Reducing energy consumption is a key policy focus for mitigating climate change. This study investigates the determinants of residential building energy efficiency, leveraging expert insights from Energy Performance Certificates (EPCs) to develop a machine learning prediction framework. Datasets from countries at distinct latitudes, the UK and Italy, are analyzed to identify potential regional variations in the factors influencing energy efficiency. Findings reveal the crucial role of factors related to heating systems and insulation materials in the determination of the building’s efficiency. Also, there is evidence of the superior ability of non-linear machine learning models to capture complex relationships between building characteristics and efficiency. A scenario analysis further demonstrates the cost-effectiveness of policies informed by machine learning recommendations. 

Dimension reduction techniques have been proposed to cope with some modelling and forecasting issues with large and complex datasets such as overfitting, over-parametrization and inefficiency. Among all dimension reduction techniques, the most widespread are principal component analysis and factor analysis. Recently, Random Projection (RP) techniques became popular in many fields due to their simplicity and effectiveness and found applications to machine learning, statistics and econometrics. The basis of the RP technique relies on the remarkable result in the Johnson-Lindenstrauss lemma, which provides some conditions to achieve an effective reduction of the size of the data, without altering their information content. This chapter reviews the most used dimensionality reduction techniques, introduces random projection methods and shows their effectiveness in time series analysis through a simulation study and some original applications to tracking and forecasting financial indexes and to predicting electricity trading volumes. Our empirical results suggest that random projection preprocessing of the data does not jeopardize the validity of inference and prediction procedures and possibly improves their efficiency.

Do sovereign credit ratings take into account physical and transition climate risks? This paper empirically addresses this question using a panel dataset that includes a large sample of countries over two decades. The analysis reveals that higher temperature anomalies and more frequent natural disasters—key indicators of physical risk—are associated with lower credit ratings. In contrast, transition risk factors do not appear to be systematically integrated into credit ratings throughout the entire sample period. However, following the Paris Agreement, countries with greater exposure to natural disasters received comparatively lower ratings, suggesting that credit rating agencies are increasingly recognizing the significance of physical risk for sovereign balance sheets. Additionally, more ambitious CO2 emission reduction targets and actual reductions in CO2 emission intensities are associated with higher ratings post-Paris Agreement, indicating that credit rating agencies are beginning to pay more attention to transition risk. At the same time, countries with high levels of debt and those heavily reliant on fossil fuel revenues tend to receive lower ratings after the Paris Agreement. Conversely, sovereigns that stand to gain from the green transition—through revenues from transition-critical materials—are assigned higher sovereign ratings after 2015.