CMStatistics 2020: Start Registration
View Submission - CMStatistics
Title: Prediction analysis for extreme conditional quantiles through panel data quantile regression with individual effects Authors:  Xuan Leng - Xiamen University (China) [presenting]
Yanxi Hou - Fudan University (China)
Abstract: The aim is to study the estimation and prediction of extreme conditional quantiles of panel data based on individual-effect models. We propose a two-stage method, where the first stage is implemented based on the panel quantile models at an intermediate level and the second stage is based on the extrapolation method for an extreme level. The method relies on a set of second-order regular variation conditions of heteroscedastic extremes, which is used for the establishment of asymptotic normality for extreme conditional quantiles. Finally, a simulation study is evaluated to show the finite performance of the estimator, and a real data analysis are conducted for the illustration of the method.