Predicting Greek mergers and acquisitions: a new approach

In this paper, we investigate the possibility of predicting takeover targets in Greece, which is an incipient market for acquisitions. Our work is based on recursive partitioning techniques, that is decision-tree models, given that takeover likelihood models (such as logit) are not robust over time (Powell, 1997). However, we adopt a new technique with respect to Espahbodi and Espahbodi (2003), the machine learning algorithm J4.8 that is a new application in the sector of mergers and acquisitions. The results show that J4.8 outperforms the classical regression tree, although the predictive accuracy is not promising.