Influencing factors and forecasting of European union allowance price: A comparative study using machine learning models

The European Union Allowance (EUA) is acknowledged as a significant mechanism to motivate EU enterprises to decrease emissions and fulfill the United Nations Sustainable Development Goals (SDGs). Investigating the influencing factors and forecasting trajectory of the EUA prices is hindered by the nonlinearity and non-stationarity characteristics of the EUA prices. This study dissects the intricate relationships of energy prices, financial market variables, and climate factors influencing the EUA prices by the use of Vector Autoregression (VAR), Granger causality tests, Impulse response function(IRF), and Newey-West OLS model. The forecasting results of EUA prices are compared using machine learning models involving BP Neural Networks (BPNN), Random Forests (RF), and Support Vector Machines (SVM) in this study. The results show that, first, energy prices, financial market variables, and climate are influencing factors for EUA prices. However, the interplay among these factors is intricate. Second, Machine learning models that incorporate BPNN, RF, and SVM offer substantial advantages in forecasting of EUA prices. Third, it has been determined that the RF model exhibits superior accuracy and stability in forecasting the EUA prices when compared to the performance of the three machine learning models. These findings underscore the capability of machine learning models to handle intricate and dynamic carbon market data, offering critical insights for lowering carbon emissions and promoting sustainable development. • Climate and non-climatic factors significantly impact European Union Allowance price. • Unexpected temperature variables strongly influence EUA price fluctuations. • Machine learning models outperform traditional methods in EUA price forecasting. • Random Forest model demonstrates superior accuracy and stability in predictions. • Study offers insights for enhancing carbon market forecasting and policy development.