Prune the state like a gardener, and let others plant

Our governments save by cutting every institution by the same rule, and they cut the levers others pay into as well. A gardener would prune instead.

A walled garden with one old fruit tree, its dead branches neatly cut away and stacked at its foot. Pruning shears with red handles rest on a bench. Young saplings have just been planted along the edge, and behind the wall a hedge is cut flat.

Each federal institution will be told how much it has to save. Until it gets there, it may replace only two of every five people who leave. That is how the De Wever government plans to save on its own civil service, according to BX1 in January. Justice, Defence, the police and the Interior are exempt. Every other institution gets the same rule.

At the end of September the Flemish government followed. It will save 750 million euro on its own administration, about 180 million of it on its own staff. It also cuts more than 500 million euro in subsidies over the coming years. In the words of the public broadcaster VRT, “Elke minister moet een inspanning leveren”, every minister has to make an effort.

I have seen this film before. In 2014 I wrote here about the cuts that followed the financial crisis. In my words then:

“cuts were being imposed across the board, with little consideration for efficiency and optimizing service performance.”

Twelve years later, the method has barely changed.

The thread

This is the third piece in a row on the same mood. First came the clowns, the politicians who sell anger because anger pays. Then came the floor we fear most, and the fading promise that society will catch you. In the floor piece I wrote one sentence that I owe you more on. “In my view, the government is not today the most efficient and effective route for every euro.”

It is also a reaction to Pieter Timmermans, the CEO of the Federation of Belgian Enterprises, VBO-FEB. Last week he and the heads of the other employers’ organisations signed a joint opinion piece. They ask for a growth plan, built on four choices. In their words:

“concurrentiekracht versterken, administratieve lasten verminderen, een slankere overheid en ondernemerschap ondersteunen.”

That means stronger competitiveness, fewer administrative burdens, a leaner government and support for entrepreneurs. A leaner government is the goal they name. They do not say how to get there.

So let me explain what I mean, and what I would do about it.

A side note first. I have worked for years on optimising government. Twice, between the early and the late 2000s, I managed the measurement bureau of the Burden Reduction Office of the Belgian federal government. That is where my view on this subject comes from.

Six governments

The Belgian constitution opens with one sentence. “België is een federale staat, samengesteld uit de gemeenschappen en de gewesten.” Belgium is a federal state, made up of communities and regions. There are three of each, plus 10 provinces and 565 municipalities. Flanders merged its community and its region, so we have six governments, each with its own parliament.

And whatever the cause, we spend a lot. In 2024 Belgian government spending came to 54.1 per cent of GDP. GDP, gross domestic product, is the value of everything the economy produces in a year. The euro area average was 49.3 per cent. Germany spent 49.0, the Netherlands 44.6. Of our three large neighbours, only France spent more, at 57.0.

Spending a lot is a choice a country can make. The question is what we get for it. The International Monetary Fund, the IMF, visited in December and put it politely:

“Public investment management would improve by coordinating planning among federal and federated entities, standardizing project selection, strengthening federal-regional cooperation […]”

It also asked for spending reviews: “Systematic spending reviews at the regional level” (IMF, December 2025).

Let me explain what that means in practice. Six governments each plan their own roads and schools. They each decide with their own rules. Nobody has to look at the whole. And when they do not agree, things simply stop. Brussels needed 613 days to form a government.

Brussels and the EU

One administration in Brussels creates jobs for Brussels without Belgium paying the wages, and it is not ours. According to the Brussels Commissioner for Europe and International Organisations, the international presence accounts for around 52,000 direct jobs. Of those, 40,514 are European Union staff. In its highest estimate, the international presence generates “up to 23.2% of regional employment”, once you count the jobs it creates around it.

Think of who works around those buildings. Somebody has to clean and guard them, and feed the people inside. Many of those jobs do not need a university degree. And I assume EU staff spend a good part of their salaries in Brussels shops and restaurants.

Now, before you get me wrong, there is a catch. EU officials pay a tax to the Union itself. The EU treaties then add this (Protocol No 7, Article 12):

“They shall be exempt from national taxes on salaries, wages and emoluments paid by the Union.”

