Insight, Just Transition

The Factory Cannot Transition Without the Town

An Italian court has upheld the shutdown of the Taranto blast furnaces. Around 8,000 people work there directly. The evidence on what happens to workers like them is unambiguous, and it is the reason retraining alone has never worked.

Line-art scene: a large steelworks with cooling towers and a blast furnace drawn in outline at the top of the frame, and beneath it, connected by fine radiating lines, a dense town of small buildings including a school, a bakery, a machine shop, a bus and rows of houses, so that the works and the town read as one structure.
The lines only run one way. That is what makes this hard.

Taranto, and the evidence that retraining was never going to be enough

An Italian court has upheld an order to shut the blast furnaces at the former Ilva steelworks in Taranto, citing environmental and safety deficiencies (as reported). The plant directly employs roughly eight thousand workers and supports a substantially larger contractor ecosystem. Italy is now considering a future built around cleaner technologies including electric arc furnaces and direct reduced iron.

The usual framing of this is coal-based steel becoming green steel, which is a technology substitution. What is actually in front of Italy is a legacy plant, a pollution liability, eight thousand direct jobs, a supplier economy, a regional identity, an unproven technology package, a financing gap and a clock.

The technology is the part that is closest to solved.


What the evidence says happens to these workers

There is a large and consistent body of research on what happens to people displaced from long-tenure industrial employment, and it is worth stating plainly because it contradicts the assumption underneath most transition planning.

Jacobson, LaLonde and Sullivan used administrative earnings data from Pennsylvania in the 1980s to follow workers who left distressed firms after long service. Their findings, published as Earnings Losses of Displaced Workers, are stark. High-tenure workers separating from distressed firms suffered long-term losses averaging twenty-five per cent per year. The losses began mounting before the separation, depended more on local labour market conditions and the former industry than on age or sex, and were large even for those who found new jobs in similar firms.

Two further findings matter more than the headline number. Almost all displaced workers found stable employment within about eighteen months. And most of the earnings loss was not unemployment. It was lower wages in the new job.

So the problem is not primarily that people cannot find work. It is that the wage attached to twenty years of specific experience does not transfer, and there is little evidence it ever comes back.

Chart, 'They found work. The earnings did not come back.' A horizontal axis running from '5 years before separation' through 'separation' to '10 years after'. A cobalt dashed line rises gently across the whole range, labelled 'expected earnings'. A coral solid line tracks it, begins to fall slightly before the separation marker, drops sharply at separation, recovers partially, and then runs flat and permanently below the dashed line, labelled 'actual earnings'. A small sky-blue marker on the recovering coral line is labelled 'stable employment found'. The persistent gap between the two lines is shaded and annotated 'this is the harm, and retraining does not address it'. Footnote: Pattern after Jacobson, LaLonde and Sullivan (1993). Schematic, not Italian data. Transitions Lab, 2026.
Employment returns fairly quickly. The old wage does not.

That is why retraining, taken alone, has such a poor record. Retraining addresses employability, and employability was not the binding constraint. A fifty-two-year-old furnace operator with a new certificate is employable and will earn substantially less, and no amount of certification changes that, because what has been lost is the value of experience in a specific production system that no longer exists.


Which is why the supplier economy is the larger number

The eight thousand direct employees are the visible figure. The contractor ecosystem around a plant of that scale is usually larger and is always more fragile.

Maintenance firms, refractory specialists, transport operators, engineering workshops, industrial cleaning, canteens, security. Many of them have one customer. Their equipment is specific to that customer's processes. Their workforce holds knowledge about that plant rather than about steelmaking generally.

When a plant closes, those firms do not downsize. They end, and their owners are usually people who invested their own capital in equipment with no second market.

An electric arc furnace employs fewer people than a blast furnace and buys a different set of services. So even the successful version of this transition, in which the plant is replaced with a cleaner one on the same site, does not restore the supplier economy. It replaces it with a smaller and differently shaped one.

