A0508
Title: Bootstrap-based forecasting in quantile autoregression
Authors: Silvia Novo - Universidad Carlos III de Madrid (Spain) [presenting]
Cesar Sanchez-Sellero - University of Santiago de Compostela (Spain)
Abstract: Novel approaches are introduced for constructing prediction intervals within quantile autoregression models, applicable both in the homoscedastic case and in more general settings. The methodology relies on quantile estimation, the use of multiplier bootstrap techniques to capture the variability in coefficient estimation, and the generation of bootstrap-based forecasts. Theoretical consistency of the proposed procedures is established. A simulation study and an empirical application are included to assess the performance of the methods in finite samples and to benchmark them against existing alternatives in the literature.
