Multivariate risk measures in the non-convex setting (Q2291757)

When evaluating the risk of a random vector \(X\), describing positions in several assets, e.g., in several currencies, it is indispensable to take into account eventual transfers between its components. If transfers are forbidden, then the risk of \(X\) is the vector of risk measures calculated for its components. A different situation arises when transfers are allowed, and then one should normally take into account transaction costs associated with them. In this case, it is natural to evaluate risks of all admissible positions which can be attained from \(X\) by exchanges and call \(X\) acceptable if there exists a transfer (a selection of the set of attainable positions) that converts \(X\) to a vector with all individually acceptable components. The attainable positions form a random closed set, which is convex for most transaction costs models. The paper under review deals with the non-convex case, which arises, e.g., if fixed transaction costs are imposed. For this, it follows the approach based on considering all selections of the portfolio and checking if one of them is acceptable. Properties and basic examples of risk measures of non-convex sets of attainable positions are presented. A particular attention is devoted to the special case of fixed transaction costs.