Supervisors: Ingo Kleindienst & Yulia Muratova
Discussants: Pernille Smith & Panos Mitkidis

Abstract
Research has recently started to explore the impact of sentiments in textual data on firm behavior and firm outcome, for example, in the context of domestic acquisitions (Yang et al, 2019) and stock returns (Sun et al, 2016). Following social learning theory environmental and personal determinants and their interaction influence human behavior (Bandura, 1977; Latham and Saari, 1979). Starting from an understanding that sentiments in texts are one specific form of environmental determinant, we argue that sentiments expressed in texts affect personal thoughts and, as an extension, behavior. Following this logic, we investigate the impact of sentiments in textual data on investor reactions to the announcement of cross-border acquisitions. Specifically, we examine investor reactions as proxied by acquirer cumulative abnormal returns (CARs) to cross-border merger and acquisition (M&A) announcements made by US firm for the period 2005 to 2022. To proxy sentiments, we rely on GDELT (Global Database of Events, Language, and Tone), a global open-source platform that monitors and analyzes media from around the world and applies natural language processing and machine learning algorithms to identify and extract sentiments from a variety of news sources. The results reveal that sentiments expressed in textual data on targets’ home-countries have a statistically significant effect on investor reactions to cross-border M&A announcements of US firms. Specifically, the more positive the sentiment expressed in textual data towards targets’ home countries, the higher the CARs. We also observe that the effect of sentiment on CARs is contingent on the global sentiment. When the global sentiment is more positive than the sentiment specific to the US and host-country dyad, CARs tend to be worse.  

Everyone is welcome!