SIMULATING MANAGEMENTS EARNINGS-PER-SHARE FORECASTS

This study attempts to simulate management's earnings-per-share forecasts using six naive time series models. The results indicate two (random walk and random walk with a drift) of the six models were significantly more accurate than the other four models in simulating management's earnings forecast. Both exhibited a tendency to consistently underforecast earnings, and it was hypothesized that this could be due to an inherent bias in the data. The results for the remaining four models were not as favorable. The two submartingales and the simple linear trend models exhibited a high degree of variability in accuracy and significant biases based on the magnitude of the forecast. The arithmetic average model appeared to be unbiased but exhibited a high degree of variability. The results of the study indicate that it may be possible to simulate management's earnings forecasts using relatively naive mathematical models.