91³ÉÈ˰涶Òô

Background

In recent years, corporates around the world have ramped up on disclosure of physical climate risks. However, it has been observed that the quality of these disclosures remains low, with significant challenges due to data quality, lack of standardization, and the complexities of translating climate models into actionable insights.

These challenges are compounded by the inherent uncertainties in climate models, especially when downscaling to local levels, which can lead to significant variability and uncertainties in risk assessment.

With regulators demanding an uplift in disclosure standards, the failure to adequately assess and disclose these risks may also lead to reputational damage, liability risk, and financial penalties. However, there are clear and present risks involved in the disclosure process; corporates that make misleading or incomplete disclosures may face lawsuits from investors, regulators, and other stakeholder who rely on this information for decision-making. This article addresses two of the main challenges associated with physical risk disclosure; the erstwhile lack of an international standard, as well as intrinsic complexity in performing robust assessments of physical risks, and offer some suggestions for improvement.

Evolution of the climate risk reporting landscape

The regulatory landscape for climate-related disclosure has evolved significantly over time, driven by the need for greater transparency around both the impact of corporates to climate and vice versa. While the Global Reporting Initiative (GRI) released the first set of global guidelines for sustainability performance in the early 2000s, specific to climate - the Task Force on Climate-related Financial Disclosure (TCFD) published a framework for companies and financial institutions to disclose climate-related financial risks in 2017. Since then, several frameworks have built on the TCFD recommendations and extended them such as the Sustainability Accounting Standards Board (SASB), the European Union’s Corporate Sustainability Reporting Directive (CSRD), the Securities and Exchange Comission (SEC). While its beginnings were mostly voluntarily based, several progressive jurisdictions in Europe, Japan, Singapore, New Zealand, and most recently Australia (through the recently passed Parliament bill) has made disclosure mandatory.

This trend towards mandatory disclosures is likely to continue.

These regulations aim to enhance transparency and accountability, enabling regulators, investors and other stakeholders to better understand and manage the impacts of climate change. The initial focus on transparency has now shifted towards compliance, with companies and financial institutions being held accountable for the accuracy and completeness of their disclosures. While the development of these disclosure frameworks is undoubtedly a positive step forward, there are notable challenges such as the lack of alignment between different disclosure frameworks, and the recommended focus on climate hazards, time horizons, and climate scenarios.

Lack of an international standard (for now)

One of the primary challenges in physical climate-related disclosures lies in the standardization of these frameworks. While frameworks like TCFD and the EU taxonomy criteria provide guidelines, their application is not yet consistent across industries or regions.

The lack of a universally accepted standard means that disclosures can vary widely in their scope, detail, and methodology, making it difficult for firms with a global footprint to disclose according to each regulator.

In response, there are ongoing efforts through the ISSB (International Sustainability Standards Board) to harmonize global standards by providing comprehensive global baseline of disclosure standards that can be adapted (for example, through the Australian Sustainability Standards Board (ASSB)) for specific jurisdictions. However, these efforts will take time and for now, disclosures remain inconsistent in the use of climate scenarios, timeframes, and hazards. This disparity is especially problematic for firms with a global footprint, prompting the funneling of resources into tailoring disclosures to a certain regulatory regime rather actually improving climate risk understanding or executing on commitments. The absence of standardization also leads an abundance of low-quality disclosures, as different companies may use different assumptions or models, with limited insights that can be gleaned from disclosed risks, as well as making benchmarking exercises challenging. As physical risk is inherently geo-location dependent, there is expected to be some degree of regional nuance, and firms with greater present-day exposure to acute physical climate risk may be expected to disclose to a relatively higher standard. This, coupled with varying data and skillsets maturity means that standardization will not occur overnight. However, there needs to be a roadmap that sets out clear expectations for corporates reporting for the very first time versus those in their n-th iteration.

Intrinsic complexity in rating, and reporting on physical climate risk

The source of physical climate risk data used for reporting originates from outputs of global climate models but is then heavily ‘processed’ by climate service providers to arrive at reporting ready end results.

A crucial challenge, but one that is not often discussed, is the underlying variability and quality of climate data.

First, climate data from different providers differ significantly, making comparisons of projections nearly impossible (Hain, Kölbel, & Leippold, 2022). Second, climate models still struggle to accurately represent extreme weather events such as extreme rainfall and heatwaves. For global climate models, this challenge primarily arises due to the coarse spatial resolution, which limits their ability to capture the small-scale processes that drive these extremes , (John, Douville, Ribes, & Yiou, 2022). Even though recent advances in high-resolution modeling have improved the understanding of localized phenomena, these models continue to face challenges in accurately simulating extreme events . Furthermore, the quality of data also varies by region. For example, broad swathes of Africa often have less robust historical data compared to North America and Europe for say bias correction purposes. This discrepancy not only skews the global understanding of climate risks but also complicates the ability of businesses to make informed decisions in these underrepresented areas. While there have been recent advances in AI applications to climate data challenges, it remains uncertain if this would materially advance our understanding of future extreme events.

