Abstract The time-series behavior of the quarterly earnings, sales and expense series of 69 firms over the 1946-74 period is examined. A Box-Jenkins time-series methodology is adopted. Based on inspection of the cross-sectional autocorrelation function, it is concluded that each series has (a) an adjacent quarter-to-quarter component and (b) a seasonal component. One-step-ahead forecasting results reveal that these two components can be successfully modelled at the individual firm level. The use of various quarterly forecasting models in security price analysis is also examined. The results are consistent with the market adjusting for seasonality in quarterly earnings in interpreting each quarter's earnings change.
G. N. Foster (Sat,) studied this question.