Short-term predictions for country-level energy demand with top-down and bottom-up stacking

Haoyang Li; Erkal Ersoy; Boulis Ibrahim; Mark E. Schaffer

doi:10.63065/ajafe.2026.1.2.003

Asian Journal of Applied Financial Econometrics (AJAFE)

Volume 1_Number 2 , Tháng 8/2026, Trang 24-47

Short-term predictions for country-level energy demand with top-down and bottom-up stacking

Haoyang Li, Erkal Ersoy, Boulis Ibrahim, Mark E. Schaffer

DOI: 10.63065/ajafe.2026.1.2.003

Ngày nhận bài: 30/04/2026

Ngày nhận bài sửa: 15/08/2026

Ngày duyệt đăng: 20/08/2026

Email: m.e.schaffer@hw.ac.uk

Đơn vị công tác:

Lượt xem: 180

Downloads: 0

Các định dạng trích dẫn

Tóm tắt

Purpose - This paper examines whether stacking (Wolpert, 1992) improves short-term forecasts of global oil consumption and electricity demand during periods of major shocks, including COVID-19, geopolitical conflicts, and trade disruptions. Given diverging trends between OECD and non-OECD economies, we evaluate forecast performance separately across these regions.

Design/methodology/approach - We combine XGBoost, random forest, elastic net, multi-layer perceptron, and a unit fixed-effects model within a stacking framework using a non-negative OLS meta-learner. We implement both a top-down approach, which models aggregate demand directly, and a bottom-up approach, which models disaggregated components and aggregates predictions, to assess the value of decomposition.

Findings - We find that stacking provides a robust benchmark, consistently matching or outperforming the best individual learner across most specifications. In the top-down setting, stacking generally delivers the strongest performance, particularly for global and non-OECD oil demand and global electricity demand. Bottom-up stacking improves accuracy for most oil sub-aggregates and performs comparably to the best base learner for electricity demand by source. Comparing strategies, bottom-up stacking performs better for electricity demand, while both approaches perform similarly for oil demand. Results by demand size suggest that retransformation bias from the log scale (Duan, 1983) and cross-country heterogeneity are particularly important for large economies. Overall, while no single model dominates across all settings, stacking provides a simple and reliable approach to improving forecast accuracy.

Originality/value - Our contribution is threefold. First, we provide a systematic comparison of stacking methods for forecasting global energy demand across OECD and non-OECD economies. Second, we evaluate the relative performance of top-down and bottom-up strategies within a common framework. Third, we examine how forecast performance varies with the scale of energy demand, highlighting the role of retransformation bias (Duan, 1983) and cross-country heterogeneity.

Từ khóa

Stacking, Energy demand, Machine learning, Top-down, Bottom-up, Prediction.

Download

Tài liệu tham khảo

Ahrens, A., E. Ersoy, V. Iakovlev, H. Li, and M. E. Schaffer (2022). An introduction to stacking regression for economists. In International Conference of the Thailand Econometrics Society, pp. 7–29. Springer.

Ahrens, A., C. B. Hansen, M. E. Schaffer, and T. Wiemann (2025). Model averaging and double machine learning. Journal of Applied Econometrics 40(3), 249–269.

Aldabbagh, A. M., A. Economou, and C. Chariton (2024). Forecasting global oil demand: Application of machine learning techniques. Technical report, OIES Paper.

Atiya, A. F. (2020). Why does forecast combination work so well? International Journal of Forecasting 36(1), 197–200.

Bhattacharyya, S. C. and G. R. Timilsina (2009). Energy demand models for policy formulation: a comparative study of energy demand models. World Bank Policy Research Working Paper (4866).

Blanc, S. M. and T. Setzer (2020). Bias–variance trade-off and shrinkage of weights in forecast combination. Management science 66(12), 5720–5737.

Breiman, L. (1996). Stacked regressions. Machine learning 24(1), 49–64.

Breiman, L. (2001). Random forests. Machine learning 45(1), 5–32.

Budescu, D. V. and E. Chen (2015). Identifying expertise to extract the wisdom of crowds. Management science 61(2), 267–280.

Chen, T. and C. Guestrin (2016). Xgboost: A scalable tree boosting system. In Proceedings of the 22nd acm sigkdd international conference on knowledge discovery and data mining, pp. 785–794.

Dees, S., P. Karadeloglou, R. K. Kaufmann, and M. Sanchez (2007). Modelling the world oil market: Assessment of a quarterly econometric model. Energy policy 35(1), 178–191.

Diebold, F. X. and R. S. Mariano (2002). Comparing predictive accuracy. Journal of Business & economic statistics 20(1), 134–144.

Duan, N. (1983). Smearing estimate: a nonparametric retransformation method. Journal of the American Statistical Association 78(383), 605–610.

Ersoy, E., H. Li, M. E. Schaffer, and T. Szendrei (2023). Stacking regression for time-series, with an application to forecasting quarterly us gdp growth. In Optimal Transport Statistics for Economics and Related Topics, pp. 131–149. Springer.

