(10)–(11), p. 9 — VaR^R_β(Y) ≤ 0 iff P(Y > ε) < β ∀ε > 0, and VaR^L_β(Y) ≤ 0 iff P(Y > 0) ≤ β
ProvedTailRiskSharing.VaRConv.display10_11_var_nonposp2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-paperp2o-v1quantilevalue-at-risk
Let be a probability measure, a random variable and . Then
and
These two characterizations of a non-positive quantile are the basic tool of the proof of Theorem 1: they turn bounds on VaRs of shares of a risk into bounds on tail probabilities.
Formalization Note is measurable (). Atomlessness is not assumed: the equivalences hold under any probability measure.
Preamble
import Mathlib import Definitions.Def_TailRiskSharing_VaRConv_Setting open MeasureTheory ProbabilityTheory
Formal statement
namespace TailRiskSharing.VaRConv
theorem display10_11_var_nonpos {Ω : Type*} [MeasurableSpace Ω]
(P : Measure Ω) [IsProbabilityMeasure P]
(Y : Ω → ℝ) (hY : Y ∈ (L0 : Set (Ω → ℝ))) (β : ℝ) (hβ0 : 0 < β) (hβ1 : β < 1) :
(VaRR P β Y ≤ 0 ↔ ∀ ε : ℝ, 0 < ε → P.real {ω | ε < Y ω} < β) ∧
(VaRL P β Y ≤ 0 ↔ P.real {ω | 0 < Y ω} ≤ β) := by sorry
end TailRiskSharing.VaRConv
Source
Liu, Mao, Wang & Wei, Inf-convolution, Optimal Allocations, and Model Uncertainty for Tail Risk Measures, SSRN 3490348 (version of January 16, 2022), p. 9, proof of Theorem 1, displays (10) and (11)
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.