Analysts' street earnings forecasts are sometimes based on GAAP earnings and sometimes based on non-GAAP earnings, which exclude various GAAP earnings components. Therefore, differences in analysts' street earnings forecastsmay capture differences in not only expected performance but also the earnings metric forecasted. We argue that analysts who forecast non-GAAP, rather than GAAP, street earnings are more likely to separately analyze earnings components. Consistent with this argument, we find that analysts who forecast non-GAAP street earnings issue relatively more accurate forecasts. We also argue that excluded earnings components often reflect negative transitory items, and that variation across analysts in the earnings metric forecasted suggests that the negative excluded items are forecasted by only a subset of analysts. Consistent with this assertion, we find that variation across analysts in the earningsmetric forecasted is associatedwith a lower consensus GAAP earnings surprise and lower stock returns around the earnings announcement. Finally, although variation in the earnings metric forecasted is a source of analyst forecast dispersion, we find that it is also incrementally associated with a lower earnings response coefficient, consistent with the existence of transitory items.We therefore find that the variation in the earnings metric forecasted is an important source of analyst forecast dispersion that predicts not only a lower earnings surprise but also a lower earnings response.
|Number of pages||20|
|State||Published - Aug 2023|
Bibliographical noteFunding Information:
History: Accepted by Brian Bushee, accounting. Funding: Financial support from the Gatton College of Business & Economics, the KPMG Fellowship at the Mendoza College of Business, and the Goizueta Business School is gratefully acknowledged. Supplemental Material: The data files are available at https://doi.org/10.1287/mnsc.2022.4553.
© 2022 INFORMS.
- street vs. GAAP
ASJC Scopus subject areas
- Strategy and Management
- Management Science and Operations Research