Bridging Science and Practice to En(in)sure Resilience in a Changing Climate

Based on their collective experiences, authors Colby K. Fisher, Elinor Benami, Michaela Dolk, Jane W. Baldwin, Inbal Becker-Reshef, Rosa I. Cuppari, Tobias Dalhaus, Andrew Hobbs, Gregor C. Leckebusch, Peter Lacovara, Adam H. Sobel, and Elizabeth Tellman, discuss how closer coordination between science and practice is essential for developing effective insurance programmes that can address inequitable protection gaps.

Extreme weather disasters have yielded more than US$4 trillion in damages since 1979 globally, of which over a third have occurred in the last decade (1). Such losses have garnered increasing attention, as exemplified by the Spring 2023 guidance issued by the US President’s Council of Advisors on Science and Technology on translating advances in scientific prediction of extreme weather risk to protect vulnerable communities (2). This and other recent publications (3, 4) reflect a growing need for science-driven solutions to address extreme natural hazards. Yet, even beyond the United States, natural hazards pose increasing threats to sustainable development and warrant attention^a. Natural hazards and extreme weather events can not only cause devastating fatalities and immediate destruction, but they also have long-lasting financial impacts that disrupt development progress or reverse development gains (6–8). Climate change introduces further challenges by shifting the frequency, severity, and geographic distribution of natural hazards such as floods, cyclones, wildfires, and droughts (9), posing threats to risk management systems not designed to deal with events of such unprecedented magnitude (10). The impacts from these hazards are typically borne most severely by those least equipped to handle them, disproportionately affecting historically marginalised communities (11).

Mitigating the financial risk of natural hazards can be addressed through risk reduction^b and management of the residual risk, which may be explicitly or implicitly accepted (retained)^c or shifted to another party (transferred, e.g., via insurance)^d. Although risk retention remains an important disaster risk finance strategy, risk transfer is increasingly considered a critical tool to dampen the financial impact of disasters and accelerate recovery times (12–14). Despite this, global reinsurers have noted that over half of the US$270 billion of global economic losses incurred in 2022 from natural disasters were uninsured, a statistic that has remained relatively consistent over the last 10 years (15). This reflects society’s “protection gap,” i.e., the financial impacts of disasters borne (i.e., retained) by society. Left unaddressed, this gap leads to disproportionate impacts on under-protected communities, often already marginalised, ultimately perpetuating a cycle of poverty and vulnerability. Thus, the current large protection gap exacerbates inequalities as it impedes progress towards improving resilience and climate adaptation.

Effective risk transfer depends on strong scientific foundations (e.g., for risk characterisation and monitoring). Although recent innovations in computational techniques and remote sensing technologies have helped unlock discoveries in hazard science, the authors’ experiences interfacing with both science and practice indicate many disconnects between the “state of practice” and the “state of the science” (16–18), resulting in missed opportunities to advance risk transfer. This paper highlights the need to strengthen bridges between science and practice, identifies weaknesses in the status quo, and suggests how scientists, practitioners, and policy makers can contribute to delivering on the promise of effective risk transfer as a vehicle for climate adaptation.

The need for strong bridges between science and practice

Building stronger, reciprocal bridges between science and practice offers opportunities to enhance financial resilience to disasters in a changing climate. Shoring up these bridges first involves understanding the roles that each party plays (Figure 1).

Figure 1: A synthesis of the roles of scientists, practitioners, and policymakers, indicating where current bridges are weak, and opportunities to strengthen them.

Roles for science in characterising and monitoring risks: Scientific research can improve risk transfer instruments in two critical ways: risk modelling and hazard monitoring. Risk modelling involves characterising the features (e.g., expected distribution, magnitude, and frequency of damages and losses) of the underlying disasters, which enables practitioners to design, price, and underwrite risk transfer instruments. Hazard monitoring involves, for example, identifying transparent, timely, and effective triggers for risk transfer instruments to provide the most value to beneficiaries. Recent scientific advances in earth observation, physical models, and forecasting can enhance the robustness of these systems, particularly in a changing climate that is rapidly eroding the adequacy of prior approaches (19). Ultimately, the dynamic nature of risk underscores the importance of enhancing connections between science and practice to ensure that risk transfer instruments match the evolving nature of the problems they seek to address.

