A two-speed construction sector: civil engineering accelerates, building waits

Public works

Public works: the engine that accelerates

Why roads, networks and asphalt run clearly above the rest of construction.

It is the part of construction least talked about, yet the one running best. Cost and activity indices for public works are rising markedly faster than those for building, and the gap has widened since the start of the year. Leading the way: roads, asphalt production and energy networks, all lines pulled by public procurement, infrastructure renewal and the energy transition. Where building waits for the private buyer to return, civil works run on order books fed by local authorities, network operators and the State. The result: within a single sector, two clearly diverging dynamics. Reading 'construction' as one homogeneous block today means hiding its most dynamic half.

🔍 Focus: why public works and building don't follow the same cycle

The difference comes down to who pays. Public works live on public procurement: local authority budgets, infrastructure plans, renewal of water, electricity and telecom networks, energy-transition projects. These are long cycles, voted in advance, largely insensitive to the mortgage rate. Building depends on private financing: a household borrowing to buy, a developer launching a project when demand is there. When credit is expensive, that engine stalls, even if the needs exist. This is why, in 2026, the same sector can show accelerating civil works and a waiting building segment: they are not plugged into the same clock. For a client, the lesson is simple: never steer a 'construction' decision on an average, always on the sub-segment that concerns you.

🧭 Key takeaways

Civil works are the sector's strong point in 2026: roads, asphalt and energy networks lead, well above building.
The fuel of this acceleration is public: infrastructure, energy transition, network renewal. It does not depend on mortgage credit.
The signal to watch: the trajectory of public budgets and infrastructure envelopes, the only real possible brake on this engine.

⚠️ The public-works engine is solid but not invulnerable: it rests on public spending. A tighter budget constraint on local authorities or the State would, with a lag, slow procurement. Second watchpoint: asphalt stays tied to bitumen, and thus to oil. A lasting rebound in crude would show up in road costs.

→ The detail of TP and building indices + forecasts in the Construction Newsletter

Building

Building: the engine looking for its second wind

Why the upstream of housing recovers slowly while the downstream stays numb.

On the building side, the picture is one of convalescence, not a rebound. The good news comes from upstream: housing authorizations, read over twelve rolling months to smooth the noise, are gradually recovering after two very low years. The bottom seems behind us. The less good comes from downstream: new-home sales stay at depressed levels, held back by still-expensive credit. In other words, permits to build are being granted faster than homes are sold. This gap between a recovering upstream and a lagging downstream is the real signature of building in 2026: the sector has stopped sinking, but it has not yet found the spark that turns a permit into a launched site and a sold home. As long as the cost of credit does not ease, this gap should persist.

🔍 Focus: why authorizations are read over twelve rolling months

Construction statistics are one of the most slippery grounds in economic data. Two traps await the hasty reader. First seasonality: a winter month is structurally quieter than a spring one, so comparing two raw months makes no sense. Then revisions: the latest published months are provisional, fed by still-incomplete collection, and therefore almost always understated on the first pass. An isolated very low month there is most often a month not yet fully caught up, not a genuine drop. The fix is simple and robust: read the twelve-month rolling total, which neutralizes the season and absorbs most of the revisions. It is this reading, and it alone, that today shows a gradual recovery in authorizations. Concluding on an isolated point means risking getting the direction wrong.

🧭 Key takeaways

Upstream recovers, downstream lags: housing authorizations edge up, but new-build sales stay low.
The lock is still credit: as long as its cost does not fall, a granted permit does not turn fast enough into a site and a sale.
The signal to watch: the cost of housing credit, which governs the shift from a convalescent upstream to a downstream that truly restarts.

⚠️ Caution on headlines: construction figures for the very latest months are provisional and will be revised, almost always upward. Do not read an isolated month, neither very low nor very high, as a trend. Only the twelve-month rolling total tells the real story of building.

→ Authorizations, starts and detailed sales in the Real Estate Newsletter

Site costs

Site costs: the bill has changed engine

Why the easing of materials will not bring your quote down.

Here is the most important shift, and the least visible. The line that had sent site costs soaring in 2022 and 2023, materials, has turned around: their indices are stabilizing, and some are even retreating sharply. One might conclude the bill will fall. The opposite is happening. The indices measuring the full production cost of a project, materials included, keep rising above inflation. The reason for the gap comes down to one word: labour. What materials give back, the cost of work takes up. Cost inflation has not disappeared, it has changed engine. For a client, it is a classic trap: to hope for a lower quote because the price of steel or copper has retreated, when the growing share of the bill now plays out elsewhere.

🔍 Focus: reading the difference between material cost and production cost

Two families of official indices tell two stories that must not be confused. Material cost indices track the price of inputs: cement, steel, copper, wood, bitumen. Production cost indices measure the full cost of delivering a work: materials, but also labour, equipment, energy and site overheads. In normal times, the two rise together. What makes 2026 instructive is that they diverge: materials calm down while production cost keeps climbing. This scissor effect has one logical culprit, the line that has taken over from raw materials: the cost of labour. For anyone signing contracts, the consequence is very concrete. Indexing a contract on a materials index when the rise comes from labour means hedging the wrong risk. The right reflex: look at the production index, not only the materials one.

🧭 Key takeaways

Materials are no longer the engine of cost inflation: they are stabilizing, some are retreating. The raw-materials surge is behind us.
Yet the site bill keeps rising: the baton has passed to labour, a far more rigid line than raw materials.
The signal to watch: the gap between material and production indices, the best gauge of the 'labour' share in your costs.

⚠️ Do not confuse easing raw materials with a falling bill. Labour is a rigid line: wages do not retreat like copper prices. The site bill will therefore stay tilted upward, even with materials calmed, with particular attention to strained trades such as finishing works and electrics.

→ Detailed material and production indices + forecasts in the Construction Newsletter

A two-speed construction sector: public works accelerate, buildings wait

This month's intuition

Public works: the engine that accelerates

🔍 Focus: why public works and building don't follow the same cycle

Building: the engine looking for its second wind

🔍 Focus: why authorizations are read over twelve rolling months

Site costs: the bill has changed engine

🔍 Focus: reading the difference between material cost and production cost

The common thread: one sector, three truths

Construction signals in summer 2026

Get our sector analyses delivered straight to your inbox

Sources & methodology