A two-stage interval-valued carbon price forecasting model based on bivariate empirical mode decomposition and error...

Area of Science:

• Environmental Economics
• Climate Change Policy
• Data Science

Background:

• Global economic development increases greenhouse gas emissions, driving climate change.
• Accurate carbon price forecasting is essential for effective carbon pricing and market stability.
• Existing forecasting models may not fully capture the complexities of carbon price dynamics.

Purpose of the Study:

• To propose and validate a novel two-stage interval-valued carbon price combination forecasting model.
• To enhance the accuracy and stability of carbon price predictions.
• To provide a robust tool for policymakers and investors in carbon markets.

Main Methods:

• Bivariate Empirical Mode Decomposition (BEMD) for decomposing carbon prices and influencing factors into interval sub-modes.
• Combination forecasting using artificial intelligence (AI) neural networks (IMLP, LSTM, GRU, CNN) in Stage I.
• Error correction using Long Short-Term Memory (LSTM) network in Stage II to refine predictions.

Main Results:

• The proposed two-stage model significantly outperforms single forecasting methods for interval-valued carbon prices.
• Stage I combination forecasting of interval sub-modes showed superior performance.
• Stage II error correction further improved forecasting accuracy and stability.

Conclusions:

• The developed model is an effective approach for interval-valued carbon price forecasting.
• The model aids policymakers in formulating emission reduction strategies.
• It assists investors in mitigating risks within carbon markets.