Measuring risk with multiple eligible assets (Q2018547)

The authors consider an abstract setting for the definition of a risk measure. \(\mathcal X\), the set of marketed claims, is an ordered, topological vector space, \(\mathcal S\), the space of eligible actions, a linear subspace of \(\mathcal X\), \(\mathcal A\), the acceptance set, a monotone subset of \(\mathcal X\) and \(\pi\) a strictly positive linear functional on \(\mathcal M\) (a property which imposes restrictions on the choice of \(\mathcal X\)). The risk measure associated with \(X\) is defined as \[ \rho(X)=\inf\{\pi(Z):Z\in\mathcal S,\;Z+X\in\mathcal A\}. \] The purpose of the paper is to study finiteness, continuity and dual representations of \(\rho\) as a function of the properties of \(\mathcal X\), \(\mathcal S\) and \(\mathcal A\). As intuition suggests, some of these properties are related to characterizing the supporting functional of some of these sets, particularly the acceptance set. The multiplicity of eligible assets is related to the possibility that \(\mathcal S\) may have dimension larger than \(1\).