Birmingham's city centre is 98.9% flats, and it's barely cheaper than the rest of Birmingham. Middlesbrough's centre is only 4.1% flats — overwhelmingly older terraced streets — and it's the single most discounted city centre in the UK, selling for 53.8% less than the city around it. A previous Homecost analysis found that flat share barely predicts the size of a centre's discount at all. This piece tests the next candidate explanation directly: how old the housing actually is.
Flat share wasn't the answer
UK City Centres Are Usually Cheaper, Not Pricier, Than the Suburbs found that 27 of the UK's 29 biggest city centres sell for less than their surrounding city — from Middlesbrough's TS1 at −53.8% down to a near-exact tie in Cambridge, with small premiums only in Oxford and Bath. That piece tested whether a bigger flat share explained a bigger discount and found almost no relationship (Pearson r = −0.12). An earlier study, City Centre Flat Share vs House Prices Across 29 UK Cities, found an equally weak link (r = −0.19) between flat share and a centre's raw price level. Whatever drives the size of a city-centre discount, it isn't simply how many flats are sold there.
Testing build-era mix instead
Using the same 29 city-centre postcode districts and their parent local authorities, this analysis pulls the latest Energy Performance Certificate on record for every matched address, both inside each centre postcode and across its wider local authority (HM Land Registry sale definitions and EPC data, fetched 27 October–3 November 2026). Each certificate's build-era band is split at 1950: "pre-1950" covers the register's "before 1900", "1900–1929" and "1930–1949" bands; "1950–1966" and every later band count as post-1950. Roughly two-thirds of the addresses in scope link to a usable certificate (1,616,465 of 2,426,243 addresses across the 29 local authorities) — the percentages below describe that certified subset, not every property on the ground. Every centre figure rests on at least 81 certified addresses (Leeds); every city-wide figure on at least 19,823 (Cambridge).
A stronger predictor, but running backwards
Across the 29 cities, a centre's raw pre-1950 share correlates with the size of its price discount at r = −0.48 — using the gap between the centre's pre-1950 share and its own city's pre-1950 share instead gives r = −0.44. Both readings put build-era mix at roughly 20–23% of the variation explained, well above flat share's 1–2%. But the direction cuts against the "historic core commands a premium" story that Bath and Oxford might suggest: centres with an older-than-their-city housing mix tend to carry bigger discounts, not smaller ones.
The 29-city comparison
| City | Centre | Centre pre-1950% (n) | City-wide pre-1950% (n) | Gap (pp) | Price vs. city-wide |
|---|---|---|---|---|---|
| Bath | BA1 | 52.2% (6,100) | 42.0% (29,913) | +10.2 | +20.0% |
| Oxford | OX1 | 42.4% (2,482) | 50.2% (21,651) | -7.8 | +7.8% |
| Cambridge | CB1 | 45.8% (9,484) | 42.5% (19,823) | +3.3 | -1.0% |
| Hull | HU1 | 9.3% (270) | 51.2% (44,498) | -41.9 | -1.6% |
| Birmingham | B1 | 6.9% (347) | 58.8% (156,296) | -51.9 | -1.7% |
| Manchester | M1 | 3.5% (171) | 60.6% (66,177) | -57.1 | -3.2% |
| Exeter | EX4 | 39.7% (9,401) | 41.0% (22,032) | -1.3 | -5.0% |
| York | YO1 | 43.4% (836) | 39.7% (37,811) | +3.7 | -5.6% |
| Norwich | NR1 | 52.4% (4,919) | 47.3% (25,400) | +5.1 | -7.3% |
| Reading | RG1 | 69.4% (6,659) | 48.7% (26,716) | +20.7 | -13.6% |
| Portsmouth | PO1 | 72.1% (3,421) | 72.8% (33,985) | -0.7 | -13.9% |
| Liverpool | L1 | 32.3% (232) | 63.8% (71,727) | -31.5 | -17.0% |
| Bristol | BS1 | 34.7% (447) | 61.5% (68,488) | -26.8 | -17.1% |
| Southampton | SO14 | 59.4% (2,668) | 51.7% (34,257) | +7.7 | -17.3% |
| Plymouth | PL1 | 47.2% (2,875) | 40.9% (45,583) | +6.3 | -17.8% |
| Stoke-on-Trent | ST1 | 65.0% (6,445) | 50.1% (46,950) | +14.9 | -18.8% |
| Wolverhampton | WV1 | 40.5% (1,844) | 43.4% (37,884) | -2.9 | -18.9% |
| Preston | PR1 | 64.7% (12,189) | 45.8% (27,833) | +18.9 | -19.0% |
| Leeds | LS1 | 6.2% (81) | 45.3% (139,639) | -39.1 | -23.5% |
| Nottingham | NG1 | 60.3% (468) | 47.3% (49,052) | +13.0 | -25.0% |
