Quantifying corporate culture using machine learning and 10-K filings

Abstract

Corporate culture plays a critical role in the success and integration of organizations, particularly during mergers and acquisitions (M&A). This dissertation aims to extract and quantify elements of corporate culture from 10-K textual data and subsequently apply these quantifications to real-world scenarios. To explore this question, a comprehensive dataset consisting of 68,855 Management Discussion and Analysis sections from 10-K Filings stemming from 12,553 companies was utilized. The study employs advanced NLP techniques, including word embedding and sentiment scoring using Term Frequency-Inverse Document Frequency (TF-IDF), to generate a culture dictionary and identify as well as quantify linguistic patterns indicative of corporate culture. The findings demonstrate that NLP-driven quantification of corporate culture can provide valuable insights for addressing cultural integration in M&A scenarios. By scoring the core cultural values innovation, integrity, quality, respect, and teamwork, stakeholders can make more informed decisions, potentially improving the success rates of M&A activities.

Date of Award	2 Jul 2024
Original language	English
Awarding Institution	Universidade Católica Portuguesa
Supervisor	Dan Tran (Supervisor)