Does a C-or-better EPC sell for more in B5 4 (Birmingham)?
No measurable EPC premium here. In postcode sector B5 4 (Birmingham), comparing like with like — same floor area, property type, build period and year of sale — the difference is +0.5% (95% range −5.6% to +7.0%): the EPC band itself is not reflected in sale prices here. The raw gap is bigger: C+ homes sold for 5% more per square metre than D–G homes (median £3,327 vs £3,169, 316 sales 2015–2026), mostly because they are different homes. Across England and Wales the like-for-like figure is +1.2%. Data: HM Land Registry Price Paid to 30 September 2026 matched to the EPC register.
From the raw gap to like-for-like
The homes are not alike: C+ homes here have a median floor area of 53 m² against 57 m² for D–G; both groups are most often built in 2003–2006.
| Comparing C+ with D–G homes… | Difference | 95% range |
|---|---|---|
| Raw gap per m² (medians) | +5% | — |
| Whole price, nothing held equal | +0.7% | −7.9% to +10.0% |
| Same floor area | +7.8% | +0.5% to +15.7% |
| + same property type | +7.8% | +0.5% to +15.7% |
| + same build period | −0.4% | −6.9% to +6.5% |
| + same year of sale | +0.5% | −5.6% to +7.0% |
The first row compares median prices per m²; the second compares whole prices; each later row adds one more thing that must be equal, ending with the like-for-like premium (regression of log price; build period from the certificate, 12 bands). The raw gap per m² and the whole-price gap differ because C+ and D–G homes differ in size.
Sold prices by EPC band in B5 4
| EPC band (last certificate before sale) | Sales | Median £ per m² |
|---|---|---|
| B | 122 | £3,635 |
| C | 149 | £3,207 |
| D | 38 | £3,195 |
| E | 7 | too few |
A median is shown when a band has at least 10 sales.
C or better vs D–G, by property type
| Type | C+ sales | C+ £/m² | D–G sales | D–G £/m² |
|---|---|---|---|---|
| Flat / maisonette | 271 | £3,327 | 45 | £3,169 |
Compared with the wider area
| Area | Sales | Like-for-like | 95% range |
|---|---|---|---|
| Sector B5 4 | 316 | +0.5% | −5.6% to +7.0% |
| District B5 | 1,582 | 0.0% | −4.4% to +4.7% |
| England and Wales | 1,000,000 (sample) | +1.2% | +0.9% to +1.5% |
Average gas use in B5 4 (2024): meters vs certificates
A rough comparison: the certificate figure is an upper-bound estimate (primary energy including lighting, converted with one factor), so treat the share as an order of magnitude.
| Postcodes with published gas data (5+ meters) | 2 (12 meters) |
|---|---|
| Median gas use per meter (median of postcodes) | 25,221 kWh |
| Gas-heated homes with an EPC | 207 |
| Energy implied by their certificates (median) | 9,855 kWh |
| Metered gas as a share of the certificate figure (median of 1 postcodes) | 157% (middle half 157%–157%) |
Certificates by band imply: C 8,610 kWh (45 homes) · D 10,835 kWh (151 homes). Across England and Wales, real use is much closer to the certificate for A–C homes than for D–G homes, so the saving an upgrade promises is usually smaller than the certificate suggests.
Certificate energy = primary energy per m² × floor area ÷ 1.13, for homes whose latest certificate shows mains gas heating; it also covers lighting and other uses, so it is an upper bound for gas. DESNZ gas is weather-corrected and covers all meters in the postcode, with or without a certificate.
316 of 497 house and flat sales in the sector since 2015 (64%) were matched to a certificate.
How this is calculated
- Every standard sale (HM Land Registry Price Paid, category A) since January 2015 is matched to the latest energy certificate of the same address (postcode + house number or name + flat) lodged from 10 years before to 3 months after the sale; sales with no certificate or an ambiguous match are left out. Across England and Wales 92% of sales match.
- Price per m² uses the certificate's total floor area (15–1,000 m²).
- Like-for-like premium: regression of the log of price on “C or better”, log floor area, property type, the certificate's build period and the year of sale. The 95% range shows how precise the estimate is; when it includes zero, the data cannot tell the premium apart from none.
Caveats
- The band is the one on the certificate before the sale; certificates are a median 0.8 years old at the sale, but a quarter are over 4 years old and some homes were improved since.
- Condition, garden, parking, views and the exact position within the sector are not known; a premium could hide in them, or they could hide one.
- Individual sales and addresses are not shown.