| Derby | DE1 | 61.0% (2,300) | 42.0% (47,402) | +19.0 | -25.0% |
| Newcastle | NE1 | 26.2% (404) | 49.8% (46,830) | -23.6 | -28.3% |
| Swansea | SA1 | 68.1% (7,486) | 45.3% (41,149) | +22.8 | -31.0% |
| Coventry | CV1 | 73.6% (3,997) | 51.7% (59,666) | +21.9 | -31.3% |
| Cardiff | CF10 | 10.6% (2,566) | 41.2% (61,181) | -30.6 | -33.9% |
| Sheffield | S1 | 89.5% (143) | 53.9% (85,357) | +35.6 | -37.3% |
| Bradford | BD1 | 93.1% (116) | 55.3% (97,300) | +37.8 | -42.6% |
| Leicester | LE1 | 62.1% (214) | 59.0% (47,092) | +3.1 | -47.8% |
| Middlesbrough | TS1 | 97.5% (4,002) | 41.7% (25,957) | +55.8 | -53.8% |
Gap is the centre's pre-1950 share minus its city-wide pre-1950 share — positive means the centre's certified stock skews older than the city around it; negative means it skews newer.
Where the pattern holds hardest
Middlesbrough's TS1 is 97.5% pre-1950 by certified stock, against 41.7% for the Middlesbrough local authority as a whole — the widest age gap of any centre in the sample, at +55.8 percentage points — and it carries the single biggest discount in the table, at −53.8%. Bradford's BD1 (93.1% vs 55.3%, +37.8pp) and Sheffield's S1 (89.5% vs 53.9%, +35.6pp) sit in the same corner: each centre is dominated by older, pre-war terraced and industrial-era housing that never went through the flat-led redevelopment its wider city later picked up, and each sells well below its own city's median.
Where it breaks
The pattern isn't universal. Leicester's LE1 carries the second-biggest discount in the table (−47.8%) with almost no age gap at all — 62.1% pre-1950 in the centre against 59.0% city-wide, a gap of just +3.1 points. Whatever drives Leicester's discount, it isn't build-era mix. At the other extreme, Birmingham's B1 (6.9% pre-1950 against a 58.8% city-wide figure, a −51.9pp gap — overwhelmingly modern flats set against an older city) and Manchester's M1 (3.5% vs 60.6%, a −57.1pp gap, the most extreme of any centre in either direction) are barely discounted at all, at −1.7% and −3.2% respectively. Oxford breaks the pattern from the premium side: its centre is actually less pre-1950 than the city around it (42.4% vs 50.2%, a −7.8pp gap) yet still commands a 7.8% premium — the opposite of what a simple "older means pricier" rule would predict.
What this does and doesn't explain
Plain pre-1950 age, on its own, is not what makes a city centre expensive — the Middlesbrough, Bradford and Sheffield cores show that an old centre can be among the cheapest property in its own city. What sets Bath and Oxford apart isn't simply that their centres are old; Bath's Georgian terraces and Oxford's collegiate core are a specific, low-density, well-preserved kind of older housing, not the ex-industrial terraced stock that dominates the biggest-discount centres. Build-era mix accounts for more of the discount than flat share does, but it still leaves close to 80% of the variation unexplained — Leicester's near-zero age gap next to a near-50% discount is the clearest reminder that no single variable tested so far, flat share or build-era, fully explains why some centres are cheap and others aren't.
What it means at street level
At Middlesbrough's TS1, recent resales on Linthorpe Road's Constantine Court have changed hands from the mid-£60,000s to the low six figures — see TS1 3GA on Homecost for sold prices on an actual street in the postcode. A buyer comparing two city centres with very different age profiles should weigh a discount against what's actually driving it locally — build quality, floor area, EPC rating and running costs can all vary with a property's age — rather than assume an older centre is automatically a bargain or a newer one automatically a premium address. Speak to a qualified adviser before acting on any buying or mortgage decision.
This analysis covers 2,426,243 Land-Registry-linked addresses across the 29 local authorities and 1,616,465 matched EPC certificates, fetched 27 October–3 November 2026. Browse more market analysis on the Homecost blog, where every figure in this piece — and the two studies it builds on — is sourced back to HM Land Registry and the national EPC register.