Measuring the extreme linkages and time-frequency co-movements among artificial intelligence and clean energy indices

Abstract

This is the first study analyzing the volatility connectedness and time-frequency interdependence between AI index and clean energy index. Specifically, we use the QVAR frequency connectedness, Wavelet Local Multiple Correlations (WLMC) and Granger causality quantile methods to check the risk spillovers and multivariate time and frequency relationships among the eight clean energy indexes and the AI index. This is over the period from December 18, 2017 to April 4, 2023. Our results show: (1) NASDAQ OMX Geothermal Index is the strongest net sender of short- and long-term shocks in the system during extreme upside market conditions. In downturn conditions, the S&P Global Clean Energy Index is the largest net shock sender. The AI Index exports shocks at all frequencies. In addition, market connectedness among markets is stronger under extreme market conditions. (2) We find that the AI Index predominantly exhibited positive co-movements with clean energy indices, primarily concentrated within the long-term frequency domain. However, they displayed robust cooperative dynamics across all frequency domains within the context of multivariate wavelet interconnections. (3) The quantile granger causality analysis revealed that below the extreme bullish threshold (0.95), the NASDAQ CTA Artificial Intelligence & Robotics index could predict changes in the risk associated with all clean energy indices. However, under extremely bullish quantile conditions, the NASDAQ CTA Artificial Intelligence & Robotics index statistically exhibited Granger causality only with respect to the NASDAQ OMX Renewable Energy Index, NASDAQ OMX Geothermal Index, and WilderHill Clean Energy Index.