Jens Grossklags - Pennsylvania State University Corporate communication
Date: May 26th 2014 Location: Eurecom - Eurecom
The systematic risk of a networked system depends to a large extent on the ways in which the network is connected. In this talk, I will discuss the connection between a network's systematic risk and its topology, using a model of risk propagation from the literature on interdependent security games, and focusing on the loss distribution in terms of the number of compromised nodes. I will show that it is NP-hard to compute this loss distribution for an arbitrary network topology. Nevertheless, it is possible to derive efficient formulae for loss distributions resulting from homogeneous, star, and E-R random topologies. Further, I will introduce a simulation algorithm for approximating the loss distribution in general. Applying the simulation methodology to the study of scale-free networks, I can find systematic risks which distinguish these networks substantively from even their own random subnets. This implies on the one hand, that a random subnet of a network with large systematic risk may still be insurable. On the other hand, the true systematic risk of a networked system may not be easily discoverable by risk assessment methods, such as incident reporting, that are based on subsampling. However, the talk also illustrates by example that this challenge can be met if some general topological features of the connection network are known.