A0470
Title: Essential economic features for European Union carbon allowance future prediction
Authors: Shuen-Lin Jeng - National Cheng Kung Univeraity (Taiwan) [presenting]
Abstract: The selection and construction of many economic features are investigated to assess their strength of predictability on the European Union carbon allowance (EUA) future price. The economic features were constructed from a group of carefully selected energy and macroeconomic indicators, including clean energy, crude oil futures, natural gas futures, coal futures, power futures, STOXX 600, and Europe's leading country stock index, with their technical extensions. The essential features are established by discovering those with the most significant influence on EUA price by comparing the results from several machine learning and statistical models, such as LASSO, AdaBoost, XGBoost, LSTM, MARS, and random forests. An empirical analysis was carried out on the selected features between January 1, 2023, and March 31, 2024. The identified essential economic features will produce significant benefits not only to the carbon-trading corporations or investors in building portfolios and trading strategies but also to the government authorities in formulating effective regulations that could achieve carbon reduction within their targeted time frame.
