Climate Risks and the Risk Premia of AI Stocks: Some Forecasting Experiments

ABSTRACT This study examines whether physical and transition climate‐risk measures predict the risk premia of AI‐related stocks, and whether incorporating these risks improves forecast performance relative to benchmark models. The findings suggest that global transition climate shocks lower the premia of AI stocks, while US climate policy increases them; AI stocks do not respond to physical climate risks. Conversely, physical climate risks lower the risk premia of conventional stocks; conventional stocks do not respond to transition climate risks. This suggests that each stock category responds uniquely to diverse climate risks. Essentially, AI and conventional stocks provide pricing stability and forestall drastic portfolio reassessment in the face of global climate transitional risks and physical risks, respectively. Since both stock options react differently to a diverse set of climate risks, they may offer valuable diversification benefits. Finally, we show that incorporating climate risks into the predictive model of AI and conventional stock premia yields higher utility gains than the benchmarks. The evidence suggests that climate‐risk measures contain useful information for forecasting AI and conventional stock premia, with implications for investors, portfolio managers, and the pricing of climate‐related risks.