Predicting European carbon emission price movements

Abstract

The European carbon emission trading market is critical in achieving planned carbon emission reduction for global sustainable growth. This paper investigates various statistical methods in forecasting the European carbon emission (CO2 hereafter) price movements. The paper builds a predictive regression model of CO2 price movements with past returns of various commodities and financial products. In the paper, 22 functional forms of five different classifiers are employed and CO2 price movements are forecast. Results indicate that the past returns of Brent crude futures, natural gas (NG), Financial Times Stock Exchange 100 (FTSE100), Deutscher Aktienindex (German stock index) 30 (DAX30), Cotation Assistée en Continu (French stock index) 40 (CAC40) and Standard & Poor's 500 (S&P500) are statistically significant in forecasting the current CO2 price movements. The authors also found that the bagged decision tree of the ensemble classifier best forecasts the CO2 price movements. The result should be relevant to firms that wish to trade European carbon emissions.

Keywords

• european carbon emission
• autoregressive
• data mining
• return prediction