Analytical indicators on physical risks

Global warming increases the likelihood of extreme weather events. The resulting damage can have a significant impact on the financial system. For example, companies affected by flooding or droughts might find it difficult to service their debts. Or collateral, such as buildings or land, might suddenly lose value. This can affect the stability of our financial system.

Our physical risk indicators take into account risks stemming from climate change-induced natural hazards and consider how these risks affect firms’ ability to pay back loans and bonds and the performance of their equity. The indicators cover coastal flooding, river flooding, wildfires, landslides, subsidence, windstorms, water stress, droughts and rainfall variation.

You will find on this page:

To place the current situation in context and explore potential future developments, our indicators incorporate climate scenarios based on Representative Concentration Pathways (RCPs). These pathways model future greenhouse gas concentration trajectories. RCP4.5 is considered to be a moderate mitigation scenario. It assumes that policies will be implemented to reduce greenhouse gas emissions. RCP8.5 assumes a high-emission scenario in which no significant actions are taken to mitigate climate change.

With the latest data release, the windstorm and water stress indicators have been improved through the integration of more reliable and accurate hazard estimates. For windstorms, a new dataset from Copernicus has been incorporated, providing more up-to-date information and extending coverage from just the winter season to year-round. The water stress indicator relies on the World Resources Institute's Aqueduct dataset, which features advanced hydrological modelling and incorporates updated climate scenarios based on the latest models.

Significant improvements have been made to better measure the potential damage from acute hazards. Previously, assessments focused mainly on companies’ tangible fixed assets, such as buildings and plants. Inventories – such as raw materials, work-in-progress goods and finished goods – have now been added, as these can also incur losses. Additionally, for flooding, building height is now taken into account, acknowledging that damage is typically concentrated in lower storeys. These improvements provide a more accurate representation of assets at risk at company level.

Compiling these indicators is subject to data-related limitations. In particular, there is a lack of suitable data to identify all relevant locations of firms’ physical assets and their vulnerability to natural hazards. Because the data are continually being improved with each release, we are only publishing them as analytical indicators for the time being.

In addition, the compilation framework for the indicators makes use of statistical methodologies that are usually applied to larger samples. Therefore, the framework should not be directly applied to single entities (e.g. a specific firm) or to single neighbourhoods. The developers of climate models have also cautioned against applying them at the local level, as the models are constructed with larger geographical areas in mind.

For further details on the methodology, data sources and limitations, please consult the Statistics Paper and the Technical annex .

Risk scores

Risk scores sort portfolio exposures according to the location of the debtor, with locations being assigned a score from 0 (no risk) to 3 (high risk).

The geographic prevalence of each hazard largely determines financial institutions’ exposure to risk. Acute hazards like wildfires and landslides tend to have a localised impact, resulting in relatively limited affected exposures overall, compared with more widespread hazards such as temperature and precipitation-related events or windstorms (Chart 1). Nevertheless, acute hazards still have the potential to cause significantly greater physical damage despite their restricted spatial extent. Currently, the methodology allows for estimating the monetary impact of floods and windstorms (see the section on indicators based on expected losses).

It is important to note that risk score categories are not directly comparable across hazard types, as they rely on differing methodological assumptions. However, they do provide valuable insights for assessing relative risk levels across countries and climate scenarios, as well as variations within individual hazards. For instance, projections under the high-emission climate scenario RCP8.5 suggest that hazards related to temperature and rainfall would pose a greater risk to financial institutions’ portfolios compared with current conditions (Chart 1, panel a). This effect is especially evident in the standardised precipitation index (SPI), which captures extremes – such as drier conditions in southern Europe and wetter conditions in northern Europe.

For three hazard types – landslides, subsidence and windstorms – only historical data are currently available, meaning they lack forward-looking dimension. Although a significant portion of portfolios are potentially exposed to these hazards, most exposures of euro area financial institutions are deemed low risk. This is particularly true for windstorms, owing to the relatively low intensity of this hazard and the quality of buildings and construction materials used in the euro area (Chart 1, panel b). At the same time, windstorms appear more prevalent in this release compared with previous assessments. This is due to an improved modelling methodology that captures lower-intensity events and, thanks to new data sources, includes additional storm types and more recent events.

