Indirect inference for ARFIMA processes

摘要

Inference for ARFIMA processes is more complicated than that for ordinary ARMA models. This has limited the application of these processes in the practice of modelling economic time series. In this paper, a procedure of indirect estimation for fractionally integrated processes is proposed and discussed. A Monte Carlo experiment shows that indirect estimators are not inferior to other estimators proposed in the literature, and could provide a reliable framework for extending fractionally integrated models in new directions and widen their applicability.