PRISM

Property tax proposals stand or fall on counting. A surcharge on £2 million homes, a Council Tax revaluation, a replacement for Stamp Duty: pricing any of them means knowing how many homes sit at each value and where. No official source publishes that. Taxes fall on every home; price data exists only for the ones that sell. So we built PRISM, an estimate of England's full residential stock by current value and region.

England has around 26 million homes; about 830,000 a year sold across 2024 and 2025, roughly 3 in every 100, and nearly all hard data on house prices comes from that small, self-selected slice. It is not a fair sample. A £350,000 semi in Leeds changes hands far more often than a £6 million house in Kensington, so counting expensive homes by counting expensive sales undercounts them, and the more expensive the home the worse the undercount.

That bias can be measured directly: compare each band's sales against the VOA's count of its homes, giving a turnover rate, sales per 100 homes per year. It is a transaction rate, not the share of homes that sold: 1.2% of sales in a single year and 2.2% across two are repeat sales of the same property.

The shape is a hump. Turnover peaks in bands D and E at about 3.7% a year, then falls: 3.1% in band G, 2.6% in band H. Band A is also low, for a different reason the audit below explains. Both ends of the market are under-represented in sales data, so a correction has to scale each cell's sales up by exactly how rarely its homes sell.

That is a reweighting, and we do it the way good opinion polling does. A pollster does not treat the thousand people who answered the phone as a miniature Britain; they reweight to match known totals for age and region. We do the same for homes, anchored to the one source covering every dwelling whether it sold or not: the Valuation Office Agency's count of homes in each Council Tax band, in every local authority. The table below is what that produces: pure counting, no fitted model in it. The rest of the page tests whether the counting deserves belief.

Two headlines. England has roughly 164,000 homes worth more than £2 million, almost two-thirds of them in London, and about 21,000 worth more than £5 million, 85% in London. At the other end, 1.4 million homes are worth under £100,000, seven in ten of them in the North East, North West and Yorkshire.

The reweighting works hardest at both ends: cheap homes sell rarely too (social housing, empty homes), so the sub-£100k rows are scaled up just as the expensive ones are. The difference is evidence: the cheap rows rest on tens of thousands of sales, the £10m-plus row on 280, so the top rows carry the most uncertainty.

PRISM is not an end in itself; it sits underneath our published estimates. It supplies the stock counts and revenue behind our High Value Council Tax Surcharge analysis: how many homes are worth over £2 million, what they are worth, and what a rate schedule would raise. The same machinery prices any value-based schedule, because it estimates the whole distribution, not just the homes that sold.

A reweighting is only as good as its cells. If homes sold or stayed put for reasons the local authority and band grid does not capture, weighting by that grid would mislead, and the top rows would be wrong in ways nobody could see.

So the second half of PRISM is an audit. We take the turnover differences you have already seen down to cell level and build a regression to explain them, drawing on Council Taxbase records of second and empty homes and Census data on who owns and who rents. The regression never touches the table above (the weights are pure counting), but it is the reason to trust them, and it explains why expensive homes come to market so rarely.

The obvious explanation for low turnover at the top is price itself: Stamp Duty bills are largest there, and moving is costly. That turns out to be a small part of the story.

Across 2,163 local authority and band combinations, the strongest predictors are not prices but who holds the homes and what the homes are. The model measures nine things about each cell: the ownership mix (social housing, private renting, second homes, long-term empties, and the share owning outright with no mortgage); the buildings (flats, detached houses, bungalows); and how much of the stock has been built since 2021.

Social housing rarely sells on the open market; that is why band A turnover is low. Rented homes sell less often than owner-occupied ones. Long-term empties barely sell at all and pile up at both ends of the market; second homes also sell less often, and cluster at the very top, exactly where the puzzle is, though that signal has weakened as the sales data improved.

Band H homes are more than four times as likely to be second homes as homes in bands A to E. Prime central London is extreme: around 16% of Westminster's band H dwellings are recorded as second homes and another 3% are empty; roughly one in five is nobody's main residence.

