Asset-level assessment of climate physical risk matters for adaptation finance

How much physical risk do investors miss when they map firms by their headquarters?

Giacomo Bressan, Anja Đuranović, Irene Monasterolo and Stefano Battiston measure how the location of productive assets shapes climate losses for investors in "Asset-level assessment of climate physical risk matters for adaptation finance".

They link 1,820 physical assets in Mexico, mainly mines, power plants and other energy facilities, to the 177 listed firms that own them and to 1,014 European investors.

They simulate hurricane damage, add chronic sector impacts, and translate both into equity values with a climate-adjusted dividend discount model. Their main conclusions include:

• Two renewable power producers with the same chronic shock of about 3% face acute hurricane shocks of almost 0% versus ~55% because their plants sit in different regions.
• Replacing plant locations with a single headquarters address shrinks estimated value at risk by a factor of 3 to 5.5 and understates combined losses by up to 70%.
• At the portfolio level, value-at-risk from tail hurricane losses is 5 to 38 times the average loss.
• For 35% of firms, equity losses under a 1-in-250-year hurricane are at least three times larger than expected annual losses, which holds across alternative discount rate and growth assumptions.
• Relying on chronic risk alone understates investor losses by up to 82%, and relying on average hurricane impacts misses close to 98% of tail losses.
• Because extreme events are expected to increase globally, even a well-diversified equity portfolio keeps substantial physical risk.

This article stresses investors should ask for plant-level location, capacity and ownership data before relying on vendor physical risk scores, which are often aggregated by firm.

It shows how investors may otherwise produce significantly misplaced risk estimates and steer scarce public and private money toward the wrong assets and policies.

The analysis covers one country, one hazard and one asset class, and it ignores firms' own adaptation measures, such as relocation or physical barriers, which are not observable from the data.