Statecraft

31 August 2026 · analysis

Governing the governance

Why the Dutch central government grew by 45 per cent in eight years without performing better, and what a salami slice does not change about that

by Jacob Huibers · Lees in het Nederlands →

I. Two reports on the same Tuesday

On Tuesday 14 April 2026 Dutch civil servants stopped work. The Central Government Cleaning Organisation struck for 72 hours, other parts for 24; there were gatherings in six cities and the House of Representatives cancelled its debates that day. At issue was the pay freeze: no collective pay rise for 160,000 central government staff in 2026.1 That same day the Advisory Division of the Council of State published its Spring Report on budgetary supervision, containing a warning to a cabinet that wants to economise on the civil service: a generic cut leads to delivery problems, and Councillor Van Zwol cited the UWV, the benefits agency already behind on its disability assessments, as an example of where such a cut goes wrong.2

Two reports from the same building, and they do not contradict each other. The apparatus striking because it gets no pay rise and the apparatus that cannot be cut generically are the same apparatus. What the two reports together make visible is that the question of how large the Dutch central government should be is asked as a question of volume, while the answer lies in the proportion.

The figures first, because they are sharper than most of the debate about them. On 30 June 2025 central government counted 160,953 full-time jobs: 18,979 in policy, 10,498 in inspectorates, 17,209 in support and 114,265 in delivery.3 In April, Marike Stellinga set that figure beside the position at the end of 2017, 110,649 full-time jobs, arriving at growth of over 45 per cent in eight years.4 Within that total the policy core at the ministries grew by more than 66 per cent, delivery by almost 39.5 The series in the Annual Report on Government Operations shows the acceleration: 131,132 FTE in 2021, 138,376 in 2022, 147,841 in 2023, 157,019 in 2024 and 160,016 at the end of 2025. Growth in 2024 was 6.2 per cent; in 2025 it was 1.9, which the report calls “levelling off”.6

The political answer to this series has been the same three times over. The Rutte II cabinet cut a central government that had already been stable in size for decades, promising it would be the better for it; in hindsight it was established that far too much had been cut.5 The Schoof cabinet set itself a structural saving of 22 per cent on the apparatus and watched that apparatus grow by another 3,934 full-time jobs in the first half of 2025.7 The Jetten cabinet adds a structural target of €1.4 billion from 2030 on top of that, together with the earlier billion roughly ten per cent of the cost of the apparatus, and the Ministry of the Interior writes in its own annual report that it is uncertain whether this will succeed, because new tasks keep arriving and nobody knows precisely how many people they require. The president of the Court of Audit summed it up on Accountability Day: getting a smaller central government under these conditions is mopping with the tap running.8

Three cabinets, three times a smaller state as the aim, three times a bigger one as the outcome. Anyone reading that as political failure is missing what there is to see. An outcome independent of the colour of the cabinet is a property of the architecture.

II. Not the volume, the proportion

Where the growth sits is known more precisely than why it is there. Stellinga catalogued the explanations the department itself gives: the recovery operations for Groningen and for the childcare benefits scandal required extra people, external hiring of IT specialists rose to repair failing systems, in 2016 the state took its cleaners back in-house, the Open Government Act and data protection require work, and there are more motions and parliamentary questions. All true, and together insufficient, because the growth sits at every ministry, including the ones that did not get any busier.5

Two people who know the apparatus from the inside offered, in the same column, an explanation that reaches further. Roel Bekker, who as programme secretary-general for Renewal of Central Government led a savings target on the civil service himself between 2007 and 2010,9 calls it worrying that the ministries grew so much faster than delivery, and states: government has not come to perform better through this growth. Paul ’t Hart, vice-chair of the Scientific Council for Government Policy and chair of the project group that produced the 2025 WRR report Deskundige overheid,10 points to the natural growth tendency of bureaucracies and to politicians who promise everything and rarely say what civil servants may stop doing. Government deals wholesale in compensation schemes, he says; out of fear of voters no cabinet dares organise disappointment any more. And then the sentence that gives this paper its title. Since the great decentralisations of 2015, tasks such as youth care, home care and district nursing have gone to municipalities. “Since then the ministries have been governing the governance. In practice that too often means throwing hard-to-implement political compromises over the fence.”5

That sentence deserves to be read slowly, because it contains a mechanism that explains the size debate and at the same time renders it invisible.

