Modelling multiperiod patterns in stock-market reactions to events, with an application to serial acquisitions

Abstract

In event studies, the now standard window of a few days may miss relevant price movements if the market’s reaction to the news announcement tends to be slow or if the initial reaction tends to be partially or wholly undone afterwards. We propose a parsimonious hybrid of splines and Almon lags to detect and classify various patterns of post-event reactions spread over many periods. The scheme can interact with one or more event characteristics (like deal size), and the resulting non-linear model can be estimated via maximum likelihood (ML).In our application, we study the returns pattern that follows takeover announcements by two leading serial acquirers, AB Inbev SA and Heineken. Our method confirms the presence of a drop-and-recovery pattern as reported in Doan and Sercu (2021), but the amplitude of the pattern shows no link with deal size. This last finding is not in line with the view that the pattern reflects a rise in uncertainty that is slowly resolved (Malatesta and Thomson, 1985).