Essays on sell-side analysts’ non-GAAP and GAAP reporting

  • M Zhu

Student thesis: Doctoral ThesisDoctor of Philosophy (PhD)


This thesis seeks to assess the practical implications of equity analysts’ non-GAAP and GAAP earnings reporting to different capital market participants. It consists of three essays that are presented as chapters.

The first empirical chapter uses hand collected data from analysts’ reports for large European banks and examines analysts’ actual and forecast GAAP and non-GAAP earnings per share (EPS). It finds that there is significant variation among sell-side analysts’ non-GAAP actual earnings measures. These measures are not easily reconcilable to firms’ reported non-GAAP earnings, GAAP earnings or to street earnings reported by I/B/E/S. By contrast, reported measures of GAAP earnings in analysts’ reports rarely differ from one another or from firms’ reported GAAP earnings. When evaluated against analysts’ own actual non-GAAP earnings measures, forecasts appear more accurate and more biased than those based on I/B/E/S. This chapter suggests that although non-GAAP earnings measures are more persistent, GAAP earnings are less vulnerable to measurement ambiguity across analysts. Therefore, whether non-GAAP or GAAP earnings are superior involves a trade-off of persistence and measurement uncertainty.

The second empirical chapter studies the relative informativeness at the earnings announcement date and post earnings announcement drift (PEAD) associated with GAAP and non-GAAP earnings surprises. Previous studies misalign GAAP actual earnings with non-GAAP forecasts to measure GAAP earnings surprises. This chapter overcomes this measurement error problem by aligning the measurement bases of both forecast and actual GAAP earnings. It finds that investors still perceive non-GAAP earnings to be more informative than GAAP earnings at the earnings announcement date. However, previously identified GAAP earnings surprises downwardly bias market responses to GAAP earnings. In addition, after correcting the measurement error, the evidence suggests that the GAAP-based PEAD is higher than the non-GAAP based PEAD, indicating that investors may not use the information contained in GAAP earnings as efficient as non-GAAP earnings.

The third empirical chapter explores analysts’ disagreement on GAAP earnings forecasts, forecast exclusions and their relationship with future stock returns. In this chapter, the non-GAAP forecasts are separated into two components: GAAP forecasts and forecast exclusions. It finds that both analysts’ disagreement on non-GAAP forecasts and disagreement on GAAP forecasts are negatively associated with future stock returns. However, a higher level of disagreement on forecast exclusions is associated with higher future stock return. Further evidence suggests that dispersion in forecast exclusions reflects the firm idiosyncratic risk and the uncertainty of fundamental firm value.

Overall, the thesis extends the literature on analysts’ non-GAAP reporting by demonstrating that analysts disagree significantly on how actual non-GAAP earnings are measured. It also complements the literature on market reactions to analysts’ forecasts and the literature on analyst dispersion anomaly by provides additional evidence on the implications of analysts’ GAAP forecasts and forecast exclusions for capital market participants.
Date of Award11 May 2021
Original languageEnglish
Awarding Institution
  • The University of Bristol
SupervisorMark A Clatworthy (Supervisor) & Tuan Q Ho (Supervisor)

Cite this