A1083
Title: Markov-switching quantile autoregression: A Gibbs sampling approach
Authors: Richard Luger - Laval University (Canada) [presenting]
Xiaochun Liu - University of Central Arkansas (United States)
Abstract: We extend the class of linear quantile autoregression models by allowing for Markov-switching effects in the location of the conditional quantiles. We also propose a Gibbs sampling algorithm for posterior inference by using data augmentation and a location-scale mixture representation of the asymmetric Laplace distribution. An estimate of the marginal likelihood is available as a by-product of the procedure, since all complete conditional densities used in the Gibbs sampler have closed-form expressions. We use the marginal likelihood estimates across different probability levels to determine the order of a stepwise re-estimation procedure which solves the well-known quantile crossing problem. Our method to enforce non-crossing quantiles can be applied in any quantile regression model with endogenous or exogenous covariates, and whether Markov-switching effects are allowed for or not.
