Calculated figure

Below B on paper

This is a calculated figure, not a count. It estimates how many large let buildings rated below EPC B would come back at B or better on a fresh certificate, using what actually happened when buildings like them were re-assessed.

Below B today

25,735

Rated before 15 June 2022

11,628

Below B on paper (calculated)

≈5,329

The evidence: what happened on re-assessment

On 15 June 2022 the national calculation methodology behind non-domestic EPCs changed, and the carbon factor used for grid electricity was cut sharply. Ratings are carbon-based, so electrically heated buildings assessed after that date score far better than the same building assessed before it. The register shows the step on the day: the median rating of electrically heated certificates fell from the low 70s to the high 50s within a fortnight, while gas-heated ones did not improve.

Method

1. Pairs of consecutive certificates for the same building, at least 180 days apart, floor area within 10%, the earlier one over 1,000 m² and rated C to G, and the same heating fuel on both.
2. For pairs either side of 15 June 2022, the share whose later certificate is A+, A or B, separately for electric and gas heating.
3. For each council: below-B buildings on pre-June-2022 certificates, by heating fuel, multiplied by those shares. Arithmetic on counts; nothing is fitted or forecast.
4. The national figure is the sum of the councils’ figures, and the build fails if it is not.

What this does not control for

• Works. The register cannot see improvements made between certificates. Pairs re-assessed before the change came back at B or better 19.8% (electric) and 17.1% (gas) of the time; that is roughly what re-assessment did when the methodology did not change, and part of the across-change rate will be works too.
• Who chooses to re-assess. Owners who expect a better rating are more likely to commission one. The baseline is also made of buildings whose owners chose to re-assess, which limits but does not remove this.
• Units and buildings. Certificates are often lodged per unit. The 2031 proposal applies to buildings over 1,000 m², and a large building let floor by floor may appear here as several smaller units.
• Letting. A building counts as let if any of its certificates was lodged for letting. That undercounts, and cannot see a let that has ended.

Two earlier versions were wrong

The first version ranked councils on below-B share standardised for building type and size. Its rank correlation with the raw share was 0.91: it was the raw count under another name. Adding methodology era and heating fuel to the standardisation left it at 0.88 and 0.79, and its top ten were all councils with 26 to 67 large buildings, where chance alone moves the share by around eight points. Neither was published. The re-assessment record is what the register shows directly, so it replaced them.

Where the gap is mostly paperwork

The working for every council is in epc-b-2031-by-council.csv, and the re-assessment counts in reassessment-record.csv.