Fatima, T., E. Xia, and M. Ahad (2019). Oil demand forecasting for china: a fresh evidence from structural time series analysis. Environment, Development and Sustainability 21(3), 1205–1224.

Friedman, J. H. (2001). Greedy function approximation: a gradient boosting machine. Annals of statistics, 1189–1232.

Fung, W., K. S. Lam, W. Hung, S. Pang, and Y. Lee (2006). Impact of urban temperature on energy consumption of hong kong. Energy 31(14), 2623–2637.

Goodfellow, I. (2016). Deep learning.

Hoerl, A. E. and R. W. Kennard (1970). Ridge regression: applications to nonorthogonal problems. Technometrics 12(1), 69–82.

Hornik, K., M. Stinchcombe, and H. White (1989). Multilayer feedforward networks are universal approximators. Neural networks 2(5), 359–366.

Hunt, L. C. and Y. Ninomiya (2003). Unravelling trends and seasonality: a structural time series analysis of transport oil demand in the uk and japan. The Energy Journal 24(3), 63–96.

IEA (2025). Oil 2025. Paris. https://www.iea.org/reports/oil-2025. Licence: CC BY 4.0.

Khairalla, M. A., X. Ning, N. T. Al-Jallad, and M. O. El-Faroug (2018). Short-term forecasting for energy consumption through stacking heterogeneous ensemble learning model. Energies 11(6), 1605.

Krogh, A. and J. Vedelsby (1994). Neural network ensembles, cross validation, and active learning. Advances in neural information processing systems 7.

Kuncheva, L. I. and C. J. Whitaker (2003). Measures of diversity in classifier ensembles and their relationship with the ensemble accuracy. Machine learning 51(2), 181–207.

Manderson, E. and T. Considine (2025). The effect of temperature on energy demand and the role of adaptation. Journal of the Association of Environmental and Resource Economists 12(3), 739–774.

Morris, M. (2001). The impact of temperature trends on short-term energy demand.

Opitz, D. and J. Shavlik (1995). Generating accurate and diverse members of a neural-network ensemble. Advances in neural information processing systems 8.

Qian, C., Y. Yu, and Z.-H. Zhou (2015). Pareto ensemble pruning. In Proceedings of the AAAI conference on artificial intelligence, Volume 29.

Reddy, S., S. Akashdeep, R. Harshvardhan, and S. Kamath (2022). Stacking deep learning and machine learning models for short-term energy consumption forecasting. Advanced Engineering Informatics 52, 101542.

Sbrana, G. and A. Silvestrini (2013). Forecasting aggregate demand: analytical comparison of top-down and bottom-up approaches in a multivariate exponential smoothing framework. International Journal of Production Economics 146(1), 185–198.

Sharma, K., Y. K. Dwivedi, and B. Metri (2024). Incorporating causality in energy consumption forecasting using deep neural networks. Annals of Operations Research 339(1), 537–572.

Spiliotis, E., F. Petropoulos, N. Kourentzes, and V. Assimakopoulos (2020). Cross-temporal aggregation: Improving the forecast accuracy of hierarchical electricity consumption. Applied Energy 261, 114339.

Tibshirani, R. (1996). Regression shrinkage and selection via the lasso. Journal of the Royal Statistical Society Series B: Statistical Methodology 58(1), 267–288.

Timmermann, A. (2006). Forecast combinations. Handbook of economic forecasting 1, 135–196.

Urban, F., R. Benders, and H. Moll (2007). Modelling energy systems for developing countries. Energy policy 35(6), 3473–3482.

Van der Laan, M. J., E. C. Polley, and A. E. Hubbard (2007). Super learner. Statistical applications in genetics and molecular biology 6(1).

Wang, X., R. J. Hyndman, F. Li, and Y. Kang (2023). Forecast combinations: An over 50-year review. International Journal of Forecasting 39(4), 1518–1547.

Widiarta, H., S. Viswanathan, and R. Piplani (2009). Forecasting aggregate demand: An analytical evaluation of top-down versus bottom-up forecasting in a production planning framework. International Journal of Production Economics 118(1), 87–94.

Wolpert, D. H. (1992). Stacked generalization. Neural networks 5(2), 241–259.

Yu, L., Y. Zhao, L. Tang, and Z. Yang (2019). Online big data-driven oil consumption forecasting with google trends. International Journal of Forecasting 35(1), 213–223.

Zhang, Y., S. Burer, W. Nick Street, K. P. Bennett, and E. Parrado-Hernández (2006). Ensemble pruning via semi-definite programming. Journal of machine learning research 7(7).

Zhou, Z.-H., J. Wu, and W. Tang (2002). Ensembling neural networks: many could be better than all. Artificial intelligence 137(1-2), 239–263.

Zou, H. and T. Hastie (2005). Regularization and variable selection via the elastic net. Journal of the Royal Statistical Society Series B: Statistical Methodology 67(2), 301–320.