The autoregressive fractionally integrated moving average (ARFIMA) model ha
s become a popular approach for analyzing time series that exhibit long-ran
ge dependence. For the Gaussian case, there have been substantial advances
in the area of likelihood-based inference, including development of the asy
mptotic properties of the maximum likelihood estimates and formulation of p
rocedures for their computation. Small-sample inference, however, has not t
o date been studied. Here we investigate the small-sample behavior of the c
onventional and Bartlett-corrected likelihood ratio tests (LRT) for the fra
ctional difference parameter. We derive an expression for the Bartlett corr
ection factor. We investigate the asymptotic order of approximation of the
Bartlett-corrected test. In addition, we present a small simulation study o
f the conventional and Bartlett-corrected LRT's. We find that for simple AR
FIMA models both tests perform fairly well with a sample size of 40 but the
Bartlett-corrected test generally provides an improvement over the convent
ional test with a sample size of 20.