Roles for practice in implementation and feedback: Although scientific approaches can contribute to the development and improvement of constituent product elements, the operational aspects of risk transfer implementation extend beyond the domain or strengths of most scientific organisations. Instead, practitioners (e.g., public, private, or non-profit organisations) are ultimately responsible for developing and implementing risk transfer products while demonstrating a clear value proposition to risk-bearers. The role of practitioners can range from discerning risk bearers’ needs to structuring, pricing, placing, and deploying operational systems^e for risk transfer products. Ultimately, practitioners often play a role of reducing the information asymmetries among potential policyholders about their risk profiles and illustrating the value that products can offer in mitigating that risk. Increasingly, governments and scholarly communities are asking practitioners to evaluate implementation effectiveness, too, believing that practitioners should be able to identify and apply relevant scientific findings to effectively deliver on the promise of what risk transfer can offer. Practitioners can also offer important feedback and insights to the scientific community by identifying key challenges they encounter during implementation and indicating gaps that scientific research can address, strengthening the offered products and improving their market fit.

Why have bridges been so weak?

Although interest in risk transfer for climate action has been growing (20), a few key challenges related to the bridges between science and practice have broadly impeded the development of risk transfer tools for multiple perils and regions across the globe, leaving many gaps in the scope of effective coverage. Based on the collective experience of the author team in seeking to address these issues for cyclones, floods, droughts, and other hazards across the planet, we have encountered the following repeated challenges:

1. Misalignment of timescales: Academic research, publication, funding, and hiring cycles often do not correspond well to the time frames of practitioners who need to make decisions and implement projects quickly (17, 21), and who may not have time to wait for new models to be developed or stress-tested. For example, traditional scientific projects are typically measured in units of a few years as part of a PhD or postdoc appointment, whereas industry often has much shorter-term deliverables distributed over project lifespans. This can generate pressure to implement “interim” and often suboptimal solutions.
2. Misalignment of incentives: Despite increasing recognition of public service or impact as a potentially valuable outcome of science, the coin of the realm among scientific researchers implies a focus on publications and grant funding. In contrast, practitioners are incentivised to focus on creating products that induce and sustain enrollment. This may, at times, be at odds with products that would most benefit marginalised populations.
3. Lack of regional demand or funding: Insufficient funding or weak “business cases” can impede the development of locally-grounded risk models or data collection that would enable appropriate risk transfer. Weak business cases can arise from contexts with high uncertainty, often in areas where limited data collection has occurred in the past, further complicating efforts to adapt models to local contexts. Poorly calibrated models that do not clearly offer value to intended beneficiaries translate to muted demand and can contaminate the reputation of risk transfer for otherwise well-designed offerings. This issue is especially problematic when many of the most difficult “business cases” are among the populations most vulnerable to shocks. From both the science and practitioner perspectives, the funding-data-demand gap can hinder efforts to address the unique challenges in underserved regions or sectors.
4. Proprietary models and paywalled publications: Many risk assessment models are proprietary, which impedes the ability of scientists to scrutinise the models, compare their performance, and contribute to improving them. As a result, even practitioners with access to model information from proprietary providers may miss out on the value that scientists can bring in helping stress-test models and assumptions. Furthermore, many academic publications are paywalled, rendering the information inaccessible to practitioners and otherwise low-resourced communities. Even with open-access publications, many studies do not distribute underlying models or data, impeding knowledge transfer efforts. Each of these factors impede information transmission and the ability to co-design projects that could otherwise serve both science and society.
5. Proprietary exposure, claims, and vulnerability data: Even in contexts with some open-source models available for assessing hazards (e.g., cyclone risk), exposure and claims data used in industry applications are typically proprietary. The restricted availability of such data impedes the development, calibration, and validation of full risk assessment models that characterise not just hazard but also exposure and ultimately, vulnerability.
6. Inertia and incumbency bias: Practitioners are often hesitant to use new models or approaches that have not yet been widely adopted by others, especially models they have not been involved in developing themselves. Understandably, adopting new practices can be costly or time consuming for practitioners without guaranteed improvements in efficiency or profit; however, this hesitance can make it difficult to introduce otherwise value-enhancing innovations. Conversely, it is often the case that existing industry expertise and advances are not translated into accessible academic research or models. The closed nature of mature catastrophe models often results in scientists using older models or simply creating their own, at times reinventing the industry standard models before they can add to them.

