p. 17 (Corollary 1 of Embrechts et al. 2018, as used) — Σ VaR^L_{εᵢ}(Xᵢ) ≥ VaR^L_{Σεᵢ}(Σ Xᵢ)
ProvedTailRiskSharing.TailConv.embrechts_cor1_varL_sump2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-paperp2o-v1quantilerisk-aggregationvalue-at-risk
Let be an atomless probability space, , , and with . Then
In the proof of Theorem 3 this is the inequality with and for , deduced there from Corollary 1 of Embrechts, Liu and Wang (2018). It shows that the left VaRs of the parts dominate the left VaR of the total at the summed level.
Formalization Note Only the case is stated. When the proof takes and , where the inequality is trivial.
Preamble
import Mathlib import Definitions.Def_TailRiskSharing_TailConv_Setting open MeasureTheory
Formal statement
namespace TailRiskSharing.TailConv
/-- Corollary 1 of Embrechts et al. (2018), as used in the proof of Theorem 3, p. 17:
`x_1 + ⋯ + x_n ≥ x` with `x_i = VaR^L_{ε_i}(X_i)` and `x = VaR^L_{Σ ε_i}(Σ X_i)`, for
`ε_i ∈ (0,1)` with `Σ ε_i < 1`. -/
theorem embrechts_cor1_varL_sum {Ω : Type*} [MeasurableSpace Ω] (P : Measure Ω)
[IsProbabilityMeasure P] (hP : TailRiskSharing.VaRConv.IsAtomless P) (n : ℕ) (hn : 1 ≤ n) (εs : Fin n → ℝ)
(hε : ∀ i, 0 < εs i ∧ εs i < 1) (hsum : ∑ i, εs i < 1) (Xs : Fin n → Ω → ℝ)
(hXs : ∀ i, Xs i ∈ Linf P) :
TailRiskSharing.VaRConv.VaRL P (∑ i, εs i) (fun ω => ∑ i, Xs i ω) ≤ ∑ i, TailRiskSharing.VaRConv.VaRL P (εs i) (Xs i) := by sorry
end TailRiskSharing.TailConv
Source
Liu, Mao, Wang & Wei, Inf-convolution, Optimal Allocations, and Model Uncertainty for Tail Risk Measures, SSRN 3490348 (version of January 16, 2022), p. 17, proof of Theorem 3 (Corollary 1 of Embrechts et al. (2018), as used)
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.