Among the homes that could sell, the strongest signal is the share of owners with no mortgage: they are older, settled, and their homes rarely trade. Then the buildings. New homes sell unusually often (every one has a first sale); bungalows turn over faster than their price suggests (elderly occupants move into care or die); flats slower.

Empty homes tell the opposite story: band A has the highest empty rate, concentrated in low-demand areas. The two ends of the hump have different causes, and the model captures both.

Here is the finding at the heart of PRISM. Measured naively, turnover falls sharply with price: a £5 million home sells a little over half as often as a £500,000 one. But account for what those homes are and who holds them and almost all of that fall turns out to be composition rather than price; across the whole range from £150,000 to £5 million the adjusted curve is close to flat.

In plain terms: places full of expensive owner-occupied houses see them come to market about as often as places full of £800,000 ones. Expensive homes are rare in the sales data mainly because fewer of them are anyone's main home, not because their owners refuse to move. The flat national curve above £2 million blends two regions that disagree: outside London the adjusted gradient stays clearly negative, inside it is strongly positive, and since almost every very expensive cell is a London borough, the national figure inherits London's answer. Fit by region, below, sets out how far apart they are.

This matters for the top rows of the table above. PRISM never assumes a £10 million house sells at some ever-smaller rate; each cell's weight just counts what happened there. The finding here justifies that: what stops expensive homes trading is mostly measurable (second homes, ownership, building type, all from published data), so weighting by cells that hold those things roughly constant is a sound way to count the homes that did not sell.

One caution: every part of England faces the same Stamp Duty schedule, so comparing places cannot tell us what the tax does to moving; the note under the regression draws that line.

Price patterns alone explain about 37% of the variation in turnover between cells. Adding the ownership mix takes that to 57%; the full model reaches 69%. That is not a memorised fit: refitted with whole local authorities held out and predicted cold, the model still explains 67%.

The sharpest test is the hump itself: the model is never told which band a cell is in, yet asked to predict national turnover band by band it reproduces the whole curve, both low ends included, within 0.3 points at every band.

Band averages can flatter a model, because errors in opposite directions cancel. So here is every cell. The big cells sit close to the line; the scatter is widest among small cells, where a handful of sales swings the rate. The correlation is 0.76.

At the top end, borough by borough, the fit is rougher and misses both ways. The model expects prime central London to sell more rarely than it does: Camden, Kensington & Chelsea and Westminster all outsold their predictions across 2024–25, each by around half a percentage point. Two of the caveats below are the likely causes: our neighbourhood-based tenure estimate probably overstates renting in prime London, and rumoured taxes on high-value homes may have pulled top-end sales forward. Birmingham and Richmond err the other way; Birmingham's band H homes sold at 1.6% against a predicted 3.0%.

The unit of observation is a local authority and band cell, 2,163 in all. Each cell's 2024–25 sales count, matched sales plus the fractional band assignments of unmatched ones, is modelled from its stock, median price and nine composition shares. There are no band dummies: any pattern across bands must come from the covariates.

The regression does not set the weights behind the table above; those are each cell's raw stock-to-sales ratio. It audits them: observable composition explains the turnover differences between cells, leaving price selection little to explain.

Three modelling notes. The quadratic in log price does no hidden work: a free dummy for each band lifts fit only from 0.69 to 0.75. Every covariate must earn its keep out of sample; bedroom mix, stock growth, pre-1919 stock and the share of residents aged 65 and over were tested and dropped, the last because it was too entangled with outright ownership to read either coefficient alone. And one honesty note on the 67%: the cross-validation folds were consulted during covariate selection, so it is marginally optimistic as a pure out-of-sample number. Fitting on one year and validating on the other is the cleaner test, now that two years are in, and it is the next check.

What the regression is not: an estimate of Stamp Duty's effect on moving. Every cell faces the same tax schedule at a given price, so a cross-section cannot separate price from tax; measuring it means watching behaviour change when the tax changes, as the transaction-tax literature does (Besley et al.; Best and Kleven; Hilber and Lyytikäinen). The flat adjusted gradient neither contradicts nor supports that evidence. The coefficients also describe places, not people: tenure and age come from neighbourhood data, so read them as area-level associations.

