Title: Cyber claim analysis through generalized Pareto regression trees with applications to insurance pricing and reserving
Authors: Sebastien Farkas - Sorbonne Universite, CNRS, Laboratoire de Probabilites, Statistique et Modelisation, LPSM (France) [presenting]
Olivier Lopez - Sorbonne Universite Paris (France)
Maud Thomas - Sorbonne University (France)
Abstract: Cyber claim databases are heterogeneous and contain extreme values. This heterogeneity is caused by the evolution of the risk but also by the evolution in the quality of data and of sources of information through time. We propose a methodology to analyze the heterogeneity of cyber claim databases using regression trees. We consider a public database considered as a benchmark for cyber event and more specifically for data breaches. Particular attention is paid to the tail of the distribution, using a generalized Pareto likelihood as splitting criterion for growing the regression tree. Combining this analysis with, on the one hand, a model for the frequency of the claims, and on the other hand, a model for loss quantification of data breaches, we develop a simple model for pricing and reserving in cyber insurance.