The short answer
Yes — a building calculated below its embodied-carbon limit at design can finish above that limit once it is built. A 2026 Finnish study from VTT and Aalto University found that choosing high-carbon product alternatives over average ones, inside the same product groups the design already specified, may raise a building’s production-stage (A1–A3) emissions by 20 to 30%. The design did not change. The structure did not change. What changed was which supplier’s version of the same board, bar, and beam actually arrived on site.
That is the gap this research measures for the first time: the distance between the building as calculated and the building as procured.
Where the swing lives
The researchers took a reference apartment building — five storeys, 1,370 m² in Helsinki, modelled in hybrid, timber, and concrete versions — and re-ran its product-stage emissions using real environmental product declarations (EPDs) for the highest- and lowest-carbon products inside each category. An average of five EPDs per category, all to the EN 15804+A2 standard.
The spreads inside a single product group are wide:
- Steel rebar: the high-carbon EPD runs +90% against the average; the low-carbon option runs −29%.
- Stone wool insulation: +100% at the top of the range, −83% at the bottom.
- Gypsum board: +53% / −56%.
- Sawn timber: +60% / −27%.
- Concrete: Finland’s low-carbon classification cuts C30/37 from 255 to 178.5 kgCO2e per cubic metre, and C35/45 from 315 to 220.5.
Rolled up to the whole building, high-carbon choices inside the specified groups added 20% (hybrid), 30% (timber), and 24% (concrete) to A1–A3 emissions versus average products. Low-carbon choices cut 17 to 23%. In other words, procurement holds a swing of roughly half the production-stage footprint, in both directions, without touching the design.
And that understates it. Building services — the single largest contributor in the reference building, at 30 to 40% of A1–A4 emissions — had to be left out of the variation entirely, because too few EPDs with genuinely lower-carbon alternatives exist yet for equipment like elevators and air handling units. When those declarations mature, the procurement swing gets bigger, not smaller.
The second front: transport and the site itself
Modules A4 and A5 — getting materials to site and running the site — swung wider still. Across transport, construction machinery, and site energy use, the study’s low-carbon scenario came in 58% below the average scenario, and its high-carbon scenario 152% above it.
The widest single lever was the electricity the site runs on: hydro power versus the standard Finnish production mix differed by a factor of 21, and a grey residual mix ran another factor of five above the average. Transport told the same story in miniature — electric trucks at 6 tonnes of CO2e for the reference project, biodiesel at 29, diesel at 59. None of these are design decisions. All of them are made by whoever plans logistics and site operations, usually after the carbon calculation is already filed.
Why Finland’s rule makes this a compliance problem, not a curiosity
The Finnish Building Act does something most regimes do not: it mandates the embodied-carbon calculation, and tests it against the legal limit value, only after construction, on the building as built. The European EPBD, by contrast, asks for the whole-life carbon assessment during design.
Under an as-built test, the sequence inverts. The design team calculates a compliant building. Procurement then makes the final product choices — in Finnish practice, routinely, inside the design framework — and the emissions are realized on site. If those choices drift high, the building crosses its limit at exactly the moment nothing can be redesigned. The failure is discovered when it is permanent.
Any jurisdiction moving toward limit values inherits some version of this exposure. The specific trigger differs — an as-built assessment, a tightening limit at a later permit stage, an EPD-based specification audited at completion — but the mechanism is identical: the number is not final when the design is. It is final when the last substitution is approved.
The two moves the study recommends
The authors’ recommendations are plain, and both are cheaper than the failure they prevent.
Design with a safety margin. If product choice alone can move A1–A3 by up to 30%, a design that lands at 97% of the limit has not complied — it has gambled. The margin belongs in the calculation from the start, sized to the spread of the products the specification actually allows.
Steer the footprint while it can still be steered. The study calls for better data collection and visibility during construction, plus a post-construction evaluation calculation. In practice that means treating substitution as a compliance act: a specified product swapped for an “equivalent” is checked against its EPD and the running total before it is approved, not discovered after it is installed. The carbon ledger stays open and visible for as long as the building is still changeable — which is precisely the procurement window.
Read the numbers the way the authors ask you to
This is one Finnish reference building, EPD samples are small, the site-operation figures were transplanted from two Norwegian cases, and the authors say plainly that their results are indicative rather than exact. What travels is not the third decimal place. What travels is the mechanism: inside every product group your specification already names, there is a high-carbon version and a low-carbon version, and the difference between them is large enough to decide whether a compliant design becomes a compliant building.
The design sets the ceiling. Procurement decides whether the building stays under it.
Source: Vainio-Kaila, T., van den Brink, R., Ala-Kotila, P., Lavikka, R., Nyqvist, R. “The effect of procurement and construction site decisions on the embodied carbon emissions of buildings: a Finnish perspective.” IOP Conference Series: Earth and Environmental Science 1672 (2026) 012051. VTT Technical Research Centre of Finland and Aalto University. Open access (CC BY 4.0).