Official Statistics

Energy Performance of Buildings Certificates: change note

Published 30 July 2026

Applies to England and Wales

The Ministry of Housing, Communities and Local Government (MHCLG) has previously informed users of changes to the Energy Performance of Buildings Certificates (EPC) quarterly statistics. This note provides information on recent improvements to the data quality leading to a reduction in the number of EPC lodgements across quarters.

The Energy Performance of Buildings Certificates quarterly statistics are based on EPCs issued for domestic and non-domestic buildings and Display Energy Certificates (DECs) issued for buildings occupied by public authorities. The certificates are lodged on the Energy Performance of Buildings Registers (“the Register”) for England and Wales.

What was done to improve the data quality?

To improve the quality of the statistical dataset we have added an extra step to the de-duplication process of the new dwellings EPC data to include certificates without an assigned UPRN.

In addition to the usual cleaning process:

• Step 3: Removes multiple certificates for new dwelling EPCs with assigned UPRNs.

New step added:

• Step 4: Using the full data set for new dwellings with and without URPNS, removes multiples using all address variables and floor level. But keep certificates where multiples are more than 4.

This step was added to the process at Q2 2026 to address possible duplicates within the data set with certificates where a UPRN was not assigned. Originally these had been excluded from the de-duplication process.

Once step 3 had been performed to remove multiples using the UPRN and address fields, the rest of the data, where UPRNs had not been assigned, were added back into the main dataset and step 4 was performed to remove multiples using just the address fields and floor level (to ensure certificates that might have the same address but be separate flats on different floors were not excluded erroneously). To account for where EPC assessors have lodged a very large number of certificates to represent new flats that may not have been built yet or allocated a UPRN, addresses with more than 4 multiples were kept in the dataset.

Then the certificate with the most recent lodgement date was retained in the data set. In rare cases where the lodgement date was the same for the duplicate certificates, the code selected one certificate at random to keep and the rest were removed. 

How this impacts the statistics

Data will look slightly lower across all quarters for new dwellings only (less than 1%). The table below shows the number of duplicate certificates removed by year for both England and Wales during the processing for Q2 2026.

Table 1: Reduction in EPC lodgements for new dwellings across years for England and Wales

Year	Number removed	Year	Number removed
2008	93	2018	97
2009	293	2019	67
2010	121	2020	65
2011	127	2021	301
2012	97	2022	316
2013	88	2023	230
2014	108	2024	150
2015	91	2025	250
2016	113	2026	347
2017	113	Total	3067

What this means for users

The statistical release published on 30 July 2026 will report on the new figures following the amendments to the data processing. Live tables will be revised historically, however previous statistical releases will not be changed.

Data will look lower across all quarters for new dwellings only.

Future work

We recognise that there may be more multiple records that have been lodged on the Register in error that the above method will not remove, for example for existing dwellings. We are consulting on whether users require the existing dwellings dataset to also be de-duplicated.

If you have any questions or comments on this process or would like to feed into the decision making, please contact us at EPBStats@communities.gov.uk.