Analysis — interpretation, not data
August 31, 2026
Employers, industry groups and governments have spent years warning that Canada and the United States are running out of skilled tradespeople. Electricians are scarce. Plumbers are scarce. Welders are scarce. Construction cannot find enough workers.
But a labour shortage is not just a report that hiring is difficult. It is a claim about a market.
If there are too few skilled tradespeople nationwide, who is bidding for them?
The national wage data’s answer is: apparently, almost nobody.
Over the past decade, trades employment expanded substantially in both Canada and the United States. Employers absorbed hundreds of thousands of additional workers. Yet real pay barely moved, occupational wage premiums deteriorated, and the wages employers advertise for many supposedly scarce trades rose more slowly than the broader market.
That is not what a national bidding war for a scarce input looks like.
The distinction matters because “shortage” has come to cover several very different problems: a worker may exist but not in the right city; a qualified worker may prefer another industry; licensing may prevent workers from moving between regions; a government wage schedule may prevent an employer from raising an offer; or a contractor may simply be unable to afford the wage required to attract another electrician.
Those can all create vacancies. They do not all mean that the economy has too few skilled tradespeople.
Difficulty buying skilled labour at the price you want to pay is not evidence that skilled labour is in short supply.
No single statistic establishes labour scarcity. Wages, employment and total labour spending have to be read together.
If demand for a type of worker rises faster than the available supply, buyers have to compete for the workers who exist. Where wages can adjust, that competition should put upward pressure on compensation. Higher prices should also attract workers into the market: people change employers, occupations or locations, apprentices enter, and firms find ways to use the newly valuable labour more efficiently.
The result can eventually be a much larger workforce. A shortage does not have to last forever.
But the sequence matters.
Demand outruns available labour. Employers bid. Relative compensation rises. Workers respond.
A market in which employment expands rapidly while the price of labour stays flat or falls relative to other work is telling a different story. It is showing that additional labour was available at roughly the prevailing price.
That is what the national trades data mostly shows.
Over the decade to 2025, Canadian construction employment rose about 21%.
Real average earnings in construction rose only 0.8%. Among building equipment contractors — the electricians, plumbers and HVAC trades most often named in shortage stories — real earnings moved −0.6%.
The Canadian economy as a whole delivered +5.5%.
The occupational median is even harder to reconcile with a generalized shortage. The median Canadian worker in trades, transport and equipment occupations saw real wages fall 0.9%, while the median Canadian worker gained 6.4%.
The United States tells the same basic story. Building equipment contractors did better in absolute terms, with real pay up 6.3%, but employment expanded roughly 35% while pay still failed to pull decisively away from the broader economy.
Population growth does not explain the result away. On a population-adjusted basis, the trades workforce did not disappear beneath a larger population. Construction added workers.
This is important. The usual shortage story is implicitly a story about a quantity that cannot expand fast enough to meet demand.
The quantity expanded.
Relative pay is a cleaner test than asking whether nominal wages rose.
In 2000, the median Canadian worker in trades, transport and equipment occupations earned 9.2% more than the median Canadian worker. By 2019 the premium had fallen to +4.6%.
By 2025 the published medians were effectively at parity: $30.61 and $30.59.
In the United States, construction and extraction occupations went from roughly parity with the median worker in 2000 (+0.6%) to 10.1% below the median.
This does not mean skilled trades are low-paid occupations. Many are not. It means their position relative to alternative work has deteriorated during the same period in which employers have repeatedly described trades labour as increasingly scarce.
That is an unusual way for a scarce input to behave.
Other high-skill occupations do not show the same erosion. US computer and mathematical occupations maintained premiums of roughly +63% to +68%. Architecture and engineering remained around +65% to +61%. Canadian sciences remained near +47%. Registered nurses retained large occupational premiums in both countries.
The Canadian construction industry still pays roughly 15% above the economy-wide average, which is an important qualification. Industry and occupation measures count different people: construction includes managers and office workers as well as tradespeople, while the broad trades occupational group includes transport and equipment operators.
