Natcat modelling - the challenges of existing models
Article information and share options
At the Catastrophe Knowledge Exchange, held at the Swiss Re Centre for Global Dialogue, the discussions identified particular challenges for modelling around correlated peril combinations like flood and wind, ranging from modelling the interactions between the perils, and how event definitions may vary to increased problems calibrating the models against losses once new ambiguity is introduced regarding hazard definitions.
In Rick Thomas (Executive Director, Willis Re) and Caspar Honegger (Head Flood, Swiss Re)'s session on Event definitions – do wordings match models?, it was determined that there a couple of ways that models can define events, and that event definitions are evolving and creating new challenges.
Further complications come from some of the new features of event definitions which allow clients optionality regarding what they include or don’t include in an event. Optionality also introduces new accumulation control and portfolio modelling challenges for reinsurers. The problem that some of these wordings are untested was also highlighted. The discussions identified some particular challenges for modelling around correlated peril combinations like flood and wind, ranging from modelling the interactions between the perils, and how event definitions may vary to increased problems calibrating the models against losses once new ambiguity is introduced regarding hazard definitions.
From a modelling perspective, the solution must be greater flexibility and transparency regarding how the models are put together with more mechanisms to aggregate and/or disaggregate different perils and “event” losses in model outputs. HD modelling will bring some of this, but it important that modelling firms allow clients increasing flexibility to change the assumptions behind the simulation of event losses within each simulated year.
What current exposure data misses
In the breakout session on What current exposure data doesn't tell us, by Balz Grollimund, The Head Earthquake, Cat Perils and Treaty Centre, Swiss Re and Mark Weatherhead, Head of Model Development at Guy Carpenter - the discussion began with three questions:
1) What is it? (occupancy, structure type, age, protection etc)
2) Where is it? (exact location, extent of site, supply chains, etc)
3) How much is it? (Sums insured, replacement values, deductibles, limits, etc)
Participants were asked if current data is adequate to address each of the three questions and how they addressed the shortcomings. Finally, they were asked if their views changed depending on the country and peril of interest.
Summary by Brian Rogers, Swiss Re Centre for Global Dialogue