The history of analysts’ forecasts is rich (Brown 1993). In some respects, it is the
successor to the time-series of earnings literature (Beaver 1970; Ball and Watts 1972). The
early literature focuses on which time-series model most accurately forecasts eamings.
Identifying the process tells us something about the general characteristics of the accounting
numbers (e.g., seasonality and adjacent quarter-to-quarter effects). Moreover, researchers
use earnings forecasts derived from these models as inputs into other forms of research
(e.g., we can use earnings forecast errors in security retums studies). Analysts‘ earnings
forecasts are natural candidates for more accurate forecasts because they can reflect a richer
information system than simply the past eamings series. One of the original purposes,
leaming about the features of the accounting system, has withered. However, the literature