Weighted V\@R and its properties (Q854285)

The author considers coherent risk measures \(\rho_\mu (X)\) defined on the space \(L^0\) of all random variables. As usual, these measures are considered on \(L^{\infty}\), so a procedure of extension is described. Then the basic facts about such coherent risk measures as Tail V\@R and Weighted V\@R, two representations of Weighted V\@R and two representations of determining set of Weighted V\@R are given and it is stressed that Weighted V\@R is better, in some sense. It is established that \(\rho_{\mu}(X+Y)<\rho_{\mu}(X)+\rho_{\mu}(Y)\) provided that \(\text{supp }\mu=[0,1]\) and \(X,Y\) are not comonotone. It is named as strict diversification property. A class of extreme measures for a coherent risk measure is introduced and it is shown that there exists a unique extreme measure that is the smallest in the convex order. The next results provide a description of risk-neutral densities that correctly price traded call options and are ``not far'' from the real-world density, the ``distance'' being measured with the help of a coherent risk measure. A Markowitz type optimization problem with restrictions on the coherent risk measure is solved in the case when the measure is Weighted V\@R and the model is complete.