A0458
Title: Conditional expectile-based risk measures
Authors: Cecile Adam - KU Leuven (Belgium) [presenting]
Irene Gijbels - KU Leuven (Belgium)
Abstract: Among the main interests in regression analysis is to explore the influence that covariates have on a variable of interest, the response. There is extensive literature on flexible mean regression, in which the targeted quantity is the conditional mean of the response given the covariates. Quantile regression is another method that aims at estimating the conditional median or other quantiles of the response variable given the covariates. An alternative to quantiles are expectiles. Expectile regression estimates the conditional expectiles of the response variable given realized values of the predictor variables. After a brief introduction to expectiles and to univariate nonparametric expectile regression, we discuss the application of expectiles in risk management. Some risk measures are defined using the expectile regression framework and the estimators of these measures are established. The performance of these conditional risk measure estimators is investigated via simulations, and is illustrated on real data.