The drop-one column measures fit, not what each covariate does to the adjusted price curve, which is the audit's key exhibit at the top end. Across alternative specifications deviance explained barely moves; the curve does:

Read the £5m column. The qualitative finding holds: raw turnover at £5m is a little over half the £500k rate, and every composition-controlled variant puts it far higher. But the precise figure leans on second homes, empties and outright ownership: the least precisely estimated covariates in the model, and the ones carrying the most ecological content. Across specifications the £5m relative runs from about 92 to 97. That range understates the real uncertainty, and we would rather say so than let the table imply otherwise. Fit on each year separately, the same model gives 84 on 2025 and 113 on 2024, with no overlap between the two years' specification ranges. Averaging stops one year's distortions setting the answer, but it produces a midpoint, not a resolution. The honest statement is that the composition-adjusted top-end gradient is close to flat, and its sign is not identified by two years of English transactions.

The national model uses one set of coefficients everywhere. That is an assumption worth testing. Every region has a hump, but not the same hump: the low-band shortfall deepens as you move south, and the top-end decline shallows.

Two questions follow. How well does the single national model work inside each region? And what does each region say fitted on its own?

The pattern is not subtle. The national model explains most of the cell-to-cell variation across the north and the midlands, less across the greater south, and least in London, where it accounts for well under half. Refitted alone, eight of the nine regions put adjusted turnover at £5m well below the £500k rate; London puts it at nearly twice.

London is not a wrinkle in the national number; above £2 million it very nearly is the national number, because almost every cell with a median in the millions is a London borough. Read the flat national curve as one region's answer carrying the rest; the honest reading of the other eight is that turnover really does fall at the top.

The London-only fit also produces coefficients no one should believe. The three tenure shares explode: social renting −4.9, private renting −4.6 and outright ownership −5.5, against roughly −2.2, −0.2 and −1.1 in the rest of England. These are not noisy estimates but precisely estimated wrong numbers. Within London the tenure shares index a single axis, inner borough against outer, and with no between-region contrast left they absorb every difference in demand and liquidity between boroughs, not just composition. Constraining social, second-home and empty shares to pure stock removal, which costs almost nothing nationally, collapses the London fit from 0.75 to 0.54.

Grouped the way regional house-price indices have moved since the 1980s (London; the greater south, meaning the South East, South West and East of England; and the rest), the same split shows plainly: the rest of the country puts the £5m relative at 57, the greater south at 22, London at 182. Effects of this kind are unlikely to be uniform across regions, bands and price levels; testing that properly, with interactions rather than separate fits, is the next piece of work.

Any cross-sectional regression invites the question of what its covariates are really measuring. Three mechanisms deserve naming:

We build PRISM from public data and our own transaction records:

• Sales and valuations: 2.0 million Land Registry transactions from January 2024 to June 2026, matched to Council Tax bands; the model counts calendar 2024 and 2025 and reports the annual average. Every price is indexed forward to Q2 2026, so a 2025 sale counts at what the home is worth now, not what it fetched then.
• Stock counts: VOA Council Tax stock of properties (CTSOP), band counts down to neighbourhood level, 31 March 2025.
• Dwelling type and build period: the same VOA release, broken down by property type and build period for every band and local authority.
• Second and empty homes: Council Taxbase statistics, band by band for every English authority, October 2025.
• Tenure and age of residents: Census 2021 owner-occupation (outright versus mortgaged), renting shares and age structure by neighbourhood, combined with the band counts.
• Geography: ONS lookups linking neighbourhoods to local authorities and constituencies.

Poisson regression across 2,163 local authority × band cells covering 98% of English stock; fit shares are deviance explained; the out-of-sample figure holds out whole local authorities in five-fold cross-validation.

We are grateful to John Muellbauer at Nuffield College, Oxford for detailed comments on the model and wider discussions through the past year, and particularly for his comments on the treatment of unmatched sales, the regional fits, and what the tenure coefficients are really measuring. Remaining errors are ours.