Would the Administrative Cost Exceed the Revenue?

The objection is the accountant's. Whatever a wealth tax is for, a tax that costs more to collect than it raises has failed on its own terms, and a charge on assets that have to be valued every year looks like a candidate. It is an empirical question. The evidence available to answer it is thinner than either side of the argument admits.

Two costs, and only one of them lands on the Exchequer

Two costs get added together in this objection that should be kept apart. What the state spends administering the tax, the staff it employs and the disputes it runs, is a charge against the revenue; every pound of it comes out of what the tax raises. What taxpayers spend complying with it, the advisers, the appraisals, the hours and the forms, is real money, and none of it comes out of the Exchequer's take. The first cost decides whether the tax pays for itself. The second decides how much friction it imposes on the people it reaches. A single figure that runs them together describes nobody's budget.

What the Commission found when it asked

The Wealth Tax Commission put the question directly. Burgherr's background paper on the costs of administering a wealth tax, published in 2020, records what came back when Norway, Spain and Switzerland were asked what their wealth taxes cost to run. Norway's Ministry of Finance held no separate reports or estimates. The Swiss Federal Tax Administration held no information on those administrative costs, because the tax there is levied by the cantons and the municipalities. Spain's central government declared that it does not hold cost information.

A costed European example the paper gives is France's ISF, its wealth tax, in 2016. For that year the paper records a central tax-authority cost close to 103 million euros: 35.3 million of management, which covers valuation, collection and litigation, and 67.6 million of control. That is the tax authority's side only. What French taxpayers spent complying is a different number, and the paper does not offer it. The split is worth a second look: by our arithmetic about two thirds of the authority's cost went on control and the smaller part on management.

So the record holds one central administration cost for one regime in one year, with no revenue figure beside it that this article can use. None of the three central authorities that answered supplied a cost figure of its own. The operating evidence settles nothing in either direction.

The taxpayer's side

The estimate that exists is for the other cost. For a well-designed UK wealth tax the Commission's central estimate of taxpayer compliance cost is 0.1 per cent of taxable wealth, with a lower bound near 0.05 per cent, which is about what taxpayers spend complying with the Annual Tax on Enveloped Dwellings, which Burgherr notes is in principle a simpler charge to administer. The denominator is taxable wealth. Against the charge itself, the paper's worked case is a 1 per cent tax, where that cost comes to 10 per cent of the tax paid. At the 2 per cent rate this series argues for, the same estimate implies about 5 per cent, and that second figure is our arithmetic from the published case rather than one the paper prints.

Five per cent of a bill spent on paying it is a real cost, and the objection is entitled to count it. The taxpayer bears it. The Exchequer's take is reduced by none of it, and the figure itself is a modelled estimate, assembled from the costs of levies that are actually collected.

The revenue being argued over

The yield in question is also a model. Advani, Hughson and Tarrant, in Fiscal Studies in 2021, estimate the revenue from an annual tax of 1.12 per cent on individual net wealth above £10 million at £10.0 billion a year on their low-avoidance scenario and £8.8 billion on their high-avoidance one. The rate is the authors' and sits below the 2 per cent this series argues for. The figures are gross of any administration cost, and they describe a population of 22,000 individuals above the threshold, which is 0.04 per cent of UK adults.

Put the two sides of the ledger next to each other and the shape of the objection changes. On the revenue side, a peer-reviewed microsimulation gives a range of £8.8 billion to £10.0 billion a year, at a rate below the one proposed. On the cost side, the record gives no comparable administration figure for any operating regime, and one central-authority cost for a different tax in a different country in a different year. The comparison the objection wants, cost to the state set against revenue, has been made for no wealth tax in operation, and the numbers above cannot supply it for one that does not exist yet.

Where the objection is right

None of that answers the question for the UK. The Commission's own conclusion is that the cost of an annual wealth tax is design-dependent and not credibly quantifiable in advance. It turns on the breadth of the base, the threshold, the valuation rules, the revaluation cycle, the size of the filing population, the dispute process and the treatment of taxpayers who cannot pay. Change any of those and the cost moves. Any confident cost figure, offered for the tax or against it, is a model output rather than a measurement, and the record holds no measurement to check one against.

There is a version of the objection that survives all of this. Valuation is where the administrative cost sits, and every asset that needs an opinion rather than a printout is a dispute waiting to be run. A regime that let the filing population grow and priced every holding fresh each year would spend heavily on exactly that, and nothing in the evidence says how heavily. The French figures show the direction: most of the authority's cost was control. The size of the same bill for a British tax, whose base and threshold are its own, is another matter.

What the design has to contain

The levers that control cost are the ones that control everything else in this series. The first is the threshold. At £10 million of individual net wealth the filing population is 22,000 people, 0.04 per cent of UK adults; every pound the line comes down by adds filers, and every filer adds assets somebody has to look at. The Commission's own headline scenario shows how far that lever travels. It was a different tax from the one this series argues for: a one-off charge of 5 per cent on net assets above £500,000 payable in five annual instalments and falling on over 8 million individuals. Filing population is one of the drivers the Commission names, and between 8 million and 22,000 it is chiefly the threshold that moves it. The second is valuation by rule rather than by argument: holdings with a published price reported at that price, holdings without one carried on a stated method and a stated cycle, so that a dispute is about inputs rather than opinions and most never reach the authority's desk. A high threshold keeps the population small. Standardised valuation keeps the disputes few. Both decisions are made before the first return is filed, and much of the cost of the tax is decided with them.

A cost that cannot be measured in advance is a reason to design carefully, and not a reason to assume the worst. Read what the Commission found when it asked.

Sources and evidence status

Burgherr, D., The costs of administering a wealth tax, Wealth Tax Commission Background Paper no. 126, 2020: the enquiries to Norway, Spain and Switzerland (Spain at fn.36), the French ISF figures for 2016 at p.18, the taxpayer compliance estimate and its ratio to the charge, and the design-dependence conclusion. Advani, A., Hughson, H. and Tarrant, H., Revenue and distributional modelling for a UK wealth tax, Fiscal Studies 42(3-4), 2021, Table 1: the revenue scenarios at 1.12 per cent above £10 million and the 22,000 headcount, with the ONS mid-2024 population estimates as the denominator for the share of UK adults. Rate and threshold of the model this series references: a policy proposal, not enacted law.

The enquiry responses and the French figures are T1 Verified Facts taken from the paper; the French figures are the central tax authority's administration cost only and are not the complete cost of running that tax. The compliance-cost figures are T3 Modelled Estimates for a well-designed tax, built from analogues in existing UK taxes, and the 5 per cent ratio at a 2 per cent rate is our own calculation from the published 1 per cent case rather than a figure the paper prints. The revenue figures are T3 Modelled Estimates at the authors' 1.12 per cent rate, gross of administration cost, and both scenarios are given because printing one alone misstates the estimate. The headcount is a T1 Verified Fact from a peer-reviewed table and the 0.04 per cent share is arithmetic on it against the ONS adult population. The design-dependence conclusion is T5 Economic Interpretation. No figure for what a UK wealth tax would cost to run, and no ratio of cost to revenue for any operating wealth tax, is asserted here, and none should be inferred.

Nothing here is tax advice.