UK Construction Labour Shortage 2026: State of the Workforce Report

The workforce is the same size as 2019 and its shape has changed

Construction supported 2,293,000 workforce jobs in the second quarter of 2026, against 2,325,000 at the end of 2019. On the headline measure the industry is 1.4% smaller than before the pandemic. The Labour Force Survey, which counts people rather than jobs and has struggled with response rates since 2020, puts employment at 2,070,000, 10% below 2019. We use workforce jobs because it is the ONS preferred measure, but the gap between the two is itself a finding: the industry does not know its own size to within 200,000 people. Women are 15% of employment, up from 13% in 2019, almost entirely in office roles.

What has changed is the shape. Self-employed jobs fell from 833,000 to 619,000, a 26% drop, while employee jobs rose by 183,000. The self-employed share of the workforce went from 36% to 27%. Three things drove it: the off-payroll working rules extended to the private sector in April 2021, tighter Construction Industry Scheme verification, and main contractors employing core trades directly after the 2021 and 2022 labour scares. A quarter of the old sole-trader pool now sits on a payroll, with employer National Insurance at 15% since April 2025.

The shortage is real for specialists, not for labour in general

The vacancy rate is the cleanest test of a shortage that official data offers, and on its own it says the shortage has gone. Construction posted 30,000 vacancies a month in 2026, 1.8 for every 100 employee jobs. That is the same rate as 2019 and well under the 2022 peak of 2.9, when 46,000 posts were open. The economy as a whole runs at 2.2. On this measure construction is now easier to recruit for than the average employer.

That conclusion is true and incomplete. The Vacancy Survey counts posts employers have decided to fill at the wage they will pay. It cannot see a package a subcontractor declines because it cannot staff it, or the self-employed at all. So we built a second gauge. The Archdesk Labour Pressure Index compares the jobs the industry would need to deliver its current output at 2019 productivity with the jobs it actually has (Exhibit 3). It stands at 104 in the latest quarter against 100 in 2019, having peaked at 107 in 2022. The two gauges agree on timing (correlation 0.67 across 46 quarters) and disagree on the level: vacancies say balance, the index says the same workforce is still carrying 4% more output than in 2019.

Our reading is that the shortage is real but uneven, and the aggregate numbers hide it. General labour is available: operatives' pay has risen with the minimum wage and nobody sponsors a visa for a labourer. Specialists are not: the occupations employers went abroad for in 2024 and 2025 are engineering technicians, electrical engineers, electrical trades, carpenters and joiners, civil engineers and supervisors (Exhibit 4), and those are the packages that are let late on the projects we manage. The aggregate has eased because new work is down 6% on last year and firms have stopped hiring, not because the specialists arrived. When output turns, and CITB expects it to turn in 2027, that relief ends and the specialist gap is what is left.

To make that testable we scored sixteen occupations on the three signals the public data offers: how many extra workers CITB says each needs relative to its size, how often employers sponsor visas for it, and whether its real pay has risen (Exhibit 5). Per head, construction project managers, civil engineers and floorers and wall tilers top the table; in absolute numbers the electrical and carpentry trades dominate sponsorship because they are the largest occupations. The scorecard is published with the index and will be re-run each quarter.

Pay rose 24% in cash and fell 5% in real terms

Construction regular pay rose from £617 a week in 2019 to £765 in 2025, a 24% cash increase that many contractors will recognise from their own labour costs. Deflated by CPIH it is a 3% cut, and by July 2026 real pay was 5% below its late-2019 level. The whole economy went the other way: 5% above. The construction premium over the average employee, 22% in 2019, was 9% in 2026.

The occupation data explains where the money went. Of the 16 construction occupations with usable ASHE estimates, 9 pay less per hour than the median UK employee (£17.96). Bricklayers earn a median £16.64, 6% less in real terms than in 2021. Electricians held their ground at £19.19. The grades that gained were the ones the National Living Wage pushed: construction operatives and scaffolders rose 7% to 11% in real terms, so the skilled premium over general site labour has halved (Exhibit 7). ASHE covers employees only, and the self-employed majority in bricklaying and carpentry may have done better on day rates; the employed record is nonetheless the one lenders, insurers and the ONS use.

Pay that falls in real terms during a supposed shortage means employers have found other ways to absorb the pressure: longer hours, older workers staying on, and margins. Exhibit 8 shows the compression trade by trade. Those absorbers are temporary. Our expectation, stated as a view rather than a forecast, is that trade pay in the sponsored occupations rises 5% to 8% in cash terms in the first year of the recovery, because the catch-up happens when the work returns rather than gradually.

