Consistent and Non-Degenerate Model Specification Tests Against Smooth Transition Alternatives

In this paper we develop a test of functional form that is consistent against any deviation from the null specification, and directs power towards a class of nonlinear "smooth transition" functional forms. Of separate interest, we provide new details regarding when and whether consistent parametric tests of functional form are asymptotically degenerate. A test of linear autoregression against smooth transition alternatives is never degenerate. Moreover, a test of Exponential STAR has surprising power and non-degeneracy attributes entirely associated with the choice of threshold. In a simulation experiment in which all parameters are randomly selected the proposed test has power nearly identical to the most powerful tests for true STAR, neural network and SETAR processes, and dominates popular tests. We apply the test to monthly U.S. output, money supply, prices and interest rates.