Analysts' GAAP earnings forecasts and their implications for accounting research

We use newly available GAAP forecasts to document that traditionally-identified GAAP forecast errors contain 37% measurement error. Correcting for this measurement error, we settle a long-standing debate regarding investor preference for GAAP versus non-GAAP earnings and provide strong evidence of a preference for non-GAAP earnings. We also revisit the use of non-GAAP exclusions to meet analysts' forecasts when GAAP earnings fall short. Results indicate that 34% of these traditionally-identified meet-or-beat firms are misidentified due to measurement error, and this error masks evidence that firms more frequently exclude transitory rather than recurring expenses for meet-or-beat purposes.