Abstract: We establish conditions for the existence and invertibility of fractionally
differenced ARIMA time series whose innovations are in the domain of attraction of an
-stable law with and consequently have infinite variance. More importantly, we
study the effect of truncation on the minimum dispersion linear predictor of based on
the infinite past We verify that the truncated predictor based on the
finite past is asymptotically efficient, and derive asymptotic bounds on the rate
of convergence to 1 of the efficiency of The bounds are shown to decay like power
functions with the rate of decay depending on the index of stability and the difference
parameter