Inequitable efficiency: Unravelling the social and built environment drivers of London's housing energy performance

This study analyses the relationships between sociodemographic factors, building characteristics, energy efficiency and environmental impact in London's residential stock (2011–2021), using 2 million Energy Performance Certificates (EPCs) and sociodemographic data. Employing generalised linear models (GLMs) and machine learning techniques, we identify three key findings. First, building age and heating system efficiency are the dominant predictors of energy performance. Second, sociodemographic factors, including household size, income and age, significantly affect retrofitting outcomes, with low-income and elderly households facing the greatest barriers. Third, longitudinal analysis shows a shift in vulnerability drivers, from neighbourhood-level deprivation in 2011 to household-level income deprivation in 2021. Model comparisons reveal stronger accuracy for GLMs than XGBoost in predicting energy grades, highlighting the potential of data-driven interpretable methods for local authorities. The policy recommendations emphasise the integration of dynamic social support with technical regulations such as Minimum Energy Efficiency Standards (MEES) to address carbon emissions while protecting vulnerable groups.

Cuicheng Zhang: Writing – review & editing, Writing – original draft, Visualization, Validation, Software, Methodology, Formal analysis, Data curation, Conceptualization. Cong Cao: Writing – review & editing, Writing – original draft, Visualization, Methodology, Formal analysis, Conceptualization. Pengyu Zhang: Writing – review & editing, Writing – original draft, Validation, Methodology, Formal analysis, Data curation. R. Michael Alvarez: Writing – review & editing, Validation, Funding acquisition. Ramit Debnath: Writing – review & editing, Writing – original draft, Validation, Supervision, Project administration, Methodology, Funding acquisition, Conceptualization.

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

The replication code is available from https://github.com/CZhang929/Code-for-EPCs-Paper.

 

In this study, publicly available secondary data are used. To acquire the integrated datasets, send a request to Cuicheng Zhang (cuicheng.zhang@eng.ox.ac.uk).