A pseudo-linear method for fractionally integrated ARMA (ARFIMA) model estimation

A pseudo-linear method for the estimation of fractionally integrated ARMA (ARFIMA) models is introduced. The method uses a long binomial series expansion of the fractional differencing operator, as well as the relationship of the AR/MA parameters and binomial expansion terms with the model´s inverse function. It is based upon a pseudo-linear formulation motivated by the fact that this relationship leads to a special-form regression problem that can be decomposed into a scalar non-linear and a multiple linear regression. The performance characteristics of the method are demonstrated via Monte Carlo experiments and comparisons with the frequency-domain maximum likelihood method