A ministry that delivers has a natural measure: the number of applications, inspections, roads, prisoners. A ministry that governs whoever delivers has no natural measure. It produces frameworks, monitoring agreements, administrative consultations, powers of direction, progress letters to parliament, and, when the delivering layer fails, compensation schemes and recovery operations. Each of those products requires people, and none of them yields a performance that measurably improves with more people. The apparatus grows where it reproduces itself. That is the pattern described in Series I Nº 06 as reproduction inwards, now at the scale of the whole central government and with the Annual Report as the measuring instrument.11

A second movement joins it, described in Nº 07 as absorbed guilt without integration. The two largest growth items of recent years, the recovery operations for Groningen and for the benefits scandal, are organisational answers to the organisation’s own failure. The organisation absorbs the guilt, places a delivery body beside itself, and grows. The failure is caught in a new unit that leaves the old units, and the way of working that caused it, undisturbed. An apparatus that grows on its own mistakes has a growth engine no savings target touches, because a savings target sees only the size and not the cause.

The third movement comes from the other side of the Binnenhof. De gedissocieerde wetgever documented how the House of Representatives has shifted its energy from legislating to reflecting: from 598 motions in 1982 to over 5,000 in 2022, eleven motions for every amendment.12 Every motion, every parliamentary question, every rapporteurship requires a civil servant on the other side to answer. The growth of the policy core is in part the mirror image of a parliament that judges more and makes less. Two bodies serving each other, growing together without either being the better for it.

What these three movements have in common is that they all sit in the proportion and none of them in the volume. A central government of 160,000 people of whom 114,000 deliver is a different organisation from a central government of 160,000 of whom 19,000 make policy, 17,000 provide support and overhead runs between 45 and 50 per cent, as the then minister informed parliament in the autumn of 2024.13 The second figure is the real figure. The first is the figure people strike over and cut.

III. Why the salami slice feeds the pattern

A generic cut, a salami slice, takes the same percentage off every layer. In doing so it preserves precisely what it ought to change: the proportion. After a ten per cent salami slice, a slightly smaller policy apparatus governs a slightly smaller delivery, with the same overhead, the same compensation schemes and the same number of motions. The Advisory Division says as much in its Spring Report: a generic saving bites in one domain where that is responsible and in another where it puts the objectives out of reach, the accumulation of cuts exceeds the limit the CPB applies, and the cabinet will have to either adjust the cuts or name concrete tasks that will lapse.2 The CPB says it in figures: it does not consider the €0.4 billion saving on core departments and delivery bodies plausible as long as it has not been specified which government tasks will lapse, notes that large savings on the government apparatus have proved too ambitious in the past, and books €0.2 billion of the €1.4 billion.14

There is a second reason the salami slice cuts the wrong way, and it is sociological. A cut that hits everyone equally is not, in practice, distributed equally. Delivery has bodies: one prison officer fewer is a cell that is not watched. The policy core has arguments: one policy officer fewer is a letter to parliament that arrives later. In every savings process I have watched from close up, the argument beats the body, because the argument is at the table and the body is not. The layer that decides on the cut is the layer that shrinks least. That is the predictable outcome of a design in which the governance decides about itself.

Nils Brunsson described organisations of this kind as organised hypocrisy: talk, decisions and actions are decoupled and each serves a different audience.15 The talk says smaller state, the decision says €1.4 billion, the action grows by 3,934 FTE in half a year. Brunsson’s point is that this is a way of surviving contradictory demands at once: the voter wants a smaller state, parliament wants an answer to every motion, delivery wants people, and the organisation gives each of them something different. In his terms it would even be functional, were it not that the apparatus meanwhile does grow and does not improve. That is what a salami slice cannot change: it is itself a form of talk presented as a decision.

Bekker knows this from experience. The savings target he led between 2007 and 2010, and the targets of Rutte I and II after it, were followed within ten years by growth from 110,649 to 160,953. He established it himself in April 2025: government has never been as large as it is now, and the 22 per cent slimming from the coalition agreement is small beer by comparison.16 A system that, after an intervention, springs back past its starting point displays what in my old field is called the hysteresis of a regime: reversing the cause does not restore the state. What I suspect, and cannot prove, is that the growth after 2010 arose precisely in the functions that had organised the cut: the coordination, the programme management, the monitoring of the shrinkage. If that is so, the salami slice is not merely ineffective. It is one of the causes of what it is fighting.

