On s-additive robust representation of convex risk measures for unbounded financial positions in the presence of uncertainty about the market model



max (-Λ(X) - βρ (Λ)) for every X X. Moreover, ρ(X) = sup  (-Λ(X) - βρ (Λ)) holds for every X X if

λx+π                                                      ΛX+π X'

and only if ρ is lower semicontinuous w.r.t. τ.

The proof may be found in Appendix B.

The aim of the paper is to improve the representation results by allowing only representing linear forms which
are in turn representable by probability measures. For notational purposes let us introduce the counterpart of
βρ
w.r.t. the probability measures on the σ-algebra σ(X) on Ω generated by X

αρ : M1 ] - ∞, ], P sup (-EP [X] - ρ(X)).

XX

Here M1 is defined to consist of all probability measures P on σ(X) such that all positions from X are P -integrable,
and
EP [X] denotes the expected value of X w.r.t. P . We shall speak of a robust representation by probability
measures from
M or a σ-additive robust representation of ρ w.r.t. M if M M1 nonvoid, and
ρ(X) = sup (-EP [X] - αρ(P)) for every X X. As an immediate consequence of Proposition 1.1 we obtain a first
PM

characterization of such representations.

Proposition 1.2 Let F be a vector space of bounded countably additive set functions on σ(X) which separates
points in X such that each X X is integrable w.r.t. any μ F. Then in the case that the set M1 (F) of all
P M1 F with EP|C = π is nonvoid

ρ(X) =  sup (-EP [X] - αρ (P)) for all X X

PM1 (F)

if and only if ρ is lower semicontinuous w.r.t. weak topology σ(X, F) on X induced by F.

Remark 1.3 Retaking assumptions and notations from Proposition 1.2, ρ admits a robust representation in terms
of
M1 (F) if F contains the Dirac measures, and if lim inf ρ(Xi) ρ(X) holds for every net (Xi)iI in X which
i
converges pointwise to some X X.

In general the lower semicontinuity of ρ w.r.t. the topology from Proposition 1.2 is not easy to verify. Therefore we
are looking for more accessible conditions. The considerations will be based on the idea to reduce the investigations
to bounded financial positions. As shown in Lemma 6.5 below, in case of
X being a Stonean vector lattice, this
may be achieved if the linear forms from the domain of
βρ are representable by finitely additive set functions
in the sense explained there. Fortunately, we might express this condition equivalently by the property that
lim
ρ(-λ(X - n)+) = ρ(0) ((X - n)+ positive part of X - n) is satisfied for every λ > 0 and any nonnegative
n→∞

X X, which is obviously true if all positions in X are bounded (cf. Proposition 6.6 below).

Before going into the development of criteria for σ-additive representations let us collect some necessary conditions.

In the case that the positions from X are essentially bounded mappings w.r.t. a reference probability measure of



More intriguing information

1. Fiscal Reform and Monetary Union in West Africa
2. The name is absent
3. PERFORMANCE PREMISES FOR HUMAN RESOURCES FROM PUBLIC HEALTH ORGANIZATIONS IN ROMANIA
4. Naïve Bayes vs. Decision Trees vs. Neural Networks in the Classification of Training Web Pages
5. The name is absent
6. A Note on Productivity Change in European Co-operative Banks: The Luenberger Indicator Approach
7. DISCUSSION: POLICY CONSIDERATIONS OF EMERGING INFORMATION TECHNOLOGIES
8. The open method of co-ordination: Some remarks regarding old-age security within an enlarged European Union
9. Cross border cooperation –promoter of tourism development
10. Economies of Size for Conventional Tillage and No-till Wheat Production