【文章內(nèi)容簡介】 le firms 17 sample firms make no voluntary disclosures 16 firms make only one voluntary disclosure 4 firms make 13 or more disclosures Descriptive statistics Descriptive statistics + 42% So Quarterly disclosures tend to convey bad news while Annual disclosures tend to convey good news. Results —— relation between forecast horizon and sign of news HI: Bad news earnings disclosures are more likely to relate to quarterly earnings releases and less likely to relate to annual EPS numbers, and conversely for good news earnings disclosures. 3749414=266 43% 57% 22% 78% 70% 30% 67% 33% Null hypothesis There is no relation between the sign of the news and forecast horizon Results —— relation between form of disclosure and sign of news 52% 39% Results —— bias caused by the Q4 disclosures Q1 Q2 Q3 More often bad earning news The Q4 disclosure can be labeled as either quarterly or annual earnings This may lead to bias of the results. Q4 The relation between the sign of the earnings news and forecast horizon remains strong Whether there are more bad news in Q4? more bad news is disclosed in the Q4 47 of the 140 bad news disclosures occur in the Q4 good news disclosure is the same more good news is disclosed in the Q4 There is no asymmetry in the result of table 2 Results —— conclusions of Hypothesis 1 Disclosed as an annual point or range forecast of EPS good news Disclosures are qualitative and related to quarterly earnings announcements. bad news ? Bad news disclosures are more likely to be motivated by a need to preempt large negative quarterly earnings surprises. ? Good news disclosures are more likely to be driven by a desire to signal more generally that the firm is doing well. H2: The probability the information conveyed by a given quarterly earnings announcement will be preempted by a voluntary corporate disclosure is higher for relatively large negative earnings surprises than for other types of earnings information. Results —— whether negative earnings are more likely to be preempted Not Preempted Preempted Other earnings Large negative earnings Results —— whether negative earnings are more likely to be preempted Quantify the news Current period earnings Prior period earnings Earnings change Numbers from the current period earnings announcement Numbers from the corresponding period earnings announcement prior year Proxy for the market expectation Stock price Stock price 60 trading days prior to the earnings announcement date Sample selected 2,647 firmquarters and 292 earnings announcements Bad news is preempted more often than other types of earnings news Total 2,647 292 (11%) The binomial test indicates that the very bad group proportion is significantly different from the other three proportions. Somehow prove that Results —— whether negative earnings are more likely to be preempted Very bad news is preempted most often Otherwise there is no relation between the magnitude of earnings news and the probability of preemption. Bad news is still not preempted very often Results —— whether negative earnings are more likely to be preempted Results —— why bad news is still not preempted very often Possible explanation Seasonalrandomwalk model uses information that is about a year old, it is likely to measure the market39。s expectation of earnings with error. Measurement error in the earnings surprise Some very bad news may be fully expected by the market, in which case there is no need for preemption. Example Logit regression The dependent variable is coded 1 if the announcement is preempted and 0 otherwise. （ 1） News: the change in the splitadjusted quarterly earnings per share( from the corresponding quarter of the prior year) deflated by stock price 60 trading days prior to the earnings announcement. Result： The coefficient on News is negative and statistically significant (the tstatistic is ), indicating that the worse the earnings news, the more likely it is to be preempted. Results Logit regression （ 2） Bad News Indicator: an indicator variable that takes the value 1 for earnings announcements of a decline in splitadjusted quarterly EPS of at least .008, and 0 otherwise. Result： the coefficient is positive and statistically significant (the tstatistic is ), indicating that, consistent with H2, large negative surprises are more often preempted than other types of earnings news. Results Logit regression （ 3） The third regression includes both independent variables. Result： the coefficient on the dummy variable remains positive and statistically significant (the tstatistic is ), while the coefficient on News is negative but not reliably different from zero. The results in tables 3 and 4 suggest that managers face an asymmetric loss function in formulating their discretionary disclosure policies, consistent with the legal liability/reputationeffects arguments Results The stock price respond to volun