People use the services energy provides – but buildings and technologies determine how much is used

Panel: 5. Energy use in buildings: projects, technologies and innovation

This is a peer-reviewed paper.

Authors:
Gesche Margarethe Heubner, UCL Energy Institute, United Kingdom
Ian Hamilton, UCL Energy Institute
Tadj Oreszczyn, UCL Energy Institute
David Shipworth, UCL Energy Institute

Abstract

Most people in the field of energy research are familiar with the phrase “Buildings don’t use energy: people do” (Janda, 2006). Whilst this is undoubtedly true, it is also true that the same people would use very different amounts of energy in different buildings. This paper addresses the question which class of variables (building factors, socio-demographics, attitudes and self-reported behaviours) contribute most to explaining energy use in buildings. Knowing the contribution of different classes of predictors would indicate what kind of variables need measuring to understand domestic energy consumption. Knowing the relative importance of different predictors can help shape the most effective policy interventions. The paper also discusses the need to collect more relevant people-related variables to give a ‘fair representation’ of the impact of behaviour. We analysed a sample of 991 households approximately representative of the English population. Using regression analysis, we estimated that building factors accounted for about 40% of the variability in energy consumption. Whilst socio-demographics alone also accounted for a substantial part of the variability (~ 25%), the joint regression explained only about 43% of the variability, a modest increase in comparison to the building-factors-only model. Attitudes on climate change and self-reported behaviours on energy added an even smaller amount of explanatory power. This finding, together with the relatively greater temporal constancy of building factors, suggests focusing on building factors to understand domestic energy consumption on a stock level. However, potential important variables such as heating temperatures and heating durations were not collected or only collected via self-report. Measuring the right variables correctly might shift the balance of explanatory power of different variables groups. The results also highlight that more than half of the variability in energy consumption cannot be explained, even not when using such a breadth of predictors.