A0160
Title: Informative priors to estimate the value-at-risk
Authors: Mario M Pizarro - Universidad de Extremadura (Spain) [presenting]
Eva Lopez Sanjuan - Universidad de Extremadura (Spain)
M Isabel Parra Arevalo - Universidad de Extremadura (Spain)
Abstract: In Risk Theory, the use of the tools provided by Extreme Value Theory is essential to estimate risk measures, but usually, only the observations that exceed a certain fixed value are taken into account for the estimation. Value at Risk (VaR) and Conditional Value at Risk (CVaR) are the most employed risk measures. A new Bayesian method, based on Metropolis-Hastings (MH) algorithm, is proposed in order to estimate VaR. The method employs informative prior distributions for the parameters of the Generalized Pareto distribution (GPD), using all the datasets and consequently, seizing all the information available. In order to compare the quality of the estimates provided, a broad simulation study is carried out for different distributions of the data. The results show that the new strategy provides better estimates for VaR than standard MH with non-informative priors.
