TL;DR
Does a new-build cost more than an older home down the road? Almost everywhere, yes — but the size of the gap swings violently, from a +305% premium to a -69% discount depending on the outcode. And once you dig into why, the headline numbers turn out to be telling you less about "new vs old" than about what kind of home developers are putting up in each place.
We took every residential sale in England and Wales over the last five years, split it into new-build and established, and computed the median-price gap for every outcode with enough of both. The 1,623 qualifying outcodes break down as 1,242 with a new-build premium and 377 with a new-build discount.
The five biggest premiums:
The five biggest discounts:
Hold those numbers loosely. By the end of this piece you'll see that almost the entire spread is mix-shift — the new homes and the old homes being compared are not the same kind of property.
How we measured it
A deliberate choice: this is a raw, all-property-types comparison within each outcode. That makes the ranking honest about *what a buyer actually sees in the local market* — but it also means the figures are dominated by the property-type mix, which we unpick below.
The biggest premiums: a brand-new estate beside an old terrace
The premium list is overwhelmingly northern, and the mechanism is consistent. These are areas where the established housing stock is cheap, dense and old — terraces, ex-council semis, the legacy of coal and steel — and where developers are now building estates of three- and four-bedroom detached and semi-detached houses on former industrial or greenfield land.
Darlington's DL3 is the extreme case: a new-build median of £546k against an established median of £135k, a 305% gap. That is not the same home costing four times as much. It is a small number of large executive new-builds on the western edge of town, sitting in the same outcode as thousands of modest Victorian terraces. The DL4 outcode around Shildon, SR8 in Peterlee and DL17 around Ferryhill tell the same County Durham story.
The North West and Yorkshire are well represented too: BB10 in Burnley (+168%), BB2 in Blackburn (+127%, on a healthy 665 new-build sales), S66 and S61 around Rotherham, and HD2 in Huddersfield (+117%). LA14 in Barrow-in-Furness and the West Cumbrian outcodes CA25 and CA26 round out the picture — places where a new estate genuinely is the most expensive housing for miles.
Two outliers break the pattern. SW1H (Westminster, +204%) and WC2B (Covent Garden, +116%) are central-London postcodes where new-build means high-spec apartments in prime developments, and the "established" comparator is a thin, mixed bag of older flats. Same direction, completely different cause.
The biggest discounts: new flats beside old houses
Now flip it. The discount list looks alarming — new-builds selling for two-thirds less than established homes — until you notice what is being built. In almost every discount outcode, the new supply is flats and apartments, while the established stock is dominated by houses.
TW15 in Ashford, Surrey (-67%): new-build median £140k, established £431k. Nobody is selling family houses for £140k in Ashford. The new-build figure is apartments; the established figure is houses. KT10 in Esher (-56%, £410k new vs £935k established), HA5 in Pinner (-56%), B91 in Solihull (-48%) and the two Hertfordshire garden-city outcodes AL8 and AL9 are all the same story: a wave of new flats in commuter towns whose existing stock is large, expensive detached and semi-detached houses.
London's discount outcodes — SW12 (Balham), E5 (Clapton), E11 (Leytonstone), IG1 (Ilford) — are the residential conversion and regeneration story: new one- and two-bed flats discounting heavily against the Victorian and Edwardian houses that define the established market.
So the discount is real in the sense that *a buyer entering these markets to buy a new home pays less than the typical established sale* — but only because they are buying a smaller, different kind of property. It is not evidence that newness destroys value.
The mix-shift problem, measured
To prove the point, we ran the same comparison like with like — new versus established within each property type, across the whole of England and Wales:
This is the real new-build premium, stripped of composition. A new flat commands a big premium over an old flat (warranties, energy efficiency, no chain, lift, concierge). A new terraced or semi commands a modest one. And a new detached house actually sells at a slight discount to an established detached — because established detached homes tend to be larger, on bigger plots, in more mature settings, and the new ones are squeezed onto tighter estate layouts.
Put those four facts together and the whole outcode ranking falls into place:
The single number to remember: the new-build premium is overwhelmingly a flats-and-warranties premium, partially offset by a detached-house discount. Everything else is geography and what the cranes happen to be building.
What this means if you're buying
Outcode dossiers for the most striking entries on both lists:
Caveats and limitations
Mix-shift dominates. This is the headline caveat, and we have spent half the article on it. The outcode rankings are not a clean measure of "what newness is worth" — they are a measure of what is being built where, relative to what already exists. The like-for-like table is the cleaner number.
Size and plot are uncontrolled. Even within a property type, we do not adjust for floor area, number of bedrooms, or plot size. New estate houses tend to be smaller per bedroom than older equivalents; this nudges the like-for-like premium down for houses.
The new-build sales premium has a known component that fades. New-builds carry a developer premium at first sale that often does not fully survive into the second-hand resale a few years later. A five-year window of first-sale-heavy new-build data can overstate the durable premium.
Outcode geography is coarse. An outcode can span very different sub-markets. We use it because it is the unit our area pages are built around and because it keeps samples large.
England and Wales only. Price Paid data covers England and Wales; Scotland and Northern Ireland use separate registers not included here.