On the estimation of non-stationary functional series tarma models

Maximum Likelihood (ML) and Multi-Stage Weighted Linear / Non-Linear Least Squares (MS-WLLS / MS-WNLS) estimation methods are formulated for non-stationary Functional Series Time-dependent ARMA (FS-TARMA) models. The methods' effectiveness, as well as their superiority over Ordinary Linear / Non-Linear Least Squares (OLLS / ONLS) type methods not accounting for innovations serial heteroscedasticity, are demonstrated via Monte Carlo experiments.