Determinants of Building-Sector CO₂ Emissions in the EU: A Combined Econometric and Machine Learning Approach

Marco Mele, Alberto Costantiello, Fabio Anobile and Angelo Leogrande ()
Additional contact information
Marco Mele: UniTE - Università degli Studi di Teramo
Alberto Costantiello: LUM - Università LUM Giuseppe Degennaro = University Giuseppe Degennaro
Angelo Leogrande: LUM - Università LUM Giuseppe Degennaro = University Giuseppe Degennaro

Working Papers from HAL

Abstract: This paper evaluates the structural, environmental, and climatic factors influencing carbon dioxide emissions from the building sector (CBE) in 27 European Union member states from 2005 to 2023. This analysis uses panel data from the World Bank and four econometric models-Random Effects, Fixed Effects, Dynamic Panel GMM, and Weighted Least Squares-coupled with machine learning and clustering to provide a robust analysis of emissions. The econometric models show that all models support a negative relationship between agriculture, forestry, and fishing value added (AFFV) and forest area (FRST), suggesting that a robust rural economy and substantial natural carbon sinks are accompanied by lower emissions in the building sector. On the other hand, water stress (WSTR), PM2.5 pollution, heating and cooling degree days, and nitrous oxide emissions (N2OP) are found to significantly, yet positively, affect CBE. Tests of diagnostic analyses support Fixed Effects and Weighted Least Squares models, whereas results from GMM models are limited by instrument validity violations. In machine learning analysis, K-Nearest Neighbors (KNN) models are found to be most diagnostic, with all performance metrics being improved, establishing a prominent role for coal electricity, water stress, agricultural intensities, and climatic factors. Subsequently, a solution with 10 clusters, selected using Bayesian Information Criteria and silhouettes, identified a set of environmental and economic characteristics based on differences between low-and high-emission groups. High-emitting groups result from agricultural intensification, pollution, and low energy efficiency, while low-emitting groups are associated with renewable energy, low pollution, and a favorable climate. This analysis, hence, presents a multifaceted assessment of building sector emissions, with climatic, structural, and energy transition patterns as driving factors for meeting decarbonization targets for the European Union.

Keywords: C38; Q56; Q41; Q54; Cluster analysis C33; Environmental and climatic drivers; Machine learning prediction; Panel data econometrics; Building-sector carbon emissions; Building-sector carbon emissions Panel data econometrics Machine learning prediction Environmental and climatic drivers Cluster analysis C33 Q54 Q41 Q56 C38 (search for similar items in EconPapers)
Date: 2025-12-12
Note: View the original document on HAL open archive server: https://hal.science/hal-05413150v1
References: Add references at CitEc
Citations:

Downloads: (external link)
https://hal.science/hal-05413150v1/document (application/pdf)

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:hal:wpaper:hal-05413150

Access Statistics for this paper

More papers in Working Papers from HAL
Bibliographic data for series maintained by CCSD ().