Speaker: Cagin Ararat, Bilkent Univeristy
Abstract: Various approaches to measure systemic risk have been proposed in the recent literature using multivariate functionals. These functionals are typically defined in terms of a so-called aggregation function, which takes into account the interconnectedness of the institutions in the network, and a univariate risk measure applied to the random output of the aggregation function. In this talk, we specialize into the Eisenberg-Noe and Rogers-Veraart network models and formulate the corresponding systemic risk measures as large-scale vector optimization problems. While the former model yields a convex problem which can be solved efficiently using non-smooth optimization techniques, the efficient frontier for the latter model turns out to be the boundary of a non-convex set.