Taxing the Symptom

There is a particular kind of policy that feels like justice while doing the opposite. The 20% VAT on private school fees, in force since January 2025, is the textbook case. It was sold as ending a tax break for the privileged. In practice it taxes the strivers, spares the genuinely advantaged, and leaves untouched the one mechanism in English education that actually launders wealth into opportunity at public expense: academic selection. The government has gone after the symptom and left the disease in place.

Who actually pays

Start with the incidence, because a flat surcharge is regressive within the group it falls on. The family that could already absorb £20,000 a year can absorb £24,000. They stay. The family stretching a single professional salary across one set of fees, often having chosen private education precisely because their child missed a selective state place, cannot. They leave. So the VAT does not bite the people whose wealth the policy was meant to reach; it bites the marginal household at the bottom of the fee-paying distribution.

The numbers bear this out. Before the policy, the government leaned on Institute for Fiscal Studies modelling suggesting an effective 15% increase would reduce private attendance by 3 to 7%, and ministers projected only a few thousand pupils moving into the state sector. The Independent Schools Council's 2026 census tells a different story: private numbers have fallen by roughly 30,000 since the policy began, with some year groups down more than 6%, and the steepest falls concentrated at the entry points — the start of prep, the start of secondary, sixth form. Those are exactly the decision moments where a marginal family does its sums and walks away. Some of that fall is demographic; a birth-rate decline is shrinking cohorts regardless. But the concentration at entry points, well above the government's own forecast, is the fingerprint of price sensitivity among the people least able to pay.

A High Court challenge to the policy was dismissed in 2025, the human-rights argument rejected. Legally the Treasury is in the clear. That is not the same as the policy being sensible.

The injustice it walks straight past

Here is what the VAT does nothing about. In the roughly 36 English local authorities that still select, a child's secondary education is decided at age eleven by a single test. The state then funds, for free and to a high standard, a premium tier — the grammar school — and allocates places by that test. The official story is that this is a ladder of merit, a route out for the clever child from a poor home.

The evidence is that it is largely the opposite. Grammar schools admit around 2.5% of pupils on free school meals against a national average of 13.2%. Roughly one in ten grammar entrants arrives from a fee-paying prep school. Research by Jerrim and Sims shows that advantaged parents invest heavily in tutoring, that this is most pronounced in selective areas and specifically in the subjects the eleven-plus tests, and that wealthier children are more likely to pass even when poorer children have reached the same attainment. At the very top of the income distribution, the chance of a grammar place approaches 80%. The Education Policy Institute finds the disadvantage gap is wider in selective areas than in comprehensive ones — 34% against 28%.

So the eleven-plus does not measure ability cleanly. It measures a blend of ability and purchased preparation, and it converts the second into a claim on the first. The state then hands the winners a free, high-quality education funded by every taxpayer — including the taxpayer whose own, un-tutored child sat the same test, missed the cut, and was pushed toward the private sector as the only remaining route to something comparable. That family funds the premium tier through general taxation, captures none of it, and is now taxed a further 20% on the substitute. The net direction of the money is upward, from the striving middle toward the already-comfortable. This is the regressive transfer the VAT was supposedly designed to reverse, and the VAT leaves it completely intact.

A test built on an idea we no longer believe

It is worth knowing where the eleven-plus came from, because its origins are also its refutation. No single person invented selection, but three sources built it. The intellectual foundation was laid by Cyril Burt, the educational psychologist whose theory held that intelligence is innate, fixed, largely inherited and measurable early — which made sorting children at eleven look like science rather than prejudice. The Norwood Report of 1943 turned that psychology into the claim that children come in three types, suited to three kinds of school. R.A. Butler's 1944 Education Act made it law, with the eleven-plus as the gate: grammar, technical, secondary modern, a child's path set by a single exam before they had finished growing.

The trouble is that the science underneath has not survived. Burt's empirical evidence — the twin studies that supposedly proved intelligence was inherited — was, after his death, widely alleged to have been fabricated. More fundamentally, the whole premise of fixed, early-measurable intelligence has been abandoned by the system that still acts on it. Modern education is built on the opposite idea: that ability is developable, that teaching and effort move the dial, that a child's capacity at ten is not their capacity at sixteen. "Growth mindset" is not a fringe notion; it is mainstream teacher training. Schools are now taught explicitly not to label children, precisely because early labels are self-fulfilling — tell a ten-year-old they are not clever and you have gone some way to making it true.

Hold those two facts together and the contradiction is stark. We have rebuilt almost the whole of education around developable ability, and rejected fixed-trait determinism nearly everywhere — and then kept one high-stakes test, designed on a discredited theory of fixed intelligence, that classifies children as able or not at the precise age when that judgement is least reliable and most damaging. The eleven-plus is not merely unfair. It is obsolete by the standards of the very system that administers it: a culture that no longer believes ability is fixed still sorts its children with a test that assumes it is.