Opportunities to create stronger bridges

In recognition of these challenges and the consequences of inadequate risk transfer, several approaches provide guiding lights that yield benefits for both science and practice. We describe several possible actions below that can build stronger bridges to advance the state of science and practice in risk transfer.

1. Re-examine and align terms, incentives, and timescales of funding and partnerships: Emerging opportunities and examples of science-practice cross-pollination can be seen through the Willis Research Network (est. 2006), the AXA Research Fund (est. 2007), the Aon Research Forum (est. 1996), and the new Gallagher Research Centre (est. 2022) (21). However, even these initiatives are relatively limited in scope or emphasise proprietary and industry-selected criteria for engagement that may advance a given company’s priorities but not the field writ large. An alternative model can be seen in the UK Centre for Greening Finance and Investment, a research collaborative created by the National Environment Research Council and Innovate UK in 2021 to accelerate global financial institutions' use of climate and environmental data to support their decisions and enhance the resilience of the financial sector more broadly. In the US, the President’s Council of Advisors on Science and Technology and the Council of Economic Advisors have suggested that federal science agencies could develop research programmes in climate risk/climate-conditioned catastrophe modelling, with a goal of building underlying scientific understanding and open source models while training future practitioners. Making these models available to the many actors who cannot afford expensive proprietary ones would establish a basic platform that the private sector could integrate and further build on (2, 3). These US or Europe focused efforts remain in active development even with significant resources to support them, suggesting that other less-resourced areas may well find such initiatives even more challenging. Expanding these funding programmes and partnerships to emerging markets and developing economies presents a strong opportunity to advance risk knowledge to support risk transfer in these regions.
   
2. Democratise data, models, publications, and platforms: Several initiatives for supporting open science and data have emerged to support risk assessment, pricing, and other decision-making processes (e.g. Oasis Loss Modeling Framework (LMF), the CLIMADA natural catastrophe impact model, the Global Earthquake Model (GEM), and OS Climate). When used well, these types of initiatives can help overcome challenges associated with proprietary data and models; however, software development and data efforts require funding and support to become viable in commercial applications. Recognising that many models used by industry are proprietary, more open model inter-comparison initiatives can help scientific researchers as well as regulators identify the differences driving results among models with highest local relevance. Similarly, directives from funding agencies to require open access publications (e.g., EU’s Horizon funding and the US guidance from OSTP in 2022) further help accelerate constructive exchange.
   
3. Improve data sharing and standardisation: Data-sharing and standardisation can help accelerate both model development and the ability to evaluate the impacts of risk transfer products in practice. For example, in 2019, the US National Flood Insurance Program’s (NFIP) released more than two million claims records dating back to 1978 and more than 10 years of transactions data aggregated to the census tract level, to aid the development of a robust flood insurance market (22). Alternatively, a “sworn status” model like what the US Census Bureau uses could provide access to credentialed researchers acting in good faith to serve public purposes for risk management while also bringing value to private sector operations. While regulators and policy can provide the impetus for such standardisation, industry actions could also generate such initiatives themselves, e.g., through the Open Data Standards initiative currently chaired by Oasis LMF. Endeavours by scientists to make the wealth of relevant data (e.g., hazard observations and forecasts or climate model projections) more usable by practitioners would bolster these efforts. Integrating FAIR (Findable, Accessible, Interoperable, and Reusable) data principles is critical. These ensure data is easily discoverable, usable across different systems, and valuable for diverse applications, enhancing research quality and collaboration. Transparency on data, models and risk transfer products can enable clear evaluations of these to inform future improvements and the development of more effective products (23), ultimately benefiting the communities that scientists and practitioners are working to protect.
   
4. Facilitate/support collaboration opportunities between science and practice to drive innovation: To achieve the development of scientific models that are useful for practitioners, inputs from both scientists and practitioners are critical. Many of the proprietary models currently accepted and used by the (re)insurance industry have benefitted from decades of feedback between model vendors and users. While the development trajectory of these models has arguably been cumbersome, slow, and lacking openness, the feedback loop between science and practice has made them usable tools. Emulating that approach but with greater speed, broader expert input, up-to-date science, open approaches, and open data could drive innovation for the next generation of models.

