SteelOnTheNet Insights Podcast

Introduction (00:00–01:00)

Welcome back to SteelOnTheNet Podcasts. I'm Dr Andrzej M Kotas.

There is one number that predicted every major steel crisis of the last fifty years — before the crisis arrived.

In 2015, global capacity utilisation fell below seventy percent, and a wave of bankruptcies and plant closures followed across Europe and North America. In 2009, it collapsed to sixty-three percent and the industry lost billions in months. The overcapacity crisis that has plagued the industry since the early 2000s is, at its core, a capacity utilisation story.

Yet despite its predictive power, capacity utilisation is one of the most misunderstood and misused metrics in the industry. The headline figures published by trade associations are frequently misleading. The comparisons between companies and countries that analysts draw from those figures are often invalid. And the threshold levels that should trigger concern differ significantly between different types of steel operation — in ways that matter enormously for investment decisions.

This episode covers all of that. What capacity utilisation really measures — including the definitions that most published figures ignore. Why the economics differ sharply between integrated blast furnace operations and electric arc furnace steelmakers. Why the metric has such strong predictive power. What current utilisation trends are telling us about the next phase of the industry's restructuring. And how companies, investors, and governments should be using this metric in practice.

Let's start with the definition — because it is more complicated than it looks.

Section 1: What Capacity Utilisation Actually Measures (01:00–04:00)

Capacity utilisation, at its simplest, is actual output divided by capacity, expressed as a percentage. If a plant can produce two million tonnes per year and produces one and a half million, utilisation is seventy-five percent. Straightforward.

Except that the denominator — capacity — is not a fixed number. It depends on several factors that are almost never disclosed in published figures, and that vary significantly between plants and between countries.

The first factor is shift pattern. A rolling mill operating two shifts per day has a very different practical capacity to the same mill operating three shifts. Moving from two shifts to three does not simply increase capacity by fifty percent — the gain depends on the specific bottleneck in the production sequence, maintenance windows, and crew availability. But the point is fundamental: nameplate capacity figures almost always assume a specific shift configuration, and that configuration is rarely stated. When a company reports capacity utilisation of eighty-five percent, you need to ask: eighty-five percent of what shift pattern?

The second factor is product mix — and this is particularly important for long product rolling mills, where it is poorly understood outside the industry.

Consider a bar and rod mill producing rebar. If that mill is rolling forty-millimetre diameter bars, each bar carries a significant weight of steel per metre of rolled length. The mill's hourly tonnage output is high. Now shift the same mill to rolling ten-millimetre diameter bars for the same number of rolling hours. You are rolling far more metres of bar — the mill is running just as fast and the crew is working just as hard — but each metre weighs a fraction of the forty-millimetre product. The tonnage output per hour drops substantially.

The ratio can be striking. A mill producing large diameter rebar — forty or fifty millimetres — might achieve two and a half to three times the tonnage per hour of the same mill rolling small diameter wire rod quality bar. Yet both are described as operating at one hundred percent utilisation in shift terms. A mill that appears to be running at full capacity on a shift basis may be delivering significantly less tonnage than its nameplate figure suggests, simply because the market is asking for lighter gauge products.

The third factor — closely related to product mix, but distinct — is batch size and order complexity. Consider two rolling mills with identical equipment and identical shift patterns. The first produces a standard commodity rebar in two or three standard sizes, all to the same chemistry. It runs in long, uninterrupted campaigns. This mill's actual rolling time as a proportion of available time is high — perhaps eighty-five to ninety percent.

The second mill serves a diverse customer base requiring twenty or thirty different chemistries and a wide range of section sizes, including small batches for niche or high-specification applications. Every chemistry change requires a purging sequence in the meltshop. Every size change on the rolling mill requires a pass schedule adjustment, a roll change, and a re-threading of the bar guide system. When a batch is only fifty or a hundred tonnes, the changeover time as a proportion of actual rolling time can be substantial — sometimes exceeding the production time itself.

This mill may report the same nameplate capacity as the first. It operates the same number of shifts. But its actual output tonnage may be thirty to forty percent lower. The capacity utilisation figure in the annual report will not reflect this structural reality.

A further definitional issue is the distinction between nameplate, practical, and effective capacity. Nameplate capacity is what the equipment was designed to produce under ideal conditions. Practical capacity is what the plant can realistically sustain over a full year, accounting for planned maintenance shutdowns and normal operational variability. Effective capacity is what the plant can actually deliver given its current workforce, maintenance state, and order book.

In my experience, the gap between nameplate and effective capacity in an ageing integrated steel plant is commonly fifteen to twenty-five percent. In a poorly maintained plant, it can be forty percent or more. When trade associations publish global capacity utilisation figures based on nameplate capacity, they are systematically understating true utilisation — and therefore overstating the apparent degree of overcapacity. Policy debates about steel overcapacity — at the OECD, in the European Commission, and in bilateral trade negotiations — are routinely conducted using figures that overstate the problem by a significant margin.

