Modeling Inflation Series with the Parametric and Semiparametric Transition Models and Comparison of Their Forecasting Performances

Abstract

Transition models are frequently used in the modeling of nonlinear series. Until recent years, it is seen that the parametric approach in which regime transitions are approximated with well-defined mathematical functions is often used. Over the last two years, nonparametric and semiparametric approaches have also been introduced into the transition models. In this case, the transition function is estimated in a nonparametric manner. In this study, the inflation series based on TUFE (1987 = 100) for the period January 1995 - December 2008, is also estimated via semi parametric transition model in addition to the parametric approach and their forecasting performances are compared.