But that distinction does not rescue the shortage claim. The occupational workers supposedly becoming impossible to find have not become more expensive relative to everybody else.
Incumbent wages can adjust slowly. Contracts, annual raises, union agreements and internal pay scales can all delay the response to a changing market.
Job postings provide a more immediate test.
Statistics Canada publishes the wages employers advertise on vacancies by occupation. These are offers for jobs employers are trying to fill now.
Between 2016 and 2025, the average advertised wage across Canadian vacancies rose 13.9% after inflation.
Electrician offers rose 5.6%.
Carpenters: 6.7%.
Welders: 7.0%.
Plumbers and HVAC mechanics performed better and roughly kept pace with the broader market. But there is no general surge in the price employers are offering for supposedly unobtainable trades labour.
That is a much more direct problem for the shortage story than slow-moving incumbent pay.
Where wage-setting is flexible, persistent difficulty filling a position should create upward pressure on the offer.
For most of these trades, the advertised offer is not pulling away from the market. It is falling behind it.
Employers may sincerely want more workers. But wanting to purchase more labour at the current price is not evidence that the economy is incapable of supplying more labour.
One objection is that perhaps skilled-trades labour markets simply do not behave this way. Maybe wages are a poor signal. Maybe even intense scarcity can exist without employers bidding against one another.
Northern Virginia provides a useful answer.
Loudoun County sits at the center of the extraordinary buildout of data centers in Northern Virginia. Demand for electrical and specialty-trades labour has become unusually concentrated there, backed by some of the world’s best-capitalized companies and contractors.
Recent reporting from the data-center boom describes exactly the behaviour the national shortage narrative predicts: contractors losing experienced workers to better-paying projects, bonuses and per diems being used to attract labour, and employers competing directly for electricians.
And the aggregate data moves in the same direction.
Loudoun County specialty-trades employment rose about 165%. At the same time, the sector’s wage premium over the average American job increased from roughly +32.4% to +39.9%.
That is not an enormous aggregate wage increase compared with some of the individual compensation figures reported from data-center projects. That is itself informative.
The spectacular anecdotes come from a small subset of workers on a small subset of projects in one exceptionally hot regional market. Once the denominator expands even to all specialty-trades employment in Loudoun County, the wage effect becomes much more modest.
But it is still visible.
This is what the mechanism looks like when employers genuinely compete for a locally scarce pool of trades labour: workers are poached, compensation rises, workers move toward the market, and employment expands rapidly.
The absence of that pattern nationally cannot be dismissed by saying trades labour markets simply do not respond to scarcity.
Loudoun shows that they do.
The national market just does not look like Loudoun.
The Permian Basin provides a different experiment.
Oilfield work around Midland, Texas pays extremely well in absolute terms. Oilfield-services compensation is about $121,000 a year, and the region experienced an enormous expansion in demand.
Workers responded.
Midland County oilfield-services employment more than doubled.
Yet the sector’s premium over the average American job fell from about +80% to +53.2%, and its real weekly wage ended below where it began.
This is not evidence that oilfield labour lacked value. Quite the opposite: employers paid unusually high wages and attracted an extraordinary quantity of labour into the region.
But once again, high demand and high wages are not the same thing as a persistent shortage.
Midland demonstrates how a market can absorb an enormous demand shock through a quantity response. Workers move toward valuable work. Headcount expands. A previously exceptional wage premium can shrink even while the industry remains very well paid.
Even one of the industries most capable of paying high absolute trades wages did not produce persistent real-wage acceleration over the period.
That matters when national shortage claims are justified with anecdotes about how much particular oilfield workers earn.
A $120,000 job can be extremely well paid while still becoming cheaper in real terms.
Quebec electricians move differently.
Between 2016 and 2025, advertised real wages for electricians in Quebec rose about 26.2%, compared with 15.8% across the provincial vacancy market.
