Estimation of Periodic Bilinear Time Series Models
Abdelouahab Bibi, Moon‐Ho Ringo Ho · Communication in Statistics- Theory and Methods · 2006
In this article, a new class of models is proposed for modeling nonlinear and nonstationary time series. This new class of models, referred to as the periodic bilinear models, has a state space representation and can be characterized by a set of recursive equations. Condition for the stationarity is presented. Procedures for parameter estimation using the cumulants of order less than four are described and the accuracy of the proposed method is demonstrated in the Monte Carlo simulations.