The new-build premium is mostly a mirage
The new-build premium is +25.0% on the headline numbers and +10.4% comparing like with like — and in 26% of the country new builds sell for less than second-hand.
New builds look 25% dearer than second-hand homes. Compare like with like and most of that gap disappears.
Arm's-length residential sales (category A), 1 April 2025 to 31 March 2026. 635,379 transactions in the window.
The evidence
New builds are 48.6% detached against 24.5% of existing stock. That mix difference, not newness, is most of the headline premium.
Read the per-type rows carefully, because the Detached, Semi-detached and Terraced rows move the other way: within a single property type, the like-for-like figure is higher than the national one, since new builds of a given type cluster in different districts from the existing stock of that type. The Flat row moves with the headline instead. The composition effect that inflates the headline is the property-type mix, and it is only visible in the "all sales" comparison — which is exactly the comparison the widely-quoted premium is based on.
| Property type | New builds | Median new | Median existing | Headline | Like-for-like |
|---|---|---|---|---|---|
| Detached | 9,444 | £413,250 | £420,000 | -1.6% | +4.8% |
| Semi-detached | 5,241 | £284,950 | £270,000 | +5.5% | +14.0% |
| Terraced | 1,800 | £270,000 | £235,000 | +14.9% | +31.3% |
| Flat | 2,934 | £407,200 | £220,000 | +85.1% | +38.3% |
| All sales | 19,419 | £356,250 | £285,000 | +25.0% | +10.4% |
Like-for-like pairs the same property type in the same postcode district, requiring at least 10 sales on each side of every pair (614 qualifying pairs).
Does it depend on where we drew the line?
No. The finding holds at every sensible minimum-sales gate.
| Minimum sales per side | Qualifying pairs | Like-for-like premium | Pairs where new builds sell for less |
|---|---|---|---|
| 5 | 1,239 | +12.4% | 24% |
| 10 | 614 | +10.4% | 26% |
| 20 | 200 | +8.6% | 27% |
| 30 | 76 | +8.6% | 26% |
How we calculated this
The headline premium divides the median new-build price by the median price of existing stock. That comparison is not like-for-like: new builds skew heavily detached, so the headline number partly measures what kind of home was built rather than what a buyer paid for newness.
To remove it we pair within a postcode district AND a property type — comparing new detached houses in a district only against existing detached houses in the same district — and require at least 10 sales on each side of every pair before it counts. We then take the median across all qualifying pairs. We report the result at four different minimum-sales gates so you can see the finding does not depend on where we drew the line.
What this can't tell you
Price Paid Data records no floor area, so within a district and property type we cannot rule out new builds being systematically larger or smaller than the existing stock they are compared against. The composition half of this finding — that the headline premium is inflated by new builds skewing detached — is measured directly and is not affected. The remaining like-for-like premium should be read as an upper bound on a genuine newness premium, not a precise estimate of one.
Source and licence
HM Land Registry Price Paid Data — arm's-length residential sales (category A), 1 April 2025 to 31 March 2026 (635,379 transactions), analysed by Offrly.
Contains HM Land Registry data © Crown copyright and database right. This data is licensed under the Open Government Licence v3.0.
Free to reproduce with attribution to Offrly and HM Land Registry. The data behind this page · Open Government Licence v3.0
Spotted an error? Tell us — we will correct it and say what changed.