Tariffs and Canadian Construction: Where the Cost Actually Lands

This article states the position as at 11 September 2026. That qualifier matters more than usual, because the measures described here have changed repeatedly over the past eighteen months and will change again.

After the United States declined to renew the USMCA in July 2026, it imposed tariffs of 50 per cent on roughly $27.6 billion of Canadian goods with effect from 22 August 2026. Canada announced matching countermeasures on 25 August, and those came into force on 8 September 2026, applying rates of 15, 25 and 50 per cent across more than six hundred tariff lines covering a comparable value of United States imports.

The construction-relevant entries are not peripheral. Flat-rolled and semi-finished steel, aluminum bar, plasterboard and plywood all attract the top rate of 50 per cent. Softwood lumber attracts 25 per cent, as do carpets and floor coverings. Plastic flooring and fixtures are at 50 per cent. Those are not specialty items. They are the inputs behind structural steel, metal deck, light-gauge framing, drywall assemblies and formwork.

This is the second round. An earlier surtax order took effect in February 2025 at 25 per cent, and most of it was repealed with effect from 1 September 2025, with the surtax on steel and aluminum retained. Anyone maintaining an escalation position on a long-running project is therefore working across at least three distinct regimes, each with its own in-force date. The in-force date is not administrative detail — as the companion article on entitlement explains, it is the fact that decides recovery.

A tariff is paid by the importer of record in the country imposing it. This sounds obvious and is routinely lost in industry commentary, which tends to treat "tariffs" as a single undifferentiated cost pressure on Canadian construction.

United States tariffs on Canadian goods are paid by United States importers. They damage Canadian exporters through lost volume and price pressure, and they matter enormously to a Canadian steel fabricator selling into the American market. But they do not appear as a line of cost on an Ontario project buying domestic material for domestic installation.

What raises an Ontario contractor's costs is the Canadian counter-measure on United States-origin material. The contractor buying American flat-rolled steel, American gypsum board or American plywood is the one whose input price moves. That is the exposure to model, and it is the exposure that a contract's taxes-and-duties provision addresses.

The practical consequence is that the first question on any package is not "are we affected by tariffs" but "what is the country of origin of the goods in this package, and when were they imported". Those are answerable questions. They are answerable from purchase orders, supplier declarations and customs documentation — and they are far more answerable now, prospectively, than they will be in eighteen months when someone is trying to reconstruct them.

Statistics Canada's building construction price index for the second quarter of 2026 recorded non-residential prices rising 1.4 per cent in the quarter and 3.5 per cent year over year — an acceleration from 0.9 per cent in the first quarter. Metal fabrications posted the largest divisional gain at 2.1 per cent, which Statistics Canada attributed explicitly to upward pressure from retaliatory tariffs already in place.

Residential moved differently: 0.5 per cent in the quarter and 2.3 per cent year over year, decelerating rather than accelerating. Toronto residential construction costs actually fell 0.8 per cent in the quarter.

Two conclusions follow, and they point in different directions from the single blended escalation rate most estimates still carry.

First, the divergence between residential and non-residential is now large enough that a single rate applied across a mixed portfolio will materially under-provision the institutional and infrastructure work. Second, and more importantly, that index closed before the 8 September measures applied. It describes the world as it was under the previous, narrower set of duties. Any escalation allowance calibrated to it is calibrated to a regime that no longer exists.

There is a temptation, in the face of that, to wait for better data. The difficulty is that a fixed-price commitment made today is priced today. Waiting for the third-quarter index to confirm what the tariff schedule already tells you is not prudence; it is accepting an unpriced risk for a quarter.

Tariff exposure is often discussed as though it raises the cost of a building by some general percentage. It does not. It raises the cost of specific commodities, which sit in specific packages, which sit at specific points in the schedule.

On a typical institutional or commercial project, the concentration is in structural steel and metal deck; light-gauge steel framing and drywall assemblies; formwork and falsework where engineered wood is used; curtain wall and architectural metals where aluminum extrusion is involved; and mechanical and electrical equipment with significant fabricated metal content.

