| Summary: | 碩士 === 國立臺灣科技大學 === 財務金融研究所 === 104 === For companies and investors, analysts’ earnings forecasts are a key issue. During forecast periods, analysts will overestimate or underestimate earnings. When the actual earnings are announced, there will be discrepancies between forecasts and actuals, which are called standardized unexpected earnings (SUE). The delayed effect of SUE may cause post-earnings announcement drifts (PEAD), and will result in positive or negative cumulative abnormal returns (CAR).
This research studies the relationship between forecast bias and SUE in order to use forecast bias to estimate SUE. Furthermore, this research also studies if there are any associations between forecast bias trends and CAR in the United States of America. If so, forecast bias could be used to estimate CAR and stocks’ trends after earnings announcements. The sample periods of quarterly earnings forecasts are between January 2010 and December 2015.
One of the results of this study shows that during forecast periods, forecast bias data and the median of forecast bias per week, fluctuate in quarterly periodical trends. In general, the median of forecast bias shows a walk-down trend which is consistent with the “sobering up” phenomenon. Furthermore, at the beginning of forecast periods, analysts tend to be more optimistic than pessimistic.
This paper also proves that there is a positive correlation between short terms initial forecasts bias and short terms CAR, while there is a negative correlation between long terms initial forecasts bias and long terms CAR. Therefore, estimating short terms CAR by short terms initial forecasts bias is more proper than estimating them by long terms’.
|
|---|
