Market (In)Attention and the Strategic Scheduling and Timing of Earnings Announcements

The following post comes to us from Ed deHaan of the Accounting Area at Stanford University; Terry Shevlin, Professor of Accounting at the University of California, Irvine; and Jake Thornock of the Department of Accounting at the University of Washington.

In our paper, Market (In)Attention and the Strategic Scheduling and Timing of Earnings Announcements, forthcoming in the Journal of Accounting and Economics, we revisit a long-standing but still unresolved question: do managers “hide” bad earnings news by announcing during periods of low market attention? Or, conversely: do managers “highlight” good earnings news by announcing earnings during periods of high market attention? We posit three necessary conditions for an effective hiding/highlighting strategy. First, to be able to hide bad news, managers must change their earnings announcement (“EA”) timing somewhat frequently. A deviation from a long-standing pattern of EA timing could attract attention to the very news the manager is trying to hide. Second, there must be variation in market attention that is predictable to the manager ex-ante—random variation in attention would not allow for strategic timing of bad or good news. Third, we must observe that managers do tend to announce more negative (positive) earnings news during periods of lower (higher) market attention. We also examine an additional potential strategy for reducing attention to bad news: by scheduling EAs with less advance notice or “lead-time.”

We construct a novel database of over 120,000 precise EA dates and times to investigate the timing of EAs by hour and by weekday. We find that firms frequently change the timing of their EAs: for example, within any given year, 81.6% of firms change their quarterly EA weekday at least once, and 25.6% change whether they report before, during, or after market hours. Although just 7.6% of all EAs happen on Fridays, 51.4% of firms have at least one Friday announcement during our sample period. Managers’ intent is unobservable, but the majority of these changes in EA times are likely made for administrative, scheduling, or other benign reasons. The high frequency of benign changes is precisely the camouflage needed for managers to occasionally switch their EA timing for strategic purposes without raising alarm, consistent with the first condition.

We turn next to the second condition: that lulls and peaks in market attention must be ex-ante predictable in order for managers to shift bad (good) news into times where the market is paying less (more) attention. We examine three specific times during which prior research has speculated that market attention differs: before versus after the close of regular trading hours; on Mondays through Thursdays versus Fridays; and on “slow” versus “busy” news days, based on the total number of firms that are reporting earnings. We employ four empirical proxies to investigate temporal variation in market attention: (i) the number of earnings-related news articles; (ii) the speed with which analysts incorporate the earnings news into future earnings forecasts; (iii) EDGAR 8-K downloads; and (iv) abnormal Google search volume. The advantage of these attention measures is that they are user-oriented, related to information processing, and measured on a timely basis.

Our findings are consistent with attention being lower (higher) after (before) market close and on busy (slow) reporting days. Specifically, EAs after trading hours (on the busiest reporting days) are associated with a 7% (12%) decrease in news articles, a 7% (4%) decrease in the speed with which analysts update forecasts, a 19% (30%) decrease in EDGAR downloads, and a 0% (2%) decrease in Google searches, after controlling for firm fixed effects, non-stationary firm characteristics, and the sign and magnitude of the earnings news. However, our analyses generally indicate that attention is actually no different on Fridays than other weekdays. We also tie together our first two conditions by showing that the act of switching EA timing (condition 1) does not draw unwanted attention to bad news (condition 2). In sum, the data are consistent with the second condition that market attention is predictably lower after hours and on busy reporting days, but the same does not hold for Fridays.

Next, we examine the validity of the third necessary condition: that managers tend to release worse (better) earnings news after market close, on Fridays, and on busy reporting days (and vice-versa). Prior research has examined whether earnings news is worse after hours and on Fridays, but the results to date have been mixed (see Patell and Wolfson 1982; Damodaran 1989; and Doyle and Magilke 2009). Using our larger sample of precise EA times, we find that unexpected earnings (i.e., IBES earnings less analyst consensus, scaled by price) are 70% lower when firms choose to report earnings after hours, 190% lower on Fridays, and 100% lower on the busiest versus slowest reporting days. We also find evidence consistent with managers both switching to periods of low expected attention to hide bad news, as well as with switching to periods of higher expected attention to highlight good news.

In a final section, we examine variation in earnings news and market attention depending on the earnings announcement scheduling “lead-time;” i.e., how far in advance of the EA the firm schedules the date and time of the forthcoming EA. Using novel data on EA scheduling announcements for roughly 55,000 firm-quarters, we find that longer (shorter) lead-times are associated with better (worse) earnings news. In addition, we find that lead-times are positively associated with two of our four attention proxies, providing some evidence that firms are able to reduce attention to bad earnings news by providing less forewarning of the EA. We also find that three-day abnormal returns around the scheduling dates are significantly negative when the forthcoming EA is scheduled for a Friday, which is consistent with investors inferring that Friday EAs tend to contain negative news. Returns are insignificant for EAs scheduled for after hours or busy days, although we do observe a negative stock price drift in the interim period between the scheduling date and EA date.

The full paper is available for download here.

Both comments and trackbacks are currently closed.