‘Imagination’ needed in a more complex world: AXA XL Reinsurance North American CEO Schiffer

AXA XL Reinsurance North America CEO Greg Schiffer used his keynote at Intelligent Insurer’s  Reinsurance Outlook USA 2026 conference to deliver a warning to the industry – in an increasingly volatile and interconnected world, data and modelling alone are no longer enough.

Schiffer  was speaking at the Reinsurance Outlook USA 2026 conference, where he argued that underwriters and reinsurers should increasingly rely not only on analytics and expertise, but also on imagination to understand emerging risk.

His remarks centred on a simple but increasingly relevant idea that the industry’s traditional tools remain essential, but they cannot fully prepare re/insurers for the kinds of systemic, fast-moving and previously unthinkable events that are now shaping the global risk environment.

Schiffer described data as the “what” of underwriting – the foundation that establishes where the industry has been and what historical patterns suggest. Technical expertise and underwriting know-how, meanwhile, provide the “so what”, helping practitioners interpret and contextualise information. But in his view, the modern risk environment increasingly requires a third capability –the ability to imagine scenarios that have never happened before.

That challenge is becoming more acute as industries and economies become more interconnected and dependent on complex systems.

To illustrate the point, Schiffer referenced the major blackout that affected Spain and Portugal in April 2025, leaving millions without electricity and disrupting transport, communications and payments systems across the Iberian Peninsula. The event demonstrated how quickly failures in one critical infrastructure system can cascade into broader economic disruption.

While the (re)insurance industry can model historical weather events or infrastructure failures, Schiffer argued that many modern risks emerge through combinations of events that traditional models struggle to anticipate fully. The issue is not simply whether an event can be predicted, but whether the industry has considered how interconnected systems may behave under stress.

He pointed to examples such as Hurricane Andrew and 9/11 as reminders that some of the most consequential losses in (re)insurance history emerged from scenarios that either were not modelled at all or were considered too remote to drive underwriting decisions.

For Schiffer, this does not mean abandoning models or data-driven underwriting. Instead, he argued for a balance between quantitative tools and experienced judgement.

He reflected on earlier debates within the (re)insurance industry around catastrophe modelling, recalling how some market participants once relied almost entirely on model output while others largely dismissed models in favour of instinct and experience. In his view, both approaches were flawed. Successful underwriting requires a balance between analytics, technical expertise and human judgement.

That judgement, he suggested, often comes from experience with real losses and difficult underwriting cycles.

Schiffer recounted lessons learned from large historical claims, including the importance of understanding exclusions, aggregation and emerging exposures that may not be fully captured in policy language or catastrophe scenarios. Underwriters, he argued, develop value not simply through technical training but through exposure to unexpected outcomes and complex claims events.

At the same time, he acknowledged that the industry faces a significant workforce challenge.

(Re)insurance is competing for talent in a labour market increasingly dominated by technology firms and digital industries. Schiffer suggested that attracting the next generation of professionals will require insurers and reinsurers to broaden the types of skills and perspectives they value.

While actuarial expertise and analytical rigour remain critical, he argued the industry also needs people capable of thinking creatively across disciplines and understanding how seemingly unrelated systems interact.

That could include drawing insights from fields such as biology, sociology or behavioural science to better understand resilience, systemic risk and organisational behaviour. The goal, he said, is to develop professionals who can connect disparate pieces of information and think beyond conventional risk frameworks.

Schiffer positioned artificial intelligence as an important tool within that process, but not a replacement for underwriting judgement.

He noted that fears around technological disruption are hardly new for the (re)insurance sector. Over the past several decades, the industry has adapted repeatedly to innovations ranging from catastrophe bonds and cat models to the internet itself. Rather than replacing underwriters, those tools changed the nature of underwriting work and forced practitioners to evolve.

AI, in his view, will likely follow a similar trajectory.

Models and AI systems can identify correlations and synthesise vast amounts of historical information, but they remain constrained by the quality and scope of the underlying data. They are inherently backward-looking to some degree, reflecting known patterns rather than genuinely unprecedented developments.

That limitation becomes increasingly important as climate change, geopolitical instability, cyber dependency and infrastructure concentration create more complex forms of risk accumulation.

Schiffer highlighted the example of Hurricane Andrew in 1992, noting that the storm developed rapidly despite forecasts suggesting a relatively quiet hurricane season. The lesson, he suggested, was not that forecasting is incorrect, but that reliance on averages and historical assumptions can create blind spots.

Similarly, he warned against simplistic interpretations of climate patterns such as El Niño. While El Niño years may historically reduce Atlantic hurricane activity, they can also increase exposure to flooding, drought or severe convective storms in other regions. The broader point was that risk correlations are becoming more dynamic and interconnected.

One area Schiffer repeatedly returned to was systemic infrastructure risk.

He encouraged delegates to think beyond traditional (re)insurance silos and consider how failures in technology or communications systems could affect industries far removed from the original trigger event. A satellite outage, for example, could disrupt shipping, aviation, agriculture and supply chains simultaneously, creating cascading economic losses across multiple sectors.

Such scenarios require underwriters to think beyond standard historical datasets and ask broader “what if” questions.

Importantly, Schiffer also argued that imagination is not solely about identifying worst-case scenarios. It can also help companies identify resilience opportunities and loss mitigation strategies.

He pointed to airports in Spain and Portugal that maintained operations during the blackout because they had invested in backup generation and automated failover systems. Likewise, he highlighted the emergence of increasingly automated “dark warehouses” using AI and predictive systems to protect food and pharmaceutical inventories during power disruptions.

For (re)insurers, he suggested, these developments underline the growing importance of understanding resilience investments alongside pure exposure metrics.

Ultimately, Schiffer’s message was that the re/insurance industry must become more comfortable thinking about risks that cannot yet be fully quantified.

Data, modelling and underwriting expertise remain indispensable. But in a world shaped by interconnected systems, emerging technologies and increasingly unpredictable events, he argued that imagination may become one of the industry’s most valuable underwriting tools.

Did you get value from this story?  Sign up to our free daily newsletters and get stories like this sent straight to your inbox.