Labour supply barely grows again before 2040

In the fourth quarter of 2021 something happened that had not occurred since Dutch statistical measurement began: open vacancies outnumbered unemployed people. Two years later that still held, with 416,000 vacancies against 366,000 unemployed in the third quarter of 2023.

This is not a snapshot of an overheated economy. DNB shows it is structural. Growth in labour supply falls quickly in the coming years and comes close to a standstill in the decade after 2030, driven by ageing and by the lower birth rates of earlier decades. The room to offset that with higher participation is limited, for the simple reason that Dutch participation is already exceptionally high by international standards.

Where the labour force grew by more than 100,000 people a year for four decades, that growth falls close to a standstill in the decade after 2030.

The consequences reach past recruitment. DNB calculates that trend growth in gross domestic product could fall to around half a percent a year, against roughly 1.5% over the past two decades. Growth per head of population comes close to a standstill in that scenario.

The interesting part is not the conclusion but the arithmetic underneath it. To hold growth per head of population at around 1%, one of the following would have to happen.

Participation would need to rise 11 percentage points by 2040, meaning 100,000 more people in work every year. That route runs into the level the Netherlands already occupies. Around 75% of people aged 15 to 74 work or look for work, and among those aged 15 to 64 it reaches 85%. Of the more than 3 million people not active, most are retired, unable to work, or studying. The unused potential among people who do want to work but are not active is estimated at just under 100,000.

Or every worker would need to work 5 more hours a week, reversing a trend running fifty years. That route runs into what people themselves want. Around half of Dutch workers hold a part-time job, and part-timers say they want on average 2 to 2.5 extra hours a week. That would lift labour volume by 3 to 4%.

Or 1.5 million additional labour migrants would be needed by 2040. An estimated 800,000 labour migrants work in the Netherlands, a figure that more than quadrupled between 2006 and 2021, averaging growth of about 40,000 a year. The arithmetic asks for 100,000 extra a year. And the countries they come from in Central and Eastern Europe are ageing faster than the Netherlands.

Or productivity growth would need to rise from 0.5% to 1.4% a year. DNB calls it the most desirable route, and at the same time one that would require reversing a decades-long trend.

What it means when demand adapts to supply is clearest in business services, where staff are not a cost line but the product itself.

In the final quarter of 2025, almost half of firms in the sector were short of people. That holds both in labour-intensive parts such as cleaning and security and among specialists: accountants, lawyers and civil-law notaries. The sharpest shortages in early 2026 sit with security guards, cleaners and gardeners.

The consequence is not that vacancies stay open longer. It is that less work gets taken on. Firms cannot meet demand, so they accept fewer new clients and pick their assignments more carefully.

The brake on growth is no longer market demand. It is whether you can staff the work.

The second adjustment runs through price. Scarcity combined with rising minimum wages and higher negotiated pay has pushed labour costs up sharply. Cleaning rates in 2025 averaged 7.5% above the year before. In staffing, accountancy and law they rose by close to 5%.

The third adjustment is automation, and it splits the sector in two. For accountants, consultants, notaries and lawyers, AI can absorb routine work. In cleaning and security, automation is far harder, so dependence on people stays high.

Here is the point that matters most for recruitment. In pure numbers the distance between supply and demand is small: against every 100 open vacancies stood roughly 114 unemployed people. In quality the distance is wide, because the skills those people hold do not line up with what those vacancies ask for.

The broader figures point the same way. Recorded unemployment stood at 3.2% in 2025, while labour market slack, which includes part-timers who want more hours and people available but not searching, stood at 9.3%. So the people exist. They are simply not visible as job seekers.

Mobility only partly solves this. Around 4.5% of the labour force changes employer each quarter, but workers move mainly between related sectors and far less from shrinking ones to growing ones.

Training is therefore the hinge, and it is unevenly distributed. In 2022, 36% of the working population took part in some form of learning. Among workers on permanent contracts that was 29%, against 57% among those on flexible contracts. For those under a year in post it was 52%, against 25% at ten years or more. Those who stay longest in one place develop least.

• Capacity becomes the sales argument. When clients turn work away because they have nobody to do it, the question is not what recruitment costs but what the missed revenue costs.
• The reserve is not on job boards. The gap between 3.2% unemployment and 9.3% slack consists largely of people not actively searching.
• Hours are a separate source of supply. Part-timers want 2 to 2.5 extra hours a week on average. At a client with ten part-timers that is nearly half a full-time role, without a single new candidate.
• Sector boundaries are the real barrier. People do change employer, but rarely sector.
• Automation changes the profile unevenly. In specialist services demand shifts toward people who can work with AI. In cleaning and security the volume of people remains the question.

The structural analysis and the scenario arithmetic come from the DNB Analysis on labour scarcity of February 2024, with reference periods between 2021 and 2023. That publication is under copyright, so it is summarised and cited here, not reproduced. The business services figures come from the ING sector analysis of 9 April 2026. The 2025 unemployment and slack figures come from Eurostat, dataset lfsi_sla_a.

Because the sources carry different reference periods, each figure is dated where it appears. Figures from 2023 do not describe the position in 2026 and are used for the structural argument.