Natural Language Processing for Exploring Culture in Finance: Theory and Applications

Abstract

Natural language processing (NLP) has found its way into financial data analysis. In this chapter, the authors argue for the prominence of taking NLP approaches to the study of culture in finance. To this end, the chapter first surveys some of the NLP algorithms, including bag-of-words, TF-IDF, sentiment analysis, cosine similarity, word embeddings, and topic models, with the illustration of their implementation in R—an open-source statistical language. Second, the authors demonstrate the usefulness of NLP to finance text mining by analyzing Warrant Buffet’s letters to Berkshire Hathaway’s shareholders from 1977 to 2019. Results show that there exist identifiable communication patterns in the letters. These findings are crucial for enhancing the understanding of the company’s corporate culture and financial decision-making. The chapter is concluded by offering future research directions and opportunities.