Physical climate risk assessment in practice: Lessons from the financial sector

This article was written by Edo Schets, Head of Climate for Sustainable Finance Solutions and Zane Van Dusen, Global Head of Risk & Investment Analytics Products at Bloomberg.

Financial firms across the globe increasingly seek to integrate physical risk into financial analysis, as extreme weather events continue to impact global markets. Recent examples from Bloomberg Intelligence highlight this growing challenge across sectors:

• A severe drought in Clyde, Texas forced the city to default on its municipal water system debt, requiring bond insurers to step in. While insurance protected investors this time, climate-related risks may increase borrowing costs for municipalities. (BI Research “Climate-Related Risk for Governments” from September 6, 2024).
• Volkswagen’s production dropped by 150,000 units in late 2023 after flooding disrupted their Slovenian supply chain, reducing quarterly margins by 0.5%. (BI Research “Global Autos ESG Outlook” from July 18, 2024).
• Property & casualty insurance brokers have seen their returns surge 201% over five years, outpacing the broader insurance index’s 45% gain. This mirrors a 90% increase in catastrophe insurance premiums, as measured by Guy Carpenter’s US Property Rate Index. (BI Research “Climate Costs From Storms Power P&Cs” from February, 2024).

At a client roundtable last year, hosted by Bloomberg together with climate analytics provider Riskthinking.AI, practitioners revealed their key challenges when assessing physical risk – from gathering granular asset-level data to translating climate impacts into financial metrics, while emphasizing the need for transparent and defendable modeling approaches.

Progress and challenges: How firms are implementing physical risk assessment

Prior to the roundtable, we surveyed 150 financial practitioners globally. The results revealed a clear pattern: while most institutions have begun incorporating physical risk analysis, few have achieved full integration into their risk frameworks. This pattern holds consistent across sectors, suggesting a broader industry-wide challenge in operationalizing climate risk assessment.

Among those who have fully integrated physical risk assessment and climate-related impacts, two key challenges emerged: obtaining granular asset-level data and converting climate insights into financial metrics.

These elements ultimately help answer critical questions about asset resilience: When faced with an extreme weather event, can a facility withstand the impact? If not, what are the potential remediation costs for the business or municipality? A recent Bloomberg News article, “America’s Factories Weren’t Built to Endure This Many Hurricanes,” illustrates this challenge through the experience of Lennox International Inc.—a 2018 tornado strike on their air conditioning and furnace plant provided a real-world case study of physical risk assessment and climate-related financial impact. 

Our roundtable discussions revealed that firms have begun to tackle the asset-level data challenge head-on — committing significant resources to manually building detailed location databases with precise latitude and longitude coordinates for thousands of physical assets. However, another persistent challenge remains: gaining broad acceptance of their physical risk modeling approaches among stakeholders. While it is paramount that methodologies are rigorously science-based, it is equally important that they are easy to explain and produce clear and intuitive outputs.

Validating the accuracy of physical risk methodologies is likewise a challenge, as there is limited historical data for backtesting. Roundtable attendees noted, however, that both internal stakeholders and global regulators appreciate this constraint as well as the uncertainty associated with forecasting weather events 5-10 years in the future (a time horizon aligned with firms’ investment strategies).

As an alternative to formal backtesting, users of physical risk assessments often focus on recent, newsworthy case studies that can demonstrate a model’s ability to capture tail events. For example, Riskthinking.AI produced a case study about how its analytics captured the 2022 Texas Ice Storms, in addition to global backtesting studies (summary provided below).  

Even firms with sophisticated physical risk assessment capabilities highlighted several remaining challenges. For example, beyond direct asset exposure, understanding supply chain vulnerabilities is crucial for comprehensive risk assessment. Notably, as organizations’ capabilities mature in assessing physical risk, a shift occurs: the primary challenge evolves from modeling physical hazards to addressing more nuanced data needs. Firms find themselves focusing not only on expanding their asset coverage but also deepening their understanding of each asset’s operational significance and resilience characteristics.

Corporate adaptability adds another layer of complexity to the analysis. A notable example comes from Schneider Electric SE, which chose to build its new switchgear and power distribution facility in Tennessee rather than expanding its Texas operations, in order to reduce its exposure to tornado risks.

The landscape of physical risk assessment presents a clear paradox: while firms have made progress in data collection and basic modeling, the harder challenges involve creating comprehensive, intuitive frameworks that capture climate impacts’ full complexity. These impacts range from asset vulnerability to supply chain resilience and adaptive capacity.

Using physical climate risk data to address industry challenges

Responding to these industry challenges, Bloomberg launched its first physical climate risk solution in Q3 2023 in partnership with Riskthinking.AI. The solution utilizes Bloomberg’s extensive asset-level database, tracking both operational and announced facilities—enabling analysts to assess not just current exposure but also companies’ strategic adaptations through their facility location decisions. The combination of Bloomberg data and Riskthinking.AI climate modelling results in a transparent and intuitive physical risk assessment approach for approximately 48,000 companies globally across 4 different climate scenarios with time horizons going to 2050.

The solution aims to address three key needs: 

1. Transparency, reliability and ease of use
2. Decision-useful metrics and time horizons
3. Holistic analysis that includes supply chain, operational significance and asset resilience

1. Transparency, reliability and ease of use

Riskthinking.AI’s approach to modelling future physical hazards prioritizes transparency and stakeholder acceptance through tangible validation. The stochastic modeling methodology has provided compelling case studies for clients. This approach leans into the uncertainty of future physical risks by utilizing all possible climate pathways. For example, using projections generated before 2023, the model successfully captured 97% of record-breaking temperature events that occurred in 2023. This type of real-world validation helps build confidence among stakeholders who prefer intuitive, demonstrable results.

2. Decision-useful metrics and time horizons

Investors are looking for physical risk data that is expressed in financial terms and can be integrated in fundamental and risk models, across near-term horizons that are relevant to investment decisions. To meet this need, Bloomberg and Riskthinking.AI have launched an updated physical risk methodology in the first quarter of 2025, that quantifies impact in terms of asset replacement value, with detailed breakdowns by specific chronic and acute risk factors such as cyclones, wildfires, extreme heat, and sea level rise. The results are available for time horizons stretching from the near term (2025) up to mid-century (2050).

3. Holistic analysis

Bloomberg is also working on further enhancements to support an increasingly sophisticated and holistic physical risk analysis. This includes supply chain vulnerability metrics that utilize Bloomberg’s extensive dataset of relationships across more than 28,000 public and 93,000 private companies, thereby automating the analysis that investors can do using SPLC tool on the Bloomberg Terminal or by using our supply chain data feed.

Additionally, Bloomberg will evolve its assessment of the operational significance of individual assets, such as recognizing that manufacturing sites are significantly more important than administrative offices for an industrial company. This enhancement will enhance our ability to generate relevant red flags on a company-level.

Furthermore, Bloomberg and Riskthinking.AI are researching approaches that account for the climate resilience of complex assets, such as underground pipelines and above ground telecommunication lines, which will boost asset type coverage. Using bespoke asset damage estimations that account for the specifics of such assets allows us to include these less common but important asset types in the analysis.

Bloomberg’s physical climate risk data is available on the Terminal and via Data License. More information is available Bloomberg Sustainable Finance Solutions. Bloomberg Terminal subscribers can also access it via ESGD <GO> and data.bloomberg.com.

Interested in learning more? Sign up to attend a Bloomberg webinar on physical risk.