. Sample selection and descriptive data
Earnings guidance directly communicates managers’ private beliefs about the future earnings implications of all their
private information. It is perhaps for this reason that providing earnings guidance increased in popularity throughout the
1990s.3 However, in recent years the practice of providing earnings guidance appears to be declining—surveys by NIRI
consistently reported around 79% of respondents as providing some form of earnings guidance through 2003 (Thompson,
2003a, 2003b) but that proportion declined to 51% in 2007 (Kelleher, 2007). Thus, the practice of providing guidance, while
still relatively widespread, appears to be declining.
4.1. Theory and empirical predictions
As noted before, theoretical models generally predict that firms will disclose good news above a certain threshold and
withhold bad news below this threshold (Dye, 1985; Verrecchia, 1983). By announcing the decision to stop providing
earnings guidance, managers are indicating that no disclosure will be made for some period of time going forward—i.e., they
are committing to a policy of non-disclosure. Such a commitment is beneficial because in future quarters, when a firm that has
committed to a policy of non-disclosure does not provide guidance, the market’s interpretation of this non-disclosure should
be less negative than for firms who do not make such a commitment.18 This reasoning is similar in spirit to the theoretical
model developed in Einhorn and Ziv (2008), who suggest that managers can enhance their reputation of being uninformed by
withholding information and thereby reduce the negative price impact of future non-disclosure. While their model is not
specifically about committing to a policy of non-disclosure, the multi-period nature of their model highlights the fact that
disclosure decisions in one period impact the market’s interpretation of non-disclosure in future periods. In other words,
managers can develop a disclosure reputation that mitigates negative inferences about non-disclosure in future periods.
Because there are likely reputational costs to a manager from breaking his/her commitment to non-disclosure, we
conjecture that managers will publicly announce the decision to stop guidance only if (1) it is unlikely that they will have good