| Summary: | ABSTRACT
Prior research has been widely documented that financial analysts earnings forecast are not consistent with rational expectations. For example, analysts do not efficiently use public information to predict earnings and they may issue forecasts that are systematically optimistic. Therefore, this dissertation, by using different regression-based tests, i.e. ordinary least squares (OLS) regression and least absolute deviation (LAD) regression, aims to test the efficiency of financial analysts forecasts.
Based on the data from UK, my results indicate that, on the assumed quadratic loss function (use OLS to test the rationality of analysts forecasts), analysts inefficient incorporate information on prior earnings, extreme past earnings changes, previous forecast revisions, prior forecast errors and past stock returns. In particular, my estimated coefficients based on OLS regression are far from their predicted values under the rational expectations hypotheses. This is consistent with prior research that analysts forecasts do not inefficiently incorporate all available information with their personal judgments (e.g. DoBondt and Thaler, 1990; Easterwood and Nutt, 1999; Mendenhall, 1991; Ali et al. 1992; Abarbanell and Bernard, 1992). In remarkable contrast, results from the assumed linear loss function (use LAD to test the rationality of analysts forecasts) show that the LAD estimated coefficients are closer to their predicted values under rational expectations hypotheses, which indicate no virtual evidence of forecast inefficiency in analysts forecasts.
|
|---|