In Investor Inattention, Firm Reaction, and Friday Earnings, authors Stefano Della Vigna and Joshua Pollet examine Friday earnings announcements and weekend distraction as a possible reason for post-earnings announcement drift. Vigna and Pollet conjecture that investors don’t react as quickly to Friday earning announcements because they become preoccupied with other upcoming weekend activities. Managers of companies with the short-term stock price in mind take advantage of the weekend distraction factor by reporting below expected earnings on Friday.
Using 127,099 earnings announcements between January 1994 and June 2005, Vigna and Pollet study characteristics of investor attention for each weekday and firms that announced on Friday as a way to understand the hypothesized difference in investor responses to different announcement dates and possible motives for a company’s choosing of an announcing date. For investor response, the authors look at short-term and long-term stock price response and trading volumes.
To evaluate the over-all stock price volitility, Vigna and Pollet study the short term (day 0 –day 1) response as well as the long term (day 2- day 75) response. For the short term analysis, reporting of earnings are associated with 12-21% lower stock price volatility, meaning the stock price doesn’t move as much in reaction as it would on another announcement date, a findings consistent with investor inattention. In addition, the trading volume for the two days surrounding the announcement is 10% lower, even after holding for variables like consistent lower than average Friday volume. However, two days after the reporting (Tuesday), stock prices respond more to Friday earning surprises.
While non-Friday announcements have 40-45% of the stock response delay, Friday announcements have 56-64% delay. The authors calculate delayed response as the response from the third day to the 75th day divided by the eventually stock response of the entire 75 days. For Friday surprises, only 56-64% of the eventual stock price is created in the first two trading days. On the aggregate level, this delayed above average reaction makes up for the unresponsiveness in the first two days, making the long term response equal to that of any other announcement date. Long term stock prices are not affected by short term investor inattention.
These findings lead the authors into the more interesting question of what are the motives of Friday earning reporting firms. Although only 5.7% of all announcements are made on Fridays, the majority of these announcements are likely to be a negative earnings surprise, meaning the announcement is below the expected or projected earnings created by Wall Street analysts. Earnings reported on Friday are 45% more likely to fail at meeting expectations. As Friday surprises are twice as likely to be negative, the authors concluded that firms release negative announcements on Fridays as part of a strategic move. With stock prices responding less to Friday announcements than other weekday announcements, the authors hypothesize that companies that look to publish on Fridays are more interested in the short term performance of the company’s stock. Firm managers acknowledge the effect investor inattention has on the stock price and use it to their advantage in the short term. However, this strategy is only short term due to the delayed rebound of investors reacting to the earning reporting. Managers interested in long-term performance usually stick to a set schedule for reporting, knowing that moving announcement to Friday won’t matter in the long-run. The market cap of companies reporting on Friday is 26% smaller, with a possible explanation being smaller companies are more concerned about the short term.
These findings help the authors explain post-earnings announcement drift.
Overall, I found this paper extremely informative with good information, good data and good explanations. While the idea of weekend distraction affecting investor response to earning reporting is intuitive, understanding the data and how the authors calculated the inattention was the interesting section. Many parts were very technical with many regressions, but the authors did a great job putting in small explanations of what each regression outcome would mean in terms of the hypothesis. In addition, whenever the authors referred to past working papers, they gave a brief description, which helped greatly to help orient the read to the relevance of the past paper’s information and findings.
However, the paper had a few assumptions in the models that I think were a stretch. Only companies that had analyst coverage were used in the data set. This problem is two fold. To calculate the earning surprise, the authors used analyst expectations. By nature, analysts can only be right 50% of the time or everyone would beat the market by following analysts’ predictions. By assuming analyst predictions were the correct expected earning, the authors take in a huge error reading. Even within analyst coverage, analysts can have varying earnings predictions. In addition, the only publicly available analyst predictions are those of the sell-side. People on the buy-side spend their time figuring out which analysts are wrong and by how much. That is how mutual funds make any money. If sell-side analysts were always correct, mutual funds wouldn’t be profitable. So using the sell-side analyst expectations isn’t the greatest method. However, I understand that using analyst estimations is the best guess of future earnings, though the method still has its own negative aspects.
The other side of using companies that have analyst coverage is that the data set is skewed, possibly changing the results. Companies that don’t have coverage are usually smaller, newer and unproven. Taking out these companies gives Vigna and Pollet a look at investor inattention to larger firms. If Vigna and Pollet are correct in thinking that smaller companies have more short-term goals and announce on Fridays, the addition of these data sets could help deepen their findings.