Beschreibung:
We discuss implementation problems and test approaches for using the full-information approach for estimating divisional betas. For both full-information and pure-play estimates, variations using capital structure adjusted betas perform better than those with raw betas, and those with betas estimated using the Mamp;M no-tax type capital structure adjustment are better than those using the Hamada type adjustment. The choice of divisional weights does not make any statistically significant difference. We find that estimates from the pure-play approach are better than those from the full-information approach