B0902
Title: Bayesian inference for generalised quantiles
Authors: Valeria Bignozzi - University of Rome Sapienza (Italy) [presenting]
Mauro Bernardi - University of Padova (Italy)
Lea Petrella - Sapienza University of Rome (Italy)
Abstract: We investigate analytical properties and financial meaning of generalised quantiles used as risk measures. We compare them with classical examples of risk measures such as Value-at-Risk (VaR), Expected Shortfall and Conditional Tail-Variance. Thanks to their elicitability property we are able to extend the concepts of quantile and asymmetric least square regression used for VaR and expectiles to generalised quantiles and provide a link with the kernel of the Asymmetric Exponential Power Distribution (AEPD). This distribution can then be used to perform Bayesian inference for the estimation of generalised quantiles.
