A Reexamination of Bias in Management Earnings Forecasts

If managers base their voluntary earnings forecasts on their best estimates of future earnings; then, on average, management earnings forecasts should be evenly distributed around firms' subsequent earnings reports. If, however, firms' forecasts tend to be more frequently above or below the actual subsequent earnings, then this systematic pattern may be the result of a systematic 'bias' in management earnings forecasts. Prior research shows that management earnings forecasts are, on average, unbiased. This study finds that short-term forecasts (three months or less forecast horizon) are pessimistically biased, while long-term forecasts (eight months or greater forecast horizon) are optimistically biased in terms of both magnitude and frequency. There exists a temporal trend in the bias. Managers start optimistic in their forecasts and then move to pessimism as the end of the year approaches. Previous claims that management earnings forecasts are unbiased could be driven by the cancellation effect that may occur between the long-term optimistic bias and the short-term pessimistic bias. In addition, our evidence suggests that the magnitude of the bias is associated with ex ante information - the unexpected earnings in the forecasts, firm size, forecast horizon, firm performance, and growth potential of the firm. Since this information is publicly available when a forecast is announced, investors and analysts may use it to predict the bias component included in the management earnings forecast.