56 Pages Posted: 16 Oct 2013 Last revised: 27 Sep 2014
Date Written: October 4, 2013
We introduce a variant of the smooth transition autoregression - the GSTAR model - capable to parametrize the asymmetry in the tails of the transition equation by using a particular generalization of the logistic function. A General-to-Specific modelling strategy is discussed in detail, with particular emphasis on two different LM-type tests for the null of symmetric adjustment towards a new regime and three diagnostic tests, whose power properties are explored via Monte Carlo experiments. Four classical real datasets illustrate the empirical properties of the GSTAR, jointly to a rolling forecasting experiment to evaluate its point and density forecasting performances. In all the cases, the dynamic asymmetry in the cycle is efficiently captured by the new model. The GSTAR beats AR and STAR competitors in point forecasting, while this superiority becomes less evident in density forecasting, specially if robust measures are considered.
Keywords: Dynamic Asymmetry, Smooth Transition, Testing, Estimation,
JEL Classification: C22, C51, C52
Suggested Citation: Suggested Citation
Zanetti Chini, Emilio, Generalizing Smooth Transition Autoregressions (October 4, 2013). CEIS Working Paper No. 294. Available at SSRN: https://ssrn.com/abstract=2340539 or http://dx.doi.org/10.2139/ssrn.2340539