So Brussels gets the jobs and the spending, while Belgium does not get the income tax on those salaries. Nobody publishes how much tax that is. What is published is what the federal government pays back. Since the sixth state reform, Brussels receives a compensation for these tax-free officials, 210.5 million euro in the 2024 budget. Another 55 million a year pays for security at European summits and for its role as a capital. That amount has not risen with prices since 2012.

Is that a good deal for Belgium? The Brussels government’s own study counts more than 162,000 jobs linked to the international presence. Take away the 52,000 direct jobs, and roughly 110,000 jobs remain around the institutions. Those do pay Belgian tax. Spread the compensation over them, and it comes to less than 2,000 euro per job a year. With a tax wedge of more than 50 per cent on an average wage, an ordinary job pays far more than that. So in my view the balance for the country is positive. For Brussels it is closer. Three in four of the people working at the international institutions live in the region and use its services. That is exactly what the compensation is for, and part of it shrinks every year it is not indexed.

Still, the lesson is useful. A public body can create work without employing the workers itself. The Commissioner’s page also counts between 10,000 and 14,000 lobbyist jobs in Brussels, and more than 700 accredited journalists. The EU employs none of them, yet they are there because the EU is. Keep that in mind, because I will come back to it.

The two in five rule

Back to the cut. A rule of two in five treats every institution the same. The department that already works lean pays the same price as the one with slack. In my view that is the opposite of what we want, because it punishes exactly the managers who did the work.

And here lies the deeper problem. In 2019 I wrote that “government administrations have no economic incentive to optimise their cost of control” (Directive versus coercive policies). That still holds, and for more than control. A public manager who delivers the same service with fewer people is rarely rewarded for it. As I see it, the saving usually just disappears into next year’s budget. So why would anyone try?

A uniform rule does not ask which tasks matter. It asks how much each department can lose without shouting too loudly. I think it also slows the economy, because good services and good investments get cut along with bad ones.

The state’s own work

So what would a gardener do? A gardener prunes, branch by branch, and asks of each one whether it still bears fruit. The cut goes where it helps the tree.

For a government, I see four things it can do directly.

The first is a spending review that looks at tasks before it looks at budgets. Do we still need this task, and if so, which government is best placed to do it? The IMF asks for exactly this. Has such a review ever ended a task? That is the test.

The second is one task, one government. Where six governments each do a bit of the same thing, pick one and let the others stop. For me, that is what the next state reform should be about. As I read them, the earlier reforms mostly moved competences between levels. This one should reduce the number of places where the same work is done twice.

The third is to let administrations keep part of what they save. Say a department finds a cheaper way to deliver a service. Let it keep a share for a year, to invest in its people or its systems. Then saving becomes something managers want to do, instead of something done to them.

The fourth is to use the people who already know where the waste is. In 2014 I argued that the agents of change “reside inside of the administrations”, and that they are called internal auditors. They are still there. Ask them which tasks can go, and listen to the answer.

Tasks for others

The second half of the job is harder, because it is about what a government chooses not to do. A government does not have to deliver every service itself to reach its goals. It can set the goal and let others carry part of the work, with an advantage in return.

Belgium has done this for a long time, often without noticing. Unemployment benefits are paid out by trade unions. The national employment office, the RVA, lists one public payment office, the HVW. The others are “uitbetalingsinstellingen verbonden met een vakbond”, payment offices linked to a union. Health insurance runs through the ziekenfondsen, the mutual health insurance funds. A gift to an approved charity earns a tax reduction of 30 per cent since 1 January 2025.

In 2019 I called these tools “directive policy”, the carrot next to the stick. The government sets the direction, rewards the behaviour it wants, and someone else pays a large part of the bill. I also called it the holy grail of policy making, because it answers one question. “How do I enact policy while making others pay for it?”

The eco-cheque is the clearest case. The government picks the goal, the employer pays, the worker chooses. I wrote about it on 30 October, together with the Flemish KMO-portefeuille. There, small companies put in more than two euro for every euro Flanders added. The federal government wants to phase out the eco-cheque. Flanders closes the KMO-portefeuille to companies from 2027.

The same Flemish budget raises the price of a service voucher from 10 to 12 euro from 1 January 2027. From the 176th voucher in a year, it costs 13 euro. A household can buy at most 300 a year. Those vouchers pay for home help, and for the jobs of the people who do it.

