ABSTRACT We exploit analysts’ forecasts of nonearnings measures and a granular accruals dataset to assess different explanations for why accruals are associated with analyst forecast errors. We evaluate four explanations: one accounting-based (the estimation error hypothesis) and three economics-based (related to investment activity, demand slowdowns, and product-market shocks). We find earnings forecast errors are stable over time, almost completely explained by revenue and cash flow errors, span multiple years, and are more pronounced in the presence of product-market shocks. Using a novel dataset of accruals not available in Compustat, we find no evidence that accruals with higher reporting discretion explain analyst forecast errors. Collectively, our evidence suggests that analysts’ forecast errors are best explained by a positive correlation between accruals and product-market shocks, indicating that these errors stem from the challenge of predicting how economic shocks affect future earnings rather than a failure to understand accounting principles or detect earnings management. Data Availability: The data used in this study are commercially available. JEL Classifications: G1; G14; M4; M41.
Davidson et al. (Sun,) studied this question.