Chart 1

Portfolio exposures of euro area financial institutions to different hazards by risk score

a) Hazards with climate projections	b) Hazards with only historical data
(left-hand scale: EUR billions; right-hand scale: percentage of portfolio)	(left-hand scale: EUR billions; right-hand scale: percentage of portfolio)

Potential exposure at risk

The potential exposure at risk (PEAR) indicator provides insight into the prevalence of a natural phenomenon and is compiled as a sum of risk scores, which range from 1 (low risk) to 3 (high risk). The PEAR indicator can serve as a summary measure, enabling comparison across various dimensions of the indicators.

For example, Chart 2 shows the effect of climate adaptation strategies for coastal flooding. Risk scores illustrate how flood defences reduce exposures from medium and high risk to low risk across different time horizons and climate scenarios in the euro area (Chart 2, panel a). By contrast, the PEAR indicator provides an aggregate metric illustrating the extent to which flood defences eliminate risk, highlighting cross-country differences – with the Netherlands providing a notable example (Chart 2, panel b).

While adaptation strategies are crucial for measuring future resilience to hazards, information is currently only available for flood defences. The flood protection dataset is partially based on technical assumptions, so the data may not reflect the actual defence structures in place. This has major implications for risk assessments – and it is prudent to evaluate risk under scenarios with and without flood defences, as these measures may be insufficient or subject to failure.

Chart 2

Potential exposure at risk for coastal flooding with and without flood defences

a) Risk scores	b) Potential exposure at risk (PEAR), RCP8.5 for 2050
(left-hand scale: EUR billions; right-hand scale: percentage of portfolio)	(EUR billions)

Normalised exposure at risk

The normalised exposure at risk (NEAR) indicator measures the losses that financial institutions are expected to incur should their debtors not be able to repay their loans following a natural event that damages their physical assets.

The indicator takes into account the intensity of a hazard using damage functions. For example, for a flood that is 1 m deep, damage amounting to 25% of a company’s physical assets is assumed, while for a flood that is over 2 m deep the share of damages will be higher. Because damage functions are not currently available for all hazards, the normalised exposure at risk indicator is only provided for river flooding, coastal flooding and windstorms. The indicator also captures the probability of a hazard occurring, so it is possible to estimate expected losses. The indicator shows losses both on an annual basis and over the remaining maturity of an instrument to highlight potential differences in the maturity structure of banks’ portfolios.

Collateral-adjusted exposure at risk

Like the normalised exposure at risk indicator, the collateral-adjusted exposure at risk (CEAR) indicator provides an estimate of expected losses within a bank’s portfolio. The difference is that this indicator also takes into account the mitigating effect of collateral pledged with a loan commitment. In terms of minimising financial losses, collateral serves as a robust mitigating factor for euro area creditors. However, it is important to note that physical collateral could be damaged following a natural disaster, and its value could fall. This is also accounted for in the calculation of the CEAR.

Chart 3 shows the expected loss indicators – NEAR and CEAR – for the loan portfolios of euro area banks. These harmonised indicators are a valuable tool for comparing the economic impact of climate change across various hazards, climate scenarios and countries.

Flood defences and collateral serve different purposes: flood defences prevent damage from occurring and thus deliver real‑economy benefits, whereas collateral primarily limits banks’ credit losses without addressing the possible underlying physical damage. Both measures play an important role in reducing overall losses. At the same time, it is worth noting that national practices vary considerably, which affects country-specific risk profiles (Chart 3, panel b).

For river flooding scenarios with flood defences, the two expected loss indicators – NEAR and CEAR – show a steady increase in risk across climate scenarios relative to the historical baseline (Chart 3, panel a). This underscores the importance of investing in flood defences to mitigate the potential degradation of such systems, particularly when combined with the intensification of physical climate risks.

Chart 3

Adaptation measures for expected loss indicators: annualised normalised and collateral-adjusted exposure at risk of euro area banks’ loan portfolios

a) River flooding with flood defences, change from historical baseline	b) River flooding, RCP8.5 for 2100

Data access

The underlying aggregated data for the analytical physical risk indicators are available as compressed csv files below.

To receive updates on methodology and data quality, subscribe to our climate data information service.