Dynamic Connectedness Among Oil, Food Commodity, and Renewable Energy Markets: Novel Perspective from Quantile Dependence and Deep Learning

The spillover effect between multiple markets is crucial in influencing the volatility of energy and food commodity prices. In this study, a time-varying QVAR model is used to analyze the impact of the outbreak of the Russo-Ukrainian war on the risk-associated system of oil, food, and renewable energy, and the DY spillover index model and wavelet coherence are introduced for result comparison. Specifically, data on oil, food commodities, and renewable energy from March 1, 2013, to January 31, 2023, are used, along with various influencing factors such as the climate change index, geopolitical risk, green bond index, and Dow Jones Industrial Average index, to analyze the dynamic correlations presented in the market from multiple perspectives. Furthermore, the ATT-CNN-LSTM model is utilized to construct a risk early warning system for oil-food-renewable energy, providing insights into the future trends of risks. Empirical results indicate that the correlation at the 0.05 and 0.95 quantiles is significantly greater than that at the conditional mean and median, indicating higher systemic risk spillover levels during extreme market conditions. We also find that the outbreak of the Russo-Ukrainian war has promoted the development of renewable energy and enhanced the correlation between food and renewable energy. The results of the extension analysis show that the predictive performance of the ATT-CNN-LSTM model is superior to the other five models. This research will contribute to more effective risk management of the energy-food system.