An is supposed to move a share price when a company reports profit different from what was expected, and that is the founding assumption of an efficient market. But investor attention is limited, and on some days dozens of companies publish at once. What happens then?
The reaction arrives late. Counting how many other companies published on the same day showed that on crowded days the price and the reacted less to a company’s own . A then appeared, continuing in the direction of the news. Available information does not enter a price at once when attention is divided.
The measure the authors built is what turns the claim into a test. Crowding is counted by the number of announcements from other companies that day, including unrelated industries. That matters: a company in another sector competes for nothing but attention, which makes it a clean gauge of attentional load.
The result is under-reaction. On crowded days the price and the trading volume reacted less to the company’s earnings surprise, and the weakening stood out especially where the surprise was large, which is precisely when attention is needed. That is the reverse of what an efficient market should do.
Then comes the drift. After the announcement the price keeps moving in the direction the news already pointed, for weeks. This strengthens the explanation: the information was neither lost nor rejected, it entered the price slowly because nobody had time to process it on the day it appeared.
And two caveats. Crowding is not assigned entirely at random, and companies can choose a publication date for reasons connected to their news, for instance to bury a weak report on a busy day. A strategy exploiting the gap carries risk and costs, and it is no promise of profit once the pattern has been published.
wat. On a day crowded with reports, the market reacted less to one company’s news and completed part of the reaction later.
wat wat
A market full of computers still depends on human and institutional attention. The fact that information was published is no guarantee that it is working yet.
How do we know?
The authors joined every earnings report, its price reaction and its trading volume to the number of other announcements published that same day.
What is still unknown
Crowding is not assigned entirely at random, and companies can choose a publication date for reasons connected to their own news.
Learn something?
Where to next
What happened to a share after more people searched for it online?markets
Does the choice of a share pass between neighbors?markets
Do market anomalies pay more when everyone is watching?markets