This paper reviews and synthesizes empirical work in accounting assessing the time-series
properties of firms exhibiting nonseasonal quarterly earnings characteristics. The focus on
nonseasonal firms is warranted given the recent trend documented by Lorek, Willinger, and
Bathke (2008) showing that 35.6% of their sample firms exhibit nonseasonal quarterly earnings
characteristics. Moreover, 43.6% of their nonseasonal firms were not covered by security
analysts. Therefore, analysts, investors, and researchers interested in obtaining quarterly
earnings forecasts for such firms must employ statistically-based forecasting models thereby
increasing the importance of this topic. Yet, use of the seasonal ARIMA models on nonseasonal
firms described by O’Hanlon (1995) results in seasonal over-differencing, parameter
redundancy, and a violation of the principle of parsimony – not to mention a significant decline
in predictive ability. Finally, the paper also discusses specific statistically-based forecasting
models (i.e., the AR (1) model and, more recently, the random-walk model) which have
improved predictive ability of quarterly earnings for such nonseasonal firms.