Risk was never truly local
There is a tendency in insurance to anchor risk to place. Flood is modelled river by river, windstorm basin by basin, earthquake fault by fault. It makes sense. Data is local, impacts are local, and decisions are often made at that level. But the consequences of those risks have never been local.
By 2007, insurers were increasingly managing exposures that spanned regions, continents and perils. The challenge went beyond simply understanding individual hazards, to understanding how different assumptions, models and vulnerabilities compared across a global portfolio. In the years following Hurricane Katrina, the limitations of viewing catastrophe risk through isolated regional lenses were becoming harder to ignore.
Building a global research network
It was in that context that the Willis Research Network took a deliberate step outward. In its second year, the network expanded beyond the UK, bringing in partners from across the US, Europe and Asia. Institutions such as the National Center for Atmospheric Research, University of Bologna, Princeton University, National University of Singapore, University of Colorado and Kyoto University joined.
20 years of the Willis Research Network
The aim was not simply to grow the network geographically. It was to create a more connected approach to catastrophe research at a time when much of the most relevant expertise remained fragmented across institutions, regions and disciplines. Many academic groups were producing world-leading research, but often in isolation from the practical questions insurers were trying to answer. Models reflected local assumptions, making comparisons across regions difficult. Bringing those perspectives together in a structured and consistent way was rarely done.
Connecting expertise rather than centralising it
This mattered in practice. A portfolio spanning multiple regions could only ever be as strong as the weakest assumption underpinning it. Differences in how hazard, vulnerability or exposure were treated were not always visible, but they shaped pricing, capital allocation and risk selection. Without a more connected view, there was always a risk of false confidence.
The approach taken by WRN was to connect, rather than centralise. Leading scientists remained embedded in their own environments, close to the data, the hazards and local conditions shaping their work. What changed was the framework around them. Through dedicated research fellows and a shared agenda, local insight could be brought into a broader conversation, tested against other regions, and translated into more directly into underwriting and portfolio decisions.
That shift allowed risk to be viewed more consistently across geographies without losing the detail that makes each region distinct. It also helped narrow a gap that had existed for some time between academic research and practical application within the insurance market.
From isolated analysis to comparative understanding
This shift also started to change how problems were framed. Rather than analysing how risk behaved in a single location, it became possible to see how similar risks compared across regions, where assumptions diverged, and which differences actually mattered. The change from isolated analysis to comparative understanding remains one of the more important developments in how risk is assessed.
Looking back, the idea feels simple: bring together the best people, wherever they are, and apply that knowledge to real-world problems.
Looking back, the idea feels simple: bring together the best people, wherever they are, and apply that knowledge to real-world problems. But it addressed a more fundamental issue that still exists today. No single institution, model or discipline has the full picture. Climate behaves differently across regions. Vulnerability varies. Loss is shaped by a combination of physical, economic and social factors. Understanding risk requires those perspectives to be brought together, not treated in isolation.
Why the challenge has intensified
The risks now shaping the market are increasingly interconnected. Flood risk is influenced by climate variability and land use. Severe convective storms, for example, remain highly local in how they develop and impact, but increasingly important at a portfolio level, making them difficult to assess consistently across regions. Supply chain disruptions extend far beyond the point of impact. Even newer areas such as AI liability or cyber risk cut across traditional boundaries in ways that are difficult to define.
Signals of change, whether in climate behaviour or loss patterns, need to be identified and interpreted earlier.
There is also a growing need to move more quickly. Signals of change, whether in climate behaviour or loss patterns, need to be identified and interpreted earlier. That requires not just access to data, but the ability to bring together different types of expertise and test emerging ideas from multiple perspectives. That shift towards a more connected, global view of risk has since been reflected more widely across the industry and in the WRNs continued support for global thinking.
Initiatives such as the Global Earthquake Model Foundation, the Global Volcano Model and Global Flood Partnership have all recognised the same underlying point: understanding risk at a global level requires shared frameworks, open collaboration and access to diverse sources of expertise. What is notable is not just their emergence, but that they reflect an idea the industry has taken time to fully embrace. In that context, a global perspective is about consistency and comparability. It allows local signals to be interpreted more accurately and understood as part of a wider system.
A model that still matters
The question now is less about whether global collaboration is needed, and more about how effectively it is being used. Access to data has improved. Collaboration is easier. Integrating insight across regions and disciplines, and translating that into better decisions, has moved forward significantly, in part through approaches like WRN that were designed to connect science and application from the outset.
The model established early in the Willis Research Network was one way of addressing that challenge. It recognised that expertise is distributed, that risk does not respect boundaries, and that connecting the two is where value is created. That remains as relevant now as it was then.
Author
Many academic groups were producing world-leading research, but often in isolation from the practical questions insurers were trying to answer.