Why Flat Floor Space Doesn't Explain City-Centre Discounts (2026)

In every one of the 29 city-centre postcodes this site tracks across England and Wales, the properties with an Energy Performance Certificate on file look nothing like the properties that are actually there. Flats make up the large majority of homes in these centres — often 90% or more of the housing stock — but in the EPC register they are badly under-represented against houses, by as little as 4 percentage points in Wolverhampton and as much as 70 points in Portsmouth. That mismatch matters for this site's own recent attempts to work out why some city centres sell for much less than their surrounding city and others barely discount at all — including the specific test this piece set out to run.

This series has already ruled out one candidate explanation and found a second, partial one. 27 of the 29 tracked centres sell below their own city's median, by as much as 53.8% in Middlesbrough. A test of what actually drives the size of that discount found a centre's flat share barely predicts it (Pearson r = -0.12), while the age of the local housing stock does a noticeably better — though still partial — job (r = -0.48, roughly a fifth of the city-to-city variation). The next candidate on the list was floor space: do smaller homes simply sell for less per property? Testing it directly surfaced the EPC coverage problem first.

The EPC register sees houses, not flats, in a city centre

For each centre, we checked what share of flats and what share of houses (detached, semi-detached and terraced pooled) have at least one Energy Performance Certificate matched to their address — regardless of when it was lodged. The gap between the two is wide everywhere, and in some cities it is extreme.

16 of 29 centres shown (widest and narrowest point-gaps); the remaining 13 fall between 13.8 and 55.0 points. Coverage counts any matched certificate against the address, current property stock, detached/semi/terraced/flat only.

Sheffield (S1) and Bradford (BD1) have the lowest absolute flat-coverage rates of the 29 — fewer than 1 in 100 flats in either centre has a certificate on record at all — even though their house-coverage rates (59.3% and 36.8%) are themselves on the low side, which is why their point-gaps sit mid-table rather than at the very top. Bradford's BD1 is 92% flats by address count but only 17% flats once you look at which addresses actually carry a usable certificate — on one verified Bradford street, none of the 59 sold flats on record carry an energy rating at all. In 17 of the 29 centres, the flat-versus-house coverage gap is wider in the centre than it is across the whole city — the shortfall concentrates precisely where flats are most common, not evenly across the register.

Why this matters beyond the EPC data itself

An Energy Performance Certificate is generated when a home is sold, let, or newly built — not on a fixed schedule — so a flat that has sat with the same owner-occupier for years has no reason to have one yet. That is a plausible mechanism, not something this data can prove on its own. What the data does show plainly is the consequence: any analysis built from the EPC register in a flat-heavy city centre is, in practice, describing the minority of houses there far more than the flats that actually dominate the street. That is a caveat worth attaching retroactively to this site's own build-era comparison above, which used the same register — its city-centre age-mix figures lean more heavily on each centre's houses than its true, mostly-flat population would suggest.

Testing floor space anyway — flats against flats, houses against houses

Given the coverage gap, comparing "all city-centre homes" against "all city-wide homes" risks measuring which type of property happened to get assessed, not genuine floor space. So this test holds property type fixed: median floor area for flats in the centre versus flats city-wide, and separately for houses, using only cities with at least 30 EPC-matched properties on each side (27 of 29 for flats; 28 of 29 for houses).

The pattern doesn't hold a direction. Middlesbrough and Leicester carry the two biggest discounts in the whole series, yet their centre flats are larger than the city-wide flat, not smaller. Newcastle's centre flats are noticeably smaller than the city-wide figure despite a middling discount. Bath's premium centre does have bigger-than-average flats, but so does heavily-discounted Leicester — the same floor-space signal points to opposite price outcomes in different cities. Across all 27 qualifying cities, the correlation between the flat floor-space gap and discount size is r = -0.11; for houses (28 qualifying cities) it is r = -0.17. Both are weaker than flat share's already-weak r = -0.12, and nowhere near build-era mix's r = -0.48.

What's left unexplained

Three candidate drivers of a city centre's discount size have now been tested on the same 29-city set: flat share (r = -0.12), build-era mix (r = -0.48), and floor space (r = -0.11 for flats, -0.17 for houses). Only build-era mix shows a relationship worth calling a partial explanation, and even that accounts for roughly a fifth of the city-to-city variation. The rest — including why Middlesbrough discounts by more than half while Cambridge barely discounts at all — is still open. Property type, age and size are the stock characteristics most readily available in Land Registry and EPC data; what's left is more likely demand-side (local job markets, student populations, buy-to-let concentration) than anything about the buildings themselves.

See how any of these 29 centres compares on the ground: the postcode tool shows individual sold prices and certificate status address by address. For the wider series, the city centre price-versus-flat-share piece and the rest of the Market Analysis section carry the earlier instalments.