Title: Inference for shape constrained quantile regression splines
Authors: Thomas Parker - University of Waterloo (Canada) [presenting]
Abstract: Inference methods are proposed for nonparametric quantile regression estimates based on series methods that are subject to shape constraints such as that the conditional quantile curves are monotone or convex in an explanatory variable. Constraints can also be imposed to maintain monotonicity in quantile level. Hypotheses such as linearity of conditional quantile curves against the alternative of convex curves are considered across quantile levels. Inference is nonstandard but can be conducted using resampling.