Bayes factors and nonlinearity: Evidence from economic time series (Q1305670)

This paper argues in favor of a Bayesian approach to evaluating evidence of nonlinearity in economic time series over the classical approach that has been dominant in the applied literature. An application is presented concerning nonlinearity in US GNP. The outline of the rest of the paper is as follows. Section 2 gives some formal definitions of Bayes factors and outlines the issue of prior sensitivity for a general class of nonlinear time series models. Section 3 discusses the advantages of a Bayesian approach to nonlinear time series modeling. Section 4 contrasts Bayesian methods to classical ones for a simple threshold model. Section 5 discusses various computational techniques for calculating marginal likelihoods and Bayes factors. Section 6 discusses the formulation of priors and carries out an empirical exercise where the Bayes factors for two nonlinear time series models of US GNP are calculated. The last section concludes.