IV. Three design choices

The three choices below are not in themselves revolutionary. They cut against current practice because they touch the proportion rather than the volume, and because they do not spare the layer that decides on the cut.

The first choice is a proportion norm instead of a volume norm. The Roemer norm, ten per cent external hiring, has achieved more in ten years than all the savings targets together, because it appears annually in the Annual Report per ministry and a deviation has to be explained. In 2024 external hiring stood at 15.4 per cent, in 2025 at 13.1, the first fall in years.6 A comparable norm for the ratio between policy core plus support on one side and delivery on the other, per ministry, reported annually, does what the salami slice cannot: it makes the growth of the governance visible as such and subject to explanation. ’t Hart’s proposal of a permanent cut of one per cent a year then acquires an address: the layers above delivery, with an exemption for any delivery body the Council of State establishes in its budgetary supervision as already unable to cope.

The second choice concerns compensation schemes. Every scheme introduced to repair a mistake or a politically untenable outcome is given, at the moment of introduction, an end date and an owner in delivery, not in policy. That organises disappointment, which ’t Hart says no cabinet dares to any more. A cabinet that publishes, alongside its programme, a list of what it will not do and which schemes it will allow to lapse, gives the ministries the one thing that really restrains growth, namely a mandate to stop. The 128-page government programme that followed the 2024 coalition agreement had an annex listing the legislation to come.12 An annex listing what would cease, I have not seen.

The third choice is a governance paragraph. Where a political compromise has been called unworkable in the delivery test, it cannot simply be decentralised or assigned to a delivery body. The department throwing it over the fence sets out in a paragraph who bridges the gap between compromise and workability: with which people, which money and which powers. This is the aikido choice in this paper. The political urge to promise is neither denied nor forced; it is redirected through a design in which promising without organising pays less than promising with an address. A minister who cannot write the paragraph has not finished the compromise.

V. For interim practice

The €3.6 billion in external hiring, fifteen per cent of the wage bill, is in part my own profession.5 That forces an uncomfortable observation: an interim manager who builds governance — programme offices, steering groups, dashboards and coordinators — adds to the pattern this paper describes, however good that governance is. The measure I apply to that is the transfer value of an assignment: what still stands once I have gone and nobody thinks about it any more. In the context of this paper that measure takes a sharp form. Does the organisation need fewer people governing other people’s work after I leave than before? If the answer is no, I have strengthened the governance of the governance, and that is precisely the product the central government already has too much of.

In the partnership where I am currently quartermaster, the first thing I count is the number of people whose work consists of organising the work of others. That figure ought to fall as an organisation matures. In central government it rose for eight years running.

VI. The open question

Central government can be smaller, and it has failed three times. The question underneath this file is who establishes whether government performs better. Bekker’s sentence, government has not come to perform better through this growth, is true and unprovable, because no performance measure for the whole exists. As long as size is the only thing measured annually, size is the only thing governed, and growth is the only thing the governance can show of itself.

An apparatus that has learned to grow on its own failure ought also to be able to shrink on its own success. Whether it can, nobody knows, because the success is not measured. That is the open question, and it is more precise than €1.4 billion.

Colophon

“Governing the governance” is a Statecraft analysis in the line of Series I, Dissociated organisations, in particular papers Nº 06 (Reproduction inwards) and Nº 07 (Absorbed guilt without integration). It takes its occasion from Marike Stellinga’s column in NRC of 18 April 2026 and the figures in the parliamentary letter on the savings target (October 2025) and the Annual Report on Government Operations 2025 (May 2026). The aikido design rule is drawn from the manuscript in preparation on interim management in the public sector.

Responses and counter-argument via Statecraft.nl.

Jacob Huibers is an interim manager with over twenty years’ experience in the Dutch public sector. He has worked as cluster manager, cluster director and quartermaster at municipalities ranging from fifty thousand to over two hundred thousand inhabitants and at regional partnerships in the social and physical domains. Statecraft is his platform for strategic reflection on public delivery, pillar IV of House of Viridian.

