Physical risk: When storms hit markets

Most acute physical risk tools overlay company facilities with hazard maps to estimate direct damage. That works for insurers pricing a site. It falls short for equity investors, because most listed companies don’t disclose where their assets are, and storm losses spread through supply chains, demand and competition. When Hurricane Harvey hit Houston in 2017, Ford and General Motors idled plants in other states because they couldn’t source plastics and chemicals.

Storm risk measured from market responses

This research tracks how companies in 37 storm-sensitive US industries performed in the 26 weeks after 45 hurricanes, floods and wildfires, each causing at least $5 billion in damage between 2000 and 2024. Each company’s return is measured against the S&P 500 over the same window, giving 11,725 company-event observations. Hierarchical Bayesian models then link the size of each event to relative performance. The financial crisis and pandemic years are excluded.

Six industries show consistent storm sensitivity

Most industries performed worse after major storms. Six show a statistically significant negative relationship with damage size: homebuilding, electric utilities, oil and gas pipelines, gas distributors, specialty insurance, and property and casualty insurance. Homebuilding is the most sensitive. At sector level, agriculture, real estate development, utilities and finance also show significant effects. The industry models explain 12% to 24% of excess return variation, which is high for a single factor in an event study.

Electric utilities carry costs without offsetting revenue

For electric utilities, the probability of a negative relationship is 99.4%, and storm damage explains 24% of excess return variation over 26 weeks. Duke Energy follows the same pattern, underperforming the S&P 500 by 10% to 30% after the largest hurricanes. Restoration costs arrive quickly while rate recovery lags, so storms bring costs with little revenue to offset them.

Coverage for every holding

Because the sensitivities come from stock prices, the method doesn’t depend on asset disclosures. Each company gets an industry model where one is robust, a sector model where it isn’t, and a pooled model otherwise. That gives portfolio-wide coverage, and the sensitivities can be linked to forward-looking climate scenarios. Location-based methods remain the right tool for site-level damage.

Download the full white paper for the industry and sector results and the electric utilities case study.