A0746
Title: An optimal index insurance framework for extreme losses
Authors: Daniel Nkameni - CREST ENSAE (Polytechnic Institute of Paris) (France) [presenting]
Abstract: The modern landscape of insurance is characterized by the emergence of new risks with destructive potential, challenging the feasibility of developing sustainable financial protection. Several administrations and regulators regularly highlight the potential uninsurability of certain risks, such as those associated with natural disasters in the context of climate change or cyber-attacks in the context of rapidly advancing artificial intelligence. To address these challenges, index-based insurance is frequently mentioned as a technical tool that may help provide coverage in these seemingly desperate situations. The construction of index insurance coverage against extreme losses is proposed, focusing on cases where these losses follow heavy-tail distributions. This index coverage is designed to activate above a certain threshold, while classical indemnity-based insurance is maintained below this threshold, leveraging the advantages of both types of coverage. The proposed methodology begins by adapting the usual utility of the wealth framework to the situation of heavy-tail losses. Subsequently, regression trees are employed to build the optimal index payout function through utility maximization. The final step involves determining the optimal threshold for the transition between the two types of coverage.