Footnotes

  1. NOS, Ambtenaren staken om nullijn, ‘schoonmakers en cipiers hebben zware baan’, 14 April 2026. ↩

  2. Council of State, Advisory Division, Voorjaarsrapportage begrotingstoezicht 2026, 14 April 2026, in particular the section “Wees preciezer met efficiencykortingen”; covering letter to the House of Representatives of the same date. ↩ ↩2

  3. Minister of the Interior and Kingdom Relations, Ontwikkelingen uitvoering taakstelling Rijksoverheid, letter to the House of Representatives of 3 October 2025, parliamentary paper 31490, no. 391 (2025D43424). The four categories total 160,953; the letter reports an increase of 3,934 FTE compared with 31 December 2024. ↩

  4. Dutch central government, Jaarrapportage Bedrijfsvoering Rijk 2017, annex to parliamentary paper 31490, no. 239, May 2018: 110,649 FTE at the end of 2017, one per cent more than in 2016. ↩

  5. Marike Stellinga, Het moet kunnen met minder ambtenaren, NRC, 18 April 2026. ↩ ↩2 ↩3 ↩4 ↩5

  6. Dutch central government, Jaarrapportage Bedrijfsvoering Rijk 2025, published 20 May 2026; press release Rijksoverheid wordt efficiënter en effectiever – minder externe inhuur, stijging aantal ambtenaren vlakt af, 20 May 2026. ↩ ↩2

  7. Dutch central government, Kabinet aan de slag met 22% structurele bezuiniging bij rijksoverheid, news release of 14 March 2025; the growth of 3,934 FTE between 31 December 2024 and 30 June 2025 from the parliamentary letter of 3 October 2025 (note 3). ↩

  8. NOS, Aantal ambtenaren groeit verder, terwijl kabinet eigenlijk fors wil bezuinigen, 20 May 2026, with the remark by Pieter Duisenberg and the passage from the Ministry of the Interior’s annual report; Binnenlands Bestuur, Efficiëntere rijksoverheid moet 1,4 miljard besparen, January 2026. ↩

  9. Parlement.com, Prof.Mr. R. (Roel) Bekker; the Renewal of Central Government programme (2007-2010) aimed at savings of €750 million and a reduction of some fifteen thousand posts. ↩

  10. WRR, Deskundige overheid, report no. 113, 8 July 2025, project group chaired by Paul ’t Hart. ↩

  11. Statecraft, Series I Nº 06, De reproductie naar binnen, 30 March 2026; Nº 07, Opgenomen schuld zonder integratie, April 2026. ↩

  12. Statecraft, De gedissocieerde wetgever, 5 May 2026, section IV, with underlying figures from the Central Information Point of the House of Representatives and PDC. ↩ ↩2

  13. Reply by Minister Uitermark (Interior) to parliamentary questions from members Bontenbal and Van Dijk (CDA), Aanhangsel Handelingen II 2024/25, no. 154; Binnenlands Bestuur, Helft banen op ministeries is overhead, 4 October 2024. ↩

  14. CPB Netherlands Bureau for Economic Policy Analysis, Analyse coalitieakkoord 2026-2030, February 2026, measures CA_170 (apparatus cut) and CA_171 (Renewal of Central Government): “De veronderstelde ombuiging van 0,4 mld euro op de kerndepartementen en uitvoeringsorganisaties acht het CPB niet plausibel, zolang niet nader is ingevuld welke overheidstaken komen te vervallen. Grote ombuigingen op het overheidsapparaat bleken in het verleden te ambitieus.” The €1.0 billion saving was treated as a generic apparatus target; beyond the maximum of €0.2 billion, the CPB holds, there is “niet verder generiek te korten”. ↩

  15. Nils Brunsson, The Organization of Hypocrisy: Talk, Decisions and Actions in Organizations, Wiley 1989; see also John W. Meyer and Brian Rowan, “Institutionalized Organizations: Formal Structure as Myth and Ceremony”, American Journal of Sociology 83 (1977). ↩

  16. Roel Bekker, De rijksdienst: niet kleiner, maar groter dan ooit!, Montesquieu Institute, 14 April 2025. ↩