Does the incorporation of climate risk enhance the predictability of stock market returns?

Purpose This paper investigates whether climate risk contains valuable information for predicting future stock market returns and examines its implications for asset pricing. Design/methodology/approach We develop a novel measure of climate risk based on the autoencoder (AE) by integrating diverse proxies for both physical and transition risks. Findings The climate risk (1) possesses strong in-sample and out-of-sample predictive ability over the 1- to 12-month horizon; (2) provides the sizable asset allocation benefits for mean-variance investors at all levels of risk aversion, even after accounting for the transaction costs; (3) achieves higher cumulative returns than the competitors, such as historical average, market benchmark and alternative climate risk, indicating its superiority in guiding investment decisions; and (4) derives its forecasting power primarily from the cash flow channel rather than the discount rate channel, shedding light on the mechanism through which climate risk affects future stock market returns. Originality/value Firstly, we develop a comprehensive measure of climate risk based on the AE, which systematically integrates eight proxies that capture both the physical and transition dimensions of climate risk. Secondly, we present the robust empirical evidence that the climate risk contains the strong predictive content for future stock market returns and delivers sizable economic benefits for mean-variance investors, thereby underscoring its practical implications in asset allocation. Finally, we shed light on the potential economic mechanisms underlying the forecasting power of climate risk by confirming that the climate risk primarily influences the stock market returns through the cash flow channel.