That is the honest version of the industrial case and it is rarely put to the region in those terms.


What the sequence has to be, and why it usually is not

The structural difficulty at Taranto is a timing problem that recurs everywhere this happens.

The court order is immediate. The environmental harm is present tense and has been for decades, which is why the order exists and why it is defensible. But the replacement technology package, direct reduced iron with electric arc furnaces, requires a financing decision, a construction period, a supply of the right iron ore inputs, an enormous quantity of firm electricity at an industrial price, and in the medium term hydrogen that does not yet exist at cost.

So the closure is fast and the replacement is slow, and the interval between them is where a regional economy either survives or does not. Everything that determines the outcome happens in that interval: whether the skilled workforce stays or disperses, whether the supplier firms survive on other work, whether the young leave.

Once a workforce disperses it does not reassemble. A plant commissioned in 2033 into a region whose skilled labour left in 2028 will import its operators, and the industrial case for building it there weakens accordingly. That is the capability question operating in reverse: capability that took decades to accumulate can be lost in a few years, and the loss is not recoverable on the same timescale.

Diagram: a horizontal timeline. At the far left a blast furnace drawn solid, with a heavy vertical coral line beside it labelled 'court order, immediate'. At the far right an electric arc furnace drawn in dashed outline, labelled 'replacement, if financed'. Between them a wide bracketed span labelled 'the interval'. Inside the span, four small icons in sequence with arrows: a group of workers labelled 'workforce disperses', a small workshop labelled 'supplier firms end', a group of figures with suitcases labelled 'out-migration', and a head with a brain labelled 'tacit knowledge'. Beneath the whole span: 'everything that decides the outcome happens here'.
Fast closure, slow replacement, and the interval between them is where a regional economy either survives or does not.

What a serious transition package would contain

Income continuity that matches the evidence, not the assumption. If the documented loss is a permanent wage reduction rather than a spell of unemployment, then a wage insurance instrument, topping up the difference between old and new earnings for a period of years, addresses the actual harm. A short unemployment benefit followed by a retraining course addresses a harm that is not the main one.

Bridge work for the supplier firms. Decommissioning, remediation, site preparation and construction of the replacement are all work, and much of it can be done by the firms that would otherwise fail. Sequencing that work to fill the interval is an explicit choice and it is usually made by default.

Early commitment on power price and supply. An electric arc furnace is an electricity business. If the industrial tariff and the firm supply are not settled early, the investment decision will not be made, and the region will have absorbed the closure without receiving the replacement. That is the same decision bar that is currently holding up two dozen European mineral projects.

Honesty about headcount. The replacement will employ fewer people. Saying so early is painful and is the precondition for the rest of the package being trusted. A promise of equivalent employment that everybody in the town knows to be false makes every other commitment suspect.


What would be worth measuring

Earnings trajectories, individually tracked, for five years. Not employment rates. Actual earnings against a matched comparison group, which is exactly what the displaced-worker literature does and which is entirely feasible with Italian administrative data.

Survival of supplier firms, and what the survivors switched to.

Out-migration by age and skill. The single best predictor of whether the replacement plant will have a workforce.

What the town was told, and when. Because the durability of any package depends on whether people believe it, and belief is set by whether the first thing they were told turned out to be true.

Taranto will be studied for the next twenty years. The most useful thing anybody could do now is start measuring, before the closure, because the comparison that matters requires a before.

The Lab works on this in local manufacturing and through field research with the workers and suppliers a transition lands on.

If you are designing a just transition package and want the harm measured as the evidence describes it, tell us what you need to know.


Sources


This is an independent insight piece by Transitions Lab. For the Lab's applied work, see Local Manufacturing & Supply Chains. See also Capability Is the Slow Part on capability that takes decades and can be lost in years, and Strategic Is Not the Same as Financeable on the investment decision that determines whether a replacement ever gets built. To discuss a study, see Contact.

Read more Articles & insights See all articles → See it in the field Case studies See all case studies →