Who leaves decides the shortage

The Netherlands employs 8,800 butchers. The trade is barely growing, yet the labour market for butchers stays tight to 2030. The reason is not new jobs but departures: 8.1% of butchers must be replaced every year, the highest replacement demand of all 112 occupation groups. The average across all occupations is 2.8% a year.

The example is not an outlier. For most tight occupations, the engine behind the vacancies is who leaves, not what grows.

The figures come from two sources joined at the end of 2025. The Research Centre for Education and the Labour Market (ROA) published its biennial forecast De arbeidsmarkt naar opleiding en beroep tot 2030 on 25 November 2025. On 18 December 2025, UWV set that forecast against its own tightness measure, the Spanningsindicator, in Ontwikkeling krapte naar beroep 2024-2030. For the first time, current tightness and its expected path to 2030 can be read per occupation group.

The join covers 112 occupation groups in the Dutch BRC classification. For 19 of them UWV computes no representative Spanningsindicator, for instance for checkout staff, 66% of whom are pupils or students. That leaves 93 scored groups.

The starting position is one-sided. In the fourth quarter of 2024 the market was loose for 1 group, average for 2, tight for 34, and very tight for 56. So 90 of the 93 scored groups sat on the tight side. For 32 of the 112 groups, the market is tight now and forecast to tighten further to 2030.

Within that picture, replacement demand does the sorting. Butchers top the list at 8.1% a year, followed by bus and tram drivers at 7.0%. At the bottom sit advisers in marketing, public relations, and sales: 1.1% a year against the 2.8% average. Their market was tight at the end of 2024 and eases slightly to 2030, despite the demand for commercial staff.

The sharpest evidence comes from occupations that shrink and stay tight anyway. The number of representatives and buyers falls by 1,900 workers from 2025 through 2030, a negative expansion demand. Yet their classification moves from very tight to slightly less tight, not to loose. Gardeners and growers lose 3,600 jobs in six years and tighten regardless, because 19,800 people must be replaced over the same period.

Job openings arise in two ways: an occupation grows, or people leave it. UWV names retirement, disability, care duties, and the switch to another occupation as the drivers of that outflow. In a country where 19.9% of workers are aged 55 to 64, the figure this library's Replacement demand started from, the second stream is structurally larger than the first.

The forecast shows it per occupation. Primary school teachers face 43,300 replacements to 2030, 4.4% a year on average, and their market stays tight. Sports instructors run at 2.1% and ease slightly. Managers of care institutions (1.7%) ease for the same reason. ICT managers grow fast, but a replacement rate of 1.2% a year outweighs the growth: on balance, their market eases too. Where outflow is low, the market cools. Where it is high, it stays tight, almost regardless of what employment does.

The two instruments do not tell the same story everywhere. Of the 24 technical occupation groups, 16 were very tight and 7 tight at the end of 2024; only construction and industry helpers sat at average. The forecast expects 12 groups to tighten further, 7 to hold roughly level, and 5 to ease, welders and sheet metal workers most of all.

That last movement has not been measured yet. UWV writes that the forecast easing for these five technical groups was, as of the third quarter of 2025, not yet visible in the Spanningsindicator.

In ICT the opposite has happened. The three scored ICT groups, database and network specialists (93,400 workers), software and application developers (371,500, the largest occupation group in the country), and ICT user support (53,500), were all very tight at the end of 2024. All three are forecast to ease slightly, mainly on low replacement demand. And for ICT user support it was already visible in the third quarter of 2025: the classification dropped from very tight to tight. Together the three groups employ 518,400 people; the fourth ICT group, radio and television technicians, has no representative indicator and is not counted here.

So forecast and measurement agree in one occupational domain and not yet in another. This paper leaves both readings standing, each with its source and its reference date.

New in the 2025 edition are regional forecasts per occupation, computed by ROA and financed by UWV. Provincial tightness is measured as the average of all four quarters of 2024, not the fourth quarter alone. Setting provincial and national figures side by side therefore compares two reference periods.

Groningen had the least tight labour market of the twelve provinces in 2024, even though 19% of its workers hold a care or welfare occupation, against 15% nationally. For doctors, therapists, and specialised nurses, ROA expects easing in Groningen, Utrecht, and Gelderland; in the other nine provinces that market stays tight or tightens further. Engineers are very tight in 11 provinces and tight only in Groningen.

The regional figures have a floor: where the number of surveyed workers in the Dutch labour force survey averages below 40 over two years, ROA publishes no forecast. Coverage per segment is something to count, not to assume.

• For a tight occupation, check replacement demand before job growth. On average 2.8% of workers leave each year; anything above that keeps an occupation hard to fill, growth or no growth.
• A shrinking occupation is not a loose occupation. Representatives and buyers lose 1,900 jobs to 2030 and stay on the tight side; gardeners and growers lose 3,600 and tighten.
• In ICT, count on real, already measured easing: ICT user support dropped from very tight to tight in the third quarter of 2025.
• In technical occupations, do not run ahead of the forecast easing. It was not yet visible in the Spanningsindicator in the third quarter of 2025.
• In care, promise regional relief only where the forecast gives it: Groningen, Utrecht, and Gelderland. For the other nine provinces there is no basis.

The forecast figures come from De arbeidsmarkt naar opleiding en beroep tot 2030, published by ROA on 25 November 2025. The report itself may not be reproduced; the underlying dataset in ROA's labour market information system sits on DataverseNL under DOI 10.34894/DVQTOG, version 4.0 of 25 November 2025, published under CC BY 4.0. The join with current tightness comes from Ontwikkeling krapte naar beroep 2024-2030, UWV, 18 December 2025. UWV states in that publication that everything in it may be reused, but only with attribution; every UWV figure here carries that attribution.

The Spanningsindicator is the ratio of open vacancies to short-term jobseekers whose unemployment benefit has run for less than six months. The forward view is ROA's indicator of future staffing bottlenecks per occupation. The national baseline is the fourth quarter of 2024; the provincial figures average the four quarters of 2024. The two reference periods are kept apart throughout this paper.

Of the 112 occupation groups, 19 are not scored because the Spanningsindicator gives no representative picture there, among them checkout staff and loaders, unloaders, and shelf stackers, where many pupils and students work. The 518,400 ICT workers is the sum of three published group counts and is presented as a sum, not as a source total. The published total of nearly 545,000 ICT workers includes radio and television technicians and is therefore not a denominator for the three scored groups.

The projections to 2030 are expectations, not measurements. Where forecast and Spanningsindicator contradict each other in the third quarter of 2025, the text says so, with the source attached. The share of workers aged 55 to 64 (19.9% in 2025) comes from Eurostat lfsa_egan22d and was first used in Replacement demand.