HIGH LEVEL QUANTILE ESTIMATIONS OF SUMS OF RISKS

Andrés Cuberos (1) , Esterina Masiello (2) , Véronique Maume-Deschamps (2)

1 SAF - Laboratoire de Sciences Actuarielle et Financière
2 ICJ - Institut Camille Jordan

The estimation of high level quantile or of the expectation over a high quantile (Value at Risk - VaR - or Tail Value at Risk - TVaR - in risk management contexts) is crucial for the insurance industry. We propose a new method to estimate high level quantiles of sums of risks. It is based on the estimation of the ratio between the VaR of the sum and the VaR of the maximum. We use results on consistently varying functions. We compare the e ciency of our method with classical ones, on several models.

Mots clés: consistently varying functions; value at risk estimation; risk aggregation

Domaines: Statistiques [math.ST]; Théorie [stat.TH]

Andrés Cuberos, Esterina Masiello, Véronique Maume-Deschamps. HIGH LEVEL QUANTILE ESTIMATIONS OF SUMS OF RISKS. 2014. ⟨hal-01020597v1⟩