Article Abstract:
A model associating unexpected earnings with risk-adjusted security returns was systematically and empirically tested for specification errors. The unexpected earnings response regression (UERRM) model was examined to detect possible nonlinearity, residual nonnormality, heteroscedasticity, omitted variables, and systematic and random coefficient variation across firms. The specification testing used three samples of earnings forecasts, obtained from IBES, 'Value Line' and COMPUSTAT, as substitutes for expected earnings. Results showed the simple regression of unexpected earnings response to have serious specification problems. There was evidence, however, that these errors may be lessened with the application of several methods, including transformations and testing for heteroscedasticity.
User Contributions:
Comment about this article or add new information about this topic:
Article Abstract:
The importance and measurement of pension variables in bond-rating decisions is examined. Measures of net pension asset are included as an independent variable in the bond rating prediction model. Measures of the pension obligation, including the obligation reported by the corporation and obligations recalculated for interest discount rates, are calculated to determine which measures affect bond ratings. Results indicate that these pension measures had no significant effect on bond ratings in 1980, but pension measures using standardized interest rates had a significant effect on bond ratings in 1981 and 1982.
User Contributions:
Comment about this article or add new information about this topic:
Article Abstract:
Cready and Shank's (CS) claims in relation to Abdel-khalik and McKeown's (AM) 1978 research on accounting changes in an efficient market are characterized as incorrect. The three claims are: that AM used an incorrect statistical technique, that AM incorrectly used the statistical technique chosen, and that CS employed a correct statistical technique. It is suggested that CS do not comprehend the nature and qualities of analysis-of-variation designs that employ a blocking factor combined with one or more crossed factors.
User Contributions:
Comment about this article or add new information about this topic: