Systemic Risk of Modelling in the Insurance Sector
Risk models are employed in the insurance business to assess the probability and size of risk events; risk models are never perfectly accurate. However, the variety of models used in the insurance sector is severely limited. This creates a danger that all insurance companies may rise and fall in tandem, making the sector brittle and creating a public welfare problem. This project aims to determine the potential gains in resilience that can result from making use of a diversity of models, even when some of the models are suboptimal. We are attacking this problem through a combination of different tools, including extreme value theory, statistical analysis and agent-based modeling. The project is developed in collaboration with the insurance company MS Amlin.
People: J. Doyne Farmer, Davoud Taghawi-Nejad, Torsten Heinrich, External Collaborators: Sam Howison, Anders Sandberg