Climate risk and renewable energy market volatility: Machine learning approach

Global climate change is a major environmental challenge, and the new-energy market is increasingly attracting investors’ attention as a key area for investment, especially because of the impact of climate change on price volatility. In this study, we investigated the impact of climate change on China’s new-energy market by introducing the following two indicators: China’s climate policy uncertainty (CEU) indices and the climate uncertainty (CU) indices. We employed Diebold–Mariano and model confidence set tests to assess the out-of-sample prediction accuracy of our model. Empirical results showed that incorporating climate risk indices significantly improved the predictive accuracy of the three deep-learning models, with the CU index performing best in a variational modal decomposition (VMD)-long short-term memory (LSTM) model. In particular, in a VMD-LSTM model with the CU index indicator, the mean absolute error values for 1-, 3-, and 5-step ahead predictions were reduced by 8.1 %, 17.3 %, and 18.4 %, and the mean squared error values were reduced by 20.3 %, 40.0 %, and 32.4 %, respectively. Finally, the empirical findings remained robust, even when considering different estimation windows (historical training periods), forecast horizons (short-term, medium-term and long-term), and the impact of the COVID-19 pandemic.