Anticipation and Returns in Event Studies

When an event is anticipated, the firm's stock return around the announcement of the event may have an inconsistent sign: a positive sign around negative news, or vice versa. We attempt to quantify the frequency of this problem, first with a brief mathematical model and simulation, then with empirical tests that support the model, employing data on analyst upgrades. We find that more than 10% of returns can have inconsistent signs, even with low levels of anticipation. Researchers should adjust their event-study data to properly account for anticipation