Several avenues exist to foster enhanced dialogue and cooperation between these two communities. Short term consultancies or secondments, along with other opportunities for in-person interaction, can help share lessons across organisations and pollinate them with new ideas to help overcome inertia and incumbency bias. For the last four years, for example, the co-authors have convened sessions on innovations in hazard science and risk transfer at the American Geophysical Union and other conferences that have attracted practice-informed scholars as well as practitioners representing insurance, non governmental organisations, and beyond. Other similar types of venues include the Microinsurance Network, the InsuResilience Global Partnership for Climate and Disaster Risk Finance and Insurance Solutions, and the Global Risk Modelling Alliance supported by the Insurance Development Forum.

Implications for policy and action

A few actor-specific activities can help advance risk transfer. Practitioners can play an important role by communicating out the key questions, challenges, and bottlenecks they face. Additionally, being open to a wider range of model providers beyond those traditionally accepted by the (re)insurance industry, and collaborating with scientists to support the development of these models, can help advance the field further. Organisations like the Centre for Disaster Protection also serve as a potentially useful model for encouraging monitoring, learning, and evaluation of the approaches used for risk transfer and expanding the solution set for practitioners. Scientists can also facilitate uptake of their work into practice by exploring options to adapt their analyses into standardised open-source modelling platforms. For example, they can develop “plug and play” modules for individual components of a risk model, enabling rapid development and adoption of scientific data and modelling advances.

Policy plays an important cross cutting role in incentivising and fostering collaboration between scientists and practitioners to enhance risk transfer and improve disaster risk management. Agencies involved in science funding or disaster risk management can support initiatives to develop relevant fundamental knowledge (e.g., risk mapping or impact evaluation). Policymakers can also encourage use of open modelling techniques and platforms through regulatory means or incentives (2). Informed by the needs of scientists and practitioners, policy can also drive improved data collection systems that enable robust – and where appropriate, spatially explicit – model development, calibration, and validation, e.g., higher density networks of weather or crop monitoring stations and water gauges, as well as efforts to track variation in vulnerability, exposure, and impact.

Ultimately, science and practice play important complementary roles in the design and execution of effective risk transfer, especially in a changing climate where quick action to the right place at the right time can be critical for mitigating the financial and social impacts of disasters. Yet, to deliver on the promise of stabilising lives and livelihoods in the wake of a disaster, shoring up the weak bridges between science and practice is critical. Even amid our focus on risk transfer, lessons learned from this domain can be more broadly applied to other disaster risk management and resilience efforts. We call upon our colleagues in science and practitioner communities – alongside the policy makers that shape our work – to push on these frontiers and take advantage of emerging opportunities to improve the ability of risk transfer products to effectively en(in)sure financial resilience to disasters in a changing climate.

Footnotes

a: While we refer to these hazards as natural, several of them may also be classified as socio natural hazards according to the Sendai Framework, as they include elements of both natural and anthropogenic factors (5).
b: For example, (non)structural measures to address hazard, exposure, and vulnerability
c: For individuals, households, or businesses who have retained risk, responding to a disaster may mean using savings, selling assets, or seeking a loan. Where such options are insufficient or unavailable, poverty levels often increase. For governments, risk retention may involve a variety of ex-ante approaches (e.g., reserve funds, contingent credit) and ex-post approaches (e.g., budget reallocations, post-disaster debt).
d: For individuals, households, or businesses, risk transfer instruments may include indemnity or parametric insurance for property, crops, livestock, or business interruption. For governments, risk transfer instruments may include insurance, reinsurance, risk pools, catastrophe bonds, and other insurance-linked securities.
e: For example, pay-out protocols and delivery system

Acknowledgments

AHS acknowledges support from the Swiss Re Foundation and Aon, IB and EB acknowledge support from NASA Cooperative Agreement: 80NSSC17K0625 (NASA Harvest). MD engaged in the preparation of this work prior to joining the World Bank. We also appreciate the contributions of many participants from four years of AGU sessions related to this topic, especially Julio Herrera Estrada and Dickie Whitaker.

