This paper provides an explanation for the puzzling phenomenon in Tieslau c
t al. (1996, Journal of Econometrics 71, 249-264) that a substantial effici
ency loss occurs if low-order autocorrelations are omitted when estimating
the differencing parameter, d. This is because for all it strictly bigger t
han I,the nth-order autocorrelation function does not depend uniquely on th
e differencing parameter. We construct a new estimator for the differencing
parameter based on the partial autocorrelation function. Comparisons of th
e asymptotic and finite-sample variance of our estimator and those of TSB a
re made. A substantial efficiency gain is achieved by our estimator as comp
ared to TSB's. (C) 2000 Elsevier Science S.A. All lights reserved. JEL clas
sification: C22.