Physical Climate Risk and Bank Lending

Physical climate risk is highly local

One of the most important lessons from recent banking research is that physical climate risk is granular. A flood does not affect every borrower in a region equally. Two firms separated by a few hundred metres can face very different outcomes because one lies inside the inundation boundary and the other does not.

The 2026 ECB working paper included in the source compilation uses this fact as the basis for a spatial regression discontinuity design. It combines AnaCredit loan-level information with high-resolution Copernicus flood maps and studies four major European flood events between 2021 and 2024. This is valuable for risk managers because it shows why country-level or sector-level climate scores can miss the actual credit transmission channel.

Schematic of a credit risk parameter chain linking exposure, default probability and loss given default, illustrating "Physical climate risk is highly local" in Physical Climate Risk and Bank Lending: What Flood Events Reveal About Credit Risk — Figure 1. Schematic of a credit risk parameter chain linking exposure, default probability and loss given default, in the context of physical climate risk is highly local.

The immediate effect can be more lending, not less

A simple intuition might suggest that banks immediately reduce credit after a disaster. The ECB evidence is more nuanced. In the quarter of the flood, lending to affected firms increased by approximately 3.5 to 5 percent, driven mainly by short-term liquidity demand. Firms needed working capital, drew existing facilities and sought funding to manage disruption. Lending then contracted by a similar magnitude in the following quarter.

For climate stress testing, this means physical shocks can create a temporary increase in utilisation before the longer-term deterioration in investment and credit quality becomes visible. Liquidity and credit models should therefore be linked. An acute event can raise both funding needs and default risk at the same time.

Credit quality deteriorates even when aggregate supply remains stable

The same ECB study finds a persistent increase in default rates on pre-existing loans to affected firms. The estimated cumulative increase was around 0.7 percentage points over two quarters, close to a doubling relative to the pre-event annual baseline reported in the paper.

This distinction is important. Stable aggregate lending does not imply that physical risk is immaterial. A bank may continue lending for relationship, recovery or liquidity-support reasons while underlying credit quality deteriorates. Monitoring should therefore separate lending volume, pricing, collateral, maturity and default performance rather than using one variable as a proxy for the whole credit response.

Overlaid physical hazard intensity layers for flood, wind and heat exposure, illustrating "Credit quality deteriorates even when aggregate supply remains stable" in Physical Climate Risk and Bank Lending: What Flood Events Reveal About Credit Risk — Figure 2. Overlaid physical hazard intensity layers for flood, wind and heat exposure, in the context of credit quality deteriorates even when aggregate supply remains stable.

Relationship banking changes the transmission mechanism

The paper also finds that main relationship banks provide more credit to affected firms while tightening collateral requirements. This is consistent with risk sharing rather than unconditional support. Existing relationships can therefore reduce short-term liquidity stress but do not eliminate the need for stronger risk controls.

For banks, this has two implications. First, relationship intensity is a relevant model feature when assessing post-disaster credit supply. Second, collateral management should be part of climate risk analysis. If a bank extends additional credit while collateral values or insurability are deteriorating, the combined PD and LGD effect can be more significant than a lending-volume analysis suggests.

Geospatial data should sit inside credit risk infrastructure

The ECB methodology demonstrates the value of combining loan-level credit data with satellite-based hazard information. A practical bank architecture could connect borrower addresses, collateral coordinates, flood or wildfire hazard layers, property information, insurance status, facility utilisation and internal rating history.

This does not mean that every bank needs to build a catastrophe model. It means that the bank should be able to map the external hazard signal to the internal exposure at a level of granularity that is meaningful for credit decisions. The correct unit may be a property, production site, warehouse, branch, supplier or transport route rather than the legal entity's registered office.

Capital ratio trajectory under baseline and adverse stress paths, illustrating "Geospatial data should sit inside credit risk infrastructure" in Physical Climate Risk and Bank Lending: What Flood Events Reveal About Credit Risk — Figure 3. Capital ratio trajectory under baseline and adverse stress paths, in the context of geospatial data should sit inside credit risk infrastructure.

From event study to risk appetite

Physical risk management should move from event reporting to portfolio decisions. The relevant questions are: where are exposures concentrated, which borrowers have weak adaptation capacity, which collateral types are hard to insure, which sectors depend on vulnerable infrastructure, and how much additional credit or liquidity may be required after an event?

A bank can then define concentration limits, collateral haircuts, enhanced underwriting rules, insurance requirements and scenario-based capital overlays. The objective is not to stop lending in exposed regions. It is to price, structure and monitor the risk using evidence that reflects the local nature of physical climate events.

Conclusion

The key requirement is decision usefulness. Climate analysis adds value when it improves risk identification, pricing, capital allocation, portfolio management or governance, while making uncertainty and model limitations explicit.

Sources

European Central Bank. Physical climate risk, credit risk and lending activity. Working Paper Series No. 3224, 2026. https://www.ecb.europa.eu

Copernicus Emergency Management Service. https://emergency.copernicus.eu

European Central Bank AnaCredit. https://www.ecb.europa.eu/stats/money_credit_banking/anacredit/html/index.en.html

Disclaimer: This article is professional risk-management analysis and is not investment, legal or regulatory advice.

Physical Climate Risk and Bank Lending: What Flood Events Reveal About Credit Risk