Working patterns, leisure dynamics, and energy consumption: a time-use analysis for Italy

ABSTRACT

Recent working time reduction policies aim at achieving a triple dividend, yet their impact is uncertain due to potential rebound effects. Saved work time may be spent on energy-intensive leisure activities, complicating the overall outcomes. In our study, we conduct a time use analysis to investigate how different work schedules affect leisure time and its associated energy consumption, using the Italian Time Use Survey from Istat. Our analysis examines how predictive factors such as employment status, gender, and variables related to the household environment affect preferences for leisure activities, employing various econometric models, including Dirichlet regression which best suits compositional data. Results suggest that working time predominantly affects leisure time allocation; employment status, gender, and the presence of a teenager in the household also play a role. From an energy usage standpoint, a time rebound effect is present, although it is not mediated by the aforementioned factors.

KEYWORDS:

• Composition effects
• energy use
• leisure preferences
• rebound effects
• time-use analysis
• working time reduction policies

Disclosure statement

No potential conflict of interest was reported by the author(s).

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/02614367.2024.2363802

Notes

1. https://www.ISTAT.it/en/archive/202524

2. Released on July 18th, 2017.

3. For this reason, full completion of the form is accounted for by a binary variable which returns 1 if the diary was properly filled.

4. See Appendix, Table 7.

5. See Appendix, Table 7.

6. These macro-categories are: essential, secondary and travelling activities.

7. To make this determination, we aggregate the coefficients associated with WT, $\mathrm{W}{\mathrm{T}}^2$, the interactions of WT with ES, Gender, and having at least a 16-year-old child, regardless of their statistical significance.

8. Except for the binary variable ‘giocom’, ‘rip’ and ‘colf’ which create problems with the computation of predictions.

Additional information

Notes on contributors

Edoardo Alberti

Edoardo Alberti is a Phd candidate at the University of Siena, Pisa and Florence (Joint Doctoral Program of Tuscan Universities). He received a bachelor’s degree in Bank, Finance and Financial Markets from University of Pisa, a master’s degree in Economics from Santa School of Advanced Studies and University of Pisa and a master’s degree in Economics, curriculum in Public Policy At Vrije Universiteit, Amsterdam. His research interests are labor economics, energy transition, sustainable development and environmental economics.

Paolo Frumento

Paolo Frumento is an Associate Professor of Statistics at the University of Pisa. Before this, he served as an Assistant Professor at Karolinska Institutet in Stockholm. His research focuses on statistical models, with a particular emphasis on quantile regression models, and their application in survival analysis and causal inference.