Ontario, Alberta and British Columbia did not show the same result.
Quebec is also institutionally unusual. Its construction labour market combines collective wage-setting with a restrictive competency and regional mobility system. Construction workers are organized into geographic labour pools, and access to work across regions has historically been constrained in ways that do not exist to the same degree elsewhere in Canada.
That does not establish that Quebec’s licensing and mobility regime caused the stronger wage response. Collective bargaining, construction demand and other provincial differences move at the same time.
But it suggests a useful hypothesis.
If employers cannot draw as freely from workers elsewhere, additional demand should be more likely to appear in the price of local labour.
Quebec is therefore not an awkward exception to the national result. It is a case worth studying precisely because its labour market is less geographically substitutable.
Recent reforms loosening regional mobility may eventually provide an even cleaner test.
Loudoun, Midland and Quebec are more informative together than separately.
Loudoun shows unusually concentrated demand producing visible bidding for particular workers.
Midland shows very high wages attracting a massive expansion in labour supply, until the quantity response overwhelms the idea of a persistently constrained workforce.
Quebec shows unusually strong electrician wage growth inside a labour market where geographic substitution is institutionally restricted.
These are the kinds of differences a functioning labour market should produce.
They also expose the problem with treating every difficult vacancy as the same phenomenon.
A contractor who cannot persuade an electrician to move to an expensive remote town without a large premium does not face the same problem as a shipyard bound to an administered wage schedule. Neither faces the same problem as a data-center contractor being outbid by another project down the road. And none of them necessarily imply that Canada or the United States has too few people capable of becoming electricians.
Calling all four a “skilled-trades shortage” throws away most of the useful information.
There are skilled tradespeople earning extraordinary incomes.
That fact is sometimes offered as evidence of a national shortage. The distributional data suggests almost the opposite conclusion.
For Canadian trades occupations, the gap between average and median pay widened from about +3.6% in 2000 to +9.3% in 2025. The upper end of the distribution has pulled away from the middle.
But the gap remains much smaller than in the economy as a whole, where the mean–median difference grew from +11.3% to +19.0%.
The high-paying niches are real: oil fields, remote industrial work, major projects, defense work, ports, data centers, extreme overtime.
They are also not representative of the national trades workforce.
Loudoun makes that especially clear. Some individual projects can produce spectacular compensation packages while the aggregate wage premium for the county’s entire specialty-trades workforce moves far less.
Anecdotes about workers earning $200,000 do not establish the market conditions facing millions of tradespeople.
They identify the places where buyers really are prepared to pay.
The demographic version of the shortage argument says the problem may not be visible yet: tradespeople are old, retirements are coming, and the workforce is about to collapse.
The current workforce does not look unusually old.
Workers aged 55 and over account for about 21% of Canadian construction, compared with 24% across the economy.
Construction has also been adding workers quickly enough to keep replenishing its workforce.
That does not mean retirements create no replacement demand. Every large occupation must replace retiring workers, and particular trades or regions can face difficult age structures.
It means the national construction workforce does not presently exhibit the unusually old age profile implied by the common “everyone is about to retire” story.
Replacement demand is real.
It is not evidence of an existing national shortage.
American construction workers quit less often than the average private-sector worker: roughly 2.1% against 2.5%.
Canadian construction job tenure is approximately in line with the all-industry average.
Construction does have unusually high layoffs and discharges. That is consistent with an industry organized around projects, contracts and seasonal work rather than unusually high voluntary departures.
Again, there are individual employers who cannot retain workers.
The national data does not show a trades workforce collectively abandoning the industry faster than other workers abandon theirs.
One way to reconcile rising employment with flat wages is straightforward: labour supply expanded.
Immigration is part of that story.
Approved temporary-foreign-worker positions in Canadian construction peaked at about 0.78% of the workforce in 2024, compared with 0.60% economy-wide. Construction’s use of the program increased substantially.
But the programme remains too small by itself to explain construction employment rising roughly 21%.