Author contributions

Conceptualization: CKF, EB, MD
Writing – original draft: EB, MD, CKF
Writing – review & editing: JWB, IB, RIC, TD, AH, GL, PL, AHS, ET, CKF, EB, MD

References

1. World Meteorological Organization. WMO Atlas of Mortality and Economic Losses from Weather, Climate, and Water Extremes (1970–2021). https://public.wmo.int/en/resources/atlas-of-mortality (2023). [Accessed May 2023]. 
2. President’s Council of Advisors on Science and Technology. Report on extreme weather. https://www.whitehouse.gov/pcast/briefing-room/2023/04/24/pcast-releases-report-on-enhancing-prediction-and-protecting-communities-against-extreme-weather-risk/  (2023). [Accessed April 2023]. 
3. Council of Economic Advisers and Office of Management and Budget. Methodologies and considerations for integrating the physical and transition risks of climate change into macroeconomic forecasting for the president’s budget.  https://www.whitehouse.gov/cea/written-materials/2023/03/14/methodologies-and considerations-for-integrating-the-physical-and-transition-risks-of-climate-change-into macroeconomic-forecasting-for-the-presidents-budget/ (2023). [Accessed March 2023]. 
4. Condon, M. Climate services: The business of physical risk. Arizona State Law Journal (2023). 
5. United Nations Office for Disaster Risk Reduction. Sendai framework terminology on disaster risk reduction. https://www.undrr.org/terminology [Accessed April 2023]. 
6. Hillier, D., Oxfam & Nightingale, K. How disasters disrupt development: Recommendations for the post 2015 development framework. Tech. Rep., Oxfam International (2013). 
7. Benson, C. Promoting sustainable development through disaster risk management. ADB Sustainable Development Working Paper Series 41, Asian Development Bank (2016). 
8. United Nations Office for Disaster Risk Reduction. Global assessment report on disaster risk reduction 2022: Our world at risk: Transforming governance for a resilient future. Tech. Rep., United Nations (2022). 
9. Mills, E. Insurance in a climate of change. Science 309, 1040–1044 (2005). 
10. Kreibich, H. et al. The challenge of unprecedented floods and droughts in risk management. Nature 608, 80–86 (2022). 
11. Wing, O. E. J. et al. Inequitable patterns of US flood risk in the Anthropocene. Nature Climate Change 12, 156–162 (2022). 
12. Otto, C., Kuhla, K., Geiger, T., Schewe, J. & Frieler, K. Better insurance could effectively mitigate the increase in economic growth losses from U.S. hurricanes under global warming. Science Advances 9, eadd6616 (2023). 
13. Clarke, D. J. & Dercon, S. Dull Disasters? How planning ahead will make a difference (Oxford University Press, 2016). 
14. Cambridge Institute for Sustainability Leadership. Mutual microinsurance and the sustainable development goals: An impact assessment following typhoon Haiyan. Tech. Rep., Cambridge Institute for Sustainability Leadership (2019). 
15. Munich Re. Natural disaster risks: Rising trend in losses https://www.munichre.com/en/risks/natural-disasters.html (2023). [Accessed April 28, 2023]. 
16. Keenan, J. M. A climate intelligence arms race in financial markets. Science 365, 1240– 1243, DOI: 10.1126/science.aay8442 (2019). 
17. Hillier, J. K. et al. Demystifying academics to enhance university–business collaborations in environmental science. Geoscience Communication 2, 1–23 (2019). 
18. Surminski, S., Barnes, J. & Vincent, K. Can insurance catalyse government planning on climate? Emergent evidence from Sub-Saharan Africa. World Development 153, 105830, DOI: 10.1016/j.worlddev.2022.105830 (2022). 
19. Steptoe, H., Souch, C. & Slingo, J. Advances in numerical weather prediction, data science, and open-source software herald a paradigm shift in catastrophe risk modelling and insurance underwriting. Risk Management and Insurance Review, 69–81 (2022). 
20. Hight, J. Using risk transfer to achieve climate change resilience. https://www.preventionweb.net/publication/using-risk-transfer-achieve-climate-change resilience (2019). [Accessed: March 2023]. 
21. Noacco, V. Democratising cat risk modelling https://www.maxinfo.io/blog/democratising cat-risk-modelling (2022). [Accessed: March 2023]. 
22. Costello, C. & Chadwick, S. The power of data transparency. https://firststreet.org/research-lab/published-research/data-transparency/ (2019). [Accessed: March 2023]. 
23. Centre for Disaster Protection. The Jamaica cat bond: Is it any good?. https://medium.com/@centrefordp/the-jamaica-cat-bond-is-it-any-good-9b87f89c52ff (2019). [Accessed: March 2023].