Section 2: BOF Versus EAF — Why the Threshold Differs (04:00–07:00)

Now I want to address something that is almost never discussed clearly in general commentary on steel utilisation — the fundamental difference between integrated blast furnace and BOF operations, and electric arc furnace steelmaking, when it comes to the economics of low utilisation.

These are not just different technologies. They have radically different cost structures, and that means the utilisation threshold at which they become loss-making is completely different.

An integrated blast furnace and BOF plant has an extremely high proportion of fixed costs.

• The blast furnace must be kept hot.
• The coke ovens must run continuously.
• The sinter plant, the hot blast stoves, the slag handling systems — all of these consume energy and labour whether steel is being produced or not.

Turning down a blast furnace is not like turning down a thermostat. A partial blowdown and restart cycle costs millions and takes weeks. A full blowdown — effectively a shutdown — means a reline when you restart, which costs tens of millions more.

The consequence is that BOF operations have a very steep cost curve below their optimal utilisation rate. The break-even point for a typical integrated mill — covering all fixed and variable costs — is generally in the range of seventy-five to eighty percent of practical capacity. Below that level, fixed cost absorption deteriorates rapidly. Below seventy percent, most integrated mills are generating negative cash margins. Below sixty-five percent, they are destroying capital with every tonne they produce.

This is why BOF producers fight so hard to maintain volume, even at the cost of margin. Selling steel below its full cost is rational in the short term if the alternative is absorbing the full fixed cost base with even lower output. It is the logic that drives the price destruction we see during downturns — and it is embedded in the cost structure of the technology itself, not in any failure of commercial judgment.

EAF operations are structurally different. An EAF melts scrap using electricity. The primary input costs — electricity and scrap — are both variable. When an EAF is not running, it is not consuming either. The fixed cost base of an EAF operation is significantly lower than an equivalent BOF operation — typically thirty to forty percent lower as a proportion of total cost.

The result is that EAF operations can be turned on and off with far greater flexibility. The utilisation threshold at which an EAF operation becomes loss-making is therefore lower — often in the range of sixty to sixty-five percent of practical capacity, compared to seventy-five to eighty percent for a BOF.

The transition from BOF to EAF now underway across Europe and North America — driven partly by decarbonisation requirements, partly by scrap availability — will therefore change the cyclical dynamics of the industry structurally. A higher proportion of EAF capacity means the industry as a whole will be better able to flex output in response to demand. That should, over time, reduce the amplitude of the price swings that have historically made steel such a difficult investment proposition.

Section 3: Utilisation as a Leading Indicator (07:00–09:30)

The reason capacity utilisation has such strong predictive power is the pricing mechanism it creates.

When utilisation falls — particularly when it falls across a region or globally rather than at a single plant — producers with the highest cost structures face an impossible choice. They can cut output, absorbing the fixed cost consequences I described. Or they can maintain volume and accept deteriorating margins. In practice, the majority choose to maintain volume, at least initially. The result is that pricing power collapses as multiple producers simultaneously prioritise volume over margin.

The price decline that follows tends to be faster and deeper than the utilisation decline that triggered it. A fall from eighty percent to seventy-five percent utilisation — which sounds modest — can produce a price decline of twenty to thirty dollars per tonne within six months, as the market absorbs the marginal volume from producers who have no rational alternative to keeping their blast furnaces running.

This is why rate of change matters more than absolute level. A market at seventy-two percent utilisation and stable is manageable. A market at seventy-five percent utilisation and falling is not.

The historical correlations are consistent. The global utilisation trough of 2001 to 2002, at around seventy-five percent, preceded the Bethlehem Steel, LTV, and National Steel bankruptcies. The 2009 trough at sixty-three percent was the worst since the 1980s and produced the deepest margin compression in decades. The 2015 to 2016 period, when Chinese capacity additions pushed global utilisation below seventy percent, triggered anti-dumping investigations across every major importing region and accelerated the European restructuring cycle.

The lag between utilisation decline and price impact is typically six to nine months. Between price decline and financial distress at the weakest producers, another six to twelve months. A careful observer watching utilisation data therefore has a twelve to twenty-one month window to anticipate the downstream consequences — in pricing, in credit quality, in acquisition opportunities.

Section 4: What Current Utilisation Is Telling Us (09:30–12:00)

I want to be careful here, because utilisation figures change and this podcast will be listened to at different points in time. What I can offer is the structural context that makes current and near-future figures interpretable.