That concentration is useful. It means the analysis is tractable — a handful of packages rather than an entire budget — and it means mitigation can be targeted. It also means the schedule matters: a package procured before an in-force date sits on one side of the risk line, and the same package procured a week later sits on the other.

This is where controls and commercial functions have to be talking to each other rather than reporting in parallel. The procurement schedule and the cost forecast need to share a work breakdown structure, so that a change in a supply commitment is visible as a movement in both. Where they are maintained separately — which is still common — a tariff-driven cost increase surfaces at month end as a variance with no obvious cause.

The cost pressure is arriving into a market that is not uniformly weak or uniformly strong, and the difference shapes how much of it can be passed on.

Residential is contracting. CMHC's summer outlook forecast housing starts falling from roughly 241,400 in 2026 to 223,400 in 2027 and 211,900 in 2028 — the first outlook in some years to project a multi-year decline rather than a single soft year. The Greater Toronto condominium market is the sharpest expression of it: new condominium sales in July 2026 ran roughly 80 per cent below the ten-year average, against standing inventory equivalent to about 36.5 months of supply.

Institutional and infrastructure work is the opposite story. Ontario building permits rose 28.5 per cent month over month in June 2026, driven by roughly $1.3 billion of approved medical facility permits in the Toronto area. Permits lead shovels by a year or more, so that volume lands through 2027 and 2028.

For anyone pricing work, the implication is uncomfortable but clear. The sectors with the most tariff-exposed content — institutional, industrial, infrastructure — are the sectors with the strongest forward pipeline, which supports pricing discipline. The residential sector, where competition for a shrinking pool of work will be fiercest, is where the temptation to absorb escalation risk in order to win work will be greatest. That is precisely the wrong combination, and it is how contractors acquire loss-making backlog.

A common assumption is that a softening market will at least deliver cheaper and more available labour, offsetting some of the material pressure. The current data does not support it.

Ontario's construction unemployment rate improved to about five per cent over the summer of 2026 — effectively full employment for the trades — even as residential starts fell. BuildForce Canada's 2026 to 2035 outlook puts Ontario's total hiring requirement at roughly 126,100 workers against approximately 92,000 retirements, leaving a projected shortfall of about 27,300. Non-residential employment is forecast to peak around 2029 and 2030 on transit, utilities and healthcare work.

The age profile is the part that should concern anyone baselining long-duration work. Over the year, Ontario construction workers under thirty declined by roughly 24,400 while workers over fifty-five increased. The entry pipeline is narrowing as the demand peak approaches.

The planning consequence is that durations set from historical productivities assume a crew that can actually be staffed. Where the trade market is tight, the same activity takes longer with a less experienced crew, and the effect compounds through a congested sequence. A schedule baselined today for completion in 2029 or 2030 should carry an explicit labour availability contingency, identified as such, rather than a general inflation allowance quietly standing in for one.

None of this is an argument for a new reporting layer. It is an argument for making a small number of things visible that are usually not.

The first is origin. A register of tariff-exposed commodities by package, with country of origin, supplier, import date and the applicable order, is a modest document that becomes the foundation of any later claim. Built prospectively it takes hours. Reconstructed after the fact, through distributors who did not itemise duty on their invoices, it may not be buildable at all.

The second is the procurement-to-schedule link. Long-lead commitments should be visible as schedule activities, so that a supply decision and its cost consequence appear in the same place.

The third is the contractual position, which most projects do not establish until they need it. Whether a contract adjusts the price for changes in duties, from what date, and on what notice, is knowable on day one. Establishing it at mobilisation costs nothing. Establishing it after the invoice arrives is how entitlement is lost — a subject the companion article takes up in detail.

The fourth is honesty in the forecast. An exposure that has been identified but not yet priced belongs in the report as an exposure, not folded into contingency where it becomes invisible. Contingency absorbs the unknown. A 50 per cent duty on a commodity you have already specified is not unknown.