A1049
Title: A signal-processing approach in cyber risk valuation
Authors: Rogemar Mamon - University of Western Ontario (Canada) [presenting]
Abstract: The cyber risk insurance market is rapidly developing due to the potentially huge losses from cyberattacks. We present such a framework for cyber risk modelling, wherein the cyberattacks occurrences and their inter-arrival and duration are captured by a regime-switching Markov model (RSMM). In this customised RSMM, the transition probabilities of the Markov chain are governed by another hidden Markov chain representing the various states of the cyber security environment. A self-calibrating mechanism is provided via filtering, and a cyber kill chain is built based on the stages of the cyberattack. With the aid of change of reference probability measures and the EM algorithm, the estimators for the transition matrix are derived. The main point of interest is the random losses from cyberattacks, which are assumed to follow a doubly truncated Pareto distribution. The Vasicek model is utilised to describe the interest rate process for the discounting of losses. The premium for a cyber security insurance contract is calculated via a simulated data set based on two pricing principles. The methodology featuring dynamic parameter estimation and flexible adjustments in modelling various risk factors widens the available pricing and cyber risk management tools.