The structural overhang from Chinese capacity additions between 2005 and 2015 has never been fully resolved. China added roughly five hundred million tonnes of capacity in a decade — more than the entire rest of the world's production base. Although Chinese utilisation has recovered from its 2015 trough, the absolute scale of Chinese capacity means that even small changes in Chinese domestic demand have large effects on export volumes and therefore on the utilisation of producers in importing regions.

There is also a data quality issue with Chinese figures specifically that I want to flag. A significant proportion of Chinese steel capacity is owned by state enterprises operating under soft budget constraints — that is, enterprises where the financial consequences of loss-making production are absorbed by the state rather than by private shareholders. The normal economic signal that drives capacity exits in market economies — sustained losses leading to closure — operates much more slowly, or not at all, in this context. Chinese utilisation figures must therefore be read with particular caution.

The second structural complication is the green steel transition. Across Europe, North America, and parts of Asia, significant new EAF and DRI capacity is being constructed or planned. These projects are real, they are funded, and many will come online in the next three to seven years. The problem is the sequencing.

New green steel capacity will be added before the legacy blast furnace capacity it is intended to replace has been permanently closed. The reasons are partly commercial — no investor wants to close a cash-generating blast furnace until the replacement is operational — and partly political, because governments are reluctant to force closures before the new jobs are in place. The result is a period, potentially lasting five to ten years, during which effective global capacity is higher than it would otherwise be, because old and new capacity coexist.

The arithmetic is uncomfortable. If announced green steel projects in Europe alone complete on current schedules, and if legacy BOF capacity does not close in parallel, effective European capacity rises before any net reduction occurs. That is a utilisation headwind of significant proportions — arriving at a time when European steel demand is already under pressure from economic conditions and the import dynamics created by the CBAM implementation period.

This is not a reason to oppose the green steel transition — it is necessary and the right long-term direction. It is a reason to be clear-eyed about the near-term utilisation and pricing consequences for the industry during the transition.

Section 5: How to Use This Metric in Practice (12:00–14:30)

Let me turn to practical application, because this is ultimately what the metric is for.

For investors and lenders evaluating a steel company acquisition or financing, the most important utilisation discipline is stress-testing. Almost every steel investment model I have reviewed assumes utilisation in the range of eighty to eighty-five percent throughout the projection period. This is broadly the average over a full cycle — but it is not what happens at the trough. Any investment that cannot service its debt at sixty-five to seventy percent utilisation for a sustained two-year period is not adequately stress-tested. The cycle will find it.

The second application is as an acquisition timing indicator. When sector utilisation is falling and approaching the seventy percent threshold, distressed assets begin to appear. When it has fallen through that threshold and has been below it for twelve months or more, the strongest assets become available — because even well-run companies run out of liquidity in a deep trough. The investors who made money buying steel assets in 2002 and 2016 were not lucky. They were watching utilisation data and had prepared capital in advance.

For governments and policy makers, capacity utilisation is the earliest available warning signal for industrial policy intervention. A sustained decline in domestic utilisation — particularly if it is falling faster than the global average — indicates a structural competitiveness problem that will not self-correct. The appropriate policy response differs significantly between a structural decline and a purely cyclical one. Distinguishing between the two requires watching utilisation alongside cost position data, not in isolation.

For steel company management, the most common mistake I observe is benchmarking current performance against current sector utilisation rather than against cycle-trough utilisation. A plant that looks competitive at eighty percent utilisation may be deeply loss-making at sixty-five percent. Understanding your own cost curve across the full utilisation range — not just at your current operating point — is the foundation of any serious strategic planning.

Conclusion: Three Rules (14:30–16:00)

Let me close with three rules I apply whenever I am using capacity utilisation data in an analysis.

Rule one: never accept a capacity figure without understanding how it was derived. Was it nameplate or practical capacity? What shift pattern does it assume? What product mix? A figure stated without these qualifications is not a reliable basis for serious analysis — and in steel, serious analysis is the only kind worth doing.

Rule two: always distinguish between BOF and EAF when assessing utilisation thresholds. The break-even utilisation rate for an integrated blast furnace mill is fifteen to twenty percentage points higher than for an equivalent EAF operation. Applying the same threshold to both leads to systematically wrong conclusions about which operations are under genuine financial threat.

Rule three: watch the rate of change, not the absolute number. A market falling from eighty percent to seventy-five percent utilisation is more dangerous than a market stable at seventy percent, because the pricing consequences come from the direction of travel — from the additional volume hitting the market as BOF producers respond to declining output with the one tool they have available, which is to keep producing.

Capacity utilisation does not predict everything. But it predicts more than any other single number in the steel industry — the direction of prices, the timing of financial distress, the emergence of acquisition opportunities, and the need for policy intervention. And most of the people who should be watching it closely are looking at the wrong version of it.

The most valuable number in steel analysis is often the one nobody agrees how to calculate.

Until next time, this is Dr Andrzej M Kotas.

Thank you for listening.