Analysis, not a count
Which councils beat their own housing stock
This page is different from the rest of the site. Every other figure here is counted directly from certificates. This one is calculated, by comparing each council against what its own housing stock would predict. The method is set out below in full, and the working is in the download, so you can check it rather than trust it.
Ranking councils on the raw share of rentals below EPC C mostly ranks them on the age of their buildings. Pre-1900 rented stock is 77% below C. Stock built in the early nineties is 40%. That 37% gap swamps any real difference in effort.
So this table asks a different question: given the housing stock a council actually has, is it doing better or worse than you would expect? A ratio of 1.00 means exactly as expected. Above 1.00 is worse than its own stock predicts. Below is better.
Worse than their housing stock predicts
| # | Council | Below C | Expected | Ratio | Raw share |
|---|---|---|---|---|---|
| 1 | Tandridge | 3,052 | 2,514 | 1.21 | 66% |
| 2 | Isle of Anglesey | 2,944 | 2,525 | 1.17 | 80% |
| 3 | Southend-on-Sea | 9,916 | 8,541 | 1.16 | 66% |
| 4 | Thurrock | 4,213 | 3,643 | 1.16 | 50% |
| 5 | Stratford-on-Avon | 3,440 | 2,986 | 1.15 | 62% |
| 6 | Castle Point | 2,094 | 1,820 | 1.15 | 65% |
| 7 | Sandwell | 9,051 | 7,874 | 1.15 | 61% |
| 8 | Cornwall | 22,874 | 19,920 | 1.15 | 65% |
| 9 | South Hams | 3,120 | 2,718 | 1.15 | 68% |
| 10 | Uttlesford | 1,792 | 1,566 | 1.14 | 63% |
Better than their housing stock predicts
Several of these look unremarkable on the raw share. Westminster is 41% below C, which reads as middling until you account for how old and dense its rented stock is.
| # | Council | Below C | Expected | Ratio | Raw share |
|---|---|---|---|---|---|
| 1 | Rushmoor | 1,697 | 2,109 | 0.80 | 35% |
| 2 | Tower Hamlets | 9,639 | 11,906 | 0.81 | 24% |
| 3 | Newcastle upon Tyne | 9,701 | 11,716 | 0.83 | 43% |
| 4 | Westminster | 15,866 | 19,154 | 0.83 | 41% |
| 5 | Hackney | 10,769 | 12,952 | 0.83 | 37% |
| 6 | Gateshead | 4,066 | 4,817 | 0.84 | 42% |
| 7 | Camden | 13,207 | 15,564 | 0.85 | 44% |
| 8 | Telford and Wrekin | 4,291 | 5,012 | 0.86 | 41% |
| 9 | North Tyneside | 3,740 | 4,360 | 0.86 | 43% |
| 10 | Kensington and Chelsea | 11,493 | 13,355 | 0.86 | 47% |
How this is calculated
This is indirect standardisation, the same technique behind a standardised mortality ratio. Nothing here is fitted or predicted; it is arithmetic on counts.
- Split every privately rented property in England and Wales into strata of construction age band, property type and built form. These are the characteristics a landlord cannot change.
- For each stratum, work out the England and Wales share rated below EPC C.
- A council's expected count is the sum, across strata, of its own number of homes in that stratum multiplied by the national rate for it.
- The ratio is actual divided by expected.
Only strata holding at least 500 properties nationally are used, because a rate from a handful of homes is noise. Those 259 strata cover 99% of the rented sector, and both the actual and the expected count are restricted to them, so the two sides of the ratio always compare like with like. Councils with fewer than 2,000 rented homes inside those strata are left out entirely.
As a check on the arithmetic: summed across England and Wales the ratio comes to exactly 1.00, which it must if expected counts are being derived correctly.
What it does not tell you
It controls for housing stock and nothing else. It does not account for local incomes, rural or urban character, how active a council has been on enforcement, or what any individual landlord has done. A ratio of 1.00 does not mean an area is in good shape; it means it is typical for its stock. An area can sit below 1.00 and still have most of its rentals failing.
Deliberately excluded from the strata: wall and roof construction, heating fuel and insulation. Those are the things retrofit work changes, so controlling for them would absorb the very difference this table exists to show.
The straight counts
Every other table on this site is a direct count with no calculation involved.