Other immigration channels matter and are more difficult to assign by industry. Open work permits, permanent migration and new entrants can all expand labour supply.
That is not a rebuttal to the argument here.
If immigration or domestic training supplied the workers needed for employment to expand without unusual wage bidding, then the labour supply expanded.
A market whose supply successfully expands with demand is not usefully described as suffering a persistent shortage simply because employers would prefer an even larger supply.
The main payroll wage series excludes self-employed workers, an important qualification in construction, where self-employment is common.
About 27.7% of Canadian construction workers are self-employed.
But the share has been stable, and the employment result barely changes when they are included: roughly +20.8% growth among payroll employees versus +20.7% counting everyone.
The missing shortage is not hiding in a mass migration from payroll employment into self-employment.
Construction can complain about rising labour costs while individual workers see little improvement in pay.
Those statements are perfectly compatible.
Total labour compensation rose substantially because the sector employed far more people. An industry that hires 20% more workers can have a much larger wage bill without increasing pay per worker by much at all.
That distinction matters because employers experience labour as a cost line.
A company whose total payroll has risen sharply may quite reasonably say that it is spending far more on labour. But that does not mean the price of a unit of labour has risen sharply.
Headcount, pay per worker and total compensation tell different stories.
The national trades story is primarily a story of more workers, not a dramatically higher price for each worker.
Nursing is useful because it demonstrates a genuinely different failure mode.
Canadian health care has very low unemployment alongside unusually high vacancy rates. Positions go unfilled and the consequences for patients are real.
Yet advertised real wages for registered nurses fell 9.5% from 2016 to 2025, while nurse vacancies surged.
That is not evidence that labour markets magically stop caring about price.
It is evidence that some buyers cannot freely change the price.
Public health budgets, negotiated wage grids, staffing rules and government finance constrain wage-setting. A hospital can desperately need another nurse while lacking authority or money to bid enough to obtain one.
Calling that situation a shortage may be entirely reasonable operationally. There is a bed and no nurse available to staff it.
But the diagnosis matters.
If the binding constraint is an administered wage that cannot clear the market, training additional workers and raising the wage are not interchangeable solutions.
The same distinction applies to skilled trades.
A federal shipyard with vacancies under an administered wage schedule can face a genuine staffing shortage even while its wage premium declines.
That does not establish a national shortage of welders.
It establishes that this employer cannot purchase as much welding labour as it needs under the price-setting system it operates inside.
The national skilled-trades shortage is difficult to find because the word has gradually stopped meaning scarcity in a labour market.
It often means:
These are all legitimate things to measure.
They are not the same thing.
And aggregating them into a single national “skilled-trades shortage” produces a policy diagnosis almost guaranteed to favour increasing the supply of workers over asking why the buyers of labour are not increasing its price.
None of this says that every trades worker is well paid.
It does not say every employer can fill every vacancy.
It does not say apprenticeships work well, licensing is optimal, geographic mismatch is trivial, housing costs do not matter, or that every region has enough electricians.
It says the much larger claim — that Canada and the United States face a broad shortage of skilled tradespeople — does not fit the observed national labour market.
Employment expanded substantially.
The trades did not become more valuable relative to other work.
Advertised wages for many supposedly scarce occupations failed to keep pace with the broader market.
The workforce is not unusually old.
Workers are not unusually eager to leave.
And where unusually intense local demand actually does collide with a limited pool of workers, as in Northern Virginia’s data-center boom, employers behave exactly as economic reasoning predicts: they bid.
That is the positive control.
If skilled tradespeople were broadly too scarce nationwide, the extraordinary thing would not be that some electricians somewhere can earn $200,000.
The extraordinary thing would be that millions of supposedly scarce workers could remain at stagnant relative wages while employers spent years insisting they could not find enough of them.
There is a simpler description.
Employers want more skilled labour than they can or will purchase at the price required to attract it.
Workers have a name for that too.
A wage shortage.