Identifying M&A Targets from Textual Disclosures: A Transformer Neural Network Approach

Nils Lohmeier, Lennart Stitz, and Paul Drecker

♦ Can textual information from firm disclosures help to identify M&A targets? We employ two state-of-the-art transformer neural networks, RoBERTa and Longformer, based on 102,987 annual reports of publicly listed US firms to estimate takeover likelihoods. We show that incorporating publicly available, standardized textual information can improve the predictability of corporate takeovers significantly in out-of-sample tests and that this information is not fully incorporated in stock prices. We use explainable artificial intelligence methods to examine the reasons for the improved predictions. Our analyses indicate that the machine learning algorithms are able to identify product offerings and firm-specific capabilities sought by acquirers.

Read it here.