months. On average we find that just over half of all companies have some news
reported each month, ranging from 30% at the beginning of the sample to 70%
currently. We find that news stocks have a 60% chance of having more news in a
subsequent month (be it good/bad news) and no-news stocks have a 30% chance.
We also find a statistically lower incidence of news reported on Fridays compared
with the rest of the working week. Along a similar vein, academic research typically finds
that Mondays and Fridays are light information days compared with Tuesdays and
Thursdays and suggests that the short-term price and volume reactions to earnings
announcements on Fridays are smaller and tend to drift more, attributed to lower
investor attention on Fridays. Interestingly, some studies find that firms tend to release
bad news after the close of trading on Fridays due to lower investor attention.
The coverage of news is an increasing function of the size of the company since larger
and more liquid companies are more likely to be in the media. Based on our sample, we
find correlations between the log market cap/6-month average daily trading volume and
news citations of around 35%. The top-size quintile each month accounts for 40% of all
news articles, while the bottom quintile accounts for just 5%. Coverage has also
benefited over time from the better collection of news by data vendors and quarterly
reporting post the introduction of IFRS in 2005.
8.4 Does news flow matter?
The challenge of running long-term event studies with news flow datasets is how to
control for the release of subsequent news items which may or may not be in the same
direction. We find that the incidence of news is not highly autocorrelated. A company
can switch from being a news winner to a news loser several times over a year. The
average proportion of stocks switching from a news winner to loser (and vice versa) is
fairly equal. This means that any news patterns are likely to be influenced by single news
events rather than the accumulated reaction to multiple items.
Our focus is therefore on the short-term reaction to news. We measure the short-term
price reaction on a sector-relative basis around news announcements which aims to
capture the informational surprise of the news. We sort returns into three groups and
categorize news flow as either positive, negative, or marginal to avoid reacting on every
reported news item and focus on the major news releases. We then measure subsequent
returns over days 2–5 and 5–10 post the announcement.
Figures 8.5 and 8.6 show the average short-term returns for different news items,
distinguishing between good and bad news. Our results show that investors react
strongly to earnings announcements and guidance news, which is not unsurprising as
investors focus more on these news items when making their assessments. The share
price reaction to other news is also important but lesser so compared with earnings-
related news. Over the next five days, the payoffs are marginally positive, but relatively
similar. Similar inferences can be made about negative news releases.
The results above show that there is a hierarchy to news citations, with earnings-
related news having a larger impact. We also consider whether news influences trading
volumes beyond the day of the release, by comparing abnormal trading volumes around
announcements. We find that, following positive news, trading volumes are significantly
higher for earnings and M&A-related news. When news releases are bad, we find
The impact of news flow on asset returns: An empirical study 219