Scholarlink

<p>The carbon emissions allowance trading market regulates greenhouse gas emissions through economic means. This study proposes a hybrid model that combines Variational Mode Decomposition (VMD) with machine learning to forecast volatility in the European Union (EU) carbon emissions allowance trading market. The model addresses the challenges posed by the non-stationary and nonlinear characteristics of the series, effectively decomposes the time series, selects the optimal forecasting model for each intrinsic mode function (IMF), and reconstructs the results. The combined model is applied to forecast the volatility of EU carbon emissions allowance futures. The empirical analysis shows that the hybrid model outperforms Gate Recurrent Unit (GRU) in one-step forecasting, with improvements of 8.32% (MSE), 6.15% (MAE), 0.99% (SMAPE), and 24.69% (RMSPE). The model also excels in multi-step forecasting, demonstrating both stability and accuracy.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Volatility forecasting for the European union carbon emissions allowance trading market based on the VMD method

  • 期刊论文

  • 发表日期：

    2025年8月20日

查看全文
  • Yuping Song,
  • Xiaolong Tang,
  • Yankun Sun

摘要

The carbon emissions allowance trading market regulates greenhouse gas emissions through economic means. This study proposes a hybrid model that combines Variational Mode Decomposition (VMD) with machine learning to forecast volatility in the European Union (EU) carbon emissions allowance trading market. The model addresses the challenges posed by the non-stationary and nonlinear characteristics of the series, effectively decomposes the time series, selects the optimal forecasting model for each intrinsic mode function (IMF), and reconstructs the results. The combined model is applied to forecast the volatility of EU carbon emissions allowance futures. The empirical analysis shows that the hybrid model outperforms Gate Recurrent Unit (GRU) in one-step forecasting, with improvements of 8.32% (MSE), 6.15% (MAE), 0.99% (SMAPE), and 24.69% (RMSPE). The model also excels in multi-step forecasting, demonstrating both stability and accuracy.