The deeper objection

There is a principle underneath the arithmetic, and it is not a fringe one. Michael Sandel's argument in The Tyranny of Merit is that sorting people by merit corrodes the common good even when the sorting is fair — it breeds hubris in the winners, humiliation in the left-behind, and dissolves the sense that a society is a shared enterprise. An eleven-plus result is that sorting in its rawest form: a public verdict, delivered to a child, of pass or fail. Make the test perfectly tutor-proof tomorrow and the deeper damage remains, because the harm is in the act of sorting young children into the clever and the rest, not only in who games it.

Michael Walzer's Spheres of Justice supplies the other half. Walzer argues that a just society blocks the conversion of advantage in one sphere into dominance in another — money may buy goods in the market, but it should not buy standing in education or citizenship. Tutoring for the eleven-plus is precisely that blocked exchange being broken. And the common school, for Walzer, is where shared membership is actually formed; a system that extracts the able into separate institutions fractures the civic space before adulthood begins.

That extraction is not a neutral act of administrative tidiness. A grammar school's results, ethos and self-regard are financed by removing the pace-setters from the school down the road — the pupils who make hard work look normal, who lift a classroom's expectations by their presence. Education research on peer effects is consistent and asymmetric: lower attainers gain more from being taught alongside high attainers than the high attainers lose from the mixing. So concentrating the top is not merely redistributive, it is destructive of value. The grammar captures a private benefit; the neighbouring comprehensive absorbs an unpriced cost — a thinner peer culture, fewer examples, a lower ceiling on aspiration. A community that constitutes itself by impoverishing the community next door is not a communitarian achievement. It is an externality in a blazer.

What the government should do instead

If the aim is fairness, the lever is selection, not fees.

End academic selection at eleven. Phase out the eleven-plus and convert the remaining grammar schools to all-ability admission, as the great majority of England did decades ago and as most high-performing school systems never abandoned. This removes the regressive subsidy at source: it stops the state funding a premium tier rationed by a contaminated test, and it returns the pace-setters to the schools that need them. It costs nothing in new spending; it is a change in admissions, not a cheque. Crucially, it is the one reform that the residual-pool evidence actually supports — the disadvantage gap is widest exactly where selection persists.

Note what this does not require. All-ability does not mean identical. Schools can and should differ in ethos, specialism and character; the false choice between selection and a grey state monoculture is just that — false. Differentiation by pace within a school, through setting by subject, is entirely compatible with a common institution. The thing to abolish is the sorting of children into separate schools by a single early test, not the recognition that children learn at different speeds.

As for the private sector: if the state has a legitimate interest there, the honest instrument is targeted, not a flat surcharge that falls hardest on the least wealthy fee-payers. Tie continued tax advantages to genuine public benefit — meaningful means-tested bursary numbers, shared facilities, sponsored places for disadvantaged children — and withdraw them where that benefit is absent. That is defensible. A blanket 20% that drives out the marginal family while the affluent shrug is not.

The objections, and why they fail

"This is really an argument against grammar schools, not for sparing private VAT." It is both, and that is the point. The same government chose to keep selection and impose the VAT. The package is the thing under indictment: it preserves the regressive state machine and taxes the people it has already excluded.

"Removing a tax break isn't a penalty." True in the abstract. But the burden of a flat surcharge is not abstract, and it does not land on the people the rhetoric names. Calling a regressive incidence "fairness" does not make it so.

"Grammars are a ladder for the bright poor child." They were, for a generation, in specific places. The current data says they are not now — 2.5% free-school-meals intake is not a ladder, it is a turnstile that mostly admits those who arrive already advantaged. The aspiration is honourable; the instrument has failed it.

The contradiction at the centre

Strip away the presentation and the policy reduces to a single contradiction. The government runs a state-funded sorting machine that converts parental wealth into a free premium education, concentrates disadvantage in the schools left behind, and tells eleven-year-olds whether they are clever or not. Then, instead of switching the machine off, it taxes the escape hatch — and bills the people fleeing the consequences of the machine it kept running.

If you want fairness in education, you do not tax the symptom. You end the selection that causes it.

Sources: Independent Schools Council 2026 census; Department for Education pupil-number data; Institute for Fiscal Studies (2023); Education Policy Institute, "Grammar schools and social mobility"; Jerrim and Sims on tutoring and selective admissions; HM Treasury policy statements on private-school VAT. Philosophical argument draws on Michael Sandel, The Tyranny of Merit, and Michael Walzer, Spheres of Justice.