And this lever is cheap to run. In 2022 Flanders paid out 81.3 million service vouchers. The subsidy came to 1,327 million euro, and users paid about 732 million on top, at 9 euro a voucher. The company that issues the vouchers cost 12.68 million euro to run it. That is about 0.6 per cent of the money that went through the system. The European Commission puts the cost of administering its regional funds at 3 to 4 per cent of what they spend. The two figures do not measure exactly the same thing, because the voucher figure leaves out the government’s own staff. Even so, the difference is a factor of five or more.

The ziekenfondsen give a second Belgian benchmark. For 2024 the government granted the five national federations of health insurance funds 1,285 million euro for their administration. The health insurance budget for that year was 42,743 million euro. That comes to about 3 per cent, close to the EU figure. And the funds do more for that money than pass it on. They also pay out sickness benefits and check what they reimburse.

In the same round where governments cut their own administrations by rule, they also cut the levers through which others share the cost. A broke state that saves on the money other people put behind its goals is saving in the wrong place.

The strongest objection

Now, let’s take a look at the other side, because it has a point.

Handing a task to someone else does not make it cheaper by magic. The organisation that gets the task will defend it, and its budget, as hard as any ministry. A tax advantage helps those who can afford to use it first. In 2019 I called that the Matthew effect: those who already have, get more. And whoever does the work, the government stays accountable for the result. If benefits were paid late, I suspect most people would blame the state before the union.

Yes, all of that is true. But it is an argument for doing it well. None of it argues for doing everything yourself. A ministry also defends its own budget, and it also serves some people better than others. As I see it, the difference is that a ministry rarely has to show what it delivers for the money.

Four questions

So before a government cuts a task, or hands one to others, I would ask the four questions I proposed in 2019.

  • Who carries the cost, and is it the one who benefits most?
  • Does the measure lead to the behaviour we want?
  • Do enough people actually use it?
  • What else does it bring, for jobs, for the region, for the budget?

Measure the answers, publish them, and stop what fails. That goes for a ministry and for a voucher alike.

A gardener does not cut every branch by the same amount. The dead wood goes, and others may plant where they grow things better. Our administrations need this. So do the people who pay for them.

Think about it.

Sources and Methods

Translations from Dutch and French are mine. All figures were checked against the sources below on 9 November 2026.

Official sources

Reporting, health insurance funds

  • De Specialist, Administratiekosten ziekenfondsen toegekend voor 2024 (26 April 2024), on the royal decree of 9 April 2024: “1.285.441.000 euro voor de vijf landsbonden”, plus 22,804,000 euro for the railway fund of HR Rail, which I left out. With it included the share is 3.06 per cent instead of 3.01. Against reimbursed care alone, 37,825 million euro in the same RIZIV budget, it is 3.4 per cent. Both figures are amounts granted and budgeted; actual spending may differ.

The EU balance

  • G. Carlier, under the direction of H. Bogaert, Finances publiques de la Région de Bruxelles-Capitale: trajectoire et dérapage budgétaire (2010-2024), CERPE, UNamur, Cahier 122 (2026). The “Compensation fonctionnaires internationaux” was 159 million euro in 2015 and 210.512 million in the 2024 initial budget; the “dotation sécurité” has been 55 million euro since 2012 and “n’est pas indexée”.
  • Brussels government, Brussels has increased its international input (23 June 2020): “up to 23% of regional employment, or more than 162,000 jobs”, with impact estimates for 2016 and 2017. The 52,000 direct jobs come from the Commissioner’s current page and refer to a later year, so the 110,000 is an approximation. No study I found subtracts the income tax exemption from the gains, and the 2,000 euro per job is my own division of the compensation, which is not the same as the full tax foregone.
  • VRT NWS, 48,000 people employed by international institutions in Brussels (2 June 2018), on figures of the Brussels statistics office: 72 per cent of those working at the international institutions live in Brussels.
  • The tax wedge of 52.5 per cent for an average single worker in 2025 comes from OECD, Taxing Wages 2026, as cited in my piece of 30 October.

Reporting

My earlier pieces