Private and public sanctions : investigating cartel decisions using Twitter data

Abstract

This thesis investigates whether the announcement of cartel decisions by the European Commission provides new information for investors and if Twitter data can be used to explain abnormal returns. The dataset consists of 39 cartel cases involving 124 different companies from January 2010 to May 2021. Using a standard event study methodology, we find evidence that supports previous studies findings and confirm that variables such as fines, geographic location, and the size of a company impact abnormal returns in relation to the European commission’s cartel decision. These variables are confirmed important by the use of single-factor regression and decision trees. The Twitter variables were not found to have any explanatory power on abnormal returns. A statistical significant cumulative abnormal return in the event window [-15,15] of -2.29% was found in the sample containing all fined companies. We also find that companies that receive immunity from the European Commission have no significant cumulative abnormal returns on average.