Probability of informed trading around scheduled and unscheduled corporate announcements

Abstract

This thesis examines how public announcement events with different characteristics affect the probability of informed trading (PI). Using the Bollen, Smith and Whaley (2004) model of inferring PI directly from trades, we investigate the differences in PI between the pre-announcement period and post-announcement period from 2002 to 2008 in the American stock market along two dimensions: whether announcements are scheduled; and other characteristics related to the content such as payment methods, offer premium (takeover announcements) and earnings surprises (earnings announcements). We first focus on unscheduled takeover announcements. Our results show that PI (and bid/ask spread) is significantly higher in the pre-announcement period compared to that in the post-announcement period. Further, we link the changes in PI to takeover announcement characteristics. We show that PI is significantly higher in the pre-event period for successful offers and cash offers as well as offers with relatively high premiums. In contrast, the changes in PI after announcement are not significant for unsuccessful offers, stock or mixed offers as well as offers with relatively low premiums. We then investigate informed trading around scheduled earnings announcements. Results indicate that PI (and bid/ask spread) before earnings announcements is not significantly higher than after the announcement. However, when breaking down the sample according to the size of the earnings surprise, we find significant incremental PI in the pre-earnings period when reported earnings contain big surprises with respect to the forecasts. Finally, we contrast PI around public announcements, conditioning on whether the announcement is scheduled or unscheduled. We find that the level of PI for scheduled earnings announcements are significantly higher than for unscheduled takeover announcements in the post-announcement period but not in the pre-announcement period. Our results are consistent with the argument that an announcement that is anticipated stimulates relatively more private information gathering, and hence the degree of information asymmetry around scheduled announcements might be higher. We also find that trading volume dries up in the pre-announcement period for scheduled earnings announcements. -- provided by Candidate.