Invited perspectives: An insurer's perspective on the knowns and unknowns in natural hazard risk modelling

Abstract

Abstract. This paper analyses how the current loss modelling framework that was developed in the 1990s to respond to Hurricane Andrew market crisis falls short in dealing with today's complexity. In effect, beyond reflecting and supporting the current understanding and knowledge of risks, data and models are used in the assessment of situations that have not been experienced yet. To address this question, we considered the (re)insurance market's current body of knowledge on natural hazard loss modelling, the fruit of over 30 years of research conducted by (re)insurers, brokers, modelling firms, and other private companies and academics in the atmospheric sciences, geosciences, civil engineering studies, and data sciences among others. Our study shows that to successfully manage the complexity of the interactions between natural elements and the customer ecosystem, it is essential that both private companies in the insurance sector and academia continue working together to co-build and share common data collection and modelling. This paper (i) proves the need to conduct an in-depth review of the existing loss modelling framework and (ii) makes it clear that only a transdisciplinary effort will be up to the challenge of building global loss models. These two factors are essential to capture the interactions and increasing complexity of the three risk drivers – exposure, hazard, and vulnerability – thus enabling insurers to anticipate and be equipped to face the far-ranging impacts of climate change and other natural events.

Invited perspectives: Safeguarding the usability and credibility of flood hazard and risk assessments

“…Little data exist to aid in assessing processes such as dam or levee breaches or the behaviour of humans in flood situations. Sometimes, deficits in individual model components are compensated by adjusting other model components where data are scarce; for instance, by adjusting vulnerability functions so that damage estimates agree with reported values (Déroche, 2023).…”