Across London's 32 boroughs (excluding the City of London, where the housing stock is too small to rank), a borough with weak Energy Performance Certificate coverage on its flats tends also to have weak coverage on its houses — a correlation of r=0.54. That link looked, on first inspection, like a genuine citywide pattern. It isn't quite. Pull out the five boroughs with the lowest flat coverage — Westminster, Kensington and Chelsea, Camden, Tower Hamlets and Hackney — and the correlation among the remaining 27 boroughs drops to r=0.36. Remove the ten lowest, and it falls further still, to r=0.20, a weak relationship by any normal reading.
The citywide picture
An EPC is lodged only when a home is sold, let, or newly built — it isn't a continuous record, so coverage is really a proxy for how often a borough's housing stock has changed hands or been built out in the EPC register's lifetime (in force since 2008, with reliable digital records from the early 2010s). Across London, 242,060 of 1,030,227 flat addresses (23.5%) and 754,171 of 1,068,756 house addresses (70.6%) have at least one certificate on file.
| Borough | Flat EPC coverage | House EPC coverage |
|---|---|---|
| Westminster | 6.7% | 62.5% |
| Kensington and Chelsea | 9.4% | 66.2% |
| Camden | 11.4% | 58.4% |
| Tower Hamlets | 14.2% | 73.1% |
| Hackney | 15.4% | 62.9% |
| Islington | 16.4% | 62.7% |
| Lambeth | 18.0% | 67.8% |
| Southwark | 18.2% | 70.7% |
| Hammersmith and Fulham | 19.3% | 70.3% |
| Kingston upon Thames | 19.7% | 71.0% |
| Greenwich | 21.9% | 73.8% |
| Haringey | 22.9% | 68.2% |
| Barnet | 23.2% | 68.1% |
| Wandsworth | 23.7% | 73.2% |
| Bromley | 23.8% | 70.7% |
| Sutton | 25.8% | 69.1% |
| Brent | 26.8% | 66.6% |
| Croydon | 27.6% | 70.3% |
| Lewisham | 28.1% | 70.8% |
| Newham | 28.6% | 77.2% |
| Richmond upon Thames | 29.8% | 70.8% |
| Ealing | 30.3% | 69.0% |
| Hounslow | 30.4% | 71.0% |
| Merton | 33.1% | 71.1% |
| Harrow | 34.0% | 68.4% |
| Havering | 36.2% | 72.8% |
| Waltham Forest | 39.9% | 72.1% |
| Redbridge | 41.2% | 69.0% |
| Barking and Dagenham | 41.2% | 78.6% |
| Hillingdon | 45.0% | 73.2% |
| Bexley | 46.4% | 70.4% |
| Enfield | 48.5% | 70.4% |
House coverage barely moves across the table — it sits in a tight 58-79% band almost everywhere. Flat coverage is the variable that actually swings, from 6.7% in Westminster to 48.5% in Enfield, a seven-fold range. That asymmetry is the first clue that the correlation is being pulled by one end of the flat-coverage scale rather than reflecting something that moves uniformly across all 32 boroughs.
What happens once the lowest-coverage boroughs are set aside
Westminster, Kensington and Chelsea, Camden, Tower Hamlets and Hackney are unusual on both axes at once: low flat coverage and, with the exception of Tower Hamlets, below-average house coverage too (Camden's 58.4% is the lowest house-coverage figure of any borough in London). Testing whether the citywide correlation survives without them shows it weakens steadily rather than holding flat or vanishing outright:
| Boroughs excluded (lowest flat coverage) | Boroughs remaining | Pearson r |
|---|---|---|
| None | 32 | 0.54 |
| Lowest 3 | 29 | 0.37 |
| Lowest 5 | 27 | 0.36 |
| Lowest 10 | 22 | 0.20 |
| Lowest 15 | 17 | 0.24 |
Two things follow from this gradient. First, the inner cluster is doing real work: cutting just the three most extreme boroughs takes a third off the correlation immediately (0.54 to 0.37), which means Westminster, Kensington and Chelsea and Camden alone account for a meaningful share of the citywide relationship. Second, the correlation does not collapse to zero once they're gone — it settles in the 0.20-0.24 range across the remaining 17-27 boroughs, a weak but real positive relationship among boroughs that are mostly much closer together on both coverage measures. The honest description of London's EPC-coverage link is therefore "part genuine weak citywide pattern, part a handful of inner boroughs at the extreme end of the flat-coverage scale" — not purely one or the other.
Why coverage clusters where it does
The mechanism is straightforward housing-stock turnover, not wealth, choice, or energy performance itself. A certificate only exists because a sale, letting, or new build has triggered one since EPCs became mandatory. Boroughs with large, often period, flat stock bought and held for long periods — the inner boroughs at the top of the flat-coverage ranking — accumulate fewer certificates over time, regardless of how efficient those homes actually are. The same turnover logic applies more weakly to houses, which is why the house-coverage column moves so little across the whole city: London's house stock turns over at a broadly similar rate borough to borough, while its flat stock does not.
Checking a specific address
This coverage pattern is about how many homes in an area have a certificate at all, not what rating any individual home holds. Homecost's postcode tool shows the recorded energy rating, where one exists, for any address — try a Westminster postcode such as W9 2PP to see the pattern in practice, or enter your own.
This piece builds directly on an earlier look at where London's flat and house EPC coverage both lag, which first identified the r=0.54 correlation, and on the full 33-borough coverage ranking that both pieces draw their figures from. The same inner-city/outer-London coverage split shows up again, in a different form, in the Manchester-Birmingham-Leeds EPC coverage mirror, where a single dominant city plays a similar outsized role in its own conurbation's figures.
Based on 2,098,983 London residential addresses matched against the national EPC register (HM Land Registry address data; EPC Open Data register, fetched 4 November 2026) — see more cost intelligence coverage on homecost.uk.