Identifiable inflow covers less than half of the requirement

CITB's outlook needs 41,200 additional workers a year to 2030, 206,000 in total. That figure already includes replacing those who leave; only about 15,084 a year of it is growth in the workforce. We tested that against the inflows the official record can count (Exhibit 9). Apprenticeship starts in construction subjects have been flat for four years, at 26,100 in 2021/22, 24,500, 24,200 and 24,590 in 2024/25, with 13,110 achievements in the latest year; the first half of 2025/26 was up 7.7%. Skilled Worker visas to construction sponsors were 1,628 in 2025, down from 3,155 in 2023, and 726 in the first half of 2026 after the July 2025 rules raised the skill threshold. Net growth in non-UK payrolled construction employment added about 3,300. Together that is 18,038 a year, 44% of the requirement.

HMRC's payroll data shows 91,000 EU nationals in construction employment in December 2025, 13% more than in December 2019, and 71,500 non-EU nationals, more than double (Exhibit 10). Non-UK nationals are 12% of the payrolled workforce, up from 9%. The widely quoted loss of 200,000 EU workers since Brexit is not visible among employees. If it occurred, it occurred among the self-employed, where the evidence is survey-based and thin. What is visible is that the route which replaced EU labour, sponsored non-EU visas, has been narrowed, and that the Temporary Shortage List which keeps construction trades eligible expires on 31 December 2026.

On the outflow side the payroll record shows a workforce that is ageing slowly rather than collapsing: 32.4% over 50 and 4.4% over 65, against 31.5% and 3.4% in 2019, with the under-25 share at 13% (Exhibit 11). The arithmetic is transparent. The 4.4% aged 65 and over will leave within a few years, and a fifth of the 28% aged 50 to 64 will follow within five, which is roughly a tenth of the employed workforce. The under-25 intake is 13%, and not all of it stays. The difference must come from other industries or from migration, and the second of those is closing.

Where it bites: region and county

National averages hide a country that is moving in two directions (Exhibit 12). Construction jobs grew 22% in the North East and 10% in the East Midlands between 2019 and 2026, and fell 21% in the West Midlands, 14% in the South West and 12% in London. Regional workforce jobs estimates are noisy, so we would not defend the decimals, but the direction is consistent with where infrastructure and public capital programmes have been spending.

The regional index (Exhibit 13) turns the jobs numbers into pressure. London reads 123, because output held up while jobs fell, and South West 115. North East reads 90 and East Midlands 92: jobs grew faster than the work. Where the index is high, London and South West on the 2025 reading, the specialist trades are the first to run short and packages go late; where it is low, the same trades are available at last year's rates.

The county map (Exhibit 14) shows why a national labour strategy is the wrong unit. Construction is 11% of all employment in North Lanarkshire and above 8% across a belt from Lanarkshire through Wigan to Essex and Thurrock. It is under 2% in the City of London, Southwark and Tower Hamlets, where the workforce is drawn in from outside every morning. The explorer below compares any regions, indexes them to a common year and downloads the series.

What it costs a contractor

The cost of the labour squeeze does not show up as a wage line. It shows up as output per job, which is 4% above 2019 and has been above 2019 for five years, and as insolvencies. 3,949 construction companies entered insolvency in England and Wales in 2025, 16% of all company insolvencies, against 3,221 in 2019. The rolling twelve-month count to August 2026 was 3,866, easing but still 20% above the 2019 level.

On Archdesk the same pressure shows at package level: labour and subcontract packages are where actual cost most often exceeds the tender allowance, and subcontractor failure is the most common reason a package is re-let. A re-let costs twice, the premium on the replacement and the programme slip while it is found. We do not yet publish those figures as a series; our platform observations here corroborate the official data, they do not replace it.

What to do about it in 2026

The findings point to six actions for a contractor bidding and building through 2027. They are listed in the order we would take them.

1. Price 2027 work with a labour catch-up, not a labour average. Tenders priced on 2025 pay data will be delivered into a market where real pay has fallen 5% and will correct. Carry a 5% to 8% allowance on the sponsored trades and a fluctuation clause on anything over twelve months.
2. Treat the shift to payroll as a retention asset. Three-quarters of construction jobs are now employee jobs. The cost is sunk; the benefit is only realised if payrolled trades are trained, progressed and kept. Measure voluntary leavers by trade monthly, not annually.
3. Fund completion, not just starts. Achievements run at about half the level of starts (13,110 against 24,590 in 2024/25). A contractor that sponsors a cohort through to completion gets a trade at 2026 rates for a decade; one that sponsors starts gets a recruitment cost.
4. Use the visa window before 31 December 2026. Construction trades stay eligible for Skilled Worker sponsorship only while the Temporary Shortage List runs. The numbers are small, but for a firm short of carpenters or bricklayers a sponsorship licence now is cheaper than agency rates in 2027.
5. Price and staff by region. The change in construction jobs since 2019 ranges over 44 points between regions (Exhibit 12). Tender labour rates and programme float should differ too. A national rate book is a subsidy from the North East to the West Midlands.
6. Forecast labour cost-to-complete monthly. The 2022 squeeze was visible in subcontract quotes six months before it reached accounts. A monthly cost-value reconciliation with a live labour cost-to-complete is the earliest warning available, and it is the one control that turns this article's index into a project decision.

