Title: Climate Risks and Stock Returns
Authors: Eugenio Carnemolla - University of Lausanne and Swiss Finance Institute (Switzerland) [presenting]
Giuseppe Vinci - Rice University (United States)
Abstract: We construct a novel measure of exposure to climate risks applying textual analysis to the firms' annual reports. The relation between weather sensitivity and word counts of the filings is estimated using a supervised machine learning algorithm. We validate our text-based measure of climate risk by using information on the firm's geographical footprint. We find that stocks of climate-sensitive firms significantly underperform stocks of climate-resilient firms, suggesting that investors underreact to climate change risks associated with natural disasters. Our results are stronger following months with attention-grabbing weather events and for geographically concentrated firms. A trading strategy exploiting weather sensitivity earns positive abnormal returns.