace, and on the thirtieth of March, twenty twenty-six, confirmed it was producing steel again, with around four hundred people hired to run it.

Waiting paid. But look closely at what changed in the interval. It was not the profitability of the asset. It was who owned it.

Section Five: What Actually Forces Closure (15:55–20:13)

So those are the three reasons a mill stays open: exit crystallises costs that operation defers, someone else is often absorbing part of the loss, and waiting for the cycle has genuine value. Between them they can hold a furnace lit for a decade past the point where it made any sense.

Which leaves the obvious question. If losses don't close mills, what does?

One piece of evidence first. We reviewed more than sixty steel insolvencies worldwide, from the nineteen-sixties to the present day. In roughly two-thirds, the assets passed to a named acquirer. And on our own estimate, fewer than one in ten has so far ended in permanent closure of the plant.

That is a Steelonthenet figure, from our own review, and I'd encourage you to treat it as an order of magnitude rather than a precise count. Insolvency, overwhelmingly, transfers a steel mill. It does not remove it. Which is why counting bankruptcies is such a poor way to forecast capacity, and why so many people have been so consistently wrong about rationalisation.

Closure arrives instead when something with a hard deadline turns up. Losses have no deadline. That is precisely why they don't force the decision.

One: cash runs out. Remember the Redcar furnace we left relit in twenty twelve. It ran three more years. Then, in October, twenty fifteen, its owner SSI UK was wound up. The coke ovens were kept alight for a matter of days while the official receiver looked for a buyer; there was none, and no funds to keep buying coal. The ovens would have been irreparably damaged if allowed to go cold, so the priority became closing them safely.

Note what happened to the decision itself. It passed from a board weighing options to a liquidator with no money and no purchaser. That is not the same thing at all. And note the arc: mothballed in twenty ten, restarted in twenty twelve, gone in twenty fifteen, demolished in twenty twenty-two. The option was real, it was exercised, and it still ran out.

Two: the spending deadline. A blast furnace campaign ends whether or not you can fund the reline. A permit falls due for renewal whether or not you can fund compliance. Either way the deferral is no longer available — you commit a very large sum, or you are closed.

It is the furnace, or the regulator's calendar, that sets the date; not the finance director. This is precisely the reasoning Tata gave at Port Talbot: the heavy end assets had reached the end of their operational life. Once that is true, continuing stops being the cheap option, and arithmetic that had held for years reverses inside a single year.

Three: an upstream failure strands the asset. Coke supply, oxygen, power connection, port access, rail. You can be perfectly viable in your own right and still be closed by the loss of a single input. In April, twenty twenty-five, the UK Parliament was recalled to pass emergency legislation giving the Secretary of State power to direct a steel undertaking's operations, after raw material stocks at Scunthorpe ran critically low and the owner — losing around seven hundred thousand pounds a day — was considering closing the furnaces. The National Audit Office judged that swift action prevented an imminent and disorderly shutdown.

Now note the second half of that story. The intervention did not remove the underlying problem. It created a new deferral — and we know what it costs, because it is published. Working capital provided to the company since that legislation stood at four hundred and eighty-four million pounds as at the fourteenth of May, twenty twenty-six.

Four: the support stops, or changes direction. Support is itself a deferral, and it is revocable. A change of government, a change of fiscal circumstances, or public funding conditioned on transition rather than survival — any of these can end the deferral as quickly as it was created.

Rohrwerk Maxhütte shows how that ends. The tube and pipe works at Sulzbach-Rosenberg in Bavaria, once among the largest in Germany, closed permanently in September, twenty twenty-five, after repeated insolvencies — around three hundred employees laid off, and no investor found, despite an unsuccessful state rescue attempt of some two hundred and fifty million euros.

Note the sequence. The plant did not close when it started losing money; it had been in difficulty repeatedly. It closed when the last route to postponement was exhausted. Two hundred and fifty million euros bought time. When the time ran out, so did the mill.

Five: ownership changes. The incumbent cannot close it. A new owner, a liquidator, or a parent in another jurisdiction can.

