Bayesian Inference for a Threshold Autoregression with a Unit Root
Melbourne Institute Working Paper No. 20/06
Date: October 2006
Author(s):
Abstract
A Bayesian approach to distinguishing between nonlinear and unit root behavior offers several practical advantages over equivalent frequentist procedures. Foremost among these advantages is the simplicity of the test. This paper compares the small sample power and size properties of a joint Bayesian test for a unit root and a threshold effect with Caner and Hansen's (2001) frequentist strategy. The results from Monte Carlo experiments indicate that the simpler Bayesian test performs at least as well as Caner and Hansen's procedure.