Do Analyst Earnings Forecasts Allow for Accounting Conservatism?

Recent studies show that accounting earnings are conservative, i.e., earnings tend to reflect bad news (negative stock returns) on a timelier basis than good news (positive stock returns) (Basu, 1997); moreover, the degree of conservatism is negatively associated with the price-to-book (P/B) ratio (Pae et al., 2003). If analysts correctly allowed for conservatism and its documented association with the P/B ratio, differences in earnings conservatism would be unassociated with analysts' forecast error. In contrast, we find that average yearend forecast error differs between good news and bad news firms, and between high and low P/B firms. We conclude that analysts' earnings forecasts do not fully incorporate the implications of earnings conservatism. We also find that forecast dispersion is greater for bad news than good news firms, and greater for low than high P/B firms, consistent with the hypothesis that accruals used to accelerate the recognition of bad news spawn disagreement about forthcoming earnings.