Smederevo, in Serbia, is the clearest case I know. U.S. Steel acquired the plant in two thousand and three. In January, twenty twelve, after the collapse in steel prices, it sold the business back to the Serbian government — for one dollar. Look at what an owner did when it wanted out. It did not close the plant. It gave it away.

The furnaces did go cold that summer. But the plant restarted in twenty thirteen, ran under state ownership, and was sold in twenty sixteen to the Chinese group HBIS. The capacity is still there.

Look at that list of five. Not one of them is "the mill lost money."

Section Six: The Lessons (20:13–23:53)

So what do you do with all this? It depends where you sit, and there are three answers.

If you own the asset, you need to know your exit number, in full, before you need it. Most boards can quote the net present value of continuing to the nearest million. Very few have ever had the closure liability stack properly costed — pensions, redundancy, decommissioning, remediation, take-or-pay, permit surrender. Until you have that figure, you cannot tell whether you are choosing to operate or are simply unable to leave. Those two positions look identical in the accounts, and they are opposite in strategy. And the number gets worse every year you decline to look at it.

If you compete with the asset, or lend to it, stop treating losses as a predictor of exit. They aren't one. Model closure against the five triggers instead. A mill losing money indefinitely is a normal feature of this industry. A mill approaching a campaign-end decision it cannot fund is the actual signal. Most rationalisation forecasts get this backwards, which is why they have been wrong in the same direction for thirty years.

And if you are making policy, it is worth being clear about what support of this kind actually buys. It buys the right to have that capacity available if conditions change. Not the obligation to keep producing forever — the right to still be there when the market turns. That is a perfectly reasonable thing for a government to want, and I want to be fair to the case for it. Steel is cyclical, security of supply is a legitimate concern, and closing capacity in a trough that would be viable in a recovery destroys something you cannot rebuild. Furnaces and skilled workforces do not come back on demand.

But if that is what you are buying, then price it as such. Because a right of that kind has a premium, it has a condition under which you would actually use it, and it has an expiry date.

Think back to Maxhütte. Two hundred and fifty million euros bought time. What it did not buy was an answer to the question of what the time was for — and when the money stopped, nothing had changed except that the plant was older and the workforce smaller.

So state the recovery conditions you are buying access to. State what the support costs each year — including what it adds to the eventual exit bill, because deferral is never free and that increment appears in no budget line anywhere. And state when the right runs out: the reline date, the permit date, the point at which the workforce has dispersed beyond recall.

Interventions that can answer those three questions are usually good ones. Interventions that cannot are subsidies with better public relations.

Conclusion (23:53–25:43)

I'll leave you with this.

We talk about overcapacity as though it were weather. Something that blows in, and passes, and blows in again — and the advice that follows is always the same. Hold on. Wait for the sun.

And the sun does come out. Prices recover, order books fill, and for two or three quarters everybody remembers why they held on. But notice what doesn't happen when the sun shines. Nothing closes then either. The good years don't remove capacity — they justify keeping it. Which means there is no weather in which this industry rationalises.

Capacity does not leave when it stops paying. It leaves when somebody finally runs out of ways to postpone the bill. And when you look back at what actually switched the lights off, it is always something with a date on it.

• A liquidator.
• A reline or a permit renewal falling due with no money behind it.
• A coke supply that fails.
• Support withdrawn.
• An owner with no local exposure.

Five events. Every one of them has a date. Losses have no date at all — and that is precisely why losses don't close mills.

So this is not a cycle. It is a condition.

And the question for any board is not whether the industry has too much capacity. It plainly does, and it has for decades. The question is narrower, and much harder. Do you know what your own exit would cost — all of it, pensions to remediation, as a single number you could put in front of your shareholders tomorrow?

Because if you don't, you are not waiting for the sun.

You are not choosing to stay open. You are simply discovering, one year at a time, that you cannot afford to leave.

Thank you for listening. This is Dr Andrzej Kotas for Steelonthenet dot com.