Outlook for 2027

CITB's central forecast has output flat in 2026 (-0.2%), then growing 1.8% in 2027 and 2.8% in 2028 (Exhibit 15), with the workforce reaching 2.68 million by 2030. Our base case takes that demand path and combines it with what this article has shown about supply: a flat workforce, a visa route for trades that closes at the end of 2026 unless extended, retirement outflows running ahead of young entrants, and real pay that has not yet corrected. On those assumptions the Labour Pressure Index returns to its 2022 level, about 107, by the end of 2027, and the vacancy rate rises back above 2.3 per 100 jobs. Trade pay catches up in one or two steps rather than gradually, and the trades to watch are the ones employers were already sponsoring visas for in 2025: electrical and mechanical trades, engineering technicians, carpenters and joiners, bricklayers and supervisors.

Two risks sit either side. Downside: public capital programmes and the housing target slip again, output stays flat, and the pay correction is delayed into 2028 at the cost of more insolvencies. Upside: housing and infrastructure ramp together, the 2022 pattern repeats with no migration release valve, and the index goes above 110 for the first time.

These are the calls we are making, with the date and the measure we will use to grade them. Each quarterly update reports which are on track.

Questions this article answers

Short, self-contained answers with the figures and sources, for readers who want one number and for anyone quoting the article.

Is there a labour shortage in UK construction in 2026?

Yes, but it is uneven. The construction vacancy rate in 2026 is 1.8 per 100 employee jobs, the same as 2019 and below the whole-economy rate of 2.2, and real pay is 5% below its 2019 level, which is not what a general shortage produces. The Archdesk Labour Pressure Index reads 104 against 100 in 2019, so the same workforce is delivering 4% more output per job. The shortage is concentrated in specialist occupations, engineering, electrical and mechanical trades, carpentry and joinery and supervision, and in the regions where the index is highest, London and South West.

How many people work in UK construction?

ONS Workforce Jobs put construction at 2,293,000 jobs in Q2 2026, of which 1,671,000 are employee jobs and 619,000 self-employed. The Labour Force Survey, which counts people rather than jobs, gives 2,070,000. CITB puts the total workforce including all occupations at 2.61 million in 2025.

How much has construction pay changed since 2019?

Regular weekly pay in construction rose from £617 in 2019 to £765 in 2025, a 24% cash increase, but after CPIH inflation it is 5% lower than in late 2019 while whole-economy pay is 5% higher. The median hourly pay of a bricklayer was £16.64 in April 2025, 6% less in real terms than in 2021; an electrician's was £19.19.

How many extra construction workers does the UK need each year?

CITB's Construction Workforce Outlook 2026–2030 puts the annual recruitment requirement at 41,200 workers a year, 206,000 over five years, including replacement of leavers. Identifiable inflows in the official record, 13,110 apprenticeship achievements, 1,628 Skilled Worker visas and about 3,300 net non-UK payroll growth, total 18,038 a year, 44% of the requirement.

Did UK construction lose 200,000 EU workers after Brexit?

Not among employees. HMRC payroll data shows 91,000 EU nationals in construction employment in December 2025, 13% more than in December 2019, and 71,500 non-EU nationals, more than double. If a loss of that size occurred it was among the self-employed, who are not in PAYE data, and no register measures it. Archdesk treats the 200,000 figure as unverified.

What is the Archdesk Labour Pressure Index?

A quarterly measure of construction labour pressure published by Archdesk. It is 100 × (the workforce jobs needed to deliver current output at 2019 output per job) ÷ (actual workforce jobs), using ONS construction output and ONS Workforce Jobs. A reading of 100 is the 2019 balance; above 100 means the workforce is carrying more output than in 2019. It peaked at 107 in 2022 and reads 104 in Q2 2026. A regional version and a trade pressure scorecard are published with it.

What will happen to construction labour in 2027?

Archdesk's base case, using CITB's forecast of 1.8% output growth in 2027 and 2.8% in 2028, is that the Labour Pressure Index returns to about 107 by the end of 2027, the vacancy rate rises above 2.3 per 100 jobs, and pay in the sponsored trades rises 5% to 8% in cash terms in the first year of recovery. Five dated predictions are published in the article and graded each quarter.

Key figures at a glance

All figures: Archdesk analysis, version 2026.09, 28 Sep 2026. Cite as shown in the methodology section.