Author(s)

Sinan Jin ¹, Siyuan Zhong ¹

Affiliation(s)

¹ School of Labor Economics, China University of Labor Relations, Beijing, China

Corresponding Author

Sinan Jin

ABSTRACT

Under the background of carbon peaking and carbon neutrality, the increasing consumption of renewable electricity by industrial users has made green electricity premium an important factor affecting enterprise operating costs and low-carbon transition decisions. However, existing studies mainly focus on historical measurement and policy discussion, while insufficient attention has been paid to the dynamic forecasting of green electricity premium and the early warning of cost burden risks. To address this issue, this paper constructs a province–industry–month spatio-temporal dataset by integrating green electricity trading data, coal-fired benchmark electricity price data, industrial cost data, energy structure data, electricity transmission data, and green certificate market data. Based on this dataset, a Spatio-Temporal Graph Attention Network for Cost Risk Warning, namely ST-GAT-CRW, is proposed. The model constructs a multi-relation graph by combining geographical adjacency, electricity transmission, and price correlation relationships, uses a Graph Attention Network to capture spatial linkages among province–industry nodes, and adopts a Transformer Encoder to model temporal dependencies of green electricity premium. Meanwhile, a multi-task learning framework is designed to simultaneously conduct green electricity premium forecasting and cost burden risk classification. Comparative experiments with Autoregressive Integrated Moving Average, Vector Autoregression, Extreme Gradient Boosting, Long Short-Term Memory, Transformer, Graph Convolutional Network-Long Short-Term Memory, and Graph Attention Network-Long Short-Term Memory show that the proposed model achieves better forecasting accuracy and stronger high-risk identification capability. The results provide a quantitative decision-support method for green electricity procurement, differentiated subsidy design, and cost-sharing mechanism optimization.

KEYWORDS

Green electricity premium; Cost burden warning; Spatio-temporal graph attention network; Industrial users; Multi-task learning

CITE THIS PAPER

Sinan Jin, Siyuan Zhong. Green Electricity Premium Forecasting and Cost Burden Warning for Chinese Industrial Users Based on a Spatio-Temporal Graph Attention Network. Information Systems and Economics (2026). Vol. 7, No.1, 118-135. DOI: http://dx.doi.org/10.23977/infse.2026.070114.

REFERENCES

[1] Hou M Z. Strategies toward carbon neutrality: comparative analysis of China, USA, and Germany[J]. Carbon Neutral Systems, 2025, 1(1): 3.
[2] Du E. Toward greenhouse gas neutrality: China's post-2030 transition pathway and policy[J]. Environmental Science and Ecotechnology, 2026, 31: 100695.
[3] Zhu L. A Study on the Environmental and Economic Benefits of Flexible Resources in Green Power Trading Markets Based on Cooperative Game Theory: A Case Study of China[J]. Energies, 2025, 18(17): 4490.
[4] Guven D, Kayalica M O. Optimizing agrivoltaic systems for sustainable energy and food production: A case study of East Thrace, Türkiye[J]. Renewable Energy, 2026: 125913.
[5] Georgiadis G P, Dimitriadis C N, Georgiadis M C. Decarbonizing the industry sector: current status and future opportunities of energy-aware production scheduling[J]. Processes, 2025, 13(6): 1941.
[6] Shahzad S, Jasińska E. Renewable revolution: A review of strategic flexibility in future power systems[J]. Sustainability, 2024, 16(13): 5454.
[7] Xie B C. Low-carbon transformation path of power mix in the Yangtze River Delta region[J]. Journal of Environmental Management, 2024, 372: 123316.
[8] Elliott R J R, Sun P, Zhu T. Energy abundance, the geographical distribution of manufacturing, and international trade[J]. Review of World Economics, 2024, 160(4): 1361-1391.
[9] Srichandra I V, Bhadra P. Community detection using graph attention networks clustering algorithm[C]//2024 IEEE 9th International Conference for Convergence in Technology (I2CT). IEEE, 2024: 1-6.
[10] Deng X. A study on coarse aggregate gradation detection based on shape factor estimation using GAT and Bayesian inference[J]. Measurement, 2026: 120980.
[11] Hadizadeh A, Tarokh M J, Ghazani M M. A novel transformer-based dual attention architecture for the prediction of financial time series[J]. Journal of King Saud University Computer and Information Sciences, 2025, 37(5): 72.
[12] Sharma A K, Verma N K. A novel vision transformer with selective residual in multihead self-attention for pattern recognition[J]. Pattern Recognition, 2025: 112497.
[13] FU Z, GAO F, LI Y, et al. How compound climate shocks reshape urban resilience and scenario-adaptive pathways: New insights from the Yangtze River Economic Belt[J]. Environmental Impact Assessment Review, 2026, 121: 108563. DOI: 10.1016/j.eiar.2026.108563.
[14] BAI X, LI Y. DS-ARO: a multi-strategy improved artificial rabbits optimization algorithm for global optimization and corporate bankruptcy prediction[J]. Scientific Reports, 2026. DOI: 10.1038/s41598-026-57561-8.
[15] CHEUNG Y, GUO Z, LI Y. Foliated generative adversarial networks for rolling bearings fault diagnosis[J]. IEEE Transactions on Instrumentation and Measurement, 2026, 75: 1-24. DOI: 10.1109/TIM.2026.3684644

Sponsors, Associates, and Links

---

• Accounting, Auditing and Finance
  
  
• Industrial Engineering and Innovation Management
  
  
• Tourism Management and Technology Economy
  
  
• Journal of Computational and Financial Econometrics
  
  
• Financial Engineering and Risk Management
  
  
• Accounting and Corporate Management
  
  
• Social Security and Administration Management
  
  
• Population, Resources & Environmental Economics
  
  
• Statistics & Quantitative Economics
  
  
• Agricultural & Forestry Economics and Management
  
  
• Social Medicine and Health Management
  
  
• Land Resource Management
  
  
• Information, Library and Archival Science
  
  
• Journal of Human Resource Development
  
  
• Manufacturing and Service Operations Management
  
  
• Operational Research and Cybernetics