A0706
Title: Constancy testing in varying coefficient models
Authors: Luis Antonio Arteaga Molina - Universidad de Cantabria (Spain) [presenting]
Juan Manuel Rodriguez-Poo - Universidad de Cantabria (Spain)
Abstract: The aim is to propose a Gaussian process (GP) approach for constancy testing in varying coefficient models. This methodology leads to a general unified framework of kernel-based tests having the following properties: (i) bootstrap tests are easy to implement in the presence of nuisance parameters (they are simple quadratic forms, and there is no need to re-estimate the nuisance parameters in each bootstrap replication); and (ii) the tests are valid under general conditions, including regularized estimators (e.g. Lasso) or parameters at the boundary of the parameter space. Neyman-orthogonal kernels are used, and an asymptotic theory and a detailed local power analysis are developed. Monte Carlo experiments and a real data application illustrate the sensitivity of tests to the dimension of covariates.
