English

Conditional nonlinear expectations

Probability 2020-01-16 v3 Risk Management

Abstract

Let Ω\Omega be a Polish space with Borel σ\sigma-field F\mathcal{F} and countably generated sub σ\sigma-field GF\mathcal{G}\subset\mathcal{F}. Denote by L(F)\mathcal{L}(\mathcal{F}) the set of all bounded F\mathcal{F}-upper semianalytic functions from Ω\Omega to the reals and by L(G)\mathcal{L}(\mathcal{G}) the subset of G\mathcal{G}-upper semianalytic functions. Let E(G) ⁣:L(F)L(G)\mathcal{E}(\cdot|\mathcal{G})\colon\mathcal{L}(\mathcal{F})\to\mathcal{L}(\mathcal{G}) be a sublinear increasing functional which leaves L(G)\mathcal{L}(\mathcal{G}) invariant. It is shown that there exists a G\mathcal{G}-analytic set-valued mapping PG\mathcal{P}_{\mathcal{G}} from Ω\Omega to the set of probabilities which are concentrated on atoms of G\mathcal{G} with compact convex values such that E(XG)(ω)=\mathcal{E}(X|\mathcal{G})(\omega)= supPPG(ω)EP[X]\sup_{P\in\mathcal{P}_{\mathcal{G}}(\omega)} E_P[X] if and only if E(G)\mathcal{E}(\cdot |\mathcal{G}) is pointwise continuous from below and continuous from above on the continuous functions. Further, given another sublinear increasing functional E() ⁣:L(F)R\mathcal{E}(\cdot)\colon\mathcal{L}(\mathcal{F})\to\mathbb{R} which leaves the constants invariant, the tower property E()=E(E(G))\mathcal{E}(\cdot)=\mathcal{E}(\mathcal{E}(\cdot|\mathcal{G})) is characterized via a pasting property of the representing sets of probabilities, and the importance of analytic functions is explained. Finally, it is characterized when a nonlinear version of Fubini's theorem holds true and when the product of a set of probabilities and a set of kernels is compact.

Keywords

Cite

@article{arxiv.1612.09103,
  title  = {Conditional nonlinear expectations},
  author = {Daniel Bartl},
  journal= {arXiv preprint arXiv:1612.09103},
  year   = {2020}
}

Comments

previous title: "Pointwise dual representation of dynamic convex expectations"