Physical risk: Built for insurers, sold to investors

Most physical climate risk tools used by investors trace back to catastrophe models built for property insurers. They estimate the cost of repairing a building after a storm, using a formula borrowed from underwriting: hazard intensity multiplied by vulnerability and exposure value. That works for pricing a policy. Portfolio managers need to know how climate events affect enterprise value.

Climate models are weakest where losses are largest

Global climate models run on grids of 50 to 200 kilometers, while the processes behind cyclones, heavy rain and floods play out over a few kilometers or less. Projections are most reliable for temperature-driven hazards such as heat and drought. They are least reliable for wind and rainfall, which drive the biggest financial losses. Downscaling can’t add information the global model doesn’t contain.

Direct damage is often the smallest part of the loss

Business interruption and supply chain disruption often outweigh property damage. Hurricane Harvey disrupted more than a third of US chemical production in 2017, and the 2011 Thailand floods cascaded through global supply chains. Incumbent methods typically leave supply chains out, and they add up facility-level losses without capturing network effects between sites and suppliers.

One store out of 2,800

When the Palisades fire destroyed a Kroger supermarket in January 2025, a facility-level model would have flagged a significant loss. The store was 0.036% of Kroger’s footprint. After an early dip alongside the S&P 500, Kroger’s stock rose about 10% over the following 12 weeks while the index fell 6%. For a company of that scale, one destroyed site barely registers. For a grocer with one to three stores, the same loss can be existential.

Market signals add an enterprise-level check

Research on US hurricanes from 1996 to 2019 found option-implied volatility rose 5–10% for firms with sites in landfall regions, with elevated uncertainty lasting up to three months. Stock and option prices reflect supply chains, competitive position and adaptive capacity that facility models can’t see. The paper argues for keeping hazard models, reported with honest uncertainty bounds, and testing their outputs against market responses and peer performance.

Download the full white paper for the Kroger case study and four steps investors can take now.

‍