Title: Exact inference in predictive quantile regressions
Authors: Sermin Gungor - University of Western Ontario (Canada) [presenting]
Richard Luger - Laval University (Canada)
Abstract: An exact simulation-based procedure is developed to test for quantile predictability at several quantile levels, jointly. The approach proceeds by combining the quantile regression $t$-statistics and uses Monte Carlo resampling techniques to control the overall significance level in finite samples. As a by-product our procedure also yields an exact distribution-free confidence interval for the persistence parameter of a first-order autoregressive model, assumed for the predictor variable. We employ the new procedure to test the ability of many commonly used variables to predict the quantiles of excess stock returns.