Container or Content: from Flood Hazards on Firms’ Physical Assets to Credit Risks

Potential floods, buildings and establishments

Non-Technical Summary

When floods strike, who bears the heaviest losses: the building’s owner, who sees the property damaged, or its occupier, whose equipment and machinery are destroyed? And from the viewpoint of banks, which risk matters most? Put differently: should we worry more about the container (the building) or the content (what’s inside)? If the Seine overflows, what will be the spillovers for the banking system? While the frequency and the severity of floods are expected to increase with climate change, will they cascade into to credit risks?  

These questions are crucial for monitoring the resilience of the financial system to climate risks and require central banks to combine and analyse very granular data. To that end, our research develops a Digital Twin of French firms. This virtual model closely represents the spatial distribution of French firms’ assets and simulates their vulnerability to potential floods. We simulate potential losses under different flood severity scenarios, that we reconcile with historical records.

The paper shows that distinguishing between the building itself and the equipment within is essential for assessing the impact of floods on firms’ balance sheets and the banking system. Specifically, we demonstrate that a significant portion of climate-related physical risk resides in geographically dispersed production premises, containing the equipment and machinery and which may be far more vulnerable. While real estate can be repaired or rebuilt, production assets—machines, tools, computers—are often more fragile and more costly when destroyed. These “content” losses are typically larger than those linked to property damage, and because ownership of the building and its equipment often differs, the risks propagate along two separate channels. We find that banks face higher credit risk from the occupiers’ channel (businesses losing equipment and defaulting) than from the owners’ channel (landlords facing property devaluation).
At present, France’s insurance system contains much of this risk for banks. But our projections show that as climate change intensifies floods, losses will rise sharply, straining the insurance scheme and creating new vulnerabilities for the financial system.

Beyond diagnosis, this work provides a tool for action. The Digital Twin tool can underpin climate-adjusted stress tests, helping supervisors spot fragile loan portfolios and design more targeted safeguards. In particular, it could guide, if needed, the calibration of bank capital requirements, whether at the level of individual institutions or the system as a whole. Finally, by simulating future floods under climate change, the model quantifies the costs of inaction — “no transition” and “no adaptation” — that could be compared to the costs of investing now in transition or resilience. This allows policymakers to weigh trade-off, with a fine identification of winners and losers at the sector/firm level.

Keywords: Floods, Financial Stability, Non-financial Corporations, Credit Risk, Climate Change

Codes JEL : G21, G32, Q54