Prior studies use a linear adaptive expectations model to describe how anal
ysts revise their forecasts of future earnings in response to current forec
ast errors. However, research shows that extreme forecast errors are less l
ikely than small forecast errors to persist in future years. If analysts re
cognize this property, their marginal forecast revisions should decrease wi
th the forecast error's magnitude. Therefore, a linear model is likely to b
e unsatisfactory at describing analysts' forecast revisions. We find that a
,lon-linear model better describes the relation between analysts' forecast
revisions and their forecast errors, and provides a richer theoretical fram
ework for explaining analysts' forecasting behaviour. Our results are consi
stent with analysts' recognizing the permanent and temporary nature of fore
cast errors of differing magnitudes. Copyright (C) 2000 John Wiley & Sons,
Ltd.