Search

Corporate takeovers — the purchases of a company by another — are significant economic events. They affect the parties involved in the transaction, inducing relevant dynamics in the enterprises' lifecycle, and are also relevant at an aggregate level for the whole economy: takeovers can produce efficiency gains and improved capital allocation, while, on the other hand, they can also increase the market power of specific companies and hamper competition. In this paper, we propose a logic-probabilistic reasoning framework to study the determinants of company takeovers and predict future ones. In particular, we model the domain of interest as a logic-based knowledge graph, where the extensional knowledge contains facts concerning company ownership structures and the characteristics of the shareholders, and the intensional knowledge encodes a set of takeover suitability criteria in the form of reasoning rules, whose conditional dependencies are modeled with a Bayesian network. Our rules are expressed in Vadalog, a language of the Datalog+/- family. Our framework revolves around a data engineering process that allows eliciting the takeover determinants from a corpus of anecdotal cases, refining and encoding them into logic rules, and finally combining their outcomes. We implement and operate the framework in the Vadalog System, a state-of-the-art reasoner, and apply it to the knowledge graph of the Italian companies of the Central Bank of Italy. We provide an extensive experimental evaluation. [ABSTRACT FROM AUTHOR]