Next, the complexity of the climate system and its representation introduces significant uncertainty within the disclosure process. This involves making assumptions about future greenhouse gas emissions, socio-economic developments, and technological advancements. While this is somewhat expressed via different climate scenarios, models often vary in their climate sensitivity to the same emissions pathway, leading to a wide range of possible outcomes. For example, the same scenario might produce markedly different projections in different models, depending on how they resolve variables like cloud cover or ocean- atmosphere-land interactions. This is particularly problematic because outputs from climate models that are of most interest to inform actionable insights in risk management are the most uncertain such as extreme precipitation. The propagation of uncertainty continues when extreme precipitation is used a key input into hydrological models to project flood risks involves assumptions about land use, soil properties, and water management paractices, all of which introduce additional marginal uncertainty. As such, the sum of (known) uncertainty at the end of the modelling chain can potentially undermine the fundamental reliability of the assessment , . Worryingly, the inclusion of uncertainty bands and their discussion is currently entirely absent from most disclosures to date, and is not stated as an explicit requirement. Instead, data products providing a very granular resolution (< 1km) for future projections provide a false impression of accuracy and lead to ill informed decisions.

Quantifying the impacts of climate change using loss models adds another layer of complexity. Loss models, which produce financial impacts based on observed climate data, are crucial for translating physical climate risks into economic terms. However, these models face significant challenges, primarly due to inherent uncertainties in climate predictions and event sets that are informed by past events. Past events may not capture all potential outcomes nor provide robust counter-factuals. For instance, accurately modelling losses of extreme events, such as flood and tropical cyclones, requires high-resolution climate data, detailed vulnerability information, and robust financial models. The uncertainties in each of these components can propagate through the modelling chain. Furthermore, loss models traditionally used in the re/insurance industry often struggle to account for compounding events over time, space and across different perils as they have been traditionally purposed to inform annual, and peril specific contracts.

Conclusion and improvement suggestions

To enhance the effectiveness of physical climate-related disclosures, there needs to be a dialogue between the regulators, academia and corporates to develop a minimum agreed standard that allows actual risk insights to be derived from reporting instead of a mere tick the box exercise. ÌýFor corporates to become a meaningful participant in this regard, a mindset shift needs to happen, recognizing that accurate and comprehensive climate-related disclosures are not just a matter of regulatory compliance but also a critical aspect of risk management in a rapidly changing environmental and regulatory landscape that ultimately delivers shareholder value. This would hopefully lead to positive ancillary changes such as development of actual in-house physical climate risk expertise rather than completely outsourcing the work to an external consultant.

Although well intentioned, the voluntary reporting regime has generally proven to be largely ineffective in terms of driving improvement in the robustness of disclosures over time, with successive rounds of reports seemingly receiving the ‘roll-forward’ treatment rather than material improvement in the breadth and depth of disclosures. Instead, to drive incremental improvement, it is essential to mandate and standardize reporting frameworks on a global basis. These frameworks need to provide a clear roadmap towards the disclosure of actual results, along with clear methodology outline and treatment of uncertainty in projections, as opposed to simply stating that climate risk assessments have been undertaken. In other words, there needs to be a roadmap that sets out clear expectations for corporates reporting for the very first time versus those in their n-th iteration.

To tackle the challenge of comparability, being transparent on climate data and methodology is important as studies have shown that data from different providers can vary substantially . Transparency on the methodology of physical climate risk assessments is crucial for ensuring credibility and trust among stakeholders. It involves clearly communicating the assumptions, models, and data sources used in the assessment process. This includes justifying for example, why climate models are used, the specific climate model(s) selected, the scenarios applied, and how specific hazards are quantified. Additionally, it is important to explain any limitations or uncertainties within the assessment to provide a comprehensive understanding of the results.

By addressing these challenges, physical climate-related disclosures can become powerful tool for enhancing real understanding of climate risks within industry, and with this new-found appreciation for risks, Ìýdrive the transition to a more resilient and sustainable economy.

Bibliography

• Doury, A., Somot, S., Gadat, S., & Corre, L. (2022). Regional climate model emulator based on deep learning: concept and first evalutation of a novel hybrid downscaling approach. Climate Dynamics.
• European Commission. (2023). EU Taxonomy Appendix A: Generic Criteria for DNSH to Climate Change Adaptation. Retrieved from European Commission:
• Hadush, M., Zhang, Y., Tian, J., Ma, N., Zhang, X., Heidari, H., & Naeem, S. (2022). An Integrated Modeling Framework in Projections of Hydrological Extremes. Surveys in Geophysics.
• Hain, L. I., Kölbel, J. F., & Leippold, M. (2022). Let's Get Physical: Comparing Metrics of Physical Climate Risk. Finance Research Letters.
• Jeantet, A., Thirel, G., Lemaitre-Basset, T., & Tournebize, J. (2023). Uncertainty propagation in a modelling chain of climate change impact for a representative French drainage site. Hydrological Sciences Journal.
• John, A., Douville, H., Ribes, A., & Yiou, P. (2022). Quantifying CMIP6 model uncertainties in extreme precipitation. Weather and Climate Extremes.
• Kudiabor, H. (2024). Google AI predicts long-term climate trends and weather — in minutes. nature.
• Lamb, C., Clderon, P., McDOnald, J., Cerrato, D., Reinhammar, D., Lonnqvist, C., . . . Foulger, A. (2022). High and Dry, How water issues are stranding assets. Swiss Federal Office for the Environment (FOEN).
• Roston, E., Karra, K., Kaufman, L., & Rangarajan, S. (2024). Bloomberg. Retrieved August 28, 2024, from
• Zhang, S., & Chen, J. (2021). Uncertainty in Projection of Climate Extremes: A Comparison of CMIP